Supplement to Science

AAAS is here connecting government to the scientic community.

As a part of its efforts to introduce fully open government, the White House is reaching out to the scientic community for a conversation around Americas national scientic and technological priorities. To enable the White Houses dialogue with scientists, AAAS launched Expert Labs, under the direction of blogger and tech guru Anil Dash. Expert Labs is building online tools that allow government agencies to ask questions of the scientic community and then sort and rank the answers they receive. On April 12, 2010, AAAS asked scientists everywhere to submit their ideas to the Obama administration and at the same time launched the rst of Expert Labs tools, Think Tank, to help policy makers collect the subsequent responses. The result was thousands of responses to the White Houses request, many of which are already under consideration by the Ofce of Science and Technology Policy. As a AAAS member, your dues support our efforts to help government base policy on direct feedback from the scientic community. If you are not already a member, join us. Together we can make a difference.

To learn more, visit aaas.org/plusyou/expertlabs

PREFACES

2 Into the Depths

Sean Sanders

3 Revisiting the Ocean's Carbon Cycle

Farooq Azam and Nianzhi Jiao

CONTENTS

CHAPTER ONE
5 The Invisible Hand Behind A Vast Carbon Reservoir
Richard Stone Farooq Azam Hiroshi Ogawa, Yukio Amagai, Isao Koike, Karl Kaiser, Ronald Benner Ronald Benner, J. Dean Paulski, Matthew McCarthy, John I. Hedges, Patrick G. Hatcher Farooq Azam and Alexandra Z. Worden Matthew D. McCarthy, John I. Hedges, Ronald Benner Javier Arstegu, Carlos M. Duarte, Susana Agust, Marylo Doval, Xos A. lvarez-Salgado, Dennis A. Hansell

7 Microbial Control of Oceanic Carbon Flux: The Plot Thickens 9 Production of Refractory Dissolved Organic Matter by Bacteria 13 Bulk Chemical Characteristics of Dissolved Organic Matter in the Ocean 16 Microbes, Molecules, and Marine Ecosystems 18 Major Bacterial Contribution to Marine Dissolved Organic Nitrogen 22 Dissolved Organic Carbon Support of Respiration in the Dark Ocean 23 Community Genomics Among Stratified Microbial Assemblages in the Ocean's Interior
Edward F. DeLong, Christina M. Preston, Tracy Mincer, Virginia Rich, Steven J. Hallam, Niels-Ulrik Frigaard, Asuncion Martinez, Matthew B. Sullivan, Robert Edwards, Beltran Rodriguez Brito, Sallie W. Chisholm, David M. Karl

31 Variability in Radiocarbon Ages of Individual Organic Compounds from Marine Sediments

Timothy I. Eglinton, Bryan C. Benitez-Nelson, Ann Pearson, Ann P. McNichol, James E. Bauer, Ellen R. M. Druffel

35 Lipid-Like Material as the Source of the Uncharacterized Organic Carbon in the Ocean
Jeomshik Hwang and Ellen R. M. Druffel Lihini I. Aluwihare, Daniel J. Repeta, Silvio Pantoja, Carl G. Johnson

38 Two Chemically Distinct Pools of Organic Nitrogen Accumulate in the Ocean

CHAPTER TWO
43 Microbial Carbon Pump and its Significance for Carbon Sequestration in the Ocean
Nianzhi Jiao and Farooq Azam Ronald Benner and Gerhard J. Herndl Michal Koblek Carol Robinson and Nagappa Ramaiah Markus G. Weinbauer, Feng Chen, Steven W. Wilhelm Craig A. Carlson, Dennis A. Hansell, Christian Tamburini

46 Bacterially Derived Dissolved Organic Matter in the Microbial Carbon Pump 49 Role of Photoheterotrophic Bacteria in the Marine Carbon Cycle 52 Microbial Heterotrophic Metabolic Rates Constrain the Microbial Carbon Pump 54 Virus-Mediated Redistribution and Partitioning of Carbon in the Global Oceans 57 DOC Persistence and its Fate After Export Within the Ocean Interior 60 Molecular Characterization of Dissolved Organic Matter and Constraints for Prokaryotic Utilization
Gerhard Kattner, Meinhard Simon, Boris P. Koch

62 Shedding Light on a Black Box: UV-Visible Spectroscopic Characterization of Marine Dissolved Organic Matter
Colin A. Stedmon and Xos Antn lvarez-Salgado Joy D. Van Nostrand and Jizhon Zhou Marie Eichinger, Jean-Christophe Poggiale, Richard Sempr

64 Application of Functional Gene Arrays (GeoChips) in Monitoring Carbon Cycling 66 Toward a Mechanistic Approach to Modeling Bacterial DOC Pathways: A Review

Front and back cover images adapted from Jiao, N. et al., Nature Reviews Microbiology, 8, 593 (2010). Front image (background) istockphoto.com/BobHemphill 2011 by The American Association for the Advancement of Science. All rights reserved. 13 May 2011

Into the Depths

Microbes. Those unassuming and unseen inhabitants, occupying almost every niche on land and sea, have once more been found to have importance far beyond their physical stature. Microbes have been found living thousands of meters above sea level on Mount Everest, as well as in the deepest oceans, withstanding extraordinary hydrostatic pressure. Many scientists are familiar with Thermus aquaticus, a thermophilic microbe that can survive temperatures upwards of 70C, discovered in a geyser in the Yellow Stone National Park and source of the first thermostable DNA polymerase, the eponymously named Taq. Microbes related to T. aquaticus even live at the deep-sea hydrothermal vents, often making use of sulfur from these vents as a source of energy. But in the water columns between the deep-sea vents and ocean surface lives an extraordinarily diverse range of interconnected microbial populations that are only now being characterized and understood. More importantly, scientists are uncovering new roles that these microbes play in the ocean carbon cycle and, consequently, in the global climate. This volume seeks to bring together both the historical and the current research on this topic. In the first chapter are presented a selection of papers published in the last 20 years in Science that have helped us understand the chemical makeup of dissolved organic matter (DOM), how the carbon contained in them enters and exits the ocean carbon cycle, and how ocean-dwelling microbes interact with this cycle. Originally suspected to only assimilate or respire DOM, both Bacteria and Archaea have been demonstrated to play a much broader and more important role, in fact being integral cogs in the carbon cycle machine. The recently described microbial carbon pump (MCP), pictured on the cover of this booklet, provides a framework that describes how microbes are also DOM creators and, more importantly, contributors to the creation of refractory DOM (RDOM), a persistent form of DOM that can survive for thousands of years, constituting a previously undescribed mechanism of carbon sequestration. The second chapter brings us right up to the present day with 10 review articles from some of the leading international scientists in a variety of fieldsincluding marine biogeochemistry, microbiology, and genomicswho put the past three decades of work in perspective and give us a glimpse of where the research might take us in the near future. Covering ground from describing possible ways to model the MCP, to the contribution of virus activity to the creation of RDOM, to discussions of the underlying biochemical and molecular mechanisms at play in the MCP, these articlesvetted and reviewed by fellow scientists within the Scientific Committee for Oceanic Research working group promise to be both enlightening and provocative. Understanding and fully characterizing the MCP provides us with critical knowledge needed to refine current models of the ocean carbon cycle so that we may better predict its response to increases in atmospheric CO2. Through this research, it should be possible to define the resilience of the ocean carbon cycle and its relationship to global climate. This will benefit climate change research as well as provide a robust scientific basis for crafting climate policy. Gaining a deeper understanding of these diverse and often unique microbial inhabitants in our oceans may also allow us to make them our allies as we face future uncertainties for our climate. Sean Sanders, Ph.D. Editor, Science/AAAS Custom Publishing

Dr. Sanders completed his undergraduate training at the University of Cape Town, South Africa, and his Ph.D. at the University of Cambridge, United Kingdom. Following postdoctoral training at the National Institutes of Health and Georgetown University, he worked for three years at a biotechnology startup before moving into editing. Dr. Sanders is currently the Editor, Custom Publishing for the journal Science and Science Careers, and Program Director for Outreach.

Revisiting the Ocean's Carbon Cycle

Whether the ocean could serve as a repository for anthropogenic CO2, as it has for other human wastes, has prompted intense interest and inquiry into ocean carbon biogeochemistry. Fully one-half of global photosynthesis occurs in the ocean. It generates huge amounts of organic matter that is acted on by diverse biological and physical forces that modify, decompose, and redistribute its constituents in the ocean space in timeframes of hours to millennia. Most carbon is rapidly respired back to CO2 by the diverse biota, while a minuscule fraction sinks, or is actively transported by biota, to the sea bottom (the biological pump). There, some of it is respired by the sediment microbes, while the remainder becomes part of the sediment. The biological pump has also been invoked in studies of the fate of anthropogenic pollutants entering the ocean, including radionuclides, to assess whether their association with sinking particles will rid us of them for long periods of time. We now know that a large fraction of photosynthetically produced organic matter in fact becomes dissolved organic matter (DOM) by a variety of physiological and trophic mechanisms; hence, we need to know the biogeochemical behavior of DOM as well. While microbes readily respire most newly produced DOM, might a small fraction be (or become) refractory and join the enormous refractory DOM (RDOM) pool present in ocean water column, with turnover times of millennia? Even minor changes in RDOM concentration could significantly affect carbon sequestration in the ocean with consequences for the climate. The origin and fate of RDOM, and the underlying mechanisms for its formation and degradation are unknown. Why marine microbes are not able to degrade RDOM is still under investigation though their inability to do so is fortunate: if all RDOM were respired and the carbon released to the atmosphere, it would double the atmospheric CO2 inventory. Looking at the other side of the equation we can ask: Do bacteria play a major role in the production of RDOM? Further, might the balance between microbial degradation of DOM and production of RDOM shift due to climate change and might the response of the RDOM pool exacerbate climate change impacts through a positive feedback loop? These are new challenges for climate change microbial ecology. Progress in elucidating the role of microbes in carbon sequestration in the ocean had been limited not only by methodological limitations, but also by the lack of a unifying biogeochemical framework. In recent years, there has been a powerful convergence of marine genomics, ecophysiology, and new tools for DOM analysis. Therefore, the time is ripe for a fresh look at this refractory and important problem. The microbial carbon pump (MCP; Jiao et al. 2010. Nat Rev Microbiol. 8, 593599) was proposed as a conceptual framework to formulate and test new hypotheses about DOM sources and sinks as well as the mechanistic bases for the regulation of microbial carbon storage in the refractory DOM pool. This research is necessarily highly interdisciplinary involving, at the minimum, microbial oceanographers, marine organic chemists, and geochemists. A SCOR (Scientific Committee for Oceanic Research) working group (WG134) joined by 26 scientists from 12 countries has been formed to address the problem. In the past two years, the working group has convened several independent workshops, as well as MCP-focused sessions in conjunction with oceanography and microbial ecology conferences. These brainstorming sessions have provided the necessary, and rapidly broadening, interdisciplinary interactions, creating much excitement and momentum. It is our hope that the articles in this book convey this sense of excitement, and an optimism that we are making progress in solving this long-standing problem in science, one which is also of considerable societal import. Farooq Azam, Ph.D. Nianzhi Jiao, Ph.D.
Dr. Farooq Azam studies microbial oceanography and marine biogeochemistry. Dr. Azam and his students have made significant contributions to our understanding of the role of microbes in the functioning of marine ecosystems and carbon cycle. Dr. Azam is a Distinguished Professor at Scripps Institution of Oceanography, University of California, San Diego and co-chairs the Scientific Committee for Ocean Research (SCOR) Working Group 134 on Microbial Carbon Pump in the Ocean. Dr. Nianzhi Jiao is Cheung Kong Chair Professor at Xiamen University, China. After receiving his Ph.D. from Ocean University of Qingdao in 1991, he continued his studies at MIT in the United States, the University of Tokyo, and the National Institute for Environmental Studies, Japan. Dr. Jiaos research focuses on microbial ecology and carbon cycling. He is a co-chair of the Scientific Committee for Ocean Research (SCOR) Working Group 134 on Microbial Carbon Pump in the Ocean.

Chapter One

NEWSFOCUS
M A R I N E B I O G E O C H E M I ST RY M A R I N E B I O G E O C H E M I ST RY

The Invisible Hand Behind The Invisible Hand Behind A Vast Carbon Reservoir A Vast Carbon Reservoir A key element of the carbon cycle is the microbial conversion of dissolved
organic carbon into inedible forms. it also serve to sequester of COdissolved ? A key element of the carbon cycle Can is the microbial conversion 2 organic carbon into inedible forms. Can it also serve to sequester CO2?

XIAMEN, CHINAFor simple sea creatures, The ocean surface is like a planetXIAMEN, CHINA simple sea The ocean is aaplanetdissolved organic carbon (DOC) thecreastaff sized set of surface lungs that inhale and XIAMEN, CHINA For For simple seais creatures, The ocean surface islike like planettures, organic carbon (DOC) is sized of lungs that inhale . As a global the of life.dissolved Much of it, however, is as unpalatable exhale CO dissolved organic carbon (DOC) is the staff sized set set of lungs that average, inhale and and 2 a global average, the staff of life. Much of however, is as exhale oceansCO take up about 2% more ofthe the as and accumulates init, the water column. 2.. As As a global average, the of chaff life. Much of it, however, is as unpalatable exhale CO 2 oceans take up about 2% more of the unpalatable as chaff and accumulates in disgas than they release. Some CO Scientists are unraveling how organic matter as chaff and accumulates in the water column. oceans take up about 2% more of 2 the disthan they release. CO the water column. are unravelsolves into the waterSome column, formin the marine food Scientists chain is converted into gas 2 disgas than they release. Some CO Scientists are unraveling how organic matter 2 into water column, forming how organic matter in is the marine food ing carbonic acid. As atmospheric forms that lessfood readily relinquish carbon in solves solves into the the water column, formin the marine chain converted into acid. As atmospheric chain is converted into forms that less read). The exislevels rise, ocean decreases, CO the form of less carbon dioxide (CO ing 2carbonic carbonic acid. AspH atmospheric forms that readily relinquish carbon in ing 2 levels rise, ocean pH decreases, a CO ily relinquish carbon in the form of carbon tence of this inedible organic carbon in the a phenomenon called acidification the form of carbon dioxide (CO2). The exis- CO22 levels rise, ocean pH decreases, ). The existence of this inedcalled acidification that dioxide (CO ocean has been known for quite some time. that could endanger corals and other 2inedible tence of this organic carbon in the phenomenon a phenomenon called acidification ible organic carbon in for thecarbon ocean has been endanger coralscorals and But its role in the global cycle has could creatures by slowing theother growth of ocean has been known quite some time. that could endanger and creaother known for quite some time. But its role in has the tures by slowing the growth carbeen recognized only recently, says Michal carbonate skeletons (see p. of 1500). But its role in the global carbon cycle creatures by slowing the growth of global carbon cycle hasrecently, been recognized only skeletons (seethe Science 18 June Koblizek, a microbiologist at thesays Institute of bonate Carbon also enters seas through been recognized only Michal carbonate skeletons (see p. 1500). recently, says a micro - 2010, p. 1500). Carbon also enters the Microbiology inMichal Trebon,Koblek, Czech Republic. the food web: During photoKoblizek, a microbiologist at the Institute of Carbon also enters the seas through biologist at the of Microbiology in seas through the During foodweb: DurNew ndings are unmasking the invisible synthesis, phytoplankton Microbiology inInstitute Trebon, Czech Republic. the food web: photoTrebon, Republic. photosynthesis, phyorganic carprocesses that suspend immensethe amounts of ing xes CO2 tophytoplankton NewCzech ndings are unmasking invisible synthesis, New findings are unmasking the invisi toplankton fixes CO bonas much as 60 carbon just below the ocean waves. Its really 2 processes that suspend immense amounts of xes CO2 to organic carorganic carbon ble processes that suspend immense amounts gigatons of carbon huge. Its comparable to all the carbon diox- to DOC doc. Jiao Nianzhi formulated bonas much as 60 carbon just below the ocean waves. Its really much 60carbon giga of carbon just the ocean waves. Its as per year, as roughly the ide in Its thecomparable air,below says Jiao Nianzhi, a microthe MCP concept based on his studies gigatons of huge. to all the carbon dioxDOC doc. Jiao Nianzhi formulated of carbon per really huge. Its comparable to all the car- tons same amount fixed bial ecologist here at Xiamen University. of an unusual kind of studies photoper year, roughly the ide in the air, says Jiao Nianzhi, a microtheAAPB, MCP concept based on his year, roughly the bon dioxide in the air, says Jiao Nianzhi, a synthetic bacteria ( left ). on land. The carbon He and others are exploring the tantalizing bial ecologist here at Xiamen University. same amount fixed of AAPB, an unusual kind of photoamount fixed microbial ecologist here at Xiamen Univer - same in tantalizing this reseris captured for prospect of sequestering CO synthetic bacteria (left). on not land. The carbon He and others are exploring the 2 TheKoblizek. carbon sity. He are exploring the tantalizvoir. Itsand too early to say whether the vast pool on long, The realization that refractory this reseris land. not says captured for prospect of others sequestering CO2 in in this reserThe realization that not new captured for ing prospect of CO2 the will respond tosequestering geoengineering, says Dennis Most marine bioDOC is arealization key element in refractory the global voir. Its too early to say whether vast pool is long, says Koblizek. The that refractory voir. Its too early to say whether the vast pool DOC is a key element in the global long, says Koblek. Hansell, a marine biogeochemist at the mass is consumed in days carbon cycle has lit a re under efforts to will respond to geoengineering, says Dennis Most new marine bioDOC is a key element in the global will respond to geoengineering, Dennis carbon cycle has are fire under new marine bioUniversity Miami in Florida. says However, he Most and returned to the in air as CO2. gure out what the stuff is and where it efforts comes Hansell, aof marine biogeochemist at the mass is consumed days carbon cycle has litlit a under efforts to Hansell, a of marine biogeochemist at the to figure outthe what the stuff is and where ishowever, consumed in days Some, ends up the. deep ocean from. says, I expect the light come on over heads Researchers now know that refractory University Miami into Florida. However, he mass and returned to the air asin CO gure out what stuff is and where it comes 2 . University of Miami in Florida. However, he and comes from. of Researchers now know that returned to ends the CO2 sink, when remains ofair dead fall to it and well experience anto ah ha!on moment. DOC consists thousands of compounds, Some, however, up as inorganisms the deep ocean says, I expect the light come over heads from. Researchers now know that refractory Some, however, ends up in the deep ocean says, I expect the light to come on over heads refractory DOC consists of thousands of the sea oor. Each year, this biological pump Data from several research cruises have such as complex polysaccharides and humic sink, when remains of dead organisms fall to DOC consists of thousands of compounds, and well experience an ah ha! moment. when remains of million dead fall to compounds, and well experience an ah ha! moment. such as polysacchadeposits roughly 300 tons of carbon yielded afrom broad-brush view of what Jiao has sink, acids. A team led bycomplex Xos Antn lvarez the sea oor. Each year, thisorganisms biological pump Data several research cruises have such as complex polysaccharides and humic seaseabed. floor. Each300 year, this biological pump rides Data a from several research cruises have and humic acids. Ade team led by Xos in the dubbed the microbial carbon (MCP): Salgado the led Instituto Investigaciones deposits roughly million tons of carbon yielded broad-brush view ofpump what Jiao has the acids. A of team by Xos Antn lvarez roughly 300 million tons of carbon yielded a broad-brush view pump ofof what Jiao deposits lvarez Salgado of Instituto de Even more massive amounts carbon Antn the microbe-driven conversion bioavailMarinas in Spain, has tracked the conin the seabed. dubbed the microbial carbon (MCP): Salgado ofVigo, the Instituto de the Investigaciones in the seabed. has dubbed the microbial carbon pump Investigaciones Marinas in Vigo, Spain, has are suspended in the water column as DOC. able organic carbon into dif cult-to-digest version of some forms of bioavailable carbon Even more massive amounts of carbon Marinas in Vigo, Spain, has tracked the conthe microbe-driven conversion of bioavailmore massive amounts ofas carbon (MCP): the microbe-driven conversion of The the conversion of somechanges forms of oceans hold an water estimated 700 billion forms known as refractory DOC. This sumto refractory carbon byof observing in areEven suspended in the column DOC. tracked able organic carbon into dif cult-to-digest version of some forms bioavailable carbon suspended in column as DOC. bioavailable organic carbon into This difficultto refractory carbon by tons of carbon asthe DOCmore than all land bioavailable mer, the European Project on Ocean Acidi - are their optical carbon properties: Humic substances The oceans hold anwater estimated 700 billion forms known as refractory DOC. sumto refractory carbon by observing changes in to-digest forms known as on refractory DOC.- The an estimated 700 all billion inre-emit their optical properbiomass puthold together (600 billion tons of observing cation is European carrying out a slate of experiments absorb UVchanges light and it as blue uotons oceans of carbon as DOCmore than land mer, the Project Ocean Acidi their optical properties: Humic substances This summer, the European Project the on tons of carbon as DOCmore all land ties: Humic substances absorb light and carbon) and as much asthan all the CO in Arctic waters that includes probing rescence at speci c wavelengths. biomass put nearly together (600 billion tons of cation is carrying out a slate of experiments absorb UV light and re-emit itUV as blue uo2 Ocean Acidification carrying out aheads slate put together (600 tons of re-emit itat as blue fluorescence at DOC specific in the airand (750 billion tons of billion carbon). About MCP. Then in October, Jiaos probing team The origins of c most refractory are carbon) nearly as much as all the CO in Arctic waters thatisincludes the biomass rescence speci wavelengths. 2 of experiments Arctic waters includes and nearly astons much as all the CO wavelengths. 95% organic carbon is carbon). bound up as to the Then opposite thermal extreme: They will carbon) a black box. Some is produced when light 2 in the of air (750 billion of About MCP. in in October, Jiaosthat team heads The origins of most refractory DOC are in the air billion tons of About probing themechanisms MCP. Thenof in October, Jiaos The origins of matter most refractory DOC are refractory DOC: the largest pool of organic explore the the MCP and CO organic near thewhen ocean sur95% of (750 organic carbon iscarbon). bound up as degrades to the opposite thermal extreme: They will a black box. Some is produced light 2 95% ofin organic carbon is pool bound up as a team heads to the opposite thermal extreme: black Some is produced when light sequestration in the equatorial Indo-Paci matter the ocean, says Farooq Azam, face. Oilbox. seeps contribute to thethe pool. The oil refractory DOC: the largest of organic explore the mechanisms of the MCP and COc degrades organic matter near ocean sur2 the largest of organic They will explore the mechanisms of the organic matter near the ocean surWarm Pool, the warmest marine waters in a microbiologist at Scripps Institution of degrades spill the Gulf of Mexico just one drastic sequestration in the equatorial Indo-Paci c refractory matter in DOC: the ocean, says pool Farooq Azam, face.in Oil seeps contribute tois the pool. The oil sequestration in the equatorial in the ocean, says Farooq Azam, MCP and CO seeps contribute to is the pool. The oil the world. The MCP will marine also be featured Oceanography in San Diego, California. In face. example ofGulf how this material is released into 2 Warm Pool, the warmest waters in matter a microbiologist at Scripps Institution of spill Oil in the of Mexico just one drastic Indo-Pacific Pool, the microbiologist Scripps Institution in the Mexico is just one next monthThe at Warm aMCP Gordon Research Confer- a the December 2009 issue of Oceanography , spill the ocean, saysof Meinhard Simon, a drastic microthe world. will also bewarmest featured Oceanography in at San Diego, California. of In example ofGulf how this material is released into marine waters in the world. The MCP will Oceanography in San Diego, California. In example of how this material is released into ence on marine microbes, and it is outlined a team led by Hansell and Craig Carlson of bial oceanographer at the University of Oldnext month at a Gordon Research Confer- the December 2009 issue of Oceanography, the ocean, says Meinhard Simon, a microalso be next month at a outlined Gordon December issue of Oceanography , the says Meinhard Simon, a of microin a paper in press at Nature Reviews Micro- the the University of California, Santa Barbara, enburg in Germany. compounds are ence on featured marine microbes, and it is a team led by 2009 Hansell and Craig Carlson of bialocean, oceanographer at Other the University OldResearch Conference on marine microbes, team led the by Hansell andmap Craig of bial oceanographer at Other the University of wildOld biology . The concept could revolutioncompiled rst global ofCarlson DOC dislikely forged in underwater vents or in in a paper in press at Nature Reviews Micro- a the University of California, Santa Barbara, enburg in Germany. compounds are and it is outlined in asequestration, paper in press at the University of California, Santa Barbara, in Germany. Other compounds are ize our view of carbon says tribution. Carbon-14 studies suggest that res and swept the sea. For theor most part, biology . The concept could revolutioncompiled the rst global map of DOC dis- enburg likely forged ininto underwater vents in wildNature Microbiology . The concept first global map of DOC distri - likely forged in underwater vents or in wildMarkus Weinbauer, a microbial oceanogrefractory compounds swirl in this microhowever, says Azam, we lack understanding ize our Reviews view of carbon sequestration, says compiled tribution.the Carbon-14 studies suggest that res and swept into the sea. For the most part, Carbon-14 studies suggest that several refrac - fires could revolutionize view of carbon and swept into the sea. Forunderstanding the most part, rapher at Laboratoire dOcanographie de bution. bial eddy for more thanswirl 6000 years, of the mechanisms of its formation or variaMarkus Weinbauer, a our microbial oceanogrefractory compounds in this microhowever, says Azam, we lack swirl in this microbial eddy however, sequestration, says Markus Weinbauer, says Azam, we lack understanding Villefranche in France. times the circulation time of the ocean. tions its magnitude and composition. rapher at Laboratoire dOcanographie de tory bial compounds eddy for more than 6000 years, several of thein mechanisms of its formation or variamore 6000 years, several times the of a microbial oceanographer at Laboratoire for thein mechanisms of its formation or variaVillefranche in France. times the than circulation time of the ocean. tions its magnitude and composition. time of the ocean. www.sciencemag.org tions in its magnitude and composition. 1476dOcanographie de Villefranche in France. 18 JUNE circulation 2010 VOL 328 SCIENCE
CREDITS: (OPPOSITE PAGE) PHOTO ILLUSTRATION SCIENCE, (IMAGE) ISTOCKPHOTO.COM/TAMMY616; (THIS PAGE) COURTESY OF JIAO NIANZHI

1476
0618NewsFocus.indd 1476

18 JUNE 2010

VOL 328 SCIENCE www.sciencemag.org

Originally published 18 June 2010

CREDITS: CREDITS: COURTESY COURTESY OF JIAO OF NIANZHI JIAO NIANZHI

5
6/14/10 5:12:41 PM

NEWSFOCUS
Azam and and others others credit credit Jiao Jiao with with a a key key Azam insight: the the recognition recognition that that microbes microbes play play insight: a dominant dominant role role in in pumping pumping bioavailable bioavailable a carbon into pool of relatively inert inert comcarbon intoa a pool of relatively pounds. Some refractory DOC hangs in compounds. Some refractory DOC hangs the upper water gets in the upper watercolumn, column,while while some gets shunted to to the the deep deep ocean ocean interior interior via via the the shunted biological pump. MCP may act act as one biological pump.The The MCP may as of the beltsbelts that transport and store one of conveyor the conveyor that transport and carbon in the deep oceans, says says Chen-Tung store carbon in the deep oceans, ChenArthur Chen, an ocean chemist Tung Arthur Chen, an carbonate ocean carbonate at National Sun Yat-sen University in Kaohsichemist at National Sun Yat-sen University ung, Taiwan. The MCP also appears to funcin Kaohsiung, Taiwan. The MCP also appears tion in deepin waters, where bacteria adapted to function deep waters, where bacteria to the high-pressure environment may have adapted to the high-pressure environment a special capacity degradeto refractory may have a special to capacity degrade DOC, says Christian Tamburini, a microbirefractory DOC, says Christian Tamburini, a ologist at the Centre de Marmicrobiologist at the dOcanologie Centre dOcanologie seille in France. de Marseille in France. It took took sharp sharp sleuthing sleuthing to to uncover uncover the the It microbial connection connection with with refractory refractory DOC. DOC. microbial In a a landmark landmark paper paper in in 2001, 2001, Hiroshi Hiroshi Ogawa Ogawa In of the the University University of of Tokyo Tokyo and colleagues colleagues of showed that marine marine microbes microbes are are able able to to showed convert bioavailable bioavailable DOC to refractory refractory convert DOC ( Science , 4 9). May 2001, 917). later, Then DOC (See page Then a p. month a month later, Zbigniew at the Zbigniew Kolber, now Kolber, at the now Monterey Monterey Bay Aquarium Institute Bay Aquarium Research Research Institute in Moss in Moss Landing, California, and colleagues Landing, California, and colleagues reported reported that in the upper open ocean, an that in the upper open ocean, an unusual unusual class of photosynthetic bacteria class of photosynthetic bacteria called AAPB called AAPB for 11% the total accounts for accounts 11% of the total of microbial microbial community (Science , p. 29 June community (Science, 29 June 2001, 2492). 2001, p.seemed 2492). AAPBs seemedeverywhere, to be plentiAAPBs to be plentiful ful everywhere, according to measurements according to measurements of infrared of infrared uorescence the microbes fluorescence from the from microbes lightlight-absorbing pigments. absorbing pigments. It turned turned out, out, though, though, that that other other organorganIt ismswere were throwing AAPB estimates isms throwing thethe AAPB estimates way way off the mark. new technique, off the mark. UsingUsing a new a technique, Jiaos Jiaos determined group determined the fluoresgroup that the that fluorescent glow cent glow of phytoplankton was the masking of phytoplankton was masking glow the of glow of the target microbes. when the target microbes. Just like Just when like the moon the moon is stars bright, stars Jiao are says. visible, is bright, less areless visible, He Jiaothe says. Heapproach put the new approach through put new through its paces in its paces in Chinas 2005, when Chinas Ocean 1 2005, when Ocean 1 research vesresearch vessel conducted campaigns to sel conducted campaigns to mark the 600th mark the 600th anniversary of Admiral He anniversary of Admiral He Zhengs historic Zhengs historic voyages. The observations voyages. The observations turned things turneddown, things upside down, says. His upside Jiao says. His Jiao group found group found that AAPBs are more abunthat AAPBs are more abundant in nutrientdantwaters in nutrient-rich than in the open rich than in thewaters open ocean, indicating ocean, indicating that AAPB population levthat AAPB population levels are linked with els arenot linked with DOC, not light. DOC, light. Next, Jiao Jiao found found that thatAAPBs AAPBs are are prone prone to to Next, viral infection, infection,and andhe he isolated isolated the the first rst phage phage viral thats specific specic for these bacteria. bacteria. Phages Phages rip rip thats apart their their hosts, hosts, spilling spilling their their guts, guts, includincludapart ing organic organic carbon, carbon, into into the the water. water. This This viral viral ing shunt acting acting on on many many marine marine bacteria bacteria may may shunt be a a significant signicant player player in the accumulation accumulation be of refractory DOC compounds in the water
Biological Sequestration of Carbon in the Sea
C02 molecules s Atmosph At here

Oc cean c

Bioavailable DOC

Mic obi Microbial bial carbon ca bo on p m mp pump Refractory Re DOC Microbes s

Phytoplankton

Main ain foo food chain ch hain Biological pump

Double-barrel pump. Each year, the biological pump deposits some 300 million tons of carbon in the deep ocean sink. Even more massive amounts are suspended in the water column as dissolved organic carbon, much of which is converted into refractory forms by the microbial carbon pump.

of refractory DOC compounds in the water column, says says Steven Steven Wilhelm, a microbiolmicrobiolcolumn, ogist at at the the University University of of Tennessee, Tennessee, KnoxKnoxogist ville. Pulling Pulling together together several several strandsthe strandsthe ville. ubiquity of ofAAPBs, AAPBs, their their low low abundance abundance but but ubiquity high turnover turnover rate, rate, the the tight tight link link to to DOC, DOC, and and high their susceptibility susceptibility to to infectionJiao infectionJiao pro protheir posed that that AAPBs AAPBs and and other other microbes microbes are are posed a key key mechanism mechanism for for the the conversion conversion of of biobioa available DOC DOC to to refractory refractory DOC. DOC.That That may may available seem counterintuitive, counterintuitive,as asmicrobes microbesdo donot not set set seem out to to produce produce refractory refractory DOC; DOC; rather, rather, the the out compounds are are a a byproduct byproduct of of their their demise. demise. compounds This process process is is not not beneficial benecial to to the the cell, cell, This says Simon. Simon. says Because the the buildup buildup of of refractory refractory DOC DOC Because in the the water water column column is is accidental, accidental, it it will will be be in a challenge challenge to to coax coax microbes microbes to to sequester sequester a more carbon. carbon. For For decades, decades, researchers researchers have have more been tinkering tinkering with with the the biological biological pump pump to to been store more more carbon carbon in in the the deep deep ocean ocean by by seedseedstore ing seas seas with with iron iron fertilizer. fertilizer. The The iron iron triggers triggers ing phytoplankton blooms blooms that that suck suck more more CO CO22 phytoplankton from the the air. air. That That should should also also drive drive more more carcarfrom bon into into the therefractory refractorypool, pool,Koblek Koblizeksays. says. bon Even tweaking tweaking the the MCP MCP could could have have a a Even profound effect. column holds on profound effect.The Thewater water column holds average 35 to of carbon from on average 3540 tomicromoles 40 micromoles of carbon refractory DOCDOC per liter. An increase of a from refractory per liter. An increase mere 2 to 2 3 micromoles per liter sock of a mere to 3 micromoles per would liter would away several billion tons of says sock away several billion tons ofcarbon, carbon, says NagappaRamaiah, Ramaiah,aamarine marine microbial ecolNagappa microbial ecolo ogist the National Institute of Oceanogragist atat the National Institute of Oceanography phy in Goa, India. to investigate in Goa, India. We We have have to investigate any anyall and all means help the excess and means to helpto sink the sink excess carbon, carbon, he says. he says. Two billion years ago, when bacteria bacteria Two ruled Earth, Earth, the the oceans oceans held held 500 500 times times as as ruled much DOC DOC as as today, today, most most likely likely generated generated much by the MCP, Jiao says. Ecosystem dynam-

RICHARD STONE

www.sciencemag.org SCIENCE VOL 328

18 JUNE 2010

1477

CREDIT: C. BICKEL/SCIENCE

by the MCP, Jiao says. Ecosystem dynamics have changed immensely since then, but but the ics have changed immensely since then, microbial sequestration potential could still the microbial sequestration potential could be huge, he argues. No chemical equilibrium still be huge, he argues. No chemical equilibwould limit limit conversion of bioavailable DOC rium would conversion of bioavailable to refractory DOC, which in turn would not DOC to refractory DOC, which in turn would exacerbate ocean acidification, Jiao, not exacerbate ocean acidification, says says Jiao, who is is planning planning pilot pilot experiments experiments this this sumsumwho mer.Ramaiah, Ramaiah,meanwhile, meanwhile, says is lookmer. says hehe is looking ingenhanced for enhanced sequestration potential in for sequestration potential in select select marine bacteria marine bacteria strains.strains. Theres no simple recipeand recipeand some some sci sciTheres entists are are not not convinced convinced that that its its feasible feasible entists or even even safe. safe. I I do do not not think think it it is is possible possible to to or enhance carbon carbon sequestration sequestration by by the the MCP. MCP. enhance We have have no handle handle on on any any controls controls of of how how We refractory DOC DOC is is generated, generated, says says Simon. Simon. refractory With the the present present knowledge, knowledge, any any sequestrasequestraWith tion effort, effort, argues argues Weinbauer, Weinbauer, could could come come tion back like like a a boomerang boomerang and and worsen worsen the the probprobback lem. At the same time, time, humans humans may may already already lem. be inadvertently inadvertentlystimulating stimulatingthe theMCP, MCP, says says be Salgado. Global Global warming warming is is increasing increasing strati stratiSalgado. cation, reducing reducingdeep deep convection, convection, and and stim stimfication, ulating microbial microbial respirationall respirationall of of which which ulating favor the the MCP, MCP, he he says. says. favor The MCP MCP concept concept should should help help address address The critical issues, such as whether oceanocean acidicritical issues, such as whether cation and warming will signi cantly alter acidification and warming will significantly carbon ux into DOC, DOC, says Azam, alter carbon fluxrefractory into refractory says who with Jiao chairs thechairs Scienti c Committee Azam, who with Jiao the Scientific on Oceanic Researchs new Researchs working group on Committee on Oceanic new the role of MCP carbon biogeochemistry. working group onin the role of MCP in carbon The upcoming research cruises should ll in biogeochemistry. The upcoming research more details of how themore MCPdetails governs carbon cruises should fill in of how cycling how it may respond to climate the MCPand governs carbon cycling and how it change. As Wilhelm notes, We are at the may respond to climate change. As just Wilhelm dawn of developing notes, We are just atthis the understanding. dawn of developing RICHARD STONE this understanding.

CREDIT: C. BICKEL/SCIENCE

Microbial Control of Oceanic Carbon Flux: The Plot Thickens

Farooq Azam hotosynthesis fixes carbon into organic matter in the ocean. Biological forces then paint intricate flux patterns for carbon in ocean space and in time, as it flows through the foodweb, becomes stored in the sediments and exchanged with the atmosphere. Predicting how these carbon flux patterns might respond to global change (or to human manipulation) is a primary reason for learning more about the workings of the oceans carbon cycle. The flux patterns are a result of intricate interactions of a diverse biota with a physically and chemically complex pool of organic matter. It now seems that things will get even more complicated before they get simpler. New fundamental findings on the roles of microbes in the fate of organic matter and, recently, on the nature of the organic matter itself (14) must be properly assimilated before we can hope to construct ecosystem models to predict the patterns of carbon flux. This impetus could lead to a powerful new synthesis. What biological forces act on photosynthetically produced organic matter in the ocean? Historically, the paradigm has been that essentially all primary production stays within the particle phase (5), it is eaten by herbivores, and the fate of carbon is determined by the grazing food chain (Fig. 1). Little dissolved organic matter is spilled for bacteria to use. It had, therefore, been implicitly assumed to be safe to ignore bacteria, protozoa, and viruses in studying the fate of organic matter they were too sparse and not active enough (5). This is now changed (58): Major fluxes of organic matter, often eclipsing the grazing food chain in quantity, move via dissolved organic matter into bacteria and the microbial loop (7, 8) (Fig. 1). Previous methods had missed >99% of microorganisms and had grossly underestimated their metabolism. Now we know from extensive field studies that in most of the ocean, organic matter flux into bacteria is a major pathway; one-half of oceanic primary production on average is channeled via bacteria into the microbial loop (7, 8)a major biological force in the ocean. Ocean basin-scale biogeochemical studies now routinely quantify organic matter fluxes

into bacteria in conjunction with other major flux pathways: grazing food chain, sinking flux, and dissolved organic matter storage. The fraction of primary production used by bacteria (Fb) is highly variable over various time and space scales (710). The magnitudes and variability of the fluxes are large enough to cause variability in flux partitioning between competing pathways (Fig. 1)the microbial loop, the grazing food chain, sinking fluxes, and storage of dissolved organic matter (8, 9). Fish production in the eastern Mediterranean was diminished by a dominant microbial loop (Fb = 0.85) (11). In an earlier study (12), the richness of the fishery in coastal Newfoundland was ascribed to uncoupling of bacteria from primary production during the spring bloom. These are tantalizing but rare success stories where descriptive studies lead to interesting insights into the bacterial control of organic matter fluxes and ecosystem dynamics. However, we lack a conceptual framework to predict variation in organic matter flux into bacteria and how it fits into the overall oceanic flux picture (for example, how the flux partitioning will respond to global change). This requires elucidating the causes and mechanisms of variability of organic matter flux into bacteria. How do bacteria interact with organic matter, and what regulates the flux of organic matter into them? It is generally thought that pelagic bacteria passively receive dissolved organic matter

leaking from the grazing food chain and diffused homogeneously. However, recent studies on the complex nature of the organic matter field (14, 13) and behavioral strategies of bacteria (14) have suggested (2) that bacteria do not use only preformed dissolved organic matter; they also attack all organic matter, even live organisms, thus liberating dissolved organic matter. Hence, bacterial attack can profoundly modify the biogeochemical behavior of organic matter in ways not inferred from measuring cumulative organic matter fluxes into bacteria. Recognition of such modification interactions could add important new variables in biogeochemical models. Our view of the organic matter in seawater has changed dramatically. The traditional dichotomy of particulate organic matter versus dissolved organic matter is being replaced by the concept of an organic matter continuum (2). Several new classes of abundant colloids, submicrometer particles, and transparent polymer particles have been discovered (1, 4, 13). This oceanic dark matter ranges in size from 20 nm to hundreds of micrometers. It has been proposed (2, 3) that pelagic bacteria experience a gel-like polymeric matrix with colloids and particles embedded as suprapolymeric hotspots (8) (Fig. 2). The interaction of bacteria with the organic matter continuum, and their behavioral response to its heterogeneity, creates microscale featuresactivity hotspotswith distinctive natures and intensities of biogeochemical transformations. Patchiness may also support high bacterial diversity. An important recent study by Chin et al. (1) demonstrated that polymers in seawater indeed form a gel. Do bacteria respond to the structure of the organic matter field? Large bacterial populations can develop on hotspotsfor example, marine snow, dead and even living algaeand

The author is at the Scripps Institution of Oceanography, University of California, La Jolla, CA 92093, USA. E-mail: fazam@ucsd.edu

Fig. 1. The microbial loop: classical version. Modern view of the pelagic foodweb, emphasizing the microbial loop as a major path for organic matter flux. Competition between the three main flux pathsgrazing food chain, microbial loop, and sinkingsignificantly affects oceanic carbon cycle and productivity. DOM, dissolved organic matter; DMS, dimethylsulfide.
Originally published 1 May 1998

constituents of cells that should only slowly cate that diagenetic processing by microorratios in specific amino acids that are found accumulate. Amino acids, neutral sugars, ganisms is rapid and critical for shaping the in the peptide component of peptidoglycan and amino sugars typically compose 70% composition and refractory nature of marine (5). Overall, these results suggest that bacof bacterial biomass (29), but these bioDOM. terial degradation sufficiently alters the in association with the dark matter. environmental context. Biogeochemical chemicals were relatively minor compoThe amino sugar muramic acid is structure of peptidoglycan so that its poly[In one study, 24 to 68% of bacteria couldderived then be DOM. considered as nents variability of bacterially It apuniquely found in the repeating disacchasaccharide component is no longer recogwere on transparent particles ( 13 )]. a consequence of adaptive responses to pears that the selective preservation of unride backbone of the bacterial cell wall nizable at the molecular level. Oceanic bacteria (14) exhibit sophisvariations. This usual (micro)environmental biochemical components of cells ticated behaviorswimming speeds shouldaccount also serve a framecould approach not, by itself, foras the rapid Table 2. Chemical composition ofper bacterially of hundreds of micrometers sec- derived DOM in the glucose and glutamate experiments. All work to understand the maintenance of formation and accumulation of bacterially are the mean of duplicate-bottle experiments, except for the data indicated by asterisks, which were ond, unusual swimming (run revermicrobial derived DOM. diversity and to make predicobtained from single bottles. The mean deviation of replicates was 0.9 M for DOC, 0.2 M for DON, 2 sals), and chemosensing. organic tions on the survival of specific bacte for C/N ratio, and 1 to 2% ofThe DOC, DON, THAS, and THAA compositions. The uncharacterized fraction Components of bacterial cells are re-in the glucose is estimated as the difference between the total DOC and the DOC accounted matter field experiment is dominantly polymeric species, including human leasedrial into the surrounding water pathogens as DOM for assuprapolymeric, THNSTHASTHAA, and between and so most pelagic the total DON and the DON accounted for as THAATHAS, such as Vibrio cholerae, in response through a variety of biological processes, except forhave the data after 1 year, in which bacteria a diverse repertoire ofTHNS was not included (data in parenthesis). The uncharacto ecosystem perturbations (21). lysis This including direct release ( 11, 31 ), viral terized fraction in the glutamate experiment is based only on THAA, because THNS and THAS were not surface-bound enzymes (proteases, framework, which includes bacteria-al( 32 ), and grazing ( 33 ). Exoenzyme activity measured (indicated by nm). glucosidases, lipases, phosphatases, gae interactions, should also be relevant is critical for the microbial utilization of nucleases) to cause its hydrolysis (15). to the prediction of algalthat blooms, includthis DOM, and it appears enzymatic Initial substrate Small-scale variation of species coming toxigenic species. Powerful new apactivity plays an important role in the forGlutamate position can significantly change Glucose the proaches are enabling to study mation of refractory DOMus that is of microsmall Chemical of enzyme activities (16). distributions bial including activisize (34 ). ecology, It is possible thatconsortial nonspecific or composition Incubation time Incubation time promiscuous activities of enzymes, thattechocPelagic bacteria also have multiphasic ties, in an ecosystem context. New (days) (days) cur with much lower efficiency than primapermeases for nutrient uptake (17), niques allow multiple interrogations ry activities (35 ), occasionally produce with nanomolar to millimolar phylogeny, metabolism, growthat the 2 4half- 7 365 2 4 9 560 fragments from cell macromolecules that and essaturation constants, suggesting their individual level. These ideas of the DOM (M) cape recognition by bacterial and adaptations to life in a patchy Concentration environapproaches should lead toenzymes a synthesis of Fig. 2.bulk The microbial loop: impressionist version. DOC 30 18 16 11 18 14 9.0 molecular-level chemicalevolution, analyses.ecology, Given ment. Considering the intimate contact bacterial adaptation, A bacteria-eye view of 13 the oceans euphotic layer. DON 1.0 0.7 0.6 1.2 3.1 2.4 1.5 0.7 this scenario, the rate of formation ofform reof their digestive system with the orand biogeochemistry, and should Seawater is an organic matter continuum, a gel of tangled DOC composition (% as) fractory DOM isintegrating dependentthe onroles the rate of ganic matter gel, motile bacteria are the a basis for of bacpolymers with embedded strings, sheets, and bundles THNS 2.5 4.9 5.9 nm nm nm nm nm microbial activity. This relatively simple ultimate swimming stomachs (18). 1.5 of fibrils3.8 and particles, including living organisms, as teria in predictive biogeochemical THAS 0.7 1.4 nm nm nm nm mechanism could be responsible for much They with hotspots. on marine snow THAA cannot avoid interacting 2.5 4.3and 3.7 2.5 Bacteria 13 (red) acting 18* 12 3.1 (black) models. of the sequestration of fixed C in the ocean. Uncharacterized 94 89 89or algae (94) 87 control 82* 88 97 primary changing organic matter. (green) can sedimentation and Bacterial action on components DON of composition productivity; diverse microniches (hotspots) can support (% as) the organic matter field 3.4 can modify high bacterial THAS 5.5 its 6.6 5.6 diversity. nm nm nm nm THAA 22 without 29 nec- 27 6.4 15 31* 33 11 character in varied ways References and Notes Uncharacterized 75organic66 67 88 also expose 85 67 microscale 89 1. C. Chin et ., Nature 391, (1998). J. I. Hedges, 1. W. R. Benner, J. al D. Pakulski, M. 568 McCarthy, essarily involving large matter fluxes could the69* alga to high 2. F. et al., Microbiol. Ecol. 28, 167 (1993). P.Azam G. Hatcher, Science 255 , 1561 (1992). C:N molar ratio into bacteria. Two examples illustrate this concentrations of remineralized nutrients, thus 3. F. Azam and D. T. C. Pratum, Smith, J. in Hedges, Particle R. Analysis in 2. M. D. McCarthy, Benner, Bulk In a study of bacterial 32 26 28 8.9 5.1 fixation. 5.7 Partial 9.3hydrolysis 13 Oceanography, S. (1997). Demers, Ed. (Springer-Verlag, Berlin, point. activity on ma- enhancing carbon Nature 390, 150 1991), pp. 213236. Total characterized 7.5 7.9 9.2 (4.6) 3.5 3.1 3.3 3.6 3. I. L. Koike L. Clark, D. Ingall, R. 242 Benner, Nature , 426 rine snow (19), the colonizing bacteria 37 grew of complex polysaccharides by bacterial attack 4. et alE. ., Nature 345, (1990) M. L. 393 Wells and Uncharacterized 41 37 (10) 5.4 6.3* 12 14 (1998). E. D. Goldberg, ibid. 353, 342 (1991) R. A. Long and F. slowly but expressed large amounts of ectohy- could produce slow-to-degrade dissolved or4. Azam, E. Tanoue, Nishiyama, THAS composition (mole %) Aquat. S. Microb. Ecol. 10,M. 213Kamo, (1996). A. Tsugita, drolases, which solubilized particulate organic ganic matter, resulting in carbon storage (high Geochim. Cosmochim. Acta 59, of 2643 (1995). 5. J. H. Steele, in The Structure Marine Ecosystems GalN 20 26 31 33 nm nm nm nm Univ. Press, MA, 1974) matter; however, because of the low carbon bacterial activity could actually produce more 5. (Harvard M. D. McCarthy, J. I. Cambridge, Hedges, R. Benner, Science 281, GlcN 55 50 56 67 nm nm nm nm 6. L. R. Pomeroy, 231 (1998). Bioscience 24, 499 (1974) J. E. Hobbie et demand of these bacteria, most slow-to-degrade dissolved organic matter). MA 26 24 hydrolysate 12 0 nm nm nm nm Appl. Environ. 33, 1225 (1977). 1, 14 6. al., P. M. Williams, E.Microbiol. R. M. Druffel, Oceanography diffused out (uncoupled solubilization), thus This could 7. F. Azam et al., Mar. Ecol. Prog. Ser. 10, 257 (1983). THAA composition (mole %) be used by bacteria at a different (1988). 8. J. J. Cole et al., ibid. 43, 1 (1988) J. A. Fuhrman and F. reducing the sinking flux. enzyme action place dissolved organic Acidic 26 Also, 27 21 30 and time 26 (for example, 12 17 28 7. J. I. Hedges et al., Org. Geochem. 31, 945 (2000). Azam, Mar. Biol. 66, 109 (1982); A. Hagstrom et al., 8. Appl. G. R. Environ. Harvey, Microbiol. D. A. Boran, A.(1979). Chesal, J. M. Tokar, Basic thought to increase 12 the efficiency 10 11the matter 11 8 6 Antarctic7 waters dur8 was of accumulating in 37, L. 805 Mar. . 12and , 119 Neutral carbon pump; 23 31 - ing 41 summer24 56 the foodweb 36 46 9. H. W. Chem Ducklow C.(1983). A. Carlson, Adv. Microb. Ecol. 12, oceanic protease27 and phospha could support during 9. 113 J. E. (1993). Brophy, D. J. Carlson, Deep-Sea Res. 36, 497 Hydrophobic 38 36 37 18 41 27 40 tase activities were much higher than those of winter). These and other examples show 18 that 10. H. (1989). W. Ducklow, Deep Sea Res. 40, 753 (1993). - and -glucosidases, and this would cause the ecosystem-level consequences of bacteria- 11. N. Williams, Science 279, 483 (1998). L. R. Pomeroy and D. Deibel, ibid. 233, 359 (1986). preferential nitrogen and phosphate reten- organic matter interactions may be pervasive 12. 13. A. L. Alldredge et al., Deep Sea Res. 40, 1131 (1993). 919 www.sciencemag.org SCIENCE VOL 292 4 MAY 2001 tion compared with carbon in the photic layer in ways not currently being quantified, or even 14. J. G. Mitchell et al., Appl. Environ. Microbiol. 61, 4436 (1995) H.-P. Grossart and F. Azam, paper presented at where it supports more carbon fixation and recognized, in ocean basin-scale biogeochemithe Ocean Sciences Meeting, San Diego, CA, 9 to 13 export. Thus, bacteria influenced the fate of cal studies. February 1998. carbon without a significant carbon demand. Thus, behavioral and metabolic responses 15. H. G. Hoppe et al., Mar. Ecol. Prog. Ser. 93, 277 (1993) J. T. Hollibaugh and F. Azam, Limnol. Oceanogr. 28, 104 Bacterial interaction with live diatoms was of bacteria to the complex and heterogeneous (1983). examined in a mesocosm (20). Bacteria colo- structure of the organic matter field at the mi- 16. E. F. DeLong et al., Limnol. Oceanogr. 38, 924 (1993) A.S. Rehnstam et al., FEMS Microb. Ecol. 102, 161 (1993); nized the diatoms, grew rapidly, and expressed croscale influence ocean basin-scale carbon J. Martinez et al., Aquat. Microb. Ecol. 10, 223 (1996). large amounts of ectohydrolase. Experiments fluxes in all major pathways: microbial loop, 17. H. Nissen et al., Mar. Ecol. Prog. Ser. 16, 155 (1984) G. G. Geesey and R. Y. Morita, Appl. Environ. Microbiol. 38, suggested that ectohydrolases of attached sinking, grazing food chain, carbon storage, 1092 (1979). bacteria prune mucus from the surface of and carbon fixation itself. However, studying 18. J. Stern, personal communication. diatoms, thereby controlling diatom stickiness such varied influences of bacteria on organic 19. D. C. Smith et al., Nature 359, 139 (1992) A. L. Alldredge and C. C. Gotschalk, Cont. Shelf. Res. 10, 41 (1990). and inhibiting aggregation. Bacterial modifi- matter, and their spatial-temporal variations, 20. D. C. Smith et al., Deep Sea Res. II 42, 75 (1995) U. Passow and A. L. Alldredge, ibid., p. 99. cation of algal surfaces, a microscale process, in piecemeal fashion will only result in a con21. P. R. Epstein, Biosystems 31, 209 (1993) R. R. Colwell, could increase the persistence and intensity ceptual patchwork without a unifying frameScience 274, 2025 (1996). of algal blooms, as well as influence the sink- work or predictive power. A unifying theme 22. I thank H.-P. Grossart, R. A. Long, L. B. Fandino, K. D. Bidle, D. C. Smith, J. T. Hollibaugh, F. Rassoulzadegan ing flux, with profound ecological and bio- should derive from applying robust principles for discussions and suggestions. Research supported by NSF grants NSF OPP96-17045 and OPP95-40851. geochemical implications. Attached bacteria of biochemical adaptation in a realistic micro-

Production of Refractory Dissolved Organic Matter by Bacteria

Hiroshi Ogawa,1* Yukio Amagai,1 Isao Koike,1 Karl Kaiser, 2 Ronald Benner2 Most of the oceanic reservoir of dissolved organic matter (DOM) is of marine origin and is resistant to microbial oxidation, but little is known about the mechanisms of its formation. In a laboratory study, natural assemblages of marine bacteria rapidly (in <48 hours) utilized labile compounds (glucose, glutamate) and produced refractory DOM that persisted for more than a year. Only 10 to 15% of the bacterially derived DOM was identified as hydrolyzable amino acids and sugars, a feature consistent with marine DOM. These results suggest that microbial processes alter the molecular structure of DOM, making it resistant to further degradation and thereby preserving fixed carbon in the ocean.

OM is the largest reservoir of fixed carbon in the ocean and is approximately equivalent to the reservoir of atmospheric CO2. The major bioelements (C, N, and P) in DOM occur in functional groups common to biopolymers found in marine organisms (13). Specific cellular components of bacteria have been identified in marine DOM (4,5), indicating that bacteria are an important source of this material. These observations indicate the predominance of biomolecules in DOM, but <25% of marine DOM has been identified as specific biochemicals (6, 7), suggesting that its molecular structure has been modified. It appears that these molecular modifications also reduce the bioavailability of the DOM. Physicochemical reactions were proposed as the dominant mechanism for the formation of molecularly uncharacterized and refractory DOM (8), but a few studies reported that microorganisms also produce DOM that is resistant to decomposition (911). In a series of experiments, we examined the bacterial utilization of simple biochemicals and traced the production of fresh DOM by the bacterial community. We used seawater cultures with natural bacterial assemblages to examine DOM production by marine bacteria (12). Culture media were prepared with organic-free artificial seawater and either glucose or glutamate as the sole C source. Incubations with glucose received ammonium and phosphate as N and P sources and an inoculum from Gulf of Mexico surface water. Incubations with glutamate received phosphate as a P source and an inocu-

1 Ocean Research Institute, The University of Tokyo, 1-15-1 Minamidai, Nakano, Tokyo 164-8639, Japan. 2 Department of Biological Sciences and Marine Science Program, University of South Carolina, Columbia, SC 29208, USA.

*To whom correspondence should be addressed. Email: hogawa@ori.u-tokyo.ac.jp

lum from Sagami Bay, Japan. Concentrations of free glucose and glutamate were measured using high-performance liquid chromatography (HPLC) (13, 14). Measurements of total organic carbon (TOC), dissolved organic carbon (DOC), and dissolved organic nitrogen (DON), were made using high-temperature combustion (15, 16). Concentrations of total hydrolyzable neutral sugars (THNS), total hydrolyzable amino sugars (THAS), and total hydrolyzable amino acids (THAA) were measured by HPLC after acid hydrolysis (17). Both experiments were conducted at room temperature (22 to 28C) with duplicate bottles under dark conditions. Glucose (208 M C) was rapidly consumed and undetectable after 2 days (Fig. 1). Within 2 days, TOC concentrations decreased by 78% due to respiration, and 7% of the initial glucose C was converted to particulate organic carbon (POC, i.e., bacterial biomass), whereas 15% was converted to DOC (Table 1). Similarly, glutamate was completely consumed within 2 days, and 66% of the added C was respired. Compared to the glucose incubations, glutamate incubations had a greater yield of bacterial biomass (22% of initial glutamate C), but a similar yield of DOC (13%). After 2 days, bacterial growth yields were 8% in the N-limited glucose incubations and 25% in the glutamate incubations, if bacterially derived DOC was excluded from yield calculations. Inclusion of bacterially derived DOC results in growth yield estimates of 22 and 35%, respectively, in the glucose and glutamate incubations. Concentrations of bacterially derived DOC decreased gradually during the next week of decomposition (Stage II in Table 1). Assuming first-order kinetics, the average decay constant during Stage II was reduced by a factor of 8 compared with the utilization of glucose and glutamate (Stage I). It is generally recognized that bacteria in natural waters rapidly utilize labile compounds, such as free amino acids
Originally published 4 May 2001

and monosaccharides, even at low (nM) concentrations (18, 19). Therefore, it is unlikely that the concentrations of remaining bacterially derived DOC (20 to 30 M C) would limit bacterial utilization. Neither ammonium nor phosphate was depleted after 1 week of incubation, suggesting nutrients were not limiting for decomposition. These results demonstrate that marine bacteria rapidly consumed labile DOM and produced DOM that was relatively resistant to decomposition. The experiments were continued to examine the long-term persistence of bacterially derived DOM. Consequently, 10.5 and 9.0 M DOC were measured after 1 and 1.5 years, respectively, of incubation in the glucose and glutamate experiments, corresponding to 5 and 7% of the initial DOC added in each experiment and to 37 and 50% of the bacterially derived DOC concentrations at the end of 2 days. Although the degradation of DOC continued throughout the incubation (Stage III in Table 1), the degradation rate was remarkably low, and the decay constants were up to 100 times lower than those in Stage II. These decay constants were similar to previous estimates from long-term degradation experiments with marine DOC from oligotrophic waters (20) and corresponded to residence times of 1.2 to 2.3 years. Thus, bacteria rapidly produced a major component of refractory DOM from labile substrates. Chromatographic characterizations of major biogenic components of bacterially derived DOM indicated that THAA, THNS, and THAS accounted for 2.5 to 4.3%, 2.5 to 5.9%, and 0.7 to 1.5% of the DOC, respectively, during the first week in the glucose experiment (Table 2). Only 6 to 11% of the DOC was identified at the molecular level. Amino acids and amino sugars are also important nitrogenous components of DON. No DON was initially present in the glucose experiment, but concentrations of 0.6 to 1.0 M DON were measured during the incubation. Marine bacteria produced DON from glucose and ammonium, and amino acids and amino sugars accounted for 22 to 29% and 3.4 to 6.6%, respectively, of the DON during the first week (Table 2). As with DOC, most (67 to 75%) of the DON was not characterized at the molecular level. The C:N ratio of the DOM was 28 to 32, indicating that bacterially derived DOM was C-rich relative to the initial substrates (glucose C:NH4-N = 20:1). The C:N ratio for the molecularly characterized DOM (THAA+THNS+THAS) was 7.5 to 9.2. In contrast, the C:N ratio of the molecularly uncharacterized DOM was 37 to 41, indicating a N-poor composition. In the glutamate experiment, only amino acids (THAA) were measured in bacterially derived DOM. The yields of THAA were 12

and 15 to 33% of the ultrafiltration (15). After 6 months, 84 3% acid compositions of bacterially derived week of the incubaof the DOC was 10 kD in size, consistent DOM rapidly shifted from a composition REPORTS rized fraction comwith the size distribution of marine DOM similar to that of microorganisms to a comOM (82 to 88% of C DOM (Ta( 90% of DOC is 10 kD) ( 15 ). position similar to that of marine acids ( THAA) were measured in bacterially DOM. The size distribution of DOM the teria (neutral 25 mole %; basic 15 mole to 18% of the DOC and 15 to 33% of the DON and uncharacterized fractions were 3.1 in to 3.5 the glutamate experiment compared with the derived DOM. The yields of THAA were glutamate experiment was investigated using %) ( 27 ). During the incubations, the amino ble 2). A similar transition was observed in The C:N during ratio of the The amino acid composition of marine the first week of the incubation. The and 5.4 to 12, respectively, suggesting that glucose experiment, reflecting the different 12 to 18% of thefraction DOC and 15 to 33% of the ultrafiltration (15derived ). After 6 months, 3% N acid of bacterially derived amino sugar compositions ofcompositions bacterially 9.3, whereas the C:N DOM is relatively rich in amino acids uncharacterized composed most of neutral most bacterially DOM was84 depleted content of initial desubstrates. By 1.5 years, DON during the (30 firstto week of the incubaof the DOC was 10 rived kD in DOM size, consistent DOM rapidly shifted fromdecreased a composition in the glucose incubations. The and uncharacterized 40 mole and in basic the DOM (82 to 88% of C and 67 to %) 85% of depleted in N relative to the initial substrate (glutamate the yields of amino acids (3.1% tion. The uncharacterized fraction comwith the size distribution of marinesugar DOMcomposition similar to of that of microorganisms to a cominitial DOM in of the DON) 3.5 andN). 5.4 to 12, (5 to was 10 mole (25, 26 ) comThe C:N ratioamino of the acids total DOM 5.1 %) C:N = 5). The yields of aminoamino acids in DOM of the DOC and 11% in the posed most of the DOM (82 to 88% of C (90% of DOC is 10 kD) (15). position similar to that of marine DOM (Tatheratios incubations wasin similar to thatexperiment, in bacterial ng that most bacteripared marine organisms, including bacto 9.3, whereas the C:N with ratios of the THAA were higher and the C:N were lower glutamate and the C:N ratio of the ble 2). A similar transition was observed in and 67 to 85% of N). The C:N ratio of the The amino acid composition of marine s depleted in N relaamino sugar compositions of bacterially detotal DOM was 5.1 to 9.3, whereas the C:N DOM is relatively rich in neutral amino acids substrate ratios (glutamate rived DOM in the glucose incubations. The of the THAA and uncharacterized (30 to 40 mole %) and depleted in basic Table 1. Carbon balances and kinetic decay constants for DOM at different stages in seawater s of amino acids were in 3.1 to 3.5 and 5.4 to 12, amino acids (5 to 10 mole %) (25, 26 ) com- initial amino sugar composition of DOM in fractions cultures with bacteria and glucose or glutamate as an initial substrate. The data are given as the the C:N respectively, ratios were suggesting the incubations was similar to that in bacterial that most bacteripared with marine organisms, including bacTable Carbon balances and kinetic decay constants for DOM at different stages inwere seawater culturesat with bacteria mean1. mean-replicate of duplicate bottle incubations. Carbon balances calculated the end te experiment com-DOM ally derived depleted in N including relaand glucose or glutamate as an initial substrate. The data are given as the mean |mean-replicate| of was each stage period, remineralized % (i.e., the loss of TOC) and remaining of %duplicate as DOCbottle or tive reflectto the initial e experiment, incubations. Carbon balances were calculated at the end of each period, including % (i.e., the loss POC substrate relative to(glutamate the initial TOC. The results of POC atstage Stage III were not remineralized available, because the Table 1. Carbon balances and The kinetic decay constants for DOMnot at available, different stages in seawater sub5). The of yields of remaining amino acids TOC) and % as in DOC POCwere relative to the initialinsignicant TOC. results of POC at Stage III were difference between TOC and or DOC statistically or no measurement of TOC was ontent of C:N initial cultures with bacteria and glucose or glutamate as an initial substrate. The data are given as the DOM were higher and the ratios were TOC because the C:N difference between and DOC were statistically insignificant or no measurement of TOC was made. made. The decay constants for DOC during each stage were bottle calculated on the assumption of calculated at the end the yields of amino mean mean-replicate of duplicate incubations. Carbon balances were The decay constants for DOC during each stage were calculated on the assumption of first-order decay kinetics. lower in the glutamate experiment comrst-order decay kinetics. of each stage period, including remineralized % (i.e., the loss of TOC) and remaining % as DOC or of the DOC and 11% pared with the glucose experiment, reflectPOC relative to the initial TOC. The results of POC at Stage III were not available, because the utamate experiment, difference between TOC and DOC were ing the different N content of initial subRemaining % statistically insignicant or no measurement of TOC was he DOM increased to constant on the assumption of made. % The decay constants for DOC during each stage Decay were calculated strates. By 1.5 years, the yields of amino Stage (days) Remineralized (day-1) s, bacterially rst-order decay kinetics. acidsderived decreased (3.1% of the DOC and 11% DOC POC characterized the in the glutamate experiment, of the at DON) Remaining % ch in C relative N. ratio of the DOM increased to Glucose (208 0 M C) and theto C:N Decay constant Stage (days) Remineralized % (day1) M is of marine origin 13. In both experiments, bacterially derived I (0 2) 78 1 15 0 7 1 DOC 1.1 0.0 POC was largely uncharacterized at the 87 1 , 21, 23), DOM molecularly II (27) 81 61 0.13 0.03 molecular rich in C relative to N. 95 0 C) Glucose III (7365) 50 (208 0 M0.0012 0.0003 rich in C relative to level N and DOM is of marine origin 6, 24 ), and Most existsoceanic as I (0 2) 78 1 15 0 71 1.1 0.0 Glutamate (132 2 M C) (21, 22 ), The refractory (20, 21, 23), molecularly II (27) 87 1 81 61 0.13 0.03 ules (1, 15, 24 ). III5 (7365) 95 0 0.0012 0.0003 uncharacterized (6, 7 ), 2) rich in C relative to N 66 I (0 13 7 22 2 5 0 1.1 0.1 scribed here demon(C:N 10 to 25)II (29) (16, 24 ), and exists as 77 2 10 1 13 1 0.14 0.08 Glutamate (132 2 M C) operties of bacterially III (9 560) 71 0.0023 0.0003 relatively small molecules (1, 15, 24 ). The 93 1 I (0 2) 66 5 13 7 22 2 1.1 0.1 lar to those of marine simple experiments described here demonstrate that the bulk properties of bacterially derived DOM are similar to those of marine
II (29) III (9 560) 77 2 93 1 10 1 71 13 1 0.14 0.08 0.0023 0.0003

f free mate , and Fig. 1. 1. Concentrations Concentrationsof offree free tures (A), free glutamate glutamate glucose ( A), free (B), and TOC, TOC,DOC, DOC,POC, POC,and and sem( B), and DON in in seawater seawater cultures cultures e (A, DON with natural assemB, D, with naturalbacterial bacterial asblages and either glucose (A, rbon semblages and either glucose C , and E ) or glutamate ( B , D, owth. ( A, C,Fand E) the or glutamate (B, and ) as sole carbon e reD, and Ffor ) asbacterial the sole growth. carbon sources ) and sources for bacterial Panels (C) and (D) growth. are rees of Panels (C) and (D) are reillustrations of panels (A) and d er(B) with magnied of illustrations of panelsscales (A) and averthe y-axis. Each point and of er( B) with magnified scales ror y-axis. bar represents the avercatethe Each point and eragebar and range of duplicateWater ror represents the average bottle experiments. Water tered and range of duplicate-bottle samples were gravity-ltered lter experiments. Water samples through a glass ber lter gluwere gravity-filtered (GF/F, Whatman) in through the gluparaa glass fiber filter (GF/F, cose experiments for separaand Whatman) in particulate the glucose extion of the and gludissolved phases. In the gluperiments for separation of water tamate experiments, water the particulate and dissolved ough samplesIn were through phases. the passed glutamate exa 0.1-m polytetrauoroethoethperiments, water samples ylene membrane lter (OmOmwere passed through a 0.1nipore, Millipore) under rer rem polytetrafluoroethylene duced pressure. The DON DON membrane filter concentration at (Omnipore, the begineginMillipore) under reduced presning of the glucose experiperisure. DON concentration mentThe was omitted because it use it was not signicantly differat the beginning of the gluifferent from zero. Initial concencose experiment was omitted ncentrations it of DOC in glucose and glutamate experiments were accounted for in measured concentrations of these compounds, indicating that no because was not significantly different from zero. Initial concentrations of DOC inthe glucose and glutamate experiments were accounted for in the measured ose and glutamate experiments were forthat in the measured concentrations of these compounds, indicating that no other sources of inaccounted the initial culture medium. concentrations of DOC these were compounds, indicating no other sources of DOC were in the initial culture medium. ere in the initial culture medium.

918 10

4 MAY 2001 VOL 292 SCIENCE www.sciencemag.org

4 MAY 2001 VOL 292 SCIENCE www.sciencemag.org

composition and refractory nature of marine DOM. The amino sugar muramic acid is uniquely found in the repeating disaccharide backbone of the bacterial cell wall

(5). Overall, these results suggest that bacterial degradation sufficiently alters the structure of peptidoglycan so that its polysaccharide component is no longer recognizable at the molecular level.

Table 2. Chemical composition of bacterially derived DOM in the glucose and glutamate experiments. All Table 2. Chemical composition of bacterially derived DOM in the glucose and glutamate experiments. All are the are the mean of duplicate-bottle experiments, except for the data indicated by asterisks, which were mean of duplicate-bottle experiments, except for the data indicated by asterisks, which were obtained from single obtained from single bottles. The mean deviation of replicates was 0.9 M for DOC, 0.2 M for DON, 2 bottles. The mean deviation of replicates was 0.9 M for DOC, 0.2 M for DON, 2 for C/N ratio, and 1 to 2% of DOC, for C/N ratio, and 1 to 2% of DOC, DON, THAS, and THAA compositions. The uncharacterized fraction DON, THAS, and THAA compositions. The uncharacterized fraction in the glucose experiment is estimated as the in the glucose experiment is estimated as the difference between the total DOC and the DOC accounted difference between the total DOC and the DOC accounted for as THNS+THAS+THAA, and between the total DON for as THNSTHASTHAA, and between the total DON and the DON accounted for as THAATHAS, and the DON accounted for as THAA+THAS, except for the data after 1 year, in which THNS was not included (data except for the data after 1 year, in which THNS was not included (data in parenthesis). The uncharacin parenthesis). The uncharacterized fraction in the glutamate based only on THAA, because THNS terized fraction in the glutamate experiment is based onlyexperiment on THAA, is because THNS and THAS were not and THAS were not measured (indicated by nm). measured (indicated by nm). Initial substrate Glucose Chemical composition 2 DOC DON THNS THAS THAA Uncharacterized THAS THAA Uncharacterized Bulk Total characterized Uncharacterized GalN GlcN MA Acidic Basic Neutral Hydrophobic 30 1.0 2.5 0.7 2.5 94 3.4 22 75 32 7.5 41 20 55 26 26 12 23 38 Incubation time (days) 4 7 365 2 Glutamate Incubation time (days) 4 13 2.4 nm nm 18* 82* nm 31* 69* 5.7 3.1 6.3* nm nm nm 12 6 56 27 9 14 1.5 nm nm 12 88 nm 33 67 9.3 3.3 12 nm nm nm 17 7 36 40 560 9.0 0.7 nm nm 3.1 97 nm 11 89 13 3.6 14 nm nm nm 28 8 46 18

Concentration of the bulk DOM (M) 18 16 11 18 0.7 0.6 1.2 3.1 4.9 1.4 4.3 89 5.5 29 66 26 7.9 37 26 50 24 27 10 27 36 DOC composition (% as) 5.9 nm nm 1.5 3.8 nm 3.7 2.5 13 89 (94) 87 DON composition (% as) 6.6 5.6 nm 27 6.4 15 67 88 85 C:N molar ratio 28 8.9 9.2 (4.6) 37 (10) 5.1 3.5 5.4

and amino sugars typic of bacterial biomass ( chemicals were relativ nents of bacterially de pears that the selective usual biochemical co could not, by itself, ac formation and accumu derived DOM. Components of bac leased into the surroun through a variety of b including direct release (32), and grazing (33). is critical for the mic this DOM, and it app activity plays an impor mation of refractory D size (34 ). It is possible promiscuous activities cur with much lower ef ry activities (35), oc fragments from macro cape recognition by ba molecular-level chemi this scenario, the rate fractory DOM is depen microbial activity. Th mechanism could be re of the sequestration of f

References and Notes

THAS composition (mole %) 31 33 nm 56 67 nm 12 0 nm THAA composition (mole %) 21 30 26 11 11 8 31 41 24 37 18 41

1. R. Benner, J. D. Pakulski, P. G. Hatcher, Science 255 2. M. D. McCarthy, T. Pratu Nature 390, 150 (1997). 3. L. L. Clark, E. D. Ingall, R. (1998). 4. E. Tanoue, S. Nishiyam Geochim. Cosmochim. Act 5. M. D. McCarthy, J. I. Hedg 231 (1998). 6. P. M. Williams, E. R. M. Dr (1988). 7. J. I. Hedges et al., Org. Ge 8. G. R. Harvey, D. A. Boran Mar. Chem. 12, 119 (1983 9. J. E. Brophy, D. J. Carlso (1989).

www.sciencemag.org SCIENCE VOL 292 4 MAY 2001

DOM increased to 13. In both experiments, bacterially derived DOM was largely uncharacterized at the molecular level and rich in C relative to N. Most oceanic DOM is of marine origin (21, 22), refractory (20, 21, 23), molecularly uncharacterized (6, 7), rich in C relative to N (C:N = 10 to 25) (16, 24), and exists as rela-

tively small molecules (1, 15, 24). The simple experiments described here demonstrate that the bulk properties of bacterially derived DOM are similar to those of marine DOM. The size distribution of DOM in the glutamate experiment was investigated using ultrafiltration (15). After 6 months, 84 3% of the DOC was <10 kD in size, consistent with the size

distribution of marine DOM (>90% of DOC is <10 kD) (15). The amino acid composition of marine DOM is relatively rich in neutral amino acids (30 to 40 mole %) and depleted in basic amino acids (5 to 10 mole %) (25, 26) compared with marine organisms, including bacteria (neutral = 25 mole %; basic = 15 mole %) (27). During

11

12. An articial seawater medium (0 1 M C) was prepared using precombusted salts (NaCl, KCl, MgSO4, CaCl2 ) with NaHCO3 and organic-free, deionized water (Milli-RO and Milli-Q UV Plus). The ltrate (0.7 m pore size GF/F) from seawa10. L. J. samples Tranvik, FEMS Microbiol. Ecol. 12,inoculum 177 (1994). ter was used as a natural (2% DOM (28). In the glucose experiment, mu- selectively preserved during diagenesis (.30 ), 11. v/v) K. Stoderegger, G. J. Herndl, Limnol. 43, of marine bacteria. Based on theOceanogr DOC content ramic acid C concentrations of 35 to 89 nM but in these are relatively minor constituents 877 (1998). the Gulf of Mexico seawater inoculum, the nal 12. cells An articial seawater medium (0 accumulate. 1 Mmedium C) was concentration of DOC in the inoculated were measured during the first week of incu- of that should only slowly prepared using precombusted salts (NaCl, KCl, prioracids, to the addition of glucose was 1 sugars 1 M. bation, and based on a typical muramic acid Amino neutral sugars, and amino MgSO4,on CaCl with content NaHCO3 Based the in and the organic-free, Sagami Bay 2 )DOC content in peptidoglycan (29), we estimate typically compose >70% of bacterial biomass deionized water (Milli-RO and Milli-Q UV Plus). seawater inoculum, the nal concentration of DOC The ltrate (0.7 m poreprior size GF/F) from seawain the inoculated medium to the addition of that peptidoglycan accounted for 1% of the (29), but these biochemicals were relatively ter samples was as glutamate was 2 used 1 M.a natural inoculum (2% C and 7.5% of the N in DOM. Muramic acid minor components of bacterially derived v/v) of marine bacteria. Based on the DOC content 13. A. Skoog, R. Benner, Limnol. Oceanogr. 42, 1803 concentrations were below detection after 1 DOM. It appears that the selective preservation in the Gulf of Mexico seawater inoculum, the nal (1997). concentration of DOCAnal. in components the inoculated medium 14. unusual P. Lindroth, K. Mopper, Chem . 51, 1667 (1979). year of incubation (Table 2), indicating that of biochemical of cells prior to the of glucose was(1992). 1 1 M. 15. H. Ogawa, N. addition Ogura, Nature 356, 696 peptidoglycan was absent in the refractory could not, by itself, account for the rapid nts. All Based on the DOC content in the Sagami 16. H. Ogawa, R. Fukuda, I. Koike, Deep-Sea Res. I Bay 46, h were DOM derived from bacteria. This finding formation and accumulation of bacterially seawater inoculum, the nal concentration of DOC 1809 (1999). DON, 2 in the inoculated medium prior to the addition of 17. THNSDOM. is the sum of seven neutral sugars (glucose, was unexpected given recent observations in derived glutamate fucose, was 2 1 M. raction galactose, rhamnose, mannose, arabinose, marine DOM of high D/L enantiomer ratios 13. Components of bacterial cells are A. Skoog, R. and Benner, Limnol. . released 42, amino 1803 and xylose), THAS is the Oceanogr sum of three ounted in specific amino acids that are found in the intosugars the surrounding DOM through (1997). [galactosamine water (GalN), as glucosamine (GlcN), THAS, 14.variety P. Lindroth, Mopper, Anal. Chem . 51, measured 1667 (1979). and muramic acid (MA)], which were by peptide component of peptidoglycan (5). a of K. biological processes, including charac15. HPLC H. Ogawa, Ogura, Nature 356 696 (1992). after N. HCl hydrolysis (3 M, 5, hours, 100C) (13, ere not Overall, these results suggest that bacterial direct release ( 11 , 31 ), viral lysis ( 32 ), 16. 28 H. ). Ogawa, R. Fukuda, I. Koike, Deep-Sea Res. I 46 , No samples had measurable mannosamine. MA 1809 (1999). degradation sufficiently alters the structure and was grazing (33after ). Exoenzyme activity is not detected 1 year. THAA is the sum of 17 17. amino THNSfor is the sum of seven sugars (glucose, acids determined by neutral HPLC (14 ) after vaporof peptidoglycan so that its polysaccharide critical the microbial utilization of this galactose, fucose, rhamnose, mannose, is arabinose, phase hydrolysis (36 ). THAA composition presentcomponent is no longer recognizable at the DOM, and it appears that enzymatic activity and xylose), and THAS is the sum of three amino ed in Table 2 as four families of amino acids: acidic sugars (GalN), glucosamine (GlcN), molecular level. plays an [galactosamine important role in basic the formation of (aspartic acid, glutamic acid), (arginine, histiand muramic acid which measured by R that E (MA)], P (serine, O Tsmall S weresize dine, lysine), neutral glycine, threonine, Our results are inconsistent with several refractory DOM is R of (34).tyIt Spatial HPLC after HCl hydrolysis (3 M, 5 hours, 100C) ( 13, rosine), and hydrophobic (-alanine, alanine, -amimechanisms that have been proposed for the is possible that nonspecific or promiscuous inositid 28 ). No samples had measurable mannosamine. MA 10. L. J. Tranvik, FEMS Microbiol. Ecol. 12, 177 (1994). nobutyric acid, methionine, phenylalanine, valine, was not of detected after 1 concentrations year. is the of 17 11. K. Stoderegger, G. J. Herndl, Limnol.matter Oceanogr , isoleucine, leucine). The of sum individual formation of refractory organic in. 43 the activities enzymes, that THAA occur with much induce amino acids by as HPLC (14) after vapor877 (1998). acids determined are available supplementary Web ocean. It is unlikely that abiotic condensation lower efficiency than primary activities ing, cy phase hydrolysis ( 36 ). THAA composition is present12. An articial seawater medium (0 1 M C) was tables at www.sciencemag.org/cgi/content/full/292/ 560 reactions or humification processes ( 8 ) were ( 35 ), occasionally produce fragments from membr ed in Table 2 as four families of amino acids: acidic prepared using precombusted salts (NaCl, KCl, 5518/917/DC1 acid, R. glutamic basic (arginine, histiMgSO4, for CaCl organic-free, 18. (aspartic J. A. Fuhrman, L. Ferguson, Mar. Ecol. Prog. Ser . 33, important producing refractory in our macromolecules that acid), escape recognition 2). Ph 2 ) with NaHCO 3 and DOM lysine), neutral (serine, glycine, threonine, tydeionized water and were Milli-Q UV Plus). 237 (1986). role in Spatial study, because the (Milli-RO incubations conducted by dine, bacterial enzymes ( and molecular-level 9.0 rosine), and hydrophobic -alanine, alanine, -amiThe ltrate (0.7 m pore size GF/F) from seawa19. J. Rich, H. W. Ducklow, D. L. Kirchman, Limnol. Ocean0.7 tidylin inositid in the dark at low temperature, concentrations chemical analyses. Given this scenario, nobutyric acid, methionine, phenylalanine, valine, ter samples was used as a natural inoculum (2% ogr. 41, 595 (1996) leucine). concentrations individual v/v) of marine bacteria. on the DOC content 20. isoleucine, N. Ogura, Biol. The 13, of 89 (1972). induce the rate of Mar. formation refractoryofDOM is genera of reactants were low, Based and the compositional amino acids areE.available as supplementary in the Gulf of Mexico seawater inoculum, the nal 21. P. M. Williams, R. M. Druffel, Nature 330, Web 246 1,4,5-t nm ing, cy dependent on the rate of microbial activity. changes were rapid (hours to days). Thus, it tables at www.sciencemag.org/cgi/content/full/292/ concentration of DOC in the inoculated medium (1987). nm glycero membr This relatively simple mechanism could be appears that biological processes were critical 5518/917/DC1 prior to the addition of glucose was 1 1 M. 2 (1997). 22. S. Opsahl, R. Benner, Nature 386, 480 1 Kiyoko Fukami, * Kazuki 3.1 18. R. L. Ferguson, Mar. Ecol. Prog. Ser. 33 , Nakao, cloned 2). Ph Based on the DOC content in DOM. the Sagami Bay 23. J. R.A. T.Fuhrman, Barber, Nature 220 , 274 (1968). responsible for much of the sequestration of for the formation of refractory Specific 4 97 237 (1986). Kataoka, Kurokawa, seawater inoculum, the nal concentration of DOC 24. R. Benner, B.Yuki Biddanda, B. Black, M. D.Manabu McCarthy, Mar. be rolediv in C in the ocean. biochemical components of cells can be fixed 19. J. Rich, H. Ducklow, D. L. Kirchman, in the inoculated medium prior to the addition of 2 Limnol. Ocean2 Chem . 57 ,W. 243 (1997). Kenji Nakamura, Motoya Katsuki, K on tidylin ogr . 41 , 595 (1996) glutamate was 2 1 M. 25. M. McCarthy, J. Hedges, R. Benner, Mar. Chem. 55, nm activat 20. N. Ogura, Mar. Biol. 13,Nobuaki 89 (1972). Yoshida,4 Tadaom genera 13. A. Skoog, R. Benner, Limnol. Oceanogr. 42, 1803 281 (1996). 11 21. Williams, M.G. Druffel, Nature 330, . 246 References and Notes (1997). 26. P. U. M. Hubberten, R. E. J. R. Lara, Kattner, Mar. Chem 45, PLC is 1,4,5-t 89 (1987). 1. R. Benner, J. D. Pakulski, M. McCarthy, J. I. Hedges, 14. P. Lindroth, K. Mopper, Anal. Chem. 51, 1667 (1979). 121 (1994). form i glycero , 480 (1997). S. Opsahl, R. Benner, Nature 386 22. Several C (PLC). isoforms have b P. G. Hatcher, Science 255, 1561 (1992). 15. H. Ogawa, N. Ogura, Nature 356, 696 (1992). 27. G. L. Cowie, J.phospholipase I. Hedges, Limnol. Oceanogr 37, 703 plecks cloned 23. (1992). R. T. Barber, Nature gametes, 220, 274 (1968). 16. H. Ogawa, R. Fukuda, I. Koike, Deep-Sea Res. I 46, 2. M. D. McCarthy, T. Pratum, J. Hedges, R. Benner, mammalian and splicing isoforms of 13 24. K. R. Benner, B. Benner, Biddanda, B. Black, M. D. Mar. 1809 (1999). Nature 390, 150 (1997). 28. Kaiser, R. Anal. Chem . 72 , McCarthy, 2566 (2000). to func be mic div 3.6 pressed testis. Here we report that male Chem 57, 243in (1997). 17. THNS is the sum of seven neutral sugars (glucose, 29. F. C. . Neidhardt, J. L. Ingraham, M. Schaechter, 3. L. L. Clark, E. D. Ingall, R. Benner, Nature 393, 426 PtdIns on 14 been disrupted either produced few small litte 25. M. McCarthy, J. Hedges, R. Benner, Mar. Chem . 55 , galactose, fucose, rhamnose, mannose, arabinose, Physiology of the Bacterial Cell: A Molecular Ap(1998). and me activat 281tilization (1996). and xylose), and THAS is the sum of three amino proach (Sinauer, Sunderland, MA, 1990), 102 studies showed that pp. insemination wit 4. E. Tanoue, S. Nishiyama, M. Kamo, A. Tsugita, 26. U. Hubberten, R. J. Lara, G. Kattner, Mar. Chem. 45, sugars [galactosamine (GalN), glucosamine (GlcN), 130. types h PLC is Geochim. Cosmochim. Acta 59 , 2643 (1995). nm signicantly fewer eggs becoming activated an 121 and muramic acid (MA)], which were measured by 30. P. G.(1994). Hatcher, E. C. Spiker, N. M. Szeverenyi, G. E. 5. M. D. McCarthy, J. I. Hedges, R. Benner, Science 281, To form i nm associated with fertilization were absent or de 27. G. L. Cowie, J. I. Hedges, Limnol. Oceanogr . 37 , 703 HPLC after HCl hydrolysis (3 M, 5 hours, 100C) (13, Maciel, Nature 305, 498 (1983). 231 (1998). tance nm plecks (1992). 28). No samples had measurable mannosamine. MA 31. A. W. Decho, Oceanogr. Mar.the Biol. acrosome Annu. Rev. 28reaction, , 73 unable to initiate an 6. P. M. Williams, E. R. M. Druffel, Oceanography 1, 14 28. (1990). K. Kaiser, R. Benner, Anal. Chem. 72, 2566 (2000). was not detected after 1 year. THAA is the sum of 17 PLC 4 to func (1988). fertilization and induced by interaction with th 29. L. F. M. C. Proctor, Neidhardt, J. L. Ingraham, M. ,Schaechter, amino acids determined by HPLC (14) after vapor32. J. A. Fuhrman, Nature 343 60 (1990). disrupt PtdIns 28 7. J. I. Hedges et al., Org. Geochem. 31, 945 (2000). These data demonstrate that PLC 4 functions in Physiology of L. the Bacterial Cell: A Molecular phase hydrolysis (36). THAA composition is present33. T. Nagata, D. Kirchman, Mar. Ecol. Prog. Ser. Ap83, 8. G. R. Harvey, D. A. Boran, L. A. Chesal, J. M. Tokar, 8 and me proach (Sinauer, Sunderland, MA, 1990), pp. 102 mammali ed in Table 2 as four families of amino acids: acidic during induced by the zona pellucida 233 (1992). Mar. Chem. 12, 119 (1983). 46 types h 130. (aspartic acid, glutamic acid), basic (arginine, histi34. R. M. W. Amon, R. Benner, Limnol. Oceanogr. 41, 41 1 9. J. E. Brophy, D. J. Carlson, Deep-Sea Res. 36, 497 18 Depart R E P(serine, ORT S 30. (1996). P. G. Hatcher, E. C. Spiker, N. M. Szeverenyi, G. E. dine, lysine), neutral glycine, threonine, tyTot Spatial and temporal activation of phosphoATG (1989). ogy and Maciel, Nature , 498 (1983). rosine), and hydrophobic (-alanine, alanine, -ami35. P. J. OBrien, D.305 Herschlag, Chem. Biol. 6, 91 (1999). tance inositide turnover enables eukaryotic cells to PH do 10. L. J. Tranvik, FEMS Microbiol. Ecol. 12, 177 (1994). 31. A. W. Decho, Biol. Annu. Rev. 28 , 73 nobutyric acid, methionine, phenylalanine, valine, Medicin 36. Tsugita, T. Oceanogr. Uchida, H. Mar. W. Mewes, T. Atake, J. BioPLCE 4 11. K. Stoderegger, G. J. Herndl, Limnol. Oceanogr. 43, (1990). isoleucine, leucine). The concentrations of individual Gene chem . 102, 1593 (1987). such as cell signalinduce various functions mutant 4 MAY 2001 877 (1998). 919 amino acids are available as supplementary Web 37. 32. We L. M. Proctor, J. A. Fuhrman, Nature 343 , 60 (1990). thank Y. Fujimoto of the Ocean Research InstiMedical disrupt ing, cytoskeletal remodeling, phagocytosis, analysi 12. An articial seawater medium (0 1 M C) was 33. tute T. Nagata, D. L. Kirchman, Prog. Ser. 83 tables at www.sciencemag.org/cgi/content/full/292/ for nutrient analyses.Mar. H.O.Ecol. acknowledges for, 8039, Ja 1, membrane traffic, and ion channel activity brain prepared using precombusted salts (NaCl, KCl, 233 (1992). 5518/917/DC1 funding from the Ministry of Education, Science ( and Sciences MgSO4, CaCl2 ) with NaHCO3 and organic-free, 34. R. M. W.of Amon, R. Benner, Limnol. Oceanogr . 41 , 41 18. J. A. Fuhrman, R. L. Ferguson, Mar. Ecol. Prog. Ser. 33, 2 ). Culture Phospholipase C (PLC) plays a crucial PLC A 4 Japan. R.B. acknowledges funding from setts, 1 Depart deionized water (Milli-RO and Milli-Q UV Plus). (1996). 237 (1986). We turnover thank J. Hedges and the Marine Organic tute, R roleNSF. in this by hydrolyzing phosphaexpress ogy and The ltrate (0.7 m pore size GF/F) from seawa35. Geochemistry P. J. OBrien, D. Herschlag, Biol. 6Washington , 91 (1999). 19. J. Rich, H. W. Ducklow, D. L. Kirchman, Limnol. Oceangroup of the Chem. University of Japan. ) Bioto tidylinositol-4,5-bisphosphate (PtdInsP 1C). H Medicin ter samples was used as a natural inoculum (2% 36. for A. Tsugita, T. Uchida, H. W. Mewes, T. Atake,2J. ogr. 41, 595 (1996) comments that improved the manuscript. Gene E v/v) of marine bacteria. Based on the DOC content *To wh chem. 102 , 1593 (1987). messengers, inositol 20. N. Ogura, Mar. Biol. 13, 89 (1972). generate two second phenot in the Gulf of Mexico seawater inoculum, the nal 37. 21 WeNovember thank Y. Fujimoto of the Ocean InstiMedical 21. P. M. Williams, E. R. M. Druffel, Nature 330, 246 mail: kf 2000;[Ins(1,4,5)P accepted 21 March 2001 ] Research and diacyl1,4,5-trisphosphate crosses 3 concentration of DOC in the inoculated medium tute for nutrient analyses. H.O. acknowledges for (1987). 8039, Ja glycerol. Eleven PLC of cDNAs have been pups w prior to the addition of glucose was 1 1 M. 2 from 3 funding the Ministry Education, Science and 22. S. Opsahl, R. Benner, Nature 386, 480 1 (1997). Sciences Kiyoko Fukami, * Kazuki Nakao, Takafumi Inoue, cloned from mammalian cells, and they can indicat Based on the DOC content in the Sagami Bay Culture of Japan. R.B. acknowledges funding from 23. R. T. Barber, Nature 220, 274 (1968). 920 4 MAY 2001 setts, A 4 5 5 Kataoka, Kurokawa, Rafael Fissore, seawater inoculum, the nal concentration of DOC We thank Hedges and the 24. R. Benner, B. Yuki Biddanda, B. Black, M. D.Manabu McCarthy, Mar. tute, re R be NSF. divided into J.A. four types ,Marine , Organic , and not in the inoculated medium prior to the addition of 2 2 3,6 Geochemistry group of the University of Washington Chem. 57 , 243 (1997). Japan. Kenji Nakamura, Motoya Katsuki, Katsuhiko Mikoshiba, on the basis of sequence homology and zygous glutamate was 2 1 M. for comments that improved the manuscript. 25. M. McCarthy, J. Hedges, R. Benner, Mar. Chem. 55 , 1 *To wh healthy activation mechanisms (3). The delta type of Nobuaki Yoshida,4 Tadaomi Takenawa 13. A. Skoog, R. Benner, Limnol. Oceanogr. 42, 1803 281 (1996). mail: kf 21 21 March the 2001 (1997). 26. U. Hubberten, R. J. Lara, G. Kattner, Mar. Chem. 45, no obv PLC isNovember thought 2000; to beaccepted evolutionarily oldest 14. P. Lindroth, K. Mopper, Anal. Chem. 51, 1667 (1979). 121 (1994). kidney, form in the mammalian PLC family, and its C (PLC). 37 isoforms have been found in male and female 15. H. Ogawa, N. Ogura, Nature 356, 696 (1992). 27. G. L.Several Cowie, J. phospholipase I. Hedges, Limnol. Oceanogr , 703 homology (PH) domain exis known 920 pleckstrin 4 MAYhomozy 2001 H. Ogawa, R. Fukuda, I. Koike, Deep-Sea Res. I 46, (1992). mammalian gametes, and splicing isoforms of PLC 4 are predominantly 12 16. 1809 (1999). 28. K. Kaiser, R. Benner, Anal. Chem. 72, 2566 (2000). paired, to function in membrane association through pressed in testis. Here we report that male mice in which the PLC 4 gene had 17. THNS is the sum of seven neutral sugars (glucose, 29. F. C. Neidhardt, J. L. Ingraham, M. Schaechter,

ng that actory inding ations tiomer found glycan at bacrs the polyrecog-

appears that biological processes were critical for the formation of refractory DOM. Specific biochemical components of cells can be selectively preserved during diagenthe the amino compositions esisincubations, (30), but these are acid relatively minor constituents ofderived cells that should only slowly of bacterially DOM rapidly shifted accumulate. Aminosimilar acids, to neutral from a composition that ofsugars, microand amino to sugars typically compose 70% organisms a composition similar to that of of bacterial (29 ),similar but these biomarine DOMbiomass (Table 2). A transition chemicals were minor compowas observed in relatively amino sugar compositions nents of bacterially DOM. It apof bacterially derivedderived DOM in the glucose pears that the selective preservation of unincubations. The initial amino sugar composiusual componentswas of similar cells tion of biochemical DOM in the incubations could not, by itself, account for the rapid to that in bacterial cells (28), whereas only formation and accumulation bacterially glucosamine and galactosamineof were detected derived DOM. at the end of the incubation (Table 2). The Components of bacterial cells reglucosamine:galactosamine ratio (2:1)are in bacleased into the DOM surrounding water terially derived at the end of as theDOM incuthrough a variety ofthe biological processes, bation was similar to ratio of these amino including direct release ( 11, 31 ), viral lysis sugars in seawater (28). (32), and grazing (33). Exoenzyme activity Recent studies have identified specific is critical for the microbial utilization of components of the cell envelopes of marine this DOM, and it appears that enzymatic bacteria in DOM from the surface and deep activity plays an important role in the forocean (4, 5). In addition to this evidence for mation of refractory DOM that is of small a bacterial source of marine DOM, laboratory size (34 ). It is possible that nonspecific or studies with radiotracers and marine bacteria promiscuous activities of enzymes, that ocnoted themuch production refractory DOM from cur with lowerof efficiency than primasimple compounds 11). Results from our ry activities (35),(9occasionally produce study support these observations, and indicate fragments from macromolecules that esthat processing by microorganisms capediagenetic recognition by bacterial enzymes and is rapid and critical for shaping the composimolecular-level chemical analyses. Given tion and refractory nature of marine DOM. this scenario, the rate of formation of reThe amino sugar muramic acid is uniquely fractory DOM is dependent on the rate of found in the repeating disaccharide backbone microbial activity. This relatively simple of the bacterial cell wall polymer commonly mechanism could be responsible for much known as peptidoglycan, and therefore of the sequestration of fixed C it inis the ocean. an excellent biomarker for peptidoglycan in

Requirement of Ph C4 for the PellucidaInduced Reaction

Requirement of Phospholipase C4 for the Zona PellucidaInduced Acrosome Reaction

Bulk Chemical Characteristics of Dissolved Organic Matter in the Ocean

Ronald Benner,* J. Dean Pakulski, Matthew McCarthy, John I. Hedges, Patrick G. Hatcher Dissolved organic matter (DOM) is the largest reservoir of reduced carbon in the oceans. The nature of DOM is poorly understood, in part, because it has been difficult to isolate sufficient amounts of representative material for analysis. Tangential-flow ultrafiltration was shown to recover milligram amounts of >1000 daltons of DOM from seawater collected at three depths in the North Pacific Ocean. These isolates represented 22 to 33 percent of the total DOM and included essentially all colloidal material. The elemental, carbohydrate, and carbon-type (by 13C nuclear magnetic resonance) compositions of the isolates indicated that the relative abundance of polysaccharides was high (~50 percent) in surface water and decreased to ~25 percent in deeper samples. Polysaccharides thus appear to be more abundant and reactive components of seawater DOM than has been recognized. issolved organic matter (DOM) in the oceans is important in the global carbon cycle (1), supports heterotrophic activity (2), and affects the penetration of light and exchange of gases at the sea surface (3). Despite its importance, relatively little is known about the composition and reactivity of marine DOM because of the lack of suitable methods for its isolation from seawater. Direct biochemical analyses of DOM in seawater typically account for less than 15% of the total mixture (4, 5), and many techniques that would provide more comprehensive structural information, such as nuclear magnetic resonance (NMR) and infrared spectroscopy, require the isolation of DOM from the much more abundant salts in seawater. The conventional method for isolation has been adsorption of acidified DOM onto nonionic XAD resins (6). However, this method typically recovers a small fraction (5 to 15%) of the total DOM in seawater, requires large manipulations of pH during isolation, and is selective for hydrophobic constituents. In contrast, tangential-flow ultrafiltration concentrates organic molecules primarily on the basis of size rather than chemical properties (7) and requires no pH adjustments that may change chemical associations and structures. Ultrafiltration therefore appears to be a more appropriate method for isolating a rep-

R. Benner and J. D. Pakulski, Marine Science Institute, University of Texas at Austin, Port Aransas, TX 78373. M. McCarthy and J. I. Hedges, School of Oceanography, University of Washington, Seattle, WA 98195. P. G. Hatcher, Fuel Science Program, The Pennsylvania State University, University Park, PA 16802. *To whom correspondence should be addressed.

resentative fraction of DOM than adsorption on XAD resins. In this report we describe the bulk chemical properties of DOM isolated by tangential-flow ultrafiltration from three depths in the open North Pacific Ocean. Samples were collected during April 1991 from Station ALOHA (2245N, 15800W) located about 100 km north of Oahu, Hawaii. Hydrographic and basic chemical and biological parameters have been monitored at this station since 1988 (8). We collected water samples in Niskin bottles fitted with Tefloncoated closure springs using a 24-bottle rosette. Samples for ultrafiltration (200 liters) were collected from three depths, 10, 765, and 4000 m, corresponding to the surface, oxygen-minimum, and deep waters (Table 1). Immediately following collection, we pumped water samples through 0.2-m pore size, prerinsed Nuclepore polycarbonate filters and subsequently through 1000-dalton cutoff (~1-nm pore size) spiralwound polysulfone filters using an Amicon DC1OL ultrafiltration system. Total processing time for each sample was ~12 hours. Dissolved organic carbon (DOC) is operationally defined here as all organic carbon passing a 0.2-m pore size filter and therefore includes all colloids <0.2 m. We measured concentrations of DOC by a high-temperature oxidation method (9) in which a Shimadzu TOC 5000 analyzer was used (10). Concentrations of DOC and volumes of water were recorded for the initial water, the ultrafiltered concentrate, and the ultrafiltrate so that a mass balance of DOC could be established (11). Most sea salts were removed from ultrafiltered concentrates by diafiltration, and the samples were stored frozen for transport to the laboratory. Samples were dried under vacuum in a Savant SpeedVac concentrator.
Originally published 20 March 1992

The concentration of DOC in surface water (82 M) was approximately two times as large as concentrations in oxygen-minimum and deep waters (Table 1). DOC profiles of this magnitude and shape have been observed at many ocean sites and interpreted to reflect the presence of labile organic matter of planktonic origin in surface water that is absent from older water at depth. Mass balance calculations indicate that all of the DOC in initial water samples was accounted for in the ultrafiltered concentrates (> 1000 daltons) and ultrafiltrates (<1000 daltons) (Table 1); thus, contamination and processing losses were minimal. We recovered 33, 25, and 22% of the total DOC from surface, oxygenminimum, and deep waters (Table 1). The observed trend of a decrease in the >1000-dalton fraction from surface water to deep water suggests that material in the colloidal size range comprised an important part of the reactive components of DOM. Our results on the size distribution of DOM are similar to those of Carlson et al. (11), who found that ~34% of the persulfateoxidizable DOC (~88 M) in North Atlantic surface waters was > 1000 daltons. However, a recent investigation by Sugimura and Suzuki (9) in the Northwestern Pacific Ocean indicated that ~85% of the total DOC (~290 M) in surface waters was >1800 daltons. Similarly, these investigators found that ~85% of the total DOC (~95 M) in Northwestern Pacific deep water (4000 m) was >1800 daltons (9). Such large differences between studies are beyond the range expected for natural variability, especially for two samples from 4000-m depth in the North Pacific Ocean, and suggest that there are systematic biases in DOM measurement or characterization. Because of uncertainty about the accuracy of seawater DOC measurements (5, 9), we also independently determined the organic carbon content of the ultrafiltered DOM (UDOM) by flash combustion (~1100C) of the dried powder in a Carlo Erba 1106 CHN analyzer of the type used to quantitatively combust a wide variety of carbonaceous substances. The weight percentages of organic carbon measured were 10 to 25% lower than those calculated from the total DOC content (Shimadzu analyzer) and mass of the isolated samples (Table 1). Thus, we found no evidence for low DOC measurements that could be attributed to incomplete oxidation of the higher molecular weight or colloidal components of our DOM samples. The atomic C/N values of the UDOM were lowest (15.3) in surface water and highest (22.5) in oxygen-minimum water (Table 2). These values lie in the middle of the range of 10 to 25 determined for bulk seawater DOM with the use of wet oxidation methods (4, 5). Humic substances isolated from

13

Table 1. Hydrographic data for samples collected at Station ALOHA (2245.0N, 1580.OW) and the concentrations of dissolved organic carbon (DOC), percentages of DOC recovered using tangential-flow ultrafiltration, and weight percentages of organic carbon (Wt%OC) in the recov-

ered isolates as determined by direct-injection, high-temperature catalytic oxidation (HTCO) and the dry weights of the recovered isolates and by flash combustion of dried powders in a CHN analyzer.

*% Initial DOC = 100 (DOCconcentrate + DOCultrafiltrate)(DOCinitial water )-1. Sea salts were not completely removed during diafiltration resulting in weight percentages of organic carbon that are lower than those expected (~45%) for pure organic matter.

Table 2. Atomic C/N ratios from elemental analysis and area percentages of the major 13C NMR peaks for DOM isolated by ultrafiltration. Functional group assignments are: C-C, carbon singly bonded to carbon; C-O, carbon singly bonded to oxygen; O-C-O, carbon singly bonded to two oxygens; C=C, carbon doubly bonded to carbon; O-C=O, carboxyl and ester carbon; C=O, carbonyl carbon.

the Pacific Ocean are carbon-rich [C/N ratio 34 (12)] relative to our UDOM samples; this difference reflects the bias of XAD resins against more polar nitrogen-containing DOM. Concentrations of dissolved organic nitrogen (DON) in the >1000-dalton size fraction (0.4 to 1.8 M; from Tables 1 and 2) agree with concentrations determined recently for total DON (2 to 6 FM) in the North Pacific Ocean with the use of both high-temperature combustion (1100C) and ultraviolet oxidation methods (13) and agree as well with average concentrations of total DON (5 to 8 M) reported for these same water samples (14). These measurements are at odds, however, with other recent estimates of DON concentrations and C/N values for marine DOM. Suzuki and colleagues (9, 15) reported average C/N values near 7 for the total DOM in surface and deep waters of the Northwestern Pacific Ocean and concentrations of higher molecular weight (> 5000 daltons) DON that were 20 to 70 times as large as our estimates for DON >1000 daltons. As discussed above, these large differences are beyond those expected for natural variability. Structural features of the isolated DOM were investigated by cross polarization magic angle spinning (CP/MAS) 13C NMR (16). The 13 C NMR spectrum of the UDOM from surface water shows a dominant peak at 72 ppm and lesser peaks at 21, 40, 100, and 176 ppm (Fig. 1). Carbohydrates are the major family of biomolecules producing CO resonances at 72 ppm and OCO resonances at 100 ppm. Unsubstituted alkyl carbons, such as methyl and methylene carbons, and amine carbons give rise to resonances between 0 and 50

ppm, and carboxyl, ester, and amide carbons produce resonances around 176 ppm. Major compositional differences were apparent between UDOM from surface water and UDOM from oxygen-minimum and deep waters (Fig. 1). Integrated areas for CO and OCO resonances indicate that there was a sharp decrease in the relative abundance of carbohydrate carbon from 54% in the UDOM from surface water to 29 and 25% in UDOM from oxygen-minimum and deep waters (Table 2). Surface-water UDOM had a threefold lower relative abundance of aromatic or olefinic carbons (in 110 to 160 ppm) than UDOM from deeper waters (Table 2). On the basis of these data and the vertical distribution of radiocarbon in DOM in the North Pacific Ocean (17), which indicates that labile and refractory components of DOM coexist in surface water, whereas refractory components dominate in deeper waters, it appears that high molecular weight polysaccharides (> 1000 daltons) made up an important part of the labile DOM in surface water. Estimates of carbohydrate concentrations in the colloidal size fraction (1 to 200 nm) of DOM ranged from ~14 M C in surface waters to ~2 M C in deeper waters (Fig. 2). The main structural features of the surface-water UDOM were very different from those of humic substances isolated by hydrophobic adsorption from seawater. The UDOM from surface water was rich in carbohydrates, whereas XAD-isolated humic substances are rich in unsubstituted alkyl and carboxyl carbons (18). Moreover, structural features of XAD-isolated DOM revealed by 1H- and 13CNMR depict a material that is compositionally

invariant with oceanic environment (18) and depth (19); these characteristics suggest that the XAD-isolated DOM is composed primarily of the older, more refractory components of DOM. There are, however, some similarities between the structural features of UDOM and humic substances from deep water. Both of the isolated fractions contained a significant aromatic or olefinic component and similar carboxyl contents, but the deepwater UDOM sample was enriched in carbohydrate and depleted in unsubstituted alkyl carbon relative to deep-water humic substances (19). We also determined the carbohydrate content of the UDOM samples using a modified MBTH assay (20) to confirm the 13C-NMR carbohydrate estimates. In the modified MBTH assay hydrochloric acid was replaced with sulfuiric acid to more completely hydrolyze resistant polysaccharides, such as cellulose and chitin (21). The modified MBTH assay indicated that carbohydrates accounted for 49, 18, and 19% of the total C in surface, oxygen-minimum, and deep-water UDOM samples. The similarity between these data and those determined by 13C-NMR confirms our estimate of carbohydrates in the UDOM samples and indicates that the modified

Fig. 1. Conventional CP/MAS 13C NMR spectra of DOM isolated by ultrafiltration (>1000 daltons) from surface water (10 m), oxygen-minimum water (765 m), and deep water (4000 m) at 2245N, 15800W in the Pacific Ocean.

14

found glycan t bacs the polyrecog-

nts. All h were DON, 2 raction ounted THAS, characere not

560 9.0 0.7 nm nm 3.1 97 nm 11 89 13 3.6 14 nm nm nm 28 8 46 18

constituents of cells that should only slowly accumulate. Amino acids, neutral sugars, and amino sugars typically compose 70% of bacterial biomass (29), but these biochemicals were relatively minor components of bacterially derived DOM. It appears that the selective preservation of unusual biochemical components of cells could not, by itself, account for the rapid formation and accumulation of bacterially derived DOM. Components of bacterial cells are released into the surrounding water as DOM through a variety of biological processes, including direct release (11, 31), viral lysis (32), and grazing (33). Exoenzyme activity is critical for the microbial utilization of this DOM, and it appears that enzymatic activity plays an important role in the formation of refractory DOM that is of small size (34 ). It is possible that nonspecific or Fig. 2. Concentrations of dissolved carbohypromiscuous activities of enzymes, that ocdrates in ultrafiltered DOM as determined by 13C cur with much lower efficiency than primaNMR (UDOM-NMR) and a modified MBTH assay ry activities (and 35),the occasionally produce (UDOM-TCHO) concentrations of total fragments from macromolecules that esdissolved carbohydrates as determined by using cape recognition bacterial enzymes and the same modified by MBTH assay (TCHO). molecular-level chemical analyses. Given this scenario, the rate of formation of reMBTH assay yielded reasonably accurate esfractory DOM is dependent on the rate of timates of total carbohydrate content. microbial activity. This relatively simple Because ~70% of marine DOM was <1000 mechanism could be responsible for much daltons (<1 nm), we do not know if the comof the sequestration of fixed C in the ocean.

position of the UDOM was representative of the total DOM. We used the modified MBTH assay to determine the total dissolved carbohydrate concentrations in the seawater samples collected for ultrafiltration. The modified MBTH assay yielded total dissolved carbohydrate concentrations of 27, 13, and 7 M C in surface, oxygen-minimum, and deep-water samples (Fig. 2). These dissolved carbohydrate concentrations corresponded to 33, 34, and 17% of the total DOC, and were similar to those in the UDOM samples. Thus, ultrafiltration recovered a representative fraction of the MBTH-reactive carbohydrates from seawater. Our measurements of total dissolved carbohydrate concentrations are higher than most earlier estimates, which indicated that carbohydrates accounted for 10 to 15% of the total DOC measured by wet oxidation methods (5, 22) although values as high as ~35% were reported for the Caribbean Sea (23). We believe that the higher concentrations of carbohydrates measured with the modified MBTH assay result from the more complete hydrolysis of structural polysaccharides (21), which, as we have shown, appear to be abundant in the upper ocean.

The application of tangential-flow ultrafiltration to seawater samples has allowed a broad chemical characterization of a relatively large fraction of marine DOM, including all colloidal material. Compared to conventional hydrophobic adsorption techniques, ultrafiltration is less chemically selective and capable of recovering a greater portion of DOM. Our initial characterizations indicated that the concentration of the >1000- dalton size fraction of seawater DOM is one-tenth of that indicated by Suzuki and colleagues (9, 15) and that the C/N values are two to three times as large. Despite these differences, we concur that high molecular weight components of DOM are reactive and suggest that polysaccharides are dominant components of this reactive material, perhaps supporting much of the heterotrophic activity in the surface ocean. Use of ultrafiltration to recover DOM from seawater for experimental manipulations and detailed molecular-level characterizations will help resolve the above differences and provide a new understanding of the sources and reactions of the organic molecules composing this major carbon reservoir.

References and Notes

1. R. Benner, J. D. Pakulski, M. McCarthy, J. I. Hedges, P. G. Hatcher, Science 255, 1561 (1992). 2. M. D. McCarthy, T. Pratum, J. Hedges, R. Benner, Nature 390, 150 (1997). 3. L. L. Clark, E. D. Ingall, R. Benner, Nature 393, 426 (1998). 4. E. Tanoue, S. Nishiyama, M. Kamo, A. Tsugita, Geochim. Cosmochim. Acta 59, 2643 (1995). 5. M. D. McCarthy, J. I. Hedges, R. Benner, Science 281, 231 (1998). 6. P. M. Williams, E. R. M. Druffel, Oceanography 1, 14 (1988). 7. J. I. Hedges et al., Org. Geochem. 31, 945 (2000). 8. G. R. Harvey, D. A. Boran, L. A. Chesal, J. M. Tokar, Mar. Chem. 12, 119 (1983). 9. J. E. Brophy, D. J. Carlson, Deep-Sea Res. 36, 497 (1989).

4 MAY 2001

919

15

Microbes, Molecules, and Marine Ecosystems

Farooq Azam and Alexandra Z. Worden ntonie van Leeuwenhoek (1632 1723), the first observer of bacteria, would be surprised that over 99% of microbes in the sea remained unseen until after Viking Lander (1976) set out to seek microbial life on Mars. Through much of the 19th and 20th centuries, microbiologists focused on life-threatening pathogenic microbes or microbes as models of how life works at the molecular level. But heightened concerns for ocean health and biodiversity, highlighted in the 2003 Pew Oceans Commissions report (1), prompted microbial explorations of the sea with state-of-the-art tools such as satellite and laser-based imaging, as well as massive genomic surveys rivaling the human genome project. This has led to a gold rush of discoveries underscoring critical influences of microbes on marine ecosystems and carrying significant implications for sustainability, global climate, and human health. The challenge is integrating these discoveries at the systems level to elucidate microbial roles in overall system resilience. Understanding the role of microbes in structuring healthy and stressed marine ecosystems will provide the mechanistic basis for prognostic models. Accomplishing this may require a new and unifying framework for conceptualizing these ecosystems. Excitement over the microbial ocean was roused by the 1977 seminal discovery by Hobbie (2) and later studies that pelagic bacteria are hugely abundant (106 ml1), accounting for most oceanic biomass and metabolism (3). They had eluded detection because most are extremely smallonly a few percent of Escherichia coli in volumeand uncultivable. Thus, in order to measure bacterial biodiversity and in situ metabolism, microbial oceanographers had to devise cultivation-independent methods. Results show that bacteria are a major biological force in the oceanic carbon cycle and ecosystem structure. Photosynthetic bacteria Prochlorococcus

F. Azam is in the Marine Biology Research Division, Scripps Institution of Oceanography, University of California, San Diego, La Jolla, CA 92093, USA. E-mail: fazam@ucsd.edu A. Z. Worden is at the Marine Biology and Fisheries Division, Rosenstiel School of Marine and Atmospheric Science, University of Miami, Miami, FL 33149, USA. E-mail: aworden@rsmas.miami.edu

and Synechococcus are the most abundant oceanic primary producers, and their sequenced genomes provide new ecological insights (46). Other discoveries confront oceanographers and conservation biologists with the probability that many functional capabilities and mechanisms remain unknown. For instance, a bacterial gene for proteorhodopsin was discovered 3 years ago (7). This pigment, thought to exist only in Archaea, converts light energy directly into an electrical gradient across the cell membrane. The gene occurs in divergent marine bacterial taxa and diverse environments (8); hence, inclusion of proteorhodopsin activity may change oceanic energy budgets. Another important example is the discovery of oceanwide distribution of anoxygenic phototrophic bacteria that contribute to oceanic energy and carbon budgets (9). Nitrogen budgets, a regulating force of carbon processing and sequestration, must also be reconsidered given new evidence for microbial N2 fixation as a common feature of oceanic systems (10, 11). Despite tremendous diversity, a single bacterial clade can dominate numerically. The -proteobacterial clade SAR11 constitutes 30% of Sargasso Sea surface cells (12), and a newly discovered cluster within the Roseobacter clade, found throughout temperate and polar regions, makes up 20% of Southern Ocean bacteria (13). Finally, massive marine prokaryotic metagenome sequencing is providing a tremendous database for discovering metabolic capabilities and new ways to conceptualize and study prokaryotic biodiversity (14). Environmental genomics is also revealing bacterial interactions with marine animals, and their influences on animal populations and ecosystem function. Corals offer an excellent example. A recent study discovered 430 novel bacterial ribotypes associated with three coral species (15), and shifts in bacterial species composition appear to underlie coral health and disease (16, 17). Appreciating this diversity in conservation efforts is important because functional redundancy cannot be assumed. Marine bacteria are but one microbial realm in the ocean for which discoveries with ecosystem consequences abound. Viruses were not studied until 1989 yet are the most abundant biological entities in the sea (107 ml1). Bacteriophages induce bacterial mortality, creating a futile carbon cycle in which
Originally published 12 March 2004

dissolved organic matter assimilated by bacteria is released via bacterial lysis and metabolized by other bacteria, enhancing upper-ocean respiration (18). Species specificity and host density-dependence lead phage to kill the winner, maintaining bacterial diversity, with implications for organic matter decomposition (19, 20). Phage diversity studies were initially restricted by the requirement for cultivated hosts. Now, cultivation-independent genomics reveal enormous diversity, including a picorna-like superfamily implicated in phytoplankton mortality, even of toxic bloomforming algae (21, 22). Furthermore, specific cyanophage distributions vary from coastal to open ocean, likely influencing Prochlorococcus or Synechococcus host distributions differentially (23). In view of microbial abundance, diversity, dynamics, and influence on ocean chemistry, ecosystem-based conservation models must explicitly include microbes, to develop a functional view that integrates microbes, macrobes, and abiotic ecosystem components. This requires going beyond biodiversity assessment, because functional diversity depends on the environmental context of microbial expression. The task concerns an age-old theme in ecology: scale, both spatial and temporal, and the integration of scales (see the figure). Microbes are no different from larger organisms in this senseone must study them at habitat scales relevant to their adaptive strategies to determine how their metabolism influences larger-scale ecosystem dynamics. For microbes this spatial scale is miniscule, from micrometers to millimeters. Because microbes influence ecosystems through molecular interactions, for example, involving cell surface receptors, permeases, or enzymes, the scale reduces to macromolecular, or nanometers. Thus, microbes ecosystem activities, whether they involve carbon cycling or pathogenesis toward marine animals, should be modeled as molecular events. Studies of biochemical interactions and fluxes have yielded valuable insights. The oceanic silicon cycle is being revised following the discovery that colonizing bacteria cause postmortem dissolution of silica from diatom cell walls (frustules). These bacteria secrete proteases that denude the silica shell of its protective protein layer and enhance silica dissolution rates (24, 25). Silicon cycling, in turn, regulates diatom productivity, which is important, for example, in carbon cycling and fisheries. Genomics opens another dimension for integrating cell biochemistry and ecosystem dynamics. Genomic exploration of two Prochlorococcus ecotypes has not only identified features with obvious roles in the relative fitness of the ecotypes in response to key environmental variables but has also

16

sociated with three coral species (15), miniscule, from micrometers to millime- (26, 27). S. pomeroyi degrades dimethylnd shifts in bacterial species composi- ters. Because microbes influence ecosys- sulfoniopropionate (DMSP) via several on appear to underlie oral health and disease 6, 17). Appreciating PERSPECTIVES is diversity in consertion efforts is impordeed essential, if we are to develop mecha- goal of elucidating all significant molec7. O. Be nt because functional 8. J. R. d ular interactions underlying the structure nistic models of ocean-basin biogeochemidundancy cannot be 1283 and functioning of a cell or an organism. cal dynamics. Microbial oceanography has 9. Z. S. sumed. made remarkable strides toward system in- The approach can be extended to marine 10. D. G Marine bacteria are Carp ventory, exposing tremendous diversity of ecosystem analysis (ecosystems biolout one microbial realm in 11. J. P. Z gy), in essence treating the ecosystem microbial capabilities, and spatial and tem12. R. M e ocean for which disporal dynamics. Although they reveal the as dynamic molecular architecture and 13. N. S veries with ecosystem (200 incredible capacity of microbes to interact appreciating real-time expression as the nsequences abound. 14. J. C. mechanistic basis of ecosystem dynamwith ocean systems, these studies do not iruses were not stud2004 address system structure or organizational ics. Although the goal of treating organ- 15. F. Ro d until 1989 yet are Ecol. processes. Biosystems exhibit variability at isms and the environment as a molecular e most abundant bio16. K. Pa all organizational scales, from gene expres- continuum is huge, the rate of progress gical entities in the 8725 sion to groups of diverse individuals, but in genomics and proteomics and its inte- 17. J. Fri a (107 ml 1). BacterioAppl all display molecular interconnectedness. gration into oceanographya field rich hages induce bacterial 18. J. A. F Analysis of marine ecosystems as bio- in computational talentpromises suc- 19. T. F. T ortality, creating a fuchemical matrices is complex, but not un- cess. Ecosystems biology offers a frame(199 e carbon cycle in like, for example, that of neurobiology and work for integrating genomic, biochemi- 20. J. A. hich dissolved organic Microbes in marine ecosystems. Highly abundant and diverse bacteria and viruses (green dots) interact with (200 atter assimilated by systems biology. Modeling efforts must in- cal, and environmental data. This frame- 21. M. B biotic and abiotic ecosystem components at the molecular, or nanometer, scale yet critically influence the funcacteria is released via tioning of complex, large-scale marine work will alsoforce unify efforts for biodivercorporate interaction mechanisms andreefs vari-( right 1425 ecosystems [e.g. endangered coral )]. An atomic micro22. A. I. acterial lysis and me- scope image of marine polysaccharides sity conservation and conservation of ability at all ecosystem scales, including ( top ), a biochemical pathways chart ( background ) and genome map of (200 bolized by other bac- Prochlorococcus provide examples desirable biogeochemical the nanometer-to-millimeter of marine mi- ecosystems of the context for analysis of realm complex as dynamic mo- states of the 23. M. B ria, enhancing upper- lecular architectures. crobes and molecules. Such an analysis can ocean. 424, 24. K. Bi cean respiration (18). now begin, but it will require a conver25. K. Bi pecies specificity and References and Notes gence of genomics and the nascent field of (200 1. Pew Foundation Ocean Commissions report, www. pathways, but only one leads to DMS; tems through molecular interactions, for exost densitydependence lead phage to expression to groups of diverse individuals, inferred biochemical bases for ecotype0312Compassff 3/5/04 2:26 PM Page 1624 microscale biogeochemistry, addressing 26. J. M. pewoceans.org/oceans/oceans_report.asp (2003). thus, the relative expression of particular cell surface receptors, perkill the winner, bacterial 1261 all display molecular interconnectedness. specific maintaining environmental impacts (5ample, ). The involving Si- but spatially explicit biochemical interactions 2. J. E. Hobbie et al., Appl. Environ. Microbiol. 33, 1225 27. M. A enzymes regulates climeases, or enzymes, the scale reduces to as DMSP-degrading versity, with implications for organic Analysis of ecosystem marine ecosystems biochemical licibacter pomeroyi genome is yielding ecoand their consequences. This (1977). J. 20 mate. Overarching microbial activities or nanometers. miatter decomposition (19 , 20). Phage di- macromolecular, 3. F.the Azam, Science 280 , 694 (1998). iswill complex, but not unlike, for examphysiological insights into bacteria producing matrices approach lead Thus, to new testable hypothe28. We t A. Dufresne et al. , Proc. Natl. Acad. Sci. U.S.A. 100, is the physical 4.organization of organic ecosystem activities, whether theysystems rsity studies were initially restricted by crobes for c ple, that of neurobiology and biology. dimethyl sulfide (DMS), a volatile compound ses and prognostic models. (2003). Baln matter in seawater,10020 allowing the visualizainvolve cycling or pathogenesis to- of e requirement cultivated hosts. Now, (26 efforts must incorporate interaction that for influences cloud formation , 27). carbon S. Modeling We propose consideration concepts 5. G. Rocap et al., Nature 424, 1042 (2003). Work P E R S P E Cgenomics T I V E S reveal tion of ecosystems as molecular architecward marine animals, should be modeled as ultivation-independent and variability at all ecosystem pomeroyi degrades dimethylsulfoniopropio- mechanisms 6. B. Palenik et al., Nature 424, 1037 (2003). that are driving systems biology with its A146 tures. Seawater is structured with crossmolecular events. normous diversity, including a picornagoal ofincluding elucidating all significant molecdeed(DMSP) essential, if several we are to develop but mecha7. O. Beja et al., Science 289, 1902 (2000). the nanometer-to-millimeter nate via pathways, only scales, linked polymers, and nanoand Acad. Sci. U.S.A. 100, Studies of ular biochemical interactions and the ke superfamily implicated in phyto-biogeochemiet al. , Proc. Natl. 8. colloids, J. R. de la Torre interactions structure nistic models of ocean-basin IMM U Nmicrobes O L O Gunderlying Yand molecules. of Such an one leads to DMS; thus, the relative expres- realm 12830 (2003). matter conmicrogels, creating an organic fluxes has have yielded valuable insights. The ankton mortality, even of toxic bloom and functioning of a cell or an organism. cal dynamics. Microbial oceanography 1662 o analysis can now revised begin, but require a a wealth sion of particular DMSP-degrading enzymes 9. Z. S. Kolber et al., Science 292, 2492 (2001). tinuum and of surfaces displayoceanic silicon cycle is being fol-it will rming algae (21, 22). Furthermore, spe- system The approach can be extended to marine made remarkable strides toward in10. D. G. Capone, J. P. Zehr, H. W. Paerl, B. Bergman, E. J. (6) rev convergence of genomics and the nascent fieldand regulates climate. Overarching thelowing microbial ing activity biodiversity hot-spots ( 3 ). the discovery that colonizing bactefic cyanophage distributions vary from Carpenter, Science 276, 1221 (1997). ecosystem analysis (ecosystems bioloventory, exposing tremendous diversity of of ing the microscale biogeochemistry, addressing activities is the physical organization of oret spatial al., Naturecontext . 412, 635 (2001). J. P. Zehr a provides cause postmortem dissolution of silica This structure 11. oastal to open ocean, likely influencing gy), in explicit essence treating the ecosystem microbial capabilities, and spatialria and tem- spatially layer ( et al. , Nature 420, 806 (2002). 12. R. M. Morris biochemical and interactions ganic matter in seawater, allowing the visualfor microbial and adaptations, diatom cell walls (frustules). These interactions rochlorococcus or Synechococcus host from as dynamic molecular architecture and poral dynamics. Although they reveal the Nature , 427 , 445 13. N. Seije, M. Simon, T. Brinkhoff, lympho their ecosystem consequences. This approach ization of ecosystems architecbacteria secrete proteases that denude the and permits ecosystem stributions differentially (23). as molecular (2004). analysis in terms appreciating real-time expression asprogthe incredible capacity of microbes to interact will an eleg lead to new testable hypotheses and tures. Seawater is structured with cross-linked and biochemical In view of microbial abundance, diversi- silica shell of its protective protein layer of molecular architecture et al. , Science, published online 4 March 14. J. C. Venter mechanistic basis of ecosystem dynamwith ocean systems, studies do not nostic sals in 2004 (10.1126/science.1093857). polymers, colloids, andthese nanomicrogels, enhance silica models. dissolution ratesJean-Pierre (24, fluxes. dynamics, and influence on ocean chem-and and Kraehenbuhl Max V. Corbett ics. Although the goal of treating organaddress system structure or organizational 15. and F. Rohwer, Seguritan, F. Azam, N. Knowlton, Mar. sals ar We propose that discoveries creating an organic matter continuum a cycling, These illustrate that 25). and Silicon in turn,consideration regulates di- of concepts ry, ecosystem-based conservation models Ecol. Prog. Ser. 243, 1 (2002). isms and the as ananoscale molecular processes. Biosystems variability at are mucos biochemical bases of ecosystem atom which is environment important, for ust explicitly include microbes, toexhibit develdriving systems biology with its goal of wealth of surfaces displaying activity andproductivity, bio16. K. Patterson et al., Proc. Natl. Acad. Sci. U.S.A. 99, continuum is huge, the rate progress all organizational scales, from gene expres-in carbon The informative, intestinal epithelium provides a (IgA) he gastrointestinal tractof of human 8725 (2002). function are and inexample, cycling fisheries. p a functional view that integrates elucidating alland significant molecular interac - tractable, diversity hot-spots (3). This mistructure provides and proteomics and its inte- 17. sion to groups of diverse individuals, but in genomics J. Frias-Lopez, A. the L. Zerkle, G. T. Bonheyo, B. W. Fouke, barrier against invasion of host tissues mune r adults contains more than differthe structure and 400 functioning a spatial context for microbial interactions and tions underlying Appl. Environ. Microbiol. 68, 2214 (2002). into oceanographya field rich all display molecular interconnectedness. gration by both pathogenic and nonpathogenic trigger ent species of bacteria. Indeed, the toof a cell or an organism. The approach can be adaptations, and permits ecosystem analysis in 18. J. A. Fuhrman, Nature 399, 541 (1999). 1623 bacteria antigen www.sciencemag.org VOL 303 talentpromises 12 MARCH 2004 in computational sucAnalysis of marine ecosystems as bio- SCIENCE bacteria ( 2 ). However, pathogenic tal weight of microflora in the human gut 19. T. F. Thingstad, R. Lignell, Aquat. Microbiol. Ecol. 13 , 19 terms of molecular architecture and biochemi- extended to marine ecosystem analysis (ecoEcosystems biology offers a framechemical matrices is complex, but not un- cess. are(1997). able to breach the gut epithelium and dendrit has been estimated be about 1 kg. Most biology), into essence treating the eco- 20. cal fluxes. J. A. Fuhrman, M. Schwalbach, Biol. Bull. 204, 192 work for integrating genomic, biochemilike, for example, that of neurobiology and systems the hosts innate immune system in a numCel of these bacteria are commensals that co(2003). system as dynamic molecular architecture and Thesebiology. discoveries illustrate that nanoscale environmental data. This systems Modeling efforts must in- cal, berM. ofBreitbart ways, et in al. some destroying such a existand peacefully with their host and frameremain 21. , Proc. cases Natl. Acad. Sci. U.S.A.gut 99, real-time expression as the mechbiochemical bases of ecosystem function are appreciating work will provided also unify efforts corporate interaction mechanisms and vari14250 (2002). cells (3). In contrast to bacterial provide harmless that they for do biodivernot stray epithelial anistic basis of ecosystem dynamics. Although tractable, informative, and indeed essential, A. I. Culley, A. S. Lang, C. A. Suttle, Nature 424, 1054 conservation andSome conservation of 22. ability at all ecosystem scales, including sity pathogens, commensals have been pre- tion ag beyond the gut lumen. commensals (2003). goal of treating organisms and the environif we are to develop mechanistic models of the desirable biogeochemical states of their the 23. the nanometer-to-millimeter realm of misumed reside in the gut lumen with lit- encoun may even confer health benefits upon Nature M. B. to Sullivan, J. Waterbury, S. W. Chisholm, as a molecular continuum is huge, the ocean-basin biogeochemical dynamics. Mi- ment ocean. crobes and molecules. Such an analysis can , 1047 (2003). or no direct interaction with the epithe- nate h host by helping to digest dietary carbohy- tle 424 ofand progress in genomicsthe and proteomics 24. crobial oceanography hasrequire made remarkable K. Bidle, F. Azam, Nature 387, 508 (1999). now begin, but it will a conver- rate lium. Recent studies, however, indicate recogn drate by maintaining appropriate 25. K. Bidle, M. Manganelli, F. Azam, Science 298, 1980 its integration into oceanographya field strides toward system inventory, exposing References and Notes gence of genomics and the nascent fieldtreof and gut commensals do interact with the cules c balance among the different types of gut that (2002). 1. Pew Ocean Commissions report, www. in Foundation computational talentpromises suc- 26. mendous diversity of microbial addressing capabilities, rich microscale biogeochemistry, al., Int. J. Syst. Evol. innate Microbiol. 53, M. Gonzalez etand gutJ.epithelium can trigger and lar pat bacteria. The correct microbial balance pewoceans.org/oceans/oceans_report.asp (2003). 1261 (2003). cess. Ecosystems biology offers a framework and spatial and temporal dynamics. Although spatially explicit biochemical interactions adaptive immune responses. In addition, cosa, m helps stimulate gutEnviron. immunity and 33 to, pre2. J. E. to Hobbie et al., Appl. Microbiol. 1225 M. A. Moran, J. M. Gonzalez, R. P. Kiene, Geomicrobiol. integrating genomic, biochemical, and en- 27. they incredible capacity of microbes and reveal their the ecosystem consequences. This for (1977). commensals can influence epithelial cell the int vent colonization by pathogenic bacteria J. 20, 375 (2003). 3. F. cause Azam, Science 280 , 694 data. This framework will disalso 28. to interactwill with ocean systems, these studies vironmental approach lead to new testable hypotheWe thank T. Hollibaugh, J. Fuhrman,by and C. Beardsley proliferation, for example, injecting are sp that diarrhea and(1998). other intestinal 4. A. Dufresne etfor al., Proc. Natl. Acad. Sci. U.S.A. 100, for critical reading of the manuscript; V. Svetlicic, E. efforts biodiversity conservation do system structure or organiza- unify ses not andaddress prognostic models. factors into the gut epithelium that block pathog orders ( 1 ). 10020 (2003). Balnois, and D. Kline for sharing unpublished images. conservation of desirable tional Biosystems exhibit vari- and We processes. propose consideration of concepts PAMPs -catenin degradation (4). These microbes 5. G. Rocap et al., Nature 424, 1042 biogeochemical (2003). Work supported by NSF (OCE132677) and NIH (RO1 of the ability at driving all organizational scales, from gene 6. B. Palenik etocean. al., Nature 424, 1037 (2003). that are systems biology with its states A146600). also enhance the nuclear export of NF- B, tors tha The authors are at the Swiss Institute for the transcription factor that controls the the im Experimental Cancer Research, CH1066 Epalinges, IMMUNOLOGY cules i production of proinflammatory chemo- 17 Switzerland. E-mail: jean-pierre.kraehenbuhl@isrec. 1662 Macpherson Uhr family unil.ch kines of by this gut issue, epithelial cells (5). and On page (6) reveal that the gut mucosacompris

0312Compassff

3/5/04

2:26 PM

Page 1624

PHOTO CREDITS: (CORAL REEF) DAVID I. KLINE; (EPIFLUORESCENCE MICROGRAPH) J. FUHRMAN; (ATOMIC FORCE MICROGRAPH) V. SVETLI I & E. BALNOIS, PUBLISHED IN V. SVETLI I ET AL.. CROAT CHEM. ACTA 79, 107 (2006); (BIOCHEMICAL PATHWAYS) ROCHE DIAGNOSTICS GmbH

Keeping the Gut Microflora at Bay

REPORTS

Major Bacterial Contribution Major Bacterial Contribution to to Marine Dissolved Organic Marine Dissolved Organic Nitrogen Nitrogen
D. McCarthy,* I. Benner Hedges, Ronald Benner Matthew D. Matthew McCarthy,* John I. Hedges, John Ronald
Next to N the pool largest pool of reduced in the ocean resides in the 2 gas, the largest of reduced nitrogennitrogen in the ocean resides in the enormous Next to N2 gas, enormous reservoir of nitrogen dissolved organic nitrogen (DON). The chemical reservoir of dissolved organic (DON). The chemical identity of most of identity this of and most of this material, and the mechanisms by fundamental which it is cycled, remain material, the mechanisms by which it is cycled, remain questions in fundamental questions in contemporary oceanography. Amino acid enantiocontemporary oceanography. Amino acid enantiomeric ratios in the high molecular weight meric ratios in the high and molecular weight of DON from surface and deep fraction of DON from surface deep water in fraction three ocean basins show substantial D enantiomers of water in three ocean basins show substantial enrichment in enrichment in D enantiomers of four amino acids. The magnitude and pattern of these /L enrichments indicate four amino acids. The magnitude and pattern ofderived these Dfrom D/L enrichments indicate that peptidoglycan remnants bacterial cell walls that peptidoglycan remnants derived from bacterial cell walls constitute a constitute a major source of DON throughout the sea. These observations suggest that major source of DON throughout the sea. These observations suggest that structural properties of specific bacterial biopolymers, and the mechanisms for their structural of specic bacterial and the mechanisms for accumulation, areproperties among the central controls on biopolymers, long-term cycling of dissolved organic their accumulation, are among the central controls on long-term cycling of nitrogen in the sea. dissolved organic nitrogen in the sea. Most of the surface ocean is characterized by low to undetectable mineral nutrient concentrations. Such oligotrophic regions are central to global geochemical cycles, accounting for almost 40% of global primary production (1). Recently, the accumulation and advected export of dissolved organic matter (DOM) from such environments has been recognized as a major pathway for C flux in the upper ocean
M. D. McCarthy and J. I. Hedges, University of Washington, School of Oceanography, Box 357940, Seattle, WA 98195, USA. R. Benner, University of Texas, Marine Science Institute, Port Aransas, TX 78373, USA. *To whom correspondence should be addressed.

(2), a process likely regulated in part by the low levels of biologically available N (3). Yet these same waters contain substantial concentrations of fixed N in dissolved organic compounds (4). The apparent lower biological accessibility of this large organic N reservoir thus represents both a major control on upper ocean carbon cycles, and a corresponding N pump fundamental to closing oceanic N budgets (3, 4). The chemical identity of DON is key to understanding the mechanisms by which it is formed and escapes remineralization. Although amino acids account for most N in organisms, only a small fraction of seawater

DON can be identified as amino acid by common hydrolytic methods (5). The vast majority of this dissolved nitrogenous material appears to be contained in amide functional groups (6), suggesting that most longlived DON is hydrolysis-resistant or composed of recalcitrant nonprotein amide-containing biochemicals. Amino acid enantiomeric ratios can provide a powerful tool for characterizing nitrogenous materials (7 ). Early work identified elevated levels of D-aspartic acid and D-alanine in the open sea, indicating that D-amino acids may serve as indicators for both DOM sources (8) and for abiotic transformation reactions (9). Limitations of earlier filtration methods, however, prevented living bacteria from being excluded as a principle source of these observations (8), and interpretation was further complicated by substantial variability in the observed magnitude and distribution of D/L ratios with methods then available (10). Recent application of large-scale tangentialflow ultrafiltration to ocean waters has allowed reproducible isolation of DOM from seawater with essentially no contamination by living organisms or particulate matter (11, 12), in quantities sufficient to determine a full range of amino acid enantiomeric ratios. We have used this method to examine amino acid D/L ratios in the high molecular weight fraction of oceanic DOM from the central Pacific Ocean, the Gulf of Mexico, and the North Sea (13). These samples include surface and abyssal depths in two oligotrophic open ocean basins, as well as a biologically dynamic coastal region. Recovery efficiencies of ultrafiltered DOM

Table 1. Bulk properties and D/Lamino acid ratios from a marine cyanobacterium and for UDOM from the Central Pacic, Gulf of Mexico, and North Sea (13). Bulk properties include ultraltered dissolved organic carbon (UDOC) and nitrogen (UDON) concentrations, and atomic carbon/nitrogen ratios (C/N)a. UDOC and UDON concentrations are derived from C/N data, the dry weight recovered, and the volume ltered. Damino acids for UDOM-hydrolyzable amino acids (14) Bulk properties and D/Lamino acid ratios DOC ( M) UDOC ( M) UDON ( M) (C/N)a Asp Glu Ser Thr Ala Tyr Met Val Phe Leu Lys

are expressed as simple enantiomeric ratios, not corrected for the racemization blanks, which are shown for comparison. Sample collection depth is indicated in meters. Asterisks indicate samples in which the D/L ratio could not be quantied reliably due to chromatographic mixtures (as indicated by GC-MS); ND (not detected), samples in which the value was less than 1.5 racemization blank. S. bacillaris, Synechococcus bacillaris (25). Gulf of Mexico 4000 m 45 8.08 0.45 18.4 0.33 0.19 0.18 0.15 0.49 ** ND 0.06 0.04 ** 0.09 10 95 21.1 1.19 17.8 0.28 0.16 0.21 ** 0.53 ND ND ND ** ND ND 2 97 9.73 0.63 15.4 0.18 0.13 0.18 ND 0.47 ND ND ND 0.11 0.14 0.08 400 m 58 1.26 0.09 14.2 0.22 0.14 0.28 0.05 0.61 0.10 ND 0.13 0.04 ** ND N. Sea 2m 76 10.1 0.71 14.2 0.17 0.12 0.15 ND 0.37 ND ND ND ** ND ND

Central Pacic Blank S. bacillaris 2 0.09 0.06 0.03 0.00 0.03 0.04 0.08 0.02 0.02 0.07 0.04 0.14 0.09 ND ND 0.38 ND ND ND 0.04 ND 0.07 82 22.2 1.33 16.8 0.39 0.23 0.19 ND 0.52 ND ND ND 0.04 0.12 ND 100 85 22.2 1.42 15.6 0.39 0.17 0.19 ND 0.54 0.10 ND 0.07 0.05 0.13 0.09 375 53 10.6 0.63 16.9 0.36 0.20 0.28 ND 0.47 ** ** ND 0.04 0.12 0.12

18

Originally published 10 July 281 1998 10 JULY 1998 www.sciencemag.org SCIENCE VOL

231

REPORTS
(UDOM) are typically 20% in deep waters and 30% in surface waters (Table 1) (12), although very large sample sizes decrease relative recoveries (13). Average C/N ratios for this sample set were 17.0 for surface and 16.5 for deep water UDOM, respectively (Table 1). Hydrolyzable amino acids in these, as well as similar samples, constitute 10 to 20% of total UDOM N (5, 6). Surface and deep UDOM samples show substantial and highly characteristic D/Lamino acid ratios for all the sampled ocean basins (Table 1 and Fig. 1). Alanine (Ala), aspartic acid (Asp), glutamic acid (Glu), and serine (Ser) consistently have D/L ratios far above blank levels (Table 1 and Fig. 2) (14). Ala shows the most pronounced enrichment, with D/L ratios near 0.5, whereas Asp, Glu, and Ser ratios generally fall between 0.2 and 0.4. No other Damino acids could be repeatably detected above blank levels (Fig. 1). The D/L ratios of these four were remarkably consistent between widely separate locations, sampling times, and ocean depths (Fig. 1) and appear to represent a fundamental signature of UDOM throughout the ocean. DAmino acids are gradually produced in ancient organic matrices by abiotic racemization and may also be formed by degradation reactions of specific protein Lamino acids (9). However, the amino acid D/L and molar signatures in UDOM are not consistent with either of these origins. Bada and Hoopes (9) suggested abiotic dehydration of Ser as the most likely explanation for the near racemic D/L-Ala ratios they observed in the deep Pacific. This mechanism should produce linked decreases in Ser and threonine (Thr) as well as a concomitant buildup of -amino butyric acid (Aba) (9). UDOM isolates, however, exhibit no increased D-Ala at depth and minimal variation in mole percent Ser or Thr (5). Similarly, no Aba was detected by gas chromatographicmass spectral (GC-MS) analysis of any sample. Although abiotic equilibration of enantiomeric forms is a well-known Damino acid source, stereochemical inversion rates at ocean temperatures should be far too slow (8, 15) to account for appreciable racemization over average DOM residence times of 4000 to 6000 years (16, 17). Even if Damino acid signatures represent a far older component of UDOM, the observed highly selective pattern of D enrichment remains inconsistent with abiotic generation. Although multiple factors affect geochemical racemization rates (18), relative rates among different amino acids generally follow a regular succession (19). Ala in particular is among the slower amino acids to racemize (20). At the observed Ala D/L ratio of 0.5, abiotically derived D/L ratios of both phenylalanine (Phe) and in particular Asp (21) should be approaching their equilibrium D/L value of 1.0 (Fig. 2). However, in UDOM values of Asp are among the lowest observed D/L ratios, and D-Phe was near blank levels (Fig. 2). Moreover, the absence of any depth trend in the D/L ratios of any amino acid (Fig. 1) argues strongly that the stereoisomeric compositions of UDOM amino acids are not generated by aging or

Fig. left). UDOM /L ratios ))Central Fig. 1 1.((left). UDOM D D/L ratiosfor for(A (A Central Pacic, Pacific, ((B B)) Gulf Gulf of of Mexico, Mexico, and and ( Sea at at various variousdepths depthsfor for the four principle amino acids. Open (C C) ) North North Sea the four principle amino acids. Open barbar for for Gulf of Mexico represents the surface sample (10 m) independently Gulf of Mexico represents the surface sample (10 m) independently processed processed with 0.2- m cartridge lters (13). Dashed lines reect average with 0.2-m filters (13 ). Dashed lines reflect average for all the D/L values in values this gure have values for allcartridge the other amino acids measured. other adjusted amino acids D/L values in by thissubtracting figure have molar been adjusted for been formeasured. racemization blanks quantities racemization blanks by subtracting molar quantities expected from hydrolytic expected from hydrolytic racemization from total molar quantities of D-acid racemization from estimates total molardo quantities of d-acid These estimates measured. These not account for measured. reverse racemization of D forms, thusracemization represent minimum Fig. 2 represent (above). original do not account forand reverse of originalratios. dforms, and thus Damino acid sources: Comparison of the magnitudes and disAlternative minimum ratios. tributions of Damino acid ratios encountered in UDOM with those from other potential sources. Hydrolysis: Average hydrolysis blank D/L ratios Fig. 2. (above). damino acid sources: Comparison ofDthe magnitudes /L ratios (solid (hatched bars) Alternative are superimposed on overall average UDOM D/L ratios from pure L bars). Average racemization blanks were and distributions of Damino acid ratios encountered in produced UDOM with those from mixtures during hydrolysis. Abiotic racemization: Relative abiotic D(hatched /L ratios other potential sources. Hydrolysis: Average hydrolysis blank D/L ratios (dark bars), based on on relative racemization rates natural waters (20), bars) are superimposed overall average UDOM D/L for ratios (solid bars). Average were estimated to correspond with a hypothetical average abiotic UDOM racemization blanks were D/L ratios produced from pure L mixtures during hydroD/L-Ala ratio of 0.5. lysis. Abiotic racemization: Relative abiotic D/L ratios (dark bars), based on relative racemization rates for natural waters (20), were estimated to correspond with a hypothetical average abiotic UDOM D/L-Ala ratio of 0.5.

232

10 JULY 1998 VOL 281 SCIENCE www.sciencemag.org

19

REPORTS
selective utilization of Lamino acids over time, but instead represent an intrinsic biochemical signature of seawater DON. The principal biochemical sources of Damino acids are peptidoglycans, the main structural component of bacterial cell walls. Peptidoglycans are heterogeneous polymers composed of an amino-sugar backbone crosslinked with peptide bridges. The bridging peptides are characterized by a number of unusual nonprotein amino acids and Damino acids. In particular, the D-enantiomeric forms of Ala, Glu, and Asp are prevalent (22). D-Ala is the most abundant Damino acid in most peptidoglycans and the only one that is universally incorporated (23). The enantiomeric patterns that characterize UDOM thus closely match those characteristic of bacterial peptidoglycans. Because free-living bacteria are abundant in ocean water, direct contamination of DON isolates must be carefully excluded (8). Our UDOM isolation protocol was designed to remove essentially all natural microorganisms, and its efficacy has been verified both by filter intercomparison and direct bacterial counts. On average, 99% of oceanic bacteria are removed by the 0.1- m pore size filters with which most samples were processed before UDOM isolation. In addition, the central Pacific samples were directly examined by epifluorescence microscopy after passage through the 0.1- m pore size prefilter, and bacterial numbers were below detection (12). Typical estimates for N per cell of oceanic bacteria (24) indicate that even a maximal limit of 10% of original microbes passing the 0.1- m pore size filter would contribute less than 1% to total UDOM amino acids. The previously mentioned lack of any depth trend in UDOM Damino acid abundance, despite a corresponding 10fold decrease in bacterial abundance (12), confirms that the observed Damino acids derive not from concentrated bacteria, but from the vastly larger pool of nonliving DOM. The observed high D/L ratios strongly suggest that peptidoglycans constitute a major component of UDOM N. Most characterized oceanic water column bacteria are Gram-negative, typified by relatively thin peptidoglycan layers (22). Although the specific bacterial sources of peptidoglycan in the open sea are unknown, the cell wall enriched fraction of a cosmopolitan oceanic cyanobacterium, Synechococcus bacillaris, gives Ala D/L ratios of 0.38, in the same general range as UDOM values (Table 1) (25). This supports prior evidence that cell wall material of marine bacteria is compositionally similar to that of terrestrial bacteria (26). The contrasting relative amounts of other Damino acids from Synechococcus suggests, as would be expected, that this bacterial species alone does not dominate UDOM sources. Because of the inherent variability among known peptidoglycan structures (27) and the poorly defined species compositions of marine bacterial communities, finite percentages of peptidoglycan in UDOM isolates cannot be quantified with certainty. However, the following rough estimate of relative N contributions can be made using broad structural information. Ala represents the most useful source tracer, because it is a main component of proteins and has the most consistent D/L ratios within studied peptidoglycans (22, 23). The D enantiomer makes up about 30% of total hydrolyzable Ala in UDOM [percent D D/(D L), thus 30% D is equivalent to a D/L ratio of 0.5], whereas pure peptidoglycans contain 40 to 50% DAla. On the basis of these Damino acid contents and average structural features of common peptidoglycans and proteinacous materials, a calculation of the relative total N contributions from these two biochemical types can be made. Such a calculation (28) indicates that peptidoglycan accounts for a similar amount of the total N (45 to 80%) as is derived from hydrolyzable protein. Thus, in terms of N, this one structural biopolymer may be at least as abundant in UDOM as conventional proteinaceous material. In fact, this could be a minimal estimate, because conventional hydrolytic methods are optimized at near 100% for proteins (29) and might be substantially less efficient for peptidoglycan residues in an environmental matrix. Methods that define protein on the basis of total hydrolyzable amino acid yields from seawater DOM would also actually include a component of peptidoglycan-derived amino acid. Such a large peptidoglycan component carries major implications regarding the chemical composition of the high molecular weight fraction of marine DOM. For example, UDOM also should be enriched in amino sugar, a major component of peptidoglycan. Such an enrichment is consistent with aldose-poor UDOM sugar composition (5), as well as the specific identification of an amino sugar component by thermal desorbtionmass spectrometry (30) and pyrolysis (31). In addition, comparison of average amino acid compositions of UDOM (5) with those of planktonic sources (32) indicates that UDOM is measurably enriched in Ala, Glu, and Ser, consistent with an additional nonproteinaceous source for these specific amino acids. Finally, evidence for an important peptidoglycan component is supported by previously mentioned observations (6) of a predominant amide component not quantifiable by conventional amino acid analysis. Although UDOM alone accounts for a substantial fraction of total oceanic DOM, the extent to which these results apply to all dissolved nitrogenous material will be determined by overall size-related compositional trends. The extent of such size-related differences within the oceanic DOM pool are not fully known. However, comparisons of many properties of UDOM and total DOM (for example, C/N ratios, stable isotopic ratios, radiocarbon content, amino acid yields, and molecular-level signatures) indicate a high degree of overall compositional similarity (5, 11, 12). Recent studies of filtered whole seawater from the surface Arctic Ocean has yielded almost identical Damino acid distributions and ratios as those in our UDOM isolates, yet from a distinctly separate ocean region (33). Earlier identification of high Damino acid concentrations in unfractionated seawaters (8, 9), though more variable, is also consistent with the presence of peptidoglycan structural remnants in smaller size classes of dissolved material. Given that N from algae, as well as most of that from bacteria, is in the form of proteins containing no Damino acids, the elevated D/L ratios in UDOM indicate that the high molecular weight nitrogenous material dissolved in seawater is enriched in bacterial cell wall material. Bacteria are widely recognized as major consumers of organic matter and in some oligotrophic ocean regions may also be the major primary producers (34). Thus, abundant potential sources for bacterial cell-wall material likely exist throughout the water column. The accumulation and environmental persistence of these materials may also be related in part to intrinsic structural properties. Structural polymers can display long-term geochemical stability (35). The interwoven polysaccharide matrix of peptidoglycan, coupled with the unusual peptide substituents and structural variability, creates a heteropolymer resistant to many common hydrolytic enzymes (36). In addition, laboratory experiments and the identification of specific bacterial membrane proteins in seawater (37, 38) suggest that more labile biochemicals may also be shielded by close association in a cell-wall matrix. The high D/L-amino acid ratios found in UDOM indicate that a substantial fraction of dissolved organic N in the sea is of bacterial origin. This result challenges the common paradigm that the enormous reservoir of oceanic dissolved material is predominantly derived from algal sources. Central oceanic ecosystems are characterized by intensive bacterial recycling of DOM, coupled with similarly dynamic bacterial removal by protozoans and viruses (39). These processes represent direct pathways for introduction of bacterial cell wall structures into dissolved and colloidal seawater pools (38, 40), where minor differences in bioreactivity may result in substantial accumulation. Bacterial predation, coupled with the intrinsic structural properties of bacterial cell wall material, may thus be among the major controls on the long-term cycling of organic N in the sea.
References and Notes

1. S. D. Killops and V. J. Killops, An Introduction to Organic Geochemistry, R. C. O. Gill, Ed. (Longman Geochemistry Series, Longman, New York, ed. 1, 1993). 2. C. A. Carlson, H. W. Ducklow, A. F. Michaels, Nature

20

www.sciencemag.org SCIENCE VOL 281 10 JULY 1998

233

3. 4. 5. 6. 7. 8. 9. 10. 3. 4. 5. 3. 4. 6. 11. 5. 7. 12. 6. 8. 13. 9. 7. 10. 8. 9. 10. 3. 4. 11. 5. 12. 6. 11. 13. 7. 12. 8. 13. 9. 10. 14.

11. 12. 13. 15. 14. 16. 17. 14. 18. 19. 20. 15. 21. 16. 22. 15. 17. 14. 16. 23. 18. 17. 19. 18.

234
20. 19.

York 10. In whole seawater hydrolysates from the central Pacic, D-Ala was assumed to derive from peptidoglycans, 14. UDOM amino acid hydrolysis was conducted at 36. H. J Lee and Bada (8) found D/L ratios of Ala and Asp to be Community respiration ( R ) rates are scaled as th and L -Ala was assumed to come from a mixture of 150C for 70 min, generally with the method of 37. E. T similar in both surface and deep waters, ranging from peptidoglycan and proteinaceous material. Relative D and Lamino Cowie and Hedges ( 29 ). Individual primary production ( P ) rates of aquatic ecosyst R E P O R T S 38. T. N 0.05 to 0.16. They also reported, however, substantial amino acid compositions in marine proteinaceous acids were quantied as pentauoropropyl isopropyl 81 ( aquatic biota as carbon dioxide sources or sink differences in Asp D/L ratio between the Atlantic and esters by gas with ame ionization sources areV. largely (32 ), and accordingly 30. J. J. Boon, Klap, invariant T. Eglinton, Org. Geochem ., in 24. J. Fuhrman, in chromatography Primary Productivity and Biogeochemi371, 405 (1994); A. F. Michaels, N. R. Bates, K. O. 39. a J. A. Pacic. In contrast, Bada and Hoopes (9), using similar aquatic ecosystems d detection ( 41 ), and peak identities were veried by the Unproductive total proteinaceous N can be approximated support as press. cal Cycles in the Sea (Plenum, New York, 1992), pp. Buesseler, C. A. Carlson, A. H. Knap, ibid. 372, 537 123 D/L-Ala ratios between 0.4 isolation methods, reported GC-MS. Analytical variability in D/L ratios of UDOM ration rate than that of productive aquatic ecosy 12 (Ala-N). Similarly, an estimate of total pep361383. (1994). 31. J. D. H. van Heemst, M. Baas, J. W. de Leeuw, R. and 0.7 in surface Pacic waters, increasing to near 40. J. A. samples was less than 15%. Racemization tidoglycan N beact derived from peptiBenner, Organic Geochemistry , common K. dioxide Oygard, Ed. P. J. L. B. Williams, Mar. Chem. 51, 17 (1995). 25. Insoluble structural and cell-wall wasblanks con(R in P ),can and as carbon sources. The racemic values of 0.8 to 1.0 in the material deep. (199 were determined bycells multiple hydrolyses of cyapure (Falch Hurtigtrykk, Oslo, Norway, 1993), pp. 694 doglycan architecture as roughly 5.7 ( D -Ala N), G. A. Jackson and P. M. Williams, Deep Sea Res. 32, centrated from lysed of the cosmopolitan 11. L R. Benner, J. D. Pakulski, as M.well D. McCarthy, J. standards. I. Hedges, M. ecosystems to become autotrophic (P 41. R ) H i amino acid mixtures, as protein 698. 223 (1985). which includes N contribution from both amino nobacterium S.Science bacillaris , as described (A. Biersmith P. G. Hatcher, 255 , 1561 (1992). istry greater for marshes than for the open sea. Alt Enantiomeric ratios in commercially available Dami32. G. L. Cowie and J. I. Hedges, Limnol. Oceanogr . 37 , sugar backbone and peptide interbridges (22). For M. D. McCarthy, J. I. Hedges, R. Benner, Mar. Chem. R. Benner, Mar. Chem ., in press). &H 12. and R. Benner, B. Biddanda, B. Black, M. D. no acidcontaining peptides could beMcCarthy, repeatablyMar. deocean expected net heterotrophic, this 703 (1992). 55, 281 (1996). UDOM D/L is ratios of 0.5 to tobe 0.6 (equivalent to 26. D. J. W. Moriarty, Oecologia 26, 317 (1977). Chem . 57 ,to 243 (1997). 42. Wec termined within 5%. 33. H. P. Fitznar, J. M. Lobbes, G. Kattner, personal M. D. McCarthy, J. I. Hedges, T. Pratum, R. Benner, 30% D ) andexcess peptidoglycan D/L ratios of 0.7 to 1.0 by the production over the remaining o ami 27. K. H. Schleifer and O. Kandler, Bacteriol. Rev . 36 , 407 13. Most samples were preltered with an Amicon DC10 15. J. L. Bada and E. H. Mann, Earth Sci. Rev. 16, 21 communication. Nature 390, 150 (1997). (equivalent to near 50% D), this calculation indiproc ultraltration system with a 0.1- m pore size poly(1972). (1980). 34. L. Campbell, H. A. Nolla, D. Vaulot, Oceanogr . K. Kvenvolden, Annu. Rev. Earth Planet. Sci. 3, 183 cates that the peptidoglycan N Limnol. is 45 to 80% as C. L sulfone hollow-ber before isolation of UDOM 28. contributions total UDOM N from make u Aquatic ecosystems cover 70% of Earths sur16. Relative P. M. Williams and E.lter, R.of M. Druffel, Nature 330 ,pro246 39 , 954 (1975). large as (1994). the proteinaceous N contribution. Bear [1 nm UDOM The rst setwere of Gulf of teinaceous material0.1 and m]. peptidoglycan esti(1987). sions a face ( 1 ) and contribute 45% of the global pri35. J. W. de Leuuw and C. Largeau, in Organic GeochemC. Lee and J. Bada, Limnol. Oceanogr. 22, 502 (1977). 29. G. L. Cowie and J. I. Hedges, Mar. Chem . 37 , 223 Mexico isolates (27N, 95W) was preltered with R al E., P O R Tof S D- and L-Ala using the proportions nec17. mated P. H. Santschi et Geochim. Cosmochim. Acta 59, istry , M. Engle and S. Yet, A. Macko, Eds. of (Plenum, New controv mary production (2). the role their biota (1992). 31 M Nucleopore 0.2- m cartridges as an Dalternate J. L. Bada and E. A. Hoopes, Nature 282, 822 (1979). essary to produce observed UDOM /L ratios. and All 625 (1995). York, 1993),V. pp. 2372. independent method. Central Pacic samples were In whole seawater hydrolysates from the central Pacic, 24. D J.-Ala Fuhrman, in Primary and Biogeochemi371, 405 (1994); A. F. Michaels, N. R. Bates, K. O. 30. J. J.global Boon, Klap, T. Eglinton, Org. aGeochem in remains subject., of pare th in the CO was assumed toProductivity derive from peptidoglycans, 2 budget 18. R. Mitterer and N. Kriausakul, Org. Geochem . 7 , 91 36. H. J. Rogers, Ann. N.Y. Acad. Sci. 235, 29 (1974). collected in April 1992 at 12S, 135W from 2 pp. to Lee and Bada 8) found D/LA. ratios of Alaibid. and Asp be cal Cycles in the New York, 1992), Buesseler, C. (A. Carlson, H. Knap, 372to , 537 press. and L-Ala was assumed toand come a mixture of R E P(Plenum, O R T P. S from debate (35). Many freshwater ecosystems act piration (1984); W. L. Sea Kimber E. Hare, Geochim. 37. Mar. Chem . 51 239 (1995). 4000 m.R. Gulf of Mexico samples were collected on similar 361383. (1994).in both surface and deep waters, ranging from 31. E. J. Tanoue, D. H. van Heemst, M. , Baas, J. W. de Leeuw, R. peptidoglycan and proteinaceous material. Relative Cosmochim. Acta . 56 , 739 (1992). sources ( 6 ); in contrast, oceanic ecosysas CO elucida two separate cruises. In August 1991, the 10-m sam38. T. Nagata and D. L. Kirchman, Adv. Microb. Ecol. 15 , 2 0.05 to 0.16. They also reported, however, substantial in V. Organic Geochemistry , K. Geochem Oygard, ., Ed. P. J. ,L. B. Williams, Mar. Chem. 51N. , 17 25. Insoluble structural and cell-wall material was con30. Benner, J. J. Boon, Klap, T. Eglinton, Org. in 24. J. Fuhrman, in Primary Productivity and Biogeochemi371 405 (1994); A. F. Michaels, R.(1995). Bates, K. O. acid compositions in marine proteinaceous 19. amino L. Bada, in Chemistry and Biochemistry of the Amino ple (1000 liters) was collected at 27N, 95W. In cyaJuly 81 (1997). D /L M. ratio the Atlantic differences in Asp sinkspp. (7, 8). tems are assumed to act Norway, as CO2 tems ac (Falch Hurtigtrykk, Oslo, 1993), 694 G. A. Jackson and P. Williams, Deep Sea Res 32, centrated from lysed cells of the cosmopolitan press. cal Cycles in the Sea (Plenum, New York, 1992), pp. Buesseler, C. A. Carlson, A.between H. Knap, ibid. 372 , . and 537 sources largely invariant (32 ), accordingly Acids , G.are C. Barrett, Ed. (Chapman &and Hall, New York, of 1995, two very large integrated samples (4000 39. J. A. Fuhrman and R. T. Noble, Limnol. Oceanogr . 40 , Pacic. In contrast, Bada and Hoopes ( 9 ), using similar 698. 223 (1985). nobacterium S. bacillaris,Nas described (A. Biersmith This has been challenged calcu(R P 361383. (1994). 31. J. assumption D. H. van Heemst, M. Baas, J. W. deby Leeuw, R. 1985), p. proteinaceous 684. the total can be 400 approximated as liters) were collected from 2 and m at regular 1236 (1995). D / L -Ala ratios between 0.4 isolation methods, reported 32. Benner, G. L. Cowie and J. I. Hedges, Limnol. . 37 , M.J.D. J. Mar. I. Hedges, Benner, Mar. Chem. and R.(Ala-N). Benner, Mar.of Chem ., in press). in Organic Geochemistry , ocean K.Oceanogr Oygard, Ed. P. L. McCarthy, B. Williams, ChemR. . 51 , 17 (1995). 25. Insoluble structural and cell-wall material was con, in Kinetics the Non-Biological Decompo20. 12 Similarly, an estimate of total peplations suggesting that the coastal may be past fiv intervals on a transect between Corpus Christi, Texas, and 0.7 in surface Pacic waters, increasing to near 40. J. A. (1992). Fuhrman and C. A. Suttle, Oceanography , 51 703 55 281 (1996). P. M. Williams, Deep Sea Res. 32, 26. tidoglycan D. J. Key W. Moriarty, Oecologia 26 , 317 (1977). (Falch Hurtigtrykk, Norway, 1993), pp. 6 694 G. ,A. Jackson centrated from lysed cells the cosmopolitan cyasition and Racemization of of Aminos Acids in Natural N can be derived from common peptiand West, Florida. The North Sea surface sample net (1993). heterotrophic ( 9Oslo, ) Lobbes, and byG.the finding that evoluti racemic values and of 0.8 to 1.0 in the deep. R. Benner, 33. H. P. Fitznar, J. M. Kattner, personal M. D. McCarthy, J. I. Hedges, T. Pratum, 698. 223 (1985). 27. doglycan K. H.also Schleifer and O. Kandler, Bacteriol. Rev .-Ala 36 , 407 Waters, J.integrated D. S. Hem, Ed. (American Chemical Society, nobacterium bacillaris as described Biersmith architecture as roughly 5.7 (A. (D N), was on a, transect across the central R. Benner, J. , D. Pakulski, M. D. McCarthy, J. I. Hedges, 41. M. H. Engel and P. E. Hare, in Chemistry and Biochemcommunication. bacterial metabolism exceeds phytoplankton fluxes Nature 390 150 (1997). 32. G. L. Cowie and J. I. Hedges, Limnol. Oceanogr. 37, Washington, DC, 1971), pp. 308 331. (1972). M. D.Hatcher, McCarthy, J. I. Hedges, R. Benner, and R. Sea Benner, Mar. Chem ., in press). which includes N contribution from amino in April 1995. Such very large both sample sizes North P. Science 255, 1561 (1992). Mar. Chem. of the Amino Acids, D. G. Vaulot, C. Barrett, Ed. (Chapman 34. istry L. Campbell, H. A. Nolla, Limnol. Oceanogr . K. G. 703 (1992). production in unproductive waters ( 10 ), which Com 55 ,Kvenvolden, 281 (1996).Annu. Rev. Earth Planet. Sci. 3, 183 21. G. A. Goodfriend, Nature 357 ,interbridges 399 (1992). 28. Relative contributions of total UDOM N from pro26. D. J. W. Moriarty, Oecologia 26 , 317 (1977). are compositionally representative, but result in subsugar backbone and peptide ( 22 ). For & Hall, New York, 1985), pp. 461 479. R. Benner, B. Biddanda, B. Black, M. D. McCarthy, Mar. 954 (1994). (1975). 33. 39 H. , P. Fitznar, J. M. Lobbes, G. Kattner, personal M. D.. McCarthy, J. I. Hedges, T. Pratum, R. Benner, 22. H. J. Schleifer Rogers, in Aspects of Microbiology [Van teinaceous and peptidoglycan were estistantial decreases in relative recovery (Table 1) due orders 27. UDOM K. H. and O. Kandler, Bacteriol. Rev . 36 ,Nos407 D/L material ratios of 0.5 to 0.6 (equivalent to Carlos M. Duarte* and Su Chem 57 , 243 (1997). 42. We thank G. Cowie for inspiration and help with 35. J. W. de Leuuw and C. Largeau, in Organic GeochemC. Lee and J., Bada, Limnol. Oceanogr. 22, 502 (1977). communication. Nature 390 150 (1997). trand Reinhold, Wokingham, UK, 1983), vol. 7,necpp. to high concentration factorsD (/ 11 ). mated using proportions of D - and -Ala (1972). 30% D ) and the peptidoglycan L ratios of L 0.7 to 1.0 amino acid analysis, for help withConsejo sample (Table Most samples preltered with an Amicon DC10 istry ,de M. Engle and S.B. A.Black Macko, (Plenum, New Centro Estudios Avanzados deEds. Blanes, J. L. Bada and were E.Annu. A. Hoopes, Nature 282, 822 (1979). 34. L. Campbell, H. A. Nolla, D. Vaulot, Limnol. Oceanogr . K. Kvenvolden, Rev. Earth Planet. Sci . 3 , 183 6 25. essary to produce observed UDOM D / L ratios. All 14. UDOM amino acid hydrolysis was conducted at to near 50% D), this calculation 28. (equivalent Relative contributions of total UDOM N fromindiproprocessing, H.pp. P. Fitznar for sharing work in progress, ultraltration system with a 0.1- m pore size polyYork, 1993), 2372. Superior de Investigaciones Cient cas, Cam de Santa tive th 39, 954 (1994). In whole seawater hydrolysates from the central Pacic, Community respiration (Rinsight, ) rates are scaled as th (1975). 150C for assumed 70 min, generally with the method of D -AlaGhuysen was derive from 23. cates J.-M. and G. to D. Shockman, inpeptidoglycans, Bacterial Memthat the peptidoglycan N is 45 to 80% as C. Lee for valuable discussions and D. teinaceous material and peptidoglycan were estisulfone hollow-ber lter, before isolation of UDOM 36. H. J. s/n, Rogers, Ann. N.Y. Acad. Sci .and 235 , 29 Geochem(1974). Ba rbara 17300 Blanes, Girona, Spain. Lee and Bada (8) found D/L Oceanogr ratios of Ala and Asp to be 35. J. W. de Leuuw and C. Largeau, in Organic open s C. Lee and J. Bada, Limnol. . 22 , 502 (1977). D and L amino Cowie and Hedges ( 29 ). Individual and L-Ala was assumed from a mixture of branes and Walls ,E L.PLeive, Ed. (Dekker, New primary production ( P ) rates of aquatic ecosyst large asusing the proteinaceous N R Oto Rcome T contribution. S Bear for guidance and support. [1 nm UDOM 0.1 m]. The rst set of Gulf of mated the proportions of D- and L-Ala York, nec37. istry E. Tanoue, Mar.and Chem 51 , 239 (1995). similar in both surface and deep waters, ranging from , M.correspondence Engle S. .A. Macko, Eds.addressed. (Plenum, New acids were quantied pentauoropropyl J. L. Bada and E. A. Hoopes, Nature 822 (1979). *To whom be peptidoglycan and material. Relative 1973), vol. 1.and duction 29. G. L. Cowie J. proteinaceous I. as Hedges, Mar. Chem . isopropyl 37 , 223 Mexico isolates (27N, 95W) was 282, preltered with essary to produce observed UDOM D/L ratios. All aquatic biota as should carbon dioxide sources or sink 38. York, T. Nagata and D.2372. L. Kirchman, Adv. Microb. Ecol. 15, 0.05 0.16. They A. also however, substantial 1993), pp. esters by gas chromatography with ame ionization In whole seawater from the central Pacic, 30. 31 J. J. Boon, V. Klap, T. Eglinton, Org. Geochem., in 371 , to 405 (1994); F. reported, Michaels, R. Bates, K. O. 24. (1992). J. -Ala Fuhrman, in Primary Productivity and BiogeochemiMarch 1998; accepted 11 June 1998 amino acid compositions in marine proteinaceous Nucleopore 0.2- hydrolysates m cartridges as N. an alternate and D was assumed to derive from peptidoglycans, 81 (1997). Unproductive aquatic ecosystems support ad D / L ratio between the Atlantic and differences in Asp 36. H. J. Rogers, Ann. N.Y. Acad. Sci . 235 , 29 (1974). detection ( 41 ), and peak identities were veried by Lee and Bada ( 8) Carlson, found Central D/L ratios of Alaibid. and 372 Asp ,were to be press. Buesseler, C. method. A. A. H. Pacic Knap, 537 cal Cycles in largely the Seainvariant (Plenum, York, 1992), pp. independent samples sources are (New 32), and and L-Ala was assumed to come from aaccordingly mixture of 39. J. A. Fuhrman and R. T. Noble, Limnol. Oceanogr.aquatic 40, Pacic. In both contrast, Bada and Hoopes (9), ranging using similar 37. E. Tanoue, Mar. Chem . M. 51 , Baas, 239 (1995). D/Lapproximated ratios of UDOM GC-MS. Analytical variability in be ration rate than that of productive ecosy similar in surface and from 234 10 JULY 1998 VOL 281 SCIENCE www.sciencemag.org (1994). 361383. 31. J. D. H. van Heemst, J. W. de Leeuw, R. collected in April 1992 at deep 12S,waters, 135W from 2 to the total proteinaceous N can as peptidoglycan and proteinaceous material. Relative (1995). D. L. Kirchman, Adv. Microb. Ecol. 15, D/L-Ala ratios between 0.4 isolation methods, reported was less than 15%. Racemization blanks 38. 1236 T. Nagata 0.05 to 0.16. They also however, substantial Benner, Geochemistry , K. dioxide Oygard, sources. Ed. P. J. L. B. Williams, Mar.reported, Chem . 51 , 17 (1995). 4000 m. Gulf of Mexico samples were collected on 25. samples Insoluble structural and cell-wall material was con(Fuhrman R inand POrganic ), and and act as carbon The 12 (Ala-N). Similarly, an of total pepamino acid compositions in estimate marine proteinaceous and 0.7 in in surface Pacic waters, increasing to near 40. 81 J. A. C. A. Suttle, Oceanography 6, 51 were determined by cells multiple hydrolyses of pure (1997). D /LIn ratio between the Atlantic and differences Asp (Falch Hurtigtrykk, Oslo, Norway, 1993), pp. 694 (P two separate cruises. August 1991, theSea 10-m samG. A. Jackson and P. M. Williams, Deep Res . 32 , centrated from lysed of the cosmopolitan cyatidoglycan N can be derived from common peptisources are largely invariant (32 ), and accordingly ecosystems to become autotrophic R ) i racemic values of 0.8 to 1.0 in the deep. (1993). L amino acid mixtures, as well as protein standards. 39. 698. J. A. Fuhrman and R. T. Noble, Limnol. Oceanogr. 40, Pacic. In liters) contrast, Bada and Hoopes (9), using In similar ple (1000 was collected at 27N, 95W. July 223 (1985). nobacterium S. bacillaris ,N as described (A.(D Biersmith doglycan architecture as roughly -Ala N), the total proteinaceous can be 5.7 approximated as R. Benner, J. D. Pakulski, M. D. McCarthy, J. I. (4000 Hedges, 41. M. H. Engel and P.J.E. Hare, in Chemistry and the Biochemgreater for marshes than for open sea. Alt Enantiomeric ratios in commercially available Dami1236 (1995). D /LR. -Ala ratios between 0.4 isolation methods, reported of 1995, two very large integrated samples 32. G. L. Cowie and I. Hedges, Limnol. Oceanogr . 37 , M. D. McCarthy, J. I. Hedges, Benner, Mar. Chem . and R. Benner, Mar. Chem ., in press). which includesSimilarly, N peptides contribution from both amino 12 (Ala-N). an could estimate of total pepP. G.0.7 Hatcher, Science 255waters, ,2 1561 of the Amino Acids G. to C. Barrett, (Chapman no acidcontaining be repeatably deand in surface Pacic increasing near 40. istry J. A.ocean Fuhrman C. A., Suttle, Oceanography 6, 51 liters) were collected from and (1992). 400 m at to regular is and expected be netEd. heterotrophic, this c 703 (1992). 55 , 281 (1996). 26. tidoglycan D. J. W. Moriarty, Oecologia 26 , 317 (1977). sugar backbone and peptide interbridges (22 ). For can be derived from common pepti& Hall, New York, 1985), pp. 461 479. R. Benner, B.aBiddanda, B. Black, M.deep. D. McCarthy, Mar. termined toN within 5%. racemic values of 0.8 to 1.0 in the (1993). intervals on transect between Corpus Christi, Texas, 33. H. P. Fitznar, J. M. Lobbes, G. Kattner, personal M. D. McCarthy, J. I. Hedges, T. Pratum, R. Benner, by the excess production over the remaining o UDOM D / L ratios of 0.5 to 0.6 (equivalent to 27. K. H. Schleifer and O. Kandler, Bacteriol. Rev . 36 , 407 doglycan architecture as roughly 5.7 ( D -Ala N), Chem . 57 , 243 (1997). 42. We thank G. Cowie for inspiration and help with 15. J. L. Bada and E. H. Mann, Earth Sci. Rev. 16, 21 R. Benner, J., D. Pakulski, M.North D. McCarthy, J. I. Hedges, 41. communication. M. H. Engel and P. E. Hare, in Chemistry and Biochemand Key390 West, Florida. The Sea surface sample Nature 150 (1997). 30% includes D) and peptidoglycan D/L ratios 0.7 amino to 1.0 (1972). amino acid analysis, B. , Black for helpEd. with sample which N contribution from of both Most samples were preltered with an Amicon DC10 (1980). P. G. Hatcher, Science 255 , 1561 (1992). istry of the Amino Acids G. C. Barrett, (Chapman was also integrated on a transect across the central 34. L. Campbell, H. A. Nolla, D. Vaulot, Limnol. Oceanogr . K. Kvenvolden, Annu. with Rev. a Earth Planet. Sci . 3,poly183 to and near D ), interbridges this calculation 28. (equivalent Relative contributions UDOM N from proprocessing, H.York, P. Fitznar for sharing work in progress, ultraltration system 0.1m McCarthy, pore size sugar backbone and peptide (22). For Aquatic ecosystems cover 70% Earths surmake u 16. P. M. Williams E. 50% R.of M.total Druffel, Nature 330 ,indi246 & New 1985), pp. 461of 479. Sea in 1995. Such very sample sizes North R. Benner, B. April Biddanda, B. Black, M. large D. Mar. 39Hall, , 954 (1994). (1975). cates that peptidoglycan is (equivalent 45 to 80% as C. Lee for G. valuable discussions and insight, and D. teinaceous material and peptidoglycan were estisulfone hollow-ber lter, before isolation of in UDOM UDOM D/L the ratios of 0.5 to N 0.6 to (1987). are compositionally representative, but result subChem . 57 , 243 (1997). 42. We thank Cowie for inspiration and help with face ( 1 ) and contribute 45% of the global prisions a 35. Bear J. W. for de Leuuw and C. Largeau, in Organic GeochemC. Lee andUDOM J. Bada, Limnol. Oceanogr . 22 , 502 (1977). large as the proteinaceous N contribution. guidance and support. [1 nm 0.1 m]. The rst set of Gulf of using proportions D- and L0.7 -Ala necD ) andthe peptidoglycan Dof / L ratios of to 1.0 17. mated P. 30% H. Santschi et al ., Geochim. Cosmochim. Acta 59 , and stantial decreases inpreltered relative recovery 1) due amino acid analysis, B. for help with sample Most samples were with 282, an (Table Amicon DC10 Carlos M. Duarte* Susana Agust istry , M. Engle and S. A. Black Macko, Eds. (Plenum, New mary production ( 2 ). Yet, the role of their biota controv J. L. Bada and E. A. Hoopes, Nature 822 (1979). 29. G. L. Cowie and J. I. Hedges, Mar. Chem . 37 , 223 Mexico isolates (27N, 95W) was preltered with essary to produce D/L ratios. All (equivalent to nearobserved 50% D), UDOM this calculation indi625 (1995). to high concentration factors (11). m pore size polyprocessing, H. P. 2372. Fitznar for sharing work in progress, ultraltration system with a 0.1York, 1993), pp. In whole seawater hydrolysates from the alternate central Pacic, (1992). budget remains a subject of 31 March 1998; accepted 11 June 1998 in the global CO pare th Nucleopore 0.2m lter, cartridges as an and was to derive from peptidoglycans, 2 cates that assumed the peptidoglycan N isGeochem 45 to 80% as C. Lee for valuable discussions and insight, and D. 18. D R.-Ala Mitterer and N. Kriausakul, Org. . 7, 91 UDOM amino acid hydrolysis was conducted at sulfone hollow-ber before isolation of UDOM 36. H. J. Rogers, Ann. N.Y. Acad. Sci. 235, 29 (1974). Lee and Bada ( 8) found Central D/L ratios of Ala samples and Asp to be independent Pacic were Community respiration (RHare, ) rates are of scaled as the two-thirds power of the gross and L-Ala was assumed to come from a mixture large as R. the proteinaceous N P. contribution. debate ( 35 ). Many freshwater ecosystems act piration Bear for guidance and support. (1984); W. L. Kimber and E. Geochim. 150C 70method. min, 0.1 generally with the method of [1 nm forUDOM m]. The rst set of Gulf of 37. E. Tanoue, Mar. Chem. 51, 239 (1995). similar in both surface and deep waters, ranging from collected inHedges April 1992 at 12S, 135W from 2 to and proteinaceous material. Cosmochim. . J. 56 739 (1992). 29. peptidoglycan G. L. Cowie Acta and I., Hedges, Mar. Chem . Relative 37 , 223 ecosystems, Dpreltered and Lamino Cowie (29).95W) Individual Mexicoand isolates (27N, was with ) rates of aquatic indicating that the role ofecosysprimary production (P 6); contrast, oceanic as elucida 38. CO T. Nagata and ( D. L. in Kirchman, Adv. Microb. Ecol. 15, 2 sources 0.05 to 0.16. They also reported, however, substantial 4000 m. Gulf of Mexico samples were collected on acid compositions in marine proteinaceous (1992). 31 March 1998; accepted 11 June 1998 acids were quantied as pentauoropropyl isopropyl Nucleopore 0.2- m cartridges as an alternate and 19. amino J. L. aquatic Bada, in Chemistry andcarbon Biochemistry of the Amino 81 are (1997). biota as dioxide sources or sinks depends its as productivity. ratio between the Atlantic and differences in cruises. Asp D/L In tems assumed on to act CO2 sinks (7, 8). tems ac two separate August 1991, the 10-m samesters by gas method. chromatography with ame ionization independent Central Pacic samples were sources largely Ed. invariant (32), accordingly Acids, G. are C. Barrett, (Chapman &and Hall, New York, 39. J. A. Fuhrman and R. T. Noble, Limnol. Oceanogr . 40, Pacic. In contrast, Bada and Hoopes (9),95W. using similar ple (1000 liters) was collected at 27N, In July Unproductive aquatic ecosystems support a disproportionately higher respiThis assumption has been challenged by calcu(R P detection 41April ), and peak at identities were veried collected ( in 1992 12S, 135W from 2 by to 1985), p. 684. the total proteinaceous N can be approximated as 1236 (1995). D/L-Ala ratios between 0.4 isolation methods, reported of 1995, two large integrated samples ratios of (4000 UDOM GC-MS. variability in D/L 4000 m.Analytical Gulf very of Mexico samples were collected on rate than that of productive aquatic ecosystems, tend to be heterotrophic , in Kinetics of the Non-Biological Decompo20. 12 ration (Ala-N). Similarly, an estimate of total peplations suggesting that the coastal ocean may be past fiv and 0.7 in surface Pacic waters, increasing to near 40. J. A. Fuhrman and C. A. Suttle, Oceanography 6 , 51 liters) were collected from 2 and 400 m at regular samples was cruises. less than 15%. Racemization blanks two separate In August 1991, the 10-m samsition Racemization of in Natural tidoglycan N be derived from Acids common pepti(Rand P ), can and act asAminos carbon dioxide sources. average P required forthe aquatic netThe heterotrophic (9) and by finding that evoluti racemic values of 0.8 to 1.0 in the deep. (1993). intervals on a transect between Corpus Texas, were determined bycollected multipleat hydrolyses ofIn pure ple (1000 liters) was 27N, Christi, 95W. July Waters, J. D. Hem, Ed. (American Chemical Society, doglycan architecture roughly 5.7 (D-Ala N),(P 41.RM. R. Benner, J. D. Pakulski, M. D. McCarthy, J.standards. I. (4000 Hedges, H. Engel and E. Hare, in Chemistry and Biochemand Key West, Florida. The North surface sample ecosystems to as become autotrophic ) is over anP. order of magnitude bacterial metabolism exceeds phytoplankton fluxes L amino acid mixtures, as well as Sea protein of 1995, two very large integrated samples Washington, DC, 1971), pp. 308 331. which includes N contribution from both amino P. G.also Hatcher, Science 255 , 1561 istry of the Amino Acids, G. of C. Barrett, Ed. (Chapman was integrated on a transect across central greater for marshes than for the open sea. Although the upper amiEnantiomeric ratios in commercially available liters) were collected from 2 and(1992). 400 mthe at D regular production in four-fths unproductive waters ( 10), which Com 21. sugar G. A. Goodfriend, Nature 357,interbridges 399 (1992). backbone and peptide ( 22 ). For & Hall, New York, 1985), pp. 461 479. Sea in April 1995. Such very large sample sizes North R. Benner, B.aBiddanda, B. Black, M. be D. McCarthy, Mar. no acidcontaining peptides could repeatably deintervals on transect between Corpus Christi, Texas, ocean is expected to be net heterotrophic, this carbon demand can be balanced 22. H. J. Rogers, in Aspects of Microbiology [Van Nosorders UDOM D / L ratios of 0.5 to 0.6 (equivalent to are compositionally representative, but result in subChem . 57 , 243 (1997). 42. We thank G. Cowie for inspiration and help with termined to within 5%.The North Sea surface sample and Key West, Florida. trand Wokingham, UK, 1983), vol. 7, 1.0 pp. 30% D ) and peptidoglycan D /L ratios ofDuarte* 0.7 to byReinhold, the excess production over the remaining one-fth of Avanzados the stantial decreases in relative recovery (Table 1) amino analysis, B. ocean. Black for withConsejo sample (Table Most samples preltered withacross an Amicon DC10 Carlos M. and Susana Agust Centro de acid Estudios de help Blanes, was integrated on a transect the central J. L. also Bada andwere E. H. Mann, Earth Sci. Rev . 16 , due 21 6 25. (equivalent to near 50% D ), this calculation indito high concentration factors ( 11 ). processing, H. P. Fitznar for sharing work in progress, ultraltration system with a 0.1pore size poly1995. Such very m large sample sizes North Sea in April Superior de Investigaciones Cient cas, Cam de Santa tive th (1980). 23. J.-M. Ghuysen and G. D. Shockman, in Bacterial Memcates ecosystems that the peptidoglycan N of is 45 to 80% as C. Lee for valuable discussions and insight, and D. UDOM aminoand acid hydrolysis was at sulfone hollow-ber lter, isolation of UDOM are compositionally representative, butconducted result in, submake up 30% of the Girona, ocean. These concluAquatic cover 70% Earths surBa as rbara s/n, 17300 Blanes, Spain. P. M. Williams E. R. M. before Druffel, Nature 330 246 open s Community respiration ( R ) rates are scaled the two-thirds power of the gross branes and Walls , L. Leive, Ed. (Dekker, New York, large as the proteinaceous N contribution. Bear for guidance and support. 150C 70 min, generally with the [1 nm for UDOM m]. recovery The rst setmethod of Gulf of stantial decreases in0.1 relative (Table 1) due (1987). Carlos M. Duarte* Agust sions are correspondence based on indirect calculations and face (1 ) and 45% of the global *To Susana whom should berole addressed. vol. contribute 1.and duction 29. 1973), G. L. Cowie J. I. Hedges, Mar. Chem . 37 , pri223 and D and Lamino Cowie and Hedges (29 ).95W) Individual Mexico isolates (27N, was preltered with primary production ( P ) rates of aquatic ecosystems, indicating that the of to high concentration factors ( 11 ). P. H. Santschi et al., Geochim. Cosmochim. Acta 59, assumptions (3 ). Here, mary production (2). Yet, the role dioxide of their biota 31 March 1998; accepted 11 June 1998 we com(1992). acids were quantied as pentauoropropyl isopropyl Nucleopore 0.2-acid m cartridges as an conducted alternate and UDOM amino hydrolysis was at aquatic biota as carbon sources controversial or sinks depends on its productivity. 625 (1995). esters by chromatography ame ionization independent method. Central with Pacic samples were Community respiration (R) are scaled as the two-thirds power of the gross remains arates subject of pare gross primary production (P) and resin the global CO2 budget 150C forgas 70 min, generally with Geochem the method of R. Mitterer and N. 1992 Kriausakul, Org. . 72 , 91 Unproductive aquatic ecosystems support a the disproportionately higherwww.sciencemag.org respi234 10 JULY 1998 VOL 281 SCIENCE detection (41 ), and identities were veried by collected in April 12S, 135W from to D and Lamino Cowie and Hedges (peak 29). at Individual primary production ( P ) rates of aquatic ecosystems, indicating that the role of debate ( 35 ). Many freshwater ecosystems act piration ( R ) rates of aquatic communities to (1984); R. W. L. Kimber and P. E. Hare, Geochim. L ratios of UDOM GC-MS. variability in D/were 4000 were m.Analytical Gulf of Mexico samples collected on ration rate than that of productive aquatic ecosystems, tend to be heterotrophic acids quantied as pentauoropropyl isopropyl Cosmochim. Acta . 56 ,In 739 (1992). biota as carbon dioxide sources elucidate or sinks depends on biota its productivity. sources ( 6); in contrast, oceanic ecosysas CO2aquatic whether the of aquatic ecosyssamples was less than 15%. Racemization blanks two separate cruises. August 1991, the 10-m sam( R P ), and act as carbon dioxide sources. The average P required for aquatic esters by gas chromatography with ame ionization J. L. (1000 Bada, in Chemistry and Biochemistry of the Amino were determined by multiple of ple liters) was collected at hydrolyses 27N, 95W. Inpure July Unproductive aquatic ecosystems a disproportionately higher sinks (7, 8support ). (P tems (R respiP) or sinks tems are assumed toto act as CO as net CO detection (41 ), and peak identities were by 2 autotrophic 2 sources ecosystems become R) acts is over an order of magnitude Acids , G. C. Barrett, Ed. (Chapman & Hall, veried New York, L amino acid mixtures, as well as protein standards. of 1995, two very variability large integrated samples (4000 D / L ratios of UDOM GC-MS. Analytical in ration rate than that of productive aquatic ecosystems, tend to be heterotrophic This assumption has been challenged by calcu( R P ). We compiled data obtained over the 1985), p. 684. greater for marshes than for the open sea. Although four-fths of the upper amiEnantiomeric ratios in commercially available liters) were collected from 2 and 400 m at D regular samples was less of than 15%. Racemization blanks (R P), and act carbon dioxide sources. The average P required for aquatic , in the Non-Biological Decompolations suggesting that theas coastal ocean may be past five decades from studies in which oxygen no acidcontaining peptides could be repeatably deintervals onKinetics a transect between Corpus Christi, Texas, ocean is expected to be net heterotrophic, this carbon demand can be balanced were determined by multiple hydrolyses of pure sition and Racemization of North Aminos Acids in Natural termined to within 5%. and Key West, Florida. The surface sample ecosystems to become (P evolution R) is over order of magnitude net heterotrophic (9) and by theautotrophic finding that was an used as a surrogate for carbon L amino mixtures, as well asSea protein standards. by the excess production over the remaining one-fth of the ocean. Waters, J.acid D. Hem, Ed. (American was integrated on a transect across the central J. L. also Bada and E. H. Mann, EarthChemical Sci. Rev .Society, 16 , 21 greater for marshes than for the open sea. Although D amiEnantiomeric ratios in commercially available bacterial metabolism exceeds phytoplankton fluxes (11). four-fths of the upper Washington, DC, 1971), pp. 308 331. Sea in April 1995. Such very large sample sizes North (1980). no acidcontaining peptides could be repeatably deocean expected be of net heterotrophic, this carbon demand can be balanced production in is unproductive waters ( 10 ), which Community metabolism varied by over four G. A. Goodfriend, Nature 357 , 399 (1992). are compositionally representative, but result in, submake up 30% of the ocean. These concluAquatic ecosystems cover to 70% Earths surP. M. Williams and E. R. M. Druffel, Nature 330 246 termined to within 5%. H. J. Rogers, in Aspects of Microbiology [Van1) Nosby the excess production over the remaining one-fth of the ocean. stantial decreases in relative recovery (Table due orders of magnitude across aquatic ecosystems (1987). Carlos M. Duarte* and Susana Agust sions are based on indirect calculations and face ( 1 ) and contribute 45% of the global priJ. L. Bada and E. H. Mann, Earth Sci. Rev. 16, 21 trand Reinhold, 7, pp. to high concentration factors UK, (11 ).1983), vol. P. H. Santschi etWokingham, al., Geochim. Cosmochim. Acta 59, (Table 1). Marshes tended ( to be morewe produc(1980). Centro de Estudios Avanzados de Blanes, Consejo controversial assumptions 3 ). Here, commary production ( 2 ). Yet, the role of their biota 6 25. amino acid hydrolysis was conducted at UDOM 625 make up other 30%aquatic of power the ocean. These concluAquatic ecosystems cover 70% of Earths surP. M.(1995). Williams and E. R. M. Druffel, Nature 330, 246 Superior de Investigaciones Cient cas, de are Santa tive than ecosystems, Community respiration ( RCam )a rates scaled as the of the gross remains subject of pare the two-thirds gross primary production (P) whereas and resin the global CO2 budget 150C for 70 min, generally with the method of J.-M. Ghuysen and D. Shockman, inGeochem Bacterial R. Mitterer and N.G. Kriausakul, Org. .Mem7, 91 (1987). Ba rbara s/n, 17300 Blanes, 45% Girona, Spain. sions sea arecommunities based on indirect calculations and face (1primary ) and contribute of the global priopen showed the lowest proD and Lamino Cowie and Hedges (29 ). Individual branes Walls , L. Leive, Ed. (Dekker, New York, production ( P ) rates of aquatic ecosystems, indicating that the role of debate ( 35 ). Many freshwater ecosystems act piration ( R ) rates of aquatic communities to (1984); R. W. L. Kimber and P. E. Hare, Geochim. P. H. Santschi et al., Geochim. Cosmochim. Acta 59, controversial assumptions ( 3 ). Here, we commary production ( 2 ). Yet, the role of their biota acids were quantied as pentauoropropyl isopropyl *To whom correspondence should be addressed. 1973), vol. 1. duction and respiration rates (Table 1). The Cosmochim. Acta . 56 , 739 (1992). aquatic biota as carbon dioxide sources elucidate or sinks depends on biota its productivity. sources ( 6 ); in contrast, oceanic ecosysas CO whether the of aquatic ecosys625 (1995). 2 esters by in gas chromatography with ame ionization remains a subject support of pare gross primary production (P resin theare global CO2 budget J. L. Bada, Chemistry and Biochemistry of the R. Mitterer and N. Kriausakul, Org. were Geochem .Amino 7, 91 Unproductive aquatic ecosystems a the disproportionately higher ). (R respiP) ) and or sinks tems assumed to act as CO tems acts as net CO2 sources detection (41 ), and peak identities by 2 sinks (7, 8 Acids, G.R. C. Barrett, Ed. (Chapman & Hare, Hall, veried New York, debate ( 35 ). Many freshwater ecosystems act piration ( R ) rates of aquatic communities to (1984); W. L. Kimber and P. E. Geochim. GC-MS. Analytical variability in D/L ratios of UDOM ration than that of productive aquatic(ecosystems, to be heterotrophic assumption has been challenged by calcuR P). We tend compiled data obtained over the 10 JULYThis 1998 VOL rate 281 SCIENCE www.sciencemag.org 21 1985), p. 684. Cosmochim. Acta . 56 , 739 (1992). sources ( 6 ); in contrast, oceanic ecosysas CO elucidate whether the biota of aquatic ecosyssamples was less than 15%. Racemization blanks 2 R P), and act as carbon dioxide The average required for , in Kinetics of the Non-Biological Decompolations ( suggesting that the coastal ocean may sources. be past five decadesPfrom studies inaquatic which oxygen J. L. Bada, in Chemistry and Biochemistry of the Amino were determined by multiple hydrolyses of pure

The CO2 Bala Unproductive A Ecosystem

The CO2 Balance of Unproductive Aquatic Ecosystems

The CO2 Balance of Unproductive Aquatic The CO2 Balance of Ecosystems Unproductive Aquatic Ecosystems The CO2 Balance of Unproductive Aquatic Ecosystems

tems are assumed to act as CO sinks (7, 8).

tems acts as net CO sources (R

P) or sinks

OCEANOGRAPHY OCEANOGRAPHY OCEANOGRAPHY

Dissolved Organic Carbon Dissolved Organic Carbon Dissolved Organic Carbon Support Support of Respiration in the Support of Respiration in the Dissolved Organic Carbon of Respiration in the Dark Ocean
Javier Arstegu,1* Carlos M. Duarte, 2 Susana Agust, 2 1 2 2 3 4 5 Javier Ar stegui, , Marylo Doval, Xos A. lvarez-Salgado, A. Hansell 1* Carlos M. 2 Susana Agust Javier Ar stegui, * Carlos Dennis M. Duarte, Duarte, Susana Agust ,2
3 3 4 4 1 2 2

BREVIA

Dark Ocean Dark Ocean in the Support of Respiration Dark OceanDennis A. Marylo lvarez-Salgado, Marylo Doval, Doval, Xose Xose A. A. A A lvarez-Salgado, Dennis A. Hansell Hansell

5 5

Javier stegui, * Carlos Duarte, DOC Susana Agust , Recent evidence that Ar dissolved organic car- M.concurrent and AOU observations col3 5 Recent evidence that dissolved organic carconcurrent4DOC and AOU observations col Marylo Doval, Xose A. A lvarez-Salgado, Dennis A. Hansell bon (DOC) is a significant component of the lected in cruises conducted throughout the bon (DOC) is a significant component of the lected in cruises conducted throughout the organic carbon flux below the photic layer of worlds oceans (fig. S1, table S1) to examine organic carbon flux below the photic layer of worlds oceans (fig. S1, table S1) to examine Recent evidence that dissolved organic carconcurrent and AOU colthe ocean (1), together with verification of the relativeDOC contribution ofobservations DOC to AOU the ocean (1), together with verification of the relative contribution of DOC to AOU bon (DOC) is a significant component of(the lected in cruises conducted throughout the high respiration rates in the dark ocean 2), and, therefore, respiration in the dark ocean. high respiration rates in the dark ocean (2), and, therefore, respiration in the dark ocean. organic carbon below the photic layer of AOU worlds oceans (fig. S1, table S1) to examine suggests that theflux downward flux of DOC may increased from an average (SE) suggests that the downward flux of DOC may AOU increased from an average (SE) the ocean (1),role together with verification of 96.3 the relative contribution of DOC AOU play a major in supporting respiration 2.0 M at the base of theto surface play a major role in supporting respiration 96.3 2.0 M at the base of the surface high respiration rates in the dark ocean ( 2 ), and, therefore, respiration in the ocean. there. Here we show, on basis of examimixed layer (100 m) to 165.5 dark 4.3 M at there. Here we show, on the basis of examimixed layer (100 m) to 165.5 4.3 M at suggests downward flux of DOC AOU increased fromthermocline an average (SE) nation of that the the relation between DOC andmay apthe bottom of the main (1000 m), nation of the relation between DOC and apthe bottom of the main thermocline (1000 m), play a oxygen major role in supporting respiration 96.3 a parallel 2.0 M at thein base the surface parent utilization (AOU), that the with decline the of average DOC parent oxygen utilization (AOU), that the with a parallel decline in the average DOC there. flux Here we show, on the of exami- from mixed layer (100 to 165.5 M1). at DOC supports 10% of basis the respiration 53.5 0.2 to m) 43.4 0.3 M 4.3 C (Fig. DOC flux supports 10% of the respiration from 53.5 0.2 to 43.4 0.3 M C (Fig. 1). nation of the relation between DOC and ap- In thecontrast, bottom of the main (1000 m), in the dark ocean. there is no thermocline significant decline in in the dark ocean. In contrast, there is no significant decline in parent oxygen utilization (AOU), that the with a parallel decline in the average DOC DOC flux supports 10% of the respiration from 53.5 0.2 to 43.4 0.3 M C (Fig. 1). in the dark ocean. In contrast, there is no significant decline in

with DOC-poor cold thermocline waters (6 ). with DOC-poor cold thermocline waters (6 ). Removal of this effect by regressing DOC Removal of this effect by regressing DOC against AOU and water temperature indicates against AOU and water temperature indicates withDOC DOC-poor cold thermocline waters 6 ). that supports only 8.4 0.3% of (the that DOC supports only 8.4 0.3% of the Removal of effect by regressing respiration in this the mesopelagic waters. DOC respiration in the mesopelagic waters. against AOU andconfirm water temperature These results that DOC indicates makes a These results confirm that DOC makes a that DOC supports only 8.4 0.3% of the the small but significant contribution toward small but significant contribution toward the respiration inof the mesopelagic waters. in the maintenance respiratory processes maintenance of respiratory processes in the These results confirm that DOCfor makes a dark ocean. DOC use accounts only dark ocean. DOC use accounts for only small contribution toward the 10%but of significant the AOU in mesopelagic waters, 10% of the AOU in mesopelagic waters, maintenance of the respiratory in the indicating that bulk of processes the respiration indicating that the bulk of the respiration dark ocean. DOC usezone accounts for only within the mesopelagic is supported by within the mesopelagic zone is supported by 10% the AOU in This mesopelagic the flux of sinking POC. estimate waters, for the the flux of sinking POC. This estimate for the indicating that bulk of the respiration contribution of the DOC oxidation to global contribution of DOC oxidation to global withinis the mesopelagic zone is supported by AOU supported by the independent findAOU is supported by the independent findthe flux sinking This estimate for for the ings thatofnet DOCPOC. production accounts ings that net DOC production accounts for contribution DOC oxidation to in global 20% of net of community production the 20% of net community production in the AOU isocean supported by the independent findsurface (7 ), that DOC export contribsurface ocean (7 ), that DOC export contribings to that DOC production accounts for utes 10 net % of the export below 500 m (8 ), utes to 10 % of the export below 500 m (8), 20% community production inisthe and thatof thenet total AOU in the deep ocean in and that the total AOU in the deep ocean is in surface ocean ), that DOCfrom export agreement with(7 that expected thecontribflux of agreement with that expected from the flux of utes to 10 % (of export below 500 m (8), sinking POC 9).the It is the DOC accumulating sinking POC (9). It is the DOC accumulating and the total in the deep ocean in as a that function of AOU primary production in is the as a function of primary production in the agreement withwhich that expected fromwith the flux of surface ocean, is exported oversurface ocean, which is exported with oversinking circulation, POC (9). It is the DOC accumulating turning that dominates the DOC turning circulation, that dominates the DOC as a function of development. primary production in the fraction of AOU Our findings fraction of AOU development. Our findings surface ocean, is exported with overindicate that thewhich POC flux, which appears to indicate that the POC flux, which appears to turning circulation, that dominates the DOC be severely underestimated by sediment traps be severely underestimated by sediment traps fraction of AOU development. Our findings in the mesopelagic zone (10), is much greater in the mesopelagic zone (10), is much greater indicate theflux POCand flux, which appears to than the that DOC supports the bulk than the DOC flux and supports the bulk be severely underestimated by sediment traps ( 90%) of the respiratory carbon demand in (90%) of the respiratory carbon demand in in the mesopelagic zone (10), is much greater the dark ocean. the dark ocean. than the DOC flux and supports the bulk References (90%) of the respiratory carbon demand in References 1. C. A. Carlson, H. W. Ducklow, A. F. Michaels, Nature the dark ocean. 1. 371 C. A. Carlson, H. W. Ducklow, A. F. Michaels, Nature , 405 (1994).
2. 2. 1. 3. 3. 4. 4. 2. 5. 5. 3. 6. 6. 4.

Fig. 1. The depth distribution of the average (mean SE of data grouped by 100-m bins, N Fig. 1.DOC The depth distribution of the average (mean SE of data grouped 100-m bins, N 9578) (full circles) concentration and AOU (empty circles) in the oceanby (A) and the relation 9578) DOC (full circles) concentration and AOU (empty circles) in the ocean (A) and the relation between DOC and AOU in the ocean (B ). The symbols represent mean SE DOC concentrations between DOC and AOU in AOU the ocean (N B). The symbols represent mean the SE DOC concentrations of data grouped bydistribution 10 M 9824). The in data (B) shows relation for the N raw Fig. 1. The depth ofbins the ( average (mean inset SE of grouped by 100-m bins, of data grouped by 10 M AOU bins ( N 9824). The inset in (B) shows the relation for the raw data ( N 9824), which is described, in the interval 0 AOU 150, by the tted regression 9578) DOC (full circles) concentration and AOU (empty circles) in the ocean (A) tted and the relation data ( N 9824), which is described, in the interval 0 AOU 150, by the regression 2 0.28, P 0.0001, 5541). Multiple equation DOC 60.3 (0.2) ocean 0.136 ( 0.003) AOU (R represent between and AOU in the ( B ). The symbols SE N DOC concentrations independent 0.28, mean P 0.0001, N 5541). Multiple equation DOC 60.3 ( 0.2) 0.136 0.003) AOU ( R2 regression analysis, using water temperature (C) as an variable, improves theraw t, of data grouped by also 10 M AOU bins (N 9824). The inset in (B) shows the relation for regression analysis, also using water temperature (C) as an independent variable, improves the t, 2 the yielding the equation DOC 48.0 ( 0.2) 0.81 ( 0.01) T 0.058 ( 0.002) AOU ( R 2 0.69, data ( N 9824), which is described, in the interval 0 AOU 150, by the tted regression yielding the equation DOC 48.0 (0.2) decrease 0.81 ( 0.01) T for 0.058 (225 0.002) AOU (R P 0.0001, N 5371). The monotonic of AOU corresponds to0.69, the 2 0.28, P 0.0001, N 5541). Multiple equation DOC 60.3 (0.2) 0.136 (0.003) AOU (RDOC P 0.0001, N 5371). The monotonic decrease of DOC forequatorial AOU 225 corresponds to the oxygen minimum zones of the Arabian Sea, Indian Ocean, and Pacic Ocean. regression analysis, also using water temperature (C) as an and independent variable, oxygen minimum zones of the Arabian Sea, Indian Ocean, equatorial Pacic improves Ocean. the t, yielding the equation DOC 48.0 (0.2) 0.81 (0.01) T 0.058 (0.002) AOU (R2 0.69, P 0.0001, N 5371). The monotonic of DOC for AOU 225 corresponds to the The contribution of DOC to pelagic decrease resDOC with increasing depth beyond 1000 m The minimum contribution of of DOC to pelagic resDOC with increasing depth beyond 1000 m oxygen zones the Arabian Sea, Indian Ocean, and equatorial Pacic Ocean. piration below the surface mixed layer can be depth (Fig. 1), indicating that DOC exported

371 , 405 P. A. del (1994). Giorgio, C. M. Duarte, Nature 420, 379 P. A. del Giorgio, C. M. Duarte, Nature 420, 379 References (2002). (2002). C. D. A. Doval, Carlson, H. Hansell, W. Ducklow, A. F. Michaels, Nature M. D. A. Mar. Chem . 68 , 249 (2000). M.T. D. Doval, D. Mar. Chem . 68 , 249 (2000). 371 ,Peltzer, 405 (1994). E. N.A. A.Hansell, Hayward, Deep Sea Res. Part II 43, E. Peltzer, N. A. Hayward, Deep Sea Res. Part 43, P. T. A. del Giorgio, C. M. Duarte, Nature 420,II379 155 (1996). 155 (2002). D. A. (1996). Hansell, C. A. Carlson, Nature 395, 263 (1998). D. A. Carlson, 395 , 263 (1998). lvarez-Salgado, M. A. D. Hansell, Doval, D. C. A. A. Hansell, Chem 68F. , 249 (2000). X. A F.Mar. F. Nature Pe rez, .A. R os, M. D. lvarez-Salgado, X. T. A.Peltzer, A F. F. Deep Pe rez,Sea A. Res. F. R os, D. E. N.. A. Part M. II 43 , Doval, Sci. Mar 65Hayward, , 1 (2001). Doval, Sci. Mar. C. 65, 1 Carlson, (2001). Global Biogeochem. 155 7. D. A.(1996). Hansell, A. 7. D. A. Hansell, Hansell, C. Carlson, Global 5. Cycles D. A. A. A. Carlson, Nature 395,Biogeochem. 263 (1998). 12 , 443C. (1998). Cycles , 443 12 6. D. X. A. A. Hansell, A lvarez-Salgado, F. F. Pe rez, F. R os, M. D. 8. in (1998). Biogeochemistry of A. Marine Dissolved 8. Organic D. A. Hansell, in Biogeochemistry of A. Marine Dissolved Doval, Sci. Mar ., 65 Matter D., 1 A.(2001). Hansell, C. Carlson, Eds. Organic Matter D.A. A.Carlson, Hansell, C. A.685715. Carlson, Eds. 7. (Academic, D. A. Hansell, C. Global Biogeochem. San ,Diego, CA, 2002), pp. (Academic, SanGlobal Diego, CA, 2002), pp.Cycles 685715. Cycles 12, 443 (1998). 9. R. A. Jahnke, Biogeochem. 10, 71 9. R. A. Hansell, Jahnke,inGlobal Biogeochem. CyclesDissolved 10, 71 8. (1996). D. A. Biogeochemistry of Marine (1996). Organic Matter , D. Hansell, C. A. Carlson, Eds. 10. J. C. Scholten et al., A. Deep Sea Res. Part II 48, 243 10. (2001). J. C. Scholten al., Deep Sea Res. II 48, 243 (Academic, Sanet Diego, CA, 2002), pp.Part 685715. (2001). 9. R. A. Jahnke, Global Biogeochem. Cycles 10, 71 Supporting Material (1996). Online Supporting Online Material www.sciencemag.org/cgi/content/full/298/5600/1967/ 10. J. C. Scholten et al., Deep Sea Res. Part II 48, 243 www.sciencemag.org/cgi/content/full/298/5600/1967/ DC1 (2001). DC1S1 Fig. Fig. S1 Supporting Online Material Table S1 Table S1 www.sciencemag.org/cgi/content/full/298/5600/1967/ DC1 Fig. S1 1 Facultad de Ciencias del Mar, Campus Universitario 1 Table S1 de Ciencias del Mar, Campus Universitario Facultad

de Tara, las Canaria, Palmas de Gran2IMEDEA Canaria, piration below the surface mixed layer can be depth (Fig. 1), indicating that DOC exported 35017 LasUniversidad Palmas de de Gran Spain. 2 inferred from the relation between DOC and with overturning circulation plays a minor 35017 Las Palmas de Gran Canaria, Spain. IMEDEA Instituto Mediterra neo de Estudios Avaninferred from the relation between DOC and with overturning circulation plays a minor (CSIC-UIB), 1 (CSIC-UIB), Instituto Mediterra neo de Esporles Estudios AvanThe contribution of DOC to pelagic resDOCin with increasing depth beyond 1000 m zados, ), Facultad de Ciencias del Mar, Campus Universitario apparent oxygen utilization (AOU, M O role supporting respiration in the ocean Miquel Marque s 21, 07190 (Islas ), apparent oxygen utilization (AOU, M O2 role in supporting respiration in the ocean zados, Miquel Marque s las 21, 07190 (Islas de Tara, Universidad de Palmas deEsporles Gran 3 piration below the surface mixed layer can 2 be interior depth (Fig. indicating DOC exported Baleares), a variable quantifying the cumulative oxygen (5).1), Assuming a that molar respiratory Spain. CCCMM ( XUGA), Centro de2Canaria, Control 3 a variable quantifying the cumulative oxygen interior (5). Assuming a molar respiratory de Baleares), Spain. CCCMM ( XUGA), Centro de Control 35017 Las Palmas de Gran Canaria, Spain. IMEDEA Calidade do Medio Marin o, Peirao de Vilaxoa n s/n, inferred fromsince the relation DOC and with overturning circulation a minor consumption a waterbetween parcel was last in quotient of 0.69, the decline in plays DOC accounts de Calidade do MedioMediterra Marin o, Peirao Vilaxoa Avann s/n, (CSIC-UIB), Instituto neo dede Estudios 4 consumption since a water parcel was last in quotient of 0.69, the decline in DOC accounts Vilagarc a de Arousa, Spain. IIM (CSIC), Instituto de apparentwith oxygen (AOU, M O role19.6 in supporting in thethe ocean contact the utilization atmosphere. However, asfor 0.4% of respiration the AOU within top Vilagarc a de Spain. 2 ), (CSIC), Instituto de zados, Miquel Marque s Eduardo 21,4IIM 07190 Esporles (Islas contact with the atmosphere. However, asfor 19.6 0.4% of the AOU within the top Investigacio ns Arousa, Marin as, Cabello 6, 36208 3 a variable quantifying therelations cumulative oxygen interior 5). Assuming a molar represents, respiratory Vigo, Investigacio ns Marin as, Eduardo Cabello 6, 36208 sessments of DOC/AOU have been 1000 m ((Fig. 1). This estimate Baleares), Spain. CCCMM ( XUGA), Centro de Control 5 Spain. RSMAS, University of Miami, Miami, FL sessments of DOC/AOU relations have been 1000 m (Fig. 1). This estimate represents, 5 Vigo, Spain. RSMAS, of Miami, Miami, FL de Calidade Medio University Marin o, Peirao de Vilaxoa n s/n, consumption sinceregions a waterof parcel was last in quotient of 0.69, the decline in DOC limited to specific the ocean (3, 4 ) however, an upper limit, because theaccounts correlaUSA. do 4 limited to specific regions of the ocean (3, 4 ) however, an upper limit, because the correla- 33149, 33149, a USA. Vilagarc de Arousa, Spain. IIM (CSIC), Instituto de contact with the atmosphere. However, asfor 19.6 0.4% of the AOU within the top and have not considered the global ocean. We tion between DOC and AOU is partly due to whomcorrespondence shouldCabello be addressed. EInvestigacio Marin as, Eduardo 6, 36208 and have not considered the global ocean. We tion between DOC and AOU is partly due to *To *To whom ns correspondence should be addressed. Esessments of DOC/AOU relations been 1000 mof (Fig. 1). This estimate represents, mail: assembled a large data set (N have 9824) of mixing DOC-rich warm surface waters Vigo, jaristegui@dbio.ulpgc.es Spain. 5RSMAS, University of Miami, Miami, FL mail: jaristegui@dbio.ulpgc.es assembled a large data set (N 9824) of mixing of DOC-rich warm surface waters limited to specific regions of the ocean (3, 4 ) however, an upper limit, because the correla33149, USA. and have not considered the global ocean. We tion between SCIENCE DOC and AOU is partly due to *To whom correspondence should be addressed. Ewww.sciencemag.org VOL 298 6 DECEMBER 2002 1967 www.sciencemag.org SCIENCE warm VOL 298 6 DECEMBER 2002 1967 mail: jaristegui@dbio.ulpgc.es assembled a large data set (N 9824) of mixing of DOC-rich surface waters

de Tara, Universidad de las Palmas de Gran Canaria,

22

www.sciencemag.org SCIENCE 298 2002 6 DECEMBER 2002 Originally published 6VOL December

1967

Community Genomics Among Stratified Microbial Assemblages in the Oceans Interior

Edward F. DeLong,1* Christina M. Preston, 2 Tracy Mincer,1 Virginia Rich,1 Steven J. Hallam,1 Niels-Ulrik Frigaard,1 Asuncion Martinez,1 Matthew B. Sullivan,1 Robert Edwards,3 Beltran Rodriguez Brito,3 Sallie W. Chisholm,1 David M. Karl4 Microbial life predominates in the ocean, yet little is known about its genomic variability, especially along the depth continuum. We report here genomic analyses of planktonic microbial communities in the North Pacific Subtropical Gyre, from the oceans surface to nearsea floor depths. Sequence variation in microbial community genes reflected vertical zonation of taxonomic groups, functional gene repertoires, and metabolic potential. The distributional patterns of microbial genes suggested depth-variable community trends in carbon and energy metabolism, attachment and motility, gene mobility, and host-viral interactions. Comparative genomic analyses of stratified microbial communities have the potential to provide significant insight into higher-order community organization and dynamics.

analyses were used to identify taxa, genes, and metabolic pathways that characterized vertically stratified microbial assemblages in the water column. Study Site and Sampling Strategy Our sampling site, Hawaii Ocean Time-series (HOT) station ALOHA (2245 N, 158W), represents one of the most comprehensively characterized sites in the global ocean and has been a focal point for time seriesoriented oceanographic studies since 1988 (22). HOT investigators have produced high-quality spatial and time-series measurements of the defining physical, chemical, and biological oceanographic parameters from surface waters to the seafloor. These detailed spatial and temporal datasets present unique opportunities for placing microbial genomic depth profiles into appropriate oceanographic context (2224) and leverage these data to formulate meaningful ecological hypotheses. Sample depths were selected, on the basis of well-defined physical, chemical, and biotic characteristics, to represent discrete zones in the water column (Tables 1 and 2, Fig. 1; figs. S1 and S2). Specifically, seawater samples from the upper euphotic zone (10 m and 70 m), the base of the chlorophyll maximum (130 m), below the base of the euphotic zone (200 m), well below the upper mesopelagic (500 m), in the core of the dissolved oxygen minimum layer (770 m), and in the deep abyss, 750 m above the seafloor (4000 m), were collected for preparing microbial community DNA libraries (Tables 1 and 2, Fig. 1; figs. S1 and S2). The depth variability of gene distributions was examined by random, bidirectional endsequencing of 5000 fosmids from each depth, yielding 64 Mbp of DNA sequence total from the 4.5 Gbp archive (Table 1). This represents raw sequence coverage of about 5 (1.8 Mbp sized) genome equivalents per depth. Because we surveyed 180 Mbp of cloned DNA (5000 clones by 36 kbp/clone per depth), however, we directly sampled 100 genome equivalents at each depth. We did not sequence as deeply in each sample as a recent Sargasso Sea survey (19), where from 90,000 to 600,000 sequences were obtained from small DNA insert clones, from each of seven different surface-water samples. We hypothesized, however, that our comparison of microbial communities collected along well-defined environmental gradients (using large-insert DNA clones), would facilitate detection of ecologically meaningful taxonomic, functional, and community trends. Vertical Profiles of Microbial Taxa Vertical distributions of bacterial groups were assessed by amplifying and sequencing small subunit (SSU) ribosomal RNA (rRNA) genes from complete fosmid library pools at each

icrobial plankton are centrally involved in fluxes of energy and matter in the sea, yet their vertical distribution and functional variability in the oceans interior is still only poorly known. In contrast, the vertical zonation of eukaryotic phytoplankton and zooplankton in the oceans water column has been well documented for over a century (1). In the photic zone, steep gradients of light quality and intensity, temperature, and macronutrient and trace-metal concentrations all influence species distributions in the water column (2). At greater depths, low temperature, increasing hydrostatic pressure, the disappearance of light, and dwindling energy supplies largely determine vertical stratification of oceanic biota. For a few prokaryotic groups, vertical distributions and depth-variable physiological properties are becoming known. Genotypic and phenotypic properties of stratified Prochlorococcus ecotypes for example, are suggestive of depth-variable adaptation to light intensity and nutrient availability (35). In the abyss, the vertical zonation of deep-sea piezophilic bacteria can be explained in part by their obligate growth requirement for elevated hydrostatic pressures (6). In addition, recent cultivation-independent (715) surveys have shown vertical zonation patterns among

1 Massachusetts Institute of Technology, Cambridge, MA 02139, USA. 2Monterey Bay Aquarium Research Institute, Moss Landing, CA 95064, USA. 3San Diego State University, San Diego, CA 92182, USA. 4University of Hawaii Honolulu, HI 96822, USA.

* To whom correspondence should be addressed. Email: delong@mit.edu

specific groups of planktonic Bacteria, Archaea, and Eukarya. Despite recent progress however, a comprehensive description of the biological properties and vertical distributions of planktonic microbial species is far from complete. Cultivation-independent genomic surveys represent a potentially useful approach for characterizing natural microbial assemblages (16, 17). Shotgun sequencing and whole genome assembly from mixed microbial assemblages has been attempted in several environments, with varying success (18, 19). In addition, Tringe et al. (20) compared shotgun sequences of several disparate microbial assemblages to identify community-specific patterns in gene distributions. Metabolic reconstruction has also been attempted with environmental genomic approaches (21). Nevertheless, integrated genomic surveys of microbial communities along well-defined environmental gradients (such as the oceans water column) have not been reported. To provide genomic perspective on microbial biology in the oceans vertical dimension, we cloned large [36 kilobase pairs (kbp)] DNA fragments from microbial communities at different depths in the North Pacific Subtropical Gyre (NPSG) at the open-ocean timeseries station ALOHA (22). The vertical distribution of microbial genes from the ocean's surface to abyssal depths was determined by shotgun sequencing of fosmid clone termini. Applying identical collection, cloning, and sequencing strategies at seven depths (ranging from 10 m to 4000 m), we archived large-insert genomic libraries from each depth-stratified microbial community. Bidirectional DNA sequencing of fosmid clones (10,000 sequences per depth) and comparative sequence
Originally published 27 January 2006

23

Vertical distributions of bacterial groups were assessed by amplifying and sequencing small

(23.58 T 1.00) (35.17 T 0.16) (0.15 T 0.05) [86] (81.4 T 11.3) (14.7 T 60.3) (43.1 T 25.1) (215.8 T 5.4) (1,986.9 T 15.4) [1,202] [1,084] [363] [79] [78] [104] [144] [84] 130 22.19 35.31 0.10 5.03 T 2.30 69 284.8 66.2 204.9 2,026.5 Table 2. HOT sample oceanographic data. Samples described in Table 1. dissolved inorganic carbon. The estimated photon fluxes for upper water (21.37 T 0.96) (35.20 0.10) (0.15 T 0.06) at (49 [90] (75.2 T 9.1) (282.9 T 270.2) (106.0 T 49.7) (206.6 6.2) (2,013.4 13.4) j2 Oceanographic parameters were T measured as specified ); values shown column samples (assuming a surface irradiance of T 32 mol quanta mT dj1 [1,139] [350] where available. Values in [86] and a light extinction [78] [68] [173] [69] (65% are those from the same CTD [980] casts as the samples, coefficient of 0.0425 mj1 ) were: 10 m 0 20.92 200 0.02 1.66 T 0.24 1,161.9 274.2 109.1 130 m 198.8 2,047.7 parentheses 18.53 are the mean T 35.04 1 SD of each core parameter during the period63 of surface), 70 T m762.5 0 1.63 (5% ofTsurface), 0 0.128 (0.4% of surface), T 1.29) (34.96 T 0.18) (0.02 T 0.02) (64.0 T 9.8) [7](0.02% of surface). [84] The mean (197.6 T 7.1) (2,042.8 T 10.5) October (18.39 1988 to December 2004, with the total number of[2] measurements 200 m 0 0.07 surface mixed-layer during the [576] in brackets. [97] [113]October 2002 sampling was 61 m. Data are [190] [125] collected for [662] each parameter shown The parameter abbreviations available at (50). *Biomass 500 7.25 34.07 ND dissolved 0.48organic T 0.23 carbon;47 derived from 28,850 2,153 triphosphate 118.0 2197.3 asare Temp., Temperature; Chl a, chlorophyll a; DOC, particulate adenosine (ATP) measurements (7.22 T 0.44) (34.06 T dissolved 0.03) [107] and (47.8 (28,460 T 2210) (2,051 T 175.7) T 18.3) (2,200.2 T 17.8) NN, nitrate plus nitrite; DIP, inorganic phosphate; DIC, T 6.3) suming a carbon:ATP ratio of 250. ND, Not(120.5 determined. [1,969] [1,769] [112] [326] [322] [505] [134] Depth Temp. Chl Biomass* DOC NN DIP Oxygen DIC 770 4.78 34.32 ND a 0.29 T 0.16 39.9 41,890 3,070 32.3 2323.8 Salinity (m) (4.86((mg/kg) ([107] mg/kg) (mmol/kg) (nmol/kg) (nmol/kg) (mmol/kg) (mmol/kg) TC) 0.21) (34.32 T 0.04) (41.5 T 4.4) (40,940 T 500) (3,000 T 47.1) (27.9 T 4.1) (2,324.3 T 6.1) [888] [773] [34] [137] [135] [275] [34] 10 26.40 35.08 0.08 7.21 T 2.68 78 1.0 41.0 204.6 1,967.6 4,000 (24.83 1.46 34.69 ND 37.5 36,560 2,558 147.8 2325.5 T 1.27) (35.05 T 0.21) (0.08ND T 0.03) [78] (90.6 T 14.3) (2.6 T 3.7) (56.0 T 33.7) (209.3 T 4.5) (1,972.1 T 16.4) (1.46 T 0.01) (34.69 T 0.00) (42.3 T 4.9) (35,970 T 290) (2,507 T 19) (147.8 T 1.3) (2,329.1 T 4.8) [2,104] [1,611] [320] [140] [126] [146] [348] [107] [262] [245] [83] [108] [104] [210] [28] 70 24.93 35.21 0.18 8.51 T 3.22 79 1.3 16.0 217.4 1,981.8 (23.58 T 1.00) (35.17 T 0.16) (0.15 T 0.05) [86] (81.4 T 11.3) (14.7 T 60.3) (43.1 T 25.1) (215.8 T 5.4) (1,986.9 T 15.4) [1,202] [1,084] [363] [79] [78] [104] [144] [84] 130 22.19 35.31 www.sciencemag.org 0.10 5.03 T 2.30 69 VOL 311 284.8 204.9 2,026.5 SCIENCE 27 JANUARY 66.2 2006 497 (21.37 T 0.96) (35.20 T 0.10) (0.15 T 0.06) [90] (75.2 T 9.1) (282.9 T 270.2) (106.0 T 49.7) (206.6 T 6.2) (2,013.4 T 13.4) [1,139] [980] [350] [86] [78] [68] [173] [69] 200 18.53 35.04 0.02 1.66 T 0.24 63 1,161.9 T 762.5 274.2 T 109.1 198.8 2,047.7 (18.39 T 1.29) (34.96 T 0.18) (0.02 T 0.02) [2] (64.0 T 9.8) [7] [84] (197.6 T 7.1) (2,042.8 T 10.5) [662] [576] [97] [113] [190] [125] 500 7.25 34.07 ND 0.48 T 0.23 47 28,850 2,153 118.0 2197.3 (7.22 T 0.44) (34.06 T 0.03) [107] (47.8 T 6.3) (28,460 T 2210) (2,051 T 175.7) (120.5 T 18.3) (2,200.2 T 17.8) [1,969] [1,769] [112] [326] [322] [505] [134] 770 4.78 34.32 ND 0.29 T 0.16 39.9 41,890 3,070 32.3 2323.8 (4.86 T 0.21) (34.32 T 0.04) [107] (41.5 T 4.4) (40,940 T 500) (3,000 T 47.1) (27.9 T 4.1) (2,324.3 T 6.1) [888] [773] [34] [137] [135] [275] [34] 4,000 1.46 34.69 ND ND 37.5 36,560 2,558 147.8 2325.5 (1.46 T 0.01) (34.69 T 0.00) (42.3 T 4.9) (35,970 T 290) (2,507 T 19) (147.8 T 1.3) (2,329.1 T 4.8) [262] [245] [83] [108] [104] [210] [28]

water column (8, 15). Unexpectedly large amounts of phage DNA were recovered in clones,ARTICLES particuRESEARCH larly in the photic zone. Also unexpected was a high proportion of BetaproteobacteriaUnexpectedly large amounts of phage DNA Table 1. HOT samples and fosmid libraries. Sample site, 22 -45 N, 158-W. All seawater samples Alteromonadaeceae ; and SAR202, SAR11, and subunit (SSU) ribosomal RNA (rRNA) genes relatively 36 kbp/clone per depth), however, we directly sequences recovered at 130 m, most were pre-filtered through a 1.6-mmat glass filter,from and collected on a 0.22-library mm filter. See ( 35each ) for like Agg47 planktonic clades (Fig. sharing 2; the fig. complete fosmid pools at sampled 100 genome equivalents eachfiber depth. were recovered in bacterial clones, particularly in similarity to protein homologs from methods. S2). Large-insert clones previously recovWe did not sequence as deeply in each sample depth (Fig. 2; fig. S3). Bacterial phylogenetic highest photic zone. AlsoDNA unexpected was a relatively ferrireducens . As expected, ered from the marine (9representa, 10) also as a recent Sargasso Sea survey (19), where from distributions were generally consistent with Rhodoferax high proportion of environment Betaproteobacteria-like Total DNA (Mbp) Depth Sample Volume filtered Total fosmid tion of Prochlorococcus and Pelagibacter -like provide a good metric -like for130 taxonomic assessment polymerase chain reactionbased 90,000 to 600,000 sequences were obtained from previous sequences recovered at m, most sharing genomic sequences was high in the photic zone. At (m) DNA insert date (liters) clones Archived Sequenced of indigenous microbes. Accordingly, a relatively rRNA surveys of ma- highest small clones, from each of seven dif- cultivation-independent similarity to protein homologs from greater depths, higher proportions of Chloroflexi large proportion of our shotgun fosmid sequences rine picoplankton ( 8 , 15 , 25 ). In surface-water ferent surface-water samples. We hypothesized, Rhodoferax ferrireducens. As expected, repre10 10/7/02 40 12,288 442 7.54 like sequences, perhaps corresponding to the comost closely matched rRNA-containing bactehowever, that 10/7/02 our comparison of microbial com- samples, sentation of Prochlorococcus -like and Pelagi70 40 12,672rRNA-containing 456 fosmids included 11.03 occurring SAR202 clade, were observed (Fig. 2). rioplankton artificial clones previously recovered from Prochlorococcus munities collected along well-defined bacter -like genomic sequences was high in the 130 10/6/02 40 environ- those 13,536 487 ; Verrucomicro6.28 Planctomycetales -like genomic (fig. DNA sequences from the marine environment S3). ; Flexibacteraceae ; Gammaproteobacteria mental gradients (using large-insert DNA photic zone. At greater depths, higher propor200 10/6/02 40 clones), biales 19,008 684 7.96 were also highlybins represented at greater depths. Taxonomic of bacterial protein (SAR92, OM60, SAR86 clades); Alphaproteowould facilitate detection of ecologically meaningtions of Chloroflexi-like sequences, homologs perhaps 500 10/6/02 80 15,264 550 8.86 All archaeal SSU rRNAcontaining fosmids found in randomly fosmid ends (Fig. 2; bacteria (SAR116, OM75 Delta- corresponding ful taxonomic, functional, and community to sequenced the cooccurring SAR202 770 12/21/03 240 trends. 11,520 415 clades); and 11.18 were identified at each depth, quantified by macfig. S4) also reflected distributional patterns proteobacteria (OM27 clade) (Fig. 2). Bacterial clade, were observed (Fig. 2). Planctomyce4,000 12/21/03 670 41,472 1,493 11.10 roarray hybridization, and previous their rRNAs sequenced Vertical Profiles of Microbial Taxa generally consistent with surveys in the groups from deeper waters included members tales -like genomic DNA sequences were also water column ( 8 , 15 ). Unexpectedly large amounts Vertical distributions of bacterial groups were of Deferribacteres; Planctomycetaceae; Acido- highly represented at greater depths. assessed by amplifying and sequencing small bacteriales; Gemmatamonadaceae; Nitrospina; of phage DNA were recovered in clones, particudepth (Fig. 2; fig. S3).oceanographic Bacterial phylogenetic planktonic bac All estimated archaeal photon SSU rRNAcontaining fosTable 2. HOT sample data. SamplesSAR202, describedSAR11, in Table and 1. Agg47 dissolved inorganic carbon. The fluxes for upper water larly in the photic zone. Also unexpected was a distributions were generally consistent terial clades (Fig. 2; fig. S2). Large-insert DNA mids wereirradiance identified each Oceanographic parameters were measured aswith specified at (49 ); values shown column samples (assuming a surface ofat 32 moldepth, quantaquantified mj2 dj1 relatively high proportion of BetaproteobacteriaTable 1. from HOT samples fosmid Sample site, 22-45 N, 158 All a seawater samples are those the same and CTD casts as libraries. the samples, where available. Values in -W.and light the extinction coefficient of 0.0425 hybridization, mj1) were: 10 and m 0their 20.92 (65% previous polymerase chain reactionbased clones previously recovered from marine by macroarray rRNAs likeof sequences recovered at 130 m, most sharing were pre-filtered through a 1.6m m glass fiber filter, and collected on a 0.22m m filter. See ( 35 ) for parentheses are the mean T 1 SD of each core parameter during the period of surface), 70 m 0 1.63 (5% surface), 130 m 0 0.128 of surface), cultivation-independent rRNA surveys of ma- environment (9, 10) also provide a good met- sequenced (figs. S5 and S6). (0.4% The general pathighest The similarity to protein homologs from methods. October 1988 to December 2004, with the total number of taxonomic measurements 200 m 0 (0.02% of surface). mean surface mixed-layer during the ric for assessment of0.07 indigenous terns of archaeal distribution we observed rine picoplankton (8, 15, 25 ). In surface-water Rhodoferax ferrireducens . As expected, collected for each parameter shown in brackets. The parameter abbreviations October 2002 sampling was 61 consistent m. Data are available at (field 50).representa*Biomass microbes. Accordingly, a relatively large prosamples, rRNA-containing fosmids included were with previous surveys Total DNA (Mbp) Depth Sample Chl a, chlorophyll Volume filtered Total fosmid tion of Prochlorococcus -like and Pelagibacter-like are Temp., Temperature; a; DOC, dissolved organic carbon; fosmid derived from particulate triphosphate (ATP) measurements asof our shotgun sequences most adenosine those from Prochlorococcus; Verrucomicro( 15, 25, 26 ). Recovery of group II plankton genomic sequences was high in the photic zone. At (m) datenitrite; DIP, dissolved (liters) inorganicportion clones Archived Sequenced N N, nitrate plus phosphate; and DIC, suming a carbon:ATP ratio of 250. ND, Not determined. Euryarchaeota genomic DNAof was greatest biales; Flexibacteraceae; Gammaproteobac- closely matched rRNA-containing bacterio- ic greater depths, higher proportions Chloroflexi 10 (SAR92,10/7/02 40 Alpha- plankton 12,288 442 7.54 artificial clones in the upper water column and declined below teria OM60, SAR86 clades); Depth Temp. Chl a Biomass* DOC previously Nrecovered N DIP Oxygen DIC like sequences, perhaps corresponding to the co70 10/7/02 40 12,672 456 proteobacteria (SAR116, Salinity OM75 clades); and marine environment S3).11.03 the photic zone.(m This distribution corrobo(m) (-C) (m g/kg) from the (mg/kg) (mmol/kg) (fig.(nmol/kg) (nmol/kg) mol/kg) (mmol/kg) occurring SAR202 clade, were observed (Fig. 2). 130 10/6/02 40(Fig. 2). 13,536 6.28 Taxonomic bins of487 bacterial protein ho- rates recent observations of ion-translocating Deltaproteobacteria (OM27 clade) Planctomycetales -like genomic DNA sequences 10 26.40 35.08 0.08 7.21 T 2.68 78 1.0 41.0 204.6 1,967.6 200 10/6/02 40included mologs 19,008 684 sequenced fosmid 7.96 found in randomly photoproteins (called proteorhodopsins), now Bacterial groups from deeper waters were T also highly represented at greater depths. (24.83 10/6/02 T 1.27) (35.05 T 0.21) [78] (90.6 T 14.3) (2.6 T 3.7) 33.7) (209.3 T 4.5) (1,972.1 T 16.4) 500 80 (0.08 T 0.03) 8.86 - (56.0 ends15,264 (Fig. 2; fig. S4) 550 also reflected distribu known to occur SSU in group II Euryarchaeota inmembers of Deferribacteres; PlanctomyceAll archaeal rRNAcontaining fosmids [2,104] [1,611] 240 [320] [140] [126] [146] [348] [107] 770 12/21/03 11,520 415 11.18 patterns generally consistent with pre- habiting the photic zone (27). Group III Eutaceae; Acidobacteriales; Gemmatamonada- tional were identified at each depth, quantified by mac70 24.93 35.21 670 0.18 8.51 T 3.22 79 1.3 16.0 217.4 1,981.8 4,000 Nitrospina; 12/21/03 41,472 ceae; Alteromonadaeceae; and vious surveys in the 1,493 water column (11.10 8, 15). ryarchaeota DNA was recovered at all depths, roarray hybridization, and their rRNAs sequenced of Deferribacteres; Planctomycetaceae; Acidobacteriales; Gemmatamonadaceae; Nitrospina;

24

www.sciencemag.org

SCIENCE

VOL 311

27 JANUARY 2006

497

RESEARCH ARTICLES
(figs. and S6). The general patterns of S5 archaeal but atS5 a much lower frequency (figs. and distribution we observed were consistent with preS6). A novel crenarchaeal group, closely relat vious field surveys (thermophilic 15, 25, 26). Recovery of group ed to a putatively Crenarchaeota planktonic Euryarchaeota genomic DNA(fig. was (II 28), was observed at the greatest depths greatest in the upper water column and declined S6). below the photic zone. This distribution corroborates recent observationsGenes of ion-translocating phoVertically Distributed toproteins (called proteorhodopsins), now known and Metabolic Pathways to occur in group II Euryarchaeota inhabiting the The depths sampled were specifically chosen photic zone (27 ). Group sequences III Euryarchaeota DNA to capture microbial at discrete was recovered at zones all depths, but at a much lower biogeochemical in the water column frequency (figs. S5 and S6). A novel crenarchaeal encompassing key physicochemical features group, closely related to a putatively thermophilic (Tables 1 and 2, Fig. 1; figs. S1 and S2). To Crenarchaeota (28), was observed at the greatest evaluate sequences from each depth, fosmid depths (fig. S6). end sequences were compared against different databases including the Kyoto EncycloVertically Distributed Genes pedia of Genes Pathways and Genomes (KEGG) (29), and Metabolic National Center for were Biotechnology The depths sampled specifically Informachosen to tion (NCBI)s Clusters of Orthologous capture microbial sequences at discrete Groups biogeo(COG) (30 ), and SEED subsystems (31). Afchemical zones in the water column encompassing ter categorizing sequences from each 1 depth key physicochemical features (Tables and in 2, BLAST searches ( 32 ) against each database, Fig. 1; figs. S1 and S2). To evaluate sequences we identified protein categories that were from each depth, fosmid end sequences were more or less well different represented in oneincluding sample compared against databases versus another, using cluster analysis (33, 34) the Kyoto Encyclopedia of Genes and Genomes and bootstrap methodologies (35). (KEGG) (29), resampling National Center for Biotechnology Cluster analyses predicted seInformation (NCBI)sofClusters of protein Orthologous quence representation identified specific Groups (COG) (30), and SEED subsystems (31). genes and metabolic traits that were differenAfter categorizing sequences from each depth in BLAST searches in (32 ) against each database, we tially distributed the water column (fig. S7). identified protein that 130 were more or In the photic zonecategories (10, 70, and m), these less well arepresented in one sample versus anincluded greater representation in sequences other, using cluster analysis (33,porphyrin 34) and bootassociated with photosynthesis; and strap resampling methodologies 35). secretion chlorophyll metabolism; type (III Cluster analyses of predicted protein sequence systems; and aminosugars, purine, propono representation identified specific genes and metaate, and vitamin B6 metabolism, relative to bolic traits that were differentially distributed in deep-water samples (fig. S7). Independent the water column (fig. S7). In the photic zone (10, comparisons with well-annotated subsystems 70, and 130 m), these included a in the SEED database (31) also showedgreater simirepresentation in sequences with pholar and overlapping trends associated (table S1), includtosynthesis; porphyrin and chlorophyll metaboing greater representation in photic zone selism; type III secretion systems; and aminosugars, quences associated with alanine and aspartate; purine, proponoate, and vitamin B6 metabolism, metabolism of aminosugars; chlorophyll and relative to deep-water samples (fig. S7). Indepencarotenoid biosynthesis; maltose transport; dent comparisons with well-annotated subsystems lactose degradation; and heavy metal ion senin the SEED database (31) also showed similar sors and exporters. In contrast, samples from and overlapping trends (table S1), including depths of 200 m and below (where there is no greater representation in photic zone sequences photosynthesis) were enriched in different seassociated with alanine and aspartate; metabolism quences, including those associated with proof aminosugars; chlorophyll and carotenoid tein folding; maltose processing and export; biosynthesis; transport; lactose methio degradanine metabolism; glyoxylate, dicarboxylate, tion; and heavy metal ion sensors and exporters. and methane metabolism; thiamine metabo In contrast, samples from depths of 200 m and lism; and type II secretion systems, relative to below (where there is no photosynthesis) were surface-water samples (fig. S7). including those enriched in different sequences, COG categories alsofolding; provided insight into associated with protein processing and differentially distributed protein functions and export; methionine metabolism; glyoxylate, dicarcategories. COGs more highly represented in boxylate, and methane metabolism; thiamine photic zone included iron-transport membrane metabolism; and type II secretion systems, relative receptors, deoxyribopyrimidine to surface-water samples (fig. S7). photolyase, diaminopimelate decarboxylase, membrane COG categories also provided insight into differentially distributed protein functions and guanosine triphosphatase (GTPase) with the categories. COGs moregene highly represented in photic lysyl endopeptidase product LepA, and zone included iron-transport membrane receptors, branched-chain amino acidtransport system
30

25

10 70

Potential temperature (C)

130

Depth (meters)

20 200

15

10 500 5 770

4000 0 34 34.5 35 35.5

Salinity

Fig. 1. Temperature versus salinity (T-S) relations for the North Pacific Subtropical Gyre at station ALOHA (22-45N, 158-W). The blue circles indicate the positions, in T-S hydrospace of the seven water samples analyzed in this study. The data envelope shows the temperature and salinity conditions observed during the period October 1988 to December 2004 emphasizing both the temporal variability of near-surface waters and the relative constancy of deep waters.

components (fig. S8). In contrast, COGs with deoxyribopyrimidine photolyase, diaminopimelate decarboxylase, membrane guanosine triphosphagreater representation in deep-water samples tase (GTPase) with the several lysyl endopeptidase gene included transposases, dehydrogenase product LepA, branched-chain acid categories, and and integrases (fig. S8).amino Sequences transport system components In conmore highly represented in (fig. the S8). deep-water trast, COGs with subsystem greater representation in samples in SEED (31) comparideep-water samples transposases, sevsons included those included associated with respiraeral dehydrogenase categories, and adenosine integrases tory dehydrogenases, polyamine (fig. S8). Sequences more highly represented in triphosphate (ATP)binding cassette (ABC) the deep-water samples in metabolism, SEED subsystem 31) transporters, polyamine and (alcomparisons included those(table associated kylphosphonate transporters S1). with respiratory dehydrogenases, polyamine adenoHabitat-enriched sequences. We estimated sine triphosphate (ATP)binding cassette (ABC) average protein sequence similarities between transporters, polyamine metabolism, and alkylall depth bins from cumulative TBLASTX phosphonate transporters (table S1). high-scoring sequence pair (HSP) bitscores, Habitat-enriched sequences. We estimated derived from BLAST searches of each depth average protein sequence similarities between all against every other (Fig. 3). Neighbor-joining depth bins from cumulative TBLASTX highanalyses of a normalized, distance matrix descoring sequence pair (HSP) bitscores, derived rived from these cumulative bitscores joined from BLAST searches of each depth against photic zone (Fig. and 3). deeper samples together in every other Neighbor-joining analyses separate clusters (Fig. 3). When we compared of a normalized, distance matrix derived from our HOT sequence datasetsjoined to previously rethese cumulative bitscores photic zone ported Sargasso Sea microbial sequences ( 19 ), and deeper samples together in separate clusters these datasets also clustered according to their (Fig. 3). When we compared our HOT sequence depth andto size fraction of origin (fig. S9). The datasets previously reported Sargasso Sea clustering in Fig. is consistent microbial pattern sequences (19),3 these datasets with also the expectation that randomly photic clustered according to their sampled depth and size zone microbial sequences average fraction of origin (fig. will S9). tend Theon clustering to be more another, than those pattern in similar Fig. 3 to isone consistent with to the exfrom the deep-sea, and vice-versa. pectation that randomly sampled photic zone We also identified those (some microbial sequences will tendsequences on average to be of which have homologs in annotated damore similar tono one another, than to those from the deep-sea, vice-versa. tabases) that and track major depth-variable enWe also identified sequencessequence (some of vironmental features. those Specifically, which have found no homologs databases) homologs only in in annotated the photic zone
VOL 311 SCIENCE

unique sequences 10, 70, and 130 m), or that track major (from depth-variable environmental features. Specifically, sequence(from homologs deepwater unique sequences 500, found 770, only4000 in the photic unique(Fig. sequences and m) werezone identified 3). To(from cat10, 70, potential and 130functions m), or deepwater egorize encoded inunique these sequences (from 500, 770, and 4000 m) were photic zone unique (PZ) or deep-water unique identified (Fig. bins, 3). To potential (DW) sequence eachcategorize was compared with functions encoded these photicdatabases zone unique KEGG, COG, and in NCBI protein in (PZ) or analyses deep-water separate (29, unique 30, 36). (DW) sequence bins, each KEGG was compared with pathways KEGG, COG, Some metabolic apand NCBI protein databases in separate analypeared more highly represented in the PZ ses (29 30, 36 ). than in, DW sequence bins, including those Some KEGG metabolic pathways appeared associated with photosynthesis; porphyrin and more highly metabolism; represented inpropanoate, the PZ than purine, in DW chlorophyll sequence bins, including those associated with and glycerphospholipid metabolism; bacte photosynthesis; porphyrin and chlorophyll metabrial chemotaxis; flagellar assembly; and type olism; propanoate, purine, and glycerphospholipid III secretion systems (Fig. 4A). All proteorhometabolism; bacterial chemotaxis; flagellar assemdopsin sequences (except one) were captured bly; and type III secretion systems (Fig. 4A). All in the PZ bin. Well-represented photic zone proteorhodopsin sequences (except one) were KEGG pathway categories appeared to reflect captured in the PZ bin. Well-represented photic potential pathway interdependencies. For zone KEGG pathway categories appeared toexreample the PZ photosynthesis bin [3% of flect potential pathway interdependencies. the For total (Fig. 4A)] contained Prochlorococcus example the PZ photosynthesis bin [3% of the like and Synechococcus -like photosystem total (Fig. 4A)] contained Prochlorococcus-like I, photosystem II,-like and cytochrome In and Synechococcus photosystem genes. I, phototandem, PZ and porphyrin and chlorophyll system II, cytochrome genes. In biosyntandem, thesis sequence bins [3.9% ofbiosynthesis the total (Fig. PZ porphyrin and chlorophyll se4A)] contained high of representation cyano quence bins [3.9% the total (Fig.of 4A)] conbacteria-like cobalamin and chlorophyll biotained high representation of cyanobacteria-like synthesis as well as photoheterotrophcobalamingenes, and chlorophyll biosynthesis genes, as like biosynthetic genes. well bacteriochlorophyll as photoheterotroph-like bacteriochloroOther probable functional interdependencies phyll biosynthetic genes. Other probable funcappear reflected in the corecovery of sequenc tional interdependencies appear reflected in the corecovery of sequences associated with chees associated with chemotaxis (mostly methylmotaxis (mostly methyl-accepting chemotaxis accepting chemotaxis proteins), flagellar bioproteins),(predominantly flagellar biosynthesis (predominantsynthesis flagellar motor and

498

27 JANUARY 2006

www.sciencemag.org

25

RESEARCH ARTICLES

Fig. 2. Taxon distributions of top HSPs. The percent top HSPs that match the taxon categories shown at expectation values of e1 10j60. Values in parentheses indicate number of genomes in each category, complete

or draft, that were in the database at the time of analysis. The dots in the lower panel tabulate the SSU rRNAs detected in fosmid libraries from each taxonomic group at each depth (35) (figs. S3 and S6).

ly flagellar motor and genes), hook protein-encoding hook protein-encoding and type III segenes), pathways and type (all III associated secretory pathways (all cretory with flagellar associated with flagellar biosynthesis) in PZ biosynthesis) in PZ (Fig. 4A). (Fig. 4A). DW sequences were enriched in several

DW sequences enriched glyoxylate in several KEGG categories,were including KEGG categories, including glyoxylate and dicarand dicarboxylate metabolism (with high boxylate metabolism (with high representation representation of isocitrate lyase and formate of isocitrate lyase and formate protein dehydrogenase dehydrogenaselike genes); folding SCIENCE VOL 311

like processing genes); protein folding and processing and (predominantly chaperone(preand dominantly chaperone like genes); protease like genes); and typeprotease II secretory genes type II secretory genes (40% were most sim( 40% were most similar to pilin biosynthesis ilar to pilin biosynthesis genes); aminophosphogenes); aminophosphonate, methionine, and

26

www.sciencemag.org

27 JANUARY 2006

499

RESEARCH ARTICLES
Fig. 3. Habitat-specific sequences in photic zone versus deep-water communities. The dendrogram shows a cluster analysis based on cumulative bitscores derived from reciprocal TBLASTX comparisons between all depths. Only the branching pattern resulting from neighbor-joining analyses (not branch-lengths) are shown in the dendrogram. The Venn diagrams depict the percentage of sequences that were present only in PZ sequences (n 0 12,713) or DW sequences (n 0 14,132), as determined in reciprocal BLAST searches of all sequences in each depth versus every other. The percentage out of the total PZ or DW sequence bins represented in each subset is shown. See SOM for methods (35). mortality in the water column (3740). The large Community Genomics and number of viral DNA sequences in our dataset Host-Virus Interactions was unexpected (Fig. 5; fig. S12), because we expected planktonic viruses to pass compo through Viruses are ubiquitous and abundant our collection filters. Previous studies using a nents of marine plankton, and influence lateral similar approach found only minimal contrigene transfer, genetic diversity, and bacterial butions from sources (19, 40). The majority mortality in viral the water column (3740 ). The of viral DNAof we captured in fosmid in clone large number viral DNA sequences our libraries apparently originates replicatdataset was unexpected (Fig. 5;from fig. S12), being viruses within planktonic infected host cells 35). cause we expected viruses to ( pass Viral DNA was highest in thestudies photic through our recovery collection filters. Previous zone, a with cyanophage-like sequences repreusing similar approach found only minimal senting 1 to 10% of all fosmid sequences contributions from viral sources (19, 40). (Fig. The 5), and 60 toviral 80%DNA of total virus sequences there. majority of we captured in fosmid Belowlibraries 200 m, apparently viral DNAoriginates made up from no more clone repthan 0.3% of all sequences at each licating viruses within infected host cellsdepth. (35). Most photic zone viral sequences Viral DNA recovery was highestshared in thehighest photic similarity to T7-like and T4-like cyanophage of zone, with cyanophage-like sequences reprethe Podoviridae and Myoviridae. This is consenting 1 to 10% of all fosmid sequences (Fig. sistent with previous studies (4042), suggesting 5), and 60 to 80% of total virus sequences a widespread distribution of these phage in the there. Below 200 m, viral DNA made up no ocean. more than 0.3% all sequences at each depth. Analyses of of 1107 fosmid mate pairs proMost photic zone viral sequences shared vided further insight into the origins of thehighviral est similarity to T7-like T4-like cyano sequences. About 67% of and the viruslike clones phage of the Podoviridae and Myoviridae. were most similar to cyanophage on at least one This is consistent with were previous studies (40to end, and half of these highly similar 42 ), suggesting widespread of cyanophage at a both termini. distribution Many of the these phage in the ocean. cyanophage clones showed apparent synteny Analyses of 1107 fosmid cyanophage mate pairs prowith previously sequenced gevided insight into the origins of the nomes further (fig. S12). About 11% of the cyanophage viral sequences. About a67% of the viruslike paired-ends contained host-derived cyanoclones were most similar to)cyanophage on at phage signature gene (43 on one terminus. least one end, and of these were highly The frequency and half genetic-linkage of phagesimilar to(but cyanophage at both termini. Many of encoded host-derived) genes we observed, including virus-derived involved in synphothe cyanophage clones genes showed apparent tosynthesis ( psbA, psbD , hli), cyanophage phosphateteny with previously sequenced scavenging genes (phoH , pstS ), of a the cobalamin genomes (fig. S12). About 11% cyanobiosynthesis gene (cobS ), and carbon metabophage paired-ends contained a host-derived lism (transaldolase ) supports widespread cyanophage signature gene their (43) on one terdistribution in natural and viral populations and minus. The frequency genetic-linkage of their probable functional importance genes to cyanophage-encoded (but host-derived) we phage replication (43, virus-derived 44). observed, including genes inIf we assume that the cyanophages DNA was volved in photosynthesis (psbA, psbD , hli ), derived from infected host cells(phoH in which phage phosphate-scavenging genes , pstS ), a were replicating, the percentage of cyanophagecobalamin biosynthesis gene (cobS ), and carinfected cells was estimated to range between 1 bon metabolism (transaldolase) supports their and 12% (35 ). An apparent cyanophage infecwidespread distribution in natural viral popution maxima was observed at 70 m, coinciding lations and their probable functional imporwith the peak virus:host ratio (Fig. 5). Although tance to cyanophage replication (43, 44). these estimates are tentative, they are consistent If we assume that the cyanophages DNA with previously reported ranges of phagewas derived from infected host cells in which infected picoplankton cells in situ (38, 45). phage were replicating, the percentage of About 0.5% of all sequences were likely cyanophage-infected cells high was sequence estimated to prophage, as inferred from simrange 1 and 12% (35). An apparent ilarity between to phage-related integrases and known cyanophage infection was observed prophage genes (35). maxima Paired-end analyses of at 70 m, coinciding with the peak virus:host viral fosmids indicated that 2.5% may be ratio (Fig. 5). Although these estimates are derived from prophage integrated into a variety tentative, they are consistent with previously of host taxa. A few clones also appear to be reported ranges of phage-infected picoplankderived from temperate siphoviruses, and a ton cellsof in putative situ (38,eukaryotic 45). number paired-end viral About shared 0.5% of all sequences were likely sequences highest sequence identity with prophage, inferred from high sequence simhomologs as from herpes viruses, mimiviruses, ilarity to phage-related integrases and known and algal viruses.

nate, methionine, and sulfur metabolism; butasulfur metabolism; ion-coupled butanoate metabolism; noate metabolism; transporters; ion-coupled transporters; and other and other ABC transporter variants (Fig.ABC 4B). transporter variants (Fig. 4B). The high The high representation in DW sequences of representation insystem DW sequences of type II type II secretion and pilin biosynthesis secretion system and pilin genes, genes, polysaccharide, and biosynthesis antibiotic synthesis polysaccharide, andgreater antibiotic synthesis suggest a potentially role for surfacesuggest a potentially greater role associated microbial processes in for the surfacedeeperassociated microbialConversely, processes in the deeperwater communities. enrichment of water communities. Conversely, enrichment bacterial motility and chemotaxis sequences of in the photic zone indicates a potentially greater bacterial motility and chemotaxis sequences in importance for mobility response in these the photic zone indicatesand a potentially greater assemblages. importance for mobility and response in these Similar differential patterns of sequence assemblages. distribution seen inpatterns COG categories (Fig. Similar were differential of sequence 4B). COGs enriched in the PZ sequence bin distribution were seen in COG categories included photolyases, iron-transport outer mem(Fig. 4B). COGs enriched in the PZ sequence brane proteins, Na-driven iron-transport efflux pumps, ABCbin included photolyases, outer type sugar-transport systems, hydrolases and membrane proteins, Na+-driven efflux pumps, acyl transferases, and transaldolases. In hydrodeeper ABC-type sugar-transport systems, waters, transposases were theand most enriched lases and acyl transferases, transaldolCOG category ( 4.5% of the COG-categorized ases. In deeper waters, transposases were the DW), enriched increasing steadily in representation most COG category (4.5% ofwith the depth from 500 m to their observed COG-categorized DW), increasingmaximum steadily at 4000 m (Fig. 4B; fig. S9). Transposases in representation with depth from 500 m to represented one of the single-most overrepretheir observed maximum at 4000 m (Fig. sented COG categories in deep waters, ac4B; fig. S9). Transposases represented one counting for 1.2% of all fosmids sequenced of the single-most overrepresented COG catfrom 4000 m (fig. S8). Preliminary analyses egories in deep waters, accounting for 1.2% of the transposase variants and mate-pair seof all fosmids sequenced from 4000 m (fig. quences indicate that they represent a wide S8). Preliminary analyses of thefamilies transposase variety of different transposase and variants and mate-pair sequences originate from diverse microbial taxa.indicate In conthat represent a wide variety of differtrast,they other highly represented COG categories ent transposase families and originate from appeared to reflect specific taxon distribution diverse microbial In contrast, other and abundances. Fortaxa. example, the enrichment highly represented COG categories of transaldolases at 70 m (Fig. 4B; fig. appeared S9) were to reflect specific taxon distribution and mostly derived from abundant cyanophage DNA abundances. For example, enrichment of that was recovered at that the depth (see discustransaldolases sion below). at 70 m (Fig. 4B; fig. S9) were

Sargasso Sea surface-water microbial semostly fromasabundant quences derived (19) shared, expected, cyanophage many more DNA that was recovered at our that photic depth zone (see homologous sequences with discussion below). sequences than those from the deep sea (fig. Sargasso Sea surface-water microbial seS10). There were 10 times as many PZ than quences (19) shared, as in expected, many DW sequences shared common withmore Sarhomologous sequences with 7 our photic zone gasso Sea samples 5 through (19 ) (fig. S10). sequences those from the deep sea three (fig. In contrast,than PZ-like sequences were only S10). 10 when timescompared as many PZ than times There higher were in DW with seDW sequences shared Sea in common with S10). Sarquences from Sargasso sample 3 (fig. The fact that Sargasso sample was collected gasso Sea samples 5 through 73 (19 ) (fig. S10). during a period of sequences winter deep-water mixing In contrast, PZ-like were only three likely higher contributes to when this higher representation times in DW compared with seof DW-like homologs. Sargasso Sea homologs quences from Sargasso Sea sample 3 (fig. S10). of our PZ sequence bin included, as expected, The fact that Sargasso sample 3 was collected sequences associated with photosynthesis; amiduring a period of winter deep-water mixing no acid transport; purine, pyrimidine and likely contributes to this higher representation nitrogen metabolism; porphyrin and chloroof DW-like homologs. Sargasso Sea homologs phyll phosphorylation; of our metabolism; PZ sequence oxidative bin included, as expected, glycolysis; and starch and sucrose metabolism sequences associated with photosynthesis; (fig. S10). amino acid transport; purine, pyrimidine and Tentative taxonomic assignments PZ or nitrogen metabolism; porphyrin andofchloro DW sequences (topoxidative HSPs from NCBIs nonrephyll metabolism; phosphorylation; dundant protein database) were also tabulated glycolysis; and starch and sucrose metabolism (fig. S11). As expected, a high percentage of (fig. S10). Prochlorococcus-like sequences was found in Tentative taxonomic assignments of PZ or PZ (5% of the total), and a greater represenDW sequences (top HSPs from NCBIs nontation of Deltaproteobacteria-like, Actinobacteriaredundant protein database) were also tabulike and Planctomycete-like sequences were lated (fig. in S11). expected, a the high percent recovered DW.As Unexpectedly, single most age ofrepresented Prochlorococcuslike sequences was highly taxon category in PZ (21% found in PZ (5% of the total), and a derived greater of all identified sequences in PZ) was representation of Deltaproteobacteria-like, from viral sequences that were captured in Actinobacteria -like and Planctomycete-like fosmid clones (fig. S11). sequences were recovered in DW. Unexpectedly, the single most Community Genomics and highly represented taxon category in PZ (21% of all identiHost-Virus Interactions fied sequences in PZ) was viViruses are ubiquitous and derived abundantfrom comporal sequences were captured in fosmid nents of marinethat plankton, and influence lateral clones (fig. S11). gene transfer, genetic diversity, and bacterial

500

27 JANUARY 2006

VOL 311

SCIENCE

www.sciencemag.org

27

RESEARCH ARTICLES

Fig. 4. Cluster analyses of KEGG and COG annotated PZ and DW sequence bins versus depth. Sequence homologs unique to or shared within the photic zone (10, 70, and 130 m) and those unique to or shared in DW (500, 770, and 4000 m) were annotated against the KEGG or COG databases with TBLASTX with an expectation threshold of 1 10j5. Yellow shading is proportional to the percentage of categorized sequences in each category. Cluster analyses of gene categories (left dendrograms) were performed with the Kendalls tau nonparametric distance metric, and the Pearson correlation was used to generate the top dendrograms relating the depth series (33, 34). Dendrograms were displayed by using self-organizing mapping with the Pearson correlation metric (33, 34). Green lines in top dendrograms show PZ sequences, blue lines DW sequences. (A) KEGG category representation versus depth. KEGG categories with a standard deviation greater than 0.4 of observed values, having at least two depths R0.6% of the total KEGG-categorized genes at each depth, are shown. For display purposes, categories 98% in more than two depths are not shown. (B) COG category representation versus depth. COG categories with standard deviations greater than 0.2 of observed values, having at least two depths R0.3% of the total COG-categorized genes at each depth, are shown.

prophage genes (35). Paired-end analyses of Ecological Implications and viral fosmids indicated that 2.5% may be deFuture Prospects rived from community prophage integrated a variety Microbial samplinginto along wellof host taxa. A few clones also appear to be characterized depth strata allowed us to identify derived from temperatetrends siphoviruses, and a significant depth-variable in gene content number of putative eukaryotic paired-end viand metabolic pathway components of oceanic ral sequences shared highest sequence identity microbial communities. The gene repertoire of with homologs from herpes viruses, mimivisurface waters reflected some of the mechanisms ruses, and algal viruses. processes and primary and modes of light-driven productivity. Environmentally diagnostic sequences Ecological Implications in surface waters includedand predicted proteins asFuture Prospects sociated with cyanophage, motility, chemotaxMicrobial community sampling along wellis, photosynthesis, proteorhodopsins, photolyases, characterized depth strata allowed us systems, to idencarotenoid biosynthesis, iron-transport tify host significant depth-variable trends in gene and restriction-modification systems. The content andofmetabolic pathway components importance light energy to these communities as reflected in their gene content was obof oceanic microbial communities. The gene vious. More ecophysiological trends can repertoire ofsubtle surface waters reflected some of be in iron transport, vitamin synthesis, the seen mechanisms and modes of light-driven flagella synthesis and secretion, and chemotaxis processes and primary productivity. Envigene distributions. These sequences data support ronmentally diagnostic inhypothesurface

waters included predicted proteins associses about potential adaptive strategies of hetated with cyanophage, motility, chemotaxis, erotrophic bacteria in the photic zone that may photosynthesis, photolyactively compete proteorhodopsins, for nutrients by swimming ases, carotenoid biosynthesis, iron-transport toward nutrient-rich particles and algae (46). In systems, toand host restriction-modification contrast surface-water assemblages, deepsystems. The importance of light energy to water microbial communities appeared more these communities as reflected in their gene enriched in transposases, pilus synthesis, protein content polysaccharide was obvious. and More subtle ecophysiexport, antibiotic synthesis, ological trends can be seen iron transport, the glyoxylate cycle, and ureain metabolism gene vitamin synthesis, flagella synthesis in and sesequences. The observed enrichment pilus, cretion, and chemotaxis gene distributions. polysaccharide, and antibiotic synthesis genes These data support hypotheses about potential in deeper-water samples suggests a potentially adaptive strategies of heterotrophic greater role for a surface-attached life bacteria style in in the photic zone that may actively compete deeper-water microbial communities. Finally, for apparent nutrients enrichment by swimming towardgenes nutrientthe of phage and restriction-modification rich particles and algae systems (46). Inobserved contrast in to the photic zoneassemblages, may indicate a greater rolemifor surface-water deep-water phage in the more productive upper crobialparasites communities appeared more enriched water column, relative tosynthesis, deeper waters. in transposases, pilus protein exAtpolysaccharide finer scales, sequence distributions we port, and antibiotic synthesis, observed also reflected genomic microvarithe glyoxylate cycle, and urea metabolism
SCIENCE VOL 311

gene sequences. The observed enrichment in ability along environmental gradients, as pilus, polysaccharide, and antibiotic syntheevidenced by the partitioning of high- and lowsis genes in deeper-water samples suggests a light Prochlorococcus ecotype genes observed potentially greater role for a surface-attached in different regions of the photic zone (Fig. 5). life style in deeper-water microbial communiHigher-order biological interactions were also ties. Finally, the apparent enrichment of phage evident, for example in the negative correlation genes and restriction-modification systems of cyanophage versus Prochlorococcus host gene observed recovery in the photic mayrelation indicate a sequence (Fig. zone 5). This begreaterthe role for phage the more tween abundance ofparasites host and in cyanophage productive upper water column, relative of to DNA probably reflects specific mechanisms deeper waters. cyanophage replication in situ. These host-parasite At finer scales, sequence distributions we sequence correlations we saw demonstrate the poobserved reflected genomic microvar tential for also observing community-level interspeiability along through environmental gradients, as cies interactions environmental genomic evidenced by the partitioning of high- and datasets. Obviously, the abundance of specific low-light Prochlorococcus ecotype genes taxa obwill greatly influence the gene observed in different regions of distributions the photic zone served, as Higher-order we saw, for example, in Prochlorococ(Fig. 5). biological interactions cus gene the photic zone. Gene were alsodistribution evident, forin example in the negative sequence distributions can reflect more than just correlation of cyanophage versus Prochlororelative abundance of specific taxa, however. coccus host gene sequence recovery (Fig. 5).

28

www.sciencemag.org

27 JANUARY 2006

501

RESEARCH ARTICLES

This relation between the abundance of host Percent total sequences per depth enabled ecological studies matures, it should Fig. 5. Cyanophage and cyanobacteria disand cyanophage DNA probably reflects spetributions in microbial community DNA. The become possible to model microbial community 0 10 20 cific mechanisms replication percentage of of cyanophage total sequences derived in from genomic, temporal, and spatial variability with 0 situ. These host-parasite sequence correlations cyanophage, total cyanobacteria, total Prochlorother environmental features. Significant future we sawococcus demonstrate the potential for observspp., high-light Prochlorococcus , lowattention will no doubt focus on interpreting ing community-level interspecies light Prochlorococcus spp., orinteractions Synechococcus the complex interplay between genes, orgathroughspp., environmental genomic datasets. from each depth. Taxa were tentatively nisms, communities and the environment, as Obviously, abundance of specific assignedthe according to the origin of top taxa HSPs in well as the properties revealed that regulate will greatly influence the gene distributions TBLASTX searches, followed by subsequent global biogeochemical cycles. Future efforts manual inspection curation. observed, as we saw, forand example, in Prochloin this area will advance our general perspective 100 rococcus gene distribution in the photic zone. on microbial ecology and evolution and eluGene sequence distributions can reflect more cidate the biological dynamics that mediate RESEARCH ARTICLES than just relative abundance of specific taxa, the flux of matter and energy in the worlds however. Some depth-specific gene distribuoceans. Fig. 5. Cyanophage and cyanobacteria distions we observed [e.g., transposases found tributions in microbial community DNA. The predominantly at greater depths (Fig. 4B; fig. percentage References of total sequences derived from and Notes 0 S8)], appear to originate from a wide variety 200 1. E.total Forbes, in Physical Atlas of ProchlorNatural Phenomena, cyanophage, cyanobacteria, total of gene families and genomic sources. These A. K. Johnston, Prochlorococcus Ed. (William Blackwood & Sons, London ococcus spp., high-light , lowgene distributional patterns seem more indicaand Edinburgh, 1856). light Prochlorococcus spp., or Synechococcus 2. P. W. Hochachka, G. N. Somero, Biochemical Adaptation tive of habitat-specific genetic or physiologispp., from each depth. Taxa were tentatively (Princeton Univ. Press, Princeton, NJ, 1984), cal trends that have spread through different assigned according to the origin of top HSPs in pp. 450495. members of the community. Community gene TBLASTX3.searches, followed subsequent G. Rocap et al., Natureby 424 , 1042 (2003). distributions and stoichiometries are differenmanual inspection and curation. 4. Z. I. Johnson et al ., manuscript submitted. RESEARCH ARTICLES 5. N. J. West et al., Microbiology 147, 1731 (2001). 100 tially propagated by vertical and horizontal 300 6. A. A. Yayanos, Annu. Rev. Microbiol. 49, 777 (1995). genetic mechanisms, physiological Percent total sequences per depth enabled ecological studies matures, it should Fig. 5. Cyanophage anddynamic cyanobacteria dis7. N. R. Pace, Science 276, 734 (1997). responses, or interspecies interactions tributions in microbial community DNA. The like become possible to model community 8. M. S. Rappe , S.microbial J. Giovannoni, Annu. Rev. Microbiol. 57, 0 10 20 percentage of The total overrepresentation sequences derived from competition. of certain genomic, temporal, and spatial variability with 369 (2003). 0 9. M. T. Suzuki et al., Microb Ecol (2005). cyanophage, total cyanobacteria, total Prochlorsequence types may sometimes reflect their other environmental features. Significant future 10. O. Be ja ` et al., focus Environ.on Microbiol. 2, 516 (2000). ococcus spp.,transmission high-light Prochlorococcus , low- withhorizontal and propagation no doubt interpreting cyanophages attention will 11. R. M. Morris, M. S. Rappe , E. Urbach, S. A. Connon, light Prochlorococcus spp., Synechococcus interplay between genes, orga- 70, 2836 in a given community. Inor our datasets, the relaJ. Giovannoni, Appl. Environ. Microbiol. cyanobacteria the complex S. 200 spp., from each depth. Taxa were tentatively 400 nisms, communities tive abundance of cyanobacteria-like psbA, (2004). and the environment, as assigned according to the origin of top HSPslargely in Prochlorococcus 12.properties S. Y. Moon-van der Staay that et al.regulate , Nature 409, 607 well as the revealed psbD, and transaldolase genes were a TBLASTX searches, followed by subsequent (2001). global biogeochemical cycles. Future efforts consequence of their horizontal transfer and Synechococcus manual inspection and curation. J. A. Fuhrman, K. McCallum, A. A. Davis, Nature 356, 148 in this area13. will advance our general perspective subsequent amplification in the viruses that 100 (1992). HL Prochlorococcus on microbial ecology and evolution elu- 89, 5685 were captured in our samples. In contrast, the 14. E. F. DeLong, Proc. Natl. Acad. and Sci. U.S.A. cidate the biological dynamics that mediate LL Prochlorococcus (1992). increase of transposases from 500 to 4000 15. M. B. Karner et al., Nature 409,worlds 507 (2001). the flux of matter and energy in the 300 m, regardless of community composition, re16. J. Handelsman, Microbiol. Mol. Biol. Rev. 68, 669 500 oceans. flected a different mode of gene propagation, (2004). likely related to the slower growth, lower pro17. E. F. Delong, Nat Rev Microbiol (2005). 18. G. W. Tyson et al., Nature 428, 37 (2004). ductivity, and lower effective population sizes References and Notes 19. J. C. Venter et of al.Natural , Science 304, 66 (2004). 200ob- productivity, and lower effective population 1. E. Forbes, in Physical Atlas Phenomena , Some depth-specific gene distributions of deep-sea microbial communities. In future we 20. S. Ed. G. Tringe et al. , Science&308 , 554 (2005). A. K. Johnston, (William Blackwood Sons, London sizes of deep-sea microbial communities. In served [e.g., transposases found in predominantly comparative studies, similar deviations en21. S. J. Hallam et al., Science 305, 1457 (2004). and Edinburgh, 1856). studies, similar deviations at greater (Fig. 4B; fig. S8)],be appear to future comparative vironmental genedepths stoichiometries might environmental processes are necessary (48). 22. D. M. R. Lukas, Deep-SeaAdaptation Res. II 43, 129 2. P. W. Hochachka, G.Karl, N. Somero, Biochemical 400 (1996). in environmental gene stoichiometries might originate from a wide variety of gene (Princeton Univ. Press, Princeton, NJ, 1984), expected to provide even further insight into families As the scope and scale of genome-enabled 23. D. M. Karl et al. , Deep-Sea Res. II 48 , 1449 (2001). pp. 450495. be studies expected to provide even further insight and genomic sources. gene distributional habitat-specific modes and These mechanisms of ecological matures, it should become R., M. Letelier al. , Limnol. 3. G. Rocap24. et al. Nature 424,et 1042 (2003). Oceanogr. 49 , 508 Percent total sequences per depth enabled ecological studies matures, it and should into habitat-specific modes mechanisms patterns seem more indicative of habitat-specific gene propagation, distribution, and mobility possible to model microbial community ge4. Z. I. Johnson (2004). et al., manuscript submitted. to propagation, model microbial community of gene distribution, and mobilgenetic physiological trends that have become spread nomic,possible temporal, and spatial variability with (27 , 47). Theseor gene ecologies could read0 10 20 25. DeLong et al. , Appl. Environ. Microbiol. 65, 5554 5. N. J. West et E. al.,F.Microbiology 147 , 1731 (2001). 300 genomic, temporal, and spatial variability with (1999). ity (27, 47 ). These gene ecologies could throughdirectly different of distributhe community. 0 be mapped 6. A. A. Yayanos, Annu. Rev. Microbiol. 49, 777 (1995). ily onmembers organismal other environmental features. Significant other environmental features. Significant future 26.Science A. Pernthaler et al. , Appl. Environ. Microbiol. 68, 661 7. N. R. Pace, 276, 734 (1997). readily will be mapped directly on organismal Community gene distributionsvariabiland stoichiomefuture attention no doubt focus on intertions and interactions, environmental 8. M. S. Rappe , (2002). S. J. Giovannoni, Annu. Rev. Microbiol. 57, attention will no doubt focus on interpreting distributions and interactions, environmental tries are differentially propagated by vertical and ity, and taxonomic distributions. preting the complex interplay between genes, 27. N. U. Frigaard et al., Nature, in press. 369 (2003). 500 complex interplay between genes, orgavariability, and taxonomic mechanisms, dynamic the physiThe horizontal study ofgenetic environmental adaptation organisms, communities and the distributions. environ28. et S. al M. F. Delwiche, 9. M. T. Suzuki ., Barns, MicrobC. Ecol (2005). J. D. Palmer, N. R. Pace, Proc. nisms, communities and environment, as The study ofthe environmental adaptation ological is responses, interspecies interactions Acad. Sci. U.S.A. , 9188 (1996). O. Be ja ` et al.Natl. , Environ. Microbiol. 2, 93 516 (2000). ment, as well as the properties revealed that 10.and and variability not new, or but our technical cyanophages wellof as the properties revealed that our regulate 29. M. et Urbach, al., Nucleic Res. 32 , D277 11. R. M. Morris, M. Kanehisa S. Rappe , E. S. A. Acids Connon, variability is not new, but technical capalike competition. The overrepresentation capabilities for identifying and tracking se- regulate global biogeochemical cycles. Future S. J. Giovannoni, Appl. Environ. Microbiol. 70, 2836we ob(2004). cyanobacteria global biogeochemical cycles. Future efforts sequences, prod Some depth-specific gene distributions bilities for identifying and tracking certain sequence types may sometimes reflect efforts in this area will advance our general quences, genes, and metabolic pathways in 400 (2004). 30. R. L.transposases Tatusov et al., BMC Bioinformatics 4, 41 (2003). in this area will advance our general perspective size served [e.g., found predominantly 100 genes, and metabolic pathways in microbial their horizontal transmission and propagation Prochlorococcus 12. S. Y. Moon-van Staay et , ,Nature 409 , 607 microbial communities is. The study of gene on perspective on microbial ecology and evolu31. R. der Overbeek etal. al. Nucleic Acids Res. 33, 5691 microbial ecology and evolution and eluat greater (2005). depths (Fig. 4B; fig. S8)], appear to futu communities is. The study of gene ecology and within a given community. In our datasets, the (2001). ecology and its relation to community metab- cidate tion and elucidate thedynamics biological dynamics that Synechococcus biological that mediate originate a wide variety of gene 13. J. A. Fuhrman, A.al. A., Davis, Nature 356,,families 148 (1990). in e 32. from S.K. F.McCallum, Altschul et J. Mol. Biol. 215 403 relation to community metabolism, interrelative abundance of cyanobacteria-like psbA , theits mediate the flux of matter and energy in the olism, interspecies interactions, and habitat- the (1992). flux of matter and energy in the worlds and genomic These distributional 33. M. J.sources. L. de Hoon et al., gene Bioinformatics 12, 1453 be HL Prochlorococcus species interactions, and habitat-specific signapsbD , and transaldolase genes were largely a worlds oceans. specific signatures is nascent. More extensive oceans. 14. patterns E. F. DeLong, Proc. Natl. Acad. Sci. U.S.A. 89, 5685 (2004). into seem more indicative of habitat-specific More extensive sequencing consequence of certainly their horizontal transfer and tures is nascent. LL Prochlorococcus (1992). 34. M. B. Eisen, P. T. Spellman, P. O. Brown, Proc. Natl. Acad. sequencing efforts are required to of g genetic or physiological trends that have spread subsequentdescribe amplification inmicrobial the viruses that efforts are certainly required to more thoroughly 15. M. B. Karner Sci. et al. , Nature , 507 (2001). U.S.A. 94,409 14863 (1998). more thoroughly natural through different members community. 16. J. Handelsman, Microbiol. Mol. Biol. of Rev. 68, 669 35. Materials and methods are the available as supporting ity describe natural microbial communities. Addi500 were captured in our samples. In contrast, the References and Notes communities. Additionally, more concerted (2004). read Community geneon distributions 200 material Science Online.and stoichiome1. m, E. Forbes, in Physical Atlasconcerted of Natural Phenomena , integrate these tionally, more efforts to increase of transposases from 500 to 4000 efforts to integrate these new data into stud17. tries E. F. Delong, Nat Microbiol (2005). 36. K. D.Rev Pruitt, T. propagated Tatusova, D. R. by Maglott, Nucleic Acids dist Res. are differentially vertical and A. K. Johnston, Ed. (William Blackwood & Sons, London new data into studies of oceanographic, bioof community composition, re18. G. W. Tyson et al. Nature 428, 37 (2004). 33 , ,D501 (2005). ies of regardless oceanographic, biogeochemical, and and Edinburgh, 1856). vari horizontal genetic mechanisms, dynamic physi19. are J. C. Venter al.G. , Science 304, FEMS 66 (2004). geochemical, and environmental processes flected a different mode of gene 37.etM. Weinbauer, Microbiol. Rev. 28, 127 2. P. W. Hochachka, G. N. Somero, Biochemical Adaptation and lower effective population Some depth-specific gene distributions we propagation, ob- productivity, T interspecies 20. ological S. G. Tringe responses, et al., Science or 308 , 554 (2005). interactions (Princeton Univ. Press, Princeton, NJ, 1984), (2004). necessary ( 48 ). As the scope and scale of genomelikely related to the slower growth, lower served [e.g., transposases found predominantly sizes of deep-sea microbial communities. In vari like competition. The overrepresentation of 21. S. J. Hallam et al. , Science 305 , 1457 (2004). pp. 450495. 29 comparative studies, similar deviations at greater depths (Fig. 4B; fig. S8)], appear to future 22. certain D. M. Karl, R. Lukas, Deep-Sea II 43 , 129 sequence types Res. may sometimes reflect bilit 3. G. Rocap et al., Nature 424, 1042 (2003). (1996). gene stoichiometries might originate from a wide variety of gene families in 4. environmental Z. I. Johnson et al., manuscript submitted.
Depth (m)
Depth (m) Depth (m)

Depth (m)

is nascent. More Mol. extensive sequencing er and tures 51. This work was supported by a grant from the Gordon and and a 16. J. Handelsman, Microbiol. Biol. Rev. 68 , 669 34. (2005). M. B. Eisen, P. T. Spellman, P. O. Brown, Proc. Natl. Acad. are certainly required to more thoroughly 41. J.Sci. es that efforts Betty Moore Foundation, NSF grants MCB-0084211, Filee, F. Tetart, C. A. Suttle, corre (2004). U.S.A. 94, 14863 (1998).H. M. Krisch, Proc. Natl. MCB-0348001, and MCB-0509923 to E.F.D., and Sci. and U.S.A. 102, 12471 (2005).as supporting 17. E. F. Delong, Natmicrobial Rev Microbiol (2005). 35. Acad. Materials methods are available natural communities. Addiast, the describe Supporti OCE-0326616 to D.M.K., and sequencing support from Breitbart, J. H. Miyake, F. Rohwer, FEMS Microbiol. 18. G. W. more Tyson et al., Nature efforts 428, 37 to (2004). material on Science Online. concerted integrate these 42. M. 000 m, tionally, www.scienc Lett. 236 , 249 (2004). the U.S. Department of Energy Microbial Genomics 19. J. C. Venter et al. , Science 304 , 66 (2004). 36. K. D. Pruitt, T. Tatusova, D. R. Maglott, Nucleic Acids Res. ulation Materials a into studies oceanographic, bio- 43. M. on, re- new Sullivan et al., PLoS Biol. 3, e144 (2005). Program. We thank Dennis Ryan for help with scripting 20. S. data G. Tringe et al. , Scienceof 308 , 554 (2005). 33B. , D501 (2005). Figs. S1 to ies. In geochemical, and are 44. gation, et al., Proc. Natl. Acad. Sci.Rev. U.S.A. , 11013 and sequence analyses, the officers and crew of the R/V 21. S. J. Hallam et al., environmental Science 305, 1457 processes (2004). 37. D. M.Lindell G. Weinbauer, FEMS Microbiol. 28101 , 127 Table S1 iations (2004). Kaimikai-O-Kanaloa and the HOT team for assistance 22. D. M. Karl, R. As Lukas, Res.scale II 43, of 129 (2004). (48). theDeep-Sea scope and genomelower necessary References assis 38. M. J. Waterbury, F. Valois, Appl. M. Environ. Microbiol. 59, 3393 47. at C. sea, R. Woese, Mol. Biol. Rev. 68, 173 (2004). 45. G. Weinbauer, I. Brettar, G. Hofle, Limnol. and L.Microbiol. Fujieki for help with oceanographic data (1996). might RESEARCH ARTICLES have (1993). 48, 1457 (2003). 48. display. E. F. DeLong, D. M. Karl, Nature 437 , 336 (2005). Oceanogr. 16 Septem J. Chapman, A. Salamov, and P. Richardson of 23. D. M. Karl et al. , Deep-Sea Res. II 48 , 1449 (2001). insight num 39. F. M. B. Sullivan et 280 al., ,Nature 424, 1047 (2003). 49. the (http://hahana.soest.hawaii.edu/hot/parameters.html). 46. Azam, Science 694 (1998). 10.1126/sc DOE Joint Genome Institute provided advice and 24. R. M. Letelier et al. , Limnol. Oceanogr. 49 , 508 anisms JANUARY 2006 VOL 311 Appl. SCIENCE www.sciencemag.org DU8 40. C. R. R. A. Woese, Edwards, F. Rohwer, Nat. Rev. Microbiol. 3,(2004). 504 50. assistance (http://hahana.soest.hawaii.edu/hot/hot-dogs). 47. Microbiol. Mol. Biol. Rev. 68, 173 38. J. Waterbury, F. Valois, Environ. Microbiol. 59, 3393 in DNA sequencing and analyses. Sequences (2004). mobiland (2005). 51. This was supported a grantwith fromaccession the Gordon and 48. E. F. DeLong, D. M. Karl, Nature 437, 336 (2005). havework been deposited in by GenBank 25. (1993). E. F. DeLong et al., Appl. Environ. Microbiol. 65, 5554 corre 41. (http://hahana.soest.hawaii.edu/hot/parameters.html). J. Filee, F. Tetart, C. A. Suttle, H. M. Krisch, Proc. Natl. Betty Moore Foundation, NSF grants 49. 39. M. B. Sullivan et al., Nature 424, 1047 (2003). numbers DU731018-DU796676 andMCB-0084211, DU800850(1999). could Acad. Sci. U.S.A. 102, 12471 (2005). MCB-0348001, and MCB-0509923 toend E.F.D., and ARTICLES 50. (http://hahana.soest.hawaii.edu/hot/hot-dogs). 40. Edwards,et F.al. Rohwer, Rev. Microbiol. Microbiol. 68 3, ,504 DU800864 corresponding toRESEARCH fosmid sequences, 26. R. A. A. Pernthaler , Appl. Nat. Environ. 661 nismal Supporti 42. This M. Breitbart, H. Miyake, FEMS OCE-0326616 to D.M.K., and sequencing support from 51. work wasJ. supported byF. a Rohwer, grant from theMicrobiol. Gordon and (2005). and accession numbers DQ300508-DQ300926 (2002). www.scien mental Lett. , Woese, 249 (2004). the U.S. Department of Energy Genomics Betty Moore Foundation, NSFMol. grants MCB-0084211, 41. Filee, F. Tetart, Suttle, H. press. M. Krisch, Proc. Natl. corresponding to SSU rRNA geneMicrobial sequences. assistance in DNA sequencing and analyses. Sequences 47. 236 C. R. Microbiol. Biol. Rev. 68, 173 (2004). 38. Waterbury, F. A. Valois, Appl. Microbiol. 59, 3393 439 ,Environ. 847 (2006). 27. J. N. U.J.Frigaard et C. al. , Nature, in Materials a s. 43. MCB-0348001, M. B.E. Sullivan etand al. ,MCB-0509923 PLoS Biol. 3, e144 (2005). Program. thank Dennisin Ryan for help with scripting to 437 E.F.D., and Sci. U.S.A. , 12471 (2005). have We been deposited GenBank with accession 48. F. DeLong, D. M. Karl, Nature , 336 (2005). 28. Acad. S. M. (1993). Barns, C. F. 102 Delwiche, J. D. Palmer, N. R. Pace, Proc. Figs. S1 to Supporting Online Material 44. D. Lindell et al. , Proc. Natl. Acad. Sci. U.S.A. 101 , 11013 and sequence analyses, the officers and crew of the R/V OCE-0326616 to D.M.K., and sequencing support from 42. M. Breitbart, J. H. Miyake, F. Rohwer, FEMS Microbiol. on and numbers DU731018-DU796676 and DU80085049. (http://hahana.soest.hawaii.edu/hot/parameters.html). 39. M. B. Sullivan et al. , Nature 424 , 1047 (2003). Natl. Acad. Sci. U.S.A. 93, 9188 (1996). m Table S1 www.sciencemag.org/cgi/content/full/311/5760/496/DC1 F 0 (2004). Kaimikai-O-Kanaloa and the HOT team for assistance the U.S. Department of Energy Microbial Genomics 236 , 249 DU800864 corresponding to fosmid end sequences, 50. (http://hahana.soest.hawaii.edu/hot/hot-dogs). 40.Kanehisa R. A. Edwards, Rohwer, Nat. Res. Rev. 32 Microbiol. 29. Lett. M. et(2004). al., F. Nucleic Acids , D277 3, 504 capaReferences Materials and Methods total spin 45. M. G. Weinbauer, I. Brettar, M. G. Hofle, Limnol. at sea, and L. Fujieki for help with oceanographic data Program. We thank Dennis Ryan help with scripting 43. M. B.(2005). Sullivan et al., PLoS Biol. 3, e144 (2005). and accession numbers DQ300508-DQ300926 51. This work was supported by afor grant from the Gordon and (2004). uences, Figs. S1 to S12 tion) is RESEARCH ARTICLES Oceanogr. , analyses, 1457 (2003). display. J. Chapman, Salamov, and sequences. P. Richardson of 16 Septem and sequence the officers and crew of the R/V 44. Lindell et al. Proc. Natl. Sci. 101 ,Proc. 11013 corresponding to A. SSU rRNA gene Betty 48 Moore Foundation, NSF grants MCB-0084211, 41. Filee, F.,Tetart, C. Bioinformatics A. Acad. Suttle, H. U.S.A. M. Natl. 30. D. R. L. J. Tatusov et al. , BMC 4,Krisch, 41 (2003). Table the S1 DOE Joint Genome Institute provided advice and crobial 46. Kaimikai-O-Kanaloa F. Azam, Science 280and ,and 694 (1998). 10.1126/s the HOT team to forE.F.D., assistance MCB-0348001, MCB-0509923 and Acad. Sci. 102, 12471 31. (2004). R. Overbeek etU.S.A. al., Nucleic Acids (2005). Res. 33 , 5691 where th References and Notes Supporting Online Material gy and and L. Fujieki help with oceanographic data from 45. G.M. Weinbauer, I. M. G. Limnol. tofor D.M.K., and sequencing support 42. Breitbart, J. Brettar, H. Miyake, F. Hofle, Rohwer, FEMS Microbiol. assistance in DNA sequencing and analyses. Sequences 38. M. J. Waterbury, F. Valois, Appl. Environ. Microbiol. 59 , 3393 47. at C. sea, R. OCE-0326616 Woese, Microbiol. Mol. Biol. Rev. 68, 173 (2004). (2005). be contr www.sciencemag.org/cgi/content/full/311/5760/496/DC1 Chapman, A. Salamov, and P. Richardson of 48, (2003). 16 September 2005; accepted 21 December the J. U.S. Department of Energy Genomics 236 , 1457 249 (2004). have been deposited in GenBank with 2005 accession (1993). 48. display. E. F. DeLong, D. M. Karl, Nature 437 ,Microbial 336 (2005). 32. Oceanogr. S. F. Lett. Altschul et al., J. Mol. Biol. 215, 403 (1990). intermixture Materials and Methods DOE Joint Genome Institute provided advice and 46. Azam, Science , 694 (1998). 10.1126/science.1120250 Program. We thank Dennis Ryan for help with scripting 43. B. Sullivan al. , PLoS Biol. 3, 12 e144 (2005). numbers DU731018-DU796676 and DU80085039. M. B. Sullivan et 280 al. ,al. Nature 424 , 1047 (2003). 49. the (http://hahana.soest.hawaii.edu/hot/parameters.html). 33. F. J. M. L. de Hoon et et , Bioinformatics , 1453 signaFigs. S1 to S12 Randall G. Hulet* which is Guthrie B. Partridge, Wenhui Li, Ramsey Liao, and sequence analyses, the officers and crewI. of Kamar, the R/V Yean-an 44. Lindell F. et Rohwer, al., Proc. Natl. Acad. Sci. U.S.A. 101, 11013 DU800864 corresponding to fosmid end sequences, 40. R. A. D. Edwards, Nat. Rev. Microbiol. 3, 504 50. (http://hahana.soest.hawaii.edu/hot/hot-dogs). (2004). Table S1 encing Kaimikai-O-Kanaloa the from HOT team for assistance and accession numbers DQ300508-DQ300926 51. This work was supported byand a grant the Gordon and located n 34. (2005). M. B.(2004). Eisen, P. T. Spellman, P. O. Brown, Proc. Natl. Acad. References andto Notes oughly sea, and L. Fujieki forgrants help with oceanographic data 45. M. F. G. Tetart, Weinbauer, I. Brettar, G. Hofle, Limnol. corresponding SSU rRNA gene sequences. 41. J. Filee, C. A. (1998). Suttle, H. M. Krisch, Proc. Natl. Bettyat Moore Foundation, NSF MCB-0084211, Sci. U.S.A. 94 , 14863 evaporat We report the observation of pairing in a gas of atomic fermions with unequal numbers of two display. J. Chapman, A. Salamov, and P.and Richardson of Oceanogr. 48 , 1457 16 September 2005; accepted 21 December 2005 Sci. and U.S.A. 102 , 12471 (2005). MCB-0348001, and MCB-0509923 to E.F.D., 35. Acad. Materials methods are(2003). available as supporting Addioptic Supporting Online Material components. Beyond a critical polarization, the gas separates into a phase that is consistent with a the the DOE Joint Genome Institute provided advice and 46. F. Azam, 280,F.694 (1998). 10.1126/science.1120250 42. M. Breitbart, J.Science H. Miyake, Rohwer, FEMS Microbiol. OCE-0326616 to D.M.K., and sequencing support from material on Science Online. mF 0fie e these www.sciencemag.org/cgi/content/full/311/5760/496/DC1 netic , 249 (2004). D. R. Maglott, Nucleic Acids Res. superfluid the U.S.paired Department ofsurrounded Energy Microbial Genomics 36. Lett. K. D.236 Pruitt, T. Tatusova, core by a shell of normal unpaired fermions. The critical polarization total spi c, bioMaterials and Methods field wit 43. M. Sullivan et al., PLoS Biol. 3, e144 (2005). Program.with We thank Dennis Ryan for help with scripting For near-zero polarization, we measured the 33,B. D501 (2005). diminishes decreasing attractive interaction. Figs. S1 to S12 iss ses are 44. et al., Proc. Natl. Acad. Sci. U.S.A. , 11013 and sequence analyses, the officers and crew of the R/V ktion) 37. D. M. Lindell G. Weinbauer, FEMS Microbiol. Rev. 28,101 127 k2) sublevels 2 are m 0 (state (where F is the Table S1 parameter b 0 0.54 T 0.05, describing the universal energy of a strongly interacting paired Fermi F where th (2004). Kaimikai-O-Kanaloa and the HOT team for assistance (2004). enomein the t total spin quantum mF is its projecReferences and Notes gas, at and good for agreement with recent data theory. These results are number relevant and to predictions of 45. M. G. Weinbauer, I. Brettar, M. G. Hofle, Limnol. sea,found and L. Fujieki help with oceanographic be cont these im tion) is created by radio frequency (rf) sweeps, Oceanogr. 48, 1457 (2003). 16 September 2005; accepted 21 December 2005 display. Chapman, Salamov, and P.and Richardson of exotic new J. phases of A. quark matter of strongly magnetized superconductors. mixture 46. F. Azam, Science 280, 694 (1998). 10.1126/science.1120250 the DOE Joint Genome Institute provided advice and where the relative number of the two states can polariza which i w.sciencemag.org *). The spin N Guthrie B. Partridge, Wenhui Li, Ramsey I. Kamar, Liao, Randall G. Hulet k beYean-an controlled by the rf power ( 20 mF 0 (state 2) sublevels (where F is the i is the located regions of a BCS superfluid ermion pairing is the essential ingredient which mixture is created atpaired a magnetic field of G, press th total spin quantum number and mF is754 its are projecevapora We report the observation of pairing in a gas of atomic fermions with unequal numbers of two the surrounded by an unpaired normal phase. Little in the Bardeen, Cooper, and Schrieffer which is is within theby broad Feshbach resonance tion) created radio frequency (rf) sweeps, maj Guthrie B. Partridge, Wenhui Li, Ramsey I. Kamar, Yean-an Liao, Randall G. Hulet* the opti components. a critical polarization, the separates into aG phase is consistent with a can I w (6 N known experimentally, of (BCS) Beyond theory of superconductivity. In gasis located near 834 (21number , that 22however, ). The mixture is where the relative ofspin thebecause two states netic superfluid paired core surrounded by a shell of normal unpaired fermions. The critical polarization mean fie ha difficulty in creating magnetized superconventional superconductors, the chemical the evaporatively cooled by reducing the depth of be controlled by the rf power ( 20 ). The spin We report the observation of pairing in a gas of atomic fermions with unequal numbers of two field wi diminishes of with decreasing interaction. near-zero polarization, measured the symmet conductors. evidence for anand FFLO phase potentials the two spinattractive states are equal. For optical Initial trap that confines it, the magmixture is created at a we magnetic field of 754 G, components. Beyond a critical polarization, the gas separates into a phase that is consistent with a the k2 are s k m 0 (state 2 ) sublevels (where F is the parameter b 0 0.54 T 0.05, describing universal of strongly interacting paired Fermi fre in awhich heavy-fermion superconductor only axial There has been greatRandall interest, however, in the F energy netic field ramped adiabatically to has a desired is a within the broad Feshbach resonance * the Guthrie B. Partridge, Wenhui by Li,aRamsey Kamar, Yean-an Liao, G. Hulet superfluid paired core surrounded shell of I. normal unpaired fermions. The critical polarization in,the total spin quantum number and m isspin its kprojecgas, and found good agreement with recent theory. These results are relevant toOpportunities predictions of P 0, w recently been reported (9 , 10 ). for is consequences of mismatched chemical potenfield within the crossover region. States 1 and F located near 834 G ( 21 , 22 ). The mixture diminishes with decreasing attractive interaction. For near-zero polarization, we measured the these im issequentially created bycooled radio frequency (rf)the sweeps, exotic new phases of in quark matter and of strongly magnetized superconductors. ktion) 400 nK investigation exotic pairing tials that may arise several important sit- experimental 2 are and independently imaged evaporatively by of reducing depth of parameter b 0 the 0.54 T 0.05, describing the energy fermions of a strongly paired Fermi We report observation of pairing in universal a gas of atomic withinteracting unequal numbers of two where polariza the relative number of the two states can decreasi expanded withthe the uations, including, for example, magnetized states in the trap bytrap absorption (20). it, Analysis of thehave optical thatdramatically confines and maggas,components. and found good agreement with recent theory. These results into are a relevant to predictions of with a Beyond a critical polarization, the gas separates phase that is consistent th Ni is be controlled by thethe rfmeasurement power (20). of The spin ing pola realization of Bose-Einstein condensuperconductors (13) and cold dense quark recent these images provides N and netic field is ramped adiabatically to a desired i exotic new phases of quark matter and of strongly magnetized superconductors. superfluid paired core surrounded by a shell of normal unpaired fermions. The critical polarizationwhich press regions of a paired BCS superfluid are ermion pairing is the essential ingredient mixture is created a magnetic field of G, P and th to (BEC)BCS crossover inN a two state matter at the core of neutron stars (4). A sate k1 P0 (at Ncrossover Nspin ), 754 where field within the region. States and 1 N2)/( 1 phase. 2 diminishes with decreasing attractive interaction. Forin near-zero polarization, we measured the polarization theFor maj surrounded by an unpaired normal Little the Bardeen, Cooper, and which is within the broad Feshbach resonance * Schrieffer k of ultracold atomic gases. Recent exchemical potential imbalance may be produced mixture Guthrie B. Partridge, Wenhui Li, Ramsey I. Kamar, Yean-an Liao, Randall G. Hulet k2the is number of atoms in state i . We exN are sequentially and independently imaged i parameter b 0 0.54 T 0.05, describing the universal(BCS) energy theory of a strongly interacting paired In Fermi Iw (6N is known experimentally, however, because of of superconductivity. located near 834demonstrated G (21 , 22). The spin mixture is of resonanc periments have both superfluidby several mechanisms, including magnetiFermi temperature, TF in terms of which regions of aare paired BCSto superfluid are ermion pairing is the essential ingredient in the the trap by absorption (, 20 ). Analysis gas, and found good agreement with recent theory. These results relevant predictions of press mean h the difficulty in creating magnetized superconventional superconductors, the chemical evaporatively cooled by reducing the depth of k1atomic molecule ) and pairing (1417 ) in Fermi zation in the of superconductors, mass , as kB TFN0 the(1113 majority spin state, state surrounded bycase an unpaired normal phase. Little ity in the Bardeen, Cooper, and Schrieffer these images provides measurement of We report the observation of pairing inand a gas atomic fermions with unequal numbers of two i and exotic new phases of quark matter of of strongly magnetized superconductors. symmet conductors. Initial evidence forit, an FFLO phase potentials of the twonumbers. spin states areisequal. the optical trap that confines and the mag2 1/3 1/3 recombi gases. We report the observation of pairing in a asymmetry, or unequal Pairing qual0 2p urN u ) Nis the (6N1) , where known experimentally, however, because (BCS) Beyond theory a ofcritical superconductivity. In gasis P 0w (N1 N2( )/( where components. polarization, the separates into a phase that is consistent withof a Iw polarization 1z 2), axial fr in a heavy-fermion superconductor has only There has been great interest, however, in the 6 netic isof ramped adiabatically to a kdesired molecula itatively altered by creating the Fermi energypolarization mismatch, Li atoms. Above an interactionmean harmonic frequency of the cylindrically the difficulty in super- polarized conventional superconductors, chemical is gas the number of atoms in state i. We exNifield superfluid paired core surroundedthe by a shell of normal unpaired fermions. The magnetized critical P , 0, w recently been reported ( 9 , 10 ). Opportunities for consequences of mismatched chemical potenk field within the crossover region. States 1 and observed and there has been speculation dependent critical polarization, we T observed a of symmetric confining potential with radialterms and conductors. Initial evidence for an FFLO phase are potentials of the two spin states are equal. For press the Fermi temperature, diminishes with decreasing attractive interaction. near-zero polarization, we measured the which regions ofconsiderable a paired BCS superfluid ermion pairing is the essential ingredient F, in 400 nK experimental investigation of exotic pairing tials that may arise in several important sitk2 areseparation sequentially and independently imaged molecule regarding the nature and relative stability of phase that is consistent with a unik axial frequencies u and u , respectively. For a heavy-fermion has only There hasin been great however, in universal the in energy , as kBTF 0 state, parameter b0 0.54 Tinterest, 0.05, describing of a strongly interacting paired Fermi surrounded by an superconductor unpaired normal phase. Little the majority spin the Bardeen, Cooper, and the Schrieffer r z state 1 decreasi states have expanded dramatically with the uations,proposed including, for example, in ,the trap by (20 ). of 5, Analysis 1/3 MBEC various Inmagnetized the Fuldeformly paired superfluid core by , 10 TF , the P 0, find that N1 , N recently been reported (9,phases. 10to ). however, Opportunities consequences of mismatched chemical poten,absorption where w 2surrounded p (giving ur2uz)1/3 is I wwe (6N gas, and found good agreement with recent theory. These results areexotic relevant predictions offor of is known experimentally, because (BCS) theory of superconductivity. In 20 1) ing pola recent realization of the Bose-Einstein condensuperconductors ( 13 ) and cold dense quark these images provides measurement of N and an upper Ferrel-Larkin-Ovchinnikov (FFLO) phase ( 2 , 3 ), an unpaired shell of the excess spin state. Bei of 400 mean nK for our trap frequencies. Because investigation of magnetized exotic pairing tials that arise several important sit- experimental harmonic frequency of the cylindrically exotic newmay phases of in quark matter and ofchemical strongly magnetized superconductors. the difficulty in creating superconventional superconductors, the P and sate (BEC)BCS crossover in a two spin state matter at the core of neutron stars ( 4 ). A polarization P confining 0 (N1 efficiency Npotential N1 spatial N2), where field of to 7 pairs possess nonzero center-of-mass momenlow the critical polarization, the size of and 2)/( decreasing evaporation with increasstates have a expanded dramatically the uations, including, for example, magnetized symmetric with radial conductors. Initial evidence for an with FFLO phase potentials of the two spin states are equal. For mixture of ultracold atomic gases. Recent exchemical potential imbalance may be produced k is the number of atoms in state i . We exN to be coo tum that translational invariance, whereas the was in agreement with expectations i gas ing polarization, there u is a correlation between recent realization of the Bose-Einstein condensuperconductors (13 ) and cold dense quark axial frequencies and u , respectively. For in abreaks heavy-fermion superconductor has only There has been great interest, however, in the r z resonanc periments have demonstrated both superfluidby Sarma several mechanisms, including magnetipress the temperature, T, in of which regions of reported a paired BCS superfluid are for ermion pairing essential ingredient 5,terms final fie (1 ), or the breached (5 ), phase for aPuniversal, strongly Fermi F,paired P and total atom number S1). sate (BEC)BCS crossover in apair two spin state matter at the core of the neutron stars (4). potenA the , N2 10 giving TF , , 0, Fermi we find that interacting N1(fig. recently been (9 , 10 ). Opportunities consequences of is mismatched chemical molecul ity ( 1113 ) and pairing ( 1417 ) in atomic Fermi zation in the case of superconductors, mass k 1, as Because k 0 of the For majority spin state, state (BEC) surrounded an have unpaired normal Little in the Bardeen, Cooper, and Schrieffer similar is speculated to gapless excitations. A gas. BTF of fields onour the low-field side mixture of by ultracold atomic gases. Recent exchemical potential may beimportant produced sit400 nK for trap frequencies. experimental investigation of phase. exotic pairing tials that may imbalance arise in several 2 pairing 1/3 is in recomb gases. We report the of a measure asymmetry, or unequal numbers. Pairing is qual)1/3 , where w 0 2 p (u the IwOur (6N is known experimentally, however, because of the (BCS)mechanisms, theory of for superconductivity. In mixed phase has also been proposed (68 ) in methods for observation producing 1 raudegenerate z) exist, resonance, real two-body bound states and periments have demonstrated both superfluidby several including decreasing evaporation efficiency with increasstates expanded dramatically with uations, including, example,magnetimagnetized 6Li atoms. molecul polarized gas of itatively altered by the Fermi energy mismatch, Above an interaction6 mean harmonic frequency of the cylindrically the difficulty in creating magnetized superconventional superconductors, the chemical BCS reg gas of fermionic Li atoms ( 18 , 19 ) and the molecules are readily formed by three-body ity ( 1113 ) and pairing ( 1417 ) in atomic Fermi zation in the case of superconductors, mass superconductors (13) and cold dense quark recent realization of the Bose-Einstein conden- ing polarization, there is a correlation between observed and there has been considerable speculation dependent critical polarization, we observed a symmetric confining potential with radial and conductors. Initial evidence for an FFLO phase potentials of the two spin states are equal. realization of the BEC-BCS crossover recombination. For the case of S1). P 0 at 0, a a 0 BCS t gases. We report the observation ofa pairing in astate asymmetry, or unequal numbers. Pairing is qualP and total atom number (fig. sate (BEC)BCS crossover in two spin matter at the core of neutron stars (4). A Department of Physics and Astronomy and Rice Quantum molecul regarding the nature and relative stability of Feshbach phase separation that is ) consistent with a uniaxial frequencies ur (and u For in a heavy-fermion superconductor has only There has been great however, the Institute, well belo resonance 17 have been described Rice University, Houston, TX 77251, USA. z, respectively. molecular Bose-Einstein condensate (MBEC) is of polarized gas of 6 itatively altered by theinterest, Fermi energy mismatch, Li atoms. Above an interactionFor fields on the low-field (BEC) side mixture of ultracold atomic gases. Recent exchemical potential imbalance may be in produced 5 MBEC various proposed exotic phases. In the Fuldeformly paired superfluid core surrounded by , no N2 , 10 spin , giving TF , and P , resonance, 0, weto find that N1 recently been reported (9, 10). for consequences mismatched chemical potenImag (20 ). An incoherent mixture observed form with detectable thermal and there has of been considerable speculation dependent critical polarization, weboth observed a previously *To whom correspondence should beOpportunities addressed. E-mail: real two-body bound states exist, periments have demonstrated superfluidby several mechanisms, including magnetian uppe Ferrel-Larkin-Ovchinnikov (FFLO) phasepairing (2, 3), of an unpaired shell of the excess spin state. kformed 400 nK for our trap frequencies. Because of 830 experimental of exotic tialszation that may arise in several important Ga randy@rice.edu the F0 , m 0 (state 1)of and the F 0Be, molecules (17 ).are the basis an estimated regarding the nature and relative stabilitysitof phase that is consistent with a unimolecules readily by three-body ityseparation (1113)investigation and pairing (1417 ) in atomic Fermi in the case of superconductors, mass F On pairs possess a nonzero center-of-mass momenlow the critical polarization, the spatial size of field of decreasing evaporation efficiency with increasstates have expanded dramatically with the uations, including, for example, magnetized MBEC condensate fraction of 9 90%, we place various proposed phases. In Pairing the Fuldeformly paired superfluid core surrounded by in a recombination. For the case of P 0 0, a gases. We report the observation of pairing asymmetry, or exotic unequal numbers. is qualto be co tum that breaks translational invariance, whereas the gas was in agreement with expectations ingupper polarization, there is a correlation between recent realization of the Bose-Einstein condensuperconductors (13 and coldenergy dense 6Li limit on the temperature T G 0.1 TF at a is Ferrel-Larkin-Ovchinnikov (FFLO) phase mismatch, (quark 2, 3), an unpaired shell of the excess spin an state. Be- an molecular Bose-Einstein condensate (MBEC) polarized gas of itatively altered by)the Fermi atoms. Above interactionfinal fie the Sarma (1), or crossover the breached pair (spin 5), phase for a universal, strongly interacting paired Fermi P and total atom number (fig. S1). sate (BEC)BCS in a two state matter at the core of neutron stars ( 4 ). A www.sciencemag.org SCIENCE VOL 311 27 JANUARY fieldobserved of 754 G to (17form ). However, the gas is expected pairs possess nonzero center-of-mass low dependent the criticalcritical polarization, the spatial size of a with no detectable thermal 200 and therea has been considerable momenspeculation polarization, we observed is speculated to have gapless excitations. A gas. For fields on the low-field (BEC) side of similar mixture ofseparation ultracold atomic gases. Recent chemical potential imbalance may be produced to be cooled further during thebasis adiabatic ramp for tum that breaks translational invariance, whereas the gas was in agreement expectations molecules (17). On the of an estimated regarding the nature and relative stability of phase that is with consistent with exa unimeasure mixed phase has also been proposed (68) in resonance, Our methods for producing a degenerate real two-body bound states exist, and periments have demonstrated both superfluidby several mechanisms, including magnetifinalMBEC fields greater 754 G of (17 ). By using the Sarma (1 ), or the breached pair (In 5),the phase for a universal, strongly interacting Fermi by condensate fraction 9 90%, we place various proposed exotic phases. Fuldeformly paired superfluid corepaired surrounded 6than gas of fermionic Li atoms (18 , 19 ) and the BCS reg molecules are readily formed by three-body ity ( 1113 ) and pairing ( 1417 ) in atomic Fermi zation in the case of superconductors, mass similar experimental methods, we previously is speculated to have gapless excitations. A gas. an unpaired shell of the excess spin state. Bean upper limit on the temperature T G 0.1TF at a Ferrel-Larkin-Ovchinnikov (FFLO) phase ( 2, 3), realization of the BEC-BCS crossover at a a 0 BCS recombination. For the case 0is 0, We report the observation of pairing in a of asymmetry, orhas unequal numbers. Pairing is qualDepartment of Physics and Astronomy and Quantum measured the order parameter of of gas in the mixed phase beencenter-of-mass proposed (68 ) in gases. Our methods for producing a Rice degenerate field of 754 G (17 ). However, the P gas expected pairs possess aalso nonzero momenlow the critical polarization, the spatial size 6 well bel Feshbach resonance (17) have been described Institute, Rice University, Houston, TX 77251, USA. molecular Bose-Einstein condensate (MBEC) is polarized gas of itatively altered by the Fermi energy mismatch, Li atoms. Above an interaction6 regime and further found good agreement withramp T 0 for gas the of fermionic atoms (18 , 19) expectations and the BCS to be cooled during the adiabatic tum that breaks translational invariance, whereas gas was inLi agreement with Imag previously (20 ). An incoherent spin mixture *To whom correspondence should be crossover addressed. E-mail: observed to form with no detectable thermal and the there has been considerable speculation dependent critical polarization, we observed a 30 Department 0 BCS theory ( 17 ), indicating that the gas was realization of the BEC-BCS at a Sarma (1), and or the breached pairQuantum (5), phase for a universal, strongly interacting paired Fermi final fields greater than k754 G (17). By using of Physics Astronomy and Rice randy@rice.edu of the F 0 , m 0 (state 1)of and the F 0 , 830 G a molecules ( 17 ). On the basis an estimated regarding the nature and relative stability of phase separation that is consistent with a uniF below the critical temperature pairing. Feshbach experimental methods,for we previously is speculated to Houston, have gapless A gas. resonance (17) have been described wellsimilar Institute, Rice University, TX 77251,excitations. USA.

Pairing and Phase Separation in a Polarized Fermi Gas

REPORTS Pairing and Phase Separation in a Polarized Fermi Pairing and Phase Separation in a Gas REPORTS Polarized Fermi Gas Pairing and Phase Separation in a REPORTS Polarized Fermi Gas

Pairing and Phase Separation in a Polarized Fermi Gas

F F F

Variability in Radiocarbon Ages of Individual Organic Compounds from Marine Sediments

Timothy I. Eglinton,* Bryan C. Benitez-Nelson, Ann Pearson, Ann P. McNichol, James E. Bauer, Ellen R. M. Druffel Organic carbon (OC) from multiple sources can be delivered contemporaneously to aquatic sediments. The influence of different OC inputs on carbon-14based sediment chronologies is illustrated in the carbon-14 ages of purified, source-specific (biomarker) organic compounds from near-surface sediments underlying two contrasting marine systems, the Black Sea and the Arabian Sea. In the Black Sea, isotopic heterogeneity of n-alkanes indicated that OC was contributed from both fossil and contemporary sources. Compounds reflecting different source inputs to the Arabian Sea exhibit a 10,000-year range in conventional carbon-14 ages. Radiocarbon measurements of biomarkers of marine photoautotrophy enable sediment chronologies to be constructed independent of detrital OC influences.

olecular-level studies of organic compounds in marine sediments can provide awealth of information on the carbon cycle in past and present-day oceans as well as information on the depositional setting and origin of organic matter. Of greatest utility are lipid biomarkers that are specific to individual or a restricted range of organisms and that are sufficiently refractory to be preserved in sediments. Three characteristic features of individual organic compounds are currently exploited by biogeochemists: precise molecular structure, absolute or relative abundance, and stable carbon isotopic composition. A fourth feature, the radiocarbon content, has been added to the list (1). The 14 C content provides a means to evaluate the source and fate of natural and anthropogenic organic compounds in the biogeosphere. Organic materials from various sources and with different ages are deposited concurrently in marine sediments. This is particularly the case near continents where fresh vascular plant debris, soil organic matter, and OC eroded from sedimentary rocks may be deposited together with autochthonous biomass synthesized in the water column. The construction

T. I. Eglinton, B. C. Benitez-Nelson, A. Pearson, Department of Marine Chemistry and Geochemistry, Woods Hole Oceanographic Institution, Woods Hole, MA 02543, USA. A. P. McNichol, National Ocean Sciences Accelerator Mass Spectrometry Facility, Woods Hole Oceanographic Institution, Woods Hole, MA 02543, USA. J. E. Bauer, School of Marine Science, College of William and Mary, Gloucester Point, VA 23062, USA. E. R. M. Druffel, Department of Earth System Science, University of California, Irvine, CA 92697, USA. *To whom correspondence should be addressed.

of carbon budgets and sediment chronologies based on 14C measurements of total organic carbon (TOC) depends on being able to accurately quantify these inputs. The ages of specific OC inputs and their influence on TOC 14C ages have remained elusive. Measurements of different compound classes in sediments have indicated that 14C contents are heterogeneous (2), but it is only at the biomarker level that these variations are fully expressed and can be attributed to specific inputs. Here we present an assessment of radiocarbon ages measured in individual biomarkers from two contrasting marine sedimentary systems. The Black Sea is a stratified anoxic marine basin, and the Arabian Sea is a highly productive system supported by intense seasonal upwelling. Organic carbonrich sediments accumulating in both seas are strongly influenced by terrigenous inputs. Numerous large rivers drain into the western Black Sea and affect the hydrography and geochemistry of this system (3), whereas eolian dust associated with seasonal monsoons is the primary mode of supply of continental OC to the Arabian Sea (4). In the Black Sea, we studied laminated OC-rich (TOC, 5.5%) sediments at a depth of 4 to 7 cm recovered by a Mk-III box corer (BC4, station 2; 4251N, 3157E, at a water depth of 2129 m) during leg 134-9 of the 1988 R/V Knorr expedition. In the Arabian Sea, we studied a sediment sample (box core 6BC, depth of 2 to 4 cm; TOC, 6.7%) obtained from the Oman upwelling region in the Arabian Sea (1748.7N, 5730.3E, at a water depth of 747 m) during leg TN041 of the R/V T.G. Thompson expedition in 1994. We measured the natural 14C content of specific biomarker compounds chosen to reflect both autochthonous and allochthonous inputs (5) in order
Originally published 8 August 1997

to evaluate the magnitude and source of age variation in these components of sedimentary organic matter. We selected a series of long chain (C3739) alkenes (compounds i in Table 1), the dominant unsaturated hydrocarbons in the Black Sea lipid extract, for radiocarbon dating as biomarkers of marine photoautotrophy. These compounds are derived from prymnesiophyte algae, such as the coccolithophorid Emiliania huxleyi, a major phytoplankter and important contributor to sinking particulate matter in the contemporary Black Sea (6). An autochthonous origin for these compounds is supported by their 13C values (Table 1), which are consistent with values for lipids from marine phytoplankton (7). The similarity in both 13C values and the conventional 14C ages (8) (Table 1) between the alkenes [average 13C = 25.5 per mil; 950 years before the present (B.P.), respectively] and TOC (13C = 24.2 per mil; 880 years B.P.) suggests that modern photoautotrophic biomass is a major component of the bulk OC in these sediments. The similarity in 14C ages between prymnesiophyte alkenes and TOC also suggests that, contrary to previous assumptions (9), older detrital OC inputs have a minimal influence on the 14CTOC age of late Holocene Black Sea sediments. Conversion of conventional radiocarbon ages to reservoir-corrected ages for marine carbon requires that the reservoir age be accounted for before calibration against tree-ring records of atmospheric 14C abundance (10). The reservoir age describes the difference in 14C activity between atmospheric CO2 and surfacedissolved inorganic carbon (DIC) in marine systems resulting from the mixing of older waters from depth into the surface layer of the ocean. For a reservoir correction of 400 years (9), the alkenes yield a revised 14C age for the depth interval of 4 to 7 cm of 550 years B.P. and a calibrated (calendar) age of 1425 A.D., consistent with published 210Pb and 14C profiles from abyssal Black Sea sediment cores. This age implies that sedimentation rates were 10 to 20 cm per thousand years during the late Holocene (9,11). We assessed directly the allochthonous OC inputs from 14C measurements of individual C29 and C31 n-alkanes (compounds v and vi, respectively) (Fig. 1A). The chain length distribution, odd over even carbon number predominance (OEP), and 13C compositions (Table 1) of these hydrocarbons are highly characteristic of leaf wax inputs from C3 vascular (land) plants (plants in which the first product of photosynthesis is a three-carbon acid) (12). The young conventional 14C ages for the C29 and C31 homologs relative to bulk OC (Fig. 2A) demonstrate that corrections for detrital contributions to Black Sea sediments should not necessarily invoke older OC inputs. In-

31

deed, accounting for the reservoir age correction for the prymnesiophyte alkenes (vascular plants fix atmospheric CO2 directly, so their 14 C compositions are not subject to any reservoir correction), we find that the n-C29 and n-C31 alkanes are similar in calendar age to the marine phytoplanktonic biomarkers. Taken together, these data indicate that relatively fresh terrestrial (land plant) debris is deposited with autochthonous OC of similar age in Black Sea sediments. In contrast, a suite of shorter chain (C23 to C27) n-alkanes (compounds ii through iv) as well as the total saturated hydrocarbon (HC) fraction exhibit significantly older 14C ages than either the vascular plant n-alkanes or TOC (Fig. 2A and Table 1). Given the similarity in 13C values to those of the longer chain alkanes, an independent source for the shorter chain n-alkanes seems unlikely. Instead, we infer that fossil hydrocarbons partly contribute to the lower molecular weight homologs. Assuming that there was simple mixing of two end members, we can substitute 14C values (8) (Table 1) into an isotopic mass balance (1) to estimate the proportion of fossil hydrocarbon required to produce the observed radiocarbon values. For a 14C-free signature (14C = 1000 per mil) for the fossil component and 66 per mil for a fresh higher plant component (n-C31 alkane, Table 1), the measured 14C value for the C23 n-alkane (153 per mil) would correspond to 9% fossil n-alkane. Similarly, given the 14C value of the saturated HC fraction (155 per mil), fossil hydrocarbons likely represent a small but significant component of this fraction. This minor contribution would have little influence on the carbon number distribution or 13C values but is clearly revealed by 14C analysis. Thermogenic hydrocarbons have been detected (13) in suspended particles from the Black Sea on the basis of a low OEP n-alkane distribution dominated by lower carbon number homologs (C23 to C27). If a similar input was responsible for the fossil hydrocarbons we observed, the C23 and C25 n-alkanes would be most strongly influenced by the fossil sources. Because abyssal Black Sea sediments are finely laminated (nonbioturbated) and the depth interval we studied is believed to predate significant human use of fossil fuels (pre-1880), we conclude that these hydrocarbons entered sediments by natural processes such as erosion of petroleum-bearing sediments (14) or seepage from depth (15). We selected two sterenes (compounds e and f) as generic markers of marine primary productivity from the Arabian Sea sediment sample because sterols, their precursor natural products (16), are biosynthesized by a wide range of phytoplankton and some zooplankton. Conventional 14C ages of these compounds (mean = 680 years B.P.) are signifi-

cantly younger than the TOC age (890 years B.P.). Consequently, in contrast to the Black Sea, the offset between sterene and TOC conventional 14C ages likely reflects contributions of older OC to the sediment. This age discrepancy is amplified further when a series of highly branched isoprenoid (HBI) alkenes (compounds a through d), characteristic diatom biomarkers (17), are examined (Fig. 2B). The C25 HBI alkenes exhibit substantially younger 14C ages (mean = 280 years B.P.) than both the sterenes and TOC. Their 14C compositions would yield present-day 14C ages after a 400-year reservoir correction is applied and are likely the result of downward mixing of surficial material containing bomb radiocarbon (from nuclear weapons testing). This mixing could result from sediment winnowing, a process that has been suggested to be a dominant control on the composition of Arabian Sea sediments (18). An alternative explanation giving rise to 14C variations in different photoautotrophic biomarkers is heterogene-

ity in the reservoir age of the DIC pool. The strong seasonal variations in phytoplankton productivity, community structure, and vertical distribution in the Arabian Sea are related to wind-driven (monsoonal) upwelling and mixing of the water column (19). The balance between the upwelling of deep waters (low 14C activity) and the physical invasion or biologically driven draw-down of atmospheric CO2 (high 14C activity) into the surface ocean may give rise to spatial and temporal variability in 14CDIC (and, consequently, planktonic 14C). Consequently, differences in 14C and 13C composition, such as those between the C25 and C30 HBI alkenes, may be a function of the ecological characteristics (such as depth of growth) of their photoautotrophic sourcethe former having been identified in Haslea sp. and the latter in Rhizosolenia sp. (20). Such discrepancies between 14C ages of phytoplanktonic markers raise the possibility that information on past variations in water column structure may be carried by the isotopic signatures of

Table 1. Isotopic composition of bulk fractions and isolated compounds. The 13C values were determined by irm-GC-MS (5), except when noted, and were calculated relative to PDB. The 14C age is the conventional radiocarbon age (8, 10). Prym., prymnesiophytes; Vas., vascular (higher) plant; Fos., fossil carbon; Diat., diatoms; Phyt., phytoplankton; Bac., bacteria; ID, compound identification; n.d., not determined (13C assumed as 25 per mil in these instances).

32

different photoautotrophic biomarkers and preserved in the sedimentary record. The hopanoid alkenes (compounds g through l), which were chosen as bacterial biomarkers (21), yielded with one exception (22) conventional 14C ages that are 300 to 400 years older than the sterenes. The structures of the individual compounds measured do not permit categoric assignment of precursor organisms; however, their relatively uniform 13C compositions (Table 1) clearly point to a marine origin. Deep-dwelling cyanobacteria

could yield older 14C ages as a result of uptake of respired DIC emanating from the O2depleted waters underlying the Oman upwelling zone (7), whereas benthic heterotrophic or chemoautotrophic bacteria may consume older carbon associated with bottom waters or sediments (23). We determined the 14C compositions of selected saturated hopanoid (compound m) and normal (compounds n through p) hydrocarbons to investigate isotopic characteristics of allochthonous sources of OC. In marked

contrast to the sterenes and HBI alkenes, these hydrocarbons display much older conventional 14C ages ranging from 7200 to 10,000 years B.P. (Fig. 2B). As for the Black Sea sample, a pronounced OEP for the C25 to C31 n-alkanes indicates that OC from higher plants was supplied to the sediments (Fig. 1B). Plant detritus is likely delivered to this region by eolian processes, primarily the Somali Jet, which entrains dust from the Arabian Peninsula and the Horn of Africa (4). Older ages for the nalkanes could be expected because the OC

Fig. 1. Partial high-resolution gas chromatograms [with flame ionization detection (FID)] of total saturated hydrocarbon fractions from (A) Black Sea core BC4 (4 to 7 cm), and (B) Arabian Sea core AS-2 (2 to 4 cm). Numbers in italics

indicate carbon chain lengths of n-alkanes. Numbers and letters in parentheses denote compounds isolated for 14C analysis from these fractions (see Table 1).

Fig. 2. Plot of conventional 14C ages (determined by AMS) for individual hydrocarbons and bulk fractions: (A) Black Sea core BC4, 4 to 7 cm, and (B) Arabian Sea core 6BC, 2 to 4 cm. See Table 1 for individual compound identifica-

tions. Note break in y-axis in (B). Error bars represent uncertainty associated with statistics of the AMS measurement. Overall error is believed to be better than 5% (in 14C) (1).

33

ents. All ch were DON, 2 fraction counted THAS, characere not

560 9.0 0.7 nm nm 3.1 97 nm 11 89

pears that the selective preservation of unusual biochemical components of cells could not, by itself, account for the rapid formation and accumulation of bacterially in dust plumes from this region are believed derived DOM. to arise from desiccated lacustrine sediments Components of bacterial cells are reand paleosols formed at times when clileased into the surrounding water asthe DOM mate in eastern Africa was less arid (before through a variety of biological processes, 6000 years B.P.) (24).(11, The hopanoid alincluding direct release 31 ), viral lysis kane (compound m) may derive from similar (32), and grazing (33). Exoenzyme activity sources (soil or the lacustrine bacteria); however, is critical for microbial utilization of 13 the similarity of its C value to the photothis DOM, and it appears that enzymatic autotrophic biomarkers suggests is forhas activity plays an important role that in the 14 a marine origin. We suspect that is the old C mation of refractory DOM that of small ages of the n -alkanes and hopane are a result size (34 ). It is possible that nonspecific or in part of the activities presence of OC. that Support promiscuous of fossil enzymes, ocfor interpretation providedthan by the lack cur this with much lower is efficiency primaof in a series of occasionally n-alkanes extending to ry OEP activities (35), produce C (Fig. 1B), indicating the presence of therfragments from macromolecules that es40 mogenic hydrocarbons. The lack of an associcape recognition by bacterial enzymes and ated unresolved envelope suggests thatGiven these molecular-level chemical analyses. this scenario, the rate of formation of refossil hydrocarbons are protected from microfractory DOM is dependent on the rate of bial degradation, possibly through association microbial activity. This relatively simple with mineral matrices (25). The overlap in mechanism envelopes could be responsible for much then-alkane from the higher plant of the sequestration of fixed C in the and fossil inputs (Fig. 1B) provides an ocean. opportunity to estimate the 14C age for the plant wax
References and Notes

contribution. For infinite 14C age petrogenic contributions of 30, 40, and 50% (26) to the C23, C27, and C29 n-alkane, respectively, application of a similar mass balance approach to that described above for the Black Sea nalkanes yields a 14C age of between 4000 and 5000 years B.P. for the land plant component, quite consistent with the timing for the onset of deterioration of climatic conditions in East Africa (24). In both sediments, the maximal variation in radiocarbon ages observed between compounds exceeds the estimated <300-year age span (based on published sedimentation rates) for either of the intervals sampled. Our results reveal that compounds of the same class and belonging to the same homologous series (Black Sea n-alkanes) can exhibit distinctly different radiocarbon ages. Compounds having similar 13C values may display significantly different 14C ages (Black Sea n-alkanes) or, conversely, compounds of similar 14C age may have various 13C values

(Arabian Sea HBI alkenes). These molecular isotopic measurements thus illustrate how different sources can affect sedimentary 14CTOC composition. Differences in 14C age can now be explained in the context of isotopic characteristics of specific source inputs and hence can be used to interpret the biogeochemical processes that govern their provenance in sediments. Two examples of long-standing issues in biogeochemistry that benefit from molecular 14 C measurements are the origin of nonzero 14CTOC ages commonly observed for marine surface sediments (27), and the development of refined chronostratigraphies free of interferences due to detrital OC inputs and selective degradation of different organic components. The latter also holds promise for other disciplines where accurate 14 C measurements are important (such as archeology) but may be compromised by the presence of extraneous carbon-containing material.

1. T. R.I. Benner, Pakulski, M. J. I. Hedges, 1. Eglinton,J.L.D. I. Aluwihare, J. E.McCarthy, Bauer, E. R. M. Druffel, A. CuO was added, and tubes were evacuated and P.McNichol, G. Hatcher, Science 255 1561 (1992). P. Anal. Chem. 68,, 904 (1996). flame-sealed. The samples were combusted (900C, 5 2. Wang, E. R. M. T. Druffel, C. Lee Giger et Nucl. hours), and the resulting CO2 was purified, measured, 2. X.-C. M. D. McCarthy, Pratum, J. W. Hedges, R.al., Benner, 13 Instr. Meth. Phys. B5, 394 (1984); X.-C. Wang, E. R. and subsequently converted to graphite by reduction Nature 390 , 150Res. (1997). 3.6 M. Druffel, C. Lee, Geophys. Res. Lett. 23, 3583 (1996). over Co catalyst in the presence of H2 for radiocarbon 3. L. L. Clark, E. D. Ingall, R. Benner, Nature 393, 426 14 3. E. Izdar and J. W. Murray, Eds., Black Sea Oceanography analysis by AMS. AMS was performed at either Woods (1998). (Kluwer, Dordrecht, Netherlands, 1991). Hole Oceanographic Institution or Lawrence Livermore 4. E. Tanoue, S. Nishiyama, M. Kamo, A. Tsugita, 4. F. Sirocko and M. Sarnthein, in Paleoclimatology and National Laboratory. Geochim. Cosmochim. Acta 59 , 2643 (1995). nm Paleometeorology: Modern and Past Patterns of Global 6. J. K. Volkman, G. Eglinton, E. D. S. Corner, T. E. V. Forsberg, 5. M. D. McCarthy, J. I. Hedges, R. Benner, Science 281 , Atmospheric Transport, M. Leinen and M. Sarnthein, Eds. Phytochemistry 19, 2619 (1980); H. A. Benli, in Particle nm 231 (1998). (Kluwer, Dordrecht, Netherlands, 1989), pp. 401433. Flux in the Ocean, E. T. Degens, E. Izdar, S. Honjo, Eds. nm 5. sediment stored frozen (20C) 1 from 6. The P. M. Williams,samples, E. R. M. Druffel, Oceanography , 14 (SCOPE/UNEP Sonderbrand, West Germany, Heft 62, the time of collection until analysis, were thawed, 1987), pp. 7787; B. J. Hay, M. A. Arthur, W. E. Dean, E. D. (1988). and a portion was dried and acidified (2N HCl) Neff, S. Honjo, Deep Sea Res. 38, 1211 (1991). 28 7. J. I. Hedges et al., Org. Geochem. 31, 945 (2000). for determination of bulk elemental and isotopic 7. K. H. Freeman, S. G. Wakeham, J. M. Hayes, Org. 8. G. R. Harvey, D.13 A. Boran, L. A. Chesal, J. M. Tokar, 8 composition [ CTOC by isotope ratio mass Geochem. 21, 629 (1994). Mar. Chem. 12, 119 (1983). 46 14 spectrometry (irMS), and CTOC by accelerator mass 8. Conventional 14C ages (in years B.P.) are calculated with 9. spectrometry J. E. Brophy, (AMS)]; D. J. Carlson, Res . 36 , 497 18 30 g Deep-Sea (equivalent dry weight, a Libby half-life of 5568 years and take into account13C (1989). Black Sea sample) or 200 g (Arabian Sea sample) fractionation (corrected to a 13C value of 25 per mil) was extracted with a Soxhlet apparatus with CH3OH but not differences in specific 14C activity of reservoirs and CH2Cl2. An aliphatic hydrocarbon fraction was or calibration to calendar years. In oceanography, 4 MAY 2001isolated from the resulting total lipid extract (TLE) by 919 14C is the per mil deviation of the 14C/12C ratio silica gel chromatography with n-C6H14 as the eluent. (13C normalized and corrected to A.D. 1950) for the After removal of elemental sulfur (with an activated sample relative to the14C/12C ratio of the absolute Cu column), saturated hydrocarbons (n-C6H14 eluent) international standard (95% of the A.D. 1950 activity were separated from unsaturated counterparts of NBS HOxI, normalized to 13C = 19 per mil). (C2H5OC2H5 eluent) by AgNO3-SiO2 column 9. G. A. Jones, A. R. Gagnon, Deep Sea Res. 41, 531 (1994). chromatography. Compound identifications were 10. M. Stuiver and H. A. Polach, Radiocarbon 19, 355 made by gas chromatographymass spectrometry (1977); M. Stuiver, G. W. Pearson, T. Braziunas, ibid. 28, (GC-MS), and 13C compositions were determined 980 (1986). by isotope ratio monitoring GC-MS (irm-GC-MS). 11. S. E. Calvert, et al., Nature 350, 692 (1991); J. Crusius, R. Individual compounds were isolated from the purified F. Anderson, Paleoceanography 7, 215 (1992). hydrocarbon fractions by preparative capillary gas 12 G. Eglinton, R. J. Hamilton, Science 156, 1322 (1967); chromatography (PCGC). Compounds were trapped B. N. Smith, S. Epstein, Plant Physiol. 47, 380 (1971); G. during PCGC in cryogenically cooled glass u-tubes, and Rieley, et al., Nature 352, 425 (1991); J. W. Collister, G. on completion of the sequence (typically about 100 Rieley, B. Stern, G. Eglinton, B. Fry, Org. Geochem. 21, repeated PCGC runs of each fraction yielded sufficient 619 (1994); S. G. Wakeham, Mar. Chem. 53, 187 (1996); quantities of compound for AMS 14C analysis), the B. R. T. Simoneit, Deep Sea Res. 24, 813 (1977). products were recovered by dissolution in CH2Cl2 and 13. S. G. Wakeham, J. A. Beier, C. H. Clifford, in (3), pp. 319 transferred to quartz combustion tubes. After some 341. of the sample was removed (for determination of 14. S. J. Rowland, J. R. Maxwell, Geochim. Cosmochim. Acta yield and purity by GC), solvent was eliminated from 48, 617 (1984). the combustion tubes under a stream of nitrogen, 15. J. E. Bauer, R. B. Spies, J. S. Vogel, D. E. Nelson, J. R.

Southon, Nature 348, 230 (1990). 16. R. B. Gagosian, S. O. Smith, C. Lee, J. W. Farrington, N. M. Frew, in Advances in Organic Geochemistry, 1979, A. G. Douglas and J. R. Maxwell, Eds. (Pergamon, Oxford, UK, 1980), pp. 407419. 17. P. D. Nichols, J. K. Volkman, A. C. Palmisano, G. A. Smith, D. C. White, J. Phycol. 24, 90 (1988); R. E. Summons, R. A. Barrow, R. J. Capon, J. M. Hope, C. Stranger, Aust. J. Chem. 46, 907 (1993). 18. T. F. Pederson, G. B. Shimmield, N. B. Price, Geochim. Cosmochim. Acta 56, 545 (1992). 19. B. Haake et al., Deep Sea Res. 40, 1323 (1993); F. Pollehne, B. Klein, B. Zeitzschel, ibid., p. 737. 20. J. K. Volkman, S. M. Barrett, G. A. Dunstan, Org. Geochem. 21, 407 (1994). 21 G. Ourisson, M. Rohmer, K. Poralla, Annu. Rev. Microbiol. 41, 301 (1987). 22. The reason for the old hopene 14C age is unclear, but it may be related to a small proportion of unresolved material that was trapped with this compound during PCGC isolation. 23. J. E. Bauer, C. E. Reimers, E. R. M. Druffel, P. M. Williams, Nature 373, 686 (1995). 24. H. A. McClure, ibid 263, 755 (1976). 25. The GC profiles for the higher plant waxes and petrogenic hydrocarbons as well as the OEP (Fig. 1 B) were used to estimate contributions from each source to individual n-alkanes. 26. R. C. Barrick, J. I. Hedges, M. L. Peterson, Geochim. Cosmochim. Acta 44, 1349 (1980). 27. K. O. Emery and E. E. Bray, Bull. Am. Assoc. Petrol. Geol. 46, 1832 (1962); G. J. Benoit, K. K. Turekian, L. K. Benninger, Estuarine Coastal Mar. Sci. 9, 171 (1979); S. Emerson et al., Nature 329, 51 (1987). 28. We thank M. Kashgarian, J. Southon, I. Proctor, and L. Osborne for assistance with AMS analyses, J. Primack and J. Hayes for irm-GC-MS analyses, and D. Repeta and J. Sachs for the sediment samples. This work was supported in part by NSF grants (OCE-94155680; OCE801015) and a Woods Hole Oceanographic Institution (WHOI) Independent Study Award. This is WHOI contribution n. 9476. 23 April 1997; accepted 13 June 1997

34

Lipid-Like Material as the Source of the Uncharacterized Organic Carbon in the Ocean?
Jeomshik Hwang and Ellen R. M. Druffel The composition and formation mechanisms of the uncharacterized fraction of oceanic particulate organic carbon (POC) are not well understood. We isolated biologically important compound classes and the acid-insoluble fraction, a proxy of the uncharacterized fraction, from sinking POC in the deep Northeast Pacific and measured carbon isotope ratios to constrain the source(s) of the uncharacterized fraction. Stable carbon and radiocarbon isotope signatures of the acid-insoluble fraction were similar to those of the lipid fraction, implying that the acid-insoluble fraction might be composed of selectively accumulated lipid-like macromolecules.

ess than 40% of sinking POC collected below the euphotic zone can be molecularly characterized (1, 2). The uncharacterized fraction constitutes an increasing proportion of POC with the depth at which it is collected, with the highest fraction in sedimentary organic carbon (1). What is the composition of the uncharacterized fraction and how is it formed? One hypothesis for the formation of the uncharacterized fraction is abiological recombination of small molecules such as amino acids and carbohydrates produced by degradation of labile organic matter (3, 4). This hypothesis has been challenged by results of 13C and 15N nuclear magnetic resonance (NMR) spectroscopy (5, 6). A second hypothesis is that biologically produced refractory compounds are selectively accumulated whereas labile compounds are remineralized (7, 8). Hydrolysis-resistant cell wallderived material has been observed in recent and ancient sediments (911). A third hypothesis involves physical protection of organic carbon by refractory organic or inorganic matrices (12, 13). A recent study explored solid-state 13C NMR spectra of plankton and sinking POC collected at shallow and deep waters (14). The similarity of the spectra led Hedges et al. to suggest that the uncharacterized fraction was the same organic material produced biologically but was protected by mineral matrices or refractory biomacromolecules. Biologically produced lipids, amino acids, and carbohydrates have distinct stable carbon isotope [13C (15)] signatures because of the

different physiological fractionation of carbon during their syntheses (16, 17 ). Therefore, comparison of the 13C signature of the uncharacterized fraction with those of other organic fractions will provide insights as to its source(s). The radiocarbon isotope [14C (18)] signatures of all organic fractions are the same when measured in plankton from the surface water because they are fractionationcorrected. The overall signature changes when carbon with a different 14C signature is incorporated and when the carbon is aged O R T S organic fractions that have (R 19E ). PTherefore, similar sources, sinks, and residence times in is expected to have similar 14C values to the ocean will have similar 14C signatures. those of DIC in surface waters14 [40 to 80 per We measured 13 C and C in C values values of mil () (22)]. The observed 14 sinking POC collected from a at depth of 3450 bulk sinking POC collected a depth of m, at a M, the 4100 m deepvalues, at the 3450 msite are (Station lower than expected bottom, 3450 N, 12300 W) 220 km west of which means that sinking POC acquired old the California coast.carbon We isolated biologically carbon from other reservoirs during important classes: to lipids, total hythe transitcompound from the surface the sampling 14 drolyzable amino acids (THAA), total hydroof depth. Thus, C can be an indicator lyzable neutral (TCHO), and diagenetic statuscarbohydrates of POC (i.e., how much old, a proxy of the uncharacterized the degraded carbon is incorporated fraction, into sinking acid-insoluble fraction (20 ). 14 The acid-insoluC values of the POC), assuming constant ble fraction that carbon. remains after organic solvent source organic The weight percentages of the character-

extraction and acid hydrolysis accounts for 70% of the uncharacterized fraction (21). Sinking POC originates mainly from dissolved inorganic carbon (DIC) in surface waters, and it reaches the deep water on a time scale of months. Therefore, bulk sinking POC is expected to have similar 14C values to those of DIC in surface waters [40 to 80 per mil () (22)]. The observed 14C values of bulk sinking POC collected at a depth of 3450 m are lower than the expected values, which means that sinking POC acquired old carbon from other carbon reservoirs during the transit from the surface to the sampling depth. Thus, 14C can be an indicator of diagenetic status of POC (i.e., how much old, degraded carbon is incorporated into sinking POC), assuming constant 14C values of the source organic carbon. The weight percentages of the characterizable fractions such as lipids, THAA, and TCHO are positively correlated with the 14C values of bulk POC, whereas that of the acid-insoluble fraction is negatively correlated (Fig. 1). Thus, the higher the fraction of acidinsoluble organic carbon, the lower the bulk POC 14C value. The 14C values of each organic fraction are plotted with respect to the bulk POC 14C to show the relative amounts of old and new carbon (Fig. 2). The 14C values of THAA and TCHO are similar to each other and close to those of surface water DIC. The 14C values of lipids and the acid-insoluble fractions are izable fractions such as lipids, THAA, and similar to each other but are lower than those TCHO are positively correlated with the 14 of THAA andof TCHO. The C values C values bulk POC, whereas thatof of lipthe 14 ids and the acid-insoluble fraction decrease at acid-insoluble fraction is negatively correlatslightly higher rates than THAA and TCHO, ed (Fig. 1). Thus, the higher the fraction of making the difference between two groups acid-insoluble organic carbon,the the lower the bigger as the bulk POC 14C values decrease. C value. bulk POC 14 14 TheThe slopes of lines imply C values C the values of each that organic fraction 14 14 will converge at or around the range of DIC C are plotted with respect to the bulk POC 14 in surface waters. The 14of C values ofnew the to C show the relative amounts old and four fractions from plankton collected at the of THAA carbon (Fig. 2). The 14C values same station were equal theother range ofclose DIC and TCHO are similar to to each and 14 14 C (23). of This indicates thatDIC. all organic fracC to those surface water The values of lipids and the acid-insoluble frac-

Department of Earth System Science, University of California Irvine, Irvine, CA 926973100, USA. E-mail: jeomshik@uci.edu (J.H.); edruffel@uci.edu (E.R.M.D.)

Fig. 1. Percent of each organic compound class (lipids, solid circles; THAA, solid triangles; TCHO, open triangles; acid-insoluble fraction, open circles) separated from several 10-day samples of sinking POC collected at a depth of 3450 m at Station M in the Northeast Pacic. The amounts of CO2 of each compound class were measured manometrically after combustion and were compared to the organic carbon content measured with an element analyzer. The uncertainty is the larger value of 1 of either the yield of standard material (replicate number 5) or a duplicate run of samples (3% for lipids, 5% for THAA, 3% for TCHO, and 1% for acid-insoluble fraction).
published 7 February Fig. 2.Originally The 14C values of each 2003 organic compound class (sym-

tions a than th values tion d THAA betwee POC lines i or aro waters from p were e This i sinkin e(s) of uble fr did TH Th lower 3), as ation those ferenc tions though vary b Th the ac ganic source acid-in reactio 35 tightly differe

ganic fractions provide constraints on its source(s). One possible scenario is that the acid-insoluble fraction is synthesized by the reaction between THAA and TCHO that are Fig. 2. The 14C values of each tightly bound with About particles. However, the (220-km offshore)]. 10% of the acidorganic compound class (symbetween the acid-insoludifferences in 13C insoluble fraction needs to be old carbon to bols are as dened in Fig. 1). The ble fraction and low THAA and TCHO ( 4) explain observed 14C values (27). Howblack arrow by the right y - axis 13 argue against the degradation-recombination 14 ever, in order to explain the low C values indicates the range of DIC C hypothesis of amino acids and a carbohydrates of the acid-insoluble fraction, much larger values in surface waters at Stafor the formation of the acid-insoluble fraction M. The 14C values are fraction (up to 50%) needs to be terrestrial 13 C-enriched fractions cannot comblank-corrected. We measured tion. Two carbon ( 27 ). Therefore, this mechanism can13 14C of an amino acid standard fraction binesatisfy to form C and 14C results unless at the not thea 13C-depleted solution (mixture of 17 amino kinetic fractionation occurs during the chemsame time. acids), D-glucose powder, and ical reactions. If kinetic fractionation occurs, It is most likely that the major part of the cod liver oil by the same meth13 C to enriched inprereone would expect ods as the samples. We comacid-insoluble fraction is be a selectively maining THAA and TCHO ( 24 ). However, pared these values with those of served part of organisms that is chemically values of the compound fractions unprocessed standards to calcuthe 13Cfrom different amino acids and carbohydrates. 14 late the C values of the were reported to remain constant for organic 14 Low C values can be explained if the major blanks. The amounts of the blank carbon of detrital aggregates, sediment floc, portion of the acid-insoluble fraction is surwere measured manometrically by combining ve or six blanks together. Blanks were higher for and sediment (0 to 20 cm) at the same locaTHAA and TCHO, because there were more steps involved in their separation. The acid-insoluble face-originated fresh carbon and only a small tion, and Wang et al. concluded that 13C fraction was not blank-corrected because the blank was smaller than 0.01 mg (0.7% of the fraction is older carbon that has aged and was signatures of each organic fraction were not sample). The uncertainty is the larger value of 1 of either standard material (replicate number 5) incorporated into the particles. The source of affected by decomposition processes of oror duplicates of samples. the oldmatter carbon can ganic (23 ). be either dissolved organic carbon (DOC) or suspended POC. Resus the acidThe differences in 13C between Fig. 3. The 13C values of each compended sedimentary organic carbon will arbe insoluble fraction and THAA and TCHO pound class (symbols are as dened in old, 13Cnonselective values will mineral be very protection similar to gue but against Fig. 1). The 13C values are blank-corthose of POC ( 28 ). Old carbon from eroded 14 as well. If the acid-insoluble fraction is simrected by the same method as C. marine-origin bedrockproduced is another possible The uncertainty is the larger value of 1 ply the same material biologically source of old suspended POC (29 ). the only of either standard material or duplias lipids, THAA, and TCHO, with cates of samples. The incorporation old carbon into difference that they are of protected by mineral particles more actively should for lipids and matrices, occurs its isotopic signature be simthe acid-insoluble fraction. Because lipids ilar to those of THAA and TCHO. Furtherare hydrophobic water-insoluble, they more, treatment ofand the acid-insoluble fraction are more likely to be adsorbed of POC with an HCl-HF solution to to particles dissolve than are THAA and TCHO, of which the mineral phases did notmost increase the are water-soluble (30 ). THAA The fact the extractable fraction of andthat TCHO acid-insoluble fraction has similar 14C values to those of lipids suggests that the 882 7 FEBRUARY 2003 VOL 299 SCIENCE www.sciencemag.org tions of sinking POC originated from modern THAA and TCHO (24). However, the 13C two fractions have similar physicochemical source(s) of carbon, but lipids and the acid- values of the compound fractions were re- properties. The similarity of 13C in lipids insoluble fraction acquired more old carbon ported to remain constant for organic carbon and the acid-insoluble fractions may suggest than did THAA and TCHO. of detrital aggregates, sediment floc, and sedi- that both fractions are synthesized by similar The 13C values of lipids are 3 to 4 low- ment (0 to 20 cm) at the same location, and biochemical pathways that are different from er than those of THAA and TCHO (Fig. 3), Wang et al. concluded that 13C signatures of those for THAA and TCHO. as expected from physiological fractionation each organic fraction were not affected by deOur results suggest that the major portion (16, 17), but approximately equal to those of composition processes of organic matter (23). of the acid-insoluble fraction is composed of the acid-insoluble fraction. The differences in The differences in 13C between the acid- lipid-like material that is biosynthesized by 13C between the organic fractions for each insoluble fraction and THAA and TCHO ar- similar pathways to the extractable lipids, but sample are constant even though 13C values gue against nonselective mineral protection as somehow is resistant to organic solvents and in bulk POC samples vary by 2.5. well. If the acid-insoluble fraction is simply acids. In fact, lipids are enriched in resistant The constant differences in 13C between the same material produced biologically as parts of organisms such as membranes, the acid-insoluble fraction and the other or- lipids, THAA, and TCHO, with the only dif- spores, and cuticles (4). Selective preservation ganic fractions provide constraints on its ference that they are protected by mineral ma- of these resistant lipids has been considered source(s). One possible scenario is that the trices, its isotopic signature should be similar as one method of kerogen formation (4). acid-insoluble fraction is synthesized by the to those of THAA and TCHO. Furthermore, The refractory cell wallderived material of reaction between THAA and TCHO that are treatment of the acid-insoluble fraction of microalgae and bacteria, known as algaenans tightly bound with particles. However, the POC with an HCl-HF solution to dissolve the and bacterans, respectively, were found to differences in 13C between the acid-insoluble mineral phases did not increase the extract- make up a large part of kerogen and refractory fraction and THAA and TCHO (4) argue able fraction of THAA and TCHO significant- organic carbon in sediments (8, 11, 3133). against the degradation-recombination hy- ly [<5% of total organic carbon (OC)] (25, They are, essentially, highly aliphatic lipids pothesis of amino acids and carbohydrates for 26). However, our data do not argue against (i.e., a large number of carbon-to-carbon the formation of the acid-insoluble fraction. the hypothesis of selective mineral protection. bonds) evidenced by release of n-alkane Two 13C-enriched fractions cannot combine Another possibility is that the acid- in- and n-alka-1-ene upon pyrolysis and by a to form a 13C-depleted fraction unless kinetic soluble fraction is protected lipids, THAA, high alkyl signal in 13C NMR spectra. A fractionation occurs during the chemical re- and TCHO and contains a small fraction of network structure built up by cross-linking actions. If kinetic fractionation occurs, one old carbon [for example, old terrestrial carbon of long hydrocarbon chains was suggested would expect 13C to be enriched in remaining considering the location of the sampling site as a protection mechanism against chemical
cate run of samples (3% for lipids, 5% for THAA, 3% for TCHO, and 1% for acid-insoluble fraction).

36

10. drates C. Flaviano, F. (2 Lehours Berre, S. Derenne, Largeau, J. rence Livermore Laboratory (LLNL). For a detailed 13 14NMR spec13 SO ,C. then (TCHO) 72% H 13 will be Even though results of SO4 then 3 3 hours hours drates (TCHO) (2 hours in in 72% H2 C values to insoluble fraction has C transported old carbon from the will be C NMR specEven though results of the Cthe NMR spectra. A lective high alkyl signal in similar 2 4,continental Connan, Org. Geochem. 22 , 759 (1994). description of the method, see (23). SO ). The hydrolyzates were eluted in 0.6 M H 2 SO 4 ). The hydrolyzates were eluted in 0.6 M H carbon troscopy and direct temperature-resolved 2 4 those of lipids suggests that the two fractions shelf or slope and from rivers in the Califorcarbon troscopy and direct temperature-resolved 11. through C. Largeau, J. W. de Leeuw, in Advances in Microbial 21. The percentage was calculated from the ratio of the network structure built up by cross-linking of ion exchange columns for the separation of through ion exchange columns for the separation of Ecology , (J. G. Ed. (Plenum, New York, 1995), acid-insoluble fraction to the uncharacterized fracnother spectrometry on samples from the have physicochemical properties. The nia THAA coast 37 ).Jones, Further study will necesnother mass spectrometry on the thewas samples from as the and TCHO. The carbon left after THAA long similar hydrocarbon chains suggested a acid- mass THAA and TCHO. The organic organic carbon leftbe after THAA vol. 14, pp. 77117. tion that is the difference between the total organic extraction was used for the acid-insoluble fraction. C ( Equatorial do not support the predomlipids and acidsimilarity of 13C sary to confirm if our observation is universal extraction was used for the acid-insoluble fraction. C (29 29). ). Equatorial Pacific doin not support thethe predomprotection Pacific mechanism against chemical atTHAA, 12. Each R. G. organic Keil, D. B. Montluc accumulation on, was F. G.acidied Prahl, J. with I.of Hedges, the characterized fractions (lipids, THAA, carbon extract 1 ml attack (of 9).lipid-like Considering that the mechanisms (14,carbon 34), and the composition of sinking POC sediments, selective the Each organic carbon extract was acidied with 1 ml to parinance material in sinking POC insoluble fractions may suggest that both in the oceans. These results should provide to parinance of lipid-like material in sinking POC tack (9). Considering that the mechanisms of n of old Nature 370 , ,549 (1994). and TCHO). PO dried under vacuum, and combusted at of 3% H 3 PO 4 , dried under vacuum, and combusted at of 3% H of lipid synthesis are similar in all organisms might vary depending on the local environlipids is likely to occur in the water column 3 4 nd the ( 14 ), the composition sinking POC 13. 850C R. A. Armstrong, C. Lee, J. I.understand Hedges, S. Honjo, S. G. 22. C. A. Masiello, E. R. M. Druffel, J. E. Bauer, Deep-Sea fractions synthesized by of similar biochemimportant information to how orndconthe ( 14, , 34 34 ), are the composition of sinking POC lipid synthesis are similar in all organisms on 2 and a 850C for for 2 hours hours with with CuO CuO and silver silver foil foil in in a sealed sealed Wakeham, Deep-Sea Res. II 49, accumulation 219 (2002). ( 17),pathways these macromolecules might (36 ). ,The percentage of lipid in sinking as the POC sinks. Selective of ment Res. II 45 617 (1998). ds are might vary depending on environwas at on tube. The CO ical that are different from those for ganic carbon is preserved and removed from ds(220are might vary refractory depending on the the local local environwas graphitized graphitized at 600C 600C on tube. The resultant resultant CO2 (17 ), these refractory macromolecules might e 2 13 14. a J. I. Hedges etlipid-like al. , Nature 409 , 801 (2001). 14 23. X.-C. Wang, E. R. M.Sea Druffel, Grifn, C.high Lee, M. 14C material measurements were perCo The be to have similar similar POC in the Arabian wasS.twice as as signatures the refractory in the water 13C in C ocean. measurements were pera Co catalyst. catalyst. The ey are ment ( ). of lipid sinking THAA and TCHO. the carbon cycle in the ey are ment (36 36 ). The The percentage of lipid sinking C in signatures be expected expected to percentage have soluble 15. formed The stable carbon isotope ratio is Sciences expressed as a Kashgarian, Geochim. Cosmochim. Acta 62, 1365 at the Ocean formed at either either the National National Ocean Sciences AccelAccelas those for the extractable lipids. The column was evidenced by the observation that that in the Equatorial Pacific ( 14 ). Also, our s than POC in the Arabian Sea was twice as high as relative ratio to the PeeFacility Dee Belemnite standard, (1998). Our suggest that the major portion s than in results the Sea was twice as high as as those forArabian the extractable lipids. The simiain ob- POC erator Spectrometry Woods erator Mass Mass Spectrometry Facility (NOSAMS), (NOSAMS), Woods 12C)sample/( 13C/12C)standard given by 13C [(13C/ 24. Y. Qian, site M. H.can Engel, S. influenced A. Macko, Chem. 101, sampling be by Geol. laterally of the nonhydrolyzable fraction accounted for increasing fractions ch that in the Equatorial Pacific 14 ). our Hole of the fraction is composed of alkanes ch are that in acid-insoluble thethe Equatorial Pacific ( ( 14 ). Also, Also, our , in are or- similarity larity of nonhydrolyzable fraction and Hole Oceanographic Oceanographic Institution Institution (WHOI), (WHOI), or or the the CenCen 1] Accelerator 1000. 201 (1992). ter for Mass Spectrometry (CAMS), Lawtransported old carbon from the continental nonprotein alkyl carbon in the pyrolyzates of sinking POC as depth sampling site can be influenced by laterally ter for Accelerator Mass Spectrometry (CAMS), LawReferences and Notes lipid-like material that is by of acidsampling site cancarbon be influenced by chemical laterally e acidacid- and nonprotein alkyl inbiosynthesized chemical compo16. rence E. T. Degens, M. Behrendt, B. Gotthardt, E. Reppmann, 25. J. Hwang, unpublished data. Livermore Laboratory (LLNL). a detailed Livermore Laboratory (LLNL). For a34 detailed 1. rence S. G. Wakeham, C. J. I. Hedges, P. For J. Hernes, L. shelf or slope and from inacid-soluble the California composition and abundance was observed in the Equatorial Pacific ( ) M. and lues transported old carbon from the continental similar pathways to thewas extractable lipids, but increased Deep-Sea Res. 15 , Lee, 11 (1968). 26. After the extraction of rivers lipid and fraclues to transported old carbon from the continental n (up to sition and abundance observed in sedidescription of see description of the the method, method, see ( (23 23). ).Acta 61, 5363 Peterson, Geochim. Cosmochim. 17. The M. J.percentage DeNiro, S. was Epstein, Science 197the , 261 (1977). tions, the acid-insoluble fraction was treated with coast ( 37 ). Further study will be necessary to in sediment from the continental margin of the Mediterranean Sea ( 35 ). actions shelf or slope and from rivers in the Califorsomehow is resistant to organic solvents and 21. calculated from ratio of the actions shelf or slope and from rivers in the CaliforTherement from the continental margin of north- 21. The percentage was calculated from the ratio of the (1997). 14C value is 14 12 the per mil deviation of the spec C/ C 18. Even The though HCl-HF (5 ml of 1.5 M and 5 ml of 13 fraction to the uncharacterized frac13 Mexico, and it will was suggested results of the C, NMR - confirm ifsolution our observation isHCl universal in50% the es. The nia coast ( Further study be necesacid-insoluble fraction to the uncharacterized fracacids. In Mexico, fact, lipids are enriched in resistant 2. acid-insoluble J. I. Hedges et al. , Org. Geochem. 31 945 (2000). es. The nia coast (37 37 ). ). Further study will be neces13C C northwestern e western and it was suggested that ratiothat relative to difference a standard. It is normalized to a HF) for 24 hours at room temperature, then was is between total tion is the theGeochim. difference between the the Acta total organic organic 3. tion J. I. that Hedges, Cosmochim. 42, The 69 acidsary to confirm if observation is and direct temperature-resolved oceans. results should importhat those components were compositionally parts of organisms such as membranes, of 25 tothe remove isotopic fractionation effect. acidsary tocomponents confirm if our our observation is universal universal dried These completely to remove HF. provide The extraction of those were compositionally sim- troscopy carbon and carbon and 14 the characterized characterized fractions fractions (lipids, (lipids, THAA, THAA, (1978). years. See (38). from the half-life of C is 5730 lipids and acid-soluble fractions was performed on t both in the oceans. These results should provide tant information to understand how organic to cell wallderived algaenans (33). mass spectrometry on the samples spores, and cuticles (4 ).algaenans Selective preservaand TCHO). t of both in the oceans. These results should provide the similar ilar to cell wall derived (33 ). Even and TCHO). 4. B. P. Tissot, D. H. Welte, Petroleum Formation and 19. E. R. M. Druffel, P. M. Williams, J. E. Bauer, Bauer, Deep-Sea J. R. Ertel, the demineralized fraction. 22. A. E. M. Druffel, J. chemimportant information to understand how oris preserved and removed from the Equatorial Pacific support the predom - carbon these lipids refractory lipid-like tion of though these resistant has been consid22. C. C. A. Masiello, Masiello, E. R. R.do M. not Druffel, J. E. E. Bauer, Deep-Sea chemimportant information to understand how ory pre- Even though these refractory lipid-like macromolOccurrence (Springer, Heidelberg, J. Geophys. Res. 97, 15639 (1992). Germany, ed. 2, 27. The fraction of old carbon was calculated by a massRes. II 45 , 617 (1998). Res. II 45 , 617 (1998). ose for ganic carbon is preserved and removed from 1984), part II, chap. 2. carbon cycle in theassuming ocean. that the acid-insoluble macromolecules have been studied only in inance of lipid-like material in sinking POC ered as one method of kerogen formation ( 4 ). ose for ganic is preserved from mically eculescarbon have been studied and onlyremoved in sediments, 20. X.-C. Sinking POCE. was collected in Grifn, a sediment trap balance equation 23. Wang, R. M. Druffel, S. C. Lee, M. 23. Wang,A. E.W. R. Scaroni, M. Druffel, S. Hatcher, Grifn, C. Lee, M. 5. X.-C. H. Knicker, P. sequencer G. Geocarbon cycle in the ocean. equipped with an automatic setOrg. to, collect fraction is composed of lipids, THAA, and TCHO. The refractory cell derived material of the carbon cycle in wall the of ocean. Kashgarian, Geochim. Cosmochim. Acta 62 1365 rbohy- the selective accumulation the lipids is likely Kashgarian, Geochim. Cosmochim. Acta 62, 1365 chem. 24 , 661 (1996). samples at 10-day periods. The ltered POC (1 m Weighted averages of the lipids, THAA, and TCHO (1998). portion microalgae known as algaenans portion ined if to occur inand thebacteria, water column the POC 6. (1998). M. McCarthy, T. Pratum, J. Hedges, R. Benner, Nature pore size) was dried at 50C, ground, and Geol. stored in a are 40 and 20.3 for 14C and 13C, respec24. Y. Qian, M. H. Engel, S. A. Macko, Chem. 101 , 24. Y. Qian, M. H. Engel, S. A. Macko, Chem. Geol. 101 , osed of and bacterans, respectively, were found to 390 , 150 (1997). osed of e fracsinks. Selective accumulation of the refractofreezer until analysis. The sample was cavitated with tively. The observed values of the acid-insoluble 201 (1992). (1992). References and Notes 7. 201 P. G. Hatcher, E. in C.aSpiker, N. M. Szeverenyi, G. of E. zed by an ultrasonicator 2data. :1 volume/volume mixture fraction are 23 and 23.5 for 14C and 13C, make up a large of kerogen and refracReferences andpart Notes zed and by on ry lipid-like material in the water column was 25. Hwang, unpublished 25. J. J. Hwang, unpublished data. 1. S. G. Wakeham, C. Lee, J. I. Hedges, P. J. Hernes, M. L. Maciel, Nature 305 , 498 (1983). methylene chloride:methanol, andacid-soluble then was centrirespectively. Assuming 500 and 27 for 1. S. G. Wakeham, C. Lee, J. I. Hedges, P. J. Hernes, M. L. ds, but 26. After the extraction of lipid and fractory organic carbon in sediments ( 8 , 11 , 31 ds, but 26. After the extraction of lipid and acid-soluble frachat has evidenced by the observation that 61 alkanes Peterson, , 8. tions, E. W. Tegelaar, J. W. de Leeuw, S. Derenne, C. Largeau, fuged. The acid-insoluble extraction was repeated four times, and 14C and 13C of terrestrial carbon, respectively, Peterson, Geochim. Geochim. Cosmochim. Cosmochim. Acta Acta 61 , 5363 5363 the fraction was treated tions, the Cosmochim. acid-insoluble fraction was(1989). treated with with nts and 33 ). They for are, essentially, highlyof aliphatic nts and (1997). rticles. accounted increasing fractions the pyGeochim. Acta 53 ,HCl 3103 the combined supernatant was used as the lipid the percentage R E PO R T S of terrestrial carbon (f ) can be (1997). HCl-HF solution (5 ml of 1.5 M and 5 ml of 50% HCl-HF solution (5 ml of 1.5 M HCl and 5 ml of 50% sistant 2. al. Geochem. 31 , (2000). extract. Roughly (weighed) of the residue was 9. HF) J. W.for de Leeuw, C.half Largeau, in Organic Geochemistry: calculated by 40 (1 f ) (500) f 23 lipids (i.e., large number of carbon-to-carsistant 2. J. J. I. I. Hedges Hedges et al., , Org. Org. Geochem. 31 , 945 945 (2000). either rolyzates ofaet sinking POC as depth increased 24 hours at room temperature, then was HF) for 24 hours at room temperature, then was 3. J. Hedges, Cosmochim. 42 hydrolyzed for total hydrolyzable amino acids (f 0.12 for 14C); 20.3 (1 f ) (27) 35. S. P branes, 3. the J. I. I. Equatorial Hedges, Geochim. Geochim. Cosmochim. Acta 42, , 69 69 dried branes, spendin Pacific (34 ) andActa the Medidried completely completely to to remove remove HF. HF. The The extraction extraction of of (1978). 19 hours) and the (THAA) (6 acid-soluble M HCl under fractions N2 gas for f 23.5 (f 0.48 for 13C). Salio (1978). lipids and was performed on eservalipids and acid-soluble fractions was performed on eservaorganic terranean Sea ( 35 ). other half for total hydrolyzable neutral www.sciencemag.org SCIENCE VOL 299 7 carbohyFEBRUARY28. 2003 4. Formation and R. M. Sherrell, M. P. Field, Y. Gao, Deep Sea Res. II 45, 883 (199 4. B. B. P. P. Tissot, Tissot, D. D. H. H. Welte, Welte, Petroleum Petroleum Formation and the the demineralized demineralized fraction. fraction. 13 onsid3 hours drates (TCHO) (2 hours in was 72% H2SO4, then Occurrence (Springer, Heidelberg, Germany, ed. onsid733 (1998). will be NMR spec36. X.-C Even though results of the C Occurrence (Springer, Heidelberg, Germany, ed. 2, 2, 27. 27. The The fraction fraction of of old old carbon carbon was calculated calculated by by a a massmassThe hydrolyzates were eluted in 0.6 M H2SO4). 1984), II, 2. (200 on ( 29. C. A. Masiello, E. R. M. Druffel, Global Biogeochem. 1984), part part II, chap. chap. 2. balance on (4 4 ). ). carbon troscopy and direct temperature-resolved balance equation equation assuming assuming that that the the acid-insoluble acid-insoluble through ioncomposed exchange columns for the separation of 5. A. P. Hatcher, Org. Cycles 15, 407 (2001). 37. K. L fraction 5. H. H. Knicker, Knicker, A. W. W. Scaroni, Scaroni, P. G. G. Hatcher, from Org. GeoGeorial fraction is is composed of of lipids, lipids, THAA, THAA, and and TCHO. TCHO. rial of of another mass spectrometry on the samples the THAA and TCHO. The organic carbon left after THAA chem. 30. S. M. Henrichs, Mar. Chem. 49, 127 (1995). Carl Weighted chem. 24 24, , 661 661 (1996). (1996). Weighted averages averages of of the the lipids, lipids, THAA, THAA, and and TCHO TCHO aenans aenans extraction and was used forfor the acid-insoluble fraction. 14 13 C (29). Equatorial Pacific do not support the predom6. T. J. R. Nature 31. T. Eglinton, Org. Geochem. 21, 721 (1994). 38. M. are 6. M. M. McCarthy, McCarthy, T. Pratum, Pratum, J. Hedges, Hedges, R. Benner, Benner, Nature C and and 13C, C, respecrespecare 40 40 and 20.3 20.3 for 14C Each organic carbon extract was acidied with 1 ml und to 390 (1997). 32. A. Garcette-Lepecq, S. Derenne, C. Largeau, I. Boulou(197 tively. und to to parinance of lipid-like material in sinking POC 390, , 150 150 (1997). tively. The The observed observed values values of of the the acid-insoluble acid-insoluble PO23 , dried under vacuum, and combusted of 3% H are 14 13 at P. G. E. M. G. bassi, A. Saliot, Org. Geochem. 31, 1663 (2000). 39. We fraction refrac7. P. G. Hatcher, Hatcher, E. C. C. Spiker, Spiker, N. N. of M. Szeverenyi, Szeverenyi, G. E. E. C and and 13C, C, fraction 3 are 4 23 and and 23.5 23.5 for for 14C refracand the (7. 14Maciel, , 34 ),Nature the composition sinking POC 850C for 2 hours with CuO and silver foil in a sealed 33. Y. Ge linas, J. A. Baldock, J. I. Hedges, Science 294, 145 and respectively. Maciel, Nature 305 305, , 498 498 (1983). (1983). respectively. Assuming Assuming 500 500 and and 27 27 for for 1 1, , 31 31 ds are might vary depending on the local C. environat 600C on tube. CO2 was graphitized 14 The resultant 13 8. Largeau, (2001). for carbon, 8. E. E. W. W. Tegelaar, Tegelaar, J. J. W. W. de de Leeuw, Leeuw, S. S. Derenne, Derenne, C. Largeau, C and and 13C C of of terrestrial terrestrial carbon, respectively, respectively, 14C 14C measurements were pera Co catalyst. The iphatic Geochim. Acta , (1989). iphatic ey are ment (36 ). Cosmochim. The percentage of lipid in sinking 34. E. C. Minor, S. G. Wakeham, C. Lee, Geochim. CosBau the percentage of terrestrial carbon ( f ) can be Geochim. Cosmochim. Acta 53 53 , 3103 3103 (1989). the percentage of terrestrial carbon ( f ) can be R E P O R T S R EGeochemistry: PORTS formed at either the (1 National Ocean Sciences Accelmochim. Acta, in preparation. 9. Leeuw, C. in R. W calculated to-car9. J. J. W. W. de Leeuw, C. Largeau, Largeau, in Organic Organic Geochemistry: calculated by by 40 40 (1 f f) ) (500) (500) f f 23 23 to-cares than POC inde the Arabian Sea was twice as high as erator Mass Spectrometry Facility (NOSAMS), Woods 14 ( f 0.12 for C); 20.3 (1 f ) (27) 35. S. Peulve , J. W. de Leeuw, M.-A. Sicre, M. Baas, A. for h ch are that in the and Equatorial Pacific 14 ). S. Also, our Principles Applications , M. H. ( Engel, A. Macko, Hole Oceanographic Institution (WHOI), or the Cenalkane f 23.5 (f 0.48 for 13C). Zhe Eds. (Plenum, New York, 1993), pp. 2372. ter for Accelerator Mass Spectrometry (CAMS), Law- 883 Saliot, Geochim. Cosmochim. Acta 60, 1239 ewww.sciencemag.org acidsampling site can be influenced by laterally SCIENCE VOL 299 7 FEBRUARY 2003 by a www.sciencemag.org VOL 299 C. 7 FEBRUARY 2003 Schn 28. R. M. Sherrell, M. P. Field, Y. Gao, DeepFor Seaa Res. II 45, 883 (1996). 10. C. Flaviano, SCIENCE F. Le Berre, S. Derenne, Largeau, J. rence Livermore Laboratory (LLNL). detailed lues to transported old carbon22 from the continental WH tra. A 733 (1998).of the method, see (23). 36. X.-C. Wang, E. R. M. Druffel, Mar. Chem. 73, 65 Connan, Org. Geochem. , 759 (1994). description LLN (2001). 29. A.percentage Masiello, E. R. M. Druffel,from Global actions shelf or slope and rivers in the 11. C. Largeau, J. W. defrom Leeuw, in Advances in CaliforMicrobial king of 21. C. The was calculated theBiogeochem. ratio of the the Cycles 15, 407fraction (2001). to the uncharacterized frac37. K. L. Smith, R. S. Kaufmann, R. J. Baldwin, A. F. J.37 G.). Jones, Ed. (Plenum, 1995), acid-insoluble es. The niaEcology coast, ( Further study New will York, be necesd as a and 30. S. M.that Henrichs, Chem. 49, 127the (1995). Carlucci, Limnol. Oceanogr. 46, 543 (2001). vol. 14, pp. 77117. tion is the Mar. difference between total organic e acidsary to confirm if our observation is universal cal atKom 31. T. Eglinton, Org. Geochem. 21 , 721 (1994). 38. M. Stuiver, H. A. Polach, Radiocarbon 19, 355 12. R. G. Keil, D. B. Montluc on, F. G. Prahl, J. I. Hedges, carbon and the characterized fractions (lipids, THAA, for t both in the oceans. These 32. A. Garcette-Lepecq, S. Derenne, C. Largeau, I. Boulou(1977). sms of Nature 370, 549 (1994).results should provide and TCHO). ed b A. Saliot,E. Org. Geochem. 31 , 1663 (2000). 39. We thank S. Grifn for guidance in laboratory work 13. R. A. Armstrong, C. Lee, to J. I. understand Hedges, S. Honjo, G. 22. bassi, C. A. Masiello, R. M. Druffel, J. E. Bauer, Deep-Sea ochemimportant information howS.oranisms and 33. Y. Ge II linas, J. A. Baldock, J. I. Hedges, Science 294, 145 and C. Masiello, X.-C. Wang, C. Lee, and A. Ingalls Wakeham, Deep-Sea Res. II 49, 219 (2002). Res. 45 , 617 (1998). ose for 14. ganic carbon is preserved and removed from might iden for discussion on the experiments; K. Smith, J. J. I. Hedges et al., Nature 409, 801 (2001). 23. (2001). X.-C. Wang, E. R. M. Druffel, S. Grifn, C. Lee, M. 1 Gerd G. * Sh the carbon cycle the ocean. natures 34. E. C. Minor, Geochim. S. G. Wakeham, C. Lee, Acta Geochim. CosR EThe PO RT S inisotope Bauer, D. Wolgast, R. Baldwin, R. Kochendoerfer, Glatts, F. Uhlman, 15. stable carbon ratio is expressed as a Kashgarian, Cosmochim. 62, 1365 1 mochim. R. Wilson, the 1 crews of theCressman, R/V New Horizon 17 S relative ratio to the Pee Dee Belemnite standard, (1998). Acta, in preparation. portion Fengand Mao, Sonya Stacey T e simi14 13 12 35. , J.H. W. de Leeuw, M.-A. Sicre, Baas, A., (f by 0.12 C); 20.3 (1 13 C/ f) (27) for help with sample collection; S. Trumbore and S. C [( C/12 C)sample/( C)standard given 13for 24. S. Y. Peulve Qian, M. Engel, S. A. Macko, Chem.M. Geol. 101 1 osed of n and Christie L. Hunter, Donald W. L 13C). R E P O R T S Saliot, Geochim. Cosmochim. Acta 60, 1239 f 1] 23.5 (f 0.48 for Zheng for shared equipment; A. McNichol, R. 1000. 201 (1992). References and 1 at NOSAMS, zed by 16. ompoSchneider, J.Maia Hayes,Carnevali, and colleagues 28. E. R.T. M. Sherrell, M. P. Notes Field, Y. Deep Sea Res. II 45, Degens, M. Behrendt, B. Gao, Gotthardt, E. Reppmann, 25. (1996). J. Hwang, unpublished data. Vincent Guerigui 14Lee, 1. S. G. Wakeham, C. J. I. Hedges, P. J. Hernes, M. L. 35. S. Peulve , J. W. de Leeuw, M.-A. Sicre, M. Baas, A. for help and with J. sample collection; S.1Trumbore and S. ( f 0.12 for C); 20.3 (1 f ) (27) shown WHOI, Southon and colleagues at CAMS, 733 (1998). but 36. E. R. M.ofDruffel, Mar. Chem. 73frac, 65 Deep-Sea Res. 15, 11 (1968). 26. X.-C. After Wang, the extraction lipid and acid-soluble nds, sedi13C). Peterson, Geochim. Cosmochim. Acta 61 , 5363 Heather Porter, Stephen M. Stratto 14C equipment; Saliot, Geochim. Cosmochim. Acta 60 , 1239 Zheng for shared A. McNichol, R. f 23.5 ( f 0.48 for measurement; A. Gagnon for LLNL, for the (2001). 17. J. DeNiro, S. E. Epstein, Science 197 , 261 (1977). 29. M. C. A. Masiello, R. M. Druffel, Global Biogeochem. tions, the acid-insoluble fraction was treated with polypep nts and north(1997). 1 1 14 14C/ Schneider, J. Jill Hayes, and at discussion NOSAMS, 28. The R. M. M. P. Field, Y. Gao, Deep of Sea Res. II 12 45 M.colleagues McCarthy for the 13C measurement; Cycles 15 , 407 (2001). C value is the per mil deviation the C, 18. Sherrell, HCl-HF solution of 1.5 M HCl and 5 ml of A. 50% 37. (1996). K. L. Smith, R. (5 S. ml Kaufmann, R. J. Baldwin, F. Wilken, Jie Tang, Jay J. Levy quence esistant 2. J. I. Hedges et al. , Org. Geochem. 31 , 945 (2000). 13C d that WHOI, and J. Southon and colleagues at CAMS, 733 (1998). 36. X.-C. Wang, E. R. M. Druffel, Mar. Chem. 73 , 65 and review of the manuscript; and C. Lee, T. to a standard. is, normalized to a 30. ratio S. M.relative Henrichs, Mar. Chem.It 49 127 (1995). HF) for 24 hours at room 46 temperature, then was Carlucci, Limnol. Oceanogr. , 543 (2001). 1 14C measurement; Milan M. Crnogorac, Sure 3. J. I. Hedges, Geochim. Cosmochim. Acta 42 , 69 LLNL, for the A. Gagnon for the num (2001). 29. C. A. Masiello, E. R. M. Druffel, Global Biogeochem. branes, Komada, S. Beaupre , and an anonymous reviewer to Org. remove isotopic 21 fractionation effect. The dried completely to Polach, remove Radiocarbon HF. The extraction of 31. of T. 25 Eglinton, Geochem. , 721 (1994). y sim38. M. Stuiver, H. A. 19, 355 1manuscript. (1978). 14C is M.the McCarthy for discussion the 13C measurement; Cycles 15 , 407 (2001). 37. K. L. Smith, R. S. Kaufmann, R.was J. Baldwin, A. on F. for helpful comments on Support5730 years. See 38). of lipids and acid-soluble fractions performed 32. half-life A. Garcette-Lepecq, S. Derenne, C. ( Largeau, I. Boulou(1977). Paolo Botti, # Janice Schindler-Hor In parti eserva. Even 4. E. B. P.M. Tissot, D. Mar. H. M. Welte, Petroleum Formation and and of theOceanography manuscript; and C. Lee, T. 30. S. M. Henrichs, Chem. 49, J. 127 (1995). Limnol. Oceanogr. 46, 543 (2001). ed byreview the Chemical Program of NSF 19. R. P. Williams, E. J. R. Ertel, the thank demineralized fraction. bassi, A.Druffel, Saliot, Org. Geochem. 31 , Bauer, 1663 (2000). 39. Carlucci, We S. Grifn for guidance in laboratory work a2 preci considJohn W. Adamson, Ad romolOccurrence (Springer, Heidelberg, ed. 2, Komada, S. Beaupre , and an anonymous reviewer and the University of California Ofce of the Pres31. T. Eglinton, Geochem. 21, 721 Germany, (1994). 294 38. H. old A. Polach, Radiocarbon 19 ,mass355 Res. 97 , 15639 (1992). 27. M. The Stuiver, fraction of carbon was calculated aIngalls 33. J. Y.Geophys. Ge linas, J.Org. A. Baldock, J. I. Hedges, Science , 145 and C. Masiello, X.-C. Wang, C. Lee, andby A. 1 negativ 1984), part II, chap. 2. for helpful comments on the manuscript. Supporton ( 4 ). ident (fellowship to Stephen J.H.). 32. Sinking A. Garcette-Lepecq, Derenne,in C. Largeau, I. Boulou(1977). ments, 20. POC was S. collected a sediment trap balance equation that the acid-insoluble B. H. Kent, ** James (2001). for discussion onassuming the experiments; K. Smith, J. 1 ed by the Chemical Oceanography 1 Program of NSF 5. equipped H. Knicker, A. W. Scaroni, P.sequencer G. Hatcher, Org. Geobassi, A. Saliot, Org. Geochem. , 1663 (2000). with automatic set to collect 39. We thank S.composed Grifn R. for guidance laboratory work fraction of lipids, THAA, and TCHO. site-spe 34. E. C. Minor, S.an G. Wakeham, C.31 Lee, Geochim. CosBauer, D.is Wolgast, Baldwin, R.in Glatts, F. Uhlman, erial of Gerd G. Kochendoerfer, * and Shiah-Yun Chen, likely the University ofaccepted California Ofce of the Preschem. 24 , J. 661 (1996). 33. samples Y. Ge linas, A. Baldock, J. I. Hedges, Science 294 at 10-day periods. The ltered POC (1, 145 m and C. Masiello, X.-C. C. Lee, and A.Horizon Ingalls Weighted averages of Wang, the of lipids, THAA, and TCHO mochim. Acta , in preparation. R. Wilson, and the crews the R/V New We report the design and total 17 September 2002; 7 January 2003 1 1 1 1 chemical synthe to two eaenans POC Feng Mao, Sonya Cressman, Stacey Traviglia, Haiyan Shao, 14 13 ident (fellowship to J.H.). 6. pore M. McCarthy, T. Pratum, J. Hedges, R. Benner, Nature (2001). size) was dried at 50C, ground, and stored in a respecfor discussion on the for experiments; K. C, Smith, J. are 40 and 20.3 C and protein (SEP), a 51-kilodalton protein-polymer -lev 1 1 1 und to 390 , 150 (1997). fractountil analysis. The sample was Geochim. cavitated with tively. D. The observed values acid-insoluble 34. freezer E. C. Minor, S. G. Wakeham, C. Lee, CosBauer, Wolgast, R. Baldwin, R. the Glatts, F. Uhlman, Christie L. of Hunter, Donald W. Low, E. Neil Cagle, chain and two (N 14 13 aminoacid polypeptide covale 7. an P. ultrasonicator G. Hatcher, E.in C.a Spiker, N. M. Szeverenyi, G. of E. were i 2 :1 volume/volume mixture C and C, fraction are 23 23.5 New mochim. Acta, in preparation. R. Wilson, and the and crews of the for R/V Horizon 17 September 2002; accepted 7 January 2003 refracmn was 1 1 1 Carnevali, Peter Keogh, Maciel, Nature 305, 498 (1983). methylene chloride:methanol, and then was centrirespectively.Maia Assuming 500 andVincent 27 forGueriguian, shown monodisperse in Fig. 1. SEPJ. contains a 166-amino-acid polymer moieties that are nega These s 11 , 31 lkanes 1 1 1 8. fuged. E. W. Tegelaar, J. W. dewas Leeuw, S. Derenne, C. Largeau, The extraction repeated four times, and 13C of terrestrial carbon, respectively, 14C andHeather Porter, Stephen M. Stratton, M. Con Wiedeke, polypeptide chain (Fig. 1A) similar to the secontrol the chemistry allowed us to synthesize sylation iphatic he pyGeochim. Cosmochim. Acta 53 , 3103 the combined supernatant was used(1989). as the lipid the percentage of terrestrial 1 carbon (f ) can 1 be 1 1 Jill Wilken, Jie Tang, Jay J. Levy, Epo Les P.) Miranda, quence of (SEP 16 but differs significantly dened structure. SEP was in homogeneo cision p shown in Fig. 1.covalent contains a 166-amino-acid half (weighed) of the residue was 9. extract. J. W. de Roughly Leeuw, C. Largeau, in Organic Geochemistry: calculated by 40 (1 f ) (500) f 23 to-carreased 1 1 hydrolyzed for total hydrolyzable amino acids the number and type of attached polymers ( 17 ).50,825 analytical techniques, with a mass of 1 Milan M. Crnogorac, Suresh Kalbag, ing: (i) polypeptide chain (Fig. 1A) similar to the seMedi(THAA) (6 M HCl under N2 gas for 19 hours) and the 1 1 two 2 pI In particular, branched polymer moieties of cell spacer spectrometry, and with a of 5.0. In and an quence of Epo ( 16 ) but differs significantly in Paolo Botti, # Janice Schindler-Horvat, Laura Savatski, other half for total hydrolyzable neutral www.sciencemag.org SCIENCE VOL 299 7 carbohyFEBRUARY 2003 883 a2 precise length and bearing a polymers total of activity eight SEP displayed biological ha 1 a core the number and type ofpotent attached ( 17 ). and drates (TCHO) (2 hours in 72% H2SO4, then 3 hours R specJohn W. Adamson, Ada Kung, negative charges (Fig. 1B) were designed for tion of action in vivo. These chemical methods ar SO ). The hydrolyzates were eluted in 0.6 M H linear p In particular, two branched polymer moieties of 2 4 1 1 solved Stephen B. H. Kent, ** James A. Bradburne through ion exchange columns for the separation of 1 1 site-specific attachment through an oxime bond design of and protein constructs with potential the Gerd G. Kochendoerfer, * Shiah-Yun Chen, each b a precise length bearing a total of eight om the THAA and TCHO. The organic carbon left after THAA 24 1 1 1 chemical synthesis to two noncoded amino acid residues [Lys We report 1 the design and total of synthetic erythropoiesis control negative charges (Fig. 1B) were designed for Feng Mao, Sonya Cressman, Stacey Traviglia, Haiyan Shao, extraction was used for the acid-insoluble fraction. edom 126 1 1 ( Optimal N -levulinyl) performance and Lys of protein ( N -levulinyl)] pharmaceuticals that thesis Each organic carbon extract was acidied with 1 ml 1 1 1 protein (SEP), a 51-kilodalton protein-polymer construct consisting of a 166These site-specific attachment through an oxime bond Gerd G. Kochendoerfer, * Shiah-Yun Chen, g POC Christievacuum, L. Hunter, Donald W. Low, E. Neil Cagle, and combusted of 3% H3PO4, dried under 24 37 is were primarily incorporated determined into the by an polypeptide appropriate chain. balfined 1 1at 1 polypeptide 1 and two to aminoacid covalently attached, branched, and sioned two noncoded amino acid residues [Lys 1 1 1 chain POC Feng Mao, Sonya Cressman, Stacey Traviglia, Haiyan Shao, Maia Carnevali, Vincent 850C for 2 hours with CuO and silver foil in a sealed Gueriguian, Peter J. Keogh, among 126 the ance These sites correspond specificity, to potency, two of and four pharmaglycoatom-b monodisperse polymer moieties that are negatively charged. The ability to dynam ( N -levulinyl) and Lys ( N -levulinyl)] that 1 1 1 nvirongraphitized at 600C on tube. The resultant CO2 was 1 1 E. Neil Cagle, 1 Christie L. Hunter, Donald W. Low,

Design and Chemical a Homogeneous Poly Erythropoiesis

Design and Chemical Synthesis of a Homogeneous Polymer-Modied Erythropoiesis Protein

Design and Chemical Synthesis of Design and Chemical Synthesis of a Homogeneous Polymer-Modied a Homogeneous Polymer-Modied Erythropoiesis Protein Erythropoiesis Protein

REPORTS

Two Chemically Distinct Pools Two Chemically Distinct of Organic Pools Nitrogen of OrganicAccumulate Nitrogen in the Ocean Accumulate in the Ocean
1 1 2 32 23 Lihini I. Aluwihare, *Aluwihare, Daniel J. Repeta, Silvio Carl G. Johnson * Daniel Lihini I. Silvio Pantoja, J. Pantoja, Repeta,

The chemical dynamics of marine dissolved organic nitrogen (DON), a reservoir featuring chemical dynamics of marine organic nitrogen (DON), a to resersurfaceThe accumulations even in areas wheredissolved nitrogen limits productivity, have yet be voir featuring surface accumulations in of areas where nitrogen limits proresolved. We exploited differences in the acid even lability amide bonds within high-molecularhave yet to be resolved. exploited differences infrom the acid lability of weightductivity, (HMW) DON to show that vertical We DON profiles result in part the presence of amidedistinct bonds within high-molecular-weight (HMW) DON to show ver-as two chemically pools of amide. Half of HMWDON in surface waters isthat present tical DON profiles result in from the presence of two chemically distinct N-acetyl amino polysaccharides. In part contrast, nearly all deep-sea HMWDON, and therepools of amide.isHalf of HMWDON in surface waters is present as Nand -acetyl fore, most HMWDON, present in amides that resist both chemical hydrolysis amino polysaccharides. In contrast, nearly all deep-sea HMWDON, and therebiological degradation. fore, most HMWDON, is present in amides that resist both chemical hydrolysis and biological degradation.
Low concentrations of inorganic nitrogen (such as nitrates and ammonia) are assumed to limit primary production over wide expanses of the surface ocean. However, in many of these areas, dissolved organic nitrogen (DON) accumulates to measurable quantities (13) despite a demonstrated role in fueling both primary and secondary production (4). Given the importance of nitrogen for limiting ocean productivity, mechanisms that regulate DON production and removal could help control both the ocean_s N balance and, consequently, the sequestration of atmospheric carbon dioxide. Processes that lead to DON accumulation in seawater are unclear, but vertical profiles show that upper ocean DON concentrations are enhanced by 30 to 50% over deep water values (5). This observation suggests a major source for DON in the upper ocean and is consistent with findings that a large fraction of inorganic N assimilated by marine phytoplankton can be returned to seawater as DON (6). However, an important step for explaining DON profiles in the ocean is to identify compositional features that differentiate DON fractions with diverse biological reactivity. Here we report evidence for major structural differences between the DON pools of the surface and the deep ocean. About 30% of DON occurs as a highmolecular-weight fraction (HMWDON) that can be sampled by ultrafiltration (6, 7). The depth profile of HMWDON is similar to that of total DON, with high near-surface concentrations. Despite a decline in HMWDON con1 Geosciences Research Division, Scripps Institution of Oceanography, La Jolla, CA 92093, USA. 2Department of Marine Chemistry and Geochemistry, Woods Hole Oceanographic Institution, Woods Hole, MA 02543, USA. 3Department of Oceanography and Fondap Copas n, Casilla 160-D, Center, University of Concepcio n, Chile. Concepcio

Carl G. Johnson2

centrations below the mixed layer, nuclear magnetic resonance (NMR) spectra and amino acid analyses imply a homogenous chemical composition throughout the water column (8, 9). Nitrogen-15 NMR (15N-NMR) spectroscopy shows that nearly all HMWDON in the ocean is chemically bound as amide functional groups (8). Amides are commonly found in living marine organisms within proteins and biopolymers of N-acetyl amino polysaccharides (N-AAPs, e.g., chitin and peptidoglycan). However, acid hydrolysis of HMWDON yields only small amounts of amino acids and amino sugars, suggesting that HMWDON is deficient in these polymers (9, 10). However, previous studies have also demonstrated that it is difficult to draw quantitative conclusions from analyses that require depolymerization of HMWDON compounds (10). Rather than relying on a hydrolysis method that efficiently depolymerizes polysaccharides

and proteins, we exploited differences in the amide bonds of proteins and N-AAPs to quantify the contribution of these biochemicals to marine DON. Amide bonds in proteins are an integral part of the peptide linkage, and amide hydrolysis depolymerizes proteins to yield amino acids (Fig. 1). Amide bonds in N-AAPs are not an integral part of the glycosidic linkage, and amide hydrolysis of N-AAPs is not directly coupled to depolymerization. Instead, amide hydrolysis de-acetylates the polysaccharide, releasing 1 mole of acetic acid for each mole of amide-N that is hydrolyzed (Fig. 1). Therefore, proteins can be quantified and distinguished from N-AAPs by the products of acid hydrolysis (amino acids and acetic acid, respectively). In both cases (proteins and N-AAPs), the hydrolysis of the amide bond produces amine-N. We used solid-state (cross-polarization magic-angle spinning) 15NNMR spectroscopy to follow the hydrolysis of HMWDON amide and quantify the conversion of amide-N to amine-N. Concurrent measurements of acetic and amino acid generation were used to partition HMWDON into proteins and N-AAPs. These experiments allowed us to construct a budget of nitrogencontaining biopolymers in marine HMWDON. High molecular weight dissolved organic matter (HMWDOM) in surface seawater is rich in carbohydrate (60 to 80% of total C) and acetate (5 to 7% of total C), as seen in the 1H-NMR spectrum for Woods Hole surface seawater (Fig. 2A) Ecarbohydrates, 5.2, 4.53.2, and 1.3 parts per million (ppm); acetate, 2 ppm^. Previous studies have shown that the chemical composition of Woods Hole HMWDOM is representative of marine HMWDOM in general (10, 11). The 15N-NMR for this sample (Fig. 2B) shows one major resonance at 124 ppm for amide-N (92% of total N) and a minor resonance at 35 ppm for amine-N (8% of total N). We hydrolyzed

*To whom correspondence should be addressed. E-mail: laluwihare@ucsd.edu

Fig. 1. Schematic showChitin Protein ing the effect of mild O acid hydrolysis on the O O R O amide linkage of proH NH teins and N-AAPs (chiN NH N tin). Mild acid hydrolysis H O H3C R' O (13) completely destroys H3C O the amide linkage (gray shaded area) in an NAAP and quantitativeacid ly releases acetic acid. O However, the glycoNH 2 O O sidic linkage remains H O O unaffected, and the R' NH2 macromolecule is not NH2 + amino acids depolymerized. Mild acid + + R hydrolysis could likeHO HOCCH3 HOCCH3 wise destroy the amide NH2 linkage in proteins (gray acetic acid O O shaded area), thereby O depolymerizing the macromolecule to release amino acids. If proteins and N-AAPs are the major biochemical components of HMWDON, then the sum of acetic and amino acids recovered after mild acid hydrolysis should equal the amount of amide destroyed during the reaction.

38

www.sciencemag.org VOL OriginallySCIENCE published 13 May308 2005 13 MAY 2005

1007

REPORTS
Woods Hole HMWDON using conditions that were considerably milder than those typically employed for peptide bond hydrolysis (12) but that quantitatively release acetic acid from N-acetyl glucosamine and chitotriose (13). As expected in the case of N-AAPs, the acid hydrolysis (13) removed acetic acid from our samples (Fig. 2C) in quantifiable yield (Table 1) (14). Concurrent with the loss of acetic acid, there was a decrease in amide-N from 92% of the total N (7.1 mmol of N) to 35%, and an increase in the amount of amine-N from 8% of the total (0.6 mmol of N) to 65% (5 mmol of N) N (Fig. 2D and Table 1). Amino acid analysis showed that some proteins were hydrolyzed to amino acids that were recovered at 13% (Table 1) of the total N (15). The amount of amide-N converted to amine-N during acid hydrolysis, as quantified by 15N-NMR spectroscopy, nearly equaled the molar sum of acetic and amino acids recovered by molecular level techniques (4.4 versus 4.3 mmol) (Table 1). The agreement between NMR spectroscopy and these molecular level analyses confirms that the hydrolyzed amide was originally present in HMWDON as proteins and NAAPs. Under stronger hydrolysis conditions (12), we were able to recover 21% (1.6 mmol of N) of the total N in the sample as amino acids. Using acetic acid as a proxy for N-AAPs, our results indicate that HMWDON in surface seawater is 43% N-AAPs (Table 1), 21% hydrolyzable protein, and 29% (2.2 mmol of N) nonhydrolyzable amide (16). The remaining HMWDON in our Woods Hole sample (8%, 0.6 mmol of N) initially present as amine was not characterized by our analyses but may be present as basic or N-terminal amino acids of proteins or as amino sugars. We obtained a similar agreement between the quantity of N converted from amide to amine and the molar sum of amino acids plus acetic acid recovered by molecular level analyses for HMWDON sampled from the North Pacific Ocean (Table 1); the agreement confirms the ubiquity of N-AAPs. Our analyses demonstrate that soluble N-AAP biopolymers contribute 26% of the carbon and 40 to 50% of the nitrogen to surface ocean HMWDOM (16). Peptidoglycan, currently assumed to dominate the oceanic reservoir of HMWDON, is rich in N-acetyl glucosamine and N-acetyl muramic acid (present in a 1:1 ratio) (9). Both sugars contain amide-N and are potential sources for the N-AAPs we quantified by our NMR experiment. The presence of acetylated amino sugars in HMWDON has been inferred previously from mass spectrometric data (17, 18), but recoveries of N-acetyl glucosamine and N-acetyl muramic acid, which make up the amide-rich glycan of peptidoglycan, are low (10, 11, 19). Even under conditions specific for peptidoglycan hydrolysis (20, 21), we were unable to detect muramic acid in our samples. Glucosamine, though present, contributed G1% of the total carbon, far lower than the 10% HMWDOM-carbon ex-

Fig. 2. The effect of mild acid hydrolysis (13) on the amide linkages of HMWDON isolated from Woods Hole surface seawater. (A) The 1H-NMR spectrum [300 MHz, D2O, d (chemical shift in ppm downfield from tetramethylsilane)] shows bound acetic acid (*, 2.0 ppm). (B) The 15N-NMR spectrum [400 MHz, solid-state, d (chemical shift in ppm downfield from liquid NH3)] before acid hydrolysis shows most nitrogen (92%) bound as amide. (C) Mild acid hydrolysis and organic extraction removes acetic acid, demonstrating complete de-acetylation. (D) This hydrolysis converts 57% of the amide-N into amine, commensurate with the amount of acetic and amino acids recovered by molecular level analyses (56%) (Table 1) and increases the total amine-N to 65%. Pyridine added to quantify losses during sample processing showed 98% recovery of HMWDON. IS, internal standard.

Table 1. Change in amide content and yields of acetic and amino acids, upon mild acid hydrolysis of HMWDON. All percentages are expressed relative to total mmol of N. Sample Woods Hole (5 m) MAB (1000 m) Hawaii (23 m) Hawaii (600 m) Total (mmol N)* 7.7 7.7 6.2 6.2 Amide (mmol N) 7.1 7.7 6.2 6.2 (92%) (100%) (100%) (100%) D Amide (mmol N)y 4.4 1.8 3.7 3.7 Acetic acid (mmol) 3.3 1.3 3.3 3.4 (43%) (17%) (53%) (55%) Amino acids (mmol) 1.0 0.8 0.5 0.4 (13%) (10%) (8%) (6%)
Acetic amino acids S (mmol)

Unhydrolyzed amide (mmol N)z 2.7 5.8 2.4 2.4 (35%) (76%) (39%) (39%)

4.3 2.1 3.8 3.8

*Values are expressed per 100 mmol of HMW dissolved organic carbon and calculated based on the C/N ratio of each sample. .D Amide is the change in the amount of amide-N after hydrolysis, as quantified by integration of the 15N-NMR spectra. In all cases, hydrolysis resulted in the loss of amide-N. -The quantity (mmol of N) of amide remaining in the 15 sample after mild (13) hydrolysis, quantified directly by N-NMR spectroscopy. Higher concentrations of HCl (12) hydrolyzed a greater percentage of the amide-N, leaving slightly less (2.2 and 5.5 mmol) amide-N unhydrolyzed in the Woods Hole and MAB samples, respectively. In order to quantify the amount of amide-N remaining after strong acid hydrolysis the Sacetic acid (mild) amino acids (strong) was first determined and then subtracted from the initial amide content of the HMWDON sample (before hydrolysis) (16). Here, acetic acid is being used as a proxy for N-AAPs.

1008

13 MAY 2005 VOL 308

SCIENCE

www.sciencemag.org

39

REPORTS
pected if all of the acetic acid in our samples was from peptidoglycan. Possible explanations for the general discrepancy between NMRderived estimates of N-AAPs and molecularlevel carbohydrate analyses are incomplete depolymerization of N-AAPs and rapid Maillard condensation reactions of hydrolysis products (22). Muramic acid also contains lactic acid, which can be quantitatively released from peptidoglycan without depolymerizing the glycan (23). For peptidoglycan, 1 mol of lactic acid will be released for every 2 mol of acetic acid. Concentrations of lactic acid (14) in our samples were G0.4% of HMWDOM-carbon, an order of magnitude less than expected if all of the acetic acid we recovered was from peptidoglycan. The low concentration of lactic acid measured in this study and the low concentration of D-amino acids previously reported for HMWDON (9) together suggest that peptidoglycan is at best only a minor component of HMWDON. In addition, the D-amino acids in HMWDON so far identified to be present in peptidoglycan (9) could be present in a number of compounds synthesized by prokaryotes and eukaryotes (24, 25). Because N-AAPs represent 40 to 50% of the HMWDON in surface waters, and given that 40 to 50% of HWMDON is removed below the mixed layer (6, 8), we hypothesize that the global decrease in HMWDON with depth in the oceans results from the selective removal of N-AAPs. To test this hypothesis, we analyzed HMWDOM collected from a depth of 1000 m in the Middle Atlantic Bight (MAB), which has radiocarbon, 1H-NMR, and molecular-level properties characteristic of deep-sea HMWDOM (10). In agreement with previous 15N-NMR spectra of deep sea HMWDON (8), the 15N-NMR spectrum of our sample shows one major resonance characteristic of amide-N (100% of total N) (Fig. 3A). Mild acid hydrolysis (13) decreased amide-N from 100% to 76% of total N and increased amine-N from undetectable levels to 24% of total N (Fig. 3B and Table 1). After acid hydrolysis, acetic acid was released from HMWDOM, as were amino acids, which increased from undetectable levels to 10% of total N. The sum of acetic plus amino acids (2.1 mmol) in the hydrolysis products was similar to the increase in amine-N (1.8 mmol of N) observed with 15N-NMR spectroscopy. Molecular-level analyses after strong acid hydrolysis showed deep-sea HMWDON was 17% N-AAPs (Table 1), 12% hydrolyzable protein, and 71% nonhydrolyzable amide (5.5 mmol of N) (16). Despite the surface water abundance, a large fraction of N-AAPs are lost during mixing into the deep ocean. Proton NMR spectra of numerous deep-sea samples showed a sharp decrease in acetate below the mixed layer, confirming the loss of NAAPs with depth (Fig. 4). The loss in N-AAPs and the relative increase in nonhydrolyzable amides will not result in any change to the 15N-NMR spectrum (8). The lack of amines in marine HMWDON implies that amine-N is more labile than amide-N. Many marine microorganisms have cell surfacebound deaminases that are capable of extracting amine-N from a variety of organic compounds (26). These enzymes could render HMW amine-N more biologically available by allowing organisms to bypass polymer hydrolysis or uptake. The eventual depth-dependent loss of N-AAPs from HMWDON suggests either that N-AAP compounds are labile or that organisms have developed a mechanism to access the nitrogen in N-AAPs without complete hydrolysis of the polymer (perhaps through cell surfacebound acetamidases). Although we were unable to depolymerize N-AAPs using acid hydrolysis, the capacity to enzymatically degrade N-AAPs may be widespread among marine microbes. In particular, the ability to hydrolyze chitin Eb (1Y 4)-(poly) N-acetyl D-glucosamine^ is notable in members of the marine aproteobacteria and the Cytophaga-Flavobacter cluster (27, 28). The widespread occurrence of chitinase activity and the ubiquity of chitinase genes in marine bacteria (29) imply that chitin-like biopolymers are important substrates in the marine environment, consistent with the abundance of N-AAPs in HMWDON. The presence of specialized bacteria could explain the ultimate removal of N-AAPs from the marine environment. More than 90% of DON is sequestered in the deep sea, and most deep-sea HMWDON

Surface

600m

Amide N (100%)

B
Pyridine (IS) Amide N (76%)

Amine N 24%
3600m

7
300 20 0 100 0 300 20 0 100 0

Fig. 3. The effect of mild acid (13) hydrolysis on the amide linkage of deep-sea HMWDON. (A) The 15N-NMR spectrum of HMWDON shows only one resonance for amide-N [400 MHz, solid state d (chemical shift in ppm downfield from liquid NH3)]. (B) Treatment of the sample with mild acid converts 24% of the nitrogen to amine but does not affect most (76%) of the nitrogen. Pyridine (IS), added as an internal recovery standard, showed 998% N recovery through the hydrolysis procedure. Deep-sea HMWDON may also include some (G13% total N) contribution from pyrroleand indole-N, which resonate between 130 and 215 ppm and overlap with the amide region in our 15N-NMR spectrum.

Fig. 4. The 1H-NMR spectra [400 MHz, D2O, d(ppm)] of HMWDOM isolated from several depths (surface, 600 m, and 3600 m) in the Pacific Ocean (31-N, 159-W; the total water column depth was 5770 m). The peak at 2.0 ppm (*) arises from acetate, which is presumed to be present in N-AAPs. The clear decrease in the relative amount of acetate with depth is interpreted as a depth-dependent loss of N-AAPs.

40

www.sciencemag.org

SCIENCE

VOL 308

13 MAY 2005

1009

DOM interactions may lead to structural HMWDON pool that exists throughout the mixing. These newly added biopolymers deriv acetylated) or amino sugar monomers were detected. Cosmochim. Acta 59, 2643 (1995). the Woods Hole and MAB samples, respectively. In (mild modifications that render proteins resistant to 33. 22. J. I. H 14. D. B. Albert, S. Martens, Mar. Chem. 56 , 27 (1997). water column. If we assume the proportion are chemically distinct from the refractory We thank A. C. Beilecki, then at Bru ker Instruments, for order to quantify the amount of amide-N remaining then 15N-NMR 23. O. H 15. M. Zhao, J.with L. Bada, J. Chromatogr. A. 690 , 55 (1995). chemical hydrolysis and unavailable to bacafter strong acid hydrolysis, the sum Sacetic acid spectroscopy; M. Pullin and of assistance N-AAPs, protein, and nonhydrolyzable HMW HMWDON R E P O R Tpool S that exists throughout the 24. Y. Na 16. D. The percentage of N inin each sample represented by (mild) mechanism amino acids (strong) was first determined and Albert for assistance the determination of acetic perce teria. This could lead to the sequesamide measured in our samples is representwater column. If we assume the proportion T. Ya N-AAPs, hydrolyzable protein, nonhydrolyzable then subtracted from the initial amide content of the acid; and the staff at the Naturaland Energy Laboratory in samp 19. K. Kaiser, R. Benner, Anal. Chem. 72 , 2566 (2000). 12. S. Henrichs, P. M. Williams, Mar. Chem. 17 , 141 (1985). tration of nitrogen in the dissolved phase ( 32 ) is not hydrolyzed by treatment with strong acids ative of Hawaii, global HMWDON, as much 25. J. S. M amide was calculated as follows: Because 1 sample mol as of of N-AAPs, protein, and nonhydrolyzable HMWDON sample (before hydrolysis). In all cases, Kona, and E. Smith for then assistance in with 20. N D. L. Popham, J. Helin, C. E. Costello, P. Setlow, J. 13. De-acetylation of HMWDON samples was performed 26. B. Pa -AAP is de-acetylated for everyand mole ofwith acegive rise to the relative hydrolysis-resistant (Table 1). Hydrolysis-resistant amides have been and percentages are expressed to total N in each collection. Supported by the Biological 80% of sugar the decrease in Chemical HMWDON assum amide measured in our samples is representBacteriol. 178 , 6451 (1996). in 1 N HCl. Samples were heated overnight at 90 C (199 tic acid released, we assumed that mmoles of Founacetic 15N-NMR. sample. observed We calculated the amount of N-AAP carbon Oceanography Programs at the National Science acety amide-N by observed in marine particulate matter and seddepth the removal -AAPs. 21. We involves modified the method provided in ( 20 ) andThe hyunder an atmosphere of N2 . The acetic acid produced ative of global HMWDON, then as much as 27. acid were equal to mmoles ofof NN in N -AAPs; hywith the C/N ratio in HMWDOM (Table 1) and dation; the Carbon Sequestration Program at the U.S. 17. M. J. J.T. B drolyzed samples (in 4 N HCl) overnight at 90 C. We during the hydrolysis was extracted from a weighed Our data show that two chemically distinct iments, where the resistance to chemical hydrolabundance of amide-N throughout the water 66, 1 drolyzable protein N was calculatedCoastal on theResearch basis of 80% of the decrease in HMWDON with assumed 8 mmol of C per mmol of N-AAP (e.g., NDepartment of Energy; the Rhinehart 1051 recovered no muramic acid and only small amounts aliquot of the hydrolysate into either ethyl ether or 28. L. the recoveries of amino N after strong acid pools ofglucosamine). organic nitrogen accumulate in the ysis has been attributed to physical sorption column suggests amides are more biologically acetyl Center of the Woods Hole acid Oceanographic Institution; depth involves the removal of N-AAPs. The 18. A. J. A. of glucosamine and galactosamine amide from surface samdichloromethane. After extraction, the presence of 29. M. T hydrolysis (12); nonhydrolyzable was detern J. J. Boon, V. A. Klap, T. I. Eglinton, Org. Geochem. 29, and the Fundacio Andes, Chile. ocean. The higher HMWDON Bioge and encapsulation (30). throughout Amide-N in DON is 17. recalcitrant than other forms of organic-N. R EP ORT abundance ofSamide-N the water ples. This modified hydrolysis method was tested on acetic acid in the concentration organic fraction of was confirmed by Appl. mined after strong acid hydrolysis (Table 1). Higher 1051 (1998). 1 in the mixed layer (relative to deep ocean val- 20 chitin oligomers, peptidoglycan (from Bacillus subtilus ), solution-state H-NMR before low molecular weight not physically protected, butmore previous experi- 18. The ubiquity of accepted amide linkages in HMWDON 30. H. Kn amino acid yields were obtained when HMWDON was column suggests amides are biologically A. Leenheer, T. I. Noyes, C. E. Rostad, M., L. Davisson, 2004; 9Chem. March 2005 12. J. S. Henrichs, P. M. Williams, Mar. Chem. 17 141 (1985). 19.December K. Kaiser, R. Benner, Anal. 72 , 2566 (2000). is not hydrolyzed by treatment with strong acids and bacterial cells (mixed laboratory culture) to ensure acids (e.g., acetic acid) were quantified. The aqueous (1997 hydrolyzed with strong acid. As most a result, after strong ues) largely reflects the presence of -AAPs, 10.1126/science.1108925 ments havethan shown that the biodegradation is not surprising, given that organic ni69HMWDON , 125 (2004). recalcitrant other forms of organic-N. 13. Biogeochemistry De-acetylation samples wasN performed 20. D. L. Popham, J.(9 Helin, C. E. Costello, P. Setlow, J. the quantitative 90%) recovery of glucosamine and fraction of theof hydrolysate was lyophilized and, as (Table Hydrolysis-resistant amides have been 31. R. G. acid hydrolysis, only 2.2 (29% of total N) and 5.5 mmol which on time scales of upper ocean rate ubiquity of1). labile compounds such asHMWDON proteins is trogen in phytoplankton is protein. However, in 1 degrade N in HCl. Samples were heated overnight at 90-C Bacteriol. 178 , 6451 (1996). 1H-NMR The of amide linkages in de-acetylated muramic acid. We assessed amino sugar shown the (Fig. 2C), no longer contained 32. E. Ta (71% of total N) of amide-N remained unhydrolyzed in observed in marine particulate matter and sedunder an atmosphere ofC N2 21. We modified contribution the method provided in chromatogra(20) and hy.added The acid produced mixing. These newly biopolymers substantially reduced by abiotic complexation recoveries high-performance acetamide. Changes in and Nacetic were assessed with thethe important ofliquid N-AAPs to upis not surprising, given that most organic niCosm Woodsby Hole and MAB samples, respectively. In drolyzed samples (in 4 N HCl) overnight at 90 C. We during the hydrolysis was extracted from a weighed 13C- and 15N-NMR spectroscopy. Under iments, where the resistance to chemical hydrolphy and of o-phthaldialdehydeare solid-state chemically distinct from the ethyl refractory to fluorescence quantify thedetection amount of amide-N remaining within marine DOM (31).is Long-term proteinperorder ocean HMWDOM, and the of 33. We t trogen in R phytoplankton protein. However, recovered no muramic and onlyresistance small amounts aliquothydrolysis of the hydrolysate into ether or R EP O TS derivatized samples (15)acid or gas chromatography these conditions, noeither muramic acid (deysis has been attributed to lead physical sorption assist after strong acid hydrolysis, the sum Schemical acetic after acid HMWDON pool that exists throughout the DOM interactions may to structural amides in deep sea HMWDOM to of glucosamine and galactosamine from surface samdichloromethane. After extraction, the presence of the important contribution of N-AAPs to upderivatizing to alditol acetates (10 ). determined and acetylated) or amino sugar monomers were detected. (mild) amino acids (strong) was first D. Al and encapsulation ( 30 ). Amide-N in DON is ples. This modified hydrolysis method was tested on acetic acid in the organic fraction was confirmed by 19. K. Kaiser, R. Benner, Anal. Chem. 72 , 2566 (2000). 12. S. Henrichs, P. M. Williams, Mar. Chem. 17 , 141 (1985). R Eocean P hydrolyzed O RHMWDOM, T Sthat water column. IfMartens, we assume the 56 proportion is not byrender treatment with strong acids modifications proteins resistant to and biological degradation, are unexpected 22. J. I. Hedges et al. , Org. Geochem. 31, content 945 (2000). 14. D. B. Albert, C. S. Mar. Chem. , 27 (1997). per and the resistance of then subtracted from the initial amide of the acid; C 1H-NMR before low molecular weight chitin oligomers, peptidoglycan (from Bacillus subtilus ), solution-state 20. D. L. Popham, J. Helin, C. E. Costello, P. Setlow, J. 13. De-acetylation of HMWDON samples was performed not physically protected, previous experi23. O. Hadja, Anal. Biochem. 60hydrolysis). , 512 (1974). 15. N M.-AAPs, Zhao, J. L. protein, Bada, J. Chromatogr. A. 690, 55 (1995). of and nonhydrolyzable (Table 1). Hydrolysis-resistant amides have been chemical hydrolysis and but unavailable to bacHMWDON sample (before In all cases, Kona results thatR. help elucidate the currency of amides in deep sea HMWDOM to chemical and bacterial cells (mixed laboratory culture) to ensure acids (e.g., acetic acid) were quantified. The aqueous Bacteriol. 178 , 6451 (1996). in 1 N HCl. Samples were heated overnight at 90 C 19. K. Kaiser, Benner, Anal. Chem. 72 , 2566 (2000). 12. S. Henrichs, P. M. Williams, Mar. Chem. 17 , 141 (1985). is not hydrolyzed by treatment with strong acids amide 24. percentages Y. Nagata, T. are Fujiwara, K. Kawaguchinagata, Y.N Fukumori, 16. The percentage of N in each sample represented by ments have shown that the biodegradation expressed relativecycle. to total in each collec measured in our samples is representobserved in marine particulate matter and sedteria. This mechanism could lead to the sequesDON in the marine nitrogen I 1 2 2 1 1 the quantitative ( 9 90%) recovery of glucosamine and fraction of the hydrolysate was lyophilized and, as 21. We modified the method provided in ( 20 ) and hyunder an atmosphere of N . The acetic acid produced 20. D. L. Popham, J. Helin, C. E. Costello, P. Setlow, J. 13. De-acetylation of HMWDON samples was performed and biological degradation, amides are unexpected T. Yamanaka, Biochim. Biophys. Acta 1379 , 76 (1998). N-AAPs, hydrolyzable protein, and nonhydrolyzable 2 C. Frankenberg, J. F. Meirink, M. van Weele, Platt, Wagner (Table 1). Hydrolysis-resistant have been sample. We U. calculated the T. amount of N-AAP carbon Ocea rate of labile compounds such as proteins is 1H-NMR (Fig. 2C), de-acetylated muramic acid. We assessed amino sugar shown in the no longer contained drolyzed samples (in 4 N HCl) overnight at 90 C. We during the hydrolysis was extracted from a weighed Bacteriol. 178 , 6451 (1996). in 1 N HCl. Samples were heated overnight at 90 C ative of global HMWDON, then as much iments, of where the resistance tothe chemical hydroltration nitrogen in the dissolved phase (32 ) n 25. with J. S. Martinez etratio al., Science 287, 1245 (2000). amide was calculated as follows: Because 1 mol as of results that help elucidate currency of the C/N in HMWDOM (Table 1) and datio observed in marine particulate matter and sed- 80% substantially reduced by to abiotic complexation recoveries by E. high-performance chromatograacetamide. Changes in and Neither were assessed with recovered no muramic acid and liquid onlyin small aliquot of the ether or 21. B. We modified the method provided (20) amounts and hyunder an atmosphere ofC Nin The acetic acid produced References and 26. Palenik, Henson, Limnol. , 1544 N -AAP sugar ishydrolysate de-acetylated for everyethyl mole of ace2.into of the decrease HMWDON with ysis has been attributed physical sorption and give rise to nitrogen the hydrolysis-resistant assumed 8 S. mmore mol ofNotes C per mmolOceanogr. of N-and AAP 42 (e.g., NDepa g DON in the marine cycle. In the past two centuries, atmospheric methane has than doubled 13 15N-NMR phy and fluorescence detection of o -phthaldialdehydesolid-state Cand spectroscopy. Under of glucosamine and galactosamine from surface samdrolyzed samples (in 4 N HCl) overnight at 90 C. We dichloromethane. After extraction, the presence of during the hydrolysis was extracted from a weighed 1. acetyl J. Abell, S. Emerson, P. Renaud, J. Mar. Res. 58, 203 iments,marine where the resistance to chemical hydrol(1997). tic acid released, weremoval assumed that mmoles of acetic within DOM (31 ).N-NMR. Long-term 15 glucosamine). Cente involves the of N -AAPs. The and encapsulation (30 ). Amide-N in proteinDON is depth amide-N observed by g derivatized samples ( 15 ) or gas chromatography after these hydrolysis conditions, no muramic acid (deples. This modified hydrolysis method was tested on acetic acid in the organic fraction was confirmed by recovered no muramic acid and only small amounts now constitutes 20% of the anthropogenic climate forcing by greenhouse aliquot of the hydrolysate into either ethyl ether or (2000). 27. J. M. Cottrell, D. L. Kirchman, Appl.Org. Environ. Microbiol. acid were equal to mmoles of N in N-AAPs; hyysis has been and attributed to lead physical sorption abundance 17. J.T. Boon, V. A. Klap, T. I. Eglinton, Geochem. 29, and t DOM interactions may to structural 1H-NMR Notes of amide-N throughout the water derivatizing to alditol acetates (10 ). acetylated) or amino monomers were detected. chitin oligomers, peptidoglycan (from Bacillus subtilus ), solution-state before low molecular weight of and galactosamine from surface samdichloromethane. After extraction, the presence of notReferences physically protected, but previous experiOur data show that two chemically distinct 2. 1051 M.,glucosamine J. (1998). Church, H. W. Ducklow, D. M. Karl, Limnol. b 66 1692 (2000). drolyzable protein N sugar was calculated on the basis of gases. Yet its sources are not well quantified, introducing uncertainties in its and (P. 30 ). Amide-N in DON is column modifications thatnitrogen render proteins to 1. J. encapsulation Abell, S. Emerson, Renaud, J. Mar. resistant Res. 58, 203 22. A. J. I. et al. Org. Geochem. 31 , 945 (2000). 14. the D. B.recoveries Albert, C. the S. Martens, Mar. Chem. 56 , 27aqueous (1997). and bacterial cells (mixed laboratory culture) to(1998). ensure acids (e.g., acid) were quantified. The ples. This modified hydrolysis method was tested on acetic acid acetic in organic fraction was confirmed by Oceanogr. 47 ,T. 1I.,(2002). 28. L.Hedges Svitil, D. L. Kirchman, Microbiol. 144 , 1299 of amino acid N after strong acid suggests amides are more biologically ments have shown that the biodegradation pools of organic accumulate in the l 18. J. A. Leenheer, Noyes, C. using E. Rostad, M. L. Davisson, 20 Decem global budget. We retrieved the global methane distribution by space1H-NMR (2000). 23. O. Hadja, Anal. Biochem. 60 , 512 (1974). 15. M. Zhao, J. L. Bada, J. Chromatogr. A. 690 , 55 (1995). the quantitative ( 9 90%) recovery of glucosamine and fraction of the hydrolysate was lyophilized and, as chitin oligomers, peptidoglycan (from Bacillus subtilus ), solution-state before low molecular weight not physically protected, previous to experi3. Biogeochemistry P. Libby, P. Wheeler, Deep-Sea 345 (1997). chemical hydrolysis and but unavailable bac29. M. T. Cottrell, D. Wood, L. Res. Yu, 44 D. , L. Kirchman, hydrolysis than (12); nonhydrolyzable amide was deter69,N. 125 (2004). 10.1126/n other forms of organic-N. rate labile compounds such asHMWDON proteins is recalcitrant ocean. higher 1H-NMR 2. M.of J.The Church, H. concentration W. Ducklow, D. of M. Karl, Limnol. borne near-infrared absorption spectroscopy. In addition to the expected 24. Y. Nagata, T. Fujiwara, K. Kawaguchinagata, Y. Fukumori, 16. The percentage of N in each sample represented by and bacterial cells (mixed laboratory culture) to ensure de-acetylated muramic acid. We assessed amino sugar shown in the (Fig. 2C), no longer contained acids (e.g., acetic acid) were quantified. The aqueous 4. Appl. N. J. Antia, P. J.Microbiol. Harrison, L. Phycologia 30, 1 Environ. 66Oliveira, , 1195 (2000). mined after strong acid hydrolysis (Table 1). Higher ments have shown that the biodegradation teria. This mechanism could lead to the sequesTheamino ubiquity of amide linkages in HMWDON Oceanogr. 47 , 1 (2002). T. Yamanaka, Biophys. Acta , 76 (1998). NAAPs, hydrolyzable protein, and nonhydrolyzable substantially reduced by abiotic complexation in the mixed layer (relative to deep ocean valrecoveries by high-performance liquid chromatograacetamide. Changes inobtained C and N were assessed with the quantitative (990%) recovery of1379 glucosamine and fraction of the hydrolysate was lyophilized and, as m (1991). 30. H. Knicker, P. Biochim. G. Hatcher, Naturwissenschaften 84, 231 acid yields were when HMWDON was latitudinal gradient, we detected large-scale patterns of anthropogenic and rate of labile compounds such as, proteins is 13 15 1H-NMR tration of nitrogen in the dissolved phase ( 32 ) 3. P. Libby, P. Wheeler, Deep-Sea Res. 44 345 (1997). 25. J. S. Martinez et al. , Science 287 , 1245 (2000). amide was calculated as follows: Because 1 mol of phy and fluorescence detection of o -phthaldialdehydesolid-state Cand N-NMR spectroscopy. Under de-acetylated muramic acid. We assessed amino shown in the (Fig. 2C), no longer contained 5. (1997). T. Berman, D. A. Bronk, Aquat. Microb. Ecol. 31sugar , 279 is not surprising, given that most organic nihydrolyzed with strong acid. As a result, after strong within marine DOM ( 31). Long-term proteinues) largely reflects the presence of N -AAPs, i natural methane emissions. Furthermore, we observed unexpectedly high 4. N. J. Antia, P. J. Harrison, L. Oliveira, Phycologia 30 , 1 26. B. Palenik, S. E. Henson, Limnol. Oceanogr. 42 , 1544 N -AAP sugar is de-acetylated for every mole of acesubstantially reduced by abiotic complexation derivatized samples ( 15 ) or gas chromatography after these hydrolysis conditions, no muramic acid (derecoveries by high-performance liquid chromatograacetamide. Changes in C and N were assessed with (2003). and give rise on totime the hydrolysis-resistant 31. R. G. Keil, D. L. Kirchman, Mar. Chem. 45, 187 (1994). acid hydrolysis, only 2.2 (29% of total N) and 5.5 mmol trogen in phytoplankton is protein. However, DOM interactions may lead to structural which degrade scales of upper ocean d 13 15N-NMR methane concentrations over tropical rainforests, revealing that emission (1991). (1997). tic acid released, we assumed that m moles of acetic derivatizing to alditol acetates ( 10 ). acetylated) or amino sugar monomers were detected. phy and fluorescence detection of o -phthaldialdehydesolid-state Cand spectroscopy. Under 6. E. D.Tanoue, A. Bronk, in Biogeochemistry of Tsugita, Marine Geochim. Dissolved 15 32. S. Nishiyana, M. Kamo, A. of total N) of amide-N remained unhydrolyzed in within marine DOM (31 ). Long-term protein- the (71% amide-N observed by N-NMR. important contribution of N -AAPs to upmodifications that render proteins resistant to 5. T. Berman, D. A. Bronk, Aquat. Microb. Ecol. 31, 279 mixing. These newly added biopolymers 27. M. Cottrell, D. Kirchman, Appl. Environ. Microbiol. acid were equal to m moles ofChem. N in N,-AAPs; hy22. J. I.T. Hedges et al. ,D. Org. 31 , 945 (2000). 14. the D. B. Albert, C. S. Martens, Mar. 56 27 (1997). derivatized samples (15 ) Geochem. or gas chromatography after these hydrolysis conditions, no muramic acid (deOrganic Matter , L. A. Hansell, C. A. Carlson, Eds. inventories considerably underestimated methane sources in these regions Cosmochim. Acta 59 , 2643 (1995). Woods Hole and MAB samples, respectively. In DOM interactions may lead to structural Our data show that two chemically distinct (2003). 66 , Hadja, 1692 (2000). drolyzable N was calculated the basis of 23. We O. Anal. Biochem. 60 512 (1974). 15. order M. Zhao, J. protein L. J. sugar Chromatogr. A.resistance 690 , 55 (1995). derivatizing to alditol acetates ( 10 ). Instruments, acetylated) orBada, amino monomers were detected. (Academic Press, San then Diego, 2002), pp. 163247. ocean HMWDOM, and the of chemical hydrolysis and unavailable to bac- per are chemically distinct from the refractory 33. thank A. Beilecki, at,CA, Bru ker for to quantify the amount of amide-N remaining Methan during the time period ofon investigation (August through November 2003). 6. D. A. Bronk, inthat Biogeochemistry of Marine Dissolved modifications render proteins resistant to amides 28. A. L. Hedges Svitil, T. D. L. defined Kirchman, Microbiol. 144 , M. 1299 (1998). the of amino acid N after strong acid 15, 22. J. I. et al. Org. Geochem. , 945 (2000). 24. Y. Nagata, Fujiwara, K.here Kawaguchinagata, Y.Pullin Fukumori, 16. after The percentage of N in each sample represented by 14. D. B.recoveries Albert, C. S. Martens, Mar. Chem. 56 ,acetic 27 (1997). pools of organic nitrogen accumulate in the the 7. assistance HMWDON is as the31 fraction of DON with strong acid hydrolysis, the sum S acid N-NMR spectroscopy; and in deep sea HMWDOM to chemical teria. This mechanism could lead to the sequesHMWDON pool that exists throughout the se Organic Matter , D. A. Hansell, C. A. Carlson, Eds. 29. M. T. Cottrell, D. N. Wood, L. Yu, D. L. Kirchman, hydrolysis ( 12 ); nonhydrolyzable amide was deterT. Yamanaka, Biochim. Biophys. Acta 1379 , 76 (1998). NAAPs, hydrolyzable protein, and nonhydrolyzable 23. O. Hadja, Anal. Biochem. 60 , 512 (1974). 15. M. Zhao, J. L. Bada, J. Chromatogr. A. 690 , 55 (1995). retained ultrafiltration membrane with pore chemical hydrolysis and unavailable to bac- and(mild) amino acids (strong) was first determined and D. Albert by for an assistance in the determination of a acetic ocean. The higher concentration of HMWDON 1 2 biological degradation, are unexpected tration of nitrogen inDiego, the dissolved phase (32) Methane water column. If we assume the proportion greenho (Academic Press, San CA, 2002), pp. 163247. Appl. Environ. Microbiol. 66 , 287 1195 (2000). mined after calculated strong acid hydrolysis (Table 1). Higher 25. J. S. Martinez et al. , Science , 1245 (2000). amide was as follows: Because 1 mol of 24. acid; Y. Nagata, T. Fujiwara, K. Kawaguchinagata, Y.to Fukumori, 16. then The percentage of after N in each sample represented by size of 1the nm. This fraction is expected have a van ) is, carbon dioxide (CO ), methane emissions are anthropogenic, the (CH C. Frankenberg, J. F. Meirink, M. We subtracted from the initial amide content of 2 the and staff at the Natural Energy Laboratory in 4 teria. This mechanism could lead to the sequesin mixed layer (relative to deep ocean val7. the HMWDON is defined here as the fraction of DON 30. H. Knicker, P. G. Hatcher, Naturwissenschaften 84 231 amino acid yields were obtained when HMWDON was 26. Kona, B. Palenik, S. E. Henson, Limnol. Oceanogr. 42 , ,1544 N -AAP sugar is de-acetylated for every mole aceT. Yamanaka, Biochim. Biophys. Acta 1379 , 76 (1998). NAAPs, hydrolyzable protein, and nonhydrolyzable results that help elucidate the currency of largest nominal molecular weight of 9 1assistance kD. and give rise to the hydrolysis-resistant of N-AAPs, protein, and nonhydrolyzable effect o HMWDON sample (before hydrolysis). In allof cases, Hawaii, and E. Smith for in sample second most important anthropogenic contributors being fossil fuel productration of nitrogen inthe the dissolved phase 32) the retained by reflects an ultrafiltration membrane with a( pore hydrolyzed with strong acid. Asthat acycle. result, after strong (1997). tic in acid released, we nitrogen assumed mmoles of acetic 25. J. Martinez et al. Science , 1245and (2000). amide was calculated as follows: Because 1 mol of ues) largely presence of N-AAPs, 8. (1997). M.S. D. McCarthy, T. ,two Pratum, J.287 I. Hedges, R. A. Benner, 15 percentages are expressed relative to total N in each collection. Supported by the Chemical Biological DON the marine In the past centuries, atmospheric methan amide-N observed by N-NMR. amide measured in our samples is represent(1, 2). greenhouse gas ( 1 ). It also has an indirect tion, ruminants, rice cultivation, and waste size of 1 nm. This fraction is expected to have a 31. R. G. Keil, D. L. Kirchman, Mar. Chem. 45 , 187 (1994). acid hydrolysis, only 2.2 (29% of total N) and 5.5 m mol 26. B. Palenik, S. E. Henson, Limnol. Oceanogr. 42 , 1544 27. M. T. Cottrell, D. L. Kirchman, Appl. Environ. Microbiol. acid were equal to m moles of N in N -AAPs; hyN -AAP sugar is de-acetylated for every mole of aceNature 390, 150 (1997). and give rise on totime the scales hydrolysis-resistant sample. We calculated the amount of N-AAP carbon Oceanography Programs at the National Science Founwhich degrade of upper ocean now constitutes 20% of the anthropogenic cl Our data show that two chemically distinct ative of global HMWDON, then as much as nominal molecular weight of 9 1 kD. 32. E. Tanoue, S. Nishiyana, M. Kamo, A. Tsugita, Geochim. (71% of total N) of amide-N remained unhydrolyzed in 66 , 1692 (2000). drolyzable protein N was calculated on the basis of (1997). tic acid released, we assumed that m moles of acetic effect on climate through chemical feedbacks handling ( 3 ). The natural source strength of 9. M. D. McCarthy, J. I. Hedges, R. A. Benner, Science with the C/N ratio in HMWDOM (Table 1) and dation; the Carbon Sequestration Program at the U.S. 1 amide-N observed by 15N-NMR. mixing. These newly added biopolymers and Notes Institut 8. M. D. McCarthy, T. Pratum, J. I.accumulate Hedges, R. A. Benner, Acta 59 , 2643 (1995). the Woods Hole and MAB samples, respectively. In 28. A. L. D.of L. Kirchman, Microbiol. 144 , 1299 (1998). recoveries of amino acid after strong acid 27. Cosmochim. M. T. Cottrell, D. L. Kirchman, Appl. Environ. Microbiol. acid were equal to m moles ofNN in N -AAPs; hygases. Yet its sources are not well quantified, in 281 , Svitil, 231 (1998). pools of organic nitrogen in the (1, References 80% of the decrease in HMWDON with assumed 8 m mol of C per m mol of NAAP (e.g., N Department Energy; the Rhinehart Coastal Research 2 ). More than 50% of present-day global CH , mainly constituted by wetlands, is 1. J. Abell, S. Emerson, Renaud, Mar. 58 , 203 Our data that two chemically distinct Heidelbe Nature 390,show 150 distinct (1997). are chemically from the refractory 33. We thank A. budget. Beilecki, then at Bru ker Instruments, for order to quantify the amount ofJ.amide-N remaining 29. M. Cottrell, D. J.N. Wood, L. Yu, D.Deep-Sea L. the Kirchman, hydrolysis (12); nonhydrolyzable amide was deter66 , T. 1692 (2000). drolyzable protein NP. was calculated onRes. the basis of 10. 4 L. I. Aluwihare, D. Repeta, R.retrieved F.Chen, Res. acetyl glucosamine). Center of the Woods Hole Oceanographic Institution; global We global methan ocean. The higher concentration of HMWDON depth involves the removal of N -AAPs. The 2 (2000). particularly uncertain, these emis15N-NMR because 9. M. D. J.nitrogen I. Hedges, R. throughout A. Benner, Science with after strong acid sum S acetic acid Section spectroscopy; Pullin and Appl. Environ. 66 , 1195 (2000). mined after strong acid hydrolysis (Table 1). Higher 28. assistance A. L. the Svitil, D. L. Microbiol. Kirchman, 144 ,M. 1299 (1998). the recoveries of hydrolysis, amino acidthe NOrg. after strong II 49 , 4421 (2002). pools ofMcCarthy, organic accumulate in the the 17. J. J. J. Boon, V. A. Klap, T. I. Eglinton, Geochem. 29 and Fundacio n Andes, Microbiol. Chile. HMWDON pool that exists 2. M. Church, H. W. Ducklow, D. M. Karl, Institute ofamino Environmental Physics, University of, borne near-infrared absorption spectroscopy. in the layer (relative to deep ocean val- 1 abundance of amide-N throughout the water 281,mixed 231 (1998). sions vary considerably in time space (mild) acids (strong) was first determined and Albert for in Naturwissenschaften the determination of acetic 30. H. Knicker, P. assistance G. D. Hatcher, 84 , 387 231, amino acid yields were obtained when HMWDON was 29. M. T. Cottrell, N.Repeta, Wood, L. F. Yu, D. and L. Kirchman, hydrolysis ( 12); nonhydrolyzable amide wasLimnol. deterlands M 11. D. L. I. Aluwihare, J. R. Chen, Nature 1051 (1998). ocean. The higher concentration of HMWDON water column. If we assume the proportion Oceanogr. 47 ,229, 1 strong (2002). Heidelberg, INF 69120 Heidelberg, Germany. 10. L. I. Aluwihare, D. amides J. Repeta, R. F.more Chen, of Deep-Sea Res. then subtracted the initial amide content of the acid; and the staff at the Natural Energy Laboratory in (1997). hydrolyzed with acid. As a result, after strong Appl. Environ. Microbiol. 66 , 1195 (2000). mined after strong acid hydrolysis (Table Higher gradient, we detected large-scale p ues) largely reflects the presence N-AAPs, column suggests are biologically 166 (1997). 3730 AE 18. J. A. Leenheer, T. from I. Noyes, C. E. Rostad, M. L.1). Davisson, 20 2004; accepted 9 March 2005 ( 4, December 5) latitudinal and available ground-based measure3. P. Libby, P. yields Wheeler, Deep-Sea , 345 (1997). in the mixed layer (relative to deep ocean val- 2 Section ofacid Atmospheric Composition, Royal Netherof N49 -AAPs, protein, and nonhydrolyzable II , 4421 (2002). HMWDON sample (before hydrolysis). In all cases, Hawaii, and E. Smith for assistance in sample 30. Kona, H. G. Knicker, P.L. G. Hatcher, Naturwissenschaften 84 , 231 31. R. Keil, D. Kirchman, Mar. Chem. 45, 187 (1994). acid hydrolysis, only 2.2 (29% ofRes. total44 N) and 5.5 mwas mol amino were obtained when HMWDON Biogeochemistry 69 , 125 (2004). 10.1126/science.1108925 natural methane emissions. Furthermore, we which degrade on time scales of upper ocean recalcitrant than other forms of organic-N. ments are sparse, albeit precise, and limitedly 4. N. J. Antia, P. J. Harrison, L.relative Oliveira, Phycologia 30 , in 1 lands Meteorological Institute, Post Office Box 201, 11. L. largely I. Aluwihare, D. in J. Repeta, R. F. Chen, 387, percentages are expressed total N in each Supported byM. the Chemical and Biological 32. collection. E. Tanoue, S. Nishiyana, Kamo, A. Tsugita, Geochim. (71% of total N) of amide-N remained unhydrolyzed (1997). hydrolyzed with strong acid. As a to result, after strong ues) reflects the presence ofNature N-AAPs, amide measured our samples is representmethane concentrations over tropical rainfor (1991). mixing. These newly added biopolymers The ubiquity of amide linkages in HMWDON 166 (1997). representative at larger scales. Better knowl3730 AEWoods DeWe Bilt, Netherlands. sample. calculated the amount of Nand -AAP carbon Programs at the National Science FounCosmochim. 59, 2643 (1995). the Hole and MAB samples, respectively. In 31. Oceanography R. G. Keil, D. Acta L. Kirchman, Mar. Chem. 45 , 187 (1994). acid hydrolysis, only 2.2 (29% of total N) 5.5 mmol 13 MAY 2005 1010 which degrade on time scales of upper ocean ative of global HMWDON, then as much as 5. K. T. Berman, D. A.ratio Bronk, Microb. Ecol. 31 Kaiser, R. Benner, Anal. Chem. 72 , 2566 (2000). 12. S.chemically Henrichs, P. M. Williams, Mar. Chem. 17 ,refractory 141 (1985). inventories considerably underestimated meth with the C/N inAquat. HMWDOM (Table 1), 279 and theA. Carbon Sequestration Program atGeochim. the U.S. are distinct from the is not surprising, given that most organic ni- 19. 33. We thank Beilecki, then at Bru ker Instruments, for order to total quantify the amount of amide-N remaining 32. dation; E. Tanoue, S. Nishiyana, M. Kamo, A. Tsugita, (71% of N) of amide-N remained unhydrolyzed in g acids (2003). mixing. These newly in added biopolymers 1559 80% of decrease HMWDON with D. L. Woods Popham, J. Helin, C. E. Costello, P. Setlow, J. 13. De-acetylation of HMWDON was performed assumed 8 m mol of CMAB per m mol NAAP (e.g.,acid N Department of Energy; theperiod Rhinehart Coastal Research assistance with after strong acid hydrolysis, theof sum Sacetic Cosmochim. Acta , 2643 (1995). of N-NMR spectroscopy; M. Pullin and the Hole and samples, respectively. In during the time investigation (August HMWDON pool that exists throughout the 20. trogen in the phytoplankton issamples protein. However, ve been 6. Bacteriol. D. A. Bronk, in Biogeochemistry Marine Dissolved 178 , acids 6451 (1996). 1 involves N HCl. Samples were heated overnight at 13 90-C MAY 2005 VOL 308 SCIENCE www.sciencemag.org 1010 depth acetyl glucosamine). of for the Woods Hole Oceanographic Institution; areinchemically distinct from the refractory 33. Center We thank A. assistance Beilecki, then at Bru ker Instruments, for (mild) amino was first determined and order to quantify the(strong) amount ofof amide-N remaining D. Albert in the determination of acetic the removal of N -AAPs. water column. If we assume the proportion the important contribution of N -AAPs to The up- 21. Organic Matter , from D. A. C. A. nd sed15 modified the method provided in content (Carlson, 20 ) and hyunder an atmosphere of N The acetic acid produced 17. We J. J. Boon, V. A. Klap, T. I.Hansell, Eglinton, Org. Geochem. 29 , and Fundacio nN-NMR Andes, Chile. then subtracted the initial amide ofEds. the assistance with acid;the and the staff at the Natural Energy Laboratory in after strong acid hydrolysis, the sum Sacetic acid spectroscopy; M. Pullin and 2.throughout HMWDON pool that exists throughout the abundance of amide-N the water (Academic Press, San Diego, CA, 2002), pp. 163247. ) and is, after dioxideinof (CO methan Methane (CH drolyzed (in (before 4 N HCl) overnight at -C. We the hydrolysis wasand extracted a weighed of during N -AAPs, protein, per ocean HMWDOM, and nonhydrolyzable the from resistance of 1051 (1998). HMWDON sample hydrolysis). In 90 all cases, Kona, Hawaii, E. Smith for assistance sample (mild) samples amino acids (strong) was first determined and D. Albert for4assistance incarbon the determination acetic hydrol2), water column. If we assume the proportion 7. recovered HMWDON is defined here as the fraction of DON column suggests amides more biologically no muramic acid and only small amounts aliquot ofdeep the hydrolysate into either to ethyl ether or 18. J. A. Leenheer, T. I. Noyes, C. E. Rostad, M. L.N Davisson, 20 December 2004; accepted 9 March 2005 percentages are expressed relative to total in each collection. Supported by Natural the Chemical and Biological then subtracted from the initial amide content of the acid; and the staff at important the Energy Laboratory in the second most anthropogenic largest amide measured in HMWDOM our are samples is representamides in sea chemical orption retained by an ultrafiltration membrane with pore of glucosamine and from surface samAfter extraction, theorganic-N. presence of of dichloromethane. N-AAPs, protein, and nonhydrolyzable Biogeochemistry 69, galactosamine 125 (2004). 10.1126/science.1108925 sample. We calculated the amount of N -AAP carbon Oceanography Programs the National Science FounHMWDON sample (before hydrolysis). In all a cases, Kona, Hawaii, and E. Smith for assistance in sample recalcitrant than other forms of 1 is expected to have 2 a 2 1also 1 greenhouse gas (1). Itat has an indirect tion, ru ative of acid global HMWDON, then as much as 19. ples. and biological degradation, are unexpected DON is size of 1 nm. This fraction This modified hydrolysis method was tested on acetic in the organic fraction was confirmed by C. Frankenberg, J. F. Meirink, M. van Weele, U. Platt, T. Wagner with the C/N ratio in HMWDOM (Table 1) and dation; the Carbon Sequestration Program at the U.S. percentages are expressed relative to total N in each collection. Supported by the Chemical and Biological 12. S. Henrichs, P. M. Williams, Mar. Chem. 17 , 141 (1985). K. Kaiser, R. Benner, Anal. Chem. 72 , 2566 (2000). g acids amide measured in our samples is representThe ubiquity of amide linkages in HMWDON 1H-NMR nominal molecular weight 9 1 kD. effect on climate through chemical feedbacks handlin 80% ofthat the help decrease in samples HMWDON with results elucidate the currency of peptidoglycan (from Bacillus subtilus ), solution-state before low molecular weight experiassumed 8 m mol of C the per mmol ofof NAAP (e.g., N Department of Programs Energy; the Coastal Research sample. We calculated amount N -AAP carbon Oceanography at Rhinehart the National Science Foun13. De-acetylation of HMWDON was performed 20. chitin D. L. oligomers, Popham, J. Helin, C. of E. Costello, P. Setlow, J. ve been ative of global HMWDON, then as much as 8. and M. D. McCarthy, T. Pratum, J. I. Hedges, R. A. is not surprising, given that most organic nibacterial cells (mixed culture) toBenner, ensure acids (e.g., acetic acid) were quantified. The aqueous acetyl glucosamine). with the C/N ratio in laboratory HMWDOM (Table 1) and Center of Woods Hole Oceanographic Institution; thethe Carbon Sequestration Program at the U.S. in 1in N HCl. Samples were heated overnight at 90 -C Bacteriol. 178 , 6451 (1996). (1, dation; 2). More than 50% of present-day global CH4, m depth involves the removal of N-AAPs. The DON the marine nitrogen cycle. In the past two centuries, atmospheric methane has more than doubled and dation Nature 390 ,m 150 (1997). nd sed80% of the decrease in HMWDON with (9 90%) recovery and fraction of the hydrolysate was lyophilized and, as n Andes, 17. the J. J. quantitative Boon, V. A. Klap, T. Eglinton, Org. Geochem. 29 , and the Fundacio assumed 8 mol of C I.per mmol of ofglucosamine NAAP (e.g., NDepartment of Energy; theChile. Rhinehart Coastal Research trogen in phytoplankton protein. However, under an atmosphere of N2 21. We modified the method provided in ( 20) and hy.is The acetic acid produced particul abundance of amide-N throughout the water now constitutes 20% of the anthropogenic climate forcing by greenhouse eins is 1 9. M. D. McCarthy, J. I. Hedges, R. A. Benner, Science de-acetylated muramic We assessed amino sugar shown in the H-NMR (Fig.extracted 2C), no longer contained 1 1051 (1998). acetyl glucosamine). Center of of the Woods Hole Oceanographic Institution; drolyzed samples (in 4 acid. N HCl) overnight at 90-C. We during the hydrolysis was from a weighed hydroldepth involves theNotes removal of -AAPs. the important contribution of NN -AAPs to The upInstitute Environmental Physics, in University of References and 281 ,Boon, 231 (1998). sions v column suggests amides are more gases. Yet its sources are not well 29 quantified, introducing uncertainties its exation recoveries by high-performance liquid chromatograacetamide. Changes in C and N were biologically assessed with accepted 18. A. Leenheer, T. I. Noyes, C. E. Rostad, M. L. Davisson, 20 December 2004; 9 March 2005 17. J. J. V. A. Klap, T. I. Eglinton, Org. Geochem. , and the Fundacio n Andes, Chile. recovered no muramic acid and only small amounts aliquot of the hydrolysate into either ethyl ether or 1. J.ocean Abell, S. Emerson, P. Renaud, J. Mar. Res. , 203 orption Heidelberg, INF 229, 69120 Heidelberg, Germany. abundance of throughout the58 water per HMWDOM, and the resistance of 13 15N-NMR 10. phy L. I.glucosamine Aluwihare, D. J., Repeta, R. F. Deep-Sea Res. and fluorescence detection ofChen, o-phthaldialdehydesolid-state C-amide-N and spectroscopy. Under Biogeochemistry 69 125 (2004). 10.1126/science.1108925 1051 (1998). of and galactosamine from surface samdichloromethane. After extraction, the presence of (4, 5) recalcitrant than other forms of organic-N. 2 global budget. We retrieved the global methane distribution by using spaceotein(2000). Section of Atmospheric Composition, Royal NetherII 49 , 4421 (2002). DON is column suggests amides are biologically samples ( 15) orC. gas chromatography after these in hydrolysis conditions, nomore muramic acid (deamides deep sea HMWDOM to chemical 18. derivatized J. A. Leenheer, T. I. Noyes, E.method Rostad, was M. L.tested Davisson, 20 December 2004; accepted 9 March 2005 ples. This modified hydrolysis on acid in the organic fraction was confirmed by ments a The ubiquity of amide linkages in HMWDON 2. acetic M. J. Church, H. W. Ducklow, D. M. Karl, Limnol. borne near-infrared absorption spectroscopy. In addition to the expected uctural lands Meteorological Institute, Post Office Box 201, 11. L. I. Aluwihare, D. J. Repeta, R. F. Chen, Nature 387 , 1H-NMR derivatizing to alditol acetates (10). Bacillus subtilus acetylated) or amino sugar monomers were detected. Biogeochemistry 69 , 125 (2004). 10.1126/science.1108925 chitin oligomers, peptidoglycan solution-state before low molecular weight experirecalcitrant than other forms of organic-N. 1 (from 2 ), 2 1 1 and biological degradation, are unexpected Oceanogr. 47 ,1 (2002). 166 (1997). represen 3730 AE De Bilt, Netherlands. C. Frankenberg, J.we F. Meirink, M. van Weele, U. Platt, T. Wagner is not given that most organic ni- 22. J. latitudinal gradient, large-scale patterns of anthropogenic and tant to I. Hedges et al., (mixed Org. Geochem. 31 ,detected 945 (2000). 14. D. B.surprising, Albert, C. S. Martens, Mar. Chem. 56 , 27 aqueous (1997). and bacterial cells laboratory culture) to ensure acids (e.g., acetic acid) were quantified. The adation 15. The of amide linkages in HMWDON 3. P.ubiquity Libby, P. Wheeler, Deep-Sea Res. 44 , 345 (1997). results that help elucidate the currency of 23. O. Hadja, Anal. methane Biochem. 60,emissions. 512 of (1974). M. Zhao, J. L.the Bada, J. Chromatogr. A. 690 , 55 (1995). the quantitative ( 9 90%) recovery glucosamine and fraction of hydrolysate was lyophilized and, as trogen in phytoplankton is protein. However, natural Furthermore, we observed unexpectedly high o bac4. not N. J.surprising, Antia, P. J. Harrison, L. Oliveira, Phycologia 30 ,1 eins is 16. is that most organic ni1H-NMR DON in the marine nitrogen cycle. 24. Y. Nagata, T. Fujiwara, K. acid. Kawaguchinagata, Y. Fukumori, percentage ofgiven N in each sample represented by In the past two centuries, atmospheric methane has revealing more than doubled and de-acetylated muramic We assessed amino sugar shown in the (Fig. 2C), no longer contained theThe important contribution of Nnonhydrolyzable -AAPs to upmethane concentrations over tropical that emission (1991). eques13 MAYrainforests, 2005 VOL 308 SCIENCE www.sciencemag.org 1010 T. Yamanaka, Biochim. Biophys. Acta 1379 , 76 (1998). NAAPs, hydrolyzable protein, and exation recoveries by high-performance liquid chromatograacetamide. Changes in C and N were assessed with trogen in phytoplankton is protein. However, now constitutes 20% of the anthropogenic climate forcing by greenhouse 5. T. Berman, D. A. Bronk, Aquat. Microb. Ecol. 31 , 279 per ocean HMWDOM, and the resistance of inventories considerably underestimated methane sources in these regions se (32) 15 25. J. S. Martinez et al., Science 287 1245 (2000). was 13 calculated as follows: Because 1 mol of phy and fluorescence detection of, o -phthaldialdehydesolid-state Cand N-NMR Under References and Notes roteintheamide important contribution ofspectroscopy. N -AAPs to up(2003). gases. Yet its sources are well quantified, introducing uncertainties in its 26. B. Palenik, S. E. Henson, Oceanogr. 42 , 1544 N -AAP sugar is de-acetylated forJ.every mole of ,aceamides in deep sea HMWDOM to chemical during the time period of not investigation (August through November 2003). derivatized samples (15) Limnol. or gas chromatography after hydrolysis conditions, no muramic acid (desistant 1. J. Abell, S. Emerson, P. Renaud, Mar. Res. 58 203 6. these D. A. Bronk, in Biogeochemistry of Marine Dissolved uctural per ocean HMWDOM, and the resistance of global budget. We retrieved the global methane distribution by using space(1997). tic acid released, we assumed that m moles of acetic derivatizing to alditol acetates ( 10 ). acetylated) or amino sugar monomers were detected. (2000). 1 2 2 1 1 andOrganic biological are Matter,degradation, D. A. Hansell, C. A. unexpected Carlson, Eds. C. Frankenberg, J.absorption F. Meirink, M. van Weele, U. Platt, T. Wagner 27. T.Hedges Cottrell, D. L., Kirchman, Appl. Environ. equal to moles of D. N M. into N -AAPs; hystant to amides in deep sea HMWDOM 22. M. J. I. et al. Org. Geochem. 31 , 945 Microbiol. (2000). 14. D. B. Albert, C. S. Martens, Mar. 56 ,chemical 27 (1997). 2. acid M. J.were Church, H. W.m Ducklow, Karl, Limnol. borne near-infrared spectroscopy. In emissions addition to the expected the (Academic Press, San Diego, CA,Chem. 2002), pp. 163247. results that help elucidate the currency of ) is, after carbon dioxide (CO ), methane are anthropogenic, Methane (CH distinct 1692 (2000). drolyzable N calculated on the of 4 Biochem. 60 2 23. 66 O., Hadja, Anal. , 512 (1974). 15. Zhao, J. protein L. J. was Chromatogr. A. 690 , 55basis (1995). Oceanogr. 47 , defined 1degradation, (2002). 1 2 2 of anthropogenic 1 1 o bacand biological are unexpected 7. M. HMWDON isBada, here as the fraction of DON latitudinal gradient, detected large-scale patterns and C. Frankenberg, J.we F. Meirink, M. van Weele, Platt, Wagner 28. L. Svitil, D. L. Kirchman, Microbiol. 144 , atmospheric 1299 (1998). recoveries of acidsample Ncycle. after strong acid DON in the marine nitrogen theA. important anthropogenic largest contributors beingT. fossil fuel producIn the past two centuries, methane has U. more than doubled and in the 24. Y.second Nagata, T. most Fujiwara, K. Kawaguchinagata, Y. Fukumori, 16. The percentage of amino N Deep-Sea in each represented by 3. the P. Libby, P. Wheeler, Res. 44 , 345 (1997). retained by an ultrafiltration membrane with a pore sequesresults that help elucidate the currency of natural methane emissions. Furthermore, we observed unexpectedly high 29. M. T. Cottrell, D.(1 N. L. Yu, D. an L. Kirchman, (P. 12J. ); Harrison, nonhydrolyzable amide was deterT. Yamanaka, Biochim. Biophys. Acta 1379 , 76 (1998). NAAPs, protein, nonhydrolyzable 4. hydrolysis N. J. Antia, L. Oliveira, Phycologia 30, 1 greenhouse gas ). Wood, It also has indirect tion, ruminants, rice cultivation, and waste now constitutes 20% of the anthropogenic climate forcing by greenhouse WDON size of 1hydrolyzable nm. This fraction is and expected to have a se (32) Environ. Microbiol. 66 , 287 1195 (2000). mined strong acid (Table 1). Higher DON in after the marine nitrogen cycle. In the past centuries, atmospheric methane hasrevealing more than doubled and 25. Appl. J. S. methane Martinez et al. , two Science , 1245 (2000). amide was calculated ashydrolysis follows: Because 1 mol of concentrations tropical rainforests, that emission (1991). References and weight Notes nominal molecular of 91 kD. effect on climate through chemical feedbacks handling (3). The uncertainties natural source strength of gases. Yet its sources are over not well quantified, introducing in its an val30. G. Hatcher, Naturwissenschaften 84 231 acid were obtained when HMWDON was 26. H. B. Knicker, Palenik, S. E. Henson, Limnol. Oceanogr. , ,1544 N sugar for mole of, ace5. T. Berman, D.is A.de-acetylated Bronk, Aquat. Microb. Ecol. 279 1. amino J. -AAP Abell, S. yields Emerson, P. Renaud, J.every Mar. Res. 58 203 sistant now P. constitutes 20% of underestimated the 42 anthropogenic climate forcing by greenhouse 8. M. D. McCarthy, T. Pratum, J. I. Hedges, R. A. 31 Benner, inventories considerably methane sources in these regions (1, (1997). 2). global More than 50% of retrieved present-day global CH4, mainly constituted by spacewetlands, is budget. We the global methane distribution by using hydrolyzed with strong acid. As that a result, after strong AAPs, (1997). tic acid released, we assumed m moles of acetic (2003). (2000). References and Notes Nature 390, 150 (1997). gases. Yet its sources are not well quantified, introducing uncertainties in its during the time period of investigation (August through November 31. Keil, D. L. Kirchman, Mar. Chem. 45 , 187 (1994). hydrolysis, 2.2 (29% ofof total N) and 5.5 m 27. R. M.G. T. Cottrell, D. L. Kirchman, Appl. Environ. Microbiol. acid were equal to m moles N in N -AAPs; hy6. D. A. Bronk, inonly Biogeochemistry of Marine Dissolved 2. acid M.Abell, J. Church, H. W. Ducklow, D. M. Karl, Limnol. particularly uncertain, these emis1. J. S. Emerson, Renaud, Mar. Res. 58 , mol 203 borne near-infrared absorption spectroscopy. In addition to because the 2003). expected ocean 9. M. D. McCarthy, J. I.P. Hedges, R.J. A. Benner, Science 1 distinct 32. E. Tanoue, S.Environmental Nishiyana, M. Kamo, A. Tsugita, Geochim. global budget. We retrieved the global methane distribution by using space(71% of total N) of amide-N remained unhydrolyzed in 66 , 1692 (2000). drolyzable protein N was calculated on the basis of Organic Matter , D. A. Hansell, C. A. Carlson, Eds. Oceanogr. 47 , 1 (2002). (2000). Institute of Physics, University of 281, 231 (1998). sions patterns vary considerably in time and latitudinal we144 detected large-scale of anthropogenic and space ymers Cosmochim. Acta 59gradient, , 2643 (1995). Woods Hole and MAB samples, respectively. In 28. A. L. borne Svitil, D. L. Kirchman, Microbiol. , 1299 (1998). the recoveries of amino acid N strong acid in the (Academic San Diego, CA, 2002), 163247. 3. P. Libby, P. Press, Wheeler, Deep-Sea 44 , pp. 345 (1997). 2. the M.I. J. Church, H. Ducklow, D.after M. Karl, Limnol. near-infrared absorption spectroscopy. In emissions addition to the expected the Heidelberg, INF 229, 69120 Heidelberg, Germany. ) is, after carbon dioxide (CO methane are anthropogenic, Methane (CH 10. L. Aluwihare, D. J.W. Repeta, R. F.Res. Chen, Deep-Sea Res. 4 2), ( 4 , 5 ) and available ground-based measurenatural methane emissions. Furthermore, we observed unexpectedly high 2 actory 33. We thank A. Beilecki, then at Bru ker Instruments, for order to quantify the amount of amide-N remaining 29. M. T. Cottrell, D. N. Wood, L. Yu, D. L. Kirchman, hydrolysis ( 12 ); nonhydrolyzable amide was deter7. HMWDON is defined here as the fraction of DON 4. N. J. Antia, P. J. Harrison, L. Oliveira, Phycologia 30 , 1 Oceanogr. 47 , 1 (2002). Section of Atmospheric Composition, Royal NetherII 49, 4421 (2002). WDON latitudinal gradient, we detected large-scale patterns of anthropogenic and the second most important anthropogenic largest contributors being fossil fuel produc15N-NMR assistance with strong acid hydrolysis, the sum acetic acid spectroscopy; M. Pullin and Appl. Environ. Microbiol. 66, 1195 (2000). mined after strong hydrolysis (Table 1).(1997). Higher ments are sparse, albeit precise, and limitedly retained by an ultrafiltration membrane with a 387 pore methane concentrations over tropical rainforests, revealing that emission (1991). 3. after P. I. Libby, P. Wheeler, Deep-Sea Res. 44, S 345 ut valthe lands Meteorological Institute, Post Office Box 201, 11. L. Aluwihare, D. J. acid Repeta, R. F. Chen, Nature , an natural methane emissions. Furthermore, we observedrice unexpectedly high (mild) amino acids (strong) was firstPhycologia determined and D. Albert for assistance in the determination of acetic greenhouse gas ( 1 ). It also has an indirect tion, ruminants, cultivation, and waste 30. H. Knicker, P. G. Hatcher, Naturwissenschaften 84 , 231 amino acid yields were obtained when HMWDON was size of 1 nm. This fraction is expected to have a 5. T. Berman, D. A. Bronk, Aquat. Microb. Ecol. 31 , 279 4. N. J. Antia, P. J. Harrison, L. Oliveira, 30 , 1 166 (1997). representative at larger Better knowl3730 AE De Bilt, Netherlands. inventories considerably underestimated methane sources in scales. these regions ortion then subtracted the initial amide content the acid; andclimate the staff concentrations at the Natural Energyover Laboratory in (1997). hydrolyzed withfrom strong acid. akD. result, after of strong AAPs, nominal molecular weight ofAs 91 methane tropical rainforests, revealing that emission (2003). (1991). effect on through chemical feedbacks handling ( 3 ). The natural source strength of the time period of45 investigation (August through November 2003). yzable sample (before hydrolysis). In all cases, Hawaii, and E. Smith forChem. assistance in sample 31. Kona, R. G.during Keil, D. L. Kirchman, Mar. , 187 (1994). acid hydrolysis, only 2.2 (29% of total N) Ecol. and 5.5 m 8. M. D. McCarthy, T. Pratum, J. I. Microb. Hedges, R. A. Benner, 6. HMWDON D. Berman, A. Bronk, in Biogeochemistry of Marine Dissolved 5. T. D. A. Bronk, Aquat. 31 , mol 279 ocean inventories considerably underestimated methane sources in these regions (1, collection. 2). More than 50% of present-day global CH percentages are expressed relative to A. total N in each Supported byM. the Chemical and Biological 4, mainly constituted by wetlands, is 32. E. Tanoue, S. Nishiyana, Kamo, A. Tsugita, Geochim. (71% of total N) of amide-N remained unhydrolyzed in resentNature 390 , 150 (1997). Organic Matter , D. A. Hansell, C. Carlson, Eds. (2003). 13 MAY 2005 VOL 308 SCIENCE www.sciencemag.org 1010 sample. lymers 41 during the time of investigation (August through November 2003). particularly uncertain, these emisWe Press, calculated the amount of Npp. -AAP carbon Oceanography at period the National Science FounCosmochim. Acta 59 , 2643 (1995). Woods Hole and MAB samples, respectively. In 9. M. D. McCarthy, J. I. Diego, Hedges, R.2002), A. Benner, Science (Academic San CA, 163247. 6. the D. A. Bronk, in Biogeochemistry of Marine Dissolved )Programs is, after carbon dioxide (CO ), methane emissions arebecause anthropogenic, the Methane (CH uch as 1 4 2 Institute of A. Environmental Physics, University of the C/N ratio in HMWDOM (Table 1) DON and dation; the Carbon Sequestration at the U.S. actory 33. We thank Beilecki, then at Bru Program ker Instruments, for order to quantify theA. amount amide-N remaining 281 , 231 (1998). 7. with HMWDON is defined here asof the of Organic Matter , D. Hansell, C. fraction A. Carlson, Eds. sions vary considerably in time and space the second most important anthropogenic largest contributors being fossil fuel producN with 15

Assessing Methane Emissions from Global Space-Borne Observations

Assessing Methane Em Global Space-Borne O

Assessing Methane Emissions from Global Space-Borne Observations

Assessing Methane Emissions from Global Space-Borne Observations

Assessing Methane Emissions from Assessing Methane Emissions from Global Space-Borne Observations Global Space-Borne Observations

Chapter Two

To cite articles from Chapter Two of this booklet, please use the following format: [Author name(s)] in Microbial Carbon Pump in the Ocean, N. Jiao, F. Azam, S. Sanders, Eds. (Science/AAAS, Washington, DC, 2011), pp. xx-xx.

42

Microbial Carbon Pump and its Significance for Carbon Sequestration in the Ocean
Nianzhi Jiao1 and Farooq Azam2 The generally accepted biogeochemical mechanism of long-term carbon sequestration in the ocean is the biological pump, which is based on the process of downward flux of particulate organic matter and its burial in the ocean bottom. The recently proposed microbial carbon pump (MCP) conceptualizes a dissolved phase sequestration mechanism based on the microbial generation of refractory dissolved organic matter (RDOM), which is resistant to biological decomposition and assimilation and thus persists in the water column. The central questions of the MCP concern the structure-specific molecular consequences of microbe and organic matter interactions. Understanding the molecular nature and global scope of the MCP calls for a multidisciplinary approach and the development of new tools to, for instance, chemically characterize RDOM in concert with microbial functional gene analysis. The MCP and the biological pump are mechanistically interconnected in the ocean carbon cycle, requiring the integration of quantitative, mechanistic, and predictive studies in order to understand ocean carbon biogeochemistry and how it interacts with climate change. oncerns over climate change due to anthropogenic CO2 have stimulated intensive research on the capacity of the ocean for long-term carbon sequestration. Historically, it has been recognized since the 1970s that solving this problem would necessitate a fundamental understanding of the functioning of the global ocean carbon cycle (1, 2). In order to address this, large multidisciplinary studies were conducted on ocean carbon biogeochemistry over the past three decades [see (1) for a historical account]. The goals of these studies have been to quantify, model, and predict the spatial-temporal patterns of downward flux of organic matter and its relationship to primary productivity and foodweb dynamics in the water column. This research led to the formulation and de-

Fig. 1. Diagram showing the MCP and its relationship with the biological pump. The MCP and the biological pump are intertwined in the ocean carbon cycle. While the majority of the primary production is in the form of POM, a portion of the fixed carbon is released as DOM into the water. This DOM together with DOM from other sources along the food chain can be partially transformed by the MCP into RDOM. During the sinking process, a great deal of POM is hydrolyzed by attached microbes and becomes DOM contributing to the MCP, while some RDOM molecules are scavenged by POM, joining the biological pump. In contrast to the exponential attenuation of the POM flux along water depth, the RDOM persists throughout the water column. The relative importance of MCP vs. the biological pump varies with environmental scenarios. [after Jiao et al., 2010 (4)]

State Key Laboratory of Marine Environmental Sciences, Xiamen University, Xiamen 361005, P. R. China (jiao@xmu.edu.cn) 2 ScrippsInstitutionofOceanography, UCSD,La Jolla CA, 92093, USA (fazam@ucsd.edu)

velopment of the paradigm of the biological pump: A variable fraction of the biogenic debris from the upper ocean escapes decomposition and respiration, sinking down to the deep sea and to the seafloor and becoming buried in the sediments. The advent of the microbial loop in the 1980s revealed that approxi-

43

Fig. 2. Three pathways of RDOM generation are summarized in the MCP: direct production of RDOM from microbial cells (center), RDOM derived from POM degradation (right), and residual DOM after microbial modification of the bulk DOM (left). In the direct pathway, the main mechanisms are exudation from active microbial cells and release form viral lysis or grazing. Viral lysis can offset as much as half of the microbial production and one-quarter of primary production in the ocean; a portion of the DOM produced by viral lysis could be RDOM (19, 20). About 25% of the ocean RDOM is estimated to be of bacterial origin (6). For the POM pathway, ectoenzymes expressed by microorganisms convert POM to DOM at rates that often exceed microbial uptake of the DOM (21, 22) and some byproducts of this hydrolytic activity could be resistant to further utilization by microbes (either free-living or attached to POM), thus becoming RDOM. In the residual DOM pathway, a part of bulk DOM can remain/become RDOM after microbial processing, such as acylheteropolysaccharide originated from products of phytoplankton and bacteria (23).

of carbon sequestration and its relationship to surface productivity. In contrast, the kinetics and mechanisms of carbon sequestration by the MCP is more complex, determined by intricate interactions of molecules and microbes (Fig. 1). This is also true of metabolic interactions of bacteria with organic particles since bacteria must first hydrolyze the particles to diffusible molecules within their microenvironments. Studies to understand MCP-mediated carbon sequestration must therefore contend with great molecular and microbial diversity and system complexityas well as minuscule, nanometer to micrometer, ecosystem scales relevant to the realms of microbes and molecules. Further, the need is to measure the production of molecules of currently unknown structures (7). Thus, MCP research will require new approaches and tools that enable studies at a molecular resolution and distinct from those used in studying particle phase sequestration by the biological pump. This also argues in favor of a dedicated focus on the MCP component of the ocean carbon cycle, while also integrating these findings with data on the biological pump.

mately one-half of the fixed carbon in fact flows through the dissolved organic matter (DOM) pathways, with most presumed to be rapidly respired or assimilated (labile DOM) by heterotrophic bacteria (now known to include Archaea) (3). However, these rapid fluxes of labile DOM occur in the presence of an enormous pool of refractory dissolved organic matter (RDOM), which accounts for >90% of the organic matter in seawater and may persist for thousands of years. Thus far, the biogeochemical behavior of RDOM, its origin, and its place in the ocean carbon cycle has been technically difficult to determine. The Microbial Carbon Pump Synthesizing the knowledge on microbial carbon cycling in the ocean, Jiao et al. (4) proposed that microbial metabolism of labile DOM and trophic interactions within the microbial loop generate RDOM. This process would result in long-term carbon sequestration in the dissolved phase throughout the water column. It could be a modulating factor in the redistribution of carbon among the Earths surface reservoirs and thus influence climate change (5). The proposed pathway (the microbial carbon pump, MCP) had not previously been explicitly included in ocean carbon cycle. It is argued that the MCP is a quantitatively significant biogeochemical pathway for RDOM generation and carbon sequestration that should be specified in ocean carbon models (4, 6). The MCP framework also enables researchers to explore the hypothesis that shifts in metabolic accessibility of RDOM to bacteria and Archaea, resulting in its respiration, could influence the global carbon cycle and climate. Microbial generation and cycling of RDOM could thus be significant variables in the ocean carbon cycle (Fig. 1). RDOM Biogeochemistry, a Molecular Problem Measurements of particle flux, elemental analysis, and primary productivity can reasonably well constrain the biological pump models

New Tools of the Trade Taken together, these considerations underscore the usefulness of the MCP as a framework for studying carbon sequestration in the dissolved phase throughout the oceans water column on both a molecular and global level. Molecular characterization of RDOM has been largely intractable thus far, but recent analytical advances [for example, ultrahigh resolution mass spectrometry] (7) are beginning to resolve this fundamental problem. While the technique has already yielded thousands of molecular formulas for DOM components, the challenge to determine the corresponding chemical structures still remains. On the microbial side of the interactions, advances in genomics, transcriptomics, and proteomics can be applied to determine the metabolic capabilities of the microbes involved and how they might modify the labile organic matter to produce RDOM. Simultaneous expression profiles of thousands of relevant genes can be determined by Q-CHIP technology (8). This technique could, in principle, be configured to detect the expression of genes relevant to the presence of specific DOM components, once their chemical structures have been determined. Developing tools to accurately measure respiration in minimally perturbed samples is an important goal, as it may elucidate how microbial metabolism partitions DOM between respiration and sequestration in changing biogeochemical scenarios in the ocean. Respiration methods that do not require physical separation of bacteria and Archaea from other plankton are being developed (9) and should provide better quantitative constraints on bacterial processing of DOM and generation of RDOM.

RDOM Sources, Mechanisms, and Constraints on Decomposition Jiao et al. (4) have argued that microbes generate RDOM as they act on diverse dissolved and particulate sources of organic matter (Fig. 2) as well as participate in foodweb trophic dynamics (Fig. 1). These interactions have the potential to generate molecular diversity that may include RDOM. Highly abundant and diverse lytic viruses may generate unique RDOM as they probably lyse most marine organisms and release (rather than digest) the cellular material (10). In addition, some RDOM may be produced de novo by the metabolic activities of phyto-

44

plankton or bacteria (mechanistically distinct from the bacteria acting on organic matter). New approaches are already emerging to quantify the overall contribution of microbes to RDOM generation (6) and future analyses need to identify the dominant pathways within the MCP and the necessary ecosystem conditions for RDOM formation. While MCP studies are mainly focused on the contribution of microbes to carbon sequestration, other biotic and abiotic RDOM sources should also be taken into consideration when carbon budget and age are concerned. For example, the old DOM from seafloor seeps (11, 12) may not necessarily be all RDOM (13). Understanding the MCP also challenges scientists to elucidate the mechanistic underpinnings of RDOM generation, which will be essential for developing refined models of carbon sequestration. There is good reason to believe that the mechanisms of RDOM generation will eventually be revealed at the biochemical level. For example, it was recently demonstrated that incomplete hydrolysis of organic matter by bacterial ectohydrolases could lead to RDOM formation (14). Further, bacterial ectohydrolases have already been shown to be central to connecting the biological pump with the MCP, particularly through particle hydrolysis (15). A question of great interest is why bacteria and Archaea do not metabolize RDOM. This question is critical for predicting long-term net carbon sequestration in the dissolved phase. Highlighting this longstanding question in the context of the MCP is stimulating research on whether the persistence of RDOM could be due to the extreme dilution of individual molecular species, making their utilization energetically unfavorable for bacteria and Archaea (7). Another unanswered question is whether there are structural constraints on RDOM decomposition. This should benefit from the development of model systems of known RDOM components (currently none is known) and bacteria.

Integration of MCP and Biological Pump The MCP and the biological pump are mechanistically intertwined in the ocean carbon cycle, a fact that necessitates integrated research strategies. For example, bacteria act on sinking particles to cause quantitatively major particulate organic matter (POM) to DOM flux, some of which may become RDOM. Conversely, sinking particles can scavenge RDOM from seawater and carry it into the carbon sequestration process of the biological pump. The interconnection between the MCP and the biological pump is also apparent from the concept that organic matter cannot be neatly divided into DOM and POM, but instead exists as a continuum of small molecules to large sinking particles (1618). Therefore, the realms of the MCP and the biological pump can overlap as microbes and other organisms interact with the organic matter continuum and among themselves to create overlapping yet distinctive patterns of carbon cycling and sequestration. This view of the organic matter continuum underscores a need for an integrated approach to further our understanding of the mechanistic and predictive nature of the carbon cycle. Further, a cultural integration of the biological pump community and the emergent MCP community could move us towards this goal. Future The unfolding climate change scenarios will demand a quantitative, mechanistic, and predictive understanding of ocean carbon biogeochemistry to guide policy and advise the global society on adaptive measures. Explicit incorporation of the MCP into the concept and models of the ocean carbon cycle will be essential and will require new emphasis on methods development and global ocean field programs. It is important that the MCP community collaborates closely with other geochemists and biogeochemists to develop an integrated view of the ocean carbon cycle and its response to climate change.

References and Notes 1. E. A. Laws, P. G. Falkowski, W. O. Smith Jr, H. Ducklow, J. J. McCarthy, Global Biogeochem. Cy. 14, 1231 (2000). 2. K. L. Denman, M. A. Pena, in The changing ocean carbon cycle: a midterm synthesis of the Joint Global Ocean Flux Study, R. B. Hanson, H. W. Ducklow, J. G. Field, Eds. (Cambridge Univ. Press, New York, 2000), pp. 469490. 3. F. Azam et al., Mar. Ecol. Prog. Ser. 10, 257 (1983). 4. N. Jiao et al., Nat. Rev. Microbiol. 8, 593 (2010). 5. P. F. Sexton et al., Nature 471, 349 (2011). 6. R. Benner, G. Herndl, in Microbial Carbon Pump in the Ocean, N. Jiao, F. Azam, S. Sanders, Eds. (Science/AAAS, Washington, DC, 2011), pp. 4648. 7. G. Kattner, M. Simon, B.P. Koch, in Microbial Carbon Pump in the Ocean, N. Jiao, F. Azam, S. Sanders, Eds. (Science/AAAS, Washington, DC, 2011), pp. 6061. 8. J. D. Van Nostrand, J. Zhou, in Microbial Carbon Pump in the Ocean, N. Jiao, F. Azam, S. Sanders, Eds. (Science/AAAS, Washington, DC, 2011), pp. 6465. 9. C. Robinson, N. Ramaiah, in Microbial Carbon Pump in the Ocean, N. Jiao, F. Azam, S. Sanders, Eds. (Science/AAAS, Washington, DC, 2011), pp. 5253. 10. M. G. Weinbauer, F. Chen, S. W. Wilhelm, in Microbial Carbon Pump in the Ocean, N. Jiao, F. Azam, S. Sanders, Eds. (Science/AAAS, Washington, DC, 2011), pp. 5456. 11. J. W. Pohlman, J. E. Bauer, W. F. Waite, C. L. Osburn, N. R. Chapman, Nat. Geosci. 4, 37 (2011). 12. M. D. McCarthy et al., Nat. Geosci. 4, 32 (2011). 13. Not all fossil components of the DOM entering the water column through seabed seeps are refractory; some are readily available for

14. 15. 16. 17. 18. 19.

20. 21. 22. 23. 24.

microbial respiration and thus do not contribute to carbon sequestration. However, being much older than the biogenic DOM, they would increase the radiocarbon age of the DOM pool. Only that DOM which is resistent to microbial respiration/decompostion, whether biogenic or fossil in origin, can persist in the water column for a long time, constituting carbon sequestration. H. Ogawa, Y. Amagai, I. Koike, K. Kaiser, R. Benner, Science 292, 917 (2001). D. C. Smith, M. Simon, A. L. Alldredge, F. Azam, Nature 359, 139 (1992). F. Azam, Science 280, 694 (1998). F. Azam, A. Z. Worden, Science 303, 1622 (2004). P. Verdugo et al., Mar. Chem. 92, 67 (2004). T. Nagata, D. L. Kirchman, in Mortality of Microbes in Aquatic Environments, Microbial Biosystems: New Frontiers, Proceedings of the 8th International Symposium on Microbial Ecology C. R. Bell, M. Brylinsky, P. Johnson-Green, Eds. (Halifax, Canada, 1999), pp. 153158. K. E. Stoderegger, G. J. Herndl, Limnol. Oceanogr. 43, 877 (1998). M. Karner, G. J. Herndl, Mar. Biol. 113, 341 (1992). T. Nagata, H. Fukuda, R. Fukuda, I. Koike, Limnol. Oceanogr. 45, 426 (2000). D. J. Repeta, T. M. Quan, L. Aluwihare, A. Accardi, Geochim. Cosmochim. Acta 66, 955 (2002). We thank the SCOR WG134 members, and all the authors and those acknowledged in the Jiao et al., Nature Reviews Microbiology (2010) for discussions and comments. This work was supported by NSFC 91028001, SOA201105021 to N. Jiao, and grants from Gordon and Betty Moore foundation Marine Microbiology Initiative and the US NSF 0962721 to F. Azam. This document is based on work partially supported by the US NFS on Oceanic Research under grant number OCE-0938349.

45

Bacterially Derived Dissolved Organic Matter in the Microbial Carbon Pump

Ronald Benner1* and Gerhard J. Herndl2 Most of the carbon fixed through photosynthesis is rapidly respired to CO2 by biota in the surface ocean, but a small fraction of this carbon is transformed into dissolved organic carbon (DOC) that persists for extended periods of time. Seawater bioassay experiments demonstrate that bacteria rapidly transform labile DOC to semilabile and refractory forms, suggesting enzymatic activity plays an important role in the transformation process. A fundamental understanding of this process has yet to be obtained, but the molecular signatures of the transformed DOC are observed throughout the ocean water column. Bacterial transformations in the microbial carbon pump (MCP) have sequestered about 10 Pg of semilabile DOC and about 155 Pg of refractory DOC in the global ocean. The annual production of semilabile and refractory DOC in the upper ocean MCP is estimated to be 0.74 to 2.23 Pg and 0.008 to 0.023 Pg, respectively. This sequestration of reduced carbon as semilabile and refractory DOC contributes to the regulation of greenhouse gases on decadal to millennial time scales and influences trace metal and nutrient availability.

bout 660 petagrams (Pg; 1015 g) of dissolved organic carbon (DOC) resides in the global ocean, but the origin, formation, structure, and reactivity of this large reservoir of reduced carbon are largely unknown (1, 2). The vast majority of this dissolved organic matter (DOM), which includes DOC as well as other dissolved organic elements, is resistant to biodegradation and comprises carbon sequestered from the atmosphere over timescales of decades to millennia. The recognition that relatively small changes in the ocean reservoir of DOC can have significant impact on the atmospheric reservoir of CO2 (1, 2) has stimulated research on the mechanisms of production of semilabile and refractory DOM in the ocean. Heterotrophic bacteria process about half of net primary production and thereby play a dominant role in the microbial carbon pump (MCP) by altering and transforming labile forms of organic matter into refractory forms that persist in the ocean (3). This article provides a short review of bioassay experiments describing the transformation of labile to refractory DOM, the use of biomarkers to trace DOC of bacterial origin in the ocean, and the size of the ocean reservoirs of semilabile and refractory DOC produced by the MCP.

Microbial Transformations of Labile to Semilabile and Refractory DOM Early bioassay experiments demonstrating the microbial production of refractory DOM from labile substrates were conducted in natural seawater samples amended with simple 14C-labeled compounds, such as glucose and leucine (4). Rapid (between two to five days) transformations of labile substrates to semilabile and refractory forms of DOM were observed in surface and deep waters incubated in the dark for several months (4, 5). Dissolved humic substances were formed (6), and the molecular weight distribution of labeled DOM at the end of the incubations was similar to that of natural marine DOM suggesting the processes occurring in these laboratory experiments were similar to those in the ocean. It was demonstrated that bacterial capsular material is an important component of the DOM produced during these experiments (7). The yields of biorefractory DOM production from simple substrates ranged from 1 to 5% (4). These studies linked refractory DOM formation with microbial transformations, but they were not able to identify the specific microbial origins, chemical forms, or mechanisms of production of refractory DOM.

1 Department of Biological Sciences and Marine Science Program, University of South Carolina, Columbia, SC 29208, USA 2 Department of Marine Biology, University of Vienna, 1090 Vienna, Austria *To whom correspondence should be addressed. E-mail: benner@mailbox.sc.edu

More recent bioassay experiments have added labile substrates, such as glucose and glutamate, as the sole carbon sources to artificial seawater inoculated with natural microbial assemblages. Bacteria rapidly used the added labile substrates and produced semilabile and refractory DOM that was chemically complex and included combined forms of neutral sugars, amino acids, and amino sugars (8). Bacteria shaped the composition of the DOM, and its refractory nature was demonstrated during long-term (1 to 1.5 yr) incubations. The DOM produced in these experiments was of similar composition and molecular weight distribution as natural marine DOM (911). Bacteria also release refractory forms of chromophoric DOM (CDOM) of varying molecular weight during the utilization of glucose (12, 13). Fluorescent components of the CDOM were resistant to photodegradation, and photochemical transformations of the DOM did not enhance its subsequent bioavailability. These experiments suggest that bacteria are a likely source of the fluorescent DOM maximum observed in the oxygen-minimum zone of the ocean (14). The mechanisms of release of DOM from bacteria have been examined in bioassay experiments using labile substrates as sole carbon sources. The direct release of dissolved D/L-enantiomers of hydrolysable amino acids and amino sugars from bacteria was observed during exponential growth of bacterial cells (15). The bioavailability of the DOM released during cell growth was highly variable and included labile, semilabile, and refractory components. The release of dissolved amino acids from bacteria during viral lysis was observed using a model system comprising a bacterial strain and a strain-specific virus (16). Combined forms of amino acids were abundant in the lysate, and the nonprotein amino acid diaminopimelic acid, a unique component of the peptide bridge in peptidoglycan, was observed. A model system using glucose as the sole carbon source and a specific bacterium and its ciliate grazer was used to investigate the release of DOM from bacteria during grazing (17). The presence of the ciliate enhanced the production of bacterially derived DOM, but the grazing process did not appear to alter the chemical composition of the resulting DOM. The vast majority of molecular masses identified as refractory DOM at the end of the experiments were produced during the first two days of the incubation before the ciliate was added to the culture. Taken together, these studies indicate that bacterially derived DOM is rapidly produced through direct release from growing cells, viral lysis, and protozoan grazing. Bacterially derived DOM ranges in bioavailability from labile to refractory. The rapid accumulation of DOM that is highly resistant to microbial degradation is both surprising and baffling. The bacterial transformation of labile to semilabile and refractory DOC occurs within hours to days at room temperature and in the dark (8). How does this happen? The exact mechanism is unclear, but the process is biologically mediated. Nonspecific enzyme activities could play

46

Fig. 1. (A) Concentrations of total dissolved D-Aspartic acid (D-Asp), D-Glutamic acid (D-Glu), and D-Alanine (D-Ala) in the water column at the Hawaii Ocean Time Series Station Aloha (20). (B) Average concentrations of DOC derived from bacteria as calculated from the yields of total dissolved D-Asp, D-Glu, and D-Ala in the water column and the yields of these D-amino acids in bioassay experiments (20).

an important role in producing chemically complex and biologically refractory DOM (8). The existence of such a mechanism would suggest that semilabile and refractory DOM are produced largely in the upper ocean where most biological production and decomposition occur. It is critical to link observations from laboratory experiments to the ocean carbon cycle, and the bioassay experiments provide this connection by demonstrating that semilabile and refractory DOM produced in the MCP include bacterial biomarkers such as the D-enantiomers of alanine (Ala), glutamic acid (Glu), and aspartic acid (Asp). These bacterial biomarkers can be used to trace the production, reactivity, abundance, and distribution of semilabile and refractory DOM in the ocean. Tracing Bacterially Derived DOC and the MCP in the Ocean The contributions of bacteria to marine DOM has gained attention over the last decade following the observation that bacteria contribute to marine dissolved organic nitrogen based on the abundance of D-enantiomers of specific amino acids (D-Ala, D-Glu, D-Asp) in marine DOM (18). The L-enantiomers of amino acids are common to all organisms and are the building blocks of proteins, whereas specific D-enantiomers are synthesized by bacteria and incorporated into a variety of unusual cell wall and membrane molecules (1921). The D-amino acids are ubiquitous in marine DOM and particulate organic matter and are most useful for tracing bacterial contributions because they can be measured directly in seawater without preconcentration, they are found in all marine bacteria and bacterially derived DOM, they occur in a variety of biochemical components of bacterial cells, and they are not known to occur in combined form in any marine organisms besides bacteria (15, 20, 2225). Other bacterial biomarkers have been observed in DOM, including muramic acid (26), diaminopimelic acid (27), and short-chain 3-hydroxy fatty acids (28), further indicating the diversity of bacterially derived compounds in seawater DOM. The concentrations and depth distribution of total dissolved D-Ala, D-Glu, and D-Asp in the water column at the Hawaii Ocean Time Series Station Aloha are shown in Fig. 1A. Concentrations of individual D-amino acids range from 4 to 12 nM in surface waters and decline by approximately 70% at a depth of 750 m. The D-amino acids removed in the upper 750 m are considered to be components of semilabile DOM

based on ventilation ages (years to decades) for these water masses (10). The concentrations of individual D-amino acids in deep waters (2500 and 4000 m) are low (1 to 4 nM) and relatively constant, indicating the refractory nature of bacterially derived DOM in the deep ocean (20). The yields of D-amino acids (D-Asp, D-Ala, D-Glu) can be used to estimate the bacterial contributions to DOC in the ocean (20). Based on these three D-amino acid biomarkers, the average concentrations of bacterially derived DOC in the water column at Station Aloha are shown in Fig. 1B. Concentrations of bacterially derived DOC range from ~7 to 25 M, indicating the MCP is a major source of semilabile and refractory DOC in the ocean. The Ocean Reservoir of Semilabile and Refractory DOC Derived from the MCP The ocean reservoir of DOC is chemically complex and has multiple origins (2932). The MCP contributes to the semilabile and refractory DOC reservoirs, and bacterial biomarkers provide an approach for tracing and quantifying the contributions of the MCP to marine DOC. The D-amino acid biomarkers have been used to estimate that about 25% of the DOC throughout the ocean water column is of bacterial origin (20). The global ocean reservoir of DOC is estimated to be approximately 660 Pg, with about 40 Pg in the form of semilabile DOC and the remaining 620 Pg in the form of refractory DOC (33). Based on these totals and the fraction of bacterially derived DOC in each reservoir, approximately 10 Pg of semilabile DOC and approximately 155 Pg of refractory DOC of bacterial origin reside in the global ocean (Fig. 2). The rates of production of semilabile and refractory DOM in the MCP can be estimated from the annual rate of heterotrophic bacterial production in the upper ocean and the efficiency of DOC release from bacteria. This approach assumes that bacteria use, alter, and transform the fixed carbon derived from primary production that is not respired and remineralized to CO2 in the euphotic zone. Furthermore, it assumes that bacterial production, a measurable quantity, is representative of these activities. Bacterial production in the upper ocean is estimated to be approximately 15% of net primary production (34), or about 7.5 PgC y-1. If it is assumed that 10 to 30% of bacterial production is released as DOC in the MCP (35), this means that 0.75 to 2.25 Pg of DOC is produced annually in the upper ocean MCP. Bioassay

47

experiments indicate that only 1 to 5% of this DOC is refractory on time scales of several months to a year, and for the purpose of this calculation we have assumed that 1%, or 0.008 to 0.023 Pg, is transformed into refractory DOC annually. Based on these estimates, the remaining component of the DOC produced annually in the upper ocean MCP, 0.74 to 2.23 Pg, is presumed to be semilabile DOC (Fig. 2). These calculations also indicate that refractory DOC production from heterotrophic bacteria in the upper ocean MCP could account for 0.015 to 0.023% of annual primary production. Semilabile DOC

production in the MCP accounts for 1.5 to 4.5% of annual primary production. For comparison, approximately 0.1% of annual primary production is buried in marine sediments via sinking particles in the biological pump (36). Bacteria are present throughout the ocean water column and in marine sediments, but the productivity of these bacteria is not well known. Likewise, the role of Archaea in the MCP also has not been considered in this analysis. Therefore, these estimates of the production of semilabile and refractory DOC in the MCP are likely to be conservative.

Fig. 2. Global ocean reservoirs of semilabile DOC (SDOC) and refractory DOC (RDOC) derived from bacteria in the microbial carbon pump. Estimated annual fluxes of the bacterial transformation of labile DOC (LDOC) into SDOC and RDOC in the upper ocean microbial carbon pump.

References and Notes 1. D. A. Hansell, C. A. Carlson, Eds., Biogeochemistry of Marine Dissolved Organic Matter (Academic Press, London, 2002). 2. H. Ogawa, E. Tanoue, J. Oceanogr. 59, 129 (2003). 3. N. Jiao, et al., Nat. Rev. Microbiol. 8, 593 (2010). 4. J. E. Brophy, D. J. Carlson, Deep Sea Res. 36, 497 (1989). 5. A. Heissenberger, G. J. Herndl, Mar. Ecol. Prog. Ser. 111, 129 (1994). 6. L. J. Tranvik, FEMS Microbiol. Ecol. 12, 177 (1993). 7. K. Stoderegger, G. J. Herndl, Limnol. Oceanogr. 43, 877 (1998). 8. H. Ogawa, Y. Amagi, I. Koike, K. Kaiser, R. Benner, Science 292, 917 (2001). 9. R. Benner, J. D. Pakulski, M. McCarthy, J. I. Hedges, P. G. Hatcher, Science 255, 1561 (1992). 10. K. Kaiser, R. Benner. Mar. Chem. 113, 63 (2009). 11. H. Ogawa, N. Ogura, Nature 356, 696 (1992). 12. G. D. Kramer, G. J. Herndl, Aquat. Microb. Ecol. 36, 239 (2004). 13. K. Shimotori, Y. Omori, T. Hama. Aquat. Microb. Ecol. 58, 55 (2009). 14. Y. Yamashita, E. Tanoue, Nat. Geosci. doi:10.1038/ngeo279 (2008). 15. N. Kawasaki, R. Benner, Limnol. Oceanogr. 51, 2170 (2006). 16. Middelboe, M., N. O. G. Jrgensen, J. Mar. Biol. Ass. U.K. 86, 605 (2006). 17. D. F. Gruber, J. P. Simjouw, S. P. Seitzinger, G. L. Taghon, Appl. Environm. Microbiol. 72, 4184 (2006). 18. M. D. McCarthy, J. I. Hedges, R. Benner, Science 281, 231 (1998). 19. Y. Asano, T. L. Lbbehsen, J. Biosci. Bioeng. 89, 295 (2000). 20. K. Kaiser, R. Benner, Limnol. Oceanogr. 53, 99 (2008). 21. K. H. Schleifer, O. Kandler, Bact. Rev. 36, 407 (1972). 22. Dittmar, T., Fitznar, H. P., G. Kattner, Geochim. Cosmochim. Acta 65, 4103 (2001). 23. B. A. Lomstein, B. B. Jorgensen, C. J. Schubert, J. Niggemann, Geochim.

Cosmochim. Acta, 70, 2970 (2006). 24. N. Kawasaki, R. Sohrin, H. Ogawa, T. Nagata, R. Benner, Aquat. Micro. Ecol. 62, 165 (2011). 25. M. T. Prez, C. Pausz, G. J. Herndl, Limnol. Oceanogr. 48, 755 (2003). 26. R. Benner, K. Kaiser, Limnol. Oceanogr. 48, 118 (2003). 27. N. O. G. Jrgensen, R. Stepanaukas, A. G. U. Pedrson, M. Hansen, O. Nybroe, FEMS Microbiol. Ecol. 46, 269 (2003). 28. S. G. Wakeham, T. K. Pease, R. Benner, Org. Geochem. 34, 857 (2003). 29. R. Benner, Chemical composition and reactivity in Biogeochemistry of marine dissolved organic matter, D. A. Hansell and C. A. Carlson Eds. (Academic press, New York, 2002), pp. 5990. 30. T. Dittmar, J. Paeng, Nat. Geosci. 2, 175 (2009). 31. N. Hertkorn, et al., Geochim. Cosmochim. Acta, 70, 2990 (2006). 32. L. A. Ziolkowski, E. R. M. Druffel, Geophys. Res. Lett. 37, doi: 1029/2010GL043963 (2010). 33. D. Hansell, C. A. Carlson, D. J. Repeta, R. Schlitzer, Oceanogr. 22, 52 (2009). 34. H. Ducklow, Bacterial production and biomass in the ocean. In Microbial Ecology of the Oceans, D. L. Kirchman Ed. (John Wiley, New York, 2000), pp. 85120. 35. T. Nagata. Production mechanisms of dissolved organic matter,. In Microbial ecology of the oceans, D. L. Kirchman, Ed. (John Wiley, New York, N.Y. 2000), pp. 121152. 36. J. I. Hedges, Mar. Chem. 39, 67 (1992). 37. This work was supported by the Scientific Committee on Oceanic Research (WG134), the U.S. National Science Foundation (0080782 and 0850653 to RB) and the Austrian Science Foundation (486-B09 and 23234-B11 to GJH).

48

Role of Photoheterotrophic Bacteria in the Marine Carbon Cycle

Michal Koblek Uncovering the ecological role of photoheterotrophic bacteria in the ocean is one of the main accomplishments of marine microbiology of the past decade. Marine photoheterotrophs include two main groups: Aerobic anoxygenic phototrophic (AAP) bacteria and proteorhodopsin (PR)containing bacteria. These organisms make up large fractions of the microbial communities inhabiting the euphotic zone of worlds oceans. In spite of similar metabolisms, AAP and PR-containing bacteria differ in their photochemistry, physiology, and ecology. The ability to use light energy appears to allow more economical utilization of dissolved organic matter, which implies that photoheterotrophic bacteria may play a unique role in the microbial carbon pump. Discovery of Photoheterotrophic Bacteria in the Ocean In 2000, two independent studies documented the widespread ability of marine bacteria to utilize light energy. In the first article, which appeared in the September 14 issue of Nature, Kolber et al. (1) reported the observation of distinct signals of bacteriochlorophyll a-containing microorganisms in the surface waters of the tropical Pacific. The recorded signals were linked with the presence of aerobic anoxygenic phototrophic (AAP) bacteria. These organisms contain bacteriochlorophyll a as the main light-harvesting pigment but, in contrast to purple nonsulfur photosynthetic bacteria, they are obligate aerobes requiring oxygen for their metabolism and growth (2, 3). Only one day later, Bj et al. (4) announced in Science the discovery of a new rhodopsin gene in DNA samples recovered from Monterey Bay in California. Rhodopsins are well-known membrane proteins serving as photoreceptor molecules in Metazoa, including humans, or as proton pumps in halophilic Archaea, but until Bjs report they were not thought to be present in Proteobacteria. For this reason the newly discovered protein was termed proteorhodopsin (PR) to make a clear distinction from the only distantly related archaeal bacteriorhodopsins and visual rhodopsins (see box on right). These and follow up studies (57) have attracted significant scientific attention. The ability to use light energy challenged the classical view of marine bacteria as strictly heterotrophic organisms fully dependent on recycling dissolved organic matter (DOM) produced by photoautotrophic phytoplankton (Fig. 1). An important turning point represented the identification of the PR gene in a SAR11 isolate, Pelagibacter ubique (8). The SAR11 clade represents the most abundant bacterial group inhabiting the upper ocean (9), which suggests that photoheterotrophy could be a common phenomenon. Light Utilization in Photoheterotrophic Bacteria The phototrophic competence of AAP and PR-containing bacteria, and the way they use light as an energy source, remains a matter of ongoing research. These organisms contain photochemically active reaction centers, but are not able to grow photoautotrophically, as they require a supply of organic carbon (3, 5, 1012). Experiments with Erythrobacter sp. NAP1 and Dokdonia sp. MED134 revealed light-enhanced CO2 incorporation, however the activity was only weak, likely reflecting enhanced anaplerotic carboxylation activity (5, 11). In AAP bacteria, light has been shown to inhibit respiration and stimulate synthesis of ATP, which indicates that photophosphorylation replaces oxidative phosphorylation (12, 13). Physiological experiments conducted with heterologously expressed PR proteins demonstrated that light exposure

MARINE PHOTOHETEROTROPHIC ORGANISMS Aerobic Anoxygenic Phototrophic (AAP) Bacteria These organisms harvest light by bacteriochlorophylls and carotenoids. The excitation energy is transferred to pheophytinquinone-type reaction centers, which drive the electron transport and ATP synthesis. In contrast to their close relatives, purple photosynthetic bacteria, AAP bacteria are strict aerobes. They do not form a compact phylogenetic group, being scattered among a number of clades of Alpha-, Beta-, and Gammaproteobacteria. Proteorhodopsin-containing Bacteria Proteorhodopsin is a pigment protein composed of the protein moiety, opsin, and a pigment cofactor, retinal. Proteorhodopsins are analogous to visual rhodopsins and bacteriorhodopsins found in Archaea. The absorption of light by retinal causes the translocation of protons across the membrane driving ATP synthesis (Photophosphorylation). PR genes were found in a number of species including Alpha- and Gammaproteobacteria, Flavobacteria, Actinobacteria, and Archaea. led to the formation of proton gradients across the membrane (4, 14), which also confirms that PR drives photophosphorylation and provides ATP for cellular metabolism. Thus, AAP and PR-containing bacteria are photoheterotrophic organisms, as they use light energy to supplement their primarily heterotrophic metabolism. The ability to use light energy could allow phototrophs to store more carbon (which would otherwise be respired) in their biomass. Indeed, light-enhanced biomass accumulation has been repeatedly confirmed in AAP bacteria (5, 15, 16), but in PR-containing organisms the situation is less clear. The first experiments showed no growth stimulation by light in Pelagibacter ubique (8). In contrast, later work with Dokdonia sp. MED134 and Vibrio sp. AND4 demonstrated enhanced growth and better survival under starvation conditions when the bacteria were exposed to light, indicating that PR provided energy for growth (17, 18). Distibution of Photoheterotrophs In his pioneering paper, Kolber et al. (1) hypothesized that the ability to utilize light might be beneficial especially in nutrient-poor environments. The first quantification of PR-containing bacteria was attempted from the frequency of PR genes in metagenomic libraries. Using samples from the Mediterranean and Red Sea, it was estimated that PR genes were present in 13% of total bacteria (19). A more direct approach was used in the waters of the North Atlantic, applying quantitative polymerase chain reaction technique. It was found that PRcontaining bacteria represented between 9 and 53% of total bacteria

Institute of Microbiology CAS, Opatovick mln, 379 81 T rebo n, Czech Republic E-mail: koblizek@alga.cz

49

Fig. 1. Schematic representation of dissolved organic matter (DOM) cycling in the upper ocean. Photosynthetic phytoplankton are photoautotrophic organisms, which harvest light energy to support inorganic carbon fixation. These so-called primary producers provide organic matter for heterotrophic and photoheterotrophic species. Photoheterotrophic bacteria utilize organic matter for growth, but are capable of using light energy for their metabolism. They are unable to perform autotrophic growth, but can grow heterotrophically in the dark. Heterotrophic bacteria (Organotrophs) depend on organic matter produced by autotrophic species both as a source of energy and as a substrate for growth.

in the Sargasso Sea, and in the more productive waters of the North Atlantic they still represented 4 to 15% of total bacteria (20). AAP bacteria can be conveniently counted by infrared epifluorescence microscopy. Despite some initial controversies, it has been found that AAP bacteria on average constitute 1 to 7% of total prokaryotes in oligotrophic areas (2124). In more productive environments such as coastal waters, shelf seas, and river estuaries, they represent up to 30% of total prokaryotes (2527). The latter data contradicts the assumption that photoheterotrophy would be more advantageous in nutrient poor areas, as AAP bacteria appear to prefer more eutrophic environments. However, Kolbers hypothesis seems to hold for PR-containing bacteria, which have indeed been shown to be abundant in oligotrophic regions (Fig. 2). Role of Photoheterotrophs in the Marine Carbon Cycle The role of photoheterotrophs in the microbial carbon pump (MCP) is complex (28). Their ability to use light energy can reduce their carbon requirement, which means that less organic matter would be respired and converted into CO2. This might be especially important in the oligotrophic regions of the ocean (29), which are generally considered to be carbon sinks (30). The additional energy from light may also help to fuel various energy-demanding processes such as active transport of substrates and nutrients across membranes, production of ectoenzymes, breakdown of complex organic molecules, and cell motility. In spite of the fact that PR genes are present in very diverse organisms (31), most of them are likely to be oligotrophic species inhabiting nutrient-scarce ocean areas, potentially giving them a unique place and role in the MCP. In contrast, AAP bacteria seem to be highly active organisms with larger cell sizes (21) and high growth rates (23). This suggests that despite their smaller numbers, AAP bacteria can process a large part of the available DOM, most likely the more readily accessible labile DOM. Our knowledge of photoheterotrophic organisms has expanded significantly in the past decade. The need for appropriate tools led to the development of many new experimental techniques. Yet, our understanding of the role photoheterotrophs play in the marine environment and organic matter cycling is still only fragmentary. The increasing number of organ-

Fig. 2. Contribution of photoheterotrophic bacteria to the microbial community (expressed as a percentage of total prokaryotes) in various marine environments with different trophic status. Proteorhodopsin-containing bacteria abundance data were taken from (20). AAP data for the subtropical Atlantic were taken from (23), for the Baltic Sea from (25), and for the Mediterranean Sea from (32).

50

FIG. 2 KOBLIZEK

isms available in laboratory cultures makes it possible to design better experiments, which, in combination with genomic data, can provide important insights into cell metabolism and photobiology. Field studies should focus on environmental factors driving the distribution and ac-

tivity of photoheterotrophic species. Manipulation and enrichment experiments can be especially beneficial in elucidating the main bottomup and top-down factors shaping the natural microbial populations and establishing the role these organisms play in the marine carbon cycle.

References and Notes 1. Z.S. Kolber, C.L. Van Dover, R.A. Niederman, P.G. Falkowski, Nature 407, 177 (2000). 2. T. Shiba, U. Simidu, N. Taga, Appl. Environ. Microbiol. 38, 43 (1979). 3. V.V. Yurkov, J.T. Csotonyi, in The Purple Phototrophic Bacteria, Advances in Photosynthesis and Respiration volume 28, Hunter, C.N., Daldal, F., Thurnauer, M.C. and Beatty, J.T. Eds. (Springer Verlag, Dordrecht, The Netherlands, 2009), pp. 3155. 4. O. Bj et al., Science 289, 1902 (2000). 5. Z.S. Kolber et al., Science 292, 2492 (2001). 6. O. Bj, E.N. Spudich, J.L. Spudich, M. Leclerc, E.F. DeLong, Nature 411, 786 (2001). 7. Bj, O. et al., Nature 415, 630 (2002). 8. S.J. Giovannoni et al., Nature 438, 82 (2005). 9. R.M. Morris et al., Nature 420, 806 (2002). 10. B.M. Fuchs et al., Proc. Natl. Acad. Sci. USA 104, 2891 (2007). 11. J.M. Gonzlez et al., Proc. Natl. Acad. Sci. USA 105, 8724 (2008). 12. M. Koblek, J. Ml coukov, Z Kolber, J. Kopeck, Arch. Microbiol. 192, 41 (2010). 13. K. Okamura, F. Mitsumori, O. Ito, K.-I. Takamiya, M. Nishimura, J. Bacteriol. 168, 1142 (1986). 14. A. Martinez, A.S. Bradley, J.R. Waldbauer, R.E. Summons, E.F. DeLong, Proc. Natl. Acad. Sci. USA 104, 5590 (2007). 15. Y. Shioi, Plant Cell Physiol. 27, 567 (1986). 16. V.V. Yurkov, H. van Gemerden, Arch. Microbiol. 159, 84 (1993). 17. L. Gmez-Consarnau et al., Nature 445, 210 (2007).

18. L. Gmez-Consarnau et al., PLoS Biol 8, e1000358, 10.1371/journal. pbio.1000358 (2010). 19. G. Sabehi et al., PLoS Biol. 3, e173, 10.1371/journal.pbio.0030273 (2005). 20. B.J. Campbell, L.A. Waidner, M.T. Cottrell, D.L. Kirchman, Environ. Microbiol. 10, 99 (2008). 21. M.E. Sieracki, I.C. Gilg, E.C. Thier, N.J. Poulton, R. Goericke, Limnol. Oceanogr. 51, 38 (2006). 22. M.T. Cottrell, A. Mannino, D.L. Kirchman. Appl. Env. Microbiol. 72, 557 (2006). 23. M. Koblek, M. Man, J. Ras, A.J. Poulton, O. Pril, Environ. Microbiol. 9, 2401 (2007). 24. N. Jiao et al., Environ. Microbiol. 9, 3091 (2007). 25. M. Man et al., Aquat. Microbial. Ecol. 45, 247 (2006). 26. Y. Zhang, N. Jiao, FEMS Microbiol. Ecol. 61, 459 (2007). 27. M.T. Cottrell, J. Ras, D.L. Kirchman, ISME J. 4, 945 (2010). 28. N. Jiao et al., Nat. Rev. Microbiol. 8, 593 (2010). 29. N. Jiao et al., ISME J. 4, 595 (2010). 30. P.A. del Giorgio, C.M. Duarte, Nature 420, 379 (2002). 31. J.C. Venter et al., Science 304, 66 (2004). 32. E. Hojerov, personal communication project P501/10/0221, AV CR project 33. This work was supported by GACR M200200903, project Algatech (CZ.1.05/2.1.00/03.0110), and the Institutional Research Concept AV0Z50200510. M.K. thanks Prof. Nianzhi Jiao and Dr. Isabel Ferrera for constructive comments, and Dr. Matthew T. Cottrell for kindly providing the PR abundance data.

51

Microbial Heterotrophic Metabolic Rates Constrain the Microbial Carbon Pump

Carol Robinson1* and Nagappa Ramaiah2 The respiration of dissolved organic matter by heterotrophic bacteria and Archaea represents the largest sink in the global marine biological carbon cycle, an important constraint on organic carbon supply, and the major driver of global elemental nutrient cycles. Direct measurement of heterotrophic production and respiration is difficult. However, the recent development of methods involving in vivo electron transport system activity, bioassay uptake of specific prokaryotic substrates, and nutrient addition incubations are poised to discern the complex interactions between metabolic rate, community structure, and organic and inorganic nutrient availability. In a changing global environment, it is important to understand how increasing sea surface temperature, melting sea ice, ocean acidification, variable dust deposition, and upwelling intensity will impact the metabolism of Bacteria and Archaea and so the balance between carbon sequestration and carbon dioxide evasion to the atmosphere. Continued and improved measures of prokaryotic production and respiration are vital components of this endeavor.

he downward flux of organic carbon from the surface ocean to depth via passive sinking of particles, active transport by animals, and mixing of dissolved organic matter (DOM) is known as the biological carbon pump (BCP). The microbial carbon pump (MCP) is a conceptual component of the BCP, used to describe the microbial production of refractory DOM (RDOM) which can be stored for millennia in the deep sea, rather than being respired to dissolved inorganic carbon and returned to the atmosphere (1). Heterotrophic bacteria and Archaea generate RDOM through degradation and transformation of particulate and dissolved organic matter, exudation, and cell lysis (2, 3). In addition to assimilation and transformation of recently produced DOM, prokaryotes also degrade older DOM (4) derived from photochemically transformed upwelled DOM (5, 6) and potentially from methane seeps (7). Hence, understanding the magnitude and variability of the production and respiration of bacteria and Archaea is important not only for quantifying the efficiency of the BCP (8) and the role of prokaryotes in regulating carbon fluxes (9), but also for constraining the flow of DOM through the MCP. The composition and lability of DOM affect the prokaryotic carbon demand [PrCD = prokaryotic production (PrP) + prokaryotic respiration (PrR)] and the prokaryotic growth efficiency (PrGE = PrP/PrCD, the proportion of the prokaryotic carbon demand used for prokaryotic production). PrGE is influenced by the availability of organic and inorganic substrates as well as the energetic costs of growth in a particular environment, and so tends to be low at times of nutrient limitation or environmental stress and higher during increased primary productivity and supply of nutrients (8, 10). The direct measurement of PrP and PrR and calculation of PrCD and PrGE is technically and interpretatively challenging. This is due to uncertainties associated with factors such as the pre-incubation separation of the heterotrophic bacterioplankton fraction from the rest of the plankton community, the different incubation times required for PrR and PrP measurements, the effect of light on PrP and PrR, the quantification of prokaryotic excretion of DOM, and the conversion factors used to derive rates of carbon production and respiration from radiolabeled thymidine or leucine incorporation and oxygen consumption (8,9). Large uncertainties in PrGE contribute significantly to the

School of Environmental Sciences, University of East Anglia, Norwich NR4 7TJ, U.K. National Institute of Oceanography, Dona Paula 403004, Goa, India *To whom correspondence should be addressed. E-mail: carol.robinson@uea.ac.uk
2

mismatch between measurements of mesopelagic microbial metabolic activity and estimates of the influx of organic carbon that could support this microbial activity (11). Recent methodological developments have the potential to reduce uncertainties in PrR and PrP determinations. For example, measurements of in vivo electron transport system activity estimated from the reduction of the tetrazolium salt INT are linearly related to in situ rates of respiration, and avoid problems associated with pre-incubation filtration and relatively long incubation times (24 hours) (12). Additionally, single cell assays that measure incorporation of selected organic compounds by specific prokaryotic groups compare and contrast the components of DOM taken up by bacteria and Archaea (13). Including these assays in time series studies can elucidate the influence of environmental factors such as light on PrP (14, 15). Climate change will likely affect precipitation, river flow, ice melt, atmospheric deposition, and the timing and strength of alongshore winds that stimulate coastal upwelling, and so may significantly change the supply of inorganic and organic substrates to marine prokaryotes. Concomitant increases in sea surface temperature and decreases in pH and carbonate ion concentration could lead to changes in phytoplankton and zooplankton community structure, subsequently impacting foodweb-derived DOC (16). In short-term experiments, prokaryotic turnover of phytoplankton-derived polysaccharides was increased at the lower pH levels projected to occur with a doubling of atmospheric CO2, with the potential to reduce carbon export and enhance respiratory CO2 production (17). Field studies and inorganic and organic nutrient bioassay experiments show PrR and PrGE in coastal regions to be either mainly controlled by the DOC pool or colimited by organic and inorganic nutrients (1820). Climate-driven increases in DOC supply may also impact the plankton community photosynthesis to respiration (P:R) ratio. When released from organic carbon limitation, heterotrophic prokaryotes can outcompete phytoplankton for inorganic nutrients, thereby decreasing the overall P:R ratio, increasing the proportion of DOC that is respired, and decreasing the amount that is sequestered (21). This review aims to highlight our incomplete understanding for the causes of variability in the PrGE and respiratory potential of heterotrophic bacteria and Archaea. As new research supports the pivotal role of these microbes in the present and future ocean (22, 23), the lack of routine measurements of PrP and PrR in relation to phylogenetic composition, as well as to DOM characterization and assimilation potential, becomes increasingly difficult to defend.

52

References and Notes 1. N. Jiao et al., Nat. Rev. Microbiol. 8, 593 (2010). 2. T. Nagata, in Microbial Ecology of the Oceans D.L. Kirchman Ed. (John Wiley & Sons, Inc. New York, ed. 1. 2000), pp. 121152. 3. D. F. Gruber, J. P. Simjouw, S. P. Seitzinger, and G. L. Taghon, Appl. Environ. Microbiol. 72, 4184 (2006). 4. J. Cherrier, J. E. Bauer, E. R. M. Druffel, R. B. Coffin, J. P. Chanton, Limnol. Oceanogr. 44, 730 (1999). 5. R. Benner, B. Biddanda, Limnol. Oceanogr. 43, 1373 (1998). 6. I. Obernosterer, B. Reitner, G. J. Herndl, Limnol. Oceanogr. 44, 1645 (1999). 7. J. W. Pohlman, J. E. Bauer, W. F. Waite, C. L. Osburn, N. R. Chapman, Nat. Geosci. 4, 37 (2011). 8. C. Robinson, in Microbial Ecology of the Oceans D.L. Kirchman Ed. (John Wiley & Sons, Inc. ed. 2. 2008), pp. 299334. 9. J. M. Gasol et al., Aquat. Microb. Ecol. 53, 21 (2008). 10. P. A. del Giorgio, J. J. Cole, in Microbial Ecology of the Oceans D. L. Kirchman Ed. (John Wiley & Sons, Inc. New York ed. 1. 2000), pp. 289325. 11. A. B. Burd et al., Deep Sea Res. II 57, 1557 (2010). 12. S. Martinez-Garca, E. Fernndez, M. Aranguren-Gassis, E. Teira, Limnol. Oceanogr. Methods 7, 459 (2009).

13. D.L. Kirchman, H. Elifantz, A. I. Dittel, R. R. Malmstrom, M. T. Cottrell, Limnol. Oceanogr. 52, 495 (2007). 14. M. J. Church, H. W. Ducklow, D.A. Karl, Appl. Environ. Microbiol. 70, 4079 (2004). 15. T. R. A. Straza, D. L. Kirchman, Aquat. Microb. Ecol. 62, 267(2011). 16. O. Hoegh-Guldberg, J. F. Bruno, Science 328, 1523 (2010). 17. J. Piontek, M. Lunau, N. Handel, C. Borchard, M. Wurst, A. Engel, Biogeosciences 7, 1615 (2010). 18. J. K. Apple, P.A. del Giorgio, ISME J. 1, 729 (2007). 19. L. Alonso-Sez et al., Limnol. Oceanogr. 52, 533 (2007). 20. S. Martinez-Garca et al., Aquat. Microb. Ecol. 416, 17 (2010). 21. T. F. Thingstad et al., Nature 455, 387 (2008). 22. J. Arstegui, J. M. Gasol, C. M. Duarte, G. J. Herndl, Limnol. Oceanogr. 54, 1501 (2009). 23. D. L. Kirchman, X. A. G. Moran, H. W. Ducklow, Nat. Rev. Microbiol. 7, 451 (2009). 24. We thank G. Herndl for constructive comments and suggestions. This work was supported by the Scientific Committee on Oceanic Research (WG134), The Royal Society (to CR), and the National Institute of Oceanography, India (to NR).

53

Virus-Mediated Redistribution and Partitioning of Carbon in the Global Oceans

Markus G. Weinbauer1,2*, Feng Chen3, Steven W. Wilhelm4** Viruses in marine systems influence biogeochemical cycles by releasing organic matter during the lysis of host cells. This process redistributes this biological carbon across a continuum of organic materials, from dissolved to particulate. Estimates of release rates of viral lysis products, studies manipulating viral effects on natural communities, and efforts using virus-host systems indicate that viral lysis changes the chemical composition, distribution, and character of organic matter. The majority of these lysis products can be assimilated rapidly by prokaryotes. Moreover, viral lysis appears to influence the formation and stability of organic particles (marine snow), altering carbon flow through the biological pump from the surface to deep ocean. While well defined, the net outcomes of these virus-mediated activities remain uncertain. The consistent experimental observation that viral lysis decreases the growth efficiency of residual prokaryotic communities further suggests that the activity of viruses generates, in part, organic matter that is biologically recalcitrant. While this clearly indicates an influence on the mineralization and transformation of organic matter, new observations across spatial and temporal gradients are needed to develop our understanding of the potential role of viruses in marine carbon cycles. iruses are the most abundant life forms in the ocean (1) and some studies suggest that they significantly impact the mortality rates of marine prokaryotes and eukaryotic algae (2, 3), releasing cellular material from infected hosts into the environment. This release of lysis products from all trophic levels as dissolved organic matter (DOM) and particulate organic matter (POM) has been termed the viral shunt (4) (Fig. 1). Viral lysis is now considered a major source of dissolved organic carbon (DOC) in marine systems, one that rivals leaching from phytoplankton, the collapse of bloom events (sometimes by programmed cell death), sloppy feeding by zooplankton, and egestion by protists. The DOC pool in the ocean contains approximately the same amount of carbon as is stored as CO2 in the atmosphere. This DOC, which would otherwise be lost from the rest of the foodweb, is primarily consumed by heterotrophic prokaryotes (Bacteria and Archaea). These prokaryotes are subsequently consumed by small grazers (protists), connecting the DOM to the foodweb via the microbial loop (5). Thus, quantifying the source(s) of DOM and the fate of prokaryotic carbon production is an important step Fig. 1. The viral shunt in marine foodwebs. Viruses divert the flow of carbon and nutrients in understanding the marine carbon cycle. Microorfrom secondary consumers (black arrows) by destroying host cells and releasing the ganisms are also involved in the formation and discontents of these cells into the pool of dissolved organic matter (DOM) in the ocean (green solution of particles (marine snow). The transfer of arrows). The electron micrograph insert shows a cyanobacterial cell in the process of lysis. carbon into the deep sea via sinking of biogenic particles is a major pathway of the biological pump (6). process driving this conversion has recently been termed the microbial Within this framework, microbial activity converts a fraction of organic matter into refractory DOM (RDOM), marine car- carbon pump (MCP) (7). In the following, we summarize the state of bon that in the oceans appears to be biologically unavailable to micro- knowledge concerning the potential role of viruses and their activity as organisms; as such, DOC is relatively old (~5,000 years of age). The a source of carbon, RDOM, and precursors for aggregate formation, as well as the implications of this activity for the biological and microbial carbon pumps (7). Viral Lysis Products and Virus-mediated Aggregation of Organic Matter The composition of virus-generated lysis products is poorly studied. Nevertheless, studies with laboratory virus-host systems (VHSs), as well as microbial communities, have shown that lysis changes the composition of DOM (8, 9). In VHSs, material from the bacterial cell wall, including various D-isomers of amino acids, glucosamine, and diaminopimelic acid have been found in the dissolved fraction (8). An
FIG. 1. WEINBAUER, CHEN AND WILHELM
1 Microbial Ecology & Biogeochemistry Group, Universit Pierre et Marie CurieParis6, Laboratoire dOcanographie de Villefranche, 06230 Villefranche-sur-Mer, France 2 CNRS, Laboratoire dOcanographie de Villefranche, 06230 Villefranche-sur-Mer, France 3 Institute of Marine Environmental Technology, University of Maryland Center for Environmental Science, Baltimore, MD 21202, USA 4 Department of Microbiology, University of Tennessee, Knoxville, TN 37996, USA To whom correspondence should be addressed. *E-mail: wein@obs-vlfr.fr **E-mail: wilhelm@utk.edu

54

accumulation of polymeric and total DOM (such as amino acids and carbohydrates) due to viral lysis has also been demonstrated (8, 9). VHS studies have also shown a promotion of the formation of submicron colloids (10). Overall, viral lysis predictably increases the DOM pool, particularly the polymeric and colloidal components. Reports of direct release rates of lysis products remain rare due to methodological challenges (for example, 11). However, research using estimations of cellular carbon:nitrogen:phosphorus quota and mortality rates has shown that viral lysis transforms significant quantities of microbial biomass into DOM and colloidal pools (12, 13). Experimental studies with VHSs (12, 1416), model communities (17) and natural communities (11, 18) indicate that the majority of these lysis products are rapidly (within days) degraded and belong to the labile DOM (LDOM) fraction. This should result in a reduced transfer of carbon to higher trophic levels and fuel the microbial part of the foodweb (19). In an elegant study, the effect of lysis products on bacterial activity was studied using a bacterial VHS and the lysis products from this VHS (15). Bacterial carbon production and enzymatic activity increased, while bacterial growth efficiency decreased in the presence of the either added viruses or lysis products. These observations can be explained by an increase in prokaryotic energy demand associated with the degradation of polymeric organic nitrogen and phosphorus from the lysate products. Subsequent research on microbial communities, manipulating the presence and absence of viruses using various experimental approaches, demonstrates a consistent pattern: Lysis increases prokaryotic respiration and decreasFig. 2. The viral shunt in the context of oceanic carbon cycling. Potential major influences of virus es growth efficiency (20, 21). This means are shown in green. Note that the grazing food chain, the microbial loop, and the viral shunt are only that more organic matter has been processed shown schematically (for more details, see Fig. 1). The role of the microbial carbon pump (MCP) is and turned into CO2 and that a greater pershown as well. This scheme assumes that aggregation of marine snow dominates in surface water, centage of the nutrients within the DOM are whereas dissolution dominates in the dark ocean. DOM, dissolved organic matter; POM, particulate mineralized (16). Within this context it is organic matter; RDOM, refractory DOM. also conceivable that enhanced prokaryotic respiration and reduced growth efficiency should prime the MCP to produce more RDOM, or at least increase the Viruses and infected cells can also be found on, as well as in, agratio of RDOM to LDOM and semilabile DOM (SDOM). This is one gregates (28). This may increase the dissolution of sinking particles area that remains ripe for experimental exploration. through the direct destruction of particle nucleating bacteria or the reAnother consequence of virus activity is the potential to influence lease of intracellular enzymes that facilitate aggregate dissolution. The the aggregation of organic matter (Fig. 2). There is experimental evi- true role of viruses in particle dissolution and aggregation is likely a dence that viruses delay the formation of phytoplankton blooms (3, 22) delicate balance of these two processes (29). and thus the formation of biotic (organic) particles (23). Similar trends have been found for prokaryotic aggregation and particle colonization Viruses and the Biological Carbon Pump (24, 25). Viruses can, however, increase the size and stability of algal- The export of carbon out of the euphotic zone by the biological pump is derived aggregates (23), for instance through the formation of colloidal a central component of marine carbon cycles (6). The location of viral material and the release of sticky intracellular contents (10). In one activity within the water column, in respect to the depth and strength mesocosm study, the termination of a phytoplankton bloom by viral of the pycnocline, is thus crucial as it influences whether the lysed malysis was associated with a large production of biological aggregates terial is recycled in the euphotic zone or exported into the deep sea. To (26, 27). describe this, a conceptual model was developed with three scenarios

FIG. 2. WEINBAUER, CHEN AND WILHELM

55

for infected cells of the bloom-forming alga Heterosigma akashiwo and viruses that infect them (30). In scenario 1, when stratification is strong or the pycnocline is deep, cells lyse above the pycnocline and lysis products remain in the mixed layer. In scenario 2, when stratification is weak and waters are shallow, infected cells will reach the benthos before lysis and lysis products remain at the sediment-water interface. In scenario 3, cells are also not retained by the pycnocline and lysis products are subject to processes in the deeper water column. The exact fate and quantity of primary production shunted into the DOM and colloidal pool will depend on the actual scenario in play. In addition to the location of viral infection in the water column, the following mechanisms may influence the biological pump: (i) The conversion of cellular material into the dissolved and colloidal forms should increase the retention time of carbon and nutrients in surface water and thus reduce export; (ii) Aggregation driven by lysis products will increase carbon export unless this aggregation also increases the buoyancy of particles (for example, by retaining gases from metabolism) and thus the retention time in the euphotic zone; (iii) Nutrient (nitrogen, phosphorus, and iron) release within lysis products may act as a feedback mechanism and stimulate primary production with unpredictable consequences for the biological pump; and (iv) If viral lysis stimulates the microbial loop and reduces carbon transfer to higher trophic levels, less carbon is available for zooplankton-mediated particle formation (for example, fecal pellets or larvacean houses); this would shift aggregation to one based on phytoplankton-derived materials. Since these mechanisms are difficult to disentangle (and others yet might not be recognized), it is still not possible to assess whether the net effect of viral activity is to prime or short-circuit the biological pump

(31). Indeed, research during the last two decades has revealed that virus activity depends strongly on spatial and temporal scales (even in the deep sea) (32, 33). Thus, the role of viruses within the MCP and the biological pump is potentially environment and condition specific. Viruses and the Microbial Carbon Pump Functionally there now also appears to be a horizontal component (the pumping of biomass into the RDOM pool) within microbial carbon cycles, which, at some level, is undoubtedly influenced by the viral shunt (7). In the euphotic zone, viral lysis of picocyanobacteria and autotrophic eukaryotic plankton will contribute, in addition to bacterial cell lysates, to the DOM and subsequently RDOM pools (and finally export). Since representative cyanobacterial VHSs are available for the euphotic zone, they may serve as good models for understanding the dispersal of virus-mediated lysis products in the natural environment. In the deep ocean (where most of the recalcitrant carbon is found), significant viral activity has been reported for benthic and pelagic systems (33, 34). Viral lysis of Bacteria and Archaea (which become more abundant in the dark ocean) should produce mainly LDOM. Since this material will be consumed rapidly (and potentially help sustain high prokaryotic production in the deep sea), lysis may increase the ratio of RDOM to LDOM and SDOM. In a similar manner to the conversion of some LDOM and SDOM into RDOM by prokaryotic activity (35), viral lysis in the deep sea should also promote RDOM production directly in the environment where long term storage occurs. Since RDOM is resistant to microbial utilization, is stored in the ocean for millennia, and accounts for more than 95% of the total DOC pool (7), alterations of RDOM dynamics within the microbial carbon pump by viral activity will undoubtedly influence the global carbon cycle.

1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18.

References and Notes . Bergh, K. Y. Brsheim, G. Bratbak, M. Heldal, Nature 340, 467 (1989). L. M. Proctor, J. A. Fuhrman, Nature 343, 60 (1990). C. A. Suttle, A. M. Chan, M. T. Cottrell, Nature 347, 467 (1990). S. W. Wilhelm, C. A. Suttle, Bioscience 49, 781 (1999). F. Azam et al., Mar. Ecol. Prog. Ser. 10, 257 (1983). R. W. Eppley, B. J. Peterson, Nature 282, 677 (1979). N. Jiao et al., Nat. Rev. Microbiol. 8, 593 (2010). M. Middelboe, N. O. G. Jorgensen, J. Mar. Biol. Ass. UK 86, 605 (2006). M. G. Weinbauer, P. Peduzzi, Mar. Ecol. Prog. Ser. 127, 245 (1995). A. Shibata, K. Kogure, I. Koike, K. Ohwada, Mar. Ecol. Prog. Ser. 155, 303 (1997). L. Poorvin, J. M. Rinta-Kanto, D. A. Hutchins, S. W. Wilhelm, Limnol. Oceanogr. 49, 1734 (2004). C. J. Gobler, D. A. Hutchins, N. S. Fisher, E. M. Cosper, S. Saudo-Wilhelm, Limnol. Oceanogr. 42, 1492 (1997). S. W. Wilhelm, M. G. Weinbauer, C. A. Suttle, W. H. Jeffrey, Limnol. Oceanogr. 43, 586 (1998). G. Bratbak, A. Jacobson, M. Heldal, Aquat. Microb. Ecol. 16, 11 (1998). M. Middelboe, N. O. G. Jrgensen, N. Kroer, Appl. Environ. Microbiol. 62, 1991 (1996). M. Middelboe, L. Riemann, C. F. Steward, W. Hannsen, O. Nybroe, Aquat. Microb. Ecol. 33, 1 (2003). J. Haaber, M. Middelboe, ISME J. 3, 430 (2009). R. T. Noble, J. A. Fuhrman, Aquat. Microb. Ecol. 20, 1 (1999).

19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36.

J. A. Fuhrman, Nature 399, 541 (1999). M. Middelboe, P. G. Lyck, Aquat. Microb. Ecol. 27, 187 (2002). C. Motegi et al., Limnol. Oceanogr. 54, 1901 (2009). C. A. Suttle, Mar. Ecol. Prog. Ser. 87, 105 (1992). P. Peduzzi, M. G. Weinbauer, Limnol. Oceanogr. 38, 1562 (1993). A. Malits, M. G. Weinbauer, Aquat. Microb. Ecol. 54, 243 (2009). L. Riemann, H.-P. Grossart, Microb. Ecol. 56, 505 (2008). C. Brussaard, B. Kuipers, M. Veldhuis, Harmful Algae 4, 894 (2005). C. Brussaard, X. Mari, J. D. L. van Bleijswijk, M. Veldhuis, Harmful Algae 4, 875 (2005). L. M. Proctor, J. A. Fuhrman, Mar. Ecol. Prog. Ser. 69, 133 (1991). M. G. Weinbauer et al., Aquat. Microb. Ecol. 57, 321 (2009). J. E. Lawrence, C. A. Suttle, Aquat. Microb. Ecol. 37, 1 (2004). C. P. Brussaard et al., ISME J. 2, 575 (2008). J. M. Rowe et al., Aquat. Microb. Ecol. 52, 233 (2008). M. G. Weinbauer, I. Brettar, M. G. Hfle, Limnol. Oceanogr. 48, 1457 (2003). R. Danovaro et al., Nature 454, 1084 (2008). H. Ogawa, Y. Amagai, I. Koike, K. Kaiser, R. Benner, Science 292, 917 (2001). The authors thank C.A. Suttle for his mentorship. They also appreciate the thorough comments of G. Bratbak. They also acknowledge the French Science Ministry (ANR-AQUAPHAGE, NANR 07 BDIV 015-06; ANRMAORY; N ANR 07 BLAN 016) for support to MGW, NSF(OCE) 0825405 and 1061352 for support to SWW and the SCOR Working Group 134.

56

DOC Persistence and Its Fate After Export Within the Ocean Interior
Craig A. Carlson1*, Dennis A. Hansell2, and Christian Tamburini3,4 Biotic and abiotic processing of organic matter originally created through photosynthesis can lead to its accumulation as dissolved organic matter (DOM) in surface waters of the ocean.Components of this dissolved material are resistant to microbial remineralization, persisting long enough to be entrained during ventilation of the ocean interior, and resulting in carbon export to great ocean depths.Upon downward mixing to mesopelagic waters (from >150 to ~1000 m), the exported DOM is susceptible to removal on both short (weeks) and long (years) time scales, with remineralization by deep microbial populations playing an important role. Recent work indicates that specific lineages of subsurface bacterioplankton respond to these export events, remineralizing the same DOM that had been largely resistant to microbial degradation at the sea surface.Free-living (unattached) microbial populations are structured vertically, with that structure defined by unique metabolic capabilities that are important when considering the fate of exported DOM and its lability. issolved organic carbon (DOC) represents the largest pool of reduced carbon in the ocean with a global inventory of approximately 662 petagrams of carbon (PgC) (1). The pool contains a myriad of organic molecules, comprising broad fractions of variable lability that turn over on time scales from minutes to millennia (2, 3). Of the approximately 50 PgC y-1 of global net primary production, greater than 50% is partitioned as DOC to be processed through the microbial foodweb (3, 4). The majority of the newly produced DOC is rapidly remineralized by heterotrophic bacterioplankton within oceans surface layer (5). However, about 20% of the annual global ocean net community production (~1.8 PgC y-1) escapes degradation long enough to be exported from the euphotic zone through overturning circulation of the water column. Once exported, the dissolved organic matter (DOM) and its remineralization byproducts travel by isopycnal pathways into the oceans interior (1, 69). Figure 1 provides evidence of export with deep water formation in the North Atlantic, where mid-latitude, warm, DOC-enriched surface waters are transported with surface currents to high latitude (Fig. 1A), at which point deep water formation transports the DOC deep into the interior (Fig. 1, B and C) where it slowly declines during southward flow. Additional processes such as the solubilization of sinking particles (10) and DOC release from vertically migrating zooplankton (11) result in further DOC introduction into the subsurface layers. Factors that regulate persistence of exported DOC include inorganic nutrients required by heterotrophic microbes (12, 13), the quality of the exported DOM (14, 15), the microbial community structure (16, 17), and the expression of proteins involved in cell growth optimization (18). The mechanisms of production that result in DOC compounds that persist from months to millennia are not well described, but both abiotic and biotic processes have been shown to transform DOM to molecular forms that resist rapid microbial degradation. Abiotic processes include adsorption of labile DOC to colloidal material (19, 20), modification of chemical bond structure by ultraviolet light exposure (1921), and the formation of compounds like melanoidins via condensation reactions (22). Biotic sources of resistant DOM may include direct exudation from phytoplankton [for example, as acylheteropolysaccharides (14),

Dept. Ecology, Evolution, and Marine Biology, University of California, Santa Barbara, CA 93106-9610, USA 2 Rosenstiel School of Marine and Atmospheric Science, University of Miami, Miami, FL 33149, USA 3 Universit de la Mditerrane, LMGEM, Centre d'Ocanologie de Marseille, Case 901, 13288 Marseille Cedex 9, France 4 CNRS/INSU, UMR 6117, LMGEM, Case 901, 13288 Marseille Cedex 9, France *To whom correspondence should be addressed. E-mail: carlson@lifesci.ucsb.edu

Fig. 1. Meridional distributions of mean dissolved organic carbon (DOC) concentrations for (A) the euphotic zone (0 to 100 m), (B) the mesopelagic zone (100 to 1000 m), and (C) the bathypelagic zone (1000 to 3000 m) within the North Atlantic Basin. Surface waters elevated in DOC at low latitudes are transported to the north with surface circulation. The surface concentrations decrease with overturning circulation, thereby increasing concentrations at depth in the north. Over time, with flow of those deep waters to the south, the exported DOC concentrations decline. Mean DOC concentrations were determined by integrating DOC stocks within each depth horizon for each hydrostation and normalizing to the depth of each depth horizon. Filled circles, open triangles, and open squares are DOC values from the Climate Variability and Predictability (CLIVAR) program's transect lines A16, A22, and A20, respectively. Note scales of the y-axes change between panels. Adapted from (37).

57

Fig. 2. Distributions of DOC (mol C kg-1) at 30 m. Meridional and zonal lines of data are observed values, while the background field is modeled [see (1) for details]. Adapted from (1).

bacterially derived cell wall material (2326), liposome-like colloids produced via microzooplankton grazing (27), and the release of metabolites (28)]. The preferential removal of specific sugars and amino acids can also transform DOM to recalcitrance (2932). The production of refractory DOM compounds via heterotrophic microbial processes has been termed the microbial carbon pump (MCP) (33). In this conceptual model, a fraction of the bioavailable organic compounds processed by heterotrophic microbes is shunted to a biologically refractory form of organic matter. Experiments with marine microbes have shown that a portion of model biolabile organic compounds are transformed to products that persist for periods of months to years (34, 35). Other studies have identified bacterial biomarkers (D-enantiomer amino acids) in the high molecular weight fraction of oceanic DOM, lending support to the hypothesis that some portion of

the refractory DOM pool is of heterotrophic prokaryotic origin (24, 25). It is estimated that approximately 23% of the bulk oceanic DOC pool (~155 PgC) (36) is derived via the MCP. The accumulation of DOC is most pronounced in surface waters of the well-stratified subtropical gyres, reaching concentrations >80 mol C kg-1 (1, 37) (Fig. 2). This accumulation is unambiguous evidence for its resistance to decay. However, DOM that is persistent at one geographical location or depth horizon can be bioavailable at another. For example, of the 1.8 PgC of DOM exported from the euphotic zone, only ~0.2 PgC survives to depths greater than 500 m (1). Some mesopelagic bacterial lineages undergo productivity pulses during or shortly following water column overturn events, presumably due to the delivery of surface-derived DOM (17, 38, 39) (Fig. 3). The remineralization of exported DOC accounts for up to half the oxygen utilized in the

Fig. 3. Contours of plot of SAR11 subclade II cell densities (E8 L-1) in the surface 300 m from 2003 through 2005 at the Bermuda Atlantic Time-series Study site. The white dashed line represents, mixed layer depth and is used to examine distribution patterns in the context of mixing and stratification. The data reported in this figure demonstrate a response of mesopelagic subclade of the bacterioplankton SAR11

during or shortly following deep convective overturn. This response is presumed in part to result from the delivery of surface derived organic matter (i.e., DOC flux) into the mesopelagic during mixing. A portion of the DOC that persisted in the surface waters appears to become available to the organisms in the mespelagic zone triggering biomass production. Adapted from (39).

58

mesopelagic zone (7, 40, 41). At >1000 m, DOC removal occurs at much slower rates, accounting for < 20% of oxygen consumption (37, 42). The response of deep microbial lineages to exported DOC indicates that microbial processes in the oceans interior are capable of metabolizing some of these persistent polymeric compounds. Microbial diversity in oceanic systems is high, although there remains uncertainty with regard to how much of this diversity is functionally significant (16, 43, 44). Vertical partitioning of major prokaryotic groups, particularly between the euphotic and aphotic regions of the oceanic water column, is well documented (16, 17, 38, 39, 45, 46) and presumably linked to gradients in the character and quantity of DOM as well as to the availability of inorganic nutrients. Genomic and transcriptomic data demonstrate the potential for marine bacterioplankton to utilize a range of DOM targets (28, 4749). DeLong et al. (2006) identified a greater number of genes, putatively involved in polysaccharide degradation, in deep microbial populations compared to those found in the surface populations (16). The genes associated with the deep microbial assemblages are better suggestive of a surface-attached lifestyle capable of degrading recalcitrant pools of organic matter (16, 28). Specific groups of heterotrophic prokaryotes have been shown to respond differently to the quality and quantity of organic substrates (38, 4954). Thus, organic compounds resistant to microbial degradation at one depth horizon may serve as substrates for deeper population of heterotrophic microbes. For example, D-enantiomeric amino acids derived from the peptidoglycan of bacterioplankton cell walls have been observed to persist within the DOM pool (24, 25). However, there is a
References and Notes 1. D. A. Hansell, C. A. Carlson, D. J. Repeta, R. Shlitzer, Oceanogr. 22, 202 (2009). 2. D. Kirchman, Limnol. Oceanogr. 28, 858 (1983). 3. C. A. Carlson, in Biogeochemistry of Marine Dissolved Organic Matter, D. A. Hansell, C. A. Carlson, Eds. (Academic Press, San Diego, 2002), pp. 911151. 4. T. Nagata, in Microbial Ecology of the Oceans, D. L. Kirchman, Ed. (WileyLiss, New York, 2000), pp. 121152. 5. F. Azam, D. C. Smith, . Hagstrm, Microb. Ecol. 28, 167 (1993). 6. G. Copin-Montgut, B. Avril, Deep Sea Res. 40, 1963 (1993). 7. C. A. Carlson, H. W. Ducklow, A. F. Michaels, Nature 371, 405 (1994). 8. M. Canals et al., Nature 444, 354 (2006). 9. J. Martn, J. C. Miguel, A. Khripounoff, Geophys. Res. Lett. 37, 13604 (2010). 10. D. Smith, M. Simon, A. L. Alldredge, F. Azam, Nature 359, 139 (1992). 11. D.K. Steinberg et al., Deep Sea Res. I 47, 137 (2000). 12. J. B. Cotner, J. W. Ammerman, E. R. Peele, E. Bentzen, Aquat. Microb. Ecol. 13, 141 (1997). 13. T. F. Thingstad, A. Hagstrom, F. Rassoulzadegan, Limnol. Oceanogr. 42, 398 (1997). 14. L. I. Aluwihare, D. J. Repeta, R. F. Chen, Nature 387, 166 (1997). 15. R. H. Benner, in Biogeochemistry of Marine Dissolved Organic Matter, D. A. Hansell, C. A. Carlson, Eds. (Academic Press, San Diego, 2002), pp. 5990. 16. E. F. DeLong et al., Science 311, 496 (2006). 17. A. H. Treusch et al., ISME J. 3, 1148 (2009). 18. S. M. Sowell et al., ISME J. 3, 931105 (2008). 19. R. G. Keil, D. L. Kirchman, Mar. Chem. 45, 187 (1994). 20. T. Naganuma et al., Mar. Ecol. Prog. Ser. 135, 309 (1996). 21. R. Benner, B. Biddanda, Limnol. Oceanogr. 43, 1373 (1998). 22. L. C. Maillard, Compt. Rend. 154, 66 (1912). 23. E. Tanoue, S. Nishiyama, M. Kamo, A. Tsugita, Geochim. Cosmochim. Acta 59, 2643 (1995). 24. M. D. McCarthy, J. I. Hedges, R. Benner, Science 281, 231 (1998). 25. R. Benner, K. Kaiser, Limnol. Oceanogr. 48, 118 (2003). 26. K. Kaiser, R. Benner, Limnol. Oceanogr. 53, 1192 (2008). 27. T. Nagata, D. L. Kirchman, Arch. Hydrobiol. 35, 99 (1992). 28. E. B. Kujawinski, Annu. Rev. Mar. Sci. 3, 567 (2011). 29. G. L. Cowie, J. I. Hedges, Nature 369, 304 (1994). 30. R. M. W. Amon, R. Benner, Limnol. Oceanogr. 41, 41 (1996). 31. A. Skoog, R. Benner, Limnol. Oceanogr. 42, 1803 (1997). 32. S. J. Goldberg, C. A. Carlson, D. A. Hansell, N. B. Nelson, D. A. Siegel, Deep Sea Res. I 56, 672 (2009). 33. N. Jiao, et al., Nat. Rev. Microbiol. 8, 593 (2010). 34. H. Ogawa, Y. Amagai, I. Koike, K. Kaiser, R. Benner, Science 292, 917 (2001). 35. C. Lnborg, X. A. lvarez-Salgado, K. Davidson, A. Miller, Est. Coast. Shelf Sci. 82, 682 (2009).

significant increase in the uptake ratio of D-aspartic acid by microbes in the meso- and bathypelagic realms of the North Atlantic (55, 56), indicating that deep microbes more readily utilize D-amino acids. These deep-sea piezophilic microorganisms display metabolic capabilities that allow them to thrive under cold and high-pressure conditions (57). Some piezophiles are capable of degrading complex organic matter (58) by modifying their gene structure and protein regulation (59). For example, in Photobacterium profundum SS9 the metabolic pathways used to degrade polymers such as chitin, pullulan, and cellulose are controlled by pressure: upregulation of proteins occurs above 28 MPa and downregulation below 0.1 MPa. Cell-specific ectoenzymatic activities are greater in deep waters compared to the surface (6064), providing additional evidence that deep-sea microorganisms are adapted to degrade more refractory pools of organic matter (65), under in situ pressure and temperature (61, 63). The production of refractory DOM can be important in ocean carbon sequestration. First, DOM that persists in the surface ocean long enough for its eventual mixing to greater ocean depths represents a carbon-rich export term, contributing to 20% of the biological carbon pump (1, 6, 7, 66). Second, a fraction of this exported DOM can persist for centuries to millennia, thus the carbon remains stored in an organic form and unavailable for exchange with the atmosphere as CO2. The degree to which the MCP shunts carbon to long-term storage as organic matter, versus brief storage associated with export and rapid, deep remineralization, is unknown. Resolving these fates and identifying causative mechanisms and controls of the MCP remain important challenges.

36. R. Benner, G. Herndl, Molecular Carbon Pump in the Ocean, N. Jiao, F. Azam, S. Sanders, Eds. (Science/AAAS, Washington DC, 2011), pp. 4648. 37. C. A. Carlson et al., Deep Sea Res. II 57, 1433 (2010). 38. R. M. Morris et al., Limnol. Oceanogr. 50, 1687 (2005). 39. C. A. Carlson et al., ISME J. 3, 283 (2009). 40. M. D. Doval, D. A. Hansell, Mar. Chem. 68, 249 (2000). 41. D. A. Hansell, in Biogeochemistry of Marine Dissolved Organic Matter, D. A. Hansell, C. A. Carlson, Eds. (Academic Press, San Diego, 2002), pp. 685716. 42. J. Arstequi, S. Agusti, J. J. Middelburg, C. M. Duarte, in Respiration in Aquatic Systems, P. del Giorgio, P. J. L. Williams, Eds. (Oxford Press, Oxford, 2005), pp. 181205. 43. M. S. Rapp, S. J. Giovannoni, Annu. Rev. Microbiol. 57, 369 (2003). 44. J. C. Venter, et al., Science 304, 66 (2004). 45. S. J. Giovannoni, M. S. Rapp, K. Vergin, N. Adair, Proc. Natl. Acad. Sci. U.S.A. 93, 7979 (1996). 46. K. G. Field et al., Appl. Environ. Microbiol. 63, 63 (1997). 47. S. J. Giovannoni et al., Science 309, 1242 (2005). 48. R. S. Poretsky, S. Sun, X. Mou, M. A. Moran, Environ. Microbiol. 12, 616 (2010). 49. J. McCarren et al., Proc. Natl. Acad. Sci. U.S.A. 107, 16420 (2010). 50. M. T. Cottrell, D. L. Kirchman, Appl. Environ. Microbiol. 66, 1692 (2000). 51. C. A. Carlson et al., Limnol. Oceanogr. 49, 1073 (2004). 52. C. Arnosti, S. Durkin, W. H. Jeffrey, Aquat. Microb. Ecol. 38, 135 (2005). 53. R. R. Malmstrom, M. T. Cottrell, H. Elifantz, D. L. Kirchman, Appl. Environ. Microbiol. 71, 2979 (2005). 54. A. D. Steen, K. Ziervogel, C. Arnosti, Org. Geochem. 41, 1019 (2010). 55. M. T. Prez, C. Pausz, G. J. Herndl, Limnol. Oceanogr. 48, 755 (2003). 56. E. Teira, H. van Aken, C. Veth, G. J. Herndl, Limnol. Oceanogr. 51, 60 (2006). 57. C. Tamburini, J. Garcin, A. Bianchi, Aquat. Microb. Ecol. 32, 209 (2003). 58. A. Vezzi S. et al., Science 307, 1459 (2005). 59. F. Lauro, D. Bartlett, Extremophiles 12, 15 (2007). 60. H. G. Hoppe, S. Ullrich, Aquat. Microb. Ecol. 19, 139 (1999). 61. C. Tamburini, J. Garcin, M. Ragot, A. Bianchi, Deep Sea Res. II 49, 2109 (2002). 62. R. Zaccone et al., J. Geophys. Res. 108, 8117 (2003). 63. C. Tamburini et al., Deep Sea Res. II 56, 700 (2009). 64. F. Baltar, J. Arstegui, J. M. Gasol, G. J. Herndl, Aquat. Microb. Ecol. 60, 227 (2010). 65. H. G. Hoppe, C. Arnosti, G. J. Herndl, in Enzymes in the Environment: Activity, Ecology and Applications, R. G. Burns, R. P. Dick, Eds. (Marcel Dekker, New York, 2002), pp. 73108. 66. C. S. Hopkinson, J. J. Vallino, Nature 433, 142 (2005). 67. This work was supported by Scientific Committee on Oceanic Research (WG134) and NSF (OCE) 0801991, 0850857 to CAC; 0752972 to DAH and CAC; and ANR-POTES, ANR-05-BLAN-0161-01 to CT.

59

Molecular Characterization of Dissolved Organic Matter and Constraints for Prokaryotic Utilization
Gerhard Kattner1*, Meinhard Simon2, Boris P. Koch1,3 The chemical structures of marine dissolved organic matter (DOM) are largely unknown. However, ultrahigh resolution mass spectrometry recently allowed the identification of thousands of different molecular formulas and revealed that DOM is composed of a huge number of relatively small molecules (200 to 800 Da). Despite this success in dissecting marine DOM molecules, it remains unclear how biomolecules are transformed into refractory DOM and how the microbial carbon pump (MCP) contributes to this process. Besides a potential general resistance of DOM to prokaryotic attack, the low concentration of individual DOM constituents in the deep ocean may prevent energy-efficient microbial DOM consumption, even though bacteria have the genetic repertoire to exploit a high variety of molecular DOM structures. Further improvements in analytical and genomic technologies that will aid in the elucidation of DOM molecule structure, source, and degradation pathways should reveal the role refractory DOM and the MCP play in the global carbon cycle. ost of the dissolved organic matter (DOM) produced in sur- drolysis) and thus the chemical structures in which they were originally face waters, such as labile proteins, carbohydrates, and lipids, embedded remain unknown; (iii) Data are often obtained following is mineralized by microbial activity, and the remaining material is transformed into semilabile DOM and finally refractory DOM (1), a process in which the microbial carbon pump (MCP) plays an important role (2). The molecular composition of labile DOM is modified rapidly during decay, creating a material for which chemical structures are largely unknown and currently indeterminable. The decay process results in a large variety of compounds with completely new chemical structures, which exist in predominantly dissolved, but also particulate, forms. These transformations are mediated by microbial, viral, and photochemical activity, and proceed continuously over periods from days to months and beyond to decades and hundreds of years. Dissolved organic carbon (DOC) in the deep ocean therefore has an average age of 4,000 to 6,000 years (3). A puzzling and still unanswered question is Fig. 1. Different mass arrays of an ultrahigh resolution FT-ICR mass spectrum (electrospray ionizawhy this DOM resists biotic and abiotic decom- tion, negative mode) of a marine dissolved organic matter (DOM) sample that was extracted using position and why it is unavailable to heterotro- a solid-phase process [sorbent PPL (17)]. From the exact masses, molecular formulas can be calcuphic prokaryotes. Without knowing the molec- lated for thousands of peaks in each sample. ular structures it is impossible to successfully tackle this question. However, it is also important to examine whether isolation and fractionation of DOM using, for example, solid-phase exthe prokaryotes in the ocean, and in particular in the dark ocean where traction and ultrafiltration, but resulting fractions represent only a small the refractory DOM accumulates, have the genetic repertoire and appro- portion of total DOM (~25% for ultrafiltration and ~40% for solidpriate conditions to express the suitable genes encoding enzymes that phase extraction). can cleave and decompose the DOM molecular structures. Appropriate Successful molecular characterization of DOM using ultrahigh resmetagenomic and metatranscriptomic methods are now available to do olution mass spectrometry (Fourier transform ion cyclotron resonance this (4, 5, 6). In contrast, tools for the structural characterization of DOM mass spectrometry; FT-ICR MS) represents a significant advance in the in the ocean are limited by a number of factors: (i) Analyses thus far field, and progress has also been made applying collision induced dishave essentially been restricted to carbohydrates, amino acids, amino- sociation FT-ICR MS (8). To date several thousand molecular formulas sugars, and lipids (7), compounds that represent only 10 to 20% of DOM have been identified (Fig. 1) and a number of these may act as potential in the near-surface and even less in the deep ocean; (ii) Because only a markers of DOM sources and modification processes. The occurrence minor portion of these molecules exist as monomers, they are usually of pseudohomologous series (addition/removal of CH2, H2, O2, or H2O analyzed only after considerable chemical treatment (for example, hy- to/from a given molecular formula) shows that there are regular patterns within the strong complexity of the molecular formulas of DOM (Fig. 2). Approximately one third of the detectable formulas are present in all marine samples and most likely represent a common refractory 1 Alfred Wegener Institute for Polar and Marine Research, Ecological Chemistry, Am background in DOM (9). The Problem: Why Microbes Cannot Decompose Refractory Deep-sea DOM It has been previously argued that concentrations of individual DOM

Handelshafen 12, D-27570 Bremerhaven, Germany 2 Institute for Chemistry and Biology of the Marine Environment, University of Oldenburg, Carl-von-Ossietzky-Str. 9-11, D-26111 Oldenburg, Germany 3 University of Applied Sciences, An der Karlstadt 8, D-27568 Bremerhaven, Germany *To whom correspondence should be addressed. E-mail: gerhard.kattner@awi.de

60

decomposition of deep-sea DOM by raising the ambient DOM concentration fivefold did not yield any microbial DOM decomposition (14) presumably because concentrations of individual compounds were still below this threshold concentration. The intrinsic stability of chemical structures of deepsea DOM compounds may further explain their recalcitrance to microbial degradation. The low proportion of hydrogen (average hydrogen to carbon ratio is ~1.25; Fig. 2) reflects a substantial proportion of stable aromatic backbones, structures known to be difficult for prokaryotes to degrade. However, there is very little mechanistic evidence for a principal molecular resistance to bacterial enzymatic attack. Most molecules are highly oxygenated (oxygen to carbon ratio of ~0.45; Fig. 2), which implies that they are not stable, per se, and should be susceptible to utilization by prokaryotes. This high oxygen content, which primarily exists in carboxylic functions [refractory carboxyl-rich alicyclic molecules (15)], reflects a high degree of polarity and may therefore require a highly specific and energy-efficient uptake system. Recent studies of the genetic potential of deep-sea prokaryotes revealed that bacteria are capable of taking up and degrading a great variety of DOM compounds including 2,4-dichlorobenzoate, which requires the cleavage of the aromatic ring (4). These observations are in line with the assumption that the various DOM compounds are biodegradable by prokaryotes. Fig. 2. FT-ICR mass spectrum (electrospray ionization, negative mode) of a solid-phase exHowever, the energetic yield of degrading these highly tracted [sorbent PPL (17)] marine deep-sea dissolved organic matter (DOM) sample. Comoxygenated compounds might not be profitable enough to mon molecular formulas in marine DOM contain carbon, hydrogen, and oxygen. Each formugain metabolic energy for growth and may require complela in the plot is represented by its molecular element ratio (hydrogen/carbon versus oxygen/ mentary consumption of other compounds with a higher carbon) and its relative peak magnitude [color scale, from higher (violet, center) to lower energetic yield. (blue, edge) intensities]. Pseudohomologous series of sequential addition or loss of molecular To obtain more detailed information on the composifragments (along the arrows) point to regularities in the molecular composition of DOM. tion of DOM, it is necessary to fractionate it into smaller Average element ratios of lipids, carbohydrates, and peptides (similar ratios for proteins) do units with similar characteristics. Attempts to do so have not overlap with the element ratios of DOM molecules. already been made using polarity or size criteria, revealing several different mass formulas found exclusively in compounds might be far below the chemoreceptive threshold for pro- individual fractions (16). Further improvements in fractionating indikaryotes thus preventing the energy-efficient uptake of DOM (10). This vidual molecules using FT-ICR MS will likely be available in the near notion could explain the slow DOM degradation and its seemingly re- future and improvements in NMR techniques, for which increasingly calcitrant behavior. Prokaryotes can exploit substrates only above a cer- little material is necessary to obtain structural information, also seem tain concentration, and their uptake systems have been shown to drive probable. concentrations below this threshold (11). High-affinity uptake systems It may be some time before unequivocal chemical structures of have a substrate affinity constant (Ks) value in the nanomolar range DOM are identified. This knowledge is crucial to enable the further in(12, 13) but considerations based on FT-ICR MS analyses of DOM im- vestigation of the microbial- and photochemical-based degradation and ply that concentrations of individual DOM compounds in the deep sea transformation of DOM and to better understand the cycling of DOM are likely far below this range. An early attempt to induce prokaryotic in the global ocean.

. 1 2. 3. 4. 5. . 6 7. . 8 9.

10.

References and Notes H. Ogawa, Y. Amagai, I. Koike, K. Kaiser, R. Benner, Science 292, 917 (2001). N. Jiao, et al., Nat. Rev. Microbiol. 8, 593 (2010). P. M. Williams, E. R. M. Druffel, Nature 330, 246 (1987). E. F. DeLong, et al., Science 311, 496 (2006). X. Mou, S. Sun, R. A. Edwards, R. E. Hodson, M. A. Moran, Nature 451, 708 (2008). J. McCarren, et al., Proc. Natl. Acad. Sci. U.S.A. 107, 16420 (2010). R. Benner, in Biogeochemistry of Marine Dissolved Organic Matter, D. A. Hansell, C. A. Carlson, Eds. (Academic Press, San Diego, 2002), pp. 5990. M. Witt, J. Fuchser, B. P. Koch, Anal. Chem. 81, 2688 (2009). B. P. Koch, M. Witt, R. Engbrodt, T. Dittmar, G. Kattner, Geochim. Cosmochim. Acta 69, 3299 (2005). H. W. Jannasch, in Direct Ocean Disposal of Carbon Dioxide, N. Handa, T. Ohsumi, Eds. (Terra Scientific Publishing Company, Tokyo, 1995), pp. 111.

11. K. Komarova-Komar, T. Egli, Micobiol. Molecul. Biol. Rev. 62, 646 (1998). 12. J. W. Ammerman, F. Azam, Science 227, 1338 (1985). 13. W. Martens-Habbena, P. M. Berube, H. Urakawa, J. R. de la Torre, D. A. Stahl, Nature 461, 976 (2009). 14. R. T. Barber, Nature 220, 274 (1968). 15. N. Hertkorn, et al., Geochim. Cosmochim. Acta 70, 2990 (2006). 16. B. P. Koch, K.-U. Ludwichowski, G. Kattner, T. Dittmar, M. Witt, Mar. Chem. 111, 233 (2008). 17. T. Dittmar, B. Koch, N. Hertkorn, G. Kattner, Limnol. Oceanogr.: Methods 6, 230 (2008). 18. We are grateful to R. Benner, C. Stedmon and S. Sanders for constructive comments and suggestions. This work was supported by the Scientific Committee on Oceanic Research (WG134) and Deutsche Forschungsgemeinschaft (TRR 51, to MS).

61

Shedding Light on a Black Box: UV-Visible Spectroscopic Characterization of Marine Dissolved Organic Matter
Colin A. Stedmon1* and Xos Antn lvarez-Salgado2 The processes that drive the microbial carbon pump (MCP) in the ocean have yet to be fully characterized. Among the topics that are central to addressing this deficit are tracing the microbial production of refractory dissolved organic matter (DOM) by the MCP, and resolving how the chemical composition and structure of DOM changes as a result. A fraction of DOM absorbs ultraviolet (UV) and visible light, while a specific subset of this subsequently exhibits a natural fluorescence. These spectroscopic properties can be used as markers for the turnover of different DOM fractions in the ocean. Recent research has linked the UV-visible characteristics of DOM to its chemical structure, carbon content, and photochemical and microbial reactivity. In addition, widespread measurements of DOM optical properties in the ocean indicate that they can be used to trace the humification processes that lead to the production of part of the refractory DOM pool. These measurements are suited to being carried out using in situ or remote platforms in the ocean and can provide high-resolution temporal and spatial DOM data. In conjunction with other techniques, this will pave the way for new insights into the possible role of the MCP in the global carbon cycle. he light absorbing and fluorescent fractions of dissolved organic matter (DOM) are referred to as colored or chromophoric DOM (CDOM) and fluorescent DOM (FDOM) (Fig. 1A). For several decades, absorption and fluorescence properties of DOM have been used to follow the spatial and temporal distribution of the bulk DOM, tracing specific humic- and protein-like fluorophores (1) and characterizing the aromaticity and relative molecular weight of DOM through optical indices derived from their spectra (24). Current research in the field is a convergence of four main interests: (i) tracing water mass mixing in estuarine, coastal and continental shelf waters; (ii) quantifying underwater ultraviolet and visible light penetration; (iii) resolving remote sensing of ocean color; and (iv) studying the impact of microbial and photochemical processes on the chemical structure of DOM. The major advantage of using the UV and visible spectroscopic properties of DOM is that it requires very small sample volumes and minimal preparation before analysis. Moreover, the high sensitivity of the current spectrofluorometers allows estimation of the turnover of specific DOM subfractions during both short- and long-term incubation experiments designed to test the susceptibility of DOM to photochemical and/ or microbial degradation (3, 5, 6). The major pitfall of this approach is that the actual compounds and processes responsible for the measured absorption and fluorescence remain poorly characterized. Resolving this presents a considerable analytical and experimental research challenge, which can in part be addressed by coupling of absorption and fluorescence measurements to molecular level characterization using mass spectrometry. In spite of this limitation, CDOM and FDOM measurements are currently being used as markers of different DOM fractions and to shed light on DOM production, turnover, and fate in the ocean and how this might impact the role of the microbial carbon pump (MCP) in the ocean carbon cycle (7) (Fig. 1B). In surface waters, autotrophic and heterotrophic activity and photochemical degradation largely control the molecular characteristics and distribution of DOM (for now putting aside the effects of the mixing of waters of different origins). This is reflected in FDOM and CDOM profiles (Fig. 2) that show sharp changes in concentration and character depending on the balance of dominating processes (biological produc-

1 Department of Marine Ecology, National Environmental Research Institute, Aarhus University, Roskilde, Denmark 2 IIM-CSIC, Instituto de Investigaciones Marinas, Consejo Superior de Investigaciones Cientficas, Vigo, Spain *To whom correspondence should be addressed. E-mail: cst@dmu.dk

tion, and combined photochemical and microbial transformation and removal). At the ocean surface, ultraviolet light penetrates and acts to either directly decompose to colorless the organic compounds, CO, and CO2, or to alter the remaining DOM, influencing its chemical composition and availability to microbes (810). This is evident from ocean profiles of CDOM and FDOM (11, 12), but is often not as clear when using other bulk DOM measurements such as dissolved organic carbon (DOC). The effects of photodegradation diminish rapidly with increasing depth as the high-energy wavelengths are attenuated. Away from the immediate surface, photic and aphotic microbial processing dominate, with subsurface maxima in absorption and protein-like fluorescence of CDOM often found in association with the productive plankton biomass (Fig. 2). In the dark ocean, DOM is continually modified by microbes and the effect of the MCP on DOM optical characteristics becomes more apparent. Entrainment of DOM produced in the photic zone through seasonal vertical mixing (13) and the constant supply of organic matter from sinking particles (14) drives much of this activity, although the latter most likely dominates. The net result is an increase in the humic character of CDOM and FDOM, measured as a relative increase in longer wavelength absorption and humic-like fluorescence signals (Fig. 2). Although not entirely the same, this parallels the humification process that occurs in soils (Fig. 1B), but with alternate precursor material. The increases observed in CDOM and FDOM in the meso- and bathypelagic ocean are correlated to apparent oxygen utilization (AOU) and nutrient remineralization (11, 12, 1518). These trends reveal the MCP in action, representing the gradual accumulation of a persistent DOM fraction that is directly linked to heterotrophic activity, with the humic CDOM/FDOM essentially representing biologically refractory metabolites. Data supporting this trend have recently been collected in all the major oceans (12, 16, 18). High-resolution measurements using in situ instruments are revealing that the techniques are sensitive enough to indicate differences in the ratio of humic fluorescence production to AOU for individual intermediate and bottom water masses. Additionally, detailed spectral characterization has shown that this humic fluorescence signal is actually composed of two independent fractions, which accumulate at different rates in the deep ocean. Armed with expanding global datasets and with the dark ocean as an incubator, the research community is gradually resolving the UV-visible spectroscopic signature of refractory DOM produced by the MCP. The combination of long ocean-mixing time scales, the absence of photochemistry, and the limited input of labile DOM makes the dark ocean an ideal environment for further studies in this area.

62

Additional widespread sampling, experimentation, and methodological refinements linking UV-visible spectroscopic character of DOM to molecular level characterization (19) and carbon content (20) stand to pave the way for enhancing the utility of these optical measurements. In conjunction with the technological advances in opFig 1. (A) Colored and fluorescent DOM (CDOM and FDOM, respectively) together represent a subfraction of the total DOM pool. Part of DOM does not absorb UV and visible light and is largely characterized as monomeric and polymeric aliphatic compounds such as carbohydrates, lipids, and aliphatic amino acids. The presence of conjugated double bonds (polyenes) results in absorption of UV and visible light. The presence of aromatic rings often also results in fluorescence. The large arrow indicates increasing aromaticity, conjugation, and carbon to hydrogen (C/H) ratio. Examples of CDOM absorption spectra and excitation-emission matrices are shown over the CDOM and FDOM domains, respectively. Structure of tryptophan, a natural amino acid, and vanillin, a constituent of lignin, are shown as examples of fluorescent CDOM. The grey lines indicate the position of their respective fluorescence excitation-emission peaks. (B) Soils of the sea: The processing of organic matter (particulate and dissolved, colored and colorless) by the microbial carbon pump (MCP) (7) not only results in formation of refractory DOM (RDOM) but also an alteration of the remaining organic matter, often referred to as humification and in principle similar to what occurs in soils. During processing by the MCP, colorless or weakly colored bioavailable DOM (BDOM) and particulate organic material (POM) is transformed into colored refractory CDOM. The large arrow indicates that during this process both the carbon-specific absorption and apparent fluorescence quantum yield increase.

tical instrumentation, this will lead to the development of improved in situ instruments measuring fluorescence or absorption at specific wavelengths or spectral bands with high spatial and temporal resolution, and designed to be routinely deployed on moorings, gliders, and profiling instrumentation.

Fig 2. Generalized vertical distribution of the quantity and quality of DOM in the ocean. Idealized vertical profiles of dissolved organic carbon (DOC) concentration, fluorescence intensity of amino acids (FDOM AA-like), and humic organic matter (Humic-like), and the concentration (solid line) and character (dotted line) of colored dissolved organic matter (CDOM). CDOM character is depicted as the slope of the absorption spectrum in the UV-visible wave-length range.

1. 2. 3. 4. 5. 6.

7. 8. 9. 10. 11. 12. 13.

References and Notes P. G. Coble, Chem. Rev. 107, 402 (2007). J. L. Weishaar, et al., Environ. Sci. Technol., 37, 4702 (2003). J. Helms, et al., Limnol. Oceanogr. 53, 955 (2008). S. A. Green, N. V. Blough, Limnol. Oceanogr. 39, 1903 (1994). C. A. Stedmon, S.S. Markager, Limnol. Oceanogr. 50, 1415 (2005). M. Nieto-Cid, X. A. lvarez-Salgado, F. F. Perez, Limnol. Oceanogr. 51, 1391 (2006). N. Jiao, et al., Nat. Rev. Microbiol. 8, 593 (2010). R. J. Kieber, L. H. Hydro, P. J. Seaton, Limnol. Oceanogr. 42, 1454 (1997). M. A. Moran, R. G. Zepp, Limnol. Oceanogr. 42, 1307 (1997). R. Benner, B. Biddanda, Limnol. Oceanogr. 43, 1373 (1998). K. Hayase, N. Shinozuka, Mar. Chem. 48, 283 (1995). N. B. Nelson, D. A. Siegel, C. A. Carlson, C. M. Swan, Geophy. Res. Lett. 37, L03610 (2010). D. A. Hansell, C. A. Carlson, D. J. Repeta, R. Schlitzer, Oceanography 22, 52 (2009).

14. J. Arstegui, J. M. Gasol, C. M. Duarte, G. J. Herndl, Limnol. Oceanogr. 54, 1501 (2009). 15. R. F. Chen, J. L. Bada, Mar. Chem. 37, 191 (1992). 16. Y. Yamashita, E. Tanoue, Nat. Geo. doi:10.1038/ngeo279 (2008). 17. Y. Yamashita, A. Tsukasaki, T. Nishida, E. Tanoue, Mar. Chem. 106, 498 (2007). 18. L. W. Cooper, et al., J. Geophys. Res. 110, G02013 (2005). 19. G. Kattner, M. Simon, B. P. Koch, Molecular Carbon Pump in the Ocean (2011), pp. 6061. 20. C. G. Fischot, R. Benner, Geophys. Res. Lett. 38, L03610 (2011). 21. This work was supported by the Scientific Committee on Oceanic Research (WG134), the Danish Council for Strategic Research (10-093903), the Galician General Direction for R+D+i (07MMA002402PR), and the Spanish Ministry of Science and Innovation (MALASPINA, grant number CSD2008 00077). M. J. Paz helped with illustrations. Discussions with G. Kattner and R. Benner have contributed to improve the manuscript.

63

Application of Functional Gene Arrays (GeoChips) in Monitoring Carbon Cycling

Joy D. Van Nostrand and Jizhong Zhou* Functional gene arrays (FGAs) provide a way, through in-depth genomic analysis, to link environmental processes with microbial communities. They comprise probes for a wide range of genes involved in functional processes of interest (e.g., carbon degradation, C- and N-fixation, metal resistance) from many different microorganisms, both cultured and uncultured. The most comprehensive FGA reported to date is the GeoChip 3.0, which includes approximately 28,000 probes covering approximately 57,000 gene variants from 292 functional gene families involved in carbon, nitrogen, phosphorus and sulfur cycling, energy metabolism, antibiotic resistance, metal resistance, and organic contaminant degradation. This technology has been used successfully to examine the effects of environmental change on microbial communities. This article provides an overview of GeoChip development and utilization with an emphasis on studies examining various aspects of carbon cycling. NA microarrays can overcome many of the limitations of other molecular techniques such as the need for polymerase chain reaction (PCR) amplification steps, which restricts the number of genes that can be detected due to limitations in primer availability and coverage. Microarrays provide information on the functional abilities of microbial communities and can also provide phylogenetic information at a higher resolution than other 16S rRNA-based methods and with fewer sampling errors. GeoChip provides a species-strain level of resolution (1) while 16S rRNA gene-based sequencing only provides a genus-species level of resolution (Zhili He, personal communication). They are less time-consuming and often cheaper to use than other methods, and have the ability to provide more accurate comparisons since they can interrogate multiple samples using the same set of probes (2). One disadvantage, however, is the inability of arrays to interrogate novel sequences. GeoChip Functional Gene Arrays Functional gene arrays (FGAs) are composed of probes for microbial genes involved in functional processes, such as nitrification or carbon degradation (35). The most comprehensive FGAs are the GeoChip arrays, which contains probes for thousands of functional genes (4, 5). The first reported FGA used PCR amplicon probes and demonstrated that such arrays are an effective tool for genomic studies of microbial communities (3). Later versions use oligonucleotide probes (1, 6), which, although less sensitive, provide higher specificity since they are easily customized to afford a more targeted probe design (7). The GeoChip has gone through several development iterations, from the first proof-of-concept array to the current version 4.0, with the ultimate goal of providing a comprehensive probe set with high specificity and able to distinguish between highly homologous genes (4, 5). GeoChip 4.0 uses the Nimblegen array format, covering 410 functional gene families including probes for carbon, nitrogen, phosphorus, and sulfur cycling, metal resistance and reduction, virulence genes, stress response, and bacterial phage genes. GeoChip Development Probe design. Gene regions selected for targeting on an array should encode the catalytic or active sites of proteins key to the functions of interest. To find these, public DNA databases are searched for suitable genes using broad search terms. Seed sequences from only those genes that have had protein identity and function experimentally determined

are used to confirm sequence identity for all downloaded sequences using HMMER alignment (8). Probes (50-mer) are then designed using CommOligo software (9) based on experimentally determined criteria (sequence homology, continuous stretch length, and free energy) (10), and screened against the GenBank database to ensure specificity. Target preparation and hybridization. The use of high quality nucleic acid starting material to query microarrays is extremely important. Purified DNA should have absorbance ratios of A260:A280 > 1.8 and A260:A230 > 1.7. The A260:A230 ratio has the most influence on hybridization success (11). Whole community genome amplification (WCGA) can increase the amount of DNA available using a small quantity of initial DNA (1 to 100 ng) and provide a sensitive (10 fg detection limit) and representative amplification (12). To examine activity, RNA can be used. Stable isotope probing is another means to evaluate activity, by hybridizing 13C-DNA after incubation with a 13Clabeled carbon source, such as biphenyl (13). Nucleic acids are then labeled with cyanine dye, purified, dried, and resuspended in a hybridization buffer. GeoChips are hybridized at 42 to 50C and 40 to 50% formamide (4, 14, 15). Issues in Microarray Application Sensitivity. The current sensitivity for oligonucleotide arrays using environmental samples is approximately 50 to 100 ng (107 cells) (1, 6, 16) or ~5% of the microbial community (16), providing coverage of only the most dominant community members. Several strategies can be used to increase sensitivity, including (i) amplifying target DNA or RNA (12, 17), (ii) increasing probe length (7, 18) [although this also decreases specificity (19)], (iii) increasing probe concentration per spot (19), (iv) decreasing hybridization buffer volume (20), (v) mixing during hybridization (21), (vi) alternate labeling (7), and (vii) reducing ozone concentrations (22). Specificity. Specificity is especially important since so many environmental microorganisms remain uncharacterized. Strategies to increase specificity include (i) modifying probe design criteria (4, 18) or (ii) increasing hybridization stringency by elevating temperature or formamide concentration. Quantitation. There is a correlation between signal intensity and DNA concentration, indicating that microarrays can provide quantitative information. Oligonucleotide probes (50-mer) have been shown to maintain a linear relationship (r=0.980.99) over a concentration range of 8 to 1000 ng (1), while RNA has a linear response over the range of 50 to 100 ng (17). Application of the GeoChip GeoChips have been used to examine a variety of microbial communities at contaminated sites (1, 4, 6, 12, 15, 23, 24), in marine environments (25), and in soils (2).

Institute for Environmental Genomics and Dept. of Botany and Microbiology, University of Oklahoma, Norman, OK USA *To whom correspondence should be addressed. E-mail: jzhou@ou.edu

64

Insights into carbon cycling. The latest GeoChips (versions 3.0 and 4.0) include coverage of many genes that are useful for studying the microbial carbon pump (26). GeoChip version 3.0 contains probes for 41 genes involved in carbon cycling, including four CO2 fixation genes and 25 genes involved in carbon degradation (labile to refractory), methanogenesis, two for methane oxidation, acetogenesis, and another eight miscellaneous enzymes. GeoChip 4.0 has expanded gene coverage to include those for bacterial lysis by viruses, which may be involved in the release of dissolved organic carbon in the oceans (26). Genes for carbon fixation and degradation in algae and other microbial eukaryotes will be included in future versions. While not directly examining carbon pumps, several GeoChip studies have provided insights into carbon cycling in the marine environment. GeoChip analysis has indicated that carbon fixation, methane oxidation, and methanogenesis may be occurring in deep-sea basalts, activities that had not previously been associated with this environment (14). In a study examining deep sea hydrothermal vent microbial communities, GeoChips detected RuBisCO genes that perform an important initial step in carbon fixation from photosynthetic microorganisms, supporting the assertion that organisms able to grow by acquiring geothermal radiation may live in environments lacking light, such as this one (27). In addition, detection of a large number of genes involved in the Calvin Benson Bassham cycle suggested that autotrophic microorganisms within the hydrothermal vent communities may principally use

this cycle rather than the Krebs cycle, providing a greater insight into carbon cycling occurring in the deep ocean (27). Previous research focusing on the effects of elevated CO2 levels in soil communities may lend some insight into how climate change can affect the global carbon cycle, including carbon pumps in the ocean. In one study, results of GeoChip 3.0 analysis following a ten-year exposure of grassland to elevated CO2 indicated that there were both overall changes in the soil microbial communitys functional structure and composition (as evidenced by distinct clustering of ambient and elevated CO2 samples in detrended correspondence analysis) as well as a significant increase in carbon fixation (28). Summary GeoChip arrays have garnered a great deal of attention and have been demonstrated to be useful for studying microbial ecology, linking microbial communities with geochemical cycling, and addressing high-level ecological questions pertaining to global climate change and ecological theory. The GeoChip provides sensitive, specific, and potentially quantitative information regarding microbial communities. However, many technical, experimental, and data analysis challenges remain, including a need to improve the sensitivity and quantitative accuracy of the arrays and to develop bioinformatic tools and techniques to analyze, evaluate, and interpret the vast amounts of data being generated.

References and Notes 1. S. M. Tiquia, et al., Biotechniques. 36, 1 (2004). 2. J. Zhou, S. Kang, C. W. Schadt, and C. T. Garten, Jr., Proc. Natl. Acad. Sci. U.S.A. 105, 7768 (2008). 3. L. Wu, et al., Appl. Environ. Microbiol. 67, 5780 (2001). 4. Z. He, et al., ISME J. 1, 67 (2007). 5. Z. He, et al., ISME J. 4, 1167 (2010). 6. S. K. Rhee, et al., Appl. Environ. Microbiol. 70, 4303 (2004). 7. V. J. Denef, et al., Environ. Micrbiol. 5, 933 (2003). 8. S. R. Eddy, Bioinformatics. 14, 755 (1998). 9. X. Li, Z. He, J. Zhou, Nucleic Acids Res. 33, 6114 (2005). 10. Z. He, L.-Y. Wu, X.-Y. Li, M. W. Fields, J.-Z. Zhou, Appl. Environ. Microbiol. 71, 3753 (2005). 11. J. Ning, et al., Appl. Microbiol. Biotechnol. 82, 983 (2009). 12. L.Wu, X. Liu, C. W. Schadt, J. Zhou, Appl. Environ. Microbiol. 72, 4931 (2006). 13. M. B. Leigh, et al., ISME J. 1, 134 (2007). 14. O. U. Mason, et al., ISME J. 3, 231 (2009). 15. P. J. Waldron, et al., Environ. Sci. Technol. 43, 3529 (2009). 16. L. Bodrossy, et al., Environ. Microbiol. 5, 566 (2003). 17. H. Gao, et al., Appl. Environ. Microbiol. 73, 563 (2007).

18. Z. He, L. Wu, M. W. Fields, J. Zhou, Appl. Environ. Microbiol. 71, 5154 (2005). 19. A. Relgio, C. Schwager, A. Richter, W. Ansorge, J. Valcrcel, Nucleic Acids Res. 30, e51 (2002). 20. L. Wu, et al., Environ. Sci. Technol. 38, 6775 (2004). 21. C. J. Schaupp, G. Jiang, T. G. Myers, M. A. Wilson, BioTechniques. 38, 117 (2005). 22. W. Branham, et al., BMC Biotechnol. 7, 8 (2007). 23. J. Xiong, et al., Appl. Environ. Microbiol. (2010), doi:10.1128/AEM.00500-10. 24. T. C. Hazen, et al., Science. 330, 204 (2010). 25. N. E. Kimes, J. D. Van Nostrand, E. Weil, J. Zhou, P. J. Morris, Environ. Microbiol. 12, 541 (2010). 26. N. Jiao, et al., Nat. Rev. Microbiol. 8, 593 (2010). 27. F. Wang, et al., Proc. Natl. Acad. Sci. U.S.A. 106, 4840 (2009). 28. Z. He, et al., Ecol. Lett. 13, 564 (2010). 29. This work was supported by ENIGMA (DE-AC02-05CH11231), Environmental Remediation Science Program, Oklahoma Applied Research Support, Oklahoma Center for the Advancement of Science and Technology, Oklahoma Bioenergy Center, the State of Oklahoma (AR062-034).

65

Toward a Mechanistic Approach to Modeling Bacterial DOC Pathways: A Review

Marie Eichinger1, 2*, Jean-Christophe Poggiale1, 2, and Richard Sempr1, 2 Dissolved organic carbon (DOC) can be mineralized into CO2 through bacterial respiration, reenter the trophic chain if bacterial biomass is grazed, or be partly stored in a recalcitrant form for millennia in the deep ocean by bacterial production during DOC degradation. This review compares mathematical models used to represent these pathways in the context of the current knowledge of bacterial metabolism, and examines how they may be related to the biological carbon pump and the microbial carbon pump. In view of the ever-changing environments in which bacteria exist, the authors propose that models of DOC degradation should be based on data from experiments in which bacteria are subjected to external constraints, such as DOC availability, temperature, and ocean depth. Further, the need for combining experimental and mechanistic modeling approaches to advance our understanding of the factors that affect oceanic DOC cycling and organic carbon storage is discussed.

odeling dissolved organic carbon (DOC) degradation remains challenging due to factors including the complex nature of DOC, the variety of microorganisms using different metabolic pathways necessary for DOC breakdown, and the various oceanic physicochemical conditions that regulate bacterial metabolism. In this review, we identify those variables (for example, DOC concentration and bacterial biomass) relevant to accurately model bacterial-driven DOC pathways and forcing factors (forces external to the system, acting upon it, for example, temperature and depth). The Multi-G model represents DOC degradation using DOC concentration as a single variable. DOC concentration decay is assumed to be at a constant rate and is therefore represented by first order kinetics (Eq. 1, Table 1) (1, 2). Multi-G-based models do not include specific biological activity (Fig. 1A), even though bacteria are recognized as the main agents of DOC degradation (3). A more accurate representation of DOC dynamics requires the inclusion of more detailed information about bacterial metabolism (4). The Monod model is an example of the next generation of model, which includes bacterial biomass as a variable (Eq. 2, Table 1). This model is the most widely used in ecosystem studies, taking into account bacterial population growth as a result of DOC consumption (57). It assumes that assimilated DOC is instantaneously converted to bacterial biomass with constant bacterial growth efficiency (BGE), and that the complementary proportion (1-BGE) is used for bacterial respiration (Fig. 1B). Although the predicted results obtained when applying the Monod model agree well with experimental observations (8), certain parameters such as BGE vary widely as a function of DOC chemical characteristics (9) and environmental conditions [temperature (10), and depth (8)]. This model fails, however, when DOC availability changes suddenly, making it inappropriate for modeling natural ecosystems that are subject to frequent environmental variations (11). Variability in DOC and nutrient quality and availability may have some effects on the physiological status of bacterial communities (12). These effects on bacterial stoichiometry have been modeled using the Droop equations, where, for example, carbon and phosphorus (13), or carbon, nitrogen, and phosphorus (14), are limiting. According to this model, population growth depends on a pool of internal nutrients inside

1 Universit de la Mditerrane, LMGEM, Centre dOcanologie de Marseille, Case 901, 13288 Marseille Cedex 9, France 2 CNRS/INSU, UMR 6117, LMGEM, Case 901, 13288 Marseille Cedex 9, France *To whom correspondence should be addressed. E-mail: marie.eichinger@univmed.fr

the bacterial cells (Fig. 1C; Eq. 3, Table 1), such as an internal reserve of organic carbon that allows cells to survive periods of starvation. This can also be taken into account using the dynamic energy budget (DEB) model (15), where population growth resulting from DOC consumption is modeled using at least two bacterial variables, the reserve and the structure of the cell (Fig. 1D; Eq. 4, Table 1). The DEB model is also appropriate for handling variable resource environments and variable bacterial stoichiometry because it can be extended to deal with systems using several substrates (15) (Fig. 1E). Furthermore, DEB considers the concept of maintenance that reflects the fact that bacteria use intracellular DOC not only for growth but also for physiological activities that do not produce new biomass (osmotic regulation, maintenance of intracellular pH, macromolecule turnover) (16, 17). The importance of such considerations in changing environments has been demonstrated in experiments in which bacteria are exposed to episodic inputs of labile DOC (LDOC), where DOC utilization for maintenance is detectable during starvation (11). Maintenance cannot be neglected, since bacterial survival under starvation conditions is a fundamental aspect of bacterial existence and something bacteria experience often (17). In some studies of marine DOC cycling, maintenance modeling is restricted to the respiration process (RB in Eq. 3, Table 1) (13, 16, 18). These models may prove useful for estimating the contribution from bacteria to the biological carbon pump, but not to the microbial carbon pump (MCP), since the latter assesses the role of microbial processes in refractory DOC (RDOC) generation and in carbon storage in the ocean (19). By contrast, in the DEB model, maintenance activity incorporates not only respiration, but also a set of processes which can include RDOC production (Fig. 1D). This RDOC contributes to the total RDOC in the ocean. Since the DEB model accounts for this RDOC production explicitly, its use is relevant for connecting the biological carbon pump and the MCP. Although RDOC production by marine bacteria has been studied both under controlled experimental conditions (2022) and in situ (23), mechanisms of this production are still unresolved. Only a few models include bacterially derived DOC at either a microbial level (11, 20, 24) or at an ecosystem level (25). A study based on pure cultures indicated that bacterial RDOC production may occur as a stress response when LDOC availability is low; this can be represented using the DEB model where, in the case of starvation, maintenance costs are mainly paid not by bacterial reserve, but rather by bacterial structure, leading to the production of RDOC (20). In order to extrapolate how long-term changes (e.g. warming) influence DOC dynamics and organic carbon storage in marine waters,

66

Fig.1. Schematic representation of fluxes in Multi-G model (A), Monod model (B), Droop model with LDOC and dissolved phosphorus (DP) as limiting nutrients (C), DEB model with LDOC as the single limiting nutrient (D), and DEB model for two limiting nutrients, here LDOC and DP (E). Only the DEB model considers bacterial RDOC production, which can contribute to the total RDOC pool in the ocean. If these models are included in larger models accounting for primary production and export of organic carbon, then all models except the Multi-G can be indirectly linked to the biological carbon pump. BB, bacterial biomass; BGE, bacterial growth efficiency; RES, reserve; STR, structure; LDOC, labile DOC; RDOC, refractory DOC; DP, dissolved phosphorus; QC, cellular carbon content; QP, cellular phosphorus content; RESC, carbon reserve; RESP, phosphorus reserve; A, assimilation; G, growth; R, respiration; M, maintenance.

mathematical models describing DOC concentration should account for the effects of these changes on the rates of DOC production and consumption. Few modeling studies consider the temperature effect on bacterial metabolic rates; some authors have used an exponential function to predict the effects of temperature changes on the rate of population growth (7), whereas others have applied the Q10 model to DOC assimilation and maintenance respiration (24) or to the population growth rate (6). The Q10 function represents one of the easiest ways to incorporate temperature dependence into a mathematical model. It estimates the change in a particular variable that would result from a temperature increase of 10C. The Arrhenius function provides a more

sophisticated way to consider temperature variations in a model, incorporating a high and low tolerance range (15). This representation might be useful in view of the wide range of temperatures in which bacteria can survive. However, to the best of our knowledge, it has never been used to model DOC generation and degradation by bacteria. Although properties of bacterial metabolism like BGE (8) and bacterial production (26) are known to vary as a function of depth, the direct effects of continuously increasing pressure are poorly understood (27). Recent data suggest that bacterial production and extracellular hydrolytic enzyme activities are generally higher under in situ pressures than at atmospheric pressures, but the reasons for this remain unclear

67

Table 1. Equations of each model described in the review. t, time; LDOC, labile DOC; SDOC, semilabile DOC; RDOC, refractory DOC; BB, bacterial biomass, Bc, bacterial biomass in carbon, Vmax maximum LDOC uptake rate; K, half-saturation constant.

(28). In view of the number of bacteria present at depth, this issue needs to be more thoroughly investigated. Similarly, insufficient information on the effects of pH on bacteria-driven DOC degradation precludes its inclusion in current models. There is some evidence that lower pH increases bacterial production and degradation, but bacterial respiration and BGE have not yet been studied with respect to ocean acidification

(3). Changes in seawater chemistry due to ocean warming and acidification are expected to enhance microbial activity and channel a greater fraction of the fixed carbon into DOC (29, 30), thus potentially increasing the importance of the MCP in the oceanic carbon flow. The relevance of the MCP in carbon cycling and storage, and how these changes might affect it should also be investigated using appropriate models.

References and Notes 1. C. S. J. Hopkinson, J. J. Vallino, A. Nolin, Deep Sea Res. II 49, 4461 (2002). 2. Y. Yamanaka, E. Tajika, Global Biogeochem. Cycles 11, 599 (1997). 3. J. Liu, M. G. Weinbauer, C. Maier, M. Dai, J. P. Gattuso, Aquat. Microb. Ecol. 61, 291 (2010). 4. F. Talin, C. Tolla, C. Rabouille, J. C. Poggiale, Acta Biotheor. 51, 295 (2003). 5. T. R. Anderson, P. J. l. B. Williams, Global Biogeochem. Cycles 13, 337 (1999). 6. J. Bendtsen, C. Lundsgaard, M. Middelboe, D. Archer, Global Biogeoch. Cycles 16, Art. No. 1127 (2002). 7. R. C. Tian, A. F. Vzina, D. Deibel, R. Rivkin, Global Biogeochem. Cycles 17 (2003). 8. M. Eichinger, J. C. Poggiale, F. Van Wambeke, D. Lefvre, R. Sempr, Aquat. Microb. Ecol. 43, 139 (2006). 9. R. M. W. Amon, R. Benner, Limnol. Oceanogr. 41, 41 (1996). 10. R. B. Rivkin, L. Legendre, Science 291, 2398 (2001). 11. M. Eichinger et al., Biogeosciences 7, 1861 (2010). 12. C. A. Carlson et al., Aquat. Microb. Ecol. 30, 19 (2002). 13. E. K. Hall, C. Neuhauser, J. B. Cotner, ISME J. 2, 471 (2008). 14. T. F. Thingstad, Mar. Ecol. Prog. Ser. 35, 99 (1987). 15. S. A. L. M. Kooijman, Dynamic energy budget theory for metabolic organisation (Cambridge University Press, Cambridge, ed. 3, 2010). 16. R. Cajal-Medrano, H. Maske, Aquat. Microb. Ecol. 38, 125 (2005). 17. R. Y. Morita, Bacteria in oligotrophic environment: starvation - survival

lifestyle (Chapman & Hall, New York, 1997) 18. J. G. Baretta-Bekker, B. Riemann, J. W. Baretta, E. Koch Rasmussen, Mar. Ecol. Prog. Ser. 106, 187 (1994). 19. N. Jiao et al., Nat. Rev. Microbiol. 8, 593 (2010). 20. M. Eichinger et al., Aquat. Microb. Ecol. 56, 41 (2009). 21. D. F. Gruber, J. P. Simjouw, S. P. Seitzinger, G. L. Taghon, Appl. Environ. Microb. 72, 4184 (2006). 22. H. Ogawa, Y. Amagai, I. Koike, K. Kaiser, R. Benner, Science 292, 917 (2001). 23. K. Kaiser, R. Benner, Limnol. Oceanogr. 53, 99 (2008). 24. L. Polimene, J. I. Allen, M. Zavatarelli, Aquat. Microb. Ecol. 43, 127 (2006). 25. L. Polimene et al., J. Geophys. Res. Oceans 112, doi:10.1029/2006JC003529 (2007). 26. T. Reinthaler et al., Limnol. Oceanogr. 51, 1262 (2006). 27. C. Tamburini et al., Deep Sea Res. II 56, 1533 (2009). 28. T. Nagata et al., Deep Sea Res. II 57 1519 (2010). 29. J. Raven et al., Ocean acidification due to increasing atmospheric carbon dioxide. (The Royal Society, London, 12, 2005). 30. U. Riebesell et al., Nature 450, 545 (2007). 31. Acknowledgements: The authors are grateful to members of the SCOR Working Group on the "Microbial Carbon Pump in the Ocean" for helpful discussions. Special thanks to N. Jiao, F. Azam, R. Benner, and G. Herndl for fruitful comments on the manuscript. This work was supported by the French CNRS/INSU and Universit de la Mditerrane.

68

Get a Career Plan that Works.

An exceptional career requires insightful planning and management. Thats where Science Careers comes in. From job search to career enhancement, Science Careers has the tools and resources to help you achieve your goals. Get yourself on the right track today and get a real career plan that works. Visit ScienceCareers.org.

ScienceCareers.org