Biological Design Principles for Synthetic Biology

Christina Agapakis
April 18, 2011
Biological design principles for synthetic biology

A dissertation presented

by

Christina Maria Agapakis

to

The Division of Medical Sciences

in partial fulfillment of the requirements

for the degree of
Doctor of Philosophy
in the subject of
Genetics

Harvard University,
Cambridge, Massachusetts

May 2011

1
© 2011 Christina Agapakis
All rights reserved

ii
Advisor: Pamela Silver! ! ! ! ! ! Christina Agapakis

Biological design principles for synthetic biology

Abstract

The ability to rationally design biological systems holds tremendous promise for

applications in medicine, manufacturing, energy, and the environment. As biological

complexity and evolution can pose threats to the ease and stability of an engineering

approach, emerging principles of biological design have urged abstraction and

standardization of biological modules with defined functions. However, the power of

biology as a design substrate lies first and foremost in the rich diversity and complexity

of evolved biological systems. Instead of flattening and eliminating such diversity, can

we instead employ our ever-deepening understanding of processes that drive diversity

and evolutionary change as tools for synthetic biology design? This dissertation

explores several such design principles and platforms for synthetic biology--protein

domains that transfer high-energy electrons, cyanobacteria, plants, and cheese serve as

physical platforms, while gene recombination, cellular cooperation, and personalization

emerge as conceptual platforms. A more integrated and biological approach to synthetic

biology has potential to lead to robust designs with multiple future applications.

iii
Acknowledgements

I am lucky to have many people to acknowledge for the help, support, and
contributions that made this work possible. I want to first thank my advisor Pam Silver,
for letting me experiment with biological design, fashion design, filmmaking, and
cheesemaking, giving me the tools and the freedom and independence to take risks on
new projects that fascinated me. Many thanks to Silver lab members past and present
who have made the lab a joy to work in and a collaborative environment for great
science. I want to thank in particular Jeff   Way for his protein whispering skills and
smart advice, Patrick Boyle for being a great baymate and partner in science, business,
and filmmaking, Devin Burrill, Mara Inniss, Danny Ducat, Jake Wintermute, Karmella
Haynes, Qingqing Wang, Bruno Afonso, and Bill Senapedis for being great friends,
collaborators, and scientific advisors, Buz Barstow, Gerald Grandl, Dave Savage, and
Caleb Kennedy for their integral roles in the hydrogen project, and Henrike
Niederholtmeyer for letting me tag along when she started injecting cyanobacteria into
zebrafish embryos and teaching me long German words.
Outside of the lab are many other wonderful people who have been
tremendously helpful in innumerable ways: The Synthetic Aesthetics team, Sissel
Tolaas, Daisy Ginsberg, Pablo Schyfter, Drew Endy, and Jane Calvert for creating
something wonderful and giving me the opportunity to be part of it. The Harvard
iGEM 2010 students and advisors, Aaron Deardon, Jonathan DeWerd, Alex Gedeon,
Jackie Quinn, Morgan Paull, Anu Raman, Mark Thielmann, Lu Wang, Julia Winn, Alain
Viel, and Tamara Brenner for their passion and hard work and Kurt Schellenberg for
teaching us everything we know about plants. Kathy Buhl and the lab ops team for
keeping everything running smoothly, Kate Hodgins, Maria Bollinger and Steve
Obuchowski in the BBS office, Jodi Moore and the FACS facility, and Jennifer Waters
and Lara Petrak at the Nikon Imaging Facility for technical help whenever I needed it,
Ramil Noche and Sean Megason for helping us create the zebrafish project, Tami
Lieberman for her microscopy skills, and Dominique Frueh and Kyle Perry for advice on
starting and finishing projects.
I have to give many thanks as well to my dissertation advisory committee,
Christopher T. Walsh, Jack Szostak, and Michael Eck for their advice and support, and
my defense committee, Michael Starnbach, Jagesh Shah, Angela DePace, and Tony
Sinskey for their thoughtful questions and comments.
Last but not least I want to thank my friends and family; special thanks to my
wonderful friends Scott Jones, Ben Wolfe, and Erin Clark, who taught me that a
delicious meal and zest for life can solve almost any problem; my parents, John and Effie
Agapakis, for their constant and tireless love and support and gentle nagging, my sister
Marina, for being a great roommate and great friend, and my husband Nick Seaver, for
everything.

iv
Table of Contents

Chapter 1: Introduction
Designing Biology..............................................................................................................1
Knowing by Making.........................................................................................................  5
Engineering With Evolution.......................................................................................... 9
Engineering Electron Transfer.................................................................................... 13
Engineering Symbiosis................................................................................................... 16
Designing Biologically....................................................................................................  20
References...........................................................................................................................21

Chapter 2: Hydrogen Metabolism Circuits in E. coli

Background........................................................................................................................28
Pathway Design................................................................................................................ 32
Materials and Methods................................................................................................... 38
Results................................................................................................................................. 45

In vitro hydrogen production from heterologously expressed hydrogenases    45 41

In vivo construction and optimization of a synthetic hydrogen-producing circuit     47

Linking plant ferredoxin function to essential sulfur metabolism    51
Isolation of synthetic hydrogen metabolism through deletion of competing reactions   52
Insulation through mutation of the hydrogenase-ferredoxin interaction surface  57
Improvement of hydrogen production activity through direct protein fusion    59
The effect of scaffolding ferredoxin and hydrogenase    62
Latent function for the N-terminal ferredoxin-like domain of the hydrogenase 66 

Discussion..........................................................................................................................69
Conclusions........................................................................................................................76
References...........................................................................................................................77

Chapter 3: Towards a synthetic chloroplast

Background.......................................................................................................................83
Materials and Methods.................................................................................................. 86
Results................................................................................................................................ 91

Zebrafish embryos injected with S. elongatus hatch and thrive 91 1

Synechococcus expressing invasin and listeriolysin invade mammalian cells      93

Replication inside mammalian macrophages 94

v
 Discussion..........................................................................................................................98
References........................................................................................................................102

Chapter 4: Personalized Genetic Engineering of Plants

iGEM................................................................................................................................105
Standardization/Personalization............................................................................... 107
Materials and Methods................................................................................................ 111
Results and Discussion.................................................................................................114

Design of BioBrick compatible vectors for Arabidopsis transformation    114

Petal color and nutritional enhancement through pigment accumulation    115
Flavor inversion   122
Garden personalization through allergen knockdown   125

Conclusions.....................................................................................................................127
References........................................................................................................................127

Chapter 5: Human cultures and microbial ecosystems

Background....................................................................................................................133
Cheesemaking................................................................................................................134 
Microbial Communities...............................................................................................136
Synthesizing Mixed Cultures.....................................................................................137
Synthetic Aesthetics..................................................................................................... 140
Cheese smellomics.........................................................................................................146
Conclusions.....................................................................................................................153
References........................................................................................................................154

Chapter 6: Conclusions 158

Appendix A

Supplemental Tables and Figures.............................................................................168

vi
Appendix B

Agapakis CM and Silver PA. “Synthetic Biology: Exploring and

Exploiting Biological Modularity Through the Design of Novel
Biological Networks.” Molecular BioSystems, 2009, 5: 704-13.......................189

Appendix C

Agapakis CM, Ducat DC, Boyle PM, Wintermute EH, Way JC, and
Silver PA. “Insulation of a Synthetic Hydrogen Metabolism Circuit in
Bacteria.” Journal of Biological Engineering, 2010, 4:3...................................201

Appendix D

Agapakis CM and Silver PA. “Modular Electron Transfer Circuits for

Synthetic Biology: Insulation of an Engineered Biohydrogen Pathway.”
Bioengineered Bugs, 2010, 1:6.................................................................................217

Appendix E

Agapakis CM, Niederholtmeyer H, Noche RR, Lieberman TD,

Megason SG, Way JC, Silver PA. “Towards a Synthetic
Chloroplast.” PLoS ONE, 2011, 6(4):e18877.......................................................224

vii
List of Figures

1.1 Historical synthetic biology................................................................................... 1

1.2 Richard Feynman’s “final chalkboard”.................................................................7
1.3 The Synthetic Kingdom........................................................................................18

2.1 Electron transfer between protein bound iron sulfur clusters.....................30

2.2 Schematic of synthetic hydrogen metabolism pathways...............................34
2.3 Insulation strategies for synthetic electron transfer pathways....................35
2.4 Sequence of modified multiple cloning sites of Novagen Duet Vectors...42
2.5 Characterization of the synthetic hydrogen production pathway...............47
2.6 Complementation of the sulfite reductase deletion, cysI, in E. coli
with a synthetic pathway.......................................................................................53
2.7 Domain structure of deleted genes with ferredoxin homology..................55
2.8 Growth of the cysI deletion strain with serial deletion of
competing reactions................................................................................................56
2.9 Insulation of the synthetic hydrogenase pathway through
deletion of competing reactions..........................................................................56
2.10 Insulation of the hydrogenase pathway through ferredoxin
binding surface mutagenesis.............................................................................58
2.11 Increase in hydrogen production from the synthetic circuit
through direct hydrogenase-ferredoxin fusion..............................................61
2.12 Effect of artificial scaffolding configuration on hydrogen production
from the synthetic circuit...................................................................................65
2.13 Hydrogenase dependent and hydrogen independent growth of
the sulfite reductase strain..................................................................................68

3.1 Three paths to endosymbiosis used in this study.............................................85

3.2 Tracking intracellular S. elongatus through zebrafish development............91
3.3 Zebrafish embryos are immediately killed by E. coli.......................................92
3.4 Invasion of CHO cells by engineered S. elongatus........................................... 94
3.5 E. coli and S. elongatus differentially affect macrophages after
phagocytosis...............................................................................................................95
3.6 S. elongatus can grow inside the macrophage cytoplasm.................................97

4.1 Responses to our survey on the genetic engineering of plants..................110

4.2 Photograph of Arabidopsis seedling resistant to glufosinate selected
on agar plates...........................................................................................................115
4.3 Tulip breaking virus causes a striped flower petals.......................................116
4.4 Modification of A. thaliana carotenoid metabolism.......................................118

viii
4.5 Growth of engineered plants in the F2 generation.....................................120
4.6 Expression of commercially synthesized, plant codon optimized
miraculin and brazzein....................................................................................... 123
4.7 Incorporation of miraculin and brazzein into plants...................................124
4.8 Two methods for RNA knockdown in plants.................................................126

5.1 Different cheese types on display at Formaggio Kitchen............................135

5.2 Cheeses isolated for smell analysis....................................................................142
5.3 Cheese curds being mixed by an artisanal cheese maker.............................143
5.4 Milk incubated with swabs of bacterial cultures from the human body..144
5.5 Human cheese.........................................................................................................144
5.6 Comparative smell-omics of the four cheeses analyzed
with headspace technology.................................................................................151

S1 Sequence of codon optimized commercially synthesized genes.................168

S2 Sequence alignment of five hydrogenases.......................................................176
S3 Alignment of Chlamydomonas reinhardtii HydA1 and HydA2.....................178
S4 GC traces of volatile compounds collected from human cheese................179

ix
List of Tables

4.1 Table of modified pORE series vectors designed to be compatible with

BioBrick DNA assembly.......................................................................................114
5.1 Cheese, bacteria, and odors.................................................................................146
5.2 Details of bacteria isolated from our cheeses................................................ 150
S1 Table of primers used...........................................................................................172
S2 List of compounds identified by GC/MS headspace technology...............184

x
Chapter 1
/ Introduction

“Design is an agent of change that enables us to understand complex changes

and problems, and to turn them into something useful.”
-Global Agenda Council on Design

“Nature doesn’t have a design problem. People do.”

-William McDonough and Michael Braungart, Cradle to Cradle

xi
Designing Biology

In the past century amid amazing advances in

our understanding of living cells, an engineering

approach to biology was developing alongside the

experimental approach. Synthetic biology has

mimicked and copied what we are able to understand

of the behavior and structures of natural cells, from

Stéphane Leduc's re-creation of biological shapes

and forms out of inks and salt solutions in 19121

(figure 1.1) to the J. Craig Venter Institute's

chemical synthesis of a whole bacterial genome in

20102. Today synthetic biology, biological

engineering, and artificial life research describe a

diverse set of programs and practices aimed at
understanding the fundamental processes of living
things through recreating, recombining, and
Figure 1.1 Historical synthetic
biology. Stéphane Leduc’s ink drops
reflected many biological processes
that could be observed microscopically
at the time, including mitosis and
cytokinesis.

simulating simplified biological systems with useful functions.

Within this diverse set of research there also exists a diverse set of approaches

to dealing with biological complexity. One branch of synthetic biology aims to create

novel biological systems that do not occur naturally, either within the context of natural

1
cells or as entirely novel self-replicating “protocells.” Unnatural nucleotide base pairs3,

amino acids4, proteins5 and cell membranes6 can add new functions to cells and provide

clues to the processes that led to the abiotic origin of life. For researchers designing

novel protein folds and functions, the diversity of proteins that exist in nature can seem

remarkably limited. Of the 20100 possible 100 amino acid sequences, only a small

fraction are ever found in cells, and as more and more protein structures are determined,

the number of new protein families and folds being discovered has plateaued. This leads

researchers to ask, “Must the toolkit of life be so restricted?”5

This approach differs significantly from the large group of modern synthetic

biologists who aim to shape biology into a predictable engineering discipline, where

design principles govern rational engineering projects. In this strategy, design

principles such as abstraction allow designers and engineers to focus on one layer of

biological complexity at a time, while decoupling and standardization of biological

modules allows for facile recombination of new gene networks. These synthetic

biologists see the diversity of genes and proteins found in nature as unnecessarily

complex, seeking to control and recombine useful parts from nature in an

understandable and industrially scalable organism. Complexity must be stripped away,

biological functions standardized, diversity flattened out. As Tom Knight, a leader of

this approach has said, “an alternative to understanding complexity is to get rid of

it” (quoted in Calvert, 20107). This approach favors a streamlined approach to cell

2
engineering and the design and understanding of minimal cells, a “chassis” that is self-

replicating with the minimal possible number of proteins, inside of which synthetic

pathways can be built without fear of cross-talk or interference from natural

metabolism2,8.

How can biology be both too complex and not complex enough for synthetic

biology? The contradiction is deeper than semantics and the fact that while they may be

called the same thing, these two instances of synthetic biology represent different

research agendas functioning at different scales of biological hierarchy. Rather, to make

sense of this contradiction, we must look to the very origins of novelty and complexity

in biological systems, how the tremendous diversity of biological forms and functions

emerges through the action of a limited set of gene and protein sequences. Novel

biological forms and functions do not emerge through the evolution of entirely new

protein sequences, but through noise in gene expression9, latency in gene and protein

function10, new connections forged between closely related and redundant genes and

proteins11,12, and symbiotic relationships between communities of organisms, both

cooperating and competing13. This noise, latency, cross-talk, and cooperation, however,

is currently a hindrance to the predictable design of synthetic biological circuits in

isolated, pure cultures.

This conflict between the emergence of evolutionary novelty and the

construction of designed novelty is particularly apparent in artificial transcriptional

3
control networks, one of the most fertile and successful areas of synthetic biology

research. Rational recombination of well-studied transcriptional circuits such as the lac

operon, tet repressor, and cI system of lambda phage has led to the design of many

emergent behaviors in Escherichia coli, including switches14, oscillators15, and logic

gates16. While these simplified transcriptional networks function as predicted to a first

approximation, stochastic fluctuations in gene expression levels17 and cross talk between

orthogonal promoters and repressors18 can significantly affect system robustness and

the ability to incorporate multiple such devices into larger systems and new contexts19.

While synthetic feedback loops or band-pass filters can limit the effect of gene

expression noise, there are fundamental limits to how much biological noise can be

suppressed20, and noise plays many crucial roles in cell regulatory systems,

differentiation, and evolution9. Furthermore, transcriptional complexities such as non-

coding RNA regulation, antisense transcription, alternative start sites, and promoter

cross-talk can have significant effects on synthetic transcription networks. Even

minimal genomes are still not perfectly streamlined for their given niche or function.

Life in its simplest free-living state is robust to changes in the environment, with latent

pathways picking up slack when random mutation knocks out a gene or the

environment changes24. Mycoplasma have some of the smallest self-replicating genomes,

but these genomes also have significant redundancies, genes that can be knocked out

without killing the cell23. These redundancies are crucial for evolution, where redundant

4
genes can be recombined and mutated, undergoing selection towards new behaviors

while their counterparts remain to preserve the required function25.

Not only are there confounding redundancies, but the transcription profile of

these “minimal” genomes has turned out to be much more complex than initially

expected. A deep transcriptional analysis of the genome-reduced bacterium Mycoplasma

pneumoniae found large amounts of non-coding antisense transcription and alternative

transcription of operons in different conditions21. These complex transcriptional

profiles of organisms heralded by synthetic biologists as particularly useful minimal

chassis organisms show how difficult it is to decouple biological parts from their

interdependent, networked functions. These complexities are important not only for

adaptability and evolvability, but for basic functions as well; attempts to remove such

transcriptional complexities from bacteriophage, “refactoring” the T7 genome for

improved understandability and engineerability led to a viable, but less robust virus and

smaller plaques22.

Knowing by Making

Despite these complex challenges in even the simplest organisms, an engineering

approach to biology can help to improve our understanding of the underlying biological

networks—how molecules, genes, proteins, cells, and organisms interact and evolve to

create biological novelty and complexity. Not only can a constructed circuit be used to

5
interrogate basic principles underlying biological networks26, but the construction of a

rationally designed and functional synthetic biological system can be seen as verification

of the accuracy of the underlying model. Jacques Loeb, a pioneer of the engineering

approach to experimental biology in the early twentieth century wrote on the value of

such an engineering approach to examining the validity of a biological theory: "the

proof of the explicability of any single life phenomenon is furnished as soon at it is

successfully controlled unequivocally through physical or chemical means or is repeated

in all details with nonliving materials" (quoted in Deichmann, 200927).

Many synthetic biologists today quote Richard Feynman with a similar sentiment

as an inspiration for their engineering approach to biology: "What I cannot create, I do

not understand" (figure 1.2). This quotation is so powerfully associated with synthetic

biology that it was even (incorrectly) quoted by encoding it into the first chemically

synthesized genome2. But Feynman was a theoretical physicist, not an engineer. His

approach to quantum physics (about which he also famously said “I think I can safely say

that nobody understands quantum mechanics”28) was not about constructing objects,

but of creating mathematical solutions to problems.

This construction of mathematical solutions, however, was not always based on

the rigor and elegance of the underlying axioms, but on cobbling together and

subsequent refining of pieces that worked29. Such constructions are common to physics,

engineering, and biology, both designed and evolved, where systems that simply function,

6
 Figure 1.2 Richard Feynman’s “final chalkboard,” where his famous inspiration to synthetic biologists
was written—”What I cannot create, I do not understand.” Courtesy of the Archives, California
Institute of Technology.

first and foremost, are combined and improved upon gradually, in cycles of designing,

building, analyzing, and editing.

Indeed, if synthetic biologists attempt to simplify and abstract biological

principles in order to engineer living cells, it seems that they also simplify the very

engineering principles themselves. In his essay “Design in synthetic biology,” Adrian

Mackenzie discusses how the stated engineering principles of synthetic biology limit

not just biology, but design itself 30:

Many existing definitions and characterisations of synthetic biology take

at face value the claim that design can be done and is done according to
principles. The fact that design is also an aggregate of practices,
spanning a highly variable set of techniques, processes, materials,

7
 abstractions and sensations, and often standing on shifting economic and
organisational ground is barely recognised....Invocations of engineering
design principles such as ‘abstraction’ or ‘decoupling’ serve as abstract,
decoupled names for the more intensive processes that reorganise how,
where and by whom things are made.

Knowing and making are critically intertwined in the process of evolving

technological design principles. Rather than focus on a fixed, limited, and limiting

understanding of biological design principles, iterative cycles of design, building, and

testing of synthetic biology modules and devices through smart, evolution-based trial

and error can help to evolve new principles for design. As our understanding of the

complex interconnectedness of biological systems improves through the analytic

aspects of the design cycle, synthetic approaches that can incorporate these processes

will likewise become more robust.

Perhaps the improvement of computational models and assembly processes for

engineering biology will allow for this iterative engineering cycle to one day be replaced

by linear design-based engineering that can be optimized and refined computationally

before being constructed31. Such a transition may reflect the similar transitions seen in

other engineering disciplines, such as aerospace engineering, where early trial and error

refined models of air dynamics that today allow for fully computational design of

machines with thousands of parts32. Evolved systems, however, have crucial differences

from designed systems that will affect such a transition. Fully rational design may not

8
be an appropriate goal, given the powerful force of evolution to refine engineered

systems and as a creative force in emergent biological novelty.

Engineering With Evolution

! As with biological complexity and the notion of design principles, evolution plays

an uneasy and complicated role in synthetic biology. Once designed, built, and tested,

further evolution of a synthetic biological pathway must be prevented, before

engineered cells mutate away from their designed function. Evolution itself, with its

directionless and irrational changes is often at odds with the rational approach to cell

engineering, with noted synthetic biologist Drew Endy even asking “What if we could

liberate ourselves from the tyranny of evolution...?”33. Evolution, however, is hardly a

tyrannical force in the design and construction of synthetic biological circuits, but

rather has played a crucial role in the selection and refinement of numerous engineered

biological parts and pathways34.

Evolutionary methods for creating diversity and selecting for function in

synthetic pathways can take many forms. In protein and metabolic engineering, there is

a long history of directed evolution to optimize enzyme35 and pathway36 function.

Screening of a panel of enzyme homologs from a diverse set of organisms can identify

an optimal starting point for further directed evolution in a given chassis organism, and

can be used to develop detailed models of structure/function relationships. This

9
synthetic metagenomics approach has been applied to optimize the P450

monooxygenase enzyme used in a pathway producing the antimalarial drug precursor

artemisinic acid in yeast37, and to identify optimal methyl halide transferase activity in

E. coli38.

Such evolutionary approaches can refine enzyme and pathway function, but also

refine our understanding of the details of enzyme and pathway function in a given

context. However, while refinement through selection of random point mutations and

selection has an important role in evolutionary history, it is not enough to explain the

origin of entirely novel biological structures and functions. Mutation leads to heritable

and selectable variation between individuals of a species, but accumulated mutation

alone is not thought to be enough to create a new species13. Rather, a recent

understanding of the processes that drive large-scale evolutionary changes have refined

neo-Darwinian thinking, and these evolutionary processes can also play a role in

developing novel synthetic biological systems with novel behaviors.

Stephen Jay Gould describes such processes that drive evolutionary creativity in

terms of  the upcycling and repurposing of consumer products for functions other than

what they were originally designed for. Just as used car tires can be recycled into sturdy

sandals in poor communities where resources for new products are limited, small

modifications or simply new connections between   existing biological parts can forge

new structures and rewire networks for new function. Starting with a limited set of

10
parts, new functions can emerge through new connections and latent functions. In

Gould’s essay “Creating the Creators,” he writes: "Organisms have no equivalent to

currency for acquiring something truly new; they can reconstruct only from their own

innards."10

Redundancy and gene duplication provide the genetic currency within organisms

that synthetic biologists can take advantage of in the construction and rewiring of

genetic, signaling, and metabolic pathways. However, the gene recycling pool is in fact

much larger than the genome of a single organism. In bacteria, individual genes,

plasmids, and chunks of chromosomes are able to cross species boundaries through

horizontal gene transfer, confusing taxonomic classification and species definitions

altogether while creating biological diversity and complexity. Transfer of whole genes

or operons into a new context can increase the fitness of the recipient in a changing

environment, for example through the transfer of antibiotic resistance genes, while the

transferred genes themselves can undergo recombination with the host genome,

mutation, and selection towards refined behavior within their new contexts as well as

new functions.*1

The ability to harness the processes of transformation and lateral gene transfer

in the 1970's created the field of genetic engineering, where genes from one organism

1* Here too we see that noise complicates the evolutionary story—noise in the expression of the master
regulator for competence in Bacillus subtilis, ComK, is required for stochastic switching between the non-
competent and competent state, where individual bacteria can pick up and incorporate DNA from the
environment39.

11
could be expressed in another for close study or biotechnological applications. Combined

with directed evolution methods to refine enzymes for their new context, genetic

engineering can create novel functions and behaviors in industrially relevant species of

microorganisms. Synthetic biology attempts to differentiate itself   from this "classical"

genetic engineering by focusing on rationally designing and transferring networks and

groups of interconnected genes and proteins that have emergent function. However,

synthetic biological networks made of individual parts with single, clearly defined

functions is simply an arithmetic expansion of genetic engineering. Understanding the

complexities of context-dependent, multi-faceted biological functions and interactions,

and being able to truly harness large-scale processes of evolutionary change holds

promise for a second, truly new wave of synthetic biology19.

Whole genome sequencing of closely related but anatomically quite different

species, such as humans and chimpanzees, has made it clear that observed species

differences cannot be explained by the relatively small differences in the sequence of

protein coding genes. Rather, changes in the gene expression levels, alternative

transcription and translation, and different network connections all play crucial roles in

such evolutionary changes11. In synthetic biology, optimization of pathway function

likewise often requires adjustment in enzyme expression levels and network

connectivity in addition to enzyme sequence. In an engineered lycopene overproduction

pathway in E. coli, automated and iterative mutation of enzyme transcription levels

12
coupled with screening for increased output can fine-tune synthetic metabolic pathways

without the need for dynamic models of enzyme kinetics or well characterized,

standardized transcriptional parts40.

Furthermore, artificial evolution can be used in the context of synthetic biology

to understand and rewire transcriptional networks. Combinatorial libraries of

transcription networks can be constructed through recombination of a small set of

interacting promoters and repressors, and screened for novel logic functions41. Such

methods could be used to rapidly develop models for how network behaviors emerge

from pathway connections, while applying selective pressure applied to a given non-

functional synthetic circuit can rapidly lead to improved connectivity and function42.

Engineering Electron Transfer

The redundancy, latency, and gene duplication that confound efforts at

streamlining biological systems create the pool of genetic material from which

organisms can successfully draw on for this kind of recycling-based evolution. An

understanding of how such processes have driven the evolution of protein-protein

interactions in eukaryotic cell signaling has allowed synthetic biologists to “rewire”

yeast signaling protein modules, dramatically altering cellular behavior12. Signaling

cascades translate chemical information from outside of the cell to a coordinated

cellular behavior through the action of highly modular protein domains that

13
phosphorylate, dephosphorylate, translocate, bind, and scaffold together signaling

proteins in complex, interconnected, and highly regulated networks43. Considering the

diversity of pathways and behaviors, the types of modules that signaling proteins are

made up from are remarkably limited, with dozens of proteins containing only a handful

of interaction domains such as the Src Homology 3 (SH3), GTPase binding domain

(GBD), or PSD95/DlgA/Zo-1 (PDZ) domains. Specificity for proper interacting

partners can emerge through negative selection of the binding surfaces through

natural44 and artificial selection45. Pathways can also be rewired through physical fusion

of different domains46,47 or rearrangement of scaffolding proteins that isolate native

pathways48. Scaffold proteins can also be deployed orthogonally, isolating metabolic

pathways in organisms that don’t naturally posses such signaling domains49.

While metabolism is typically not considered to be modular, complex metabolic

electron transfer pathways likely evolved through similar processes of gene duplication,

recombination, and mutation of iron-sulfur containing proteins, with the interaction

between two electron carrying proteins or domains largely controlled by electrostatic

forces50. Taking advantage of the modular nature of iron sulfur protein interactions,

several groups have recombined redox proteins in vitro in order to engineer novel

electron transfer pathways51. Much of the focus in previous work has been to create

physical interfaces between electrodes and enzymes52, joining electron transfer proteins

with electron-generating proteins through “molecular Lego” in vitro53.!

14
 A key electron transfer protein that has played a crucial role in the evolution and

synthetic construction of such pathways is ferredoxin. Ferredoxins are a family of small

(~11kD) soluble proteins that coordinate at least one iron sulfur cluster through a

conserved cysteine motif. They have been proposed to have been among one of the

earliest proteins in evolution due to their simple structures, basic functions, and limited

amino acid content54. The simplest ferredoxins are found in anaerobic bacteria and

coordinate two [4Fe-4S] clusters. These short bacterial proteins (~55aa) with two

cysteine-based iron-sulfur cluster binding motifs likely resulted from the duplication of

a still smaller ancestral gene55. Ferredoxins from higher plants, typically coordinating

only one [2Fe-2S] cluster, likely branched off in evolution from the bacterial

ferredoxins through a second duplication event56. Such gene duplication, fusion,

horizontal gene transfer, and drift has led to the existence of ferredoxin-like domains in

many other iron-sulfur cluster containing oxidoreductases. Indeed, the X-ray structure

of other important electron transfer proteins such as the [FeFe]-hydrogenase from

Clostridium pasteurianum shows that the hydrogenase electron transfer domain is made

up of three smaller iron-sulfur cluster binding domains, one with homology to bacterial

ferredoxins, one to plant-type ferredoxins, and one unique linker domain57. As the

common currency of electron transfer proteins, ferredoxins are excellent parts for the

design of novel electron transfer pathways and novel modular electron transfer

proteins.

15
 There are many appealing applications for such engineered electron transfer

systems!in vitro, such as miniaturized biofuel cells, biocatalysts, and biosensors51. These

approaches, however, do not take advantage of the self-assembly and self-regenerating

abilities of live cells. Engineered cellular pathways in vivo have the potential to impact

our understanding of cellular electron transfer systems in live cells and may provide

self-renewing platforms for the continuous production of fuels and other useful

molecules53.

In Chapter 2, I describe the design, construction, analysis, insulation, and

optimization of in vivo electron transfer pathways centered on the interaction between

ferredoxin and the ferredoxin dependent [FeFe]-hydrognease. These hydrogen

metabolism circuits can be engineered to produce hydrogen as a potentially renewable

source of energy or consume hydrogen as a source of metabolic reducing equivalents.

Such synthetic metabolic networks must be both integrated into cellular metabolism

and insulated from competing electron transfer reactions in order to ensure proper

function. Post-hoc testing of insulation strategies and directed evolution provide clues

to how electron transfer pathways evolve and function, and uncover latent function in

the hydrogenase enzyme.

16
Engineering symbiosis

Not only do bacteria live in complex communities where genetic material can be

exchanged, driving the creation of evolutionary novelties, but the microbial

communities themselves, with their complex interactions between different species,  can

have drastic effects on evolution. While the evolution of cooperation and altruism are

often seen as paradoxical events in the course of natural selection, where competition

between individuals seeking to pass their genes to the next generation is seen as the

primary force of evolution, endosymbiosis has been recognized as a powerful force of

evolutionary change, driving gene exchange between hosts and symbionts58, the

development of communities suitable to new ecological niches59, and even the origin of

the eukaryotic kingdom60. Modern endosymbiotic relationships between bacteria and

eukaryotic organisms reflect a remarkable diversity in how widely disparate species can

interact in positive ways, from nutritional symbiosis between Buchnera and aphids61,

nitrogen fixation by Rhizobia in plant root nodules, to photosynthetic symbiosis between

algal chloroplasts and sea slugs62.

Symbiotic relationships are widespread in nature and have recently begun to be

exploited in synthetic biological networks of increasing complexity63. Multi-component

synthetic-ecological systems have been developed for hydrogen production through

metabolic engineering64 and for the production of other useful metabolites38.

Communication between cells has also been engineered for multiple applications,

17
including pattern formation65 and oscillators66. Engineered communities have also been

useful as a generalized model of cooperation and competition in microbial populations67

and two-species metabolic modeling has been used in the identification of cooperating

variants of E. coli68. If cooperative assemblages of bacteria and endosymbiosis drove

the evolution of the eukaryotic kingdom, what kind of cooperation will drive the

creation of the design driven synthetic kingdom (figure 1.3)?

Figure 1.3 The Synthetic Kingdom, by Alexandra Daisy Ginsberg. The eukaryotic kingdom arose
through endosymbiotic relationships; what will it take for synthetic biology to create entirely new
species, a new kingdom of life characterized by the synthetic networks and devices that make it up?
©Alexandra Daisy Ginsberg, 2009. Reproduced with permission from the artist.

18
 There is a fine line between the pathological and beneficial in natural

endosymbiotic events. Many endosymbiotic relationships that exist in nature are

hypothesized to have begun through the acquisition of resistance to predation—

bacterial resistance to lysosomal digestion by amoeba after phagocytosis or eukaryotic

resistance to bacterial infection after intracellular invasion69. Replicating these events in

the laboratory may lead to a partial endosymbiosis. Symbiosis is generally thought to

refer to a mutualistic relationship where both partners benefit, but the term can be

construed rather broadly; Lynn Margulis paraphrases de Bary’s original 1879 definition

of symbiosis as simply the “protracted physical associations among organisms of

different species, without respect to outcome”70.

In Chapter 3 I explore three paths for entry of photosynthetic bacteria into

animal cells that would satisfy this broad definition of symbiosis--direct microinjection

into zebrafish embryos to explore the in vivo dynamics in a whole animal mode,

engineering with invasin from Yersinia pseudotuburculosis (inv) and listeriolysin O from

Listeria monocytogenes (llo) to allow invasion of mammalian endothelial cells, and

endocytosis of inv and llo engineered strains by macrophages.

Nearly eighty years ago, photosynthetic algae were explored as symbionts for

cells grown in tissue culture, as a method for renewing and replenishing growth media

with oxygen and nutrients while removing waste products and carbon dioxide71. More

recently, photosynthetic symbiosis in tissue culture was explored as a method for

19
understanding the nutritional requirements of host and symbiont72. We sought to

explore the behavior of the photosynthetic bacteria Synechococcus elongatus inside

eukaryotic cells as a platform for engineered photosynthetic endosymbiosis and found

that cyanobacteria have little apparent effect on their host cells and can divide in the

macrophage cytoplasm. Further engineering of metabolite production and secretion73

in such endosymbiotic strains has the potential to lead to true mutualistic relationships

between photosynthetic bacteria and mammalian cells, essentially creating artificial,

engineerable, animal chloroplasts.

Designing Biologically

As designers and engineers we often seek to control natural processes, to

rationalize the “imperfect abundance” of nature74. But in biology and in design,

diversity can be a strength, with complex and irrational assemblages of species, things,

and processes maintaining ecosystems and economies in complex and changing

environments. In chapters 4 and 5 I discuss two projects that explore the diversity of

microbial ecosystems and the potential for personalization in synthetic biology and

genetic engineering design.

Chapter 4 focuses on the concept of personalized genetic engineering of plants, a

project done in collaboration with the Harvard University undergraduate team

participating in the International Genetically Engineered Machines competition

20
(iGEM). In industrial agriculture and, more recently, in biotechnology, monocultures of

single species engineered in a “one size fits all” framework are becoming the norm.

While it is increasingly recognized that a diverse ecosystem is a strong and resilient

ecosystem, monocultures are becoming pervasive as high-yield genetically engineered

crops and seed patents control the market. Can new tools and concepts from synthetic

biology make it easier to rapidly design, test, and improve new engineered plants

personalized for individual gardens and specific cultural and environmental ecosystems?

We explored several engineered personalizations in Arabidopsis thaliana: deletion of

common allergen genes, metabolic engineering of pigments for new colors and

increased nutritional value, and altered flavors.

Diverse ecosystems also play an key role in chapter 5, which describes a project

focused on the microbial ecology of cheese and the human body, done during my

Synthetic Aesthetics residency in collaboration with Sissel Tolaas. Art and design,

beyond superficially affecting the surface of engineering projects, can have a crucial role

in the engineering process, asking new questions, establishing far-reaching

collaborations, and creating something new in biology’s own terms. Collaborative,

interdisciplinary projects that embrace the diversity and complexity of biology can not

only question but improve on the simplifying, purifying, and rationalizing force of

synthetic biology, towards a biological design that can maintain difference and diversity

and query the basic biology of living ecosystems. When we design biologically, we use

21
 the tools and the strengths found in the principles of biology, where complexity and

evolution go hand in hand in both designed and evolved systems.

References

1! Fox Keller, Evelyn, Making sense of life: explaining biological development

with Models, Metaphors, and Machines (Cambridge, Massachusetts.
Harvard University Press, 2002).
2! Gibson, Daniel G et al., Creation of a Bacterial Cell Controlled by a
Chemically Synthesized Genome. Science (2010).
3! Benner, Steven A, Hutter, Daniel, and Sismour, A Michael, Synthetic
biology with artificially expanded genetic information systems. From
personalized medicine to extraterrestrial life. Nucleic Acids Res Suppl (3),
125 (2003).
4! Voloshchuk, Natalya and Montclare, Jin Kim, Incorporation of unnatural
amino acids for synthetic biology. Molecular bioSystems 6 (1), 65 (2010).
5! Fisher, Michael A et al., De Novo Designed Proteins from a Library of
Artificial Sequences Function in Escherichia Coli and Enable Cell Growth.
PLoS ONE 6 (1), e15364 (2011).
6! Zhu, TF and Szostak, JW, Coupled growth and division of model protocell
membranes. J Am Chem Soc 131 (15), 5705 (2009).
7! Calvert, J, Synthetic biology: constructing nature? Sociological Review 58
(s1), 95 (2010).
8! Jewett, Michael C and Forster, Anthony C, Update on designing and
building minimal cells. Current Opinion in Biotechnology, 1 (2010).
9! Eldar, Avigdor and Elowitz, Michael B, Functional roles for noise in genetic
circuits. Nature 467 (7312), 167 (2010).
10! Gould, S.J, Creating the creators. Discover 17 (10), 42 (1996).
11! Carroll, Sean B, Evolution at Two Levels: On Genes and Form. PLoS Biol 3
(7), e245 (2005).
12" Bhattacharyya, Roby P, Reményi, Attila, Yeh, Brian J, and Lim, Wendell A,
Domains, Motifs, and Scaffolds: The Role of Modular Interactions in the
Evolution and Wiring of Cell Signaling Circuits. Annu Rev Biochem 75 (1),
655 (2006).
13! Margulis, L and Sagan, D, Acquiring genomes: a theory of the origins of
species. books.google.com (2003).
14! Gardner, T S, Cantor, C R, and Collins, J J, Construction of a genetic
toggle switch in Escherichia coli. Nature 403 (6767), 339 (2000).

22
15! Elowitz, M B and Leibler, S, A synthetic oscillatory network of
transcriptional regulators. Nature 403 (6767), 335 (2000).
16" Tamsir, Alvin, Tabor, Jeffrey J, and Voigt, Christopher A, Robust
multicellular computing using genetically encoded NOR gates and
chemical ‘wires’. Nature, 1 (2010).
17! Murphy, K. F et al., Tuning and controlling gene expression noise in
synthetic gene networks. Nucleic Acids Research, 1 (2010).
18! Lee, S.K et al., Directed evolution of AraC for improved compatibility of
arabinose-and lactose-inducible promoters. Applied and Environmental
Microbiology 73 (18), 5711 (2007).
19! Purnick, Priscilla E. M and Weiss, Ron, The second wave of synthetic
biology: from modules to systems. Nat Rev Mol Cell Biol 10 (6), 410
(2009).
20! Lestas, Ioannis, Vinnicombe, Glenn, and Paulsson, Johan, Fundamental
limits on the suppression of molecular fluctuations. Nature 467 (7312), 174
(2010).
21! Guell, M et al., Transcriptome Complexity in a Genome-Reduced
Bacterium. Science 326 (5957), 1268 (2009).
22! Chan, Leon Y, Kosuri, Sriram, and Endy, Drew, Refactoring bacteriophage
T7. Mol Syst Biol 1 (1), E1 (2005).
23! Glass, John I et al., Essential genes of a minimal bacterium. Proc Natl
Acad Sci USA 103 (2), 425 (2006).
24" Cornelius, S and Lee…, J, Dispensability of Escherichia coli's latent
pathways. Proceedings of the … (2011).
25! Zhang, J, Evolution by gene duplication: an update. Trends in Ecology &
Evolution 18 (6), 292 (2003).
26! Burrill, D.R and Silver, P.A, Making cellular memories. Cell 140 (1), 13
(2010); Drubin, David A, Way, Jeffrey C, and Silver, Pamela A, Designing
biological systems. Genes Dev 21 (3), 242 (2007); Bashor, Caleb J,
Horwitz, Andrew A, Peisajovich, Sergio G, and Lim, Wendell A, Rewiring
Cells: Synthetic Biology as a Tool to Interrogate the Organizational
Principles of Living Systems. Annual review of biophysics (2010).
27! Deichmann, U, Chemistry and the Engineering of Life Around 1900:
Research and Reflections by Jacques Loeb. Biological Theory 4 (4), 323
(2009).
28! Feynman, R, Leighton, RB, and Sands, ML, The Feynman Lectures of
Physics. (2006).
29! O'Malley, M.A, Making knowledge in synthetic biology: Design meets
kludge. Biological Theory 4 (4), 378 (2009).
30! Mackenzie, Adrian, Design in synthetic biology. BioSocieties 5 (2), 180
(2010).
31! Tyo, Keith EJ, Kocharin, Kanokarn, and Nielsen, Jens, Toward Design-
based Engineering of industrial microbes. Curr Opin Microbiol, 1 (2010).

23
32! Mazzotti, M, Knowledge as social order: rethinking the sociology of Barry
Barnes. books.google.com (2008); CARLSON, R, Biology is technology the
promise, peril, and business of engineering. lavoisier.fr (2009).
33! Specter, M, A life of its own. The New Yorker September 28 (2009).
34! Haseltine, Eric L and Arnold, Frances H, Synthetic gene circuits: design
with directed evolution. Annual review of biophysics and biomolecular
structure 36, 1 (2007); Porcar, Manuel, Beyond directed evolution:
Darwinian selection as a tool for synthetic biology. Syst Synth Biol, 1
(2009).
35! Farinas, E T, Bulter, T, and Arnold, F H, Directed enzyme evolution. Curr
Opin Biotechnol 12 (6), 545 (2001).
36! Chatterjee, R and Yuan, L, Directed evolution of metabolic pathways.
Trends Biotechnol 24 (1), 28 (2006).
37! Ro, Dae-Kyun et al., Production of the antimalarial drug precursor
artemisinic acid in engineered yeast. Nature 440 (7086), 940 (2006).
38! Bayer, TS et al., Synthesis of methyl halides from biomass using
engineered microbes. J. Am. Chem. Soc 131 (18), 6508 (2009).
39! Maamar, H, Raj, A, and Dubnau, D, Noise in gene expression determines
cell fate in Bacillus subtilis. Science 317 (5837), 526 (2007).
40! Wang, Harris H et al., Programming cells by multiplex genome engineering
and accelerated evolution. Nature 460 (7257), 894 (2009).
41" Guet, Călin C, Elowitz, Michael B, Hsing, Weihong, and Leibler, Stanislas,
Combinatorial synthesis of genetic networks. Science 296 (5572), 1466
(2002).
42! Yokobayashi, Yohei, Weiss, Ron, and Arnold, Frances H, Directed
evolution of a genetic circuit. Proc Natl Acad Sci USA 99 (26), 16587
(2002).
43! Lim, W, The modular logic of signaling proteins: building allosteric
switches from simple binding domains. Curr Opin Struct Biol (2002).
44! Zarrinpar, Ali, Park, Sang-Hyun, and Lim, Wendell A, Optimization of
specificity in a cellular protein interaction network by negative selection.
Nature 426 (6967), 676 (2003).
45! Skerker, J et al., Rewiring the Specificity of Two-Component Signal
Transduction Systems. Cell 133 (6), 1043 (2008).
46! Dueber, John E, Yeh, Brian J, Chak, Kayam, and Lim, Wendell A,
Reprogramming control of an allosteric signaling switch through modular
recombination. Science 301 (5641), 1904 (2003).
47! Yeh, Brian J et al., Rewiring cellular morphology pathways with synthetic
guanine nucleotide exchange factors. Nature 447 (7144), 596 (2007).
48! Park, Sang-Hyun, Zarrinpar, Ali, and Lim, Wendell A, Rewiring MAP kinase
pathways using alternative scaffold assembly mechanisms. Science 299
(5609), 1061 (2003); Bashor, C. J, Helman, N. C, Yan, S, and Lim, W. A,
Using Engineered Scaffold Interactions to Reshape MAP Kinase Pathway
Signaling Dynamics. Science 319 (5869), 1539 (2008).

24
49! Dueber, John E et al., Synthetic protein scaffolds provide modular control
over metabolic flux. Nat Biotechnol 27 (8), 753 (2009).
50! Gou, Ping et al., Higher order structure contributes to specific differences
in redox potential and electron transfer efficiency of root and leaf
ferredoxins. Biochemistry 45 (48), 14389 (2006).
51! Gilardi, G, Fantuzzi, A, and Sadeghi, S J, Engineering and design in the
bioelectrochemistry of metalloproteins. Curr Opin Struct Biol 11 (4), 491
(2001).
52! Wong, T S and Schwaneberg, U, Protein engineering in
bioelectrocatalysis. Current Opinion in Biotechnology 14 (6), 590 (2003).
53" Sadeghi, Sheila J et al., Engineering artificial redox chains by molecular
‘Lego’. Faraday Disc. 116, 135 (2000).
54! Hall, D O, Cammack, R, and Rao, K K, Role for ferredoxins in the origin of
life and biological evolution. Nature 233 (5315), 136 (1971).
55! Otaka, E and Ooi, T, Examination of protein sequence homologies: IV.
Twenty-seven bacterial ferredoxins. J Mol Evol 26 (3), 257 (1987).
56! Otaka, E and Ooi, T, Examination of protein sequence homologies: V. New
perspectives on evolution between bacterial and chloroplast-type
ferredoxins inferred from sequence evidence. J Mol Evol 29 (3), 246 (1989).
57! Peters, J W, Lanzilotta, W N, Lemon, B J, and Seefeldt, L C, X-ray crystal
structure of the Fe-only hydrogenase (CpI) from Clostridium pasteurianum
to 1.8 angstrom resolution. Science 282 (5395), 1853 (1998).
58! Kleine, Tatjana, Maier, Uwe G, and Leister, Dario, DNA Transfer from
Organelles to the Nucleus: The Idiosyncratic Genetics of Endosymbiosis.
Annu. Rev. Plant Biol. 60 (1), 115 (2009).
59! Pelz, O et al., Towards elucidation of microbial community metabolic
pathways: unravelling the network of carbon sharing in a pollutant-
degrading bacterial consortium by immunocapture and isotopic ratio mass
spectrometry. Environ Microbiol 1 (2), 167 (1999).
60! Gray, MW, Burger, G, and Lang, BF, Mitochondrial evolution. Science 283
(5407), 1476 (1999); Howe, C J et al., The origin of plastids. Philos Trans R
Soc Lond, B, Biol Sci 363 (1504), 2675 (2008).
61! Douglas, AE, Nutritional interactions in insect-microbial symbioses: aphids
and their symbiotic bacteria Buchnera. Annual Review of Entomology 43
(1), 17 (1998).
62! Rumpho, M E, Summer, E J, and Manhart, J R, Solar-powered sea slugs.
Mollusc/algal chloroplast symbiosis. Plant Physiol 123 (1), 29 (2000).
63! Chin, Jason W, Modular approaches to expanding the functions of living
matter. Nat Chem Biol 2 (6), 304 (2006).
64! Waks, Z and Silver, P. A, Engineering a Synthetic Dual-Organism System
for Hydrogen Production. Appl Environ Microbiol 75 (7), 1867 (2009).
65! Basu, Subhayu et al., A synthetic multicellular system for programmed
pattern formation. Nature 434 (7037), 1130 (2005).

25
66! McMillen, David, Kopell, Nancy, Hasty, Jeff, and Collins, J J, Synchronizing
genetic relaxation oscillators by intercell signaling. Proc Natl Acad Sci USA
99 (2), 679 (2002).
67! Gore, Jeff, Youk, Hyun, and Van Oudenaarden, Alexander, Snowdrift game
dynamics and facultative cheating in yeast. Nature 459 (7244), 253 (2009);
Shou, Wenying, Ram, Sri, and Vilar, Jose M G, Synthetic cooperation in
engineered yeast populations. Proc Natl Acad Sci USA 104 (6), 1877
(2007).
68! Wintermute, Edwin H and Silver, Pamela A, Emergent cooperation in
microbial metabolism. Molecular Systems Biology 6, 1 (2010).
69! Cirillo, J D, Exploring a novel perspective on pathogenic relationships.
Trends Microbiol 7 (3), 96 (1999).
70! Margulis, L, Words as battle cries--symbiogenesis and the new field of
endocytobiology. Bioscience 40 (9), 673 (1990).
71! Buchsbaum, R and Buchsbaum, M, An Artificial Symbiosis. Science 80
(2079), 408 (1934).
72! Taylor, D L, Artificially Induced Symbiosis Between Marine
Flagellates and Vertebrate Tissues in Culture. J. Protozool. 25 (1), 77
(1978).
73! Niederholtmeyer, H et al., Engineering synthesis and export of hydrophilic
products from cyanobacteria. Appl Environ Microbiol, 1 (2010).
74! McDonough, W and Braungart, M, Cradle to Cradle. (2002).

26
Chapter 2
/ Synthetic Hydrogen Metabolism in E. coli

“Hydrogen has always been the fuel of the future, and it looks like it always
will be.”
-Gary Kendall
Background2

Genetic and metabolic circuits created in synthetic biology are able to perform

valuable technological functions and test basic science hypotheses by mimicking natural

genetic networks in a simplified context1. These pathways often are intended to function

orthogonally to existing cellular behaviors, creating a framework for studying!in

vivo!behaviors with the isolated simplicity of!in vitro!studies4. Much of the work in

synthetic biology has focused on understanding and recombining cellular information

processing in this way, integrating environmental signals into a novel, coordinated

biological behavior. Here, the indirect analogy to computational information processing

serves as a simplifying and abstracting framework for understanding and engineering

such complex systems. Recently, synthetic biology and biological engineering have

begun to deal with the physical electronic properties of cells and membrane channels5

and more complex circuit design made up of many biological parts from many

organisms6. The engineering of cellular components that manage and transfer electrical

charge can lead to novel methods of integrating electronic devices and biological

2* Work presented in this chapter was published in the following papers:

1. Agapakis CM, Ducat DC, Boyle PM, Wintermute EH, Way JC, and Silver PA. “Insulation of a Synthetic
Hydrogen Metabolism Circuit in Bacteria.” Journal of Biological Engineering, 2010, 4:3. (Attached as
Appendix C)
2. Agapakis CM and Silver PA. “Modular Electron Transfer Circuits for Synthetic Biology: Insulation of an
Engineered Biohydrogen Pathway.” Bioengineered Bugs, 2010, 1:6. (Attached as Appendix D)
3. Barstow B, Agapakis CM, Boyle PM, Grandl G, Silver PA, and Wintermute EH. “Linking Hydrogenase
Activity to Essential Metabolism.” Manuscript in review.

28
systems7, as well as the design of complex redox pathways within cells for the

production of reduced molecules such as fuels.

Electron transfer pathways in biology are crucial for the reduction of inorganic

chemicals into biologically functional molecules and the metabolic breakdown of organic

molecules as a source of energy. Electrons readily tunnel quantum-mechanically when

two protein-bound iron-sulfur clusters are brought into close contact, with an optimal

tunneling distance of 14 Å. Because they must be bound to proteins rather than freely

diffusing, iron-sulfur bound electrons are unique metabolites with fascinating

opportunities for engineering in vivo control. Many electron transfer proteins have been

studied in detail and described in terms of engineering principles8 and have been

recombined both in vitro9 and in vivo,10 making them ideal parts for synthetic biology.

Electrons typically tunnel from lower to higher potential iron-sulfur clusters in

what can loosely be thought of as an amorphous electronic circuit (figure 2.1). The

photosystems of plants and the catabolic enzymes of oxidative respiration or

fermentation introduce high-energy electrons into cellular metabolism, and can therefore

be thought of as metabolic “batteries.” These low potential electrons are then passed to

proteins or cofactors that can hold electrons and transfer them to other oxidoreductase

enzymes, which use them to generate a measurable output, such as reduction of protons,

sulfur, nitrogen, oxygen, and carbon, or the hydroxylation of steroids11.

29
 Because oxidoreductase enzymes are reversible and these circuits are made of

freely diffusing cellular components, the electrical potential of the protein bound iron-

sulfur cluster and the nature of the protein-protein interactions will determine the

direction and the rate of the electron “current” traveling through the pathway

components. Most!E. coli!redox enzymes, including the [NiFe]-hydrogenases, are

electronically coupled to the cofactor NADH, with a reducing potential of -320 mV.

Many enzymes from plants and anaerobic bacteria, however, are partnered with the

protein-based electron carrier ferredoxin, which can have a significantly lower reducing

Photosystem I

-1300

-520
Redox potential

Hydrogenase
PFOR
-420 H2
Bacterial type Fd Leaf-type Fd

-345
Sulfite Reductase Root-type Fd

-285

Figure 2.1 Electron transfer between protein-bound iron sulfur clusters. Electrons are
transferred between protein-bound iron sulfur clusters, typically from lower to higher redox
potential. Reducing equivalents from the photosystem or pyruvate ferredoxin oxidoreductase
(PFOR) can be transferred via ferredoxin (Fd) to the hydrogenase. Hydrogen can serve as a
source of reducing equivalents, transferring electrons to sulfite reductase via Fd.

30
potential, typically close to that of the H2/H+ pair (-420 mV)12. As a class of proteins

with different regulatory and structural properties that bind iron-sulfur clusters of

variable potential, ferredoxins represent a tremendous toolbox for synthetic biology

applications. Furthermore, while ferredoxin activity has been well characterized in many

photosynthetic pathways, the function of ferredoxin in E. coli13, though essential14,

remains poorly understood. Here, synthetic biology, through engineering and functional

testing of ferredoxin-based electron transfer pathways, may be useful in elucidating the

natural biology of bacterial iron-sulfur cluster protein systems.

Ferredoxin-dependent [FeFe]-hydrogenases, with midpoint potential of -420 mV,

can serve as both a source and sink for electrons in artificial electron transfer pathways.

Natural hydrogen metabolism pathways in a variety of prokaryotic and eukaryotic

species can either consume hydrogen as a source of low potential electrons, or produce

hydrogen as a sink for reducing equivalents generated during anaerobic fermentation or

photosynthesis. Synthetic hydrogen production pathways in E. coli have potential to lead

to industrial scale source of hydrogen fuel and interrogate the basic biology and

evolution of electron metabolism. The potential of synthetic hydrogen consumption

pathways is less obvious but equally powerful; hydrogen consumption can provide

reducing equivalents for numerous synthetic metabolic pathways, while linking hydrogen

consumption to essential metabolism can be used to establish a genetic selection for

improved hydrogenase function.

31
Pathway design

We designed synthetic electron transfer circuits in E. coli that replace native

NADH-dependent hydrogen consumption15 and production activity16 with [FeFe]-

hydrogenases from a diverse range of species. Due to ferredoxins’ low reducing

potential, [FeFe]-hydrogenases thermodynamically favor hydrogen production, and their

high hydrogen production activity, conserved structure, and relatively simple maturation

pathway make [FeFe]-hydrogenases excellent enzymatic modules for recombinant

expression in a synthetic!system. Furthermore, heterologously expressed [FeFe]-

hydrogenases from several species have been characterized in vitro2 and in vivo 10.

Previous research using hydrogenases within engineered electron transfer

pathways have relied heavily on in vitro approaches. For example, hydrogenase enzymes

have been explored as tools for hydrogen fuel production by purified enzyme cocktails in

vitro17. Alternatively, the hydrogenase active site has been modeled synthetically for use

in fuel cells as a catalytic center that does not require rare metals to function18. While

these in vitro systems are inherently insulated from natural metabolism, they suffer from

the same drawbacks as other in vitro enzymatic processes in the difficulty of production

and purification, and lack of robustness of the living cell. Metabolic19 and protein

engineering20 of natural hydrogen production pathways in E. coli have led to

improvements in hydrogen yield, but these methods are limited in the substrate

specificity of the native [NiFe]-hydrogenases21. Synthetic pathways expressing

32
hydrogenases along with exogenous electron donors and carriers can be used to

supplement and optimize hydrogen production from E. coli, as well as improve our

understanding of electron transfer pathways.

Heterologous expression of [FeFe]-hydrogenase alone is sufficient for small,

measurable hydrogen production from E. coli in vivo, and this hydrogen production is

increased with the co-expression of ferredoxins from several organisms10.

Overexpression of a ferredoxin oxidoreductase can link hydrogen production to cellular

metabolism22. Pyruvate-ferredoxin oxidoreductase (PFOR) from several anaerobic or

microaerobic species of microorganisms reduces ferredoxin as it breaks down pyruvate

to acetyl-CoA. Coexpression of PFOR, ferredoxin, and [FeFe]-hydrogenase therefore

couples the breakdown of glucose with the establishment of a reduced ferredoxin pool

(Figure 2.2A). As [FeFe]-hydrogenases are efficient electron acceptors from ferredoxin,

co-expressed [FeFe]-hydrogenases within this system can reset the redox state of the

ferredoxin pool, thereby completing the circuit, and providing a readout of electron flux

through the pathway as a function of hydrogen production.

At the higher end of the redox spectrum, several enzymatic pathways in E. coli

translate NADH reducing power to required metabolic functions, including glutamate,

nitrite, and sulfite reduction. Expression of a ferredoxin dependent sulfite reductase

(SiR) from corn Zea mays, can complement the deletion of the native, NADH dependent

sulfite reductase, cysI. Here again, heterologous expression of the[FeFe]-hydrogenase

33
resets the ferredoxin pool by providing a source of reducing power from hydrogen gas

(Figure 2.2B).

Figure 2.2 Schematic of synthetic hydrogen metabolism pathways A.) Pyruvate metabolism
to acetyl-CoA in E. coli with a PFOR-ferredoxin-hydrogenase synthetic pathway. Native
enzymes are indicated in black: PFL, formic acid dependent pyruvate formate lyase, PDH,
NADH dependent pyruvate dehydrogenase, heterologous enzymes in blue. B.) Sulfite
reduction to sulfide typically proceeds through an NADH dependent sulfite reductase enzyme,
cysI. We replaced this native function with a ferredoxin dependent sulfite reductase (SiR) from
corn, Zea mays.

Synthetic biology circuits such as these hydrogen metabolism pathways must be

integrated into cellular metabolism in order to function, but must also be insulated from

competing reactions in order to optimize output and ensure proper behavior. Natural

electron transfer pathways are insulated in several ways that can be adapted for use in

synthetic circuits (Figure 2.3). Elimination of competing reactions, through gene

deletion, transcriptional regulation, or spatial separation into different subcellular

compartments, can isolate proteins from competition for electrons. For example, the

34
expression of hydrogenase genes from many organisms are transcriptionally regulated

by the presence of oxygen23, likely in order to prevent competition for electrons with

aerobic metabolism.

Figure 2.3 Insulation strategies

for synthetic electron transfer
pathways; deletion of competing
reactions, optimization of
binding surfaces, direct protein-
protein fusion, and localization to
a synthetic protein scaffold.

Isolation of electron transfer pathways can also evolve through physical changes to

proteins, either through point mutations that alter interaction surfaces between two

redox partners, or adaptations that co-localize interaction partners to one another or to

secondary scaffolding complexes. Ferredoxins are characteristically acidic, and

electrostatic forces have been shown to be important for the interactions between

ferredoxins and some oxidoreductases, in particular the hydrogenase from

Chlamydomonas reinhardtii24,25. Mutational analyses of acidic surface residues in

ferredoxin show that different residues are important for binding to different

oxidoreductases26. However, electrostatic forces have also been shown to not be

important in other ferredoxin-paired reactions, notably that of bacterial ferredoxin and

the clostridial hydrogenase9. In these cases, random transient interactions between

proteins may be sufficient to initiate the extremely fast electron tunneling between iron-

sulfur proteins, and dynamic interactions must be regulated in a different way. As such,

the coevolution of protein interaction surfaces has been postulated to play a role in the

35
control of bacterial signal transduction pathways27 and is likely involved in the evolution

of many iron-sulfur containing proteins capable of interacting with ferredoxins28.

Isolation of iron-sulfur proteins through physical scaffolding, either in the

mitochondrial or chloroplast membranes or through direct protein fusion, is thought to

play a large role in the evolution of complex electron transfer pathways as well as the

iron-sulfur proteins themselves. For example, larger [FeFe]-hydrogenases, such as those

from Clostridium species contain several “ferredoxin-like” domains29. It is speculated that

these domains arose through ancestral gene fusions, enhancing hydrogenase interaction

with other ferredoxins, and providing an electron transport channel towards the

hydrogenase active site.

As they are common throughout evolutionary history, ferredoxins can be

functionally expressed heterologously in a wide host range30, and have been shown to be

amenable to functional gene fusion in vitro31. The genomes of many organisms include

several putative ferredoxin proteins, but one ferredoxin is particular is typically ideal for

a given oxidoreduction reaction32, and a diversity of ferredoxins are typically expressed

in the leaves of higher plants to account for the complex range of electron transfer

reactions occurring in these organisms33. The specification of each ferredoxin to its

particular application may be due to its electronic potential, determined by the amino

acid composition surrounding the iron-sulfur cluster, or due to the surface

complementarity between ferredoxin and the cognate oxidoreductase34. A synthetic

36
metagenomics35 or directed evolution36 approach to the development of engineered iron-

transfer pathways in synthetic biology could identify the ideal ferredoxin for each desired

application.

We created and characterized two synthetic electron transfer circuits. First, a

hydrogen production pathway that couples hydrogen evolution with the breakdown of

glucose in E. coli via heterologous expression of PFOR, ferredoxin, and [FeFe]-

hydrogenase. Using the modular nature of our synthetic circuit, we recapitulated several

of the aforementioned isolation strategies for electron transfer pathways. We reproduced

spatiotemporal isolation through the deletion of competing iron-sulfur proteins and

explored the interaction surface of the hydrogenase and ferredoxin, testing four

mutations of surface amino acids of the [FeFe]-hydrogenase from Chlamydomonas

reinhardtii previously predicted to improve ferredoxin binding24. We synthetically

modeled physical scaffolding of electron transfer proteins, both through direct protein

fusion of the Clostridium acetobutylicum hydrogenase and ferredoxins with flexible peptide

linkers, and through connection of hydrogenase and ferredoxin to a heterologous protein

scaffold37. All of these insulation strategies significantly affected the function of our

synthetic circuit, in many cases increasing total hydrogen production. The highest

improvement was seen with direct protein-protein fusion of the hydrogenase and

ferredoxin, with an optimal linker length increasing hydrogen production by up to four

fold. This method is easily transferrable to other synthetic electron transfer pathways

37
and may provide clues to understanding the evolution of complex electron transfer

proteins with multiple ferredoxin-like domains.

Second, we designed hydrogen consumption circuit where oxidation of hydrogen

by the [FeFe]-hydrogenase was coupled to essential metabolism through sulfite

reduction. Sulfite reduction to sulfide is required for production of the amino acids

cysteine and methionine in minimal media without added amino acids. Insulation was

required to prevent “short circuiting” of the pathway, where electrons can enter the

pathway through a source other than the hydrogenase and bypass the requirement of the

synthetic circuit, allowing for growth even in the absence of a functional hydrogenase.

Deletion of competing reactions was sufficient to eliminate short circuit growth in the

absence of the hydrogenase. This insulation, however, was insufficient to prevent short

circuiting through a previously unknown latent function in the hydrogenase—while

growth was hydrogenase dependent, the cells could grow on minimal media in the

absence of hydrogen. This suggests an electron transfer role for the hydrogenase that is

not tied to its catalytic function.

Materials and Methods

Plasmids and cloning

All cloning was done in E. coli DH5a. Hydrogenase genes from Chlamydomonas

reinhardtii, the ferredoxin I gene from Spinacia olearcea, and the ferredoxin- NADPH

38
reductase (FNR) gene from Zea mays were commercially synthesized by Codon Devices

(Cambridge, MA), codon optimized for expression in Saccharomyces cerevisiae and

acceptable for use in E. coli for wide applicability (see Appendix A, figure S1 for

nucleotide sequences). Hydrogenases from Clostridium acetobutylicum and Clostridium

saccharobutylicum were cloned from plasmids received from Matthew Posewitz (National

Renewable Energy Laboratory, Golden, CO). Hydrogenase genes HydA and HydB were

cloned from Shewanella oneidensis using colony PCR of bacterial cultures from Colleen

Hansel (Harvard University, Cambridge, MA). Thermotoga maritima HydA was cloned

from genomic DNA provided by Kenneth Noll (University of Connecticut, Storrs, CT).

PFOR and ferredoxin 32 were cloned from Clostridium acetobutylicum genomic DNA

(ATCC, Manassas, VA). Zea mays ferredoxin (GenBank ACA34367), Spinacia olearcea

ferredoxin (GenBank AAA34028), and Zea mays sulfite reductase (GenBank BAA23641)

were cloned from genomic DNA isolated from corn using DNeasy Plant Mini Kit

(Qiagen, Valencia, CA). PFOR from Desulfovibrio africanus was isolated from plasmid

pLP138 provided by Laetitia Pieulle (Centre National de la Recherche Scientifique,

Marseille, France) and ydbK was obtained through colony PCR of E. coli BL21.

Plasmids for tetracycline-responsive expression of synthetic scaffold proteins were

provided by John Dueber37 (University of California, Berkeley), and ferredoxin/

hydrogenase fusions to metazoan GBD, SH3, and PDZ ligand domains were constructed,

bridged by flexible (Gly4Ser)x linkers. Oligonucleotides for metazoan ligand domains and

39
linker sequences were purchased from Integrated DNA Technologies, with ligand

sequences identical to those previously reported37.

Site directed mutagenesis of the Clostridium and Shewanella hydrogenase genes to

remove restriction sites needed for cloning and for alteration of C. reinhardtii HydA

ferredoxin binding surface was carried out using the Quikchange Multi-Site Directed

Mutagenesis Kit according to manufacturer instructions (Stratagene, La Jolla, CA).

Cloning and mutagenesis primers are listed in Appendix A, table 1 (Integrated DNA

Technologies, Coralville, IA)

Cloning of hydrogenase-ferredoxin fusion proteins was done using BioBrick

standard assembly 38 and subsequently cloned into Novagen Duet vectors (Novagen,

Gibbstown, NJ) whose multiple cloning sites were mutated to accept BioBrick parts

(figure 2.4). Strep-II tag (WSHPQFEK), (Gly4Ser) 2, and (Gly4Ser) 4 oligonucleotides

with flanking BioBrick sites were purchased from Integrated DNA Technologies. Longer

(Gly4Ser) linkers were made through BioBrick fusion of multiple linker sequences or

through PCR amplification from other chimeric proteins39.

Protein expression

All protein expression and hydrogenase activity assays were performed in E. coli

BL21 (DE3). Cells were transformed with modified pCDF-duet with C. reinhardtii

HydEF in MCS1 and C. reinhardtii HydG in MCS2, and with modified pACYC-duet with

C. acetobutylicum PFOR or E. coli ydbK in MCS1 or Desulfovibrio africanus PFOR cloned

40
into the downstream NdeI and AvrII sites of MCS2. Hydrogenase/ferredoxin pairs were

transformed either in each multiple cloning site of modified pET-duet, or for the S.

oneidensis hydrogenase HydA in MCS1, HydB in MCS2, and ferredoxin in MCS1 of

modified pCOLA-duet. To compare in vitro hydrogen production using maturation

factors from Clostridium acetobutylicum, we used plasmids provided by Matthew Posewitz

(pCDF-duet with CaHydE in MCS1, CaHydF in MCS2 and pET-duet with CaHydA in

MCS1 and CaHydG in MCS2 2). Artificial scaffolds were expressed from pJD plasmids

provided by John Dueber, previously described in Dueber et. al. 2009 37.

E. coli Gene Deletion

Hydrogenase knockout ("hycE, "hyaB, "hybC) and "fpr, "ydbK, "hcr, "yeaX, "hcaD,

or "frdB single deletion strains were made by sequential P1 transduction from the Keio

collection 40 into BL21(DE3) "tonA, followed by removal of the KanR marker by

standard procedures.

SDS-Page and Western Blotting

E. coli cells were lysed with Bacterial Protein Extraction Reagent (B-PER, Pierce,

Rockford, IL), protein samples were normalized using the Bradford assay (Bio-Rad,

Hercules, CA), diluted into SDS-PAGE loading buffer and loaded onto a 4-20% Tris/

glycine/SDS acrylamide gel. a-Strep-tag II antibody (HRP-conjugated, Novagen,

Gibbstown, NJ) or a-ferredoxin primary antibody (Agrisera, Vännäs, Sweden) and a-

Rabbit IgG secondary antibody were used.

41
 Figure 2.4 Sequence of modified multiple cloning sites of Novagen Duet Vectors. 5’
phosphorylated oligonucleotide inserts (Integrated DNA Technologies, red text) were inserted
between Nco I and Afl II sites in MCS1 and Nde I and Avr II of MCS2 of Novagen Duet Vectors
pET-Duet, pACY-duet, pCDF-duet, and pCOLA-duet for heterologous expression of up to eight
BioBrick sequences.

Hydrogen production assays

Bacterial cultures were grown aerobically for two hours until reaching an OD600 of

approximately 0.15 in LB media with appropriate antibiotic (50 µg/ml ampicillin, 25 µg/

ml spectinomycin, 25 µg/ml kanamycin, and/or 12.5 µg/ml chloramphenicol) in 40 ml

glass serum vials, induced with 1mM IPTG (and 2 µg/ml anhydrous tetracycline when

relevant for the induction of scaffold proteins) and sparged with argon. For the methyl

viologen assay, adapted from King et. al. 2, vials were sparged for 2 hours and then lysed

with assay buffer containing 50 mM Tris pH 7.0, 50% B-PER, 10mM methyl viologen

(Sigma, St. Louis, MO) and 50mM sodium dithionite (Fisher, Pittsburgh, PA), the vials

were capped with rubber septa, the cells were vortexed and allowed to rock overnight.

Hydrogen concentration in the headspace gas was measured by gas chromatography

(Shimadzu GC-14A). In vivo hydrogen production assays were performed in a similar

fashion, except that cultures were supplemented with 0.5% glucose at the time of IPTG

42
induction, sparged for 30 minutes and simply capped and shaken overnight at 37°C

before measuring headspace gas composition. Glucose curves were measured in cells

pretreated overnight with 1mM IPTG then immediately diluted into LB + variable

glucose + 1mM IPTG, then sparged and grown overnight. All hydrogen production

values were normalized to an OD600 of 0.15.

Colony growth assays under custom atmospheres

Cells were grown to saturation in induction media. Residual nutrients were

removed by 3x washing with PBS. Cell densities were measured with a hemocytometer

and diluted to a final concentration of 1 cell/!L. Fifty microliters (" 50 cells) were

dispensed onto selective plates and transferred Vacu-Quick jars (Almore International).

The jars were filled with 15% H2, O2 varying from 0-12.5%, and a balance of nitrogen

to a total internal pressure of 1 atm. Incubation was carried out at 37° C for 72 hours.

Plates were photographed in an inverted camera stage. Colonies were identified and sized

with an image analysis script implemented in MATLAB (MathWorks). Each data point

represents size data collected from roughly 50 individual colonies.

Hydrogenase mutagenesis and selection

Mutation libraries were created through error-prone PCR amplification of the

hydrogenase genes with the GeneMorph II Random Mutagenesis Kit (Agilent

Technologies). Exact primer sequences are indicated in the supplementary vector

43
sequence files. The PCR products were digested with NotI and SpeI (for caHydA) or

NcoI and NotI (for csHydA), purified, and ligated into a modified pCDFDuet-1 vector

(Novagen). A target frequency of 4.5 mutations/kb was sought according to the

manufacturer's instructions and confirmed by sequencing several clones plated

nonselectively.

Host cells were transformed with the mutant hydrogenase libraries by

electroporation and recovered in SOC media for 1 hour at 37° C. Following recovery, the

cells were washed three times with Phosphate Buffered Saline (PBS) and plated on

selective media. Approximately 107 unique transformants were applied to each plate, as

assessed by serial dilution of the recovered cells. Selection plates were then transferred

to Vacu-Quick jars prepared as above with a selective atmosphere containing 10% O2 and

15% H2. Incubation proceeded at 37° C for 4 days.

Selection restreaks were performed using Petri dishes with internal divisions to

block the diffusion of metabolites from neighboring streaks. Soft plastic inoculating

loops were used to prevent marring of the agar surface and maintain a uniformly oxic

environment.

Homology modeling

Homology model of C. reinhardtii HydA1 was made using the SWISS-MODEL41

server with the Clostridium pasteurinium HydA X-ray structure (1FEH 29) as a template.

44
Results

In vitro hydrogen production from heterologously expressed hydrogenases

To create a synthetic electron metabolism circuit with hydrogenase as the terminal

electron acceptor, we first investigated the activity of various hydrogenase genes

heterologously expressed in the presence of appropriate maturation factors. We adapted

a previously established in vitro hydrogenase activity assay2, and measured hydrogen

production from crude lysates of bacteria expressing hydrogenases and maturation

factors from several species in the presence of a chemical electron donor, methyl

viologen. Previous reports have shown that the hydrogenase maturation factors from C.

reinhardtii, HydEF and HydG, are unstable when heterologously expressed in E. coli 2,

likely due to the genes’ high GC content, while the maturation factors from Clostridium

acetobutylicum were able to mature [FeFe]-hydrogenases from a wide range of species.

Using commercially synthesized, codon optimized maturation factors from C. reinhardtii

we were able to alleviate the instability of the gene constructs. We found that in vitro

hydrogen production from the Clostridium acetobutylicum hydrogenase was identical when

coexpressed with the synthetic maturation factors or with HydE, HydF, and HydG from

C. acetobutylicum . All subsequent experiments were performed using the optimized C.

reinhardtii maturation factors.

We compared the in vitro hydrogen production of [FeFe]-hydrogenases

from!Clostridium acetobutylicum, Clostridium saccharobutylicum, Chlamydomonas reinhardtii,

45
Shewanella oneidensis, and Thermotoga maritima, all of which are homologous in their

catalytic domain (Appendix A, figure S2).!All hydrogenases except HydA from

Thermotoga maritima could be expressed at a high level in E. coli (figure 2.5A), and were

functional in vitro (figure 2.5B). Hydrogen levels increased linearly for the first several

hours of measurement, and we found that levels of hydrogen gas in the headspace after

overnight incubation correlated to the relative rate of hydrogenase activity during this

linear phase. Our overnight in vitro results agree with previous reports of in vitro

hydrogen production rates, with the hydrogenases from Clostridium species producing the

highest levels of hydrogen2. The heterologously expressed hydrogenase from Shewanella

oneidensis is functional at relatively low levels in vitro when both subunits are coexpressed

in E.coli with maturation factors from C. reinhardtii.

The in vitro assay is useful to test and compare the activities of heterologously

expressed hydrogenase genes, but as the assay uses an exogenous reducing agent, it does

not provide information on the electron flux within normal metabolic pathways in vivo.

To measure electron flux in vivo as a function of hydrogen production, hydrogenase

activity must be integrated into a functional electron transfer pathway. Construction of a

system where hydrogenase activity depends on electron transfer from ferredoxin would

allow for comparison to in vitro data to provide information on hydrogenase behavior and

hydrogenase-ferredoxin interaction dynamics.

46
In vivo construction and optimization of a synthetic hydrogen-producing circuit

To produce hydrogen in vivo from glucose, the [FeFe]-hydrogenase was

coexpressed with its required maturation factors, ferredoxin, and pyruvate-ferredoxin

oxidoreductase (PFOR) from different species. In this heterologous circuit, PFOR

oxidizes pyruvate to acetyl-CoA, reducing ferredoxin, which then transfers the electron

Figure 2.5 Characterization of the synthetic hydrogen production pathway. A.) Western blot of
Strep-II tagged hydrogenase expression. B.) In vitro hydrogen production from E. coli strains
expressing various hydrogenases, measured by the methyl viologen in vitro assay 2. C.a. = C.
acetobutylicum, C.s. = C. saccharobutylicum, C.r. = C. reinhardtii, S.o. = Shewanella oneidensis. C.)
Glucose-dependence of hydrogen production. Here and below, in vivo and in vitro hydrogen
production values are in units of µmol hydrogen/ml of E. coli culture, normalized to an OD600 of
0.15 unless otherwise stated. Assays were performed in triplicate, with error bars indicating
standard deviation. D.) In vivo hydrogen production from E. coli strains expressing all combinations
of the four hydrogenases vs. three ferredoxins from C. acetobutylicum, Spinacia olearcea (Sp.o), and
Zea mays (Zm). E.) In vivo hydrogen production from the C. acetobutylicum hydrogenase paired with
combinations of three ferredoxins and three PFOR genes.

47
to the hydrogenase. In normal E. coli metabolism, the oxidative breakdown of pyruvate

to acetyl-CoA is performed either aerobically by the pyruvate dehydrogenase complex,

reducing NAD+, or anaerobically by pyruvate formate lyase, generating formate (Figure

2.2A). PFOR functions in certain anaerobic bacteria and in eukaryotic parasites that

possess hydrogenosomes, organelles evolutionarily related to the mitochondrion that

generate a proton gradient through the production of hydrogen gas42. PFOR is an

attractive electron source for a synthetic hydrogen production circuit as overexpression

of a putative E. coli PFOR homolog, ydbK, increases in vivo hydrogen production by

heterologously expressed [FeFe]-hydrogenase and ferredoxin22,!PFOR purified

from!Clostridium pasteurianum!has been shown to reduce a number of ferredoxins!in

vitro9, and functional PFOR from!Desulfovibrio africanus!has been recombinantly

expressed in!E. coli43.

Consistent with the establishment of a synthetic electron transport circuit in vivo,

we observed high levels of glucose-dependent hydrogen production upon coexpression

of PFOR, hydrogenase and its maturation factors, and ferredoxin all from Clostridium

acetobutylicum in an E. coli strain lacking endogenous hydrogenases (hycE, "hyaB, "hybC,

figure 2.5C). Hydrogen production was again measured after overnight incubation, as we

found that hydrogen production in vivo from glucose was exhausted after 16 hours (data

not shown). We were unable to detect hydrogen production in the parental strain of E.

coli with the native hydrogenases deleted. Removal of any individual pathway component

48
from the synthetic circuit drastically reduced in vivo hydrogen production. However, as

has been previously reported, there was a small background level of hydrogen

production from expression of hydrogenase and maturation factors alone44. Consistent

with previous results10, we found this background hydrogen production was slightly

increased upon overexpression of ferredoxin in addition to hydrogenase, indicating that

there are!E. coli!proteins capable of reducing both hydrogenases and plant-type

ferredoxins, several candidate proteins of which we deleted in the following section

(figure 2.7).

The hydrogenase-ferredoxin-PFOR pathway constitutes a modular system, where

each element can be exchanged with homologous genes from different organisms. By

coexpressing pathway enzymes from diverse microorganisms, we were able to compare

the relative interaction strengths of four hydrogenases, three ferredoxins (figure 2.5D),

and three PFORs (figure 2.5E). All ferredoxins were able to transfer electrons between

PFOR and hydrogenases from different species with varying levels of efficiency.

In vivo hydrogen production from circuits expressing each of the four

hydrogenases (C. acetobutylicum, C. saccharobutylicum, C. reinhardtii, and S. oneidensis)

followed the same trend as the in vitro experiments, with the highest hydrogen

production observed with the Clostridial hydrogenases (figure 2.5D). The relative

interaction and electron transfer rates for hydrogenase and ferredoxin were explored by

comparing the in vivo hydrogen production of circuits made up from all pairwise

49
combinations of the four hydrogenases and ferredoxin from C. acetobutylicum, Spinacea

olearcea, and Zea mays and the PFOR from C. acetobutylicum (figure 2.5D). All

hydrogenases produced the highest output when co-expressed with bacterial type 2-

[4Fe-4S] ferredoxin from Clostridium acetobutylicum, with a potential of -420mV32.

Intermediate levels of hydrogen were produced using leaf-type [2Fe-2S]-ferredoxin I

from spinach, S. olearcea (-420 mV45) while the homologous root-type ferredoxin III from

corn, Z. mays (-345 mV45) led to significantly lower in vivo hydrogen levels in all cases.

Interestingly, the difference in hydrogen production from circuits expressing bacterial

versus plant-type ferredoxins was more significant for hydrogenases from bacterial

species. Hydrogenase from C. reinhardtii, which naturally pairs with plant-type

ferredoxins, produced similar levels of hydrogen when co-expressed with ferredoxin

from C. acetobutylicum or S. olearcea (figure 2.5D).

The interaction of overexpressed PFOR from C. acetobutylicum, D. africanus, or the

PFOR homolog ydbK from E. coli with the three ferredoxins was compared in a similar

fashion in circuits containing the C. acetobutylicum hydrogenase (figure 2.5E).

Overexpression of PFOR from C.acetobutylicum and ydbK from E. coli led to similar

levels of hydrogen production, although surprisingly, the highest levels of hydrogen

produced from ydbK occurred when ydbK was coexpressed with plant-type ferredoxin

from S. olearcea. Overall, the highest levels of hydrogen production were seen with the

50
PFOR from D. africanus, coexpressed with the hydrogenase and ferredoxin from C.

acetobutylicum.

Linking plant ferredoxin activity to essential sulfur metabolism

The hydrogenase enzyme is reversible, able to catalyze the production and

consumption of hydrogen. When the hydrogenase acts as the metabolic “battery,”

breaking down hydrogen, electrons enter metabolism at a significantly higher potential

than from other sources, but this electron potential is enough to drive several essential

enzymatic reductions (figure 2.1). We sought to couple hydrogenase catalyzed

breakdown of hydrogen to this essential metabolism in order to establish a stringent

genetic selection on hydrogenase function for directed evolution.

Deletion of the E. coli NADH-coupled sulfite reductase, cysI, produced no

discernable phenotype in rich media, but halted growth on minimal media without

reduced sulfur in the form of the amino acids cysteine and methionine. In the presence of

oxygen, expression of the ferredoxin-coupled sulfite reductase from Z. mays (zmSiR) and

ferredoxin from corn or spinach, Spinacia olearcea, was insufficient to complement the

deletion, indicating that aerobically there is no cross talk between the plant SiR and

ferredoxins and the E. coli redox pool. To create a pool of electrons that can interact

with the plant ferredoxins and zmSiR, we coexpressed ferredoxin-NADPH-reductase

(FNR) from Z. mays. FNR transfers electrons from the oxidation of NADPH to

51
ferredoxin, linking the synthetic sulfite reductase pathway to E. coli metabolism and

rescuing growth.

FNR serves as a positive control in aerobic conditions because the hydrogenase

active site is irreversibly inactivated in the presence of oxygen and will not complement

the deletion of cysI. Anaerobically, the hydrogenase should be able to take the place of

FNR, maintaining a pool of reduced ferredoxin and complementing the deletion of cysI.

However, anaerobically the coexpression of zmSiR and ferredoxin alone is sufficient for

near wild-type levels of growth (figure 2.6). As with the hydrogen production circuit,

where small amounts of anaerobic hydrogen production were measured without co-

expression of PFOR, some factor in E. coli metabolism was able to reduce the plant-type

ferredoxin and complement the cysI deletion. Large-scale transcriptional changes in

response to change in oxygen levels have been well characterized in E. coli and likely play

a role in activating pathways that can interact with and reduce plant-type ferredoxins46.

To build a hydrogenase-dependent sulfite reductase pathway anaerobically, we therefore

had to insulate the synthetic circuit from native electron metabolism, preventing such

“short-circuiting”.

Isolation of synthetic hydrogen metabolism through deletion of competing reactions

Natural biological electron transfer circuits are insulated to prevent electron leaks

that can cause damage by creating oxygen radicals and insulated from one another to

52
prevent short circuiting47. We sought to insulate both our hydrogen producing circuit

from competing metabolism to improve levels of hydrogen production and the hydrogen

consumption circuit to ensure that growth in minimal media was coupled to hydrogenase

function rather than cross talk from E. coli metabolism. Insulation strategies such as

these may also be generalizable to other pathways, allowing us to better understand

natural biological pathway isolation, a priority for the design of synthetic metabolic

pathways.

Figure 2.6 Complementation of the sulfite reductase deletion, cysI, in E. coli with a synthetic
pathway. In oxygen, sulfite reductase from Z. mays is insufficient to complement the deletion,
requiring coexpression of ferredoxin from S. olearcea, and FNR from Z. mays to link sulfite
reduction to the E. coli redox pool. In anaerobic conditions, expression of sulfite reductase
and ferredoxin alone is sufficient for growth in minimal media, indicating cross talk between
anaerobically expressed E. coli electron transfer pathways and plant ferredoxins. Figure by
Buz Barstow and Edwin Wintermute, reproduced here with permission.

53
 Although our constructed pathways are made up of genes that are divergent from

E. coli metabolic enzymes, given the non-specific electrostatic interactions that mediate

many ferredoxin interactions48, native iron-sulfur proteins may interact with the proteins

of the heterologous pathway. This is evidenced by the background hydrogen production

in strains expressing only heterologous hydrogenases and ferredoxins and the

background growth observed anaerobically in the sulfite reductase pathway. Deletion of

these potentially competing redox interaction partners should improve pathway function.

To address these issues, we deleted six genes identified through their homology to plant-

type ferredoxins or ferredoxin oxidoreductases that still allowed for viability (fpr,

flavodoxin:NADP+ reductase 10; ydbK, the putative PFOR homolog 10; hcr, an NADH

oxidoreductase; yeaX, a predicted oxidoreductase; hcaD, ferredoxin:NAD+ reductase; and

frdB, fumarate reductase (figure 2.7).

Deletion of these six genes was performed sequentially in the "cysI "tonA

background strain and anaerobic growth in minimal media was measured with either the

full synthetic pathway of FNR, Fd, and zmSiR or coexpression of only zmSiR and Fd.

The full complement of knockouts, including deletion of the native hydrogenases, "hycE,

"hyaB, "hybC, dramatically decreased the background level of anaerobic growth

observed without FNR, with deletion of ydbK providing the single largest drop in

background growth (figure 2.8)

54
 Figure 2.7 Domain structure of deleted genes with ferredoxin homology. Fd-ferredoxin; fpr-
flavodoxin:NADP+ reductase; hcr-NADH oxidoreductase; yeaX-predicted oxidoreductase; hcaD-
ferredoxin:NAD+ reductase; frdB-fumarate reductase; ydbK-putative pyruvate-ferredoxin
oxidoreductase. Genes were identified by BLAST homology search of the Escherichia coli genome
against Spinacia olearcea ferredoxin I. Domain structure schematized from NCBI conserved domain
search.

These six deletions were tested individually in a "hycE, "hyaB, "hybC, "tonA

background while expressing hydrogenase from C. acetobutylicum and maturation factors

from C. reinhardtii, ferredoxin from S. olearcea, with or without co-expression of PFOR

from D. africanus. Deletion of fpr and ydbK have been previously shown to slightly

decrease the background level of hydrogenase activity in vivo10. We found that only the

ydbK deletion had any significant effect on hydrogen production compared to the

hydrogenase knockouts alone. The background level of hydrogen production from HydA

and ferredoxin expressed alone was decreased by half in the ydbK deletion strain,

whereas hydrogen production from the full pathway with the D. africanus PFOR was

55
 Figure 2.8 Growth of the cysI deletion
strain anaerobically complemented with
the full synthetic pathway (dark blue
bars) or with only zmSiR and soFd
(light blue bars) with sequential
knockout of E. coli proteins with
homology to ferredoxin
oxidoreductases. Figure by Edwin
Wintermute, reproduced with
permission.

Figure 2.9 Insulation of hydrogenase pathway through deletion of competing

reactions. Relative hydrogen production of different knockout strains compared to
parent strain (!hycE, !hyaB, !hybC) expressing hydrogenase alone (dark blue bar),
hydrogenase and ferredoxin only (yellow bars) or the full PFOR-ferredoxin-
hydrogenase pathway (green bars).

56
increased by 1.4 fold (figure 2.9). This is consistent with our finding that overexpression

of ydbK led to high levels of electron transfer when co-expressed with ferredoxin from

spinach, indicating that endogenous ydbK is able to disrupt the synthetic electron

transfer pathway.

Insulation through mutation of the hydrogenase-ferredoxin interaction surface

As an independent strategy of insulating the hydrogen production pathway, we

attempted to insulate the pathway from competing electron metabolism through

modification of the interaction surface of the hydrogenase and ferredoxin by rational

protein design. Co-evolution of interacting protein pairs for preferential binding

between natural partners likely plays a large role in the isolation of natural pathways49,

and this principle has been used in designing synthetic signal transduction systems27. To

test whether mutations in a component of the artificial pathway specifically enhanced

activity by improving the ferredoxin-hydrogenase interaction, we compared in vivo

activity, which is ferredoxin-dependent, with in vitro activity, in which a chemical

reducing agent, methyl viologen, drives hydrogen production. This in vitro assay thus

measures ferredoxin-independent hydrogen production, reflecting the activity and

expression level of the hydrogenase itself.

The interaction between ferredoxins and Clostridial hydrogenases is poorly

characterized, with evidence that more than simple electrostatic reactions may play an

57
role in mediating the transfer of electrons 9. In contrast, the interaction surface between

the hydrogenase from Chlamydomonas reinhardtii and its cognate ferredoxin has been

extensively modeled in silico, with evidence that this interaction has a strong electrostatic

component 50. Long et al 24 proposed a structural model for the interaction between the

hydrogenase HydA2 from Chlamydomonas reinhardtii and its cognate [2Fe-2S] ferredoxin,

suggesting several mutations that might enhance this interaction due to improved charge

complementarity (figure 2.10A). Ferredoxin is rich in negatively charged residues, and

the mutations, E5K, P2K, M119K, or D126K are designed to increase the positive charge

of the hydrogenase binding surface. We tested these mutations using HydA1 from C.

Figure 2.10 Insulation of the

hydrogenase pathway through
ferredoxin binding surface
mutagenesis. Mutation of the
ferredoxin interaction surface of
hydrogenase improves circuit
function. A.) Homology model of C.
reinhardtii hydrogenase with
mutated residues highlighted in
cyan and spinach ferredoxin X-ray
structure (1A703) with negatively
charged residues highlighted in
yellow. Iron-sulfur clusters and the
hydrogenase catalytic cluster are
highlighted in orange. B.) Relative
in vitro and in vivo hydrogen
production for wild type and
mutated C. reinhardtii hydrogenase.
Mutants D126K and E5K, which
enhance charge-complementarity at
the putative interaction surface,
show a specific enhancement of in
vivo activity relative to activity
changes seen in the ferredoxin-
independent in vitro assay.

58
reinhardtii, and ferredoxin from spinach, both of which are closely related to the proteins

studied in the in silico model (Appendix A, figure S2).

We found that two mutations in HydA1, D126K and E5K, improved in vivo

hydrogen production while in vitro these mutations showed less or no effect (figure

2.10B). As in vitro hydrogen production was closely correlated to hydrogenase expression

level (data not shown), the improvement in in vitro hydrogen production that was seen

for E5K may be the result of increased protein expression.

Improvement of hydrogen production activity through direct protein fusion

Many metabolic electron transfer pathways are insulated through the physical

scaffolding of protein components in the membrane, for example in the electron

transport chain of mitochondria or chloroplasts. Additionally, some electron transport

proteins themselves are built from combinations of modular electron binding domains,

including the hydrogenase from Clostridium29. We sought to isolate our synthetic

pathway and improve hydrogen production through physically linking the hydrogenase

and ferredoxin in order to increase the chance of binding and electron transfer between

the desired partners. We were able to show improved function of the artificial pathway

through genetic fusion of the hydrogenase and ferredoxin. Chimeras between ferredoxin

and various ferredoxin reductases have been shown to be functional in vitro31, with

improved electron transfer rates presumably due to the increased local concentration of

59
reduced ferredoxin. We fused the hydrogenase from C. acetobutylicum (figure 2.11A) with

ferredoxin from spinach (figure 2.11B & C) or from C. acetobutylicum (figure 2.11D) using

flexible glycine-serine linkers of various lengths.

Our hydrogenase-spinach ferredoxin chimeric proteins were all expressed at a high

level in E. coli (figure 2.11B) and active at identical levels in vitro (figure 2.11C),

regardless of the orientation of the fusion or linker length, consistent with the lack of a

requirement for ferredoxin in the in vitro assay. In vitro data for fusions with the C.

acetobutylicum ferredoxin followed a similar trend (data not shown). However, in vivo

activity of the fusion proteins when coexpressed with the PFOR from D. africanus

depended on linker length as well as overall configuration. Fusions of the spinach

ferredoxin to the hydrogenase C-terminus displayed decreased function when the linker

is very short, a nearly five-fold improvement at intermediate length linkers, and activity

falling gradually to the level of that of separate proteins at longer lengths. This length

dependent behavior is consistent with models of changes in local concentration of

reactants due to enzyme fusion51. Protein fusion to the C-terminus of the spinach

ferredoxin, whether the hydrogenase or a short protein tag abrogated in vivo

hydrogenase activity entirely (data not shown), likely due to the proximity of the

ferredoxin C-terminus to the iron-sulfur cluster31.

The behavior of the fusion protein is similar with the ferredoxin from C.

acetobutylicum, although the bacterial ferredoxin is equally active when fused at either

60
Figure 2.11 Increase in hydrogen production from the synthetic circuit through direct
protein fusion of spinach ferredoxin and C. acetobutylicum hydrogenase. A.) Schematic model of
the protein fusion, here showing the hydrogenase fused to the spinach ferredoxin N-terminus
(N-termini highlighted in blue, C-termini in green, and iron-sulfur clusters in orange). B.)
Hydrogenase-ferredoxin fusion proteins are highly expressed and are the predicted size for
the chimera, as indicated by western blotting with an anti-ferredoxin antibody. C.) Behavior of
fusion with spinach ferredoxin to hydrogenase C-terminus. D.) Behavior of fusion with
Clostridium ferredoxin.

61
terminus (figure 2.11C). When fused to the hydrogenase C-terminus, the linker-length

dependent activity displays similar characteristics to the spinach ferredoxin.

Interestingly, when fused to the hydrogenase N-terminus, in vivo activity with shorter

linker lengths is improved. The putative ferredoxin-binding region is on the N-terminal

domain of the hydrogenase (figure 2.11A), which includes all of the F-clusters that

transfer electrons from the surface to the active-site H-cluster. Fusion with a short linker

at the C-terminus makes it impossible to reach the N-terminal domain, resulting in the

decreased activity compared to unfused proteins. At the N-terminus, however, a short

linker still allows for interaction between the hydrogenase and its fused ferredoxin,

resulting in increased activity.

The effect of scaffolding ferredoxin and hydrogenase

In signal transduction, complex protein scaffolds isolate pathways by localizing

pathway components into a complex, directing the flow of information. These scaffolds

can be rewired in their natural contexts in eukaryotic cells in order to alter the output

behavior of signaling cascades 52. Synthetic, modular scaffold proteins have been

implemented in E. coli in order to direct flux in synthetic metabolic pathways, improving

pathway output by up to 77-fold 37. These synthetic scaffolds are built from modular

scaffold domains from eukaryotic signal transduction—PDZ, SH3, and GBD domains—

which tightly bind cognate ligand proteins that can be fused to any protein of interest

62
for localization to the scaffold. We imported these scaffold designs into our synthetic

pathway and found that scaffolding of the hydrogenase and ferredoxin dramatically

affected the function of the pathway.

Scaffolding of metabolic pathways with small-molecule intermediates can lead to a

“pipeline” effect, where the increase in the local concentration of upstream intermediates

can speed up the reaction. This can significantly affect the pathway output, particularly

when the chemical intermediates of the reaction pathway are toxic to the cell and

increased throughput can lead to cell death if the intermediate is not rapidly converted

by the next enzymatic step in the pathway37. Instead of small molecule intermediates,

our synthetic pathway relies on protein-protein interactions, as is the case in many signal

transduction cascades. By channeling electron transfer through scaffolded interactions,

the flux through the synthetic circuit can potentially be significantly increased in an

insulated manner. Because the tertiary structures of the synthetic scaffolds have not

been determined, however, it is also possible that the requirement of protein-protein

interaction for pathway function may lead to a decrease in pathway flux due to non-

optimal insulation of interacting partners. We sought to characterize the effects of

different scaffold structures, ligands, and linker lengths on the function of the synthetic

PFOR-ferredoxin-hydrogenase circuit with ferredoxin and hydrogenase localized to the

synthetic scaffold. All experiments were performed using the PFOR from D. africanus,

and ferredoxin and hydrogenase from C. acetobutylicum.

63
 The scaffold is an artificial protein built up of several modular peptide binding

domains. The GBD binding domain is at the N-terminus, followed by the SH3 domain

and the PDZ domain at the C-terminus. The number of domains in each case is variable,

and in Dueber et. al., the ratio of the domains to one another made significant differences

in flux depending on the stoichiometry of the reactions in the synthetic pathway37.

Although multiple diffusion-limited metabolic pathways could be enhanced using this

design37, artificially scaffolded redox pathways have not yet been investigated. While we

were primarily interested in exploring how different configurations of scaffold domains

and linker lengths affected the interaction of the redox proteins and therefore hydrogen

output, we measured a three-fold improvement of hydrogen production in the scaffolded

vs. non-scaffolded conditions when all of the proteins were expressed off of the lower

activity pTet promoter, leading to a decrease in the absolute value of hydrogen

production when compared to the Duet vectors (data not shown).

The ratio of the different scaffold domains, the ligand bound to the pathway

components, and the length of the linker between the ligand and the ferredoxin protein

all had significant effects on the output of the synthetic circuit (figure 2.12). Because

ferredoxin and hydrogenase need to physically interact for the circuit to function,

suboptimal configurations for the protein scaffold could conceivably sequester these

proteins from one another. Indeed, we found that hydrogen output was decreased when

the hydrogenase and ferredoxin were bound farther from one another along the length of

64
scaffold (figure 2.12A). Furthermore, the length of the linker connecting ferredoxin with

the SH3 ligand also significantly affected the ability of the hydrogenase and ferredoxin

to interact while bound to the scaffold. Increasing the linker length from five amino acids

! '
456789:5,.*,;4<=>?89<@

&" 0'1 -./ 213 &"

A?7BB<6=

#$% " #$% " +# *# &#

.*7A5
C1,
-./,,,,,0'1 +#,,,,,*#,,,,&#
B5445=<D9@ C1,",E9@=9@F,=<G79@,
A?7BB<6=,;<A989<@ 69@H54,65@F8I,JK77L,

( 1
0'1 -./ 213
&"
456789:5,.*,;4<=>?89<@

&)*

&)&
&"
#$% "
*)&

+)&

&)* &)& *)& +)& &)& *)& /)& +)&

C1).*7A5,4789< C1).*7A5,4789<

Figure 2.12 Effect of artificial scaffolding configuration on hydrogen production from the
synthetic circuit. A) Positional effects of ferredoxin targeting to artificial scaffold on hydrogen
production. B) Circuit efficiency is dependent upon length of flexible linker connecting ferredoxin
(FD) to scaffolding. C) Modulation of ferredoxin to hydrogenase ratio on scaffold affects hydrogen
production, with decreasing yield observed at higher ferredoxin:hydrogenase ratios. D.) Direct
fusion of ferredoxins to one another yields diminishing hydrogen production with increased
numbers of fused ferredoxin proteins.

65
to twenty led to a 3-5 fold increase in hydrogen output from the scaffolded circuit (figure

2.12B). Linkers of intermediate length produced intermediate pathway output.

The ratio of GBD, SH3, and PDZ domains that made up the synthetic scaffold also

significantly affect the function of the pathway in some cases. The stoichiometry of the

hydrogen production reaction requires two ferredoxins for the reduction of a single

hydrogen molecule, so it might be expected that scaffolds that localize more ferredoxin

molecules will increase flux through the hydrogenase. Unfortunately, however, PFOR is

substrate limited53, with increasing concentrations of ferredoxin leading to decreased

enzymatic activity. Increasing the ratio of ferredoxin:hydrogenase bound to the scaffold

decreases the output from the synthetic pathway (figure 2.12C). This effect was also seen

when ferredoxin genes were fused in tandem without a scaffold, with output from the

hydrogen production circuit steadily dropping with each added ferredoxin (figure 2.12D).

This strategy may, however, increase pathway flux in other synthetic electron transfer

pathways when the electron donor is not substrate limited.

Latent function for the N-terminal ferredoxin-like domain of the hydrogenase

Insulation of the synthetic electron transfer pathways helps ensure high activity

of the hydrogen production pathway, while deletion of competing reactions prevents

short-circuiting of the suflite reductase pathway. With background anaerobic growth of

66
the cysI deletion strain thus eliminated, it is possible to finally test whether the

hydrogenase can provide reducing equivalents to ferredoxin through the consumption of

hydrogen gas. Indeed, in the full knockout strain, growth was hydrogenase dependent

and oxygen sensitive, with growth decreasing rapidly as oxygen in the controlled

atmosphere was increased, and was almost entirely halted at 10% oxygen (figure 2.13A).

However, while growth was hydrogenase dependent it was not hydrogen dependent,

and similar oxygen sensitive growth was observed both in the presence and in the

absence of hydrogen in the atmosphere (figure 2.13A). The hydrogenase, rather than

reducing ferredoxin with electrons from the breakdown of hydrogen is somehow

mediating electron transfer from the E. coli electron pool to the plant ferredoxins

through an unknown cofactor or enzyme. While the [FeFe]-hydrogenase from T.

maritima has been shown to synergistically accept electrons from ferredoxin and

NADH54, no NADH oxidoreduction activity has been reported for the Clostridial

hydrogenases.

While the kinetics of inactivation in oxygen were different for the non-hydrogen

dependent growth, the two pathways are difficult to separate without knowledge of the

source of the cross-talk. We were interested in whether the oxygen sensitivity of the

hydrogen independent function was due to the oxygen sensitivity of the hydrogenase

itself or due to the transcriptional activation of E. coli electron transfer proteins in an

anaerobic environment. We mutated the hydrogenase from C. acetobutylicum and C.

67
saccharobutylicum through error-prone PCR and selected a large library of mutants in the

full knockout strain on minimal media in 10% oxygen. Many colonies grew in this

selection, and the hydrogenase carrying plasmids were able to complement the pathway

in naive cells, indicating that the oxygen sensitivity is hydrogenase based.

Of the 110 nonsilent mutations identified in 23 plasmid-linked mutants that passed

the selection, the majority eliminated catalytic activity of the hydrogenase in the

Figure 2.13 Hydrogenase dependent and hydrogen independent growth of the suflite reductase strain
uncovers latent function of the hydrogenase ferredoxin-like domain. A.) Growth of the knockout
strain can be complemented anaerobically with the hydrogenase from C. acetobutylicum (caHydA) or C.
saccharobutylicum (csHydA) both in the presence (red line) and absence (blue line) of hydrogen in the
atmosphere. As the hydrogenase is inactivated by increasing levels of oxygen, growth decreases in
both cases, with different kinetics. B.) Selection of oxygen tolerant mutants in minimal media and 10%
hydrogen yields a panel of mutations that by and large eliminate hydrogenase catalytic function,
including several truncations before the catalytic domain. Tha majority of missense mutations cluster
on the surface of the proteins.

68
hydrogen production direction, both in vivo and in vitro. Nine of the mutations were

premature truncations of the protein, and four were truncated before the catalytic site.

These data indicate that the N-terminal ferredoxin-like electron transfer domain of the

hydrogenase alone is able to link E. coli electron metabolism with plant ferredoxins. The

mutations also predominantly cluster on the surface residues of the proteins, indicating

that perhaps the mutations alter how the hydrogenase is interacting with E. coli electron

transfer proteins (figure 2.13B).

Discussion

Metabolic engineering to produce biofuels must necessarily involve the redirection

of reducing equivalents into the fuel molecule and away from other cellular metabolites.

In cells, reducing equivalents are primarily stored in iron-sulfur cluster proteins and in

small molecules such as NADPH, NADH, and FADH2. While the small molecules can

freely diffuse through the cell and interact with a wide variety of enzymes, iron-sulfur

proteins can be isolated through the techniques of metabolic and protein engineering. In

the experiments described here, we applied four approaches to controlling electron flow

out of the iron-sulfur cluster protein ferredoxin: deletion of potential interaction

partners, enhancement of interaction by engineering of a protein surface, and increasing

the local concentration of interacting proteins using a flexible peptide linker or

attachment to a scaffold protein.

69
 These insulation approaches can increase flux through the synthetic electron

transfer pathway, increasing hydrogen production and preventing short circuiting,

ensuring flux only through the pathway. The hydrogen production pathway consists of

the proteins pyruvate-ferredoxin oxidoreductase (PFOR), ferredoxin, and a hydrogenase

(expressed in the presence of hydrogenase maturation factors). This pathway produces a

theoretical maximum of two molecules of hydrogen per input glucose, and still allows

acetyl-CoA production from pyruvate. We characterized the relative efficacy of hydrogen

production using various combinations of PFOR, ferredoxin, and hydrogenase molecules

from different species, and found that PFOR from D. africanus in combination with

ferredoxin and hydrogenase from C. acetobutylicum was the most active pathway, predicted

in part by previous in vitro data55. In the reverse direction, cell growth was coupled to

flux through the synthetic pathway through deletion of the NADH dependent E. coli

sulfite reductase, cysI, complemented with a ferredoxin dependent pathway consisting of

sulfite reductase from Z. mays and plant-type ferredoxin. Reducing equivalents were

provided to ferredoxin via NADPH through coexpression with ferredoxin-NADPH

reductase (FNR) or through the [FeFe]-hydrogenase.

To direct electron flow between ferredoxin and hydrogenase, we first deleted genes

encoding six other proteins with which PFOR and/or ferredoxin might interact. Of

these, only deletion of ydbK, encoding a putative E. coli PFOR, resulted in enhanced

hydrogen production, and in the absence of the PFOR from D. africanus, deletion of ydbK

70
resulted in a decrease in the background level of hydrogen. This decrease in background

in the forward direction correlated to a significant drop in background anaerobic growth

of the cysI deletion strain expressing only zmSiR and ferredoxin. These results provide

further evidence that ydbK is a functional PFOR that can interact with a variety of

electron acceptors, particularly the ferredoxin from corn and spinach22.

As a converse approach, we addressed whether the levels of hydrogen production

could be enhanced by improving the binding between ferredoxin and hydrogenase in

vivo. In this case we used the C. reinhardtii hydrogenase, whose interaction with plant-

type ferredoxin has been computationally modelled by Long et al., who suggested

mutations that could enhance this interaction24. Several mutations that improve charge

complementarity between spinach ferredoxin and this hydrogenase were found to

enhance electron transfer between these proteins in vivo, as inferred by increased

hydrogen production. These results indicate that the activity of the hydrogenase is

limited, in part, by its ability to interact with ferredoxin; i.e. that the collision and

docking of these proteins is not effectively in excess.

Finally, we used two different methods to increase the relative local concentration

of ferredoxin and hydrogenase: direct fusion by a flexible glycine/serine-rich linker, and

indirect fusion by attachment of these proteins to interaction modules that bind to a

common scaffold. Each approach significantly improved hydrogen production in a

strongly configuration-dependent manner in vivo, as expected if the ferredoxin and

71
hydrogenase were primarily interacting in cis. For example, when the ferredoxin and

hydrogenase were attached by a linker too short to allow cis interaction, hydrogen

production was relatively low, but increased significantly when the linker was long

enough to allow interaction. As the attaching linker was further lengthened, hydrogen

production decreased gradually, consistent with the two proteins occupying a sphere of

increasing volume and decreasing relative concentration. As a tactic for metabolic

engineering, protein fusion and/or scaffolding is particularly useful with iron-sulfur

cluster proteins, because their electrons must be transferred protein-to-protein—no

small molecule carriers of reducing equivalents are generated that might diffuse away.

The iron-sulfur proteins in our synthetic circuit present a modular system, with

proteins from disparate species able to interact and transfer electrons. Such modular

systems are valuable for further synthetic biological manipulation and experimentation.

The synthetic pathways here present a relatively simple method for the analysis of

activities and electron transfer properties of hydrogenases, ferredoxins, SiR, FNR, and

PFOR genes from any number of species, or other engineered synthetic electron

transfer proteins. These in vivo data are a valuable complement to in vitro binding

constants and kinetic parameters of the enzymes and will be useful in further designing

and optimizing microbial systems for hydrogen production.

Such synthetic biological systems can also be used to better understand biological

electron transfer systems. The role of ferredoxins in E. coli metabolism is poorly

72
characterized, with ferredoxins performing many unknown but required functions in the

cell. Here we tested deletions of six iron-sulfur proteins expected to interact with

ferredoxins, many of which are previously uncharacterized. While only one gene

deletion ("ydbK) strongly affected the pathways, further combinatorial deletions may

affect hydrogen production in different ways, or may affect other synthetic electron

transfer pathways. Other deletions of iron-sulfur oxidoreductases and combinations

thereof may lead to a more complete understanding of electron transfer systems in the

E. coli cytoplasm, as well as the development of a host strain for expression of

heterologous electron transfer pathways for synthetic biology. Such a strain would have

to retain the ability to mature iron-sulfur clusters but limit the function of proteins that

can interact with ferredoxins and ferredoxin oxidoreductases to ensure optimal electron

flux through the synthetic pathway. Such specialized strains of E. coli may be optimized

for other types of synthetic pathway designs and may be better equipped for industrial

purposes than proposed “minimal” cells56, as they would retain many of the mechanisms

that allow for robust growth and protein expression.

As it stands, the insulation provided by combinatorial deletion of the six proteins

we identified in this study was sufficient to establish a genetic selection for flux through

the hydrogenase. However, while growth of the cysI deletion was hydrogenase

dependent, it was not hydrogen dependent, indicating that the hydrogenase itself, made

up of a recombinant fusion of smaller iron-sulfur proteins is able to transfer electrons

73
through interaction with the E. coli redox pool and and the plant ferredoxins of the

synthetic pathway. This latent function was also oxygen sensitive, and mutation and

selection in 10% oxygen selected for primarily non-catalytic hydrogenases, including

truncations before the catalytic domain, and identified a panel of surface mutations.

Further analysis of these mutations and experiments with the hydrogen

production pathway can also be used to further analyze protein-protein interaction

surfaces for electron transfer, including for mutagenic studies to determine the binding

surface on the Clostridial hydrogenases, which is poorly understood. An improved

understanding of the electron transfer surface between the hydrogenase and ferredoxin

would significantly affect our picture of how electron transfer pathways co-evolved;

whether specific ferredoxins evolved for interaction with specific enzymes or whether

electron transfer is regulated in other ways. Considering the ability of ferredoxins from

many distant species to interact with various hydrogenases, the impact of further

binding surface optimization may be negligible, or may require co-evolution of

complementary mutations on both binding partners to result in highly specific

interactions.

Iron-sulfur proteins are also uniquely suited to isolation techniques that involve

physical scaffolding. Electrons tunnel between proteins in close proximity, so direct

protein fusion improves the local concentration of electron transfer proteins and thus

improves the electron transfer rates. This is in contrast to other synthetic metabolic

74
pathways with small molecule intermediates, whose diffusion through the cellular

environment is much faster, limiting the potential improvement by protein fusion. This

method can be easily adapted to other electron transfer pathways in a modular,

extensible manner. Moreover, the dependence of hydrogenase activity upon scaffold

design and/or the links between the ligands and scaffold illustrates that synthetic redox

pathways can be coupled through interaction with a common adaptor protein in order to

modulate electron flux through the system. Unlike reactions with diffusible

intermediates, scaffolding of redox partners requires that the scaffold design allow

sterically unhindered interaction between bound protein to enable electrons to tunnel

between closely apposed iron-sulfur clusters. Incorrect design may tether redox partners

in a manner that constrains them and prevents electron transfer, as we observed when

hydrogenase and ferredoxin were bound to domains on distal ends of the scaffold.

Through attention to scaffold design, further optimization may significantly improve

hydrogen production through the synthetic circuit, as well as provide a template for

future scaffolding of other electron transfer pathways.

Biological hydrogen production is a promising and well-studied system for

sustainable energy production. The insulation approaches presented here are widely

applicable to other biological hydrogen systems, from in vitro enzymatic pathways17,

where protein fusions would likely improve kinetic rates, to methods for boosting natural

hydrogen production in heterotrophic57 and photosynthetic organisms. Photosynthetic

75
systems in particular may benefit from the insulation strategies presented here, as

ferredoxins are the primary electron carrier in photosynthetic organisms and

competition for electrons from other metabolic pathways is strong. Improvement of

protein binding in plant-type ferredoxins may be useful in such systems, as well as the

applicability of protein fusion or pathway scaffolding to a wide range of biological

systems.

Conclusions

Electron transfer systems such as our hydrogenase pathway are an untapped

resource for synthetic biology, which seeks to design biological pathways as predictably

as electronic circuits 60. Electrons are unique metabolites whose movement in biological

systems occurs by quantum-mechanical tunneling between protein-bound cofactors such

as iron-sulfur clusters. As a result, escape by diffusion into an aqueous phase is avoided,

offering distinctive opportunities for control. The circuit described here moves electrons

from higher to lower energy, while performing work in the form of hydrogen

production. The rationally constructed insulation of the pathway through elimination of

side reactions, interaction surface optimization, and protein fusion indicate that all three

methods are viable for synthetic circuit design and all strategies may play a role in the

evolution of complex isolated circuits in natural metabolism. This type of synthetic-

76
biological analysis may yield insights into natural mechanisms for controlling electron

flow, and may provide new approaches for metabolic engineering and bioenergy.

References

1! Benner, Steven A and Sismour, A Michael, Synthetic biology. Nat Rev

Genet 6 (7), 533 (2005).
2! King, Paul W, Posewitz, Matthew C, Ghirardi, Maria L, and Seibert,
Michael, Functional studies of [FeFe] hydrogenase maturation in an
Escherichia coli biosynthetic system. J Bacteriol 188 (6), 2163 (2006).
3! Binda, C et al., Structure of the mutant E92K of [2Fe-2S] ferredoxin I from
Spinacia oleracea at 1.7 A resolution. Acta Crystallogr D Biol Crystallogr
54 (Pt 6 Pt 2), 1353 (1998).
4! Filipovska, A and Rackham, O, Building a Parallel Metabolism within the
Cell. ACS Chem Biol (2008).
5! Han, Xue et al., Millisecond-Timescale Optical Control of Neural
Dynamics in the Nonhuman Primate Brain. Neuron 62 (2), 191 (2030).
6! Purnick, Priscilla E. M and Weiss, Ron, The second wave of synthetic
biology: from modules to systems. Nat Rev Mol Cell Biol 10 (6), 410
(2009).
7! Wong, T S and Schwaneberg, U, Protein engineering in
bioelectrocatalysis. Current Opinion in Biotechnology 14 (6), 590 (2003).
8! Page, C C, Moser, C C, Chen, X, and Dutton, P L, Natural engineering
principles of electron tunnelling in biological oxidation-reduction. Nature
402 (6757), 47 (1999).
9! Moulis, J M and Davasse, V, Probing the role of electrostatic forces in the
interaction of Clostridium pasteurianum ferredoxin with its redox partners.
Biochemistry 34 (51), 16781 (1995).
10! Veit, Andrea, Akhtar, M. Kalim, Mizutani, Taeko, and Jones, Patrik R,
Constructing and testing the thermodynamic limits of synthetic NAD(P)
H:H 2pathways. Microbial Biotechnology 1 (5), 382 (2008).
11! Coghlan, V M and Vickery, L E, Expression of human ferredoxin and
assembly of the [2Fe-2S] center in Escherichia coli. Proc Natl Acad Sci
USA 86 (3), 835 (1989).
12! Ghirardi et al., Hydrogenases and Hydrogen Photoproduction in Oxygenic
Photosynthetic Organisms (*). Annu Rev Plant Biol 58, 71 (2007).
13! Knoell, H E and Knappe, J, Escherichia coli ferredoxin, an iron-sulfur
protein of the adrenodoxin type. Eur J Biochem 50 (1), 245 (1974).

77
14! Baba, Tomoya et al., Construction of Escherichia coli K-12 in-frame,
single-gene knockout mutants: the Keio collection. Mol Syst Biol 2, 2
(2006).
15! Laurinavichene, T V and Tsygankov, A A, H2 consumption by Escherichia
coli coupled via hydrogenase 1 or hydrogenase 2 to different terminal
electron acceptors. FEMS Microbiol Lett 202 (1), 121 (2001).
16! Sawers, R G, Formate and its role in hydrogen production in Escherichia
coli. Biochem Soc Trans 33 (Pt 1), 42 (2005).
17! Ye, Xinhao et al., Spontaneous High-Yield Production of Hydrogen from
Cellulosic Materials and Water Catalyzed by Enzyme Cocktails.
ChemSusChem 2 (2), 149 (2009).
18! Aguirre de Carcer, Inigo, DiPasquale, Antonio, Rheingold, Arnold L, and
Heinekey, D Michael, Active-site models for iron hydrogenases: reduction
chemistry of dinuclear iron complexes. Inorganic chemistry 45 (20), 8000
(2006).
19! Yoshida, Akihito et al., Efficient induction of formate hydrogen lyase of
aerobically grown Escherichia coli in a three-step biohydrogen production
process. Appl Microbiol Biotechnol 74 (4), 754 (2007).
20! Maeda, Toshinari, Sanchez-Torres, Viviana, and Wood, Thomas K, Protein
engineering of hydrogenase 3 to enhance hydrogen production. Appl
Microbiol Biotechnol, 10 (2008).
21" Akhtar, M and Jones, P, Engineering of a synthetic hydF–hydE–hydG–
hydA operon for biohydrogen production. Anal Biochem 373 (1), 170
(2008).
22! Akhtar, M. Kalim and Jones, Patrik R, Construction of a synthetic YdbK-
dependent pyruvate:H2 pathway in Escherichia coli BL21(DE3). Metab
Eng 11 (3), 139 (2009).
23! Richard, D J et al., Transcriptional regulation in response to oxygen and
nitrate of the operons encoding the [NiFe] hydrogenases 1 and 2 of
Escherichia coli. Microbiology (Reading, Engl.) 145 ( Pt 10), 2903 (1999).
24! Long, H, King, P, Ghirardi, M, and Kim, K, Hydrogenase/Ferredoxin
Charge-Transfer Complexes: Effect of Hydrogenase Mutations on the
Complex Association. The journal of physical chemistry A (2009).
25! Chang, King, Ghirardi, and Kim, Atomic Resolution Modeling of the
Ferredoxin:[FeFe] Hydrogenase Complex from Chlamydomonas
reinhardtii. Biophys J (2007).
26" Coghlan, V M and Vickery, L E, Site-specific mutations in human
ferredoxin that affect binding to ferredoxin reductase and cytochrome
P450scc. J Biol Chem 266 (28), 18606 (1991); Lelong, C, Sétif, P,
Lagoutte, B, and Bottin, H, Identification of the amino acids involved in
the functional interaction between photosystem I and ferredoxin from
Synechocystis sp. PCC 6803 by chemical cross-linking. J Biol Chem 269
(13), 10034 (1994).
27! Skerker, J et al., Rewiring the Specificity of Two-Component Signal
Transduction Systems. Cell 133 (6), 1043 (2008).

78
28! Akashi, T et al., Comparison of the electrostatic binding sites on the
surface of ferredoxin for two ferredoxin-dependent enzymes, ferredoxin-
NADP(+) reductase and sulfite reductase. J Biol Chem 274 (41), 29399
(1999).
29! Peters, J W, Lanzilotta, W N, Lemon, B J, and Seefeldt, L C, X-ray crystal
structure of the Fe-only hydrogenase (CpI) from Clostridium pasteurianum
to 1.8 angstrom resolution. Science 282 (5395), 1853 (1998).
30! Seaton, B L and Vickery, L E, Expression of human ferredoxin in
Saccharomyces cerevisiae: mitochondrial import of the protein and
assembly of the [2Fe-2S] center. Arch Biochem Biophys 294 (2), 603
(1992); Davasse, V and Moulis, J M, Design and functional expression in
Escherichia coli of a synthetic gene encoding Clostridium pasteurianum 2
[4Fe-4S] ferredoxin. Biochem Biophys Res Commun 185 (1), 341 (1992).
31! Aliverti, A and Zanetti, G, A three-domain iron-sulfur flavoprotein obtained
through gene fusion of ferredoxin and ferredoxin-NADP+ reductase from
spinach leaves. Biochemistry 36 (48), 14771 (1997).
32! Guerrini, Olivier et al., Characterization of Two 2[4Fe4S] Ferredoxins from
Clostridium acetobutylicum. Curr Microbiol 56 (3), 261 (2008).
33! Hanke, Guy Thomas, Kimata-Ariga, Yoko, Taniguchi, Isao, and Hase,
Toshiharu, A post genomic characterization of Arabidopsis ferredoxins.
Plant Physiol 134 (1), 255 (2004).
34! Fukuyama, Keiichi, Structure and function of plant-type ferredoxins.
Photosyn Res 81 (3), 289 (2004).
35! Bayer, TS et al., Synthesis of methyl halides from biomass using
engineered microbes. J. Am. Chem. Soc 131 (18), 6508 (2009).
36! Haseltine, Eric L and Arnold, Frances H, Synthetic gene circuits: design
with directed evolution. Annual review of biophysics and biomolecular
structure 36, 1 (2007).
37! Dueber, John E et al., Synthetic protein scaffolds provide modular control
over metabolic flux. Nat Biotechnol 27 (8), 753 (2009).
38! Phillips, I and Silver, P, A New Biobrick Assembly Strategy Designed for
Facile Protein Engineering. dspace.mit.edu (2006).
39! Cironi, P, Swinburne, I. A, and Silver, P. A, Enhancement of Cell Type
Specificity by Quantitative Modulation of a Chimeric Ligand. Journal of
Biological Chemistry 283 (13), 8469 (2008).
40! Baba, Tomoya et al., Construction of Escherichia coli K-12 in-frame,
single-gene knockout mutants: the Keio collection. Mol Syst Biol 2,
2006.0008 (2006).
41! Guex, N and Peitsch, M C, SWISS-MODEL and the Swiss-PdbViewer: an
environment for comparative protein modeling. Electrophoresis 18 (15),
2714 (1997).
42! Boxma, Brigitte et al., An anaerobic mitochondrion that produces
hydrogen. Nature 434 (7029), 74 (2005).
43! Pieulle, L, Magro, V, and Hatchikian, E C, Isolation and analysis of the
gene encoding the pyruvate-ferredoxin oxidoreductase of Desulfovibrio

79
 africanus, production of the recombinant enzyme in Escherichia coli, and
effect of carboxy-terminal deletions on its stability. Journal of
Bacteriology 179 (18), 5684 (1997).
44" Akhtar, M. Kalim and Jones, Patrik R, Deletion of iscR stimulates
recombinant clostridial Fe–Fe hydrogenase activity and H2-accumulation
in Escherichia coli BL21(DE3). Appl Microbiol Biotechnol, 10 (2008).
45! Yonekura-Sakakibara, K et al., Analysis of reductant supply systems for
ferredoxin-dependent sulfite reductase in photosynthetic and
nonphotosynthetic organs of maize. Plant Physiol 122 (3), 887 (2000).
46! Iuchi, S and Weiner, L, Cellular and molecular physiology of Escherichia
coli in the adaptation to aerobic environments. J Biochem 120 (6), 1055
(1996).
47! McCord, J M, The evolution of free radicals and oxidative stress. Am J
Med 108 (8), 652 (2000).
48! Gou, Ping et al., Higher order structure contributes to specific differences
in redox potential and electron transfer efficiency of root and leaf
ferredoxins. Biochemistry 45 (48), 14389 (2006).
49! Zarrinpar, Ali, Park, Sang-Hyun, and Lim, Wendell A, Optimization of
specificity in a cellular protein interaction network by negative selection.
Nature 426 (6967), 676 (2003).
50! Chang, Christopher H, King, Paul W, Ghirardi, Maria L, and Kim, Kwiseon,
Atomic resolution modeling of the ferredoxin:[FeFe] hydrogenase complex
from Chlamydomonas reinhardtii. Biophys J 93 (9), 3034 (2007); Long, H
et al., Brownian Dynamics and Molecular Dynamics Study of the
Association between Hydrogenase and Ferredoxin from Chlamydomonas
reinhardtii. Biophysical Journal, 30 (2008).
51! Krishnamurthy, V et al., Dependence of effective molarity on linker length
for an intramolecular protein-ligand system. J Am Chem Soc 129 (5),
1312 (2007).
52! Bashor, C. J, Helman, N. C, Yan, S, and Lim, W. A, Using Engineered
Scaffold Interactions to Reshape MAP Kinase Pathway Signaling
Dynamics. Science 319 (5869), 1539 (2008); Park, Sang-Hyun, Zarrinpar,
Ali, and Lim, Wendell A, Rewiring MAP kinase pathways using alternative
scaffold assembly mechanisms. Science 299 (5609), 1061 (2003).
53! Uyeda, K and Rabinowitz, J C, Pyruvate-ferredoxin oxidoreductase. IV.
Studies on the reaction mechanism. J Biol Chem 246 (10), 3120 (1971).
54! Schut, G. J and Adams, M. W. W, The Iron-Hydrogenase of Thermotoga
maritima Utilizes Ferredoxin and NADH Synergistically: A New
Perspective on Anaerobic Hydrogen Production. J Bacteriol, 27 (2009).
55! Pieulle, L et al., Structural and kinetic studies of the pyruvate-ferredoxin
oxidoreductase/ferredoxin complex from Desulfovibrio africanus. Eur J
Biochem 264 (2), 500 (1999).
56! Forster, Anthony C and Church, George M, Towards synthesis of a
minimal cell. Mol Syst Biol 2, 10 (2006).

80
57! Maeda, Sanchez-Torres, and Wood, Enhanced hydrogen production from
glucose by metabolically engineered Escherichia coli. Appl Microbiol
Biotechnol (2007).
58! Zhu, Y, Eiteman, M A, Altman, R, and Altman, E, High Glycolytic Flux
Improves Pyruvate Production by a Metabolically Engineered Escherichia
coli Strain. Appl Environ Microbiol 74 (21), 6649 (2008).
59! Penfold, David W, Sargent, Frank, and Macaskie, Lynne E, Inactivation of
the Escherichia coli K-12 twin-arginine translocation system promotes
increased hydrogen production. FEMS Microbiol Lett 262 (2), 135 (2006).
60! Andrianantoandro, Ernesto, Basu, Subhayu, Karig, David K, and Weiss,
Ron, Synthetic biology: new engineering rules for an emerging discipline.
Mol Syst Biol 2, 14 (2006).

81
Chapter 3
/ Towards a Synthetic Chloroplast

“"Life did not take over the globe by combat, but by networking.”
-Lynn Margulis and Dorion Sagan

82
Background3

In eukaryotes, electron transfer pathways such as those engineered in the

previous chapter are primarily relegated to the energy harnessing endosymbiotic

organelles, the mitochondria1, hydrogenosomes2, and chloroplasts3. In each organelle,

complex electron transfer pathways in the matrix and membranes create energy for the

cell, through electrons captured from sugars or from sunlight. Endosymbiotic in origin,

there has even been a recent push to classify mitochondria as actually bacterial rather

than a part of the eukaryotic whole4. Without these ancient bacterial relationships, the

eukaryotic cell is impossible, and it is impossible to know what synthetic endosymbiotic

relationships will be able to accomplish when synthetic biology can harness the creative

forces of evolutionary change.

Recent advances in synthetic biology have moved towards engineering of

symbiotic microbial consortia5 and taking advantage of invasive behaviors for

engineering. Invasive bacteria cause several deadly infectious diseases in humans, caused

by intracellular pathogens such as Yersinia, Listeria monocytogenes, and enteroinvasive E.

coli6. Recent work in biological engineering and synthetic biology has focused on the

development of non-infectious but invasive and deadly bacteria that target and destroy

only specific cell types for disease treatment, particularly cancer7,8, or for delivery of

peptide9 or nucleotide based vaccines10 and RNA interference gene therapy11.

3*Work presented in this chapter was published in Agapakis CM, Niederholtmeyer H, Noche RR, Lieberman
TD, Megason SG, Way JC, Silver PA. “Towards a Synthetic Chloroplast.” PLoS ONE, 2011, 6(4):e18877
(Attached as Appendix E).

83
 There is a small but fascinating literature on engineering beneficial symbiosis

between photosynthetic microorganisms or isolated chloroplasts. Nutritional and gas

exchange between mammalian tissue culture and algae has been observed12 and

explored as a method for replenishing nutrients in culture, creating an artificial

ecosystem. Chloroplasts isolated from algae can be naturally incorporated into the

digestive system of the sea slug Elysia chlorotica, allowing it to survive for months

through photosynthesis13. Isolated chloroplasts have been shown to divide in an

artificial environment14 and enter the cells of higher plants15 and even mammalian

cells16.

We aimed to re-enact the establishment of endosmbiosis between larger cells

and free-living photosynthetic bacteria that precipitated the evolution of chloroplasts

and the eukaryotic algae. We explored three routes into the cell (figure 3.1). First,

microinjection of the cyanobacteria Synechococcus elongatus into zebrafish embryos

allowed us to study in vivo dynamics of the bacteria in a whole animal model. Second,

we engineered S. elongatus with invasin from Y. pseudotuburculosis (inv) and listeriolysin

from L. monocytogenes. Expression of the virulence factor invasin in non-invasive

bacteria is sufficient to allow for uptake of the bacteria by a mammalian cell into the

endosome. Expression of listeriolysin allows for bacterial escape from the endosome

into the mammalian cytoplasm. Listeriolysin also causes cells endocytosed by the

84
Figure 3.1 Three paths to endosymbiosis used in this study. A.) Direct microinjection of S. elongatus into
zebrafish embryos allow exploration of in vivo dynamics of bacteria inside animal cells. B.) Invasion of
mammalian cells through heterologous expression of invasin and listeriolysin O. C.) Phagocytosis of
bacteria by macrophages. Bacteria subsequently escape from the endosomal compartment through
expression of listeriolysin O.

mammalian macrophage to escape digestion, and we observed that engineered S.

elongatus could replicate in the mammalian macrophage cytoplasm.

Macrophages can take up and phagocytose many different species of bacteria.

However, most species of bacteria, including many pathogens, are unable to replicate in

the cytoplasm of mammalian cells, and the precise mechanism of growth inhibition is

unknown and a matter of controversy17. In contrast, non-pathogenic Bacillus subtilis

expressing heterologous hemolysin has been shown to escape phagosome digestion by

macrophages and divide in the mammalian cytoplasm18. However, microinjection studies

have found that only those species that naturally divide in the cytoplasm were able to

replicate, with even intravacuolar pathogens unable to divide19. To our knowledge, such

85
experiments have not been attempted with photosynthetic bacteria or other

autotrophs.

Relationships between species can create enormous adaptive, evolutionary

changes in behavior and structure, driving the creation of new species and new

kingdoms. Importantly this change can happen on the scale of several years, not several

million. In a fascinating recent example of serendipitous engineering, amoeba infected

with a naturally occurring parasitic bacteria and carefully cultivated over several years

eventually became dependent on the bacterial symbiont, showing that endosymbiosis

can be established quite rapidly under the right conditions20. The symbiosis of the

originally harmful pathogen with its host in a sufficient environment drove changes in

both the host and the symbiont, creating an new endosymbiotic whole that couldn’t be

divided again. An understanding of the processes that drive these symbiotic changes in

the context of synthetic biology will be crucial to engineering multi-species synthetic

biology systems greater than the sum of their parts.

Materials and Methods

Cells and media

E. coli DH5" was used for all plasmid manipulation using standard procedure. S.

elongatus PCC 7942 was cultured in BG-11 medium 21 at 30°C and illuminated by strong

86
light. CHO and J774 cells were maintained using standard procedure in F-12 medium

(Invitrogen) for CHO cells and RPMI 1640 medium (Invitrogen) for J774 cells. All

media contained L-glutamine and were supplemented with 10% FBS (HyClone) and 1%

Penicillin/Streptomycin Mix (Invitrogen). For culturing cells during infections outside

of the controlled 5% CO2 atmosphere, Leibovitz’s L-15 medium without phenol red

(Invitrogen) was used for all cell lines, supplemented with 10% FBS for all cell types and

0.069 mg/ml proline for CHO cells.

Plasmids and DNA construction

The invasin gene from Yersinia pseudotuburculosis was subcloned from the pAC-

TetInv plasmid 7 provided by Chris Voigt (University of California, San Franscisco)

and listeriolysin was amplified from Listeria monocytogenes genomic DNA provided by

Heather Kamp (Harvard Medical School, Boston MA). Invasin DNA was amplified with

primers adding a SpeI site upstream and NotI and XbaI sites downstream Listeriolysin

was amplified with primers adding a SpeI site, a ribosome binding site, and a short

spacer for cloning downstream of invasin and NotI and XbaI site downstream. Primer

sequences are provided in Appendix A. Invasin and listeriolysin were then sequentially

subcloned into the pNS3 vector for homologous recombination into Synechococcus

neutral site 3 22.

The pNS3-invllo vector was incubated overnight in the dark with a culture of S.

elongatus PCC 7942 cells washed in 10mM NaCl, and integration into the neutral site

87
was selected using BG11 plates containing 1.5% Noble Agar and 12.5 !g/ml

chloramphenicol. Expression of invasin and listeriolysin was induced with 100!M

IPTG for 24 hours.

Zebrafish injection

Zebrafish embryos in the one-cell stage were injected with a solution containing

mRNA for expression of membrane GFP (mGFP) and bacteria. Needles were pulled on

a Sutter P2000 laser needle puller from Drummond glass capillary tubes. Eggs were

injected with 2.3 nl of injection solution using a!Nanoject. The injection solution

consisted of!injection !buffer (50 mM NaCl, 1 mM Tris pH!8, 0.1 mM EDTA and 0.1%

Phenol Red) and contained 40 ng of mGFP mRNA and 1 !l of a dense bacterial

suspension!per 10 !l of injection!solution. The bacterial suspension (S. elongatus or E.

coli) was prepared by spinning down 1 ml of an overnight E. coli culture or a dense

24-48h old, exponentially growing S. elongatus culture. The cells were resuspended in 1

ml of injection buffer without Phenol Red and again pelleted. The supernatant was

removed; the cells were mixed and always used fresh for the injections.

Embryos were raised in egg water (0.3 g/L Instant Ocean, 75 mg/L

CaSO4)!slightly !shaded from the light in the cyanobacterial incubator at 30°C. Egg

water was changed as needed. To follow individual embryos over time, the!embryos

were!separated from !each other in 12‐well plates.

88
 Development of the embryos injected with bacteria was monitored with a

fluorescence dissecting-microscope. For confocal imaging the embryos were

dechorionated and placed in imaging molds made from 1% (w/v) agarose!in egg water.

Mounted embryos were imaged in an upright Zeiss LSM 710 confocal microscope.The

embryo was!submerged in eggwater containing 1X Tricaine solution (10X Tricaine

solution: 0.1% (w/v) Tricaine and 10 mM Tris in egg water adjusted to pH 7 with

NaOH) for anesthetization.

Mammalian cell invasion assay

For infections of mammalian cells with bacteria, induced bacteria were washed

and transferred from their culture medium into PBS (137 mM NaCl, 2.7 mM KCl, 10

mM Na2HPO4, 2 mM KH2PO4). Bacterial suspensions in PBS were set to the same OD

and 100 !l of this suspension were added per 2 ml of cell culture medium per well of

12-well tissue culture dishes containing the mammalian cells. L-15 medium during the

infection did not contain antibiotics. After the treatment of the cells with S. elongatus for

3 hours to overnight, the cells were washed with PBS three times and the medium was

replaced by L-15 containing 100 !g/ml gentamicin, an antibiotic that does not cross the

mammalian cell membrane. During and after infections of the mammalian cells with

bacteria, the cultures were kept at 30°C in atmospheric CO2 levels. For S. elongatus, cells

were illuminated with fluorescent lamps from both sides of the tissue culture plate.

89
 For time-course of S. elongatus infection in macrophages, 10,000 J774 cells were

plated per well of a 96-well plate in L-15 media and were allowed to attach to the

bottom overnight at 37°C in the dark. Following attachment, 10 microliters of wild

type or inv/llo engineered S. elongatus diluted to OD750 of 0.025-0.4 in PBS were

added to each well and incubated at 30°C in the light for six hours. Each well was then

washed in PBS and the media was replaced with L-15 containing 100 !g/ml gentamicin

and plates were incubated at 30°C in the light. One plate was removed every 24 hours

and fixed in 3% paraformaldehyde, cells were permeablized in 0.01% Triton-X in PBS

and stained with DAPI. Plates were stored at 4°C in the dark and imaged at the same

time using fluorescence microscopy.

FACS analysis of mammalian cells with intracellular bacteria

After 24 hour infection, CHO cells were washed in PBS, trypsinized, and

resuspended in FACS buffer (PBS supplemented with 1%FBS). Cells were sorted with a

BD FACSAria cell sorter based on red channel fluorescence. Cells positive for red

fluorescence were gently reattached to glass-bottomed tissue culture dishes with

concanavalin A and imaged with confocal microscopy.

90
Results

Zebrafish embryos injected with S. elongatus hatch and thrive

Photosynthetic

symbiosis exists in several

underwater species, such as

the sea slug Elysia

chlorotica, which

incorporates the

chloroplasts from algae that

it feeds on into the cells of

its intricately branched

digestive system, allowing

it to survive for months

photoautotrophically 13.

While such a complex

symbiotic relationship likely

Figure 3.2 Tracking intracellular S. elongatus through
zebrafish development. Two photon microscopy images of the evolved over millions of
anterior of the zebrafish embryo at A.) Day 1 post injection,
B.) Day 2, C.) Day 3, D.) Day 4, and dissecting microscope years, we were interested in
images of embryos E.) Day 8, F.) Day 12 post injection.
Zebrafish cell membranes are outlined in green, with red
autofluorescent bacteria visible in cells throughout the replicating the first step of
embryo, including the eye (yellow arrows) and brain (white
arrows). Red autofluorescence gradually decreased over the an underwater
course of experimental observations, but remained visible in
the brain of the young zebrafish even after 12 days.

91
photosynthetic symbiosis and exploring the in vivo dynamics of photosynthetic bacteria

in a developing animal. We chose zebrafish embryos as they are easy to microinject, well

studied, and are clear, allowing light to penetrate.

Up to ten million bacteria were injected into zebrafish embryos at the single cell

stage to track the relationship between the vertebrate and bacterial cells through

development. Red autofluorescent bacteria were found intracellularly throughout the

embryo during development, including throughout the brain and even the lens (figure

3.2A-D) with no discernible phenotype. Synechococcus survived inside the embryo’s

cells for up to twelve days (figure 3.2F), at which time the experiment was terminated as

the fish began to develop pigment that

would block light to the intracellular

bacteria.

In stark contrast, injecting E. coli

cells killed the embryo within two hours

(figure 3.3). This rapid death occurred

even when the E. coli were UV killed

prior to injection (figure 3.3C) and when

the Lipid A production was attenuated by
deletion of msbB (figure 3.3D), a
modification shown to decrease incidence
Figure 3.3 Zebrafish embryos are immediately
killed by E. coli. A.) Zebrafish embryo two hours
after injection of S. elongatus. Cells appear red
due to phenol red injection. B.) Injection of E.
coli led to drastic morphological changes in the
embryo after two hours, and this change was
observed with E. coli cells that were C.) UV
killed, or D.) #msbB mutants.

92
of septic shock by tumor-targeting Salmonella23. These data point to other surface

markers that can cause the death seen in the E. coli injected embryos and a benign role

for the non-pathogenic S. elongatus in vivo.

Synechococcus expressing invasin and listeriolysin invade mammalian cells

We also sought to explore a more physiological model of intracellular invasion

than direct microinjection. The bacterial virulence factors encoding mammalian cell

invasion—invasin from Yersinia pseudotuburculosis24—and escape from the lysosomal

compartment—listeriolysin O from Listeria monocytogenes25 have been identified, cloned,

and shown to confer invasive behavior to non-pathogenic bacterial species. We inserted

invasin and listeriolysin as a tandem operon in S. elongatus neutral site 322 and incubated

induced S. elongatus cells with CHO cells at 50%-80% confluence overnight at 30°C in

bright light.

While expression of invasin alone is sufficient for high efficiency invasion by E.

coli for multiple mammalian cell types in culture (HeLa, U2OS, CHO7), expression of

both invasin and listeriolysin was required for invasion of CHO cells by S. elongatus, a

result previously reported for engineered E. coli cells invading liver cancer cells11. The

invasion efficiency was such that 4.8% of mammalian cells were positive for the red

channel autofluorescence from intracellular photosynthetic bacteria as measured by

fluorescence-activated cell sorting (FACS) analysis (figure 3.4A). Sorted cells were

93
imaged with confocal microscopy (figure 3.4B), confirming intracellular localization

with approximately one bacterial cell per CHO cell.

Figure 3.4 Invasion of CHO

cells. A.) S. elongatus
engineered with invasin and
listeriolysin are able to
invade CHO cells at a higher
efficiency than S. elongatus
harboring the empty vector
or invasin alone. Cells
positive for red fluorescence
were sorted by FACS and B.)
observed under confocal
microscopy, showing
intracellular localization of
at least one bacterial cell per
CHO cell in the majority of
the cells observed.

Replication inside mammalian macrophages

Bacteria can also enter cells through phagocytosis, and escaping digestion by the

lysosome is a prerequisite for pathogenic and symbiotic growth. Macrophages are a

crucial part of the mammalian immune system, seeking out, engulfing and digesting

foreign bodies and bacteria. The immortal mouse macrophage cell line J774 will quickly

engulf large numbers of bacterial cells in culture. We therefore incubated plates of 50%

confluent macrophages with varying concentrations of bacterial cells for one hour at

37°C for E. coli and six hours at 30°C for S. elongatus. As with zebrafish embryos,

engulfed E. coli cells will quickly kill their host macrophages (figure 3.5A), while even

94
 Figure 3.5 E. coli and S.
elongatus differentially affect
macrophages after
phagocytosis. Large scale
granulation is observed
when macrophages take up
E. coli that is A.) not
expressing llo and to an even
greater extent with B.) E.
coli expressing llo off of the
inducible lac promoter of the
pNS3 vector. In contrast,
macrophages displayed
similar morphology two days
after infection with C.) empty
vector S. elongatus, D.) S.
elongatus expressing inv and
llo, and E.) macrophages
untreated with bacteria but
maintained at 30$ in bright
light for two days.

high numbers of S. elongatus cells will remain inside J774 for

several days with relatively little effect.

E. coli highly expressing listeriolysin were observed to kill macrophages faster

than wild type E. coli (figure 3.5B). However, after two days of incubation with

Synechococcus, similar levels of cell death were observed in macrophages with

Synechococcus with only the empty vector integrated (figure 3.5C), those expressing

95
invasin and listeriolysin (figure 3.5D), and with macrophages without any bacteria

(figure 3.5E).

Non-pathogenic bacteria that have been engineered with listeriolysin O to escape

the macrophage endosome have been shown to replicate in the cytoplasm18. However,

there are many factors in the mammalian cytoplasm speculated to be involved in

preventing bacterial growth, a fact suggested by the extremely small number of

intracellular pathogens able to divide in the presumably nutrient-rich cytoplasm 17. S.

elongatus requires little external metabolic input and grows at a relatively fast rate at

intracellular carbon dioxide concentrations (one division every 8-12 hours). In addition,

as we have shown (above) S. elongatus has a special relationship to eukaryotic

antimicrobial systems as it is able to peacefully coexist with animal cells. As such, it is

expected that S. elongatus engineered with listeriolysin O will be able to divide in the

mammalian cytoplasm. In the dark, S. elongatus phogocytosed by macrophages will

rapidly lose red channel autofluorescence over the course of 12 hours, indicating death

(figure 3.6A). In the light, wild type S. elongatus autofluorescence will more slowly

decrease in intensity over several days (figure 3.6B, top row). S. elongatus engineered

with invasin and listeriolysin, able to escape lysosome digestion, showed marked

increase in autofluorescence in the first two days post-infection, with the number of

autofluorescent bacteria decreasing only after 3 days (figure 3.6B, bottom row).

96
Figure 3.6 S. elongatus can grow inside the macrophage cytoplasm. A.) Time lapse microscopy of
macrophages infected with +inv+llo S. elongatus kept in the dark shows the gradual decrease in red
autofluorescence over the course of 12 hours. In contrast, when kept in the light, B.) empty vector S.
elongatus autofluorescence is observed to gradually decrease over the course of several days (top row), while a
significant increase in density of red S. elongatus autofluorescence was observed in macrophages infected with
inv llo S. elongatus for two days post-infection (bottom row). This fluorescence was observed to decrease after
the third day of infection. C.) This change in fluorescence over time can be quantified as a change in
background subtracted mean fluorescence in ImageJ and averaged over replicates. Empty vector (blue line)
and +inv+llo S. elongatus (red line) show marked differences in growth when infected at similar densities of
1-2 bacterial cells per macrophage. D.) +inv+llo S. elongatus displayed infection density dependent growth
rates in macrophages. Each line shows change in mean fluorescence in cells infected at a single starting
density, ranging in multiples of two from fewer than one cell per macrophage to 1-4 bacteria per
macrophage. E.) Macrophage cell counts were variable across replicates and over the course of the
experiment but displayed no significant difference between macrophages infected with empty vector S.
elongatus at low (green line) or high density (blue line), or +inv+llo S. elongatus at low (red line) or high
(yellow line) density. ( F.) When infected at low density of fewer than one bacteria per macrophage, S.
elongatus division was observed in over 18 hour time-lapse fluorescent microscopy in approximately 1% of
macrophages observed, in particular those cells that contained more than one bacterial cell due to stochastic
fluctuations.
97
 The rate of division in the macrophage cytoplasm was quantified for varying

densities of S. elongatus in 96 well plates. Mean, background subtracted fluorescence

was averaged for triplicate infections. At similar starting density, there is marked

contrast between empty vector (blue) and +inv+llo S. elongatus (figure 3.6C). Rates of

division were correlated to S. elongatus infection densities. At the lowest concentrations,

with fewer than one bacterial cell per macrophage, the engineered strain is digested

more slowly than wild type S. elongatus, but does not show large-scale evidence of

division, but as infection is density is doubled, the rate of growth increases and begins

to level off at the highest density (1-4 bacteria per macrophage, figure 3.6D).

Differences in S. elongatus growth did not correlate with decrease in macrophage cell

counts over time, which remained variable but consistent between infection densities

over time (figure 3.6E). Even at the lowest infection density, +inv+llo S. elongatus

division was observed in approximately 1% of cells tracked with time lapse microscopy

(figure 3.6F).

Discussion

Complex relationships between many different species of organisms characterize

the biological world, but the details of these symbiotic relationships have proven

difficult to untangle through reductionist experimentation. Simplified, engineered multi-

species relationships can provide a framework for studying natural symbiotic

98
relationships 26. We show that photosynthetic bacteria can be engineered to invade and

divide inside mammalian cells for use as a platform for further engineering or study of

evolutionary dynamics of endosymbiosis.

Natural endosymbiosis between photosynthetic organisms and animal species

occurs in many marine species such as corals and sponges, whose simple body plans and

high surface-to-volume ratios make such associations valuable27.While these marine

photosymbioses have been studied for many years, the first evidence of a facultative

photosynthetic endosymbiosis was only recently discovered between the embryo of the

spotted salamander and green algae28. Algal-salamander associations had previously

been observed extracellularly29, with gas exchange between the algae and salamander

shown to be beneficial but not required for the developing embryo30. These newly

discovered intracellular interactions occur only in the embryo, with algae dying by the

time the larvae begins to feed and no evidence of vertical transmission from the

underground-dwelling adult salamander.

Such natural events show how rare bacterial-vertebrate endosymbiosis is, as well

as how benign photosynthetic associations can be when they are established. We used a

synthetic approach to developing photosynthetic associations with animals cells, finding

that injecting S. elongatus into the zebrafish embryo does not affect fish development. As

in the natural endosymbiosis, the photosynthetic cells slowly died, but remained in the

animal for several weeks.

99
 There are no known mammalian endosymbioses, and the mammalian

cytoplasmic environment also remains poorly studied, with little understood about

virulence factors that promote pathogenic intracellular growth in the handful of species

able to replicate in the mammalian cytoplasm or mammalian factors that prevent

growth and target pathogens for destruction17. Furthermore, while our understanding

of microbial metabolism in isolated pure cultures is deep for commonly studied

organisms, the metabolism of bacterial communities is currently probed primarily

through metagenomics techniques that remain limited31. The metabolism of symbiotic

bacteria or pathogens living in vivo is likewise poorly characterized, with evidence that

in vivo metabolism differs significantly from that in pure laboratory culture32.

Intracellular pathogens and symbionts alike must be able to take nutrients from

their cellular hosts. Genes such as Hpt, the glucose-6-phosphate translocase from L.

monocytogenes, allow for assimilation of host sugars33. Studies with auxotrophic strains

of L. monocytogenes show the extent of the external metabolic requirement for

intracellular division32,34. In contrast, S. elongatus requires very little input from its

external environment besides light, carbon dioxide, and a small number of salts and

minerals. This minimal requirement may be central to our observation that

photosynthetic bacteria can replicate inside the macrophage. Indeed, non-pathogenic

photosynthetic autotrophs seem to have a privileged relationship with eukaryotic cells,

100
able to coexist and even grow inside with relatively little damage to the host cell

compared to even non-virulent E. coli.

Nutrient and gas exchange between tissue culture cells and photosynthetic algae

have been demonstrated previously12, but engineering a mutualistic metabolic

endosymbiosis remains extremely difficult due to the sheer metabolic requirement of

immortalized cells in culture. Based on concentrations of glucose and fructose secreted

by engineered strains of S. elongatus22 we estimate that each CHO cell would require on

the order of 25 cyanobacterial cells to sustain growth and J774 macrophages would

require approximately 14,000 bacteria per cell to provide adequate glucose supply,

numbers significantly higher than those observed in our experimental analysis.

Relationships based on other secreted metabolites, small molecules, or enzymes may

prove to be adequate for establishing an engineered mutualism. Additionally,

improvements in the secretion of essential nutrients will further this approach.

Just as synthetic biology can be used to query the principles underlying complex

signaling or transcriptional networks, a synthetic approach can be used to uncover the

complex dynamics underlying symbiotic relationships. Engineering of microbial

communities and (endo)symbioses between different species has tremendous potential as

a tool for synthetic biology, where growth is limited by the complexity of combining

modular genetic devices in a cellular context35. Communities of cells working together

can achieve results that pure cultures cannot do. An engineerable photosynthetic

101
symbiont can provide a light-controlled, orthogonal platform for engineering of animal

cells.

References

1! Gray, MW, Burger, G, and Lang, BF, Mitochondrial evolution. Science 283
(5407), 1476 (1999).
2! Boxma, Brigitte et al., An anaerobic mitochondrion that produces
hydrogen. Nature 434 (7029), 74 (2005).
3! Howe, C J et al., The origin of plastids. Philos Trans R Soc Lond, B, Biol
Sci 363 (1504), 2675 (2008).
4! Pallen, Mark J, Time to recognise that mitochondria are bacteria? Trends
in Microbiology, 1 (2010).
5! Brenner, K, You, L, and Arnold, F, Engineering microbial consortia: a new
frontier in synthetic biology. Trends Biotechnol 26 (9), 483 (2008).
6! Kenny, B and Valdivia, R, Host–microbe interactions: bacteria. Curr Opin
Microbiol 12 (1), 1 (2009).
7! Anderson, J Christopher, Clarke, Elizabeth J, Arkin, Adam P, and Voigt,
Christopher A, Environmentally controlled invasion of cancer cells by
engineered bacteria. J Mol Biol 355 (4), 619 (2006).
8! Forbes, Neil S, Engineering the perfect (bacterial) cancer therapy. Nat Rev
Cancer, 1 (2010).
9! Bermudes, David, Zheng, Li-mou, and King, Ivan C, Live bacteria as
anticancer agents and tumor-selective protein delivery vectors. Current
opinion in drug discovery & development 5 (2), 194 (2002).
10! Sizemore, D R, Branstrom, A A, and Sadoff, J C, Attenuated Shigella as a
DNA delivery vehicle for DNA-mediated immunization. Science 270
(5234), 299 (1995).
11! Xiang, Shuanglin, Fruehauf, Johannes, and Li, Chiang J, Short hairpin
RNA-expressing bacteria elicit RNA interference in mammals. Nat
Biotechnol 24 (6), 697 (2006).
12! Buchsbaum, R and Buchsbaum, M, AN ARTIFICIAL SYMBIOSIS. Science
80 (2079), 408 (1934); Taylor, D L, Artificially Induced Symbiosis Between
Marine Flagellates and Vertebrate Tissues in Culture. J. Protozool. 25 (1),
77 (1978).
13! Rumpho, M E, Summer, E J, and Manhart, J R, Solar-powered sea slugs.
Mollusc/algal chloroplast symbiosis. Plant Physiol 123 (1), 29 (2000).
14! Ridley, S M and Leech, R M, Division of chloroplasts in an artificial
environment. Nature 227 (5257), 463 (1970).
15! Bonnett, H.T and Eriksson, T, Transfer of Algal Chloroplasts into
Protoplasts of Higher Plants. Planta 120, 71 (1974).
16! Nass, M M, Uptake of isolated chloroplasts by mammalian cells. Science
165 (898), 1128 (1969).

102
17! Ray, K, Marteyn, B, Sansonetti, PJ, and Tang, CM, Life on the inside: the
intracellular lifestyle of cytosolic bacteria. Nature Reviews Microbiology 7
(5), 333 (2009).
18! Bielecki, J, Youngman, P, Connelly, P, and Portnoy, D A, Bacillus subtilis
expressing a haemolysin gene from Listeria monocytogenes can grow in
mammalian cells. Nature 345 (6271), 175 (1990).
19! Goetz, M et al., Microinjection and growth of bacteria in the cytosol of
mammalian host cells. Proc Natl Acad Sci USA 98 (21), 12221 (2001).
20! Jeon, KW, Change of Cellular" Pathogens" into Required Cell
Componentsa. Annals of the New York Academy of Sciences 503
(Endocytobiology), 359 (1987).
21! Stanier, R Y, Kunisawa, R, Mandel, M, and Cohen-Bazire, G, Purification
and properties of unicellular blue-green algae (order Chroococcales).
Bacteriol Rev 35 (2), 171 (1971).
22! Niederholtmeyer, H et al., Engineering synthesis and export of hydrophilic
products from cyanobacteria. Appl Environ Microbiol, 1 (2010).
23! Low, KB et al., Lipid A mutant Salmonella with suppressed virulence and
TNFα induction retain tumor-targeting in vivo. Nat Biotechnol 17 (1), 37
(1999).
24! Isberg, R R, Voorhis, D L, and Falkow, S, Identification of invasin: a
protein that allows enteric bacteria to penetrate cultured mammalian cells.
Cell 50 (5), 769 (1987).
25! Cossart, P et al., Listeriolysin O is essential for virulence of Listeria
monocytogenes: direct evidence obtained by gene complementation.
Infect Immun 57 (11), 3629 (1989).
26! Wintermute…, E, Dynamics in the mixed microbial concourse. Genes &
Development (2010).
27! Venn, A A, Loram, J E, and Douglas, A E, Photosynthetic symbioses in
animals. J Exp Bot 59 (5), 1069 (2008).
28! Kerney R, Kim E, Hangarter RP, Heiss AA, Bishop CD, Hall BK
Intracellular invasion of green algae in a salamander host. Proc Natl Acad
Sci USA, 108 (14), (2011)
29! Gilbert, PW, The alga-egg relationship in Ambystoma maculatum, a case
of symbiosis. Ecology 25 (3), 366 (1944).
30! Tattersall, GJ and Spiegelaar, N, Embryonic motility and hatching success
of Ambystoma maculatum are influenced by a symbiotic alga. Can. J.
Zool. 86 (11), 1289 (2008).
31! Tyson, GW et al., Community structure and metabolism through
reconstruction of microbial genomes from the environment. Nature 428
(6978), 37 (2004).
32! Muñoz‐Elías, EJ and McKinney, JD, Carbon metabolism of intracellular
bacteria. Cellular microbiology 8 (1), 10 (2006).
33! Chico-Calero, I et al., Hpt, a bacterial homolog of the microsomal
glucose-6-phosphate translocase, mediates rapid intracellular proliferation
in Listeria. Proc Natl Acad Sci USA 99 (1), 431 (2002).

103
34! Stritzker, J et al., Growth, virulence, and immunogenicity of Listeria
monocytogenes aro mutants. Infection and immunity 72 (10), 5622 (2004);
Marquis, H, Bouwer, HG, Hinrichs, DJ, and Portnoy, DA, Intracytoplasmic
growth and virulence of Listeria monocytogenes auxotrophic mutants.
Infection and immunity 61 (9), 3756 (1993).
35! Purnick, Priscilla E. M and Weiss, Ron, The second wave of synthetic
biology: from modules to systems. Nat Rev Mol Cell Biol 10 (6), 410
(2009).

104
!

Chapter 4
/ Personalized Genetic Engineering of Plants

“One can imagine organic crops biotically engineered...to be as nutritious and

delicious as science can make them, and to invite further refinement by the
growers. Along with genetic BioBricks, let there be AgriBricks to finesse crop
genomes for local ecological and economic fitness.”
-Stewart Brand, Whole Earth Discipline
iGEM

!In the previous chapter, the “parts” that made up the synthetic system spanned

multiple biological scales—genetic elements such as promoters and ribosome binding

sites (RBS), protein coding genes for invasin and listeriolysin, plasmids that integrate

into the S. elongatus genome, and whole cells and animals, all interacting in complex

ways, allowing us to probe evolutionary dynamics of the host-symbiont relationship.

Many of these parts are generalizable, sharable between different organisms in different

contexts. For example, expression of invasin allows many species of non-pathogenic

bacteria to invade mammalian cells, a function that has been incorporated into several

synthetic biology devices2. Genetic control elements, both transcriptional and

translational—promoters, ribosome binding sites, terminators, RNA aptamers3—are

also generalizable, if not always to different species, often to different networks and

contexts within the same cell type.

Standardization of these generalizable parts for ease of assembly and testing in

different contexts has been a primary goal of many participants in the recent wave of

synthetic biology. Such efforts at standardization may one day make the dream of

engineering biology through assembly of well characterized, off-the-shelf parts a

reality. Coupled to an ethos of sharing based loosely but explicitly on the open source

software movement, the Registry of Standard Biological Parts and the BioBricks

106
foundation go a long way towards achieving this goal. Building on shared parts with

assembly standards that take some of the trail and error out of the art of gene cloning

can allow for relatively rapid prototyping, even by undergraduate students on short

timescales.

This is the foundation for the International Genetically Engineered Machines

competition (iGEM), an undergraduate synthetic biology competition that takes place

every year at the Massachusetts Institute of Technology. Students design and build

synthetic biology projects based on standardized and interchangeable parts, for which

they can draw on and build off of the parts that already exist in the registry, returning

new parts in kind at the end of the summer. This chapter describes the work of the

Harvard University undergraduate iGEM team—Aaron Deardon, Jonathan DeWerd,

Alex Gedeon, Jackie Quinn, Morgan Paull, Anu Raman, Mark Thielmann, Lu Wang,

Julia Winn—that I mentored over the summer of 2010, as well as continuing work on

the project in our lab with the other team mentors Devin Burrill, Patrick Boyle, and

Mara Inniss.

Standardization/Personalization

Even within the iGEM community and the Registry of Standard Biological

Parts, the push to standardization in synthetic biology has been hardly standard. For

gene cloning, one of the most straightforward aspects of biological design, there are

107
currently five different BioBrick DNA assembly standards accepted by the Registry.

Each of these standards can allow for flexible and relatively rapid DNA construction

and easier sharing between members of the community. The specifics of the standard,

what restriction enzymes are used, the size and amino acid composition of the linker

sequence left between two attached parts4 change, but the concept stays the same.

Recent efforts have expanded the concept of the BioBrick to be even easier to assemble

through PCR based methods5, more amenable to protein fusions6, and to be useful for

engineering of a number of organisms, including bacteria, yeast, mammalian cells, and

plants.

In this dissertation I have argued against standardization as stifling to the

diversity and complexity of biological systems, but in current approaches to plant

genetic engineering, industrial interests, regulatory challenges, and gene patents have

significantly limited the diversity of research into engineering methods and

applications. A monoculture approach to industrial agriculture that threatens food

security, public health, and ecosystem stability reflects the monopolizing forces on the

agricultural engineering industry.

Of the approximately 250,000 plant species that have been identified, 75,000 are

edible and 7,000 are cultivated, but just 5 make up the vast majority of agricultural

output—wheat, rice, soy, corn, and rape7. Of these vast monocultures, an increasingly

large percentage is genetically modified to be resistant to the weed killing herbicide

108
glyphosate (marketed as RoundUp), with 90% of soy and 70% of corn.

Not only does biodiversity contribute to the strength and stability of an

ecosystem8, but studying diverse organisms can improve our understanding of complex

metabolism9. Can standardization be used not as a way to make biology uniform, but as

a stepping stone to further diversity in biological engineering solutions, with parallel,

shared efforts in different organisms with different goals? Will the open source

approach of synthetic biology be able to increase access to the practices and potential

benefits of plant genetic engineering? Can facile and rapid assembly of DNA-based

synthetic biology devices expand the potential for genetic engineering suited to local

environments and cultures over a “one size fits all” engineering of high-yield crops?

Our project begins with this personalization. If we were going to grow a garden,

what modifications would we want to make? What modifications would members of our

local community be interested in? We interviewed sixty visitors to the Harvard

Farmer’s Market to begin to get an idea of what modifications would interest our

neighbors as we were brainstorming. We chose to focus on three main aspects of garden

personalization, alteration of nutrient content and flower petal color through

accumulation of pigments such as lycopene and beta-carotene, modification of flavor by

addition of sweet tasting proteins miraculin and brazzein, and deletion of common

allergen proteins. Of the 60 people we interviewed, the highest percentage favored

engineering to improve nutrient content in plants, while only 5% of respondents

109
favored deletion of allergens (figure 4.1). However, of the very small number of

respondents who had food allergy, 100% were in favor of eliminating their specific

allergen. This is the heart of personalization, not everyone will agree on what

characteristics are altered, but each person can find foods that are specially tailored to

their needs. The work presented in the following chapter shows the feasibility and

applicability of several such small-scale personalization projects in the model organism,

Arabidopsis thaliana, but aim for a future personalization of many different plants that

individuals could perhaps grow in their own home gardens. This iGarden approach

Figure 4.1 Responses to our survey on personalized genetic engineering of plants. Sixty visitors
to the Harvard Farmers’ Market were asked whether they would feel comfortable changing
certain plant characteristics through genetic engineering, and the percentage of respondents
responding favorably is plotted for each category.

110
would allow for a diversity of plants grown locally and sustainably, engineered for

personalization to individual, community, and local environmental needs using easily

transferable, standardized techniques.

Materials and Methods

Plasmids and cloning

All gene assembly was performed in E. coli DH5" using BioBrick assembly

standard 216, and all parts described here were submitted to the BioBrick registry. The

Arabidopsis thaliana pORE series transformation vectors were provided by The

Arabidopsis Information Resource (TAIR) and engineered to support BioBrick cloning

through PCR based methods (vectors listed in Table 4.1 and all primers and

oligonucleotides used are listed in Appendix A, table S1). pORE Open Series vectors O1

and O2 were digested with SpeI and SacII and ligated with an annealed oligonucleotide

insert with NheI and SacII overhangs containing the BioBrick Multiple Cloning Site

(MCS) to create vectors V1 and V2. pORE Expression Series vectors E3 and E4 were

likewise digested with HindIII and SpeI and ligated with an insert PCR amplified from

the expression vectors containing HindIII site upstream from the ENTCUP2 promoter

and the BioBrick MCS and an NheI site downstream to create vectors V3 and V4.

pORE Reporter Series vectors R1, containing the gusA reporter, and R3, containing the

smgfp reporter (soluble modified green fluorescent protein), were digested with HindIII

111
and SpeI and ligated with inserts containing the reporter gene PCR amplified with

primers containing a HindIII site followed by the BioBrick MCS upstream and NheI

downstream, yielding vectors V5 and V6.

Artificial microRNA constructs targeting Bet v 1, LTP1, Lut2, %-Ohase1 and

Lyc1 were constructed through PCR assembly10 on vector RS300, kindly provided by

Kirsten Bomblies and Detlef Weigel, with primers designed using the Web MicroRNA

Designer11 (Appendix A, table S1). Constructs for intron containing hairpin RNA

(hpRNA)12 were made through sequential assembly of BioBrick parts PCR amplified

from the A. thaliana genomic DNA isolated from wild type A. thaliana using the Qiagen

DNeasy genome isolation kit. Sense and antisense primers for hpRNA construct parts

for Ger3, Bet v 1, and LTP1, and primers for the PDK intron are provided in Appendix

A, Table S1.

Flavor proteins brazzein and miraculin were codon optimized for expression in

A. thaliana, commercially synthesized by Mr. Gene (Regensburg, Germany), and

assembled with the pENTCUP2 promoter and NosT transcriptional terminator with

BioBrick assembly. Completed constructs were subcloned from the BioBrick assembly

vector V0120 to the BioBrick modified pORE vectors through digestion with EcoRI and

PstI.

112
Plant Transformation

Agrobacterium mediated transformation was performed according to previously

reported techniques13. Briefly, Agrobacterium tumefaciens was made electrocompetent by

washing in cold sterile water and resuspending in 10% glycerol. Vector DNA was

grown up in E. coli DH5" and plasmid purified, dialyzed to remove excess salt, and

electroporated into Agrobacterium. Transformed colonies selected on kanamycin were

grown in YEB medium and spread onto YEB agar plates and allowed to form a lawn.

Lawns were scraped and suspended in a solution of 20% YEB, 4% sucrose (w/v), and

0.024% Silwet L-77 surfactant (Helena Chemical Company, Collierville, TN). Wild type

Arabidopsis thaliana flowers were dipped in the agrobacterium solution. Seeds were

collected from mature plants and selected on 1X Murashige & Skoog media with 0.7%

agar supplemented with 5mg/L glufosinate or 50 !g/ml kanamycin.

qRT-PCR

Whole cell RNA was collected using the plant RNEasy kit (Qiagen) and cDNA

was synthesized with the SuperScript III First-Strand synthesis kit (Invitrogen).

Quantitative RT-PCR was performed with primer pairs amplifying 100 base pair

amplicons within the target genes to identify expression of heterologous genes or

knockdown of endogenous genes. All primers are listed in Appendix A, Table S1.

113
Results and Discussion

Design of BioBrick compatible vectors for Arabidopsis transformation

We modified six of the pORE series vectors14 to be compatible with BioBrick

DNA assembly (Table 4.1). This set of vectors provides versatility in the resistance

marker for selection in plants, either glufosinate (pat) or kanamycin (nptII), gene

expression from the constitutive pENTCUP2 promoter, or in-frame fusion with a visual

reporter, either gusA or soluble modified GFP (smgfp).

Vector BioBrick Registry Bacterial Plant promoter reporter

Name Resistance Resistance
Marker

V1 BBa_K382000 Kan pat none none

V2 BBa_K382001 Kan nptII none none

V3 BBa_K382002 Kan pat ENTCUP2 none

V4 BBa_K382003 Kan nptII ENTCUP2 none

V5 BBa_K382004 Kan nptII none gusA

V6 BBa_K382005 Kan nptII none smgfp

Table 4.1 Table of modified pORE series vectors designed to be compatible with BioBrick DNA
assembly.

We transformed our personalization constructs into Arabidopsis on the modified

vectors via agrobacterium mediated transformation13 and observed approximately a

0.1% transformation efficiency based on selection on plates containing glufosinate or

kanamycin (figure 4.2).

114
 Figure 4.2 Photograph of Arabidopsis seedling resistant to glufosinate selected on agar plates.
Approximately 0.1% of seedlings were transformed with the vector and became resistant to either
kanamycin or glufosinate.

Petal color and nutritional enhancement through pigment accumulation

Ornamentals were amongst the first plants altered through hybridization

methods, for improved durability after cutting, altered scents, and different flower petal

colors15. Flower color can inspire true passions and panics; 17th century Holland saw an

outbreak of “tulipomania” caused by the variegated colors of tulips infected with tulip

breaking virus, with bulbs selling for exorbitant prices16. Tulip breaking virus interferes

115
 with the production of anthocyanins, a

primary pigment in flower petals, leading

to a striped appearance (figure 4.3). Similar

“broken” phenotypes have been achieved

through years of careful breeding and,

much more recently, genetic engineering of

anthocyanin pathways (extensively

reviewed in Tanaka et. al. and references

therein17). Indeed, RNA interference (itself

a likely viral defense mechanism) was

discovered when overexpression of

Figure 4.3 Tulip breaking virus causes a striped
appearance in petals, where pigment loss occurs
in infected cells. The Semper Augustus, pictured chalcone synthase (CHS), a gene in the
here from the Great Tulip Book published c.
1640, is famous for being the most expensive anthocyanin pathway expected to increase
tulip sold during the “tulipomania” of 17th
century Holland.
color production surprisingly led to a

decrease in flower petal pigment18.

The first instance of metabolic engineering for altered flower petal color was in

1987 with the introduction of an anthocyanin pathway gene, dihydroflavonol 4-

reductase from corn, into a Petunia hybrida mutant lacking any flower pigment19. Since

then, anthocyanin pathway engineering is the primary path for altered flower

pigments20, although engineering of plant carotenoids has also successfully altered

116
flower petal color21.

Carotenoids also provide color to flowers and are derived from isoprenoids, a

valuable starting point for many metabolic engineering projects22, with many medicinal

applications as well as being precursors of many molecules that are valuable sources of

biofuels. While other plant pigments including the anthocyanins and betalains are

dispensable for plant function, carotenoids are precursors to required chemicals and

have crucial photoprotective function.

Carotenoids are responsible for the color of many ornamental flowers, including

marigold, daffodil, rose, lily, freesia, and daisy15. Pigmented carotenoids all derive from

lycopene, which has a deep pinkish red color and is responsible for the red color of

tomatoes. Desaturation and isomerization of the colorless precursor neurosporene by

carotenoid isomerase and &-carotene desaturase leads to the production of all-trans-

lycopene, which is further cyclized and processed in two parallel pathways by lycopene

'-cyclase or lycopene %-cyclase (figure 4.4). Both ends of the lycopene molecule can be

cyclized, and two %-cyclizations leads to production of %-carotene. Two '-cyclizations

are rare, but one '- and one %-cyclization produces "-carotene, a precursor of the

orange pigment lutein23. Mutations in lycopene '-cyclase in A. thaliana are designated

lut2, for lutein deficient, and show that lutein metabolism is not required for plant

function24.

In arabidopsis, anthocyanins provide the main source of color in the brown of

117
Figure 4.4 Modification of A. thaliana carotenoid metabolism. Pigmented carotenoid synthesis
proceeds from the desaturation of colorless neurosporene to lycopene, which is red. Knockdown of
Lycopene epsilon cyclase (LUT2, green Xs) prevents flux from lycopene to (- and '-carotene,
building up lycopene. Knockdown of lycopene beta cyclase likewise blocks flux from lycopene to )-
and ", %, and )-carotene (LYC, blue Xs). Orange %-carotene builds up upon knockdown of
carotenoid beta-ring hydroxylase (%OH, purple Xs), blocking flux to %-cryptoxanthin, zeinoxanthin,
and downstream pigments.

118
the seed coat and purple of the leaves25. Arabidopsis flowers are white, and we sought to

explore the possibility of altering flower petal color through accumulation of

carotenoid pigments. Deletion of metabolic genes can channel flux to desired metabolic

pathways leading to accumulation of useful metabolites26. In Arabidopsis, mutation of

carotenoid isomerase leads to build up of tetra-cis-lycopene, the intermediate product of

the &-carotene desaturase reaction, and to an orange-yellow color of the Arabidopsis

seedlings27. We tested three individual downstream knockdowns of carotenoid

biosynthesis genes (figure 4.4) through artificial microRNA interference11,28 (figure 4.8):

knockdown of lycopene '-cyclase, mutations in which are known to accumulate %-

cyclized carotenoids24 and/or lycopene %-cyclase, a precursor to %-carotene, can

potentially lead to a buildup of lycopene and increased red color. Knockdown of the

carotenoid %-ring hydroxylase, which further processes "- and %-carotenoids can lead to

a buildup of these orange colored pigments.

Many factors besides pigment accumulation affect flower petal color, including

the presence of other pigments, cell shape, and vacuolar pH15. While the petals of F1

generation transformed Arabidopsis remained white, the plants grew to a smaller height

and were significantly more sensitive to environmental conditions, with all LYC-

knockdown plants not surviving to flowering stage. In the F2 generation, plants with

LUT2 or %-*+ knockdown displayed highly variable growth and were on average

smaller and slower growing than sibling plants (figure 4.5). Overexpression of

119
carotenoid %-ring hydroxylase has been shown to enhance the Arabidopsis stress

repsonse, so knockdown is likely to render the plant more sensitive to photo-damage29.

Tuning of the degree of gene expression knockdown, further combinatorial

knockdown, targeting of other genes in the carotenoid pathway, such as carotenoid

isomerase, which is known to build up lycopene pigments, or overexpression of

upstream genes may sufficiently increase the buildup of pigments. Furthermore,

A B
Braz LTPa β-OH LTPa LUT2 LTPh β-OH Braz Ger Mir

Braz Mir LTPa LTPh Ger3 LUT2 β-OH

Ger LUT2 LUT2 Braz β-OH LUT2 Braz LTPa LTPh Braz

Figure 4.5 Growth of engineered plants in

the F2 generation. A.) Photographs of the
trays where plants are grown shows dramatic
differences between growth of different
engineered plants. Each column holds plants
grown from individual seeds collected from
the same parent plant. Plants that had bolted
at the time of the photograph, approximately
one month after sprouting, are isolated from
β-OH β-OH Braz Mir Braz Mir Braz Mir Braz LTPa
neighboring plants with a plastic sheath. B.)
Relative growth of the different strains. Each
plant was scored on a 0-10 scale for growth
and averaged for each engineered strain.
Plants expressing knockdown constructs are
highly variable and on average have
significantly decreased growth compared to
strains where miraculin or brazzein is
heterologously expressed.

120
targeting of gene knockdown to the flower petals may decrease toxicity of the

knockdown of the essential carotenoid biosynthesis genes while targeting the pigment

accumulation to the flower petals30.

As carotenoids are valuable antioxidants and precursors for the biosynthesis of

required micronutrients such as vitamin A, manipulation of plant carotenoid

biosynthesis pathways also has tremendous potential for the improvement of nutrient

availability. Significant previous work has increased carotenoid production in crop

plants and in E. coli, primarily through addition of biosynthetic pathways and metabolic

engineering, for example in the production of “golden rice”31 or metabolic engineering

of lycopene production pathways in E. coli32.

With BioBrick compatible vectors that allow for rapid design cycles of synthetic

biology by those with even minimal training, including undergraduates, there is large

potential for the engineering of plant pigment metabolism, either through RNA

interference based methods33 or through construction and overexpression of pigment

production pathways. Coupled to the fine-tuning possibilities allowed by synthetic

biology, through techniques such as ribosome binding site tuning33, these techniques

can allow for the predictable engineering of a wide range of flower color and nutrient

level personalizations.

121
Flavor inversion

Our BioBrick method allows for rapid testing of many personalizations in

parallel, so we concurrently sought to modify the taste of Arabidopsis, specifically

increasing the sweetness of the plant without altering sugar content. There are several

naturally occurring proteins that are 500-4000 time sweeter than sugar by weight.

Brazzein, monellin, thaumatin, pentadin, mabinlin, and curculin are sweet proteins found

in a variety of African and South Asian fruits, with no sequence similarity or common

features between them34. Brazzein, isolated from the West African fruit Pentadiplandra

brazzeana is the smallest of these proteins with only 54 amino acids, has a high heat

tolerance, surviving heating at 80° Celsius for several hours, and has been previously

expressed heterologously in E. coli34, Z. mays35, and Lactobacillus lactis36.

Miraculin, isolated from the berries of the West African plant Richadella

dulcifica, does not taste sweet on its own, but binds to taste receptors on the tongue and

causes sour foods to taste sweet, creating a “taste inverter.” 1!M of miraculin is

sufficient to activate this inversion, where 20mM citrate corresponds to the sweetness of

0.3M sucrose37. Miraculin is a glycosylated homodimer that has been heterologously

expressed in lettuce38, tomato39, and even E. coli37, indicating that glycosylation is not

required for functional expression.

Brazzein and miraculin genes were commercially synthesized and codon

optimized for expression in Arabidopsis. Codon usage was acceptable for expression in

122
! >)!$5#"*76 =)!$5#"*76
ABCD ) E ) E
4:%?@7%)%

Figure 4.6 Expression of

commercially synthesized, plant
5$67!"8$%!,'5$(-$*-$

codon optimized miraculin and

3 ,*"*+,-$+)= >)!$5#"*,( brazzein in E. coli and S. cerevisiae.
A.) Miraculin and B.) brazzein
BioBricks were expressed in E. coli
off of BioBrick modified pET-duet
=)!$5#"*,(
expression vector1 with either N- or
09: C- terminal Strep-II tag for western
,*"*+,-$+)> blot analysis or YFP tag. Miraculin
0 10 20 340 expression was undetectable
!"#$%&'(!)"*+,-!"'*%.#"*/ through western blot, with relatively
" low fluorescence detectable. In
>)!$5#"*76 =)!$5#"*76 contrast, brazzein was expressed
ABCD ) E ) E quite highly, with decreased
3:%?@7%)% expression observed through
western blot and fluorescence
30%?@7%)
measurements with tagged on the
N-terminus. C.) Brazzein expression
5$67!"8$%!,'5$(-$*-$

< was observed in yeast off of the

=)!$5#"*,(
constitutive TEF promoter and the
2 copper inducible CUP1 promoter
upon the addition of [XXX] copper.
;
In yeast brazzein showed a
>)!$5#"*,( significantly higher molecular
4
,*"*+,-$+ weight compared to expression in E.
coli, likely due to glycosylation.
0 10 20 340
!"#$%&'(!)"*+,-!"'*%.#"*/

#
IB
E= B
I
H
C
CG
)=
F

;0%?@7%)% !
4:%?@7%)% !

E. coli and Saccharomyces cerevisiae, so we first tested for expression in these

microorganisms to verify the constructs. Constructs were tagged at either the N- or C-

terminus with the Strep-II tag for western blot analysis or with 2xYFP for fluorescent

measurement. Miraculin showed very little expression, undetectable on the western blot

123
and only a small increase in YFP fluorescence over uninduced background (figure 4.6A).

Miraculin expression in E. coli has been shown to be difficult and extraction to be

dependent on specific extraction conditions, so this was perhaps a factor in our

expression experiments37. Brazzein, however, was highly expressed, with significantly

higher expression when tagged at the C-terminus vs. the N-terminus(figure 4.6B).

Brazzein was also highly expressed in yeast off of the constitutive TEF

promoter or the inducible CUP1 promoter (figure 4.6C). However, the molecular weight

of the Strep-II tagged brazzein observed by western blot in yeast was significantly

higher than that found in E. coli, approximately 35 kDa vs. approximately 12 kDa. This

shift is likely due to glycosylation of the brazzein protein40, the machinery for which is

absent in E. coli.

Untagged miraculin and brazzein

! 4"( #$* "#0 "#5 "#6 %4"( "#5 #$*
$% ) ! ! ! $ ! )
were transformed into A. thaliana under
011)23 /-.

!"#$%&'"( )#$**+"( control of the pENTCUP2 promoter

" !"#5 )#$* $%4"( )#$* !"#5 $%4"( and the NosT transcriptional
011)23 ,-.
terminator on the V1 BioBrick vector.
!"#$%&'"( )#$**+"(

Seeds were selected on glufosinate.

Figure 4.7 Incorporation of heterologous DNA for
miraculin and brazzein and expression of miraculin
These plants were often significantly
was verified with PCR amplification of genomic DNA
or from reverse trasncribed whole cell RNA with
brazzein or miraculin specific primers(Appendix A, more vigorous and fast-growing than
Table S1).
their cousins that had genes knocked

124
down (figure 4.5). While incorporation of both the miraculin and brazzein genes into

the plant genome was verified with PCR, only expression of miraculin RNA was

detected through PCR amplification of reverse-transcribed whole-cell RNA (figure 4.7).

Garden personalization through allergen knockdown

Millions of people suffer from food allergies, from relatively mild and localized

symptoms of oral allergy syndrom (OAS)41 to life-threatening anaphylaxis. While the

mechanisms accounting for the development of IgE mediated food allergy is not known,

the specific protein antigens that cause allergic reactions are known for many plants.

Oral allergy syndrome, or pollen-food allergy, is caused by allergic cross-

reaction between proteins found in pollens and fruits and vegetables. Such panallergen

proteins cause a syndrome of allergy symptoms against a large group of fruits and

vegetables. Lipid Transfer Protein, LTP 1, is recognized as a dangerous panallergen in

many plants 42, including arabidopsis43, and has been linked to anaphylaxis44. Birch

pollen allergy translates to allergy to many fruits and vegetables that contain homologs

to the pollen protein Bet v 145 46 Germin-type proteins are also common panallergens,

including Ger347. We envision fruits and vegetables genetically engineered to not

express these allergen proteins, allowing individuals to tailor a personalized garden to

their allergies.

125
 ! "
$%#!&'&(&')"*+,%-%&", $%#!&'&(&')"*+,%-%&",
;<<41 ;<<41 ;<41 ;<41
!"#!" !"#$%" .#&'!"#!" 6:;<<*='%=,'2>
-'678 -'678?

-678*&,.#!2,'1&

-678*&,.#!2,'1&

!19'2'#3
'#&,%#*!19'2'#3

2%-19"&"@*0.',1'#*A%,*678'*&.,3"&'#3

4.!"*1.','#3 /(,&0",*1,%2"!!'#3*
45*678'*-.20'#",5

Figure 4.8 Two methods for RNA knockdown in plants (not to scale). A.) intron-containing hairpin
RNA (hpRNA) is constructed from three BioBrick parts PCR amplified from the Arabidopsis genome:
a 300 bp region identical to the 5’-3’ sequence of the target gene, the PDK intron, and the 300bp
complementary to the sense strand. When expressed from a constitutive promoter, the
complementary sense and antisense strands of the mRNA bind and the intron is spliced, leaving a
double stranded RNA hairpin that is further processed by the plant RNA interference machinery. B.)
Artificial microRNA proceeds through a similar mechanism. 30 base pair complementary micro RNA
sequences are spliced together into a short hairpin RNA upon transcription of the mRNA.

Many groups have successfully knocked down allergen protein expression in

plants48,49, including Bet v 1 in strawberry50, Ara h 2 in peanut51, the apple protein Mal

d 152, and the tomato protein Lyc e 148. While deletion of allergen genes from the

genome would safely remove expression of the allergen, such full deletion is difficult in

plants due to the preponderance of multiple gene isoforms. Knockdown of allergen

genes with RNA interference is better than deletion from the genome because a single

126
RNA interference construct can effectively target multiple mRNA isoforms.

Improvements in zinc-finger nuclease mediated engineering at targeted locations in the

plant genome will likely make full deletion of isoforms more feasible53.

We tested two methods for targeted RNA interference, intron-containing hairpin

RNA (hpRNA)12,54, and artificial microRNA (amiRNA)11,28 (figure 4.8). hpRNA is a

modularly designed hairpin, where 300 base pairs identical to the mRNA target bind to

their reverse complement, separated by a short hairpin of approximately 30 base pairs

created when the intron between the sense and antisense parts is spliced out. These

parts can be constructed through PCR amplification from the Arabidopsis genome and

assembly with the BioBrick standard. Artificial miRNA is less modular, as it takes as a

starting point a naturally existing miRNA that is spliced by the RNA processing

machinery of the plant to make a short targeted RNA hairpin. Artificial miRNA can

therefore be much more specific to a single gene or isoform, with fewer off target effects

than hpRNA, which can target any of the 300 base pairs of sequence homology. The

specificity can also be a problem, we were unable to target the Arabidopsis Ger3 gene

using the web-based designer for targeting sequences with artificial microRNA

constructs.

As with knockdown of carotenoid pathway genes, the allergen knockdowns

significantly affected the growth of the plants. No seeds expressing Bet v 1 knockdown

constructs grew, and Ger3 and LTP knockdown showed highly variable and decreased

127
growth (figure 4.5). This variability is likely a result of the variability in the efficiency

of gene knockdown by RNA interference, which can vary significantly from plant to

plant and construct to construct. Preliminary quantitative real time PCR analysis on

RNA extracted from leaves of plants that grew large enough to harvest showed variable

knockdown of allergen genes. The largest fold change was seen with LTP artificial

mRNA (~4-fold decrease in expression) and with Ger3 knockdown (~10-fold, data not

shown).

Conclusions

Plant biotechnology has potential to affect production of not only food, but also

fuel sources, medicines, and materials7. Speeding the design cycle through standardized

methods such as those advocated by the synthetic biology approach can diversify the a

plants that are engineered and possible applications. We present a framework and

preliminary results towards personalization of plants for health, safety, and fun.

References

1! Agapakis, C M, Ducat, D C, Boyle, P M, Wintermute, E H, Way J C, Silver,

P A. Insulation of a synthetic hydrogen metabolism circuit in bacteria.
Journal of Biological Engineering 4 (3), (2010).
2! Isberg, R R, Voorhis, D L, and Falkow, S, Identification of invasin: a
protein that allows enteric bacteria to penetrate cultured mammalian cells.
Cell 50 (5), 769 (1987); Anderson, J Christopher, Clarke, Elizabeth J,

128
 Arkin, Adam P, and Voigt, Christopher A, Environmentally controlled
invasion of cancer cells by engineered bacteria. Journal of molecular
biology 355 (4), 619 (2006); Xiang, Shuanglin, Fruehauf, Johannes, and
Li, Chiang J, Short hairpin RNA-expressing bacteria elicit RNA
interference in mammals. Nature Biotechnology 24 (6), 697 (2006).
3! Isaacs, Farren J, Dwyer, Daniel J, and Collins, James J, RNA synthetic
biology. Nature Biotechnology 24 (5), 545 (2006).
4! Anderson, J Christopher et al., BglBricks: A flexible standard for biological
part assembly. 1 (2010).
5! Gibson, D G, Synthesis of DNA fragments in yeast by one-step assembly
of overlapping oligonucleotides. Nucleic Acids Research, 7 (2009).
6! Phillips, I and Silver, P, A New Biobrick Assembly Strategy Designed for
Facile Protein Engineering. dspace.mit.edu (2006).
7! Kern, M, Food, feed, fibre, fuel and industrial products of the future:
challenges and opportunities. Understanding the strategic potential of
plant genetic engineering. Journal of Agronomy and Crop Science 188 (5),
291 (2002).
8! Tilman, D, Biodiversity: population versus ecosystem stability. Ecology 77
(2), 350 (1996).
9! Winkel-Shirley, B, It Takes a Garden. How Work on Diverse Plant Species
Has Contributed to an Understanding of Flavonoid Metabolism. Plant
Physiology 127 (4), 1399 (2001).
10! Schwab, R, Highly Specific Gene Silencing by Artificial MicroRNAs in
Arabidopsis. THE PLANT CELL ONLINE 18 (5), 1121 (2006).
11! Ossowski, Stephan, Schwab, Rebecca, and Weigel, Detlef, Gene
silencing in plants using artificial microRNAs and other small RNAs. The
Plant Journal 53 (4), 674 (2008).
12! Helliwell, Chris and Waterhouse, Peter, Evaluation Study, 2003.
13! Logemann, Elke, Birkenbihl, Rainer P, Ulker, Bekir, and Somssich, Imre E,
An improved method for preparing Agrobacterium cells that simplifies the
Arabidopsis transformation protocol. Plant Methods 2, 16 (2006).
14! Coutu, Catherine et al., Comparative Study, 2007.
15! Grotewold, E, The genetics and biochemistry of floral pigments. Annual
Review of Plant Biology 57, 761 (2006).
16! Lesnaw, JA and Ghabrial, SA, Tulip breaking: past, present and future.
Plant disease 84 (10), 1052 (2000).
17! Tanaka, Y, Katsumoto, Y, Brugliera, F, and Mason, J, Genetic engineering
in floriculture. Plant cell, tissue and organ culture 80 (1), 1 (2005).
18! Napoli, C, Lemieux, C, and Jorgensen, R, Introduction of a Chimeric
Chalcone Synthase Gene into Petunia Results in Reversible Co-
Suppression of Homologous Genes in trans. THE PLANT CELL ONLINE 2
(4), 279 (1990).
19! Meyer, P, Heidmann, I, Forkmann, G, and Saedler, H, A new petunia
flower colour generated by transformation of a mutant with a maize gene.
(1987).

129
20! Nakatsuka, Takashi et al., Genetic engineering of novel flower colour by
suppression of anthocyanin modification genes in gentian. JI of Plant
Physiology 167 (3), 231 (2010); Tanaka, Y, Tsuda, S, and Kusumi, T,
Metabolic engineering to modify flower color. Plant and Cell Physiology 39
(11), 1119 (1998).
21! Suzuki, Sakae et al., Flower color alteration in Lotus japonicus by
modification of the carotenoid biosynthetic pathway. Plant Cell Reports 26
(7), 951 (2007).
22! Chang, Michelle C Y and Keasling, Jay D, Production of isoprenoid
pharmaceuticals by engineered microbes. Nature Chemical Biology 2 (12),
674 (2006); Misawa, Norihiko, Pathway engineering for functional
isoprenoids. Current Opinion in Biotechnology, 1 (2011).
23! Cunningham Jr, FX et al., Functional Analysis of the [beta] and [epsilon]
Lycopene Cyclase Enzymes of Arabidopsis Reveals a Mechanism for
Control of Cyclic Carotenoid Formation. THE PLANT CELL ONLINE 8 (9),
1613 (1996).
24! Pogson, B et al., Arabidopsis carotenoid mutants demonstrate that lutein
is not essential for photosynthesis in higher plants. THE PLANT CELL
ONLINE 8 (9), 1627 (1996).
25! Shirley, BW et al., Analysis of Arabidopsis mutants deficient in flavonoid
biosynthesis. The Plant Journal 8 (5), 659 (1995).
26! Kennedy, Caleb J, Boyle, Patrick M, Waks, Zeev, and Silver, Pamela A,
Systems-level engineering of nonfermentative metabolism in yeast.
Genetics 183 (1), 385 (2009); Burgard, Anthony P, Pharkya, Priti, and
Maranas, Costas D, Evaluation Study, 2003.
27! Park, H, Identification of the Carotenoid Isomerase Provides Insight into
Carotenoid Biosynthesis, Prolamellar Body Formation, and
Photomorphogenesis. THE PLANT CELL ONLINE 14 (2), 321 (2002).
28! Schwab, Rebecca et al., Specific effects of microRNAs on the plant
transcriptome. Developmental cell 8 (4), 517 (2005).
29! Davison, PA, Hunter, CN, and Horton, P, Overexpression of β-carotene
hydroxylase enhances stress tolerance in Arabidopsis. Nature 418 (6894),
203 (2002).
30! Hill, TA et al., Discrete spatial and temporal cis-acting elements regulate
transcription of the Arabidopsis floral homeotic gene APETALA3.
Development (Cambridge, England) 125 (9), 1711 (1998).
31! Beyer, Peter et al., Golden Rice: introducing the beta-carotene
biosynthesis pathway into rice endosperm by genetic engineering to
defeat vitamin A deficiency. The Journal of nutrition 132 (3), 506S (2002);
Paine, Jacqueline A et al., Improving the nutritional value of Golden Rice
through increased pro-vitamin A content. Nature Biotechnology 23 (4), 482
(2005).
32! Farmer, WR and Liao, JC, Improving lycopene production in Escherichia
coli by engineering metabolic control. Nature Biotechnology 18 (5), 533
(2000); Wang, Harris H et al., Programming cells by multiplex genome
engineering and accelerated evolution. Nature 460 (7257), 894 (2009).

130
33! Mansoor, S et al., Engineering novel traits in plants through RNA
interference. Trends in Plant Science 11 (11), 559 (2006).
34! Assadi-Porter, F, Efficient Production of Recombinant Brazzein, a Small,
Heat-Stable, Sweet-Tasting Protein of Plant Origin. Archives of
biochemistry and biophysics 376 (2), 252 (2000).
35! Lamphear, Barry J et al., Expression of the sweet protein brazzein in
maize for production of a new commercial sweetener. Plant Biotechnology
Journal 3 (1), 103 (2004).
36! Berlec, Aleš et al., Expression of the sweet-tasting plant protein brazzein
in Escherichia coli and Lactococcus lactis: a path toward sweet lactic acid
bacteria. Applied Microbiology and Biotechnology 73 (1), 158 (2006).
37! Matsuyama, T et al., Functional Expression of Miraculin, a Taste-Modifying
Protein in Escherichia Coli. Journal of biochemistry 145 (4), 445 (2009).
38! Sun, Hyeon-Jin, Cui, Min-Long, Ma, Biao, and Ezura, Hiroshi, Functional
expression of the taste-modifying protein, miraculin, in transgenic lettuce.
FEBS Letters 580 (2), 620 (2006).
39! Hirai, Tadayoshi et al., Production of Recombinant Miraculin Using
Transgenic Tomatoes in a Closed Cultivation System. Journal of
Agricultural and Food Chemistry 58 (10), 6096 (2010).
40! CARLSON, A and ARMENTROUT, R, ENHANCED PRODUCTION AND
PURIFICATION OF A NATURAL HIGH INTENSITY SWEETENER. WO
Patent WO/2010/053,636 (2010); MILES, L and LOUIE, M, SWEETENER
PREPARATIONS AND METHODS OF USE. WO Patent WO/
2010/030,999 (2010).
41! Amlot, PL et al., Oral allergy syndrome (OAS): symptoms of IgE‐mediated
hypersensitivity to foods. Clinical &amp; Experimental Allergy 17 (1), 33
(1987).
42! Salcedo, G et al., Plant non-specific lipid transfer proteins as food and
pollen allergens. Clinical and experimental allergy : journal of the British
Society for Allergy and Clinical Immunology 34 (9), 1336 (2004).
43! Chardin, H et al., Lipid transfer protein 1 is a possible allergen in
Arabidopsis thaliana. International archives of allergy and immunology 131
(2), 85 (2003).
44! San Miguel-Moncín, M et al., Lettuce anaphylaxis: identification of a lipid
transfer protein as the major allergen. Allergy 58 (6), 511 (2003).
45! Jenkins, John A et al., Comparative Study, 2005.
46! Kleine-Tebbe, Jörg et al., Case Report, 2002; Pastorello, E A et al.,
Allergenic cross-reactivity among peach, apricot, plum, and cherry in
patients with oral allergy syndrome: an in vivo and in vitro study. The
Journal of allergy and clinical immunology 94 (4), 699 (1994).
47! Jensen-Jarolim, E et al., Allergologic exploration of germins and germin-
like proteins, a new class of plant allergens. Allergy 57 (9), 805 (2002).
48! Singh, Mohan B and Bhalla, Prem L, Genetic engineering for removing
food allergens from plants. Trends in Plant Science 13 (6), 257 (2008).
49! Gallo, Maria and Sayre, Richard, Removing allergens and reducing toxins
from food crops. Current Opinion in Biotechnology 20 (2), 191 (2009).

131
50! Hjernø, Karin et al., Down-regulation of the strawberry Bet v 1-
homologous allergen in concert with the flavonoid biosynthesis pathway in
colorless strawberry mutant. PROTEOMICS 6 (5), 1574 (2006).
51! Dodo, Hortense W et al., Alleviating peanut allergy using genetic
engineering: the silencing of the immunodominant allergen Ara h 2 leads
to its significant reduction and a decrease in peanut allergenicity. Plant
Biotechnology Journal 6 (2), 135 (2008).
52! Gilissen, Luud J W J et al., Silencing the major apple allergen Mal d 1 by
using the RNA interference approach. The Journal of allergy and clinical
immunology 115 (2), 364 (2005).
53! Petolino, Joseph F et al., Zinc finger nuclease-mediated transgene
deletion. Plant Molecular Biology 73 (6), 617 (2010); Lloyd, A, Plaisier, CL,
Carroll, D, and Drews, GN, Targeted mutagenesis using zinc-finger
nucleases in Arabidopsis. Proceedings of the National Academy of
Sciences of the United States of America 102 (6), 2232 (2005).
54! Wesley, S V et al., Construct design for efficient, effective and high-
throughput gene silencing in plants. The Plant journal : for cell and
molecular biology 27 (6), 581 (2001).

132
Chapter 5
/Human cultures and microbial ecosystems

“For in the eighteenth century there was nothing to hinder bacteria busy at
decomposition, and so there was no human activity, either constructive or
destructive, no manifestation of germinating or decaying life that was not
accompanied by stench.”
-Patrick Süskind, Das Parfum
Background

Cheese is an everyday artifact of microbial artistry. Cheesemaking cultivates

complex cultures of bacteria and fungi to protect milk from dangerous bacterial

spoilage, a practice that developed long before bacteria had been discovered. In a

biological world completely surrounded by rich communities of microorganisms, but a

cultural world that often emphasizes antisepsis, cheeses and other microbe-rich foods lie

at the heart of a post-Pasteurian debate over the positive impact of microbes on our

health and happiness1.

With rising antibacterial resistance and appreciation for how bacteria maintain

our digestive2 and immune3 health, attempting to strike a balance between cultivating

helpful bacteria and keeping dangerous bacterial infections at bay is more important

than ever. Biotechnology and synthetic biology will likely play a role in developing

healthy bacterial communities, with designer bacterial ecosystems engineered to

improve human and environmental health. However, before we can have the

domesticated biotechnology that scientists like Freeman Dyson predict4, we must first

re-domesticate the microbes that have evolved with us over many thousands of years.

Cheesemaking, microbial ecosystems, and biotechnology each present examples

of complex mixed cultures. All bring diverse groups of lifeforms together into intricate

ecologies of competition and collaboration, impacting our culture and our environment.

Heather Paxson, an anthropologist who studies the microbial politics of artisanal

133
cheesemaking writes on the interactions between human cultures and microbes, “To

speak doubly of cheese cultures—bacterial and human—is thus no idle pun.”1 What can

these different cultures offer each other? Can scientists and biological engineers learn

from human cultures as readily as they do from microbial cultures? Indeed, can those

oft-battling “two cultures”5 of the arts and sciences work together through something

as simple as cheese to ease the friction at the interface of human and bacterial cultures?

Cheesemaking

Cheese begins when Lactobacillus bacteria naturally present in raw milk or added

as a starter culture break down lactose into high concentrations of lactic acid. The low

pH curdles the milk, separating the liquid whey from the curds made of the milk’s fat

and protein. Rennet, an enzyme mixture found in the stomach lining of young veal and

certain types of mold, is added to further break down the milk proteins, hardening the

curds so that they can be pressed, washed, aged, and processed.

Different cheeses are distinguished by the source and quality of the milk (cow,

sheep, goat; grass-fed, raw, low-fat) and how they are aged and processed, but they can

also be separated by the microbes involved in their production. Swiss cheese’s

characteristic holes come from carbon dioxide produced by the bacteria

Propionibacterium freudenreichii rather than L. lactis, while different species of fungus

134
Figure 5.1 Different cheese types on display at Formaggio Kitchen, Cambridge MA. Photo by the author.

from the genus Penicillium give us not only the antibiotic penicillin, but also provide

blue cheese with its stinky blueness and brie cheese its soft white rind.

Most cheese rinds don’t just harbor single species, however, but complex

biofilms, communities of bacteria and fungi that deposit themselves on the cheese

surface and grow in dense layers as the cheese ages in dark, humid caves. Cheeses are

washed, brined, and stored in different ways to cultivate unique microbial communities

and thus unique cheese flavors.

135
Microbial communities

In these and other microbial communities we see most clearly that no microbe is

an island. Bacteria and fungi in the cheese rind communicate with each other and share

nutrients in intricate ways that we are only beginning to understand. Beyond cheese,

every surface around us—the soil, air, and water, and even our own bodies—is host to

complex ecosystems of interacting microbes. Cheese rind offers a simplified system in

which to study complex microbial interactions, a model that Rachel Dutton, a Bauer

Fellow at the Harvard FAS Center for Systems Biology, is using in order to uncover

details of how microorganisms cooperate in nature. With only tens of species working

together instead of the hundreds or thousands that could be present in more complex

environments, cheese rind offers us the ability to combine different scales of

experimental understanding. Large-scale gene sequencing can be mixed with more

classical microbiological experiments on one or more bacterial species working in

isolated cultures, providing a deeper understanding of the biology of microbial

communities.

Studying bacteria in pure cultures has allowed microbiologists to untangle

thousands of the chemical reactions that make up the cell’s metabolism over many

decades. Despite exhaustive study, however, almost one third of the more than 4000

genes in the E. coli genome still have unknown functions6. Many of these genes seem to

be entirely unnecessary to the cells growing in isolated laboratory conditions—single

136
deletions from the E. coli genome has no effect on how well the cells can grow in rich

media7. It’s likely that many of these seemingly unnecessary genes are actually used by

the bacteria in their more natural context, surrounded by, competing, and

communicating with hundreds of other microbial strains and species8.

99% of these other strains can’t be isolated and grown in culture at all: they

need the dynamic microbial environment to live, making it hard to understand how they

are individually contributing to the microbial ecosystem. Metagenomic sequencing of

the microbes present in places as diverse as the Sargasso Sea9 to the human elbow

crease10 has allowed us to identify much of the microbial diversity in these

environments but many ecological details are missing. Projects from systems biology

like Dutton’s analysis of cheese can add a valuable layer of complexity to what we can

learn from sequencing alone.

Synthesizing mixed cultures

Synthetic biology too can address the complexities of how microbes work

together in mixed cultures. “Synthetic” can refer to things that aren’t found in nature

but also can refer to how those things are made; synthesis brings two or more things

together to make something new. Synthetic biology experiments putting cells or cell

components together in a new biological context can provide us with clues about how

biological systems work in nature or provide tools for new biological experiments.

137
Wendell Lim and Michael Elowitz address this new frontier for the study of biology in

“Build life to understand it” a recent commentary in Nature11:

Conventionally, biologists have sought to understand life as it exists.

Increasingly, however, from stem-cell reprogramming to microbial
factories, researchers are both describing what is and exploring what
could be. An analogous shift occurred in physics and chemistry,
especially in the nineteenth century. Like biology, these fields once
focused on explaining observed natural processes or material, such as
planetary motion or ‘organic’ molecules. Now they study physical and
chemical principles that govern what can or cannot be, in natural and
artificial systems, such as semiconductors and synthetic organic
molecules.

Systems and synthetic biology can thus inform each other, developing stronger

models of complex biological systems as part of the design cycle12. When probing

microbial cooperation, mathematical models from systems biology can be used to

understand how synthetic mixtures of different strains of bacteria can share

metabolites in a harsh environment, providing experimental data that can strengthen

the foundational models13. By bringing together different engineered strains and even

different species we can explore how microbes communicate in nature and create new

functions impossible for species working alone14.

Moreover, as a relatively new field that combines the efforts of engineers and

biologists, synthetic biology itself shows how complex mixtures of academic cultures

can potentially lead to something larger than the sum of its parts. Like different

bacterial species that each contribute a unique ability to the function of the community,

each researcher brings their own viewpoint, their own approach to the development of

138
the field. These different viewpoints make synthetic biology what it is, but can also lead

to “culture clashes” as different groups learn to communicate and work together. Lim

and Elowitz describe such clashes in their article:

Although traditional disciplinary boundaries are dissolving, the cultural

differences between scientists and engineers remain strong. For
biologists, genetic modification is a tool to understand natural systems,
not an end in itself. Thus, making biological systems ‘engineerable’ — a
goal of engineers in the field of synthetic biology — can seem pointless.
Many biologists wonder why engineers fail to appreciate the intricate,
beautiful and sophisticated designs that occur naturally. Engineers are
often equally perplexed by biologists. Why are they so obsessed about the
details of one particular system? Why don’t they appreciate the value of
replacing a complex and idiosyncratic system with a simpler, more
modular and more predictable alternative? These misunderstandings can
make for fascinating conversations, but they can also prevent mutually
beneficial synergies.

Here too we can learn from microbial communities, where competition plays an

important role. Spirited debates on what “counts” as synthetic biology and what

research will be most valuable can be useful for distinguishing the new field and

developing a strong research program, only as long as we don’t let such debates distract

from positive work being done by members of the group. Often these debates also

highlight how difficult it is to separate out individual strands from a complex

community. It’s not just engineers on one side and biologists on the other but rather all

sorts of blurred in-betweens—in between science and technology, pure and applied

research, organic and electronic. C.P. Snow warns us in The Two Cultures and the

Scientific Revolution “The number 2 is a very dangerous number: that is why the dialectic

139
is a dangerous process. Attempts to divide anything into two ought to be regarded with

much suspicion.”5 Splitting a complex issue into just two opposing and independent

factions can be as dangerous and limiting as a biological monoculture. By encouraging

debate and collaboration from many sides and intermediate interests we can build

stronger communities.

Indeed, engineers and biologists aren’t the only people with strong and

complicated interests in the future of synthetic biology. With the recent publication of

the President’s Bioethics Commission report on synthetic biology, the future of this

new field and its technological, economic, political, social implications are being

discussed from many points of view. How does synthetic biology fit into that

suspiciously binary split of science and culture? How can we incorporate other people’s

voices and concerns into new scientific and technological developments? How can we

design a positive and just synthetic biology for the future?

Synthetic Aesthetics

I had the tremendous opportunity to work explicitly in-between the two cultures

of art and science in order to address some of these questions about future of synthetic

biology. As a Synthetic Aesthetics resident I spent a month learning and working with

artists, designers, and social scientists, trying to find a common ground from which to

build a better synthetic biology.

140
 Synthetic Aesthetics residencies pair synthetic biologists with artists for a

month of work in the lab and the studio, and I had the pleasure of working with Sissel

Tolaas, someone who describes herself not as an artist, but as a “professional in-

betweener.” Her work on smell, how we communicate about and through odors is as

much chemistry as it is art. She combines a powerful ability to identify smells with

specialized knowledge of what combination of molecules will exactly replicate that

particular complex scent. Together in our individual lab spaces we explored the ways

that we both isolate and recreate the natural world through biological or chemical

means, our methods, our goals, our intentions.

In cheese we found a perfect “model organism.” Stinky and full of bacteria,

cheese has a lot to offer someone who studies difficult smells and someone who studies

bacteria. Cheesemaking is itself a culture/science hybrid, an art form forged out of

biological materials, creating a cultural object treasured by culinary cultures through

millennia.

As a scientific “model organism,” not only does cheese provide a simplified

microbial community in which to study bacterial interactions, but cheeses closely

resemble a simplified human microbiome. Milk curdling lactic acid bacteria are common

on the insides of humans, in the mammalian gut and in raw milk, while the rinds of

many stinky cheeses are washed with salt water during aging to cultivate microbes

suited to the salty and moist environment of human skin. Descriptions of human body

141
 Figure 5.2 Cheeses isolated for smell analysis. Photo by the Pablo Schyfter, reproduced with permission.

odors often overlap with those of cheese15; Propionibacterium used to make Swiss cheese

is a major contributor to the smell of the human armpit16 and Limburger cheese offers a

remarkably close substitute for the smell of human feet, an attractant for certain species

of mosquito17.

We were fascinated by the similarities between cheese and human

microbiodiversity and curious about the historic origin of cheese microflora. Given the

physicality of cheesemaking (figure 5.3), we speculated on the human origins of many

of the unique cheese flavors. To explore this hypothesis and to foreground the

microbiology of our food and bodies, we sought out to make cheeses with starter

cultures isolated from the human body. Swabs from hands, feet, noses, and armpits were

142
Figure 5.3 Cheese curds being mixed by an artisanal cheese maker in Vermont. Photograph by Rachel
Dutton, reproduced here with permission.

inoculated into fresh, pasteurized, organic whole milk (figure 5.4) and incubated

overnight at 37° Celsius. The milk curds were then strained and pressed, yelding unique

smelling fresh cheeses (figure 5.5). Eight cheeses were produced in total for further

study, with bacterial origins from the bodies of the Synthetic Aesthetics team.

The odor and flavor of different cheeses emerges from a complex interaction of

the proteins and enzymes in milk, the metabolism of the starter culture bacteria, and

the population of bacteria and fungi involved in the ripening18. In modern industrial

cheese production these species are often added intentionally as carefully balanced

143
Figure 5.4 Milk incubated with swabs of bacterial cultures from the human body

Figure 5.5 Human cheese. Unique cheese made by straining milk curdled through the metabolism of
bacteria isolated from different body parts.

144
starter cultures to pasteurized milk, but historical and artisanal cheese production allow

(ed) for a wide range of species, milk quality, and flavors19. Variety in cheese production

emerged as a result of geographic distinction, with different regions producing their

own “authentic” cheese20, linking unique combinations of the terroir of the milk, the

craft of the cheese production, and bacterial populations that remains powerful even in

the face of industrialized food production1.

Our cheeses are no different, and though they were made from the same milk

using the same processes, varied widely in texture, color, and odor due to their different

microbial sources. The cheeses were at once artistic and scientific objects, challenging

the observer to confront the microbiological aspects of their food and their body, while

offering a unique medium in which to study the interactions of microbes and the

volatile compounds that they produce. We thus analyzed the cheeses from both artistic

and scientific points of view, breaking down the microbial populations and the odor

profiles of each cheese. Volunteer human noses gave detailed descriptions of the

cheeses’ smells (table 5.1) and the smells of bacterial cultures isolated from each cheese

(table 5.2). To achieve a wide descriptive range of odors we solicited opinions from

profession odor artists, cheese store employees, and colleagues from both art and science

(n=15).

145
 Source Bacteria Isolated Odors
Hand-1 Providencia vermicola yeast, ocean salt, sour old
Morganella morganii cheese, feet
Proteus mirabilis

Foot-1 Providencia vermicola sweat, big toe nail, cat feet,

Morganella morganii sweet, milky, orange juice in the
Proteus mirabilis fridge too long, fungus, buttery
cheese, soapy, light perfume

Armpit-1 Providencia vermicola Feta cheese, turkish shop,

Morganella morganii nutty, fruity, fishy
Proteus mirabilis

Nose-2 Providencia vermicola cheesy feet, cow, cheese

Morganella morganii factory, old subway station,
Proteus mirabilis toilet cleaner

Armpit-2 Enterococcus faecalis neutral, perfumed, industrial,

Hafnia alvei synthetic, fermentation, car
pollution, burning, sharp,
chemical

Armpit-3 Micobacterium lactium neutral, sour, floral, smooth,

Enterococcus faecalis yogurt
Bacillus pumilus
Bacillus clausii

Foot-5 Providencia vermicola yeast, jam, feet, putrid, sour,

Proteus mirabilis rotten

Armpit-4 Enterococcus faecalis yogurt, sour, fresh cream,

butter, whey

Table 5.1 Cheeses, bacteria, and odors. The eight cheeses of human origin described, the bacterial species
isolated to single colonies on LB from the cheese, and the odor descriptions of each one.

Cheese smellomics

We analyzed the bacterial community thriving in the cheeses with 16S ribosomal

RNA sequencing (table 5.1). Samples of each of the eight final cheeses were streaked

onto LB plates and different species were colony purified. Genomic DNA was isolated

using the Qiagen DNEasy kit and 16S RNA was PCR amplified with universal primers

(forward: 5’ GGT TAC CTT GTT ACG ACT T 3’, reverse: 5’-AGA GTT TGA TCC

146
TGG CTC AG-3’). Approximately 1.3 kilobase fragments were gel purified and

sequenced to identify species origin. We identified several species of bacteria:

Providencia vermicola, Morganella morganii, Proteus mirabilis, Enterococcus faecalis

Hafnia alvei, Micobacterium lactium, Bacillus pumilus, and Bacillus clausii. Many of the

identified species have been found in metagenomic sequencing of isolates from the

human body, as well as in standard cheeses (table 5.2). In particular, Proteus vulgaris,

closely related to P. mirabilis, is found on cheeses and noted for its strong aroma, E.

faecalis is a lactic acid bacteria commonly found in raw milk and cheese, and H. alvei is an

enterobacteria common in cheese and added as a secondary culture in certain artisanal

cheeses for a cauliflower-like flavor. B. pumilus has been identified in cheese spoilage, as

has M. lactium, an actinobacterium that is also commonly found on unspoiled cheese and

closely related to bacteria commonly found on washed rind cheeses (Rachel Dutton,

personal communication). The cross-over between bacteria found offering a tantalizing

hint at how our bacterial symbionts have come to be part of our culinary cultures, how

bacterial and human cultures co-evolve.

Of the bacteria isolated from the cheeses, P. mirabilis had the most powerful

unpleasant odor (table 5.2). Cheeses that did not contain P. mirabilis were identified as

the nicest and most neutral smelling, all of which were of armpit origin. The diversity

we observed in isolated colonies, however, was not sufficient to explain the difference in

the smells between cheeses. Several of the cheeses had identical bacterial populations

147
but profoundly different odors. Many species that are present on the human body will

not be able to grow in the milk and cheese, and furthermore, many of the species that

can grow together in the milk will not grow in isolated colonies on LB agar, creating a

bottleneck in the identification of the microbial diversity. Deeper sequencing from

whole genomic DNA isolated from the cheeses themselves will likely identify much of

the missing diversity from these preliminary experiments.

Before the advent of facile gene sequencing technology, identification of

microbial species was challenging, and impossible for the vast majority of unculturable

microbes. In the 1960s there was significant effort made to identify bacterial species by

the identify of the volatile compounds they produced, in particular as a diagnostic of

bacterial infection21.While many of the volatile compounds identified in GC/MS

headspace analyses are common to multiple species, there are often unique traces that

allow for the identification of individual species based on scent22,23. While the clinical

value has diminished, the importance of bacterial volatile in flavor production in wine

and cheese production, there has been significant work in the identification 24 and

cataloguing25 of bacterial odors26.

GC/MS headspace analysis was performed on four of our cheeses by

collaborators at International Flavors and Fragrances to identify the chemical

composition of the cheese volatiles (Figure 5.6, GC traces provided as Appendix A

figure S5 and identified compounds are indexed in table S2). This “electronic nose”

148
technology is practiced by perfumers, flavorists, and food scientists to decompose,

identify, and recreate naturally occurring odors. Tolaas’s work in recreating smells of

cities, spaces, and bodies, employs a mixture of this technical identification of odorants

and biological identification and subjective descriptions by people with trained and

untrained noses. As in other synthetic disciplines, the recreation of complex odors and

comparison between the real odorant and the synthetic model27 will verify the accuracy

and strength of the modeling technology. Synthetic chemical models of the odor of

Swiss cheese verify the identification of a handful of compounds that make up the

primary odor and flavor of Swiss cheese28.

Many of the compounds that specifically contribute to the odor of different cheeses

has been analyzed using gas chromatography headspace technology29 (extensively

reviewed in Urbach18). Fresh cheese odors include strong doses of diacetyl and

acetaldehyde18, while several ketones and alcohols contribute to the smell of different

types of mature cheeses29. Several such compounds appeared in our volatile headspace

analysis, with the most pleasant and cheese-like smelling cheese, Armpit-3, containing

the most cheese-associated ketones. Furthermore, both hand-1 and foot-5 contained

isovaleric acid, a compound found in Swiss cheese30 and human axillary odor 16.

149
 Bacteria Appearance Odors Also found
Providencia vermicola Shiny white colonies sharp, vinegar, chlorine, Gastrointestinal Tract
swimming pool, sweet,
floral, tulip

Morganella morganii Shiny yellow colonies E. coli, pungent, rotting Skin, Airways, predatory
fish, dog breath, barn, ground beetle digestive
monkey house at the tract, wallaby cloaca,
zoo frog skin, pea aphid,
metal working fluids and
aerosols, histamine
production in cheese

Proteus mirabilis/ Fast-moving biofilm that Putrid, foul, Urogenital tract, skin,
vulgaris creates bullseye airways, swine manure,
appearance when cheese volatiles
spread over petri dish.

Enterococcus faecalis Small white colonies LB, not much of a smell, Blood, diabetic wound
chlorine, pool bathroom microbiota, raw milk,
cheese

Hafnia alvei Dense and fluffy streakssour, salty, corn tortillas, Gastrointestinal Tract,
old leather couch, human skin
musty, gym mats microbiome, feces of
the pygmy loris, yellow
catfish stomach,
cauliflower flavor
additive for cheese
production

Micobacterium lactium bright yellow small hard crumbly cheese Skin, Irish washed-rind
opaque colonies cheese

Bacillus pumilus fluffy yellow deep fried chicken, fried Soil, cheese spoilage
fat, cheedar cheese,
cheese-its, brie cheese

Bacillus clausii white colonies stinky cheese, stinging, Skin

bleach, alcohol

Table 5.2 Description of appearance and odor of bacterial species isolated from the human cheeses and
identification of other locations where these species have been found through metagenomic sequencing
analysis.

150
A

Figure 5.6 Comparative smell-omics of the four cheeses analyzed with headspace technology. A.) The
largest difference in odor and bacterial diversity was between Armpit-3 (pink) and Foot-5 (blue). Armpit-3
had the most pleasing, cheese-like odor, and also contains the largest number of ketones previously identified
as being involved in cheese odor (see Appendix A, table S2 for details). B.) Hand-1 (yellow) and Nose-2
(green) were the most similar in odor and had identical bacterial content, but still had many differences in
volatile compounds identified by headspace GC/MS.

151
 Coupled with current metagenomic sequencing technologies, analysis of the

volatile compounds produced by bacteria in isolated or mixed cultures can be used to

elucidate complex metabolic and biosynthetic pathways. Currently only a small number

of the biosynthetic routes for bacterial volatile compounds is known completely, in

particular that of geosmin, a compound responsible for the earthy and musty smell of

cellars, as well as off flavors in contaminated water and wines and the peaty flavor of

whisky31. Metagenomic sequencing can contribute to an understanding of community

metabolism, and linking this information to analysis of volatile compound metabolomics

can improve our understanding of global metabolism and ability to engineer novel

flavors.

Flavors and fragrances are emerging as a valuable application of synthetic

biology, where enzymatic production of a wide range of chemicals and compounds can

be produced through more environmentally conscious biosynthetic conversion of

glucose rather than chemical conversion of petrochemicals32. Several iGEM teams have

worked towards producing fragrant compounds in E. coli, including MIT’s production

of wintergreen and banana scent33 and Art Science Bangalore’s attempt at producing

geosmin, also responsible for the romantic smell of fresh rain. Synthetic biology has

also taken advantage of the power of volatiles as agents of inter- and intra-species

communication34. Bacterial volatiles have been shown to limit growth of other bacterial

species, fungi22,35, and plants31. Other volatiles have been shown to promote growth of

152
Arabidopsis in some cases36, influence cytokine production in mammalian tissues37.

Moreover, “bacterial olfaction” was recently discovered in certain species of Bacillus

where volatile ammonia produced by one strain affected biofilm formation in another38.

Engineering of bacteria and yeast to produce volatile chemicals and other strains with

volatile responsive promoters allows for the design of synthetic interkingdom

mutualistic relationships, with great potential for the establishment and synchronization

of multispecies synthetic devices34.

Conclusions

Having broken down and analyzed the mixed cultures of our cheeses, we can

begin to re-assemble the parts and re-synthesize the questions we started with. Will

cheese or the way we eat it change as we learn more about microbial communities and

can better engineer them? Will our relation to our food and our bodies change with an

increased appreciation for the millions of non-human cells that make up our personal

ecosystem? Can we design biology better with an appreciation for all the mixed cultures

involved, both human and microbial?

153
References

1! Paxson, H, Post-Pasteurian Cultures: The Microbiopolitics of Raw-Milk

Cheese in the United States. Cultural Anthropology 23 (1), 15 (2008).
2! Kaper, James B and Sperandio, Vanessa, Bacterial cell-to-cell signaling in
the gastrointestinal tract. Infection and immunity 73 (6), 3197 (2005).
3! Cerf-Bensussan, Nadine and Gaboriau-Routhiau, Valérie, The immune
system and the gut microbiota: friends or foes? Nature Publishing Group
10 (10), 724 (2010).
4! Dyson, F, Our biotech future. New York Review of Books 54 (12) (2007).
5! Snow, CP and Percy, C, The two cultures and the scientific revolution.
(1960).
6! Keseler, I M et al., EcoCyc: A comprehensive view of Escherichia coli
biology. Nucleic Acids Research 37 (Database), D464 (2009).
7! Baba, Tomoya et al., Construction of Escherichia coli K-12 in-frame,
single-gene knockout mutants: the Keio collection. Molecular Systems
Biology 2, 2 (2006).
8! Wintermute, E and Silver, P, Dynamics in the mixed microbial concourse.
Genes & Development 24, 2603 (2010).
9! Venter, J Craig et al., Environmental genome shotgun sequencing of the
Sargasso Sea. Science (New York, N.Y.) 304 (5667), 66 (2004).
10! Grice, E A et al., Topographical and Temporal Diversity of the Human Skin
Microbiome. Science (New York, N.Y.) 324 (5931), 1190 (2009).
11! Elowitz, M and Lim, W, Build life to understand it. Nature 468, 889 (2010).
12! Smolke, Christina D and Silver, Pamela A, Informing Biological Design by
Integration of Systems and Synthetic Biology. Cell 144 (6), 855 (2011).
13! Wintermute, Edwin H and Silver, Pamela A, Emergent cooperation in
microbial metabolism. Molecular Systems Biology 6, 1 (2010).
14! Brenner, K, You, L, and Arnold, F, Engineering microbial consortia: a new
frontier in synthetic biology. Trends in biotechnology 26 (9), 483 (2008).
15! Troccaz, M et al., Gender-Specific Differences between the
Concentrations of Nonvolatile (R)/(S)-3-Methyl-3-Sulfanylhexan-1-Ol and
(R)/(S)-3-Hydroxy-3-Methyl-Hexanoic Acid Odor Precursors in Axillary
Secretions. Chemical Senses 34 (3), 203 (2008).
16! Leyden, JJ et al., The microbiology of the human axilla and its relationship
to axillary odor. Journal of Investigative Dermatology 77 (5), 413 (1981).
17! Knols, B G and De Jong, R, Limburger cheese as an attractant for the
malaria mosquito Anopheles gambiae s.s. Parasitology today (Personal
ed.) 12 (4), 159 (1996).
18! Urbach, G, The flavour of milk and dairy products: II. Cheese: contribution
of volatile compounds. International Journal of Dairy Technology 50 (3), 79
(1997).
19! Shapin, S, Cheese and Late Modernity. London Review of Books, 1
(2008).
20! Olson, NF, The impact of lactic acid bacteria on cheese flavor. FEMS
Microbiology Letters 87 (1-2), 131 (1990).

154
21! Cherry, WB, The role of gas chromatography in the clinical microbiology
laboratory. The Journal of Infectious Diseases (1969).
22! Kai, Marco, Effmert, Uta, Berg, Gabriele, and Piechulla, Birgit, Volatiles of
bacterial antagonists inhibit mycelial growth of the plant pathogen
Rhizoctonia solani. Archives of Microbiology 187 (5), 351 (2006).
23! Henis, Y and Gould, JR, Detection and identification of bacteria by gas
chromatography. Applied and Environmental Microbiology 14 (4) 513
(1966).
24! Dickschat, Jeroen S et al., Biosynthesis of Volatiles by the Myxobacterium
Myxococcus xanthus. ChemBioChem 5 (6), 778 (2004); Wilkins, K,
Volatile metabolites from actinomycetes. Chemosphere (1996).
25! Dunkel, M et al., SuperScent--a database of flavors and scents. Nucleic
Acids Research 37, D291 (2009).
26! Schulz, Stefan and Dickschat, Jeroen S, Bacterial volatiles: the smell of
small organisms. Natural Product Reports 24 (4), 814 (2007).
27! Elmore, SJ, Thompson, K, and Howard, C, Comparison of the aroma of a
food with its gas chromatographic headspace profile using multivariate
analysis. Food Quality and Preference 5, 151 (1994).
28! Preininger, M and Warmke, R, Identification of the character impact flavour
compounds of Swiss cheese by sensory studies of models. Z Lebensm
Unters Forsch 202, 30 (1996).
29! Arora, G, Cormier, F, and Lee, B, Analysis of odor-active volatiles in
Cheddar cheese headspace by multidimensional GC/MS/sniffing. Journal
of Agricultural and Food Chemistry 43 (3), 748 (1995).
30! Thierry, A, Maillard, MB, and Yvon, M, Conversion of L-leucine to
isovaleric acid by Propionibacterium freudenreichii TL 34 and ITGP23.
Applied and Environmental Microbiology 68 (2), 608 (2002).
31! Dunkel, M et al., Bacterial volatiles and their action potential. Nucleic Acids
Research 37 (Database), D291 (2009).
32! Misawa, Norihiko, Pathway engineering for functional isoprenoids. Current
Opinion in Biotechnology, 1 (2011).
33! Smolke, Christina D, Building outside of the box: iGEM and the BioBricks
Foundation. Nature Biotechnology 27 (12), 1099 (2009).
34! Weber, Wilfried, Daoud-El Baba, Marie, and Fussenegger, Martin,
Synthetic ecosystems based on airborne inter- and intrakingdom
communication. Proceedings of the National Academy of Sciences of the
United States of America 104 (25), 10435 (2007).
35! Wheatley, R E, The consequences of volatile organic compound mediated
bacterial and fungal interactions. Antonie van Leeuwenhoek 81 (1-4), 357
(2002).
36! Ryu, Choong-Min et al., Bacterial volatiles promote growth in Arabidopsis.
Proceedings of the National Academy of Sciences of the United States of
America 100 (8), 4927 (2003).
37! Kurita-Ochiai, T, Fukushima, K, and Ochiai, K, Volatile Fatty Acids,
Metabolic By-products of Periodontopathic Bacteria, Inhibit Lymphocyte

155
 Proliferation and Cytokine Production. Journal of Dental Research 74 (7),
1367 (1995).
38! Nijland, Reindert and Burgess, J Grant, Bacterial olfaction. Biotechnology
Journal 5(9) 974 (2010).
39! Haddad, Rafi et al., Predicting Odor Pleasantness with an Electronic
Nose. PLoS Computational Biology 6 (4), e1000740 (2010).

156
Chapter 6
/Conclusions

“So it's the first living self-replicating cell that we have on the planet whose
DNA was made chemically and designed in the computer. So it has no genetic
ancestors. Its parent is a computer.”
-Craig Venter, on the first chemically synthesized genome

"The price of metaphor is eternal vigilance."

-Arturo Rosenblueth and Norbert Wiener
 In creating a new discipline, synthetic biologists have used a variety of analogies

and metaphors to firmly ground the field in the language of engineering and computer

science. Standardizable, modular biological components are isolated and recombined in

cellular “chassis,” typically Escherichia coli, creating novel devices and circuits that with

functions reminiscent of electrical engineering components. In particular, biological

circuits have been built to behave like toggle switches,1 oscillators,2 memory loops,3 and

logic gates4.

DNA sequences are at once the software and the hardware of this new

engineering discipline, the blueprints and the building materials of synthetic biology

devices. BioBrick parts are physical objects that are manipulated, characterized,

standardized, and shared, but they are also codes that translate to a function when they

are inside of a living cell. Nearly 70 years ago, before the structure of DNA was know,

quantum physicist Erwin Schroedinger described this duality of the genetic material in

his meditation on the physics of biology, What is Life? He writes that the chromosomes

“are law-code and executive power—or, to use another simile, they are architect’s plan

and builder’s craft—in one”5.

Evelyn Fox Keller traces this complex history of the language and metaphors

used to describe DNA in her book Refiguring Life: Metaphors of Twentieth-Century

Biology. As she follows the development of the life sciences in the twentieth century

from organismal to molecular to systems biology, she identifies how the metaphors that

158
surround biological experimentation shaped the course of research and vice-versa. She

writes: “Scientists usually assume that only their data and theories matter for scientific

progress, that how they talk about these data and theories does not matter, that it is

irrelevant to their actual work. But in introducing this particular way of talking, the

first generation of American geneticists provided a conceptual framework that was

critically important for the future course of biological research”6. Discussion of the

action and function of genes to create life drove the primacy of genetics and molecular

biology in the life sciences. The focus on genotype led to many fundamental discoveries,

but created a paradigm that made it difficult to study the biological contexts and

networks in which those genes are embedded, the contexts that gave the genes their

function. While new DNA nanotechnology experiments give these codes their own

function, without a living cell DNA is mostly inert.

Recent advances in systems biology have fostered an emphasis on a more holistic

model of cells, with genes interacting with other cellular components in complex and

context-dependent networks to create the behaviors of a living cell. However, many of

the old tensions that Keller describes remain fresh in synthetic biology, and nowhere is

this easier to see than in the language surrounding the recent announcement that the J.

Craig Venter Institute had synthesized a full bacterial genome that was able to “boot up”

when transformed into a closely related species7. In news reports, talk of the “synthetic

cell” and even “synthetic life” overtook the more technically precise term “chemically

159
synthesized genome.” The genome became the defining object of life; in this context,

only the code is celebrated, not the host cell that gave it life.

The emphasis on code and on biology as information makes analogies of

synthetic biology to computer science all the more compelling. Craig Venter, in an

interview about the synthetic genome with CNN describes the work by saying “It is

software.. It’s DNA software”8. In the extended analogy of synthetic biology to

computer engineering, hierarchies of biological parts correspond to hierarchies in

computer design, both hardware and software. The hardware is the physical layer of

proteins and genes imagined as transistors that can be assembled into logic gates, all the

way up to complex interaction of cells in tissues and cultures as interacting nodes in the

world wide web9. The software layer of the gene as code runs underneath: the genome

is the operating system, the gene is a module within that larger code.

Metaphors can be powerful tools to help in understanding and communicating

complex systems, but it is important to understand the limits of metaphor, to not

confuse one object for another. When the genome is recast as software made of modular

bundles that can be swapped, computational design with tools such as TinkerCell10 or

GenoCAD11 can gain popularity, with almost no biological evidence that the designs will

work in their physical (“hardware”) manifestation inside of a cell. In many cases, the

abstraction of biological data has gone so far as to obscure all biological reality. The

160
metaphor of DNA as architect and builder has progressed to the extent that it is

neither.

In looking to previous technological developments for metaphors and analogies

that can apply to synthetic biology, we must remember not only the mature technologies

where standards and principles direct the predictable, often computational, combination

of off-the-shelf parts into new devices, but the long and context-dependent

technological histories from which those parts and principles arose. The history of

flight offers just such an analogy. In airplane design trial and error and deep analysis of

failed attempts, models, and prototypes in field tests and wind tunnels led to the

evolution of functional models of air dynamics and turbulence. These models today

allow for the entirely computational design of incredibly complicated machines made of

thousands of parts12. While it is possible that one day biology will follow a similar

trajectory, the path biological engineers will follow and the models and design principles

we will use will by necessity be different and likely unique to biology.

In biology, these new models are currently being built at the intersection of

synthetic biology and systems biology13. The picture that emerges from these cycles of

design, experimentation, and modeling is not one of rigid, computational-like hierarchy,

but of complex, dynamic, and context-dependent systems. Sometimes the emergent

behavior of isolated genetic systems is easily modeled, understood, and redesigned, as in

the many successful instantiations of synthetic biology devices. Where these systems

161
fail is typically not at the scale of the “parts,” the interacting behavior of the

transcription factors and enzymes themselves, but on the scale of the “chassis,” the

complexities and cross-talk of the whole cell that is needed to turn those parts into

functional “devices.” The projects presented in this dissertation have attempted to

identify and utilize design principles at the transition between scales, to ground

biological design in evolutionary and ecological principles and strategies.

The hydrogen production and consumption pathways designed in Chapter 2

most closely reflect the synthetic biology ideals of abstraction, modularity, and circuit

design. Protein domains that transfer electrons are modules that can be swapped

between organisms and between enzymes, functioning in their new contexts as parts of

a larger whole. I use the word “circuit” as a metaphor to describe these pathways within

the context of synthetic biology, but the metaphor here also reflects the electrical

nature of these biological systems. These circuits are dependent on their host cell not

only for expression, but also for a connection to the cellular metabolism as a source or

sink of the electrical potential running through them. While they must link to the rest

of the cell’s metabolism, the pathways must also be as orthogonal as possible to prevent

“short circuiting” and ensure the highest possible function of the enzymatic pathway.

We tested “insulation” strategies for these synthetic circuits that were based on

the evolution of metabolic or signaling pathways made of redundant and cross-reacting

components—spatio-temporal control of competing reactions through scaffolding or

162
deletion or improvement of the interaction between proper partners through mutation

of the protein-protein interaction surfaces or direct fusion of the interacting partners

into one polypeptide chain. All of these methods significantly affected the levels of

hydrogen produced by the synthetic pathway and demonstrated opportunities for

control of synthetic biological systems in vivo.

The information that emerged from our synthetic pathways also fed back into

our basic understanding of the function of the hydrogenase in the cellular context. In

attempting to isolate the hydrogen consumption pathway from cellular metabolism we

identified a hydrogen-independent electron transfer function for the ferredoxin-like

domain of the clostridial hydrogenases. This latent function is likely a consequence of

the ancient fusion of ferredoxin-like electron transfer domains with the catalytic domain

of the hydrogenase, showing us how recombination between protein domains can have a

large impact on enzyme function.

A new understanding of   the evolution of   novel biological forms in the early

twentieth century inspired Jacques Loeb to take on engineering biology. He wrote in

1912, “It was perhaps not the least important of   Darwin’s services to science that the

boldness of   his conceptions gave to the experimental biologist courage to enter upon

the attempt of   controlling at will the life phenomena of   animals, and of   bringing about

effects which cannot be expected in nature”14. Evolution still has an impact on what

163
synthetic biologists can imagine possible today, at the scale of protein domains or whole

cells.

In the serial endosymbiosis theory, whole bacterial cells are the units of

evolutionary change. Engulfment of a bacteria and subsequent exchange of genetic and

metabolic materials established the energy harvesting eukaryotic organelles—the

mitochondria, chloroplast, and hydrogenosome. In Chapter 3 we attempted to re-create

the establishment of the chloroplast by engineering the relationship between the

cyanobacteria Synechococcus elongatus and eukaryotic cells. New endosymbiotic

relationships have the potential to be valuable tools for synthetic biology, as well as

tools for exploring the evolutionary dynamics of symbiosis.

The tools of synthetic biology have also precipitated a tremendous expansion in

the range of people who have access to high level biological engineering technologies.

Programs such as the International Genetically Engineered Machines (iGEM)

competition bring synthetic biology to students around the world, in many cases to

high schools, colleges, and universities that do not have a research faculty or significant

laboratory resources. This broadening of participation in biotechnology inspires many

to think about what will be possible when biological technologies can be practiced

locally at small scales to solve local problems. Chapter 4 outlines the work of the

Harvard iGEM team that I mentored during the summer of 2010, which explored the

possibility of using the dispersed and open source model of iGEM’s synthetic biology

164
for plant engineering. Arabidopsis thaliana was engineered with multiple

“personalizations,” to decrease the expression of allergen proteins, to alter pigment

metabolism, and to express sweet-tasting proteins.

Synthetic biology also opens up fascinating opportunities for truly

interdisciplinary collaboration between biologists and engineers, computer scientists,

physicists, social scientists, policy makers, lawyers, and artists. These collaborations

have been part of the discussions of government regulation of synthetic biology

research, as well as part of imagining the possible futures enabled by synthetic biology

technology. In designing biology for the future, artists and designers can work with

scientists and engineers to question, provoke, and imagine synthetic biology

possibilities. These futures are not bound to the metaphors of any one past technology,

but encourage creativity about the nature and place of technology and the fundamentals

of living cells. David Drubin and Pamela Silver close their 2007 review of synthetic

biology with an equally open-ended look to the future: “Ultimately, the rate-limiting

factor for the future development of synthetic biology may actually be human creativity.

Future ‘technology development’ may, in fact, consist of novel ways of thinking about

life that allow us to build things that are truly new.”15 These positive futures may come

about, but only with evolution and diversity in projects, principles, platforms, and

participants.

165
References

1! Gardner, T S, Cantor, C R, and Collins, J J, Construction of a genetic

toggle switch in Escherichia coli. Nature 403 (6767), 339 (2000).
2! Elowitz, M B and Leibler, S, A synthetic oscillatory network of
transcriptional regulators. Nature 403 (6767), 335 (2000).
3! Ajo-Franklin, Caroline M et al., Evaluation Study, 2007.
4! Win, M N and Smolke, C D, Higher-Order Cellular Information Processing
with Synthetic RNA Devices. Science (New York, N.Y.) 322 (5900), 456
(2008).
5! Schroedinger, Erwin, What Is Life? (Cambridge University Press,
Cambridge, England, 1944).
6! Fox Keller, Evelyn, Refiguring Life: Metaphors of Twentieth-Century
Biology. (Columbia Univeristy Press, New York, 1995).
7! Gibson, Daniel G et al., Creation of a Bacterial Cell Controlled by a
Chemically Synthesized Genome. Science (2010).
8! CNN, Scientist: 'We didn't create life from scratch', Available at http://
articles.cnn.com/2010-05-21/health/venter.qa_1_synthetic-chromosome-
genetic-code-synthetic-life?_s=PM:HEALTH, (2010).
9! Andrianantoandro, Ernesto, Basu, Subhayu, Karig, David K, and Weiss,
Ron, Synthetic biology: new engineering rules for an emerging discipline.
Molecular Systems Biology 2, 14 (2006).
10! Chandran, D and Bergmann, F, Computer-aided design of biological
circuits using TinkerCell. Bioengineered Bugs (2010); Chandran, D,
Bergmann, F, and Sauro, H, TinkerCell: modular CAD tool for synthetic
biology. Journal of Biological Engineering 3 (1), 19 (2009).
11! Czar, Michael J, Cai, Yizhi, and Peccoud, J, Writing DNA with
GenoCAD™. Nucleic Acids Research 37, W40 (2009).
12! Mazzotti, M, Knowledge as social order: rethinking the sociology of Barry
Barnes. books.google.com (2008); CARLSON, R, Biology is technology
the promise, peril, and business of engineering. lavoisier.fr (2009).
13! Smolke, Christina D and Silver, Pamela A, Informing Biological Design by
Integration of Systems and Synthetic Biology. Cell 144 (6), 855 (2011).
14! Deichmann, U, Chemistry and the Engineering of Life Around 1900:
Research and Reflections by Jacques Loeb. Biological Theory 4 (4), 323
(2009).
15! Drubin, David A, Way, Jeffrey C, and Silver, Pamela A, Designing
biological systems. Genes &amp; Development 21 (3), 242 (2007).

166
Appendix A
/Supplementary tables and figures

167
Figure S1

Sequence of codon optimized commercially synthesized genes. Gene names listed

as Organism.Gene Name; GenBank Accession Number (Cr=Chlamydomonas reinhardtii,
So=Spinacia olearcea)

>Cr.HydA1;AAL23572
gaattcgcggccgcttctagagctgcaccagccgcagaagctcctttgtctcatgttcaacaggccttag
ccgagcttgcaaaaccaaaggatgaccctactagaaaacacgtatgtgtccaagtggccccagctgttag
ggtagcaattgctgaaacacttggtttggcccctggagcaaccactccaaagcagttagctgagggccta
agaaggcttggttttgatgaagtgttcgacacattgtttggagccgatttaaccataatggaagagggct
cagaattgttacatagactaactgaacaccttgaggcacatcctcactccgacgaaccattgcctatgtt
cacaagttgctgtccaggttggatcgctatgttagaaaaaagctatcctgatctaattccatacgtgagc
tcatgcaagtcccctcaaatgatgttggccgcaatggttaaaagttatttagctgagaagaaaggtatag
ccccaaaggatatggtaatggtcagcatcatgccatgtaccagaaaacaatctgaagcagacagggattg
gttttgcgttgacgctgatcctactcttagacagttggatcatgtgattacaaccgttgagttaggaaat
atattcaaggaaagaggcatcaacctagccgaacttccagagggtgaatgggacaatcctatgggagtag
gttcaggcgcaggtgtcttgtttggaactacaggcggcgtgatggaagctgctttaaggactgcctacga
gctattcaccggtacaccattgcctagattatcccttagtgaagttaggggaatggatggtattaaagaa
actaacattaccatggtaccagcacctggctctaagtttgaggaattgttaaaacatagagctgccgcaa
gagctgaagccgcagctcacggaacaccaggtcctctagcatgggacggcggtgctggattcactagcga
ggatggtaggggcggcataacattgagagtcgccgttgcaaatggattaggtaacgctaaaaagcttatc
accaaaatgcaagccggcgaagcaaagtatgattttgtggagattatggcttgtccagccggatgtgttg
gtggaggcggacaacctagatcaactgacaaagcaataacacagaagaggcaagctgccctatacaattt
ggatgaaaaatccactttaagaagaagtcatgaaaacccatctatcagggagctttatgacacctacttg
ggtgaacctttaggtcacaaggcacatgaactattgcacacacattatgtagctggcggagtcgaggaaa
aagatgaaaagaaaactagtagcggccgctgcag

>Cr.HydEF;AAS92601
gaattcgcggccgcttctagagctgcacatgcctctgcttcaaaagcaactccagatgttcctgtagacg
atcttccacctgcccacgctagagcagccgtcgccgcagctaataggagagccagggcaatggcttccgc
cgaagcagctgccgagacattaggtgactttctaggacttggcaagggtggattgagtccaggcgcaacc
gctaacttagatagagaacaagtgctaggtgttcttgaggccgtatggagaaggggtgacttgaatttag
aaagagcattgtatagccatgctaacgccgtcactaataaatactgtggaggcggtgtgtattacagagg
attagttgagttctctaacatttgccagaatgattgttcatattgcggtataaggaacaatcaaaaggag
gtatggagatacacaatgcctgtcgaagaagttgtggaggttgcaaaatgggccctagaaaacggcatca
ggaatattatgcttcagggtggagaacttaagaccgagcaaagattagcttacctagaagcctgtgtaag
agcaataagggaggaaactacacaattggatttagaaatgagagctagagccgcatccaccactacagct
gaggccgcagctagtgcacaggctgacgccgaagcaaaaaggggtgaaccagagcttggcgtcgtggtta
gcttgtctgtaggtgaattacctatggaacaatacgagagactatttagagctggagccaggagatatct
tatcaggattgaaacctcaaatccagatttgtacgcagctttacaccctgaaccaatgtcctggcatgcc
agagtcgagtgcctaagaaacttgaagaaagcaggttatatgttaggcactggagttatggtgggccttc
ctggccaaacattgcacgacttagctggtgatgttatgttctttagggatataaaggccgacatgatcgg
aatgggtccattcattactcagcctggcaccccagcaacagataaatggactgctctatacccaaatgct
aacaagaatagtcatatgaaatctatgtttgacttgaccacagccatgaacgcattagtaagaattacta
tgggtaatgtcaacataagcgctacaaccgcccttcaagcaatcattcctactggaagagaaatagccct
agagaggggtgccaatgtggttatgccaatcttgacacctactcagtatagagaatcataccaattatat
gaaggcaagccatgtattaccgatacagcagtacaatgtagaaggtgccttgatatgagattgcattccg
tcggaaaaaccagtgctgccggtgtttggggtgaccctgcatctttcttacacccaatagtgggcgttcc
tgtaccacatgatctatcatctcctgctttggccgcagctgccagcgcagactttcacgaggtcggagct
ggtccatggaaccctatcaggttagaaagacttgttgaagtgccagatagataccctgatccagacaatc

168
Figure S1 continued from previous page
atggtaggaaaaaggccggcgcaggaaaaggcggcaaggctcacgattcccatgacgatggagatcatga
cgatcaccatcaccatcatggtgccgcaccagctggtgccgcagctggcaaaggaaccggtgccgcagct
attggcggcggagccggtgctagcagacagagagtagctggcgccgcagctgcctcagcaaggttgtgtg
ctggagccagaagagcaggtagggtcgttgcttctcctctaagaccagccgcagcttgcaggggtgtggc
cgttaaggcagctgctgccgcagctggcgaggacgccggagcaggtacaagcggtgtaggctccaatatt
gtcaccagtcctggaatagcttcaaccacagcccacggtgttccaagaatcaacattggcgtgttcggag
taatgaatgcaggtaaatctactttagtcaacgctttggcccaacaagaagcatgtatagttgatagcac
ccctggtacaactgctgacgtcaagaccgttcttttagaactacatgcattgggcccagctaaattactt
gatacagccggattggatgaggtaggtggtctaggcgacaagaaaagaaggaaggcattaaatactttga
aagaatgcgatgtcgctgttcttgtggtagacaccgatacagccgcagctgccatcaaatccggaagatt
agcagaggccctagaatgggaaagtaaggtcatggagcaggctcacaaatacaacgtttcacctgtgttg
ttattgaatgtaaagagcagaggccttccagaagcccaagcagccagcatgctagaagccgttgcaggca
tgttagatccttccaaacagattccaaggatgtcattggacttagcttctactcctcttcatgagagaag
tacaataactagcgcctttgtcaaggaaggagcagttaggtcctcaagatacggtgctccactacctggt
tgtttgccaagatggtctttaggcaggaacgccagattgcttatggtgattccaatggatgcagaaaccc
ctggaggtagactattaaggccacaagctcaagtaatggaggaagccatcagacactgggcaacagtctt
gagtgttagattagacttggatgctgccaggggtaaacttggccctgaagcatgtgagatggaaagacag
aggttcgatggagtaattgctatgatggagagaaatgacggtccaactctagttgtgaccgattctcaag
ccatagacgtcgttcatccttggacattagatagatcctcaggcaggccattggtgcctatcactacctt
tagtattgcaatggcttatcaacagaacggaggtagacttgatccatttgtagaaggcctagaagcctta
gagacattgcaagacggcgatagagtcttaatatctgaagcatgcaatcataataggatcacttcagctt
gtaacgacattggaatggttcaaatacctaataagttggaagctgcacttggtggtaaaaagctacagat
tgagcacgctttcggcagagaatttccagaattagagtctggaggtatggatggcttgaaacttgccatc
cattgcggaggttgtatgattgatgcacaaaagatgcagcaaagaatgaaagacctacacgaagctggtg
tacctgttaccaactatggcgtgttctttagctgggccgcatggccagatgctttaaggagagccttgga
accttggggagtcgagcctccagttggtacacctgcaactccagctgccgcacctgctaccgccgcatcc
ggtgtgactagtagcggccgctgcag
>Cr.HydG;AAS92602
gcggccgcttctagaactgctcatggtaaagcatctgccacaagagaatatgctggagattttttgccag
gcaccactatttcacacgcatggtccgttgagagggaaacacatcacagatacaggaatcctgccgagtg
gataaacgaagctgcaatccataaggccttagaaaccagtaaagctgacgcacaagatgctggtagagta
agagagattctagccaaggcaaaagaaaaggctttcgtcactgaacacgccccagtgaatgcagagagca
aatctgaatttgttcagggacttacattggaagagtgtgctaccttaataaacgtagactcaaataacgt
cgaactaatgaatgagatcttcgatactgcccttgcaattaaggaaaggatatatggcaacagagtggtt
ttgtttgctcctttatacatcgccaatcattgcatgaacacatgtacctattgcgcattcagatccgcta
ataaaggtatggaaaggagtattttgactgacgatgatttaagagaggaagtagccgcactacaaaggca
gggtcatagaaggattcttgctttgacaggagaacacccaaagtacacttttgacaatttcttacatgct
gtcaacgttatagccagcgtgaaaaccgagcctgaaggctctatcaggagaattaatgttgaaatcccac
ctctatcagtatccgatatgagaaggttgaagaacacagacagtgtcggtacttttgtgttattccaaga
gacctatcacagagatacatttaaagttatgcatccatctggacctaagagcgatttcgactttagagta
cttactcaagatagggcaatgagagctggtttggacgatgtcggcatcggtgccttatttggactatacg
attataggtacgaagtttgtgcaatgcttatgcactcagaacatttggagagagaatataatgctggtcc
acatacaatttccgtgcctagaatgaggccagccgacggcagtgagttatctatagcacctccataccca
gttaacgatgctgacttcatgaagctagtagcagtcttgagaatcgctgtgccttataccggtatgattt
tatcaactagagaatctccagaaatgaggagcgcccttttgaaatgcggaatgtcccagatgagtgcagg
ttcaagaacagatgttggcgcttaccacaaggatcatactttatctaccgaggccaatctaagcaaattg
gcaggacaatttacattacaagacgaaagacctactaacgaaattgtaaagtggcttatggaggaaggtt
atgtcccatcctggtgtaccgcttgttacaggcagggcagaacaggtgaagatttcatgaatatatgcaa
agccggagacatccacgatttttgtcatcctaacagtctattgactttacaagagtatcttatggattac
gcagacccagatttgaggaagaaaggtgaacaggttattgctagagagatgggccctgacgcctcagaac
cattatctgcacaaagcagaaagaggctagaaagaaaaatgaagcaagtgttggagggtgaacatgatgt
ttatttaactagtagcggccg

169
Figure S1 continued from previous page

>So.Fd;1704156A
gcggccgcttctagagctgcatataaagttactttggtaacaccaaccggtaatgtcgaatttcaatgtc
ctgatgacgtgtacattttagacgccgctgaggaagagggaatagatctaccatattcttgcagagcagg
ctcatgttccagttgcgccggtaagcttaaaactggaagcttgaaccaggatgaccaatctttcttagat
gatgaccagatcgatgaaggctgggttctaacatgtgctgcataccctgtatcagacgtcaccattgaaa
ctcataaggaggaagaacttacagccactagtagcggccg

>Brazzein (BBa_K382020)
ATGCAAGATAAGTGTAAAAAAGTGTATGAGAACTATCCTGTGAGTAAATGCCAATTGGCA
AACCAGTGCAATTATGATTGTAAACTCGATAAGCACGCTAGGAGTGGAGAGTGTTTCTAT
GATGAGAAGAGGAACCTCCAGTGTATCTGTGATTATTGTGAGTAT

>Miraculin (BBa_K382021)
ATGAAGGAGCTTACCATGCTTTCACTTTCTTTTTTCTTCGTGTCTGCTCTTTTGGCTGCT
GCTGCTAACCCTCTTTTGTCTGCTGCTGATTCTGCTCCAAACCCTGTTCTCGATATCGAT
GGAGAGAAATTGAGAACCGGAACAAACTATTATATCGTGCCTGTGCTTAGAGATCACGGT
GGAGGACTCACTGTTAGTGCTACTACTCCAAACGGAACCTTCGTGTGTCCACCTAGAGTT
GTTCAGACTAGGAAGGAAGTGGATCATGATAGACCACTCGCTTTTTTCCCTGAAAATCCT
AAAGAGGATGTTGTTAGAGTTTCTACCGATTTGAACATCAACTTTTCTGCTTTCATGCCT
TGTAGATGGACCTCTTCAACTGTGTGGAGACTCGATAAGTATGATGAGTCTACCGGACAG
TATTTCGTGACTATCGGAGGAGTGAAGGGTAATCCTGGTCCTGAGACTATTAGTTCTTGG
TTTAAAATCGAGGAGTTCTGTGGATCTGGTTTCTATAAACTTGTGTTTTGCCCAACTGTG
TGTGGATCTTGTAAAGTGAAATGTGGTGATGTGGGAATCTATATCGATCAAAAGGGAAGG
AGGAGACTTGCTTTGTCTGATAAGCCTTTCGCTTTCGAGTTCAACAAAACCGTTTATTTC

>pENTCUP2 plant specific promoter (BBa_K382022)

GGGATCTTCTGCAAGCATCTCTATTTCCTGAAGGTCTAACCTCGAAGATTTAAGATTTAA
TTACGTTTATAATTACAAAATTGATTCTAGTATCTTTAATTTAATGCTTATACATTATTA
ATTAATTTAGTACTTTCAATTTGTTTTCAGAAATTATTTTACTATTTTTTATAAAATAAA
AGGGAGAAAATGGCTATTTAAATACTAGCCTATTTTATTTCAATTTTAGCTTAAAATCAG
CCCCAATTAGCCCCAATTTCAAATTCAAATGGTCCAGCCCAATTCCTAAATAACCCACCC
CTAACCCGCCCGGTTTCCCCTTTTGATCCATGCAGTCAACGCCCAGAATTTCCCTATATA
ATTTTTTAATTCCCAAACACCCCTAACTCTATCCCATTTCTCACCAACCGCCACATAGAT
CTATCCTCTTATCTCTCAAACTCTCTCGAACCTTCCCCTAACCCTAGCAGCCTCTCATCA
TCCTCACCTCAAAACCCACCGGA

>NosT plant specific terminator (BBa_K382023)

GATCGTTCAAACATTTGGCAATAAAGTTTCTTAAGATTGAATCCTGTTGCCGGTCTTGCG
ATGATTATCATATAATTTCTGTTGAATTACGTTAAGCATGTAATAATTAACATGTAATGC
ATGACGTTATTTATGAGATGGGTTTTTATGATTAGAGTCCCGCAATTATACATTTAATAC
GCGATAGAAAACAAAATATAGCGCGCAAACTAGGATAAATTATCGCGCGCGGTGTCATCT
ATGTTACTAGATC

>NosT plant specific terminator + stop codon (BBa_K382024)

TGAGATCGTTCAAACATTTGGCAATAAAGTTTCTTAAGATTGAATCCTGTTGCCGGTCTT
GCGATGATTATCATATAATTTCTGTTGAATTACGTTAAGCATGTAATAATTAACATGTAA
TGCATGACGTTATTTATGAGATGGGTTTTTATGATTAGAGTCCCGCAATTATACATTTAA
TACGCGATAGAAAACAAAATATAGCGCGCAAACTAGGATAAATTATCGCGCGCGGTGTCA
TCTATGTTACTAGATC

170
Table S1 Primers used

Chapter 2 PCR primer

Gene GenBank Primer

Accession

Ca HydA 5' CAC3230 CCTTTCTAGA ATGAAAACAATAATCTTAAA

Ca HydA 3' AAGGCTGCAGCGGCCGCTACTAGT TTCATGTTTTGAAACATTTTTA

Cs HydA 5' CAU09760 CCTTGAATTCGCGGCCGCATCTAGA

ATGATAAACATAGTAATTGATGAAA

Cs HydA 3' AAGGCTGCAGCGGCCGCTACTAGT

TTTATTGTATTTTAAGTGTAATAAT

So HydA 5' SO_3920 CCTTGAATTCGCGGCCGCATCTAGA

ATGACAACGACAACTTATCAACCAG

So HydA 3' AAGGCTGCAGCGGCCGCTACTAGT AATGTTACCCAGCCATGAAGA

GCCT

So HydB 5' SO_3921 CCTTGAATTCGCGGCCGCATCTAGA

ATGAACAAGAAAAAACACCTATTTG

So HydB 3' AAGGCTGCAGCGGCCGCTACTAGT

AGAGCTTAATTTGGTGCGATCGACA

Tm HydA 5' AF044577 CCTTGAATTCGCGGCCGCATCTAGA

ATGAAAATTTACGTTGATGGAAGAGAAGTTAT

Tm HydA 3' AAGGCTGCAGCGGCCGCTACTAGT

TCAGCCATTTTTCGAAAGCTCCTCCAGCACCTTCTC

Ca PFOR 5' CAC2229 CCTTGAATTCGCGGCCGCATCTAGA ATGAAAAAAATGAAAACTATGG

Ca PFOR 3' AAGGCTGCAGCGGCCGCTACTAGT TTATTGATTAGCTAATCTTT

Da PFOR 5' CAA70873 AAACATATGGGAAAGAAAATGATGACG

Da PFOR 3' AAACCTAGGCTACTTCTTCGTCCGCTTGC

Ec ydbK 5' CAQ31878 CCTTGAATTCGCGGCCGCATCTAGA ATTACTATTGACGGTAATGGCG

Ec ydbK 3' AAGGCTGCAGCGGCCGCTACTAGT

TTAATCGGTGTTGCTTTTTTCCGC

Ca Fd 5' CAC0303 CCTTGAATTCGCGGCCGCATCTAGA ATGGCATATAAAATAACAGA

Ca Fd 3' AAGGCTGCAGCGGCCGCTACTAGT CTCTTGAACTGGAGCTCCTA

Zm Fd 5' EU956042 CCTTGAATTCGCGGCCGCATCTAGA

ATGGCTGTATACAAGGTGAAGCTTG

Zm Fd 3' AAGGCTGCAGCGGCCGCTACTAGT
CAGGTCGCCTTCCTTGTGGGTGTGG

171
 Table S1 continued from previous page

" Chapter 2 mutagenesis primers

Ca mut. 5' GTGGACAATGTTCCAGAAGAGAAAATTGTGAGTTCCTTAAACTTG

Cs mut. 5' GCAGCAAACTTAGCTGAGTTCATGAATAGCGG

SoB mut. CGCGGTAGAAAAGTACTACAAAGAGTTCGGTGGCGAGCCATTAGGA

CATATGTCCC

CrHydA P2K CCGCTTCTAGAGCTGCAAAGGCCGCAGAAGCTCCTTTGTC

CrHydA D126K CGCTATGTTAGAAAAAAGCTATCCTAAGCTAATTCCATACGTGAGC

CrHydA E5K CCGCTTCTAGAGCTGCACCAGCCGCAAAGGCTCCTTTGTC

CrHydA M119K CCAGGTTGGATCGCTAAGTTAGAAAAAAGCTATCC

!
" Chapter 3 PCR primers

invasin 5' ZP_02317525 CGCAACTAGTATGGTTTTCCAGCCAATCAG

invasin 3' CTGCAGCGGCCGCTAGCTCTAGATTATATTGACAGCGCACAGA

listeriolysin 5' CAA42639 CGCAACTAGTAGGAGGAAAAACATATGAAAAAAATAATGCTAGTTTT

listeriolysin 3' CTGCAGCGGCCGCTTCTAGATTATTCGATTGGATTATCTA

" Chapter 4 vector primers and oligonucleotides

O1 and O2 insert 5' GGGAATTCGCGGCCGCTTCTAGAACTAGTAGCGGCCGCTGCAGG

O1 and O2 insert 3' CTAGCCTGCAGCGGCCGCTACTAGTTCTAGAAGCGGCCGCGAATTCCCGC

R1 insert PCR actacgaagcttgaattcgcggccgcttctagaattagaggactagtagcggccgctgcagagagctcatgttac

primer 5' gtcctgtagaaacc

R1 insert PCR caagaGCTAGCAaaaggtacctcattgtttgc

primer 3'

R3 insert PCR actacgaagcttgaattcgcggccgcttctagaattagaggactagtagcggccgctgcagagagctcatggcg

primer 5' agtaaaggagaagaac

R3 insert PCR atctGCTAGCttttggtaccttatttgtatag

primer 3'

E3 and E4 insert cgtgatAAGCTTgggatcttctgcaagcatctc

PCR 5'

E3 and E4 insert caagaGCTAGCctgcagcggccgctactagtcctctaattctagaagcggccgcgaattctccggtgggtttt

PCR 3' gaggtgag

172
Table S1 Continued from previous page
Chapter 4 artificial microRNA primers

RS300 BioBrick 5' CCTTGAATTCGCGGCCGCATCTAGA CCCACAAACACACGCTCGGA

RS300 BioBrick 3' AAGGCTGCAGCGGCCGCTACTAGT CCCCATGGCGATGCCTTAAA

LTP1 miRNA sense gaTCTAACTATGTATAGGACCACtctctcttttgtattcc

LTP1 miRNA antisense gaGTGGTCCTATACATAGTTAGAtcaaagagaatcaatga

LTP1 miRNA * sense gaGTAGTCCTATACAAAGTTAGTtcacaggtcgtgatatg

LTP1 miRNA *antisense gaACTAACTTTGTATAGGACTACtctacatatatattcct

Betv1 miRNA sense gaTTGGTTATATATATCGCACCTtctctcttttgtattcc

Betv1 miRNA antisense gaAGGTGCGATATATATAACCAAtcaaagagaatcaatga

Betv1 miRNA * sense gaAGATGCGATATATTTAACCATtcacaggtcgtgatatg

Betv1 miRNA *antisense gaATGGTTAAATATATCGCATCTtctacatatatattcct

Lut2 miRNA sense gaTAACTTCCTAGTAACCCGCATtctctcttttgtattcc

Lut2 miRNA antisense gaATGCGGGTTACTAGGAAGTTAtcaaagagaatcaatga

Lut2 miRNA * sense gaATACGGGTTACTACGAAGTTTtcacaggtcgtgatatg

Lut2 miRNA *antisense gaAAACTTCGTAGTAACCCGTATtctacatatatattcct

Ohase1 miRNA sense gaTCGTTTAAATCGAAGAGACGCtctctcttttgtattcc

Ohase1 miRNA antisense gaGCGTCTCTTCGATTTAAACGAtcaaagagaatcaatga

Ohase1 miRNA * sense gaGCATCTCTTCGATATAAACGTtcacaggtcgtgatatg

Ohase1miRNA*antisense gaACGTTTATATCGAAGAGATGCtctacatatatattcct

Lyc miRNA sense gaTAATTATCCAACGACTTGCAGtctctcttttgtattcc

Lyc miRNA antisense gaCTGCAAGTCGTTGGATAATTAtcaaagagaatcaatga

Lyc miRNA * sense gaCTACAAGTCGTTGCATAATTTtcacaggtcgtgatatg

Lyc miRNA *antisense gaAAATTATGCAACGACTTGTAGtctacatatatattcct

173
Table S1 Continued from previous page
Chapter 4 hpRNA primers

pdk intron 5' CCTTGAATTCGCGGCCGCATCTAGA CCAATTGGTAAGGAAATAAT

pdk intron 3' AAGGCTGCAGCGGCCGCTACTAGT TTCGAACCCAATTTCCCAAC

LTP1 sense 5' CCTTGAATTCGCGGCCGCATCTAGA ATGGCTGGAGTGATGAAGTT

LTP1 sense 3' AAGGCTGCAGCGGCCGCTACTAGT CAGCTGCACGGCCAGCGTTG

LTP1 antisense 5' AAGGCTGCAGCGGCCGCTACTAGT ATGGCTGGAGTGATGAAGTT

LTP1 antisense 3' CCTTGAATTCGCGGCCGCATCTAGA CAGCTGCACGGCCAGCGTTG

Bet v 1 sense 5' CCTTGAATTCGCGGCCGCATCTAGA ATGATAGAAGAGGAGATTGA

Bet v 1 sense 3' AAGGCTGCAGCGGCCGCTACTAGT TACCATATTAAAAAATACGT

Bet v 1 antisense 5' AAGGCTGCAGCGGCCGCTACTAGT ATGATAGAAGAGGAGATTGA

Bet v 1 antisense 3' CCTTGAATTCGCGGCCGCATCTAGA TACCATATTAAAAAATACGT

Ger3 sense 5' CCTTGAATTCGCGGCCGCATCTAGA ATGAAGATGATAATCCAAAT

Ger3 sense 3' AAGGCTGCAGCGGCCGCTACTAGT GATGACACCACCACCGGCTA

Ger3 antisense 5' AAGGCTGCAGCGGCCGCTACTAGT ATGAAGATGATAATCCAAAT

Ger3 antisense 3' CCTTGAATTCGCGGCCGCATCTAGA GATGACACCACCACCGGCTA

Chapter 4 qRT-PCR primers

LTP 5' CGGCCCTGGTCTCAACGCTG

LTP 3' ACGTGTTGCACTTGGTGTTGAACCT

GER3 5' GGCCTCGGCACAGCTGGAAA

GER3 3' GCCAAGGCCATTGATGCCTGC

LUT2 5' ACGTCATTCCCTGCAGGCTTGC

LUT2 3' CCTCAACCTCCACGCCGTATGC

β-OH 5' TGCTTTGGCGCCGGGTTAGG

β-OH 3' AGGGACGTCGGCGATGGGAC

BRAZ 5' AATTGGCAAACCAGTGCAA

BRAZ 3' TCACAGATACACTGGAGGTTCC

MIR1 5' CGTGACTATCGGAGGAGTGAAGGGT

MIR1 3' CCACACACAGTTGGGCAAAACACA

174
Table S1 Continued from previous page

MIR2 5' CGGAGGAGTGAAGGGTAATCCTGGT

MIR2 3' ACAAGATCCACACACAGTTGGGCA

MIR3 5' TGTGTTTTGCCCAACTGTGTGTGGA

MIR3 3' GACAAAGCAAGTCTCCTCCTTCCC

Actin 5' TGTGCCAATCTACGAGGGTTT

Actin 3' TTTCCCGCTCTGCTGTTGT

175
Figure S2
Sequence alignment of five hydrogenases Protein sequences of Clostridium acetobutylicum
(Ca), Clostridium saccharobutylicum (Cs), Chlamydomonas reinhardtii (Cr), and Thermotoga maritima
(Tm) HydA and Shewanella oneidensis HydB + HydA aligned using ClustalW. Catalytic site
binding area highlighted in bold with critical cysteine residues in red.

Ca ---MKTIIINGVQFNTDEDTTILKFARDNNIDISALCFLNNCNNDINKCEICTVEVEG-T 56
Cs ---MINIVIDEKTIQVQENTTVIQAALANGIDIPSLCYLNECGN-VGKCGVCAVEIEGKN 56
Cr ------------------------------------------------------------
So MNKKKHLFAEDSFFLSRRKFMAVGAAFVAALAIPIGWFT--------------------S 40
Tm ---MKIYVDGREVIINDNERNLLEALKNVGIEIPNLCYLSEASIYG---ACRMCLVEING 54

Ca GLVTACDTLIEDGMIINTNSDAVNEKIKSRISQLLDIHEFKCGPCNRRENCEFLKLVIKY 116
Cs NLALACITKVEEGMVVKTNSEKVQERVKMRVATLLDKHEFKCGPCPRRENCEFLKLVIKT 116
Cr ------------------------------------------------------------
So KLERRNEYIKARSQGLYKDDSLAKTRVSHANPAVEKYYKEFGGEPLGHMSHELLHTHFVD 100
Tm QITTSCTLKPYEGMKVKTNTPEIYEMRRNILELILATHNRDCTTCDRNGSCKLQKYAEDF 114

Ca KARASKPFLPKDKTEYVDERSKSLTVDRTKCLLCGRCVNACGKNTETYAMKFLNKNGKTI 176
Cs KAKANKPFVVEDKSQYIDIRSKSIVIDRTKCVLCGRCEAACKTKTGTGAISICKSESGRI 176
Cr ------------------------------------------------------------
So RTKLSSMTTTTYQPGEIQG---LIKINASKCKGCDACKQFCPTHAINGASGAVHS----- 152
Tm GIRKIR--FEALKKEHVRDESAPVVRDTSKCILCGDCVRVCEEIQGVGVIEFAKRGFESV 172

Ca IGAEDEKCFDDTNCLLCGQCIIACPVAALSE-KSHMDRVKNALNAPEKHVIVAMAPSVRA 235
Cs VQATGGKCFDDTNCLLCGQCVAACPVGALTE-KTHVDRVKEALEDPNKHVIVAMAPSIRT 235
Cr ------------------APAAEAPLSHVQQALAELAKPKDDPTRKHVCVQ--VAPAVRV 40
So --------IDEDKCLSCGQCLINCPFSAIEETHSALETVIKKLADKNTTVVGIIAPAVRV 204
Tm VTTAFDTPLIETECVLCGQCVAYCPTGALSI-RNDIDKLIEALES-DKIVIGMIAPAVRA 230
.* . : : . . * :**::*.

Ca SIGELFNMGFGVDVTGKIYTALRQLGFDKIFDINFGADMTIMEEATELVQRIEN------ 289
Cs SMGELFKLGYGVDVTGKLYASMRALGFDKVFDINFGADMTIMEEATEFIERVKN------ 289
Cr AIAETLGLAPGATTPKQLAEGLRRLGFDEVFDTLFGADLTIMEEGSELLHRLTEHLEAHP 100
So AIGEEFGLGTGELVTGKLYGAMNQAGF-KIFDCNFAADLTIMEEGSEFIHRLHANVKGEA 263
Tm AIQEEFGIDEDVAMAEKLVSFLKTIGFDKVFDVSFGADLVAYEEAHEFYERLKK------ 284
:: * : : . . :: :. ** ::** *.**:. **. *: .*:

Ca --NGPFPMFTSCCPGWVRQAENYYPELLNNLSSAKSPQQIFGTASKTYYPSISGLDPKNV 347
Cs --NGPFPMFTSCCPAWVRQVENYYPEFLENLSSAKSPQQIFGAASKTYYPQISGISAKDV 347
Cr HSDEPLPMFTSCCPGWIAMLEKSYPDLIPYVSSCKSPQMMLAAMVKSYLAEKKGIAPKDM 160
So NAG-PLPQFTSCCPGWVRYLETRYPALLPNLSTAKSPQQMAGTVAKTYGAKVYQMQPENI 322
Tm --GERLPQFTSCCPAWVKHAEHTYPQYLQNLSSVKSPQQALGTVIKKIYARKLGVPEEKI 342
. :* ******.*: * ** : :*: **** .: *. . : :.:

Ca FTVTVMPCTSKKFEADRPQME------------KDGLRDIDAVITTRELAKMIKDAKIPF 395
Cs FTVTIMPCTAKKFEADREEMY------------NEGIKNIDAVLTTRELAKMIKDAKINF 395
Cr VMVSIMPCTRKQSEADRDWF---------CVDADPTLRQLDHVITTVELGNIFKERGINL 211
So FTVSVMPCTSKKLEASRPEFNSAWQYHQEHGANSPSYQDIDAVLTTREMAQLLKLLDIDL 382
Tm FLVSFMPCTAKKFEAEREEHEG----------------IVDIVLTTRELAQLIKMSRIDI 386
. *:.**** *: **.* :* *:** *:.:::* * :

Ca AKLEDSEADPAMGEYSGAGAIFGATGGVMEAALRSAKDFAENAELEDIEYKQVRGLNGIK 455
Cs ANLEDEQADPAMGEYTGAGVIFGATGGVMEAALRTAKDFVEDKDLTDIEYTQIRGLQGIK 455
Cr AELPEGEWDNPMGVGSGAGVLFGTTGGVMEAALRTAYELFTGTPLPRLSLSEVRGMDGIK 271

So ANTAEYQGDSLFSEYTGAGTIFGTTGGVMEAALRTAHKVLTGTEMAKLEFEPVRGLKGVK 442
Tm NRVEPQPFDRPYGVSSQAGLGFGKAGGVFSCVLSVLNEEIG---IEKVDVKSPE--DGIR 441
. * . : ** ** :***:...* . : :. . .*::

176
Figure S2 continued from previous page
Ca EAEVEINNNKYN---------------------------------------------VAV 470
Cs EATVEIGGENYN---------------------------------------------VAV 470
Cr ETNITMVPAPGSKFEELLKHRAAARAEAAAHGTPGPLAWDGGAGFTSEDGRGGITLRVAV 331
So SASVSLFDTELN---------------------------------------QDVTVNVAV 463
Tm VAEVTLKDGTSFKG--------------------------------------------AV 457
: : : **

Ca INGAS-NLFKFMKSGMINEKQYHFIEVMACHGGCVNGGGQPHVNPKDLEKVDIKKVRASV 529
Cs INGAA-NLAEFMNSGKILEKNYHFIEVMACPGGCVNGGGQPHVSAKEREKVDVRTVRASV 529
Cr ANGLG-NAKKLITKMQAGEAKYDFVEIMACPAGCVGGGGQPRSTDKA-----ITQKRQAA 385
So VHDMGNNIEPVLRDVMAGTSPYHFIEVMNCAGGCVNGGGQP-----------IEGKGSSW 512
Tm IYGLG-----KVKKFLEERKDVEIIEVMACNYGCVGGGGQPYPNDSR-----IREHRAKV 507
. . : . .::*:* * ***.***** :

Ca LYNQDEHLSKRKSHENTALVKMYQNYFGKPGEGRAHEILHFKYKK--------------- 574
Cs LYNQDKNLEKRKSHKNTALLNMYYDYMGAPGQGKAHELLHLKYNK--------------- 574
Cr LYNLDEKSTLRRSHENPSIRELYDTYLGEPLGHKAHELLHTHYVAGGVEEKDEKKTSSGR 445
So LGNI-------------------------------------------------------- 516
Tm LRDTMGIKSLLTPVENLFLMKLYEEDLKD--EHTRHEILHTTYRPRRRYPEKDVEILPVP 565
* :

Ca ------------------------------------------------------------
Cs ------------------------------------------------------------
Cr C----------------------------------------------------------- 446
So ------------------------------------------------------------
Tm NGEKRTVKVCLGTSCYTKGSYEILKKLVDYVKENDMEGKIEVLGTFCVENCGASPNVIVD 625

Ca --------------------
Cs --------------------
Cr --------------------
So --------------------
Tm DKIIGGATFEKVLEELSKNG 645

177
Figure S3

Alignment of Chlamydomonas reinhardtii HydA1 and HydA2. ClustalW alignment of

HydA1 and HydA2 with mutations predicted to improve the binding between HydA2
and ferredoxin (Long et. al., 2009) highlighted in red.

HydA1 APAAEAPLSHVQQALAELAKPKDDPTRKHVCVQVAPAVRVAIAETLGLAPGATTPKQLAE 60
HydA2 -ATATDAVPHWKLALEELDKPKDG-GRKVLIAQVAPAVRVAIAESFGLAPGAVSPGKLAT 58
.:* .:.* : ** ** ****. ** : .************::******.:* :**

HydA1 GLRRLGFDEVFDTLFGADLTIMEEGSELLHRLTEHLEAHPHSDEPLPMFTSCCPGWIAML 120

HydA2 GLRALGFDQVFDTLFAADLTIMEEGTELLHRLKEHLEAHPHSDEPLPMFTSCCPGWVAMM 118
*** ****:******.*********:******.***********************:**:

HydA1 EKSYPDLIPYVSSCKSPQMMLAAMVKSYLAEKKGIAPKDMVMVSIMPCTRKQSEADRDWF 180

HydA2 EKSYPELIPFVSSCKSPQMMMGAMVKTYLSEKQGIPAKDIVMVSVMPCVRKQGEADREWF 178
*****:***:**********:.****:**:**:**..**:****:***.***.****:**

HydA1 CVDADPTLRQLDHVITTVELGNIFKERGINLAELPEGEWDNPMGVGSGAGVLFGTTGGVM 240

HydA2 CVS-EPGVRDVDHVITTAELGNIFKERGINLPELPDSDWDQPLGLGSGAGVLFGTTGGVM 237
**. :* :*::******.*************.***:.:**:*:*:***************

HydA1 EAALRTAYELFTGTPLPRLSLSEVRGMDGIKETNITMVPAPGSKFEELLKHR-------- 292

HydA2 EAALRTAYEIVTKEPLPRLNLSEVRGLDGIKEASVTLVPAPGSKFAELVAERLAHKVEEA 297
*********:.* *****.******:*****:.:*:******** **: .*

HydA1 AAARAEAAAHGTPG-PLAWDGGAGFTSEDGRGGITLRVAVANGLGNAKKLITKMQAGEAK 351

HydA2 AAAEAAAAVEGAVKPPIAYDGGQGFSTDDGKGGLKLRVAVANGLGNAKKLIGKMVSGEAK 357
***.* **..*: *:*:*** **:::**:**:.**************** ** :****

HydA1 YDFVEIMACPAGCVGGGGQPRSTDKAITQKRQAALYNLDEKSTLRRSHENPSIRELYDTY 411

HydA2 YDFVEIMACPAGCVGGGGQPRSTDKQITQKRQAALYDLDERNTLRRSHENEAVNQLYKEF 417
************************* **********:***:.******** ::.:**. :

HydA1 LGEPLGHKAHELLHTHYVAGGVEEKDEKKTSSGRC 446

HydA2 LGEPLSHRAHELLHTHYVPGGAEADA--------- 443
*****.*:**********.**.* .

178
 hand-1
Figure S4

squalene

isoamyl alcohol

acetic
acid 9-hexadecenoic
isopropyl
acid
indole palmitate
limonene dioctyl
siloxane myristic diethylhexyl
ether
diethyl acid palmitic phthalate
isovaleric acid cholest-5-
phthalate acid hexadecyl
en-3-ol
nonanal dodecane cinnamate octadecanoate
Gas chromatograph traces of volatile compounds collected from human cheeses.

180
 2,6,10,15,19,23-
nose-2

hexamethyltetracosane

isopropyl
palmitate

dioctyl
ether squalene
Figure S4, continued from previous page

ethyl 3-methyl
acetic valerate
acid
isoamyl hexadecanol
alcohol siloxane cinnamate
ethyl limonene hentriacontane
butyrate 1-eicosene
ethyl diethyl diethylhexyl
phthalate hexadecyl
acetate butyric phthalate
myristic 9-octadecanoate
acid nonanal
tridecane acid

181
 squalene
armpit-3

2,6,10,15,19,23-
hexamethyltetracosane
acetyl
Figure S4, continued from previous page

isopropyl
methyl palmitate
carbinol
isopropyl palmitic
laurate acid
isoamyl
alcohol c12
hydrocarbon hexadecanol cinnamate
diethyl
toluene limonene
2-pentanone siloxane phthalate cholest-5-en-3ol
2-heptanone dioctyl
tetradecane ether diethylhexyl
ethyl 2-ethyl hexadecyl
acetate tridecane myristic phthalate
hexanol nonanal 9-octadecanoate
acid

182
 foot-5

squalene

dioctyl
ether
Figure S4, continued from previous page

2,6,10,15,19,23-
isopropyl hexamethyltetracosane
palmitic palmitate
siloxane tritriacontane
limonene acid diethylhexyl
phthalate neobee
1,3-butylene diethyl component
9-hexadecenoic hexadecyl
glycol phthalate oleic acid
decanal acid 9-octa-
ethyl toluene 2-ethyl cinnamate decanoate
acetate myristic
hexanol nonanal tridecane
acid

183
 hand-1 nose-2 armpit-3 foot-5
Area % Identification Area % Identification Area % Identification Area % Identification
alcohols 3.87% ethanol 0.4547 ethanol 0.6723 ethanol 0.2627 ethanol
0.33% isopropyl alcohol 0.191 isopropyl alcohol 0.3065 isopropyl alcohol 0.1365 isopropyl alcohol
0.1363 isobutanol
0.0567 acetol 0.0379 acetol
Table S2

0.70% propanol 0.0367 propanol

0.31% butanol 0.0208 butanol 0.0755 butanol 0.0785 butanol
0.59% 1-methoxy-2-propanol 0.0693 1-methoxy-2-propanol 0.0606 1-methoxy-2-propanol 0.338 1-methoxy-2-propanol
0.60% acetyl methyl carbinol 0.0699 acetyl methyl carbinol 6.4009 acetyl methyl carbinol 0.227 acetyl methyl carbinol
11.41% isoamyl alcohol 2.3131 isoamyl alcohol 2.0496 isoamyl alcohol
0.15% 2-methylbutanol 0.0199 2-methylbutanol 0.0696 2-methylbutanol 0.7039 2-methylpentanol
0.0867 1,3-butanediol
0.5613 1,3-butylene glycol
0.0256 methyl carbotil
0.33% amyl alcohol 0.029 amyl alcohol
0.92% 2,3-butylene glycol 0.0723 2,3-butylene glycol
2.15% 1,3-butylene glycol 0.3007 1,3-butylene glycol
0.24% 2-hexenol 0.0299 2-hexenol
0.15% methionol (3-(methylthio)propanol
0.076 hexanol 0.0742 hexanol
0.1236 glycerin 0.5117 glycerin
0.27% phenol 0.0299 phenol 0.0472 phenol 0.1999 phenol
0.23% 2-ethylhexanol 0.2812 2-ethylhexanol 0.6071 2-ethylhexanol 0.5846 2-ethylhexanol
0.09% eucalyptol 0.0266 eucalyptol 0.1549 eucalyptol 0.057 eucalyptol
0.03% octanol 0.0362 octanol 0.0809 octanol 0.0484 octanol
0.14% phenylethyl alcohol 0.051 phenylethyl alcohol 0.1327 phenylethyl alcohol 0.07 phenylethyl alcohol
0.07% linalool 0.0933 linalool 0.1661 linalool 0.1966 linalool
0.03% isomenthol 0.0901 isomenthol
0.0642 butyl carbotol 0.1854 butyl carbitol
0.02% eugenol 0.0147 eugenol 0.0317 eugenol
0.0151 geosmin
0.02% dodecanol 0.0137 methyl paraben 0.0302 methyl paraben
0.0584 dodecanol 0.0848 dodecanol
0.0383 3,5-di-tertiarybutylphenol 0.0835 3,5-di-tertiarybutyl phenol
0.09% hexadecanol 0.4436 hexadecanol 3.6188 hexadecanol 0.6665 hexadecanol
0.1413 octadecanol 0.4724 octadecanol 0.2118 octadecanol
0.65% cholest-5-en-3-ol 0.704 cholest-5-en-3-ol 1.1291 cholest-5-en-3-ol 1.5652 cholest-e-en-3-ol

aldehydes 0.29% acetaldehyde 0.1366 acetaldehyde 0.8142 acetaldehyde 0.1719 acetaldehyde

5.20% isobutyraldehyde 0.1053 isobutyraldehyde
0.0459 isovaleraldehyde 0.0232 isovaleraldehyde
0.89% hexanal 0.0727 hexanal 0.1306 hexanal 0.1861 hexanal
0.0324 furfural 0.0653 furfural
0.0208 heptanal 0.0405 heptanal 0.0509 heptanal
0.09% benzaldehyde 0.1228 benzaldehyde 0.1763 benzaldehyde 0.2309 benzaldehyde
0.08% octanal 0.4659 octanal 0.1481 octanal 0.1788 octanal
analysis of standard cheese volatiles are highlighted in yellow.

0.37% nonanal 0.343 nonanal 0.8469 nonanal 0.8871 nonanal

0.9212 decanal 0.7408 decanal
0.06% anisaldehyde 0.0372 anisaldehyde 0.0835 anisaldehyde 0.0827 anisaldehyde
0.02% 2-decenal 0.0167 2-decenal 0.0368 2-decenal 0.0285 2-decenal
0.0298 4-hydroxybenzaldehyde
0.0163 dodecanal 0.096 dodecanal

esters 0.451 ethyl acetate 0.5777 ethyl acetate 0.5734 ethyl acetate
0.06 isobutyl acetate 0.0459 isobutyl acetate
0.14% ethyl butyrate 0.9016 ethyl butyrate 0.0804 ethyl butyrate 0.0642 ethyl butyrate
0.90% butyl acetate 0.3932 butyl acetate 0.1249 butyl acetate 0.2024 butyl acetate
0.0239 ethyl 2-methylbutyrate 0.0282 ethyl 2-methylbutyrate
0.80% isoamyl acetate 0.0641 isoamyl acetate 0.0246 isoamyl acetate 0.0256 isoamyl acetate
listed for each cheese, separated by compound type. Compounds found in GC/MS
by International Flavors and Fragrances. Relative area of peaks and identification is
Comparative smell-omics of four human cheeses. Headspace analysis was performed

184
 0.0191 butyl cellosolve 0.0459 butyl cellosolve 0.1035 butyl cellosolve
0.0126 ethyl hexanoate
0.05% ethyl-3-hydroxybutyrate
0.6144 ethyl 3-methyl valerate
0.02% c-3-hexenyl acetate 0.0399 cis-3-hexenyl acetate 0.0332 cis-3-hexenyl acetate
0.07% hexyl acetate 0.0926 hexyl acetate 0.0864 hexyl acetate
0.02% isoamyl butyrate 0.0523 cis-rose oxide 0.0546 cis-rose oxide 0.0524 cis rose oxide
0.05% rose oxide isomer 0.0136 trans rose oxide
0.02% benzyl acetate 0.0295 benzyl acetate 0.064 benzyl acetate 0.0515 benzyl acetate
0.0314 2-ethylhexyl acetate 0.0675 2-ethylhexyl acetate 0.0687 2-ethylhexyl acetate
0.00% ethyl octanoate 0.0165 ethyl octanoate
0.028 methyl salicylate 0.0679 methyl salicylate
0.0177 ethyl octanoate 0.0051 ethyl octanoate
0.0042 alpha-terpinyl methyl ether 0.0039 alpha-terpinyl methyl ether
0.0121 linalyl acetate 0.0303 linalyl acetate 0.0388 linalyl acetate
0.10% anethole 0.0299 anethole 0.0499 anethole 0.0506 anethole
0.04% 4-methylpentyl butyrate
0.064 isobornyl acetate 0.1076 isobornyl acetate 0.1023 isobornyl acetate
0.02% glyceryl triacetate 0.0202 glyceryl triacetate 0.0392 glyceryl triacetate 0.0398 glyceryl triacetate
0.0144 alpha-terpinyl acetate
0.0391 ethyl decanoate 0.0574 ethyl decanoate 0.0056 ethyl decanoate
0.01% dimethyl phthalate 0.0146 dimethyl phthalate 0.0252 dimethyl phthalate 0.0327 dimethyl phthalate
1.01% diethyl phthalate 1.0061 diethyl phthalate 2.2907 diethyl phthalate 1.5956 diethyl phthalate
0.21% isopropyl laurate 0.1769 isopropyl laurate 0.4302 isopropyl laurate 0.2714 isopropyl laurate
0.02% triethyl citrate 0.012 triethyl citrate 0.0245 triethyl citrate 0.0247 triethyl citrate
0.99% dioctyl ether (t) 8.8121 dioctyl ether 1.2741 dioctyl ether 13.209 dioctyl ether
0.03% 2-methylheptyl benzoate 0.0338 2-methylheptyl benzoate 0.0709 2-methylheptyl benzoate 0.0333 2-methylheptyl benzoate
0.0899 tria-(1-chloro-2-propyl)phosphate
0.0452 ethyl tetradecanoate 0.0228 ethyl tetradecanoate
0.0368 2-ethylhexyl salicylate
0.11% isopropyl myristate 0.163 isopropyl myristate 0.2106 isopropyl myristate 0.6087 isopropyl myristate
0.12% diisobutyl phthalate 0.1236 diisobutyl phthalate 0.244 diisobutyl phthalate 0.2072 diisobutyl phthalate
0.12% ethyl hexadecanoate
Table S2 continued from previous page

0.0198 ethyl pentadecanoate 0.036 ethyl pentadecanoate

0.0955 dibutyl phthalate 0.1281 dibutyl phthalate 0.1805 dibutyl phthalate
0.1816 ethyl hexadecanoate 0.1008 ethyl hexadecanoate
0.46% isopropyl palmitate 10.843 isopropyl palmitate 4.8043 isopropyl palmitate 1.8836 isopropyl palmitate
0.0248 ethyl heptadecanoate 0.0431 ethyl heptadecanoate 0.0322 ethyl heptadecanoate
0.20% 2-ethylhexyl 4-methoxycinnamate isomer 0.388 2-ethylhexyl-4-methoxy cinnamate isomer 0.5861 2-ethylhexyl 4-methoxycinnamate isomer
0.1449 ethyl oleate
0.0538 ethyl 9-octadecenoate
0.11% ethyl octadecanoate 0.1209 ethyl octadecanoate (stearate) 0.2065 ethyl octadecanoate 0.1231 ethyl octadecanoate
0.39% 2-ethylhexyl 4-methoxy cinnamate isomer 0.7097 2-ethylhexyl-4-methoxy cinnamate isomer 1.863 2-ethylhexyl 4-methoxycinnamate isomer 0.7142 2-ethylhexyl 4-methoxycinnamate isomer
0.2465 2-ethylhexyl adipate
0.0591 dihydroabietate isomer (floralyn)
1.18% di-ethylhexyl phthalate 0.5267 diethylhexyl phthalate
0.48% 13-docosenate
0.0715 tetradecyl benzoate 0.0814 tetradecyl benzoate
0.7761 diethylhexyl phthalate 2.2449 diethylhexyl phthalate
0.0547 hexadecyl nonanoate
0.0507 octadecyl nonanoate 0.143 octadecyl nonanoate
0.1946 tetradecyl tetradecanoate
0.3342 tetradecyl 9-tetradecenoate
0.2726 tetradecyl 9-hexadecenoate
0.2794 hexadecyl tetradecanoate 0.3791 hexadecyl tetradecanoate
0.9506 alpha-tocopheryl acetate 0.5936 alpha-tocopheryl acetate
0.6483 tetradecyl 9-octadecenoate 0.3793 tetradecyl 9-octadecenoate
0.2922 alpha-tocopheryl acetate
0.6844 hexadecyl palmitate
0.7298 hexadecyl 9-hexadecenoate

185
1.1987 hexadecyl hexadecanoate 0.4304 hexadecyl hexadecanoate
0.68% hexadecyl 9-octadecenoate 1.3616 hexadecyl 9-octadecenoate 2.0392 hexadecyl 9-octadecenoate 1.335 hexadecyl 9-octadecenoate
 0.21% hexadecyl octadecanoate 0.5582 hexadecyl octadecanoate 1.3413 hexadecyl octadecanoate 0.4875 hexadecyl octadecanoate

acids 11.85% acetic acid 17.395 acetic acid 1.1553 acetic acid 0.515 acetic acid
1.46% propionic acid 0.0571 propionic acid
0.53% isobutyric acid 0.0816 isobutyric acid
0.77% butyric acid 3.5795 butyric acid 0.0441 butyric acid 0.0417 butyric acid
0.0678 valeric acid
7.77% isovaleric acid 0.0212 isovaleric acid
0.24% 2-methylbutyraic acid
0.39% hexanoic acid 0.6767 hexanoic acid 0.2048 hexanoic acid 0.0803 hexanoic acid
0.06% heptanoic acid 0.0949 heptanoic acid 0.2217 heptanoic acid 0.1375 heptanoic acid
0.02% 2-ethylhexanoic acid 0.0321 2-ethylhexanoic acid 0.1043 2-ethylhexanoic acid 0.0575 2-ethylhexanoic acid
0.09% benzoic acid 0.0467 benzoic acid 0.1641 benzoic acid
0.33% octanoic acid 0.5947 octanoic acid 0.3262 octanoic acid 0.2262 octanoic acid
0.06% nonanoic acid 0.0691 nonanoic acid 0.1018 nonanoic acid 0.0721 nonanoic acid
0.09% decanoic acid 0.1202 decanoic acid 0.0964 decanoic acid 0.0904 decanoic acid
0.18% dodecanoic acid 0.0646 dodecanoic acid 0.2248 dodecanoic acid 0.1226 dodecanoic acid
0.0571 11-tetradecenoic acid
0.71% tetradecanoic acid (myristic acid) 0.4763 tetradecanoic acid (myristic) 1.2338 tatredecanoic acid (myristic) 1.4953 tetradecanoic acid (myristic)
0.33% pentadecanoic acid 0.1575 pentadecanoic acid 0.333 pentadecanoic acid 0.8473 pentadecanoic acid
0.40% 9-hexadecenoic acid 0.2523 9-hexadecenoic acid 0.718 9-hexadecenoic acid 1.3923 9-hexadecenoic acid
1.53% hexadecanoic acid (palmitic) 1.4536 hexadecanoic acid (palmitic) 5.4809 hexadecanoic acid (palmitic) 4.926 hexadecanoic acid (palmitic)
0.0978 heptadecanoic acid
0.0737 linoleic acid 0.1531 linoleic acid
0.24% oleic acid 0.4198 oleic acid 1.7842 oleic acid 1.827 oleic acid
0.5773 octadecanoic acid (stearic) 0.3941 octadecanoic acid (stearic)

ketones 0.09% 1-hydroxy-2-propanone

0.60% methyl ethyl ketone 0.0578 methyl ethyl ketone
0.0384 2-heptanone
0.061 6-methyl-5-hepten-2-one 0.0914 6-methyl-5-hepten-2-one
0.0421 n-methyl-2-pyrrolidinone
Table S2 continued from previous page

0.5184 acetone 0.4461 acetone

0.5072 diacetyl
0.109 2-butanone 0.0325 2-butanone
0.1494 3-methylpentanone
1.4013 2-pentanone
0.0384 2-hexanone 0.0148 2-hexanone
0.189 3-hydroxy-2-pentanone
0.0881 2-hydroxy-2-pentanone
0.0397 2(5h)-furanone
0.0578 cyclohexanone 0.0611 cyclohexanone
1.3928 2-heptanone 0.0152 2-heptanone
0.0925 2-octanone
0.094 delta-valerolactone
0.06% acetophenone 0.1344 acetophenone 0.1682 acetophenone 0.1558 acetophenone
0.03% camphor 0.0287 camphor 0.0367 camphor
0.1845 8-nonen-2-one
0.5865 2-nonanone
0.06% menthone 0.057 menthone 0.0999 menthone 0.0982 menthone
0.04% p-mentha-6,8-dien-2-one
0.0098 carvone 0.0153 carvone 0.0173 carvone
0.01% gamma-nonalactone 0.0173 gamma-nonalactone 0.0246 gamma-nonalactone 0.0308 gamma-nonalactone
0.0093 alpha-ionone
0.03% geranyl acetone 0.0306 geranyl acetone 0.0762 geranyl acetone 0.0621 geranyl acetone
0.0306 2,5-di-tertiarybutylbenzophenone 0.0636 2,4-ditertiarybutyl benzophenone 0.0314 2,6-di-tertiarybutyl benzophenone
0.02% beta-ionone 0.1586 beta-ionone 0.0112 beta-ionone 0.0253 beta-ionone
0.2754 hexadecyl tetradecanoae
0.0155 hedione

186
0.0125 2-heptadecanone
 hydrocarbons 0.34% isopentane 0.305 isopentane 0.3693 isopentane 0.2327 isopentane
0.1127 isohexane
0.2011 hexane 0.1014 hexane
0.0361 methyl-cyclopentane 0.0528 methylcyclopentane
0.0357 2-methylpentane
0.1327 3-methylpentane 0.1311 3-methylpentane
0.0149 cyclohexane 0.0431 cyclohexane
0.1095 isoheptane 0.3013 c7 hydrocarbon
0.76% heptane 0.1462 heptane 0.0078 heptane 0.504 c7 hydrocarbon
0.0926 2,2-diemthylhexane 0.0126 cyclohexane
0.0528 c-7 hydrocarbon 0.259 c7 hydrocarbon 0.1514 c7 hydrocarbon
0.1012 methyl cyclohexane 0.2655 c8 hydrocarbon
0.36% toluene 0.4021 toluene 0.6075 toluene 1.0941 toluene
0.0174 1-octene
0.17 ethyl benzene 0.0512 ethyl benzene 0.0587 ethyl benzene
1.06% xylene isomer 0.0965 xylene isomer 0.0749 xylene isomer 0.33 xylene isomer
0.0413 styrene 0.1068 styrene 0.0601 styrene
0.0534 xylene isomer 0.0712 xylene isomer 0.1074 xylene isomer
0.0172 nonane 0.0294 c9 hydrocarbon 0.0351 nonane
0.0207 isopropyl benzene
0.0125 cumene
0.05% alpha-pinene 0.0903 alpha-pinene 0.1201 alpha-pinene 0.224 alpha-pinene
0.0293 camphene 0.0161 camphene
0.075 beta-pinene 0.0882 beta-pinene 0.119 beta-pinene
0.2648 trimethyl benzene isomer 0.1108 trimethyl benzene isomer 0.5066 trimethyl benzene isomer
0.06% decane 0.0496 decane 0.0883 decane
0.04% myrcene 0.0826 myrcene 0.1151 myrcene
0.0343 1,2-dichlorobenzene
0.0893 c10 hydrocarbon
0.08% delta-3-carene 0.0592 delta-3-carene 0.0999 delta-3-carene 0.1459 delta-3-carene
0.08% p-cymene 0.1249 p-cymene 0.2744 p-cymene 0.2039 para-cymene
Table S2 continued from previous page

1.78% limonene 1.7636 limonene 2.7012 limonene 3.9166 limonene

0.06% gamma-terpinene 0.0625 gamma-terpinene 0.0779 gamma-terpinene 0.1613 gamma-terpinene
0.04% alpha-para-dimethyl styrene
0.13% undecane 0.1302 undecane 0.2447 c11 hydrocarbon 0.2344 undecane
0.11% naphthalene 0.0742 naphthalene 0.277 naphthalene 0.2326 naphthalene
1.75% siloxane 2.3877 siloxane 2.4369 siloxane 5.4838 siloxane
0.02% 1-dodecene 0.0278 1-dodecene 0.76 c12 hydrocarbon 0.0484 1-dodecene
0.32% dodecane 0.3061 dodecane 0.0493 c13 hydrocarbon 0.4333 dodecane
0.02% 3,6-dimethyl-undecane (t) 0.0229 3,6-dimethyl-undecane 0.0351 3,6-dimethylundecane
0.02% 4-methylpentyl cyclohexane
0.0479 2-methylnaphthalene 0.0719 2-methylnaphthalene 0.0836 2-methyl naphthalene
0.30% tridecane 0.3288 tridecane 0.9444 tridecane 0.4502 tridecane
0.02% heptyl cyclohexane 0.0278 heptyl cyclohexane (t) 0.0525 heptyl cyclohexane 0.03 heptyl cyclohexane
0.13% siloxane 0.1536 siloxane 0.1869 siloxane 0.2644 siloxane
0.03% longicyclene 0.0281 longicyclene 0.0341 longicyclene 0.0074 longicyclene
0.0512 c14 hydrocarbon
0.0144 isolongifolene 0.0089 isolongifolene
0.02% longifolene 0.0438 longifolene 0.0868 longifolene
0.22% tetradecane 0.2005 tetradecane 0.5213 tetradecane 0.3939 tetradecane
0.03% 2,5-di-tert-butyl benzophenone
0.10% pentadecane 0.0817 pentadecane 0.1521 pentadecane
0.0157 alpha-bergamotene
0.1705 c15 hydrocarbon
0.0208 calamenene 0.0054 calamenene
0.0362 c16 hydrocarbon
0.04% hexadecane 0.0354 hexadecane 0.082 hexadecane 0.0602 hexadecane
0.5068 1-eicosene 0.5804 1-eicosene
0.06% tricosane 0.042 tricosane 0.0861 tricosane
0.0654 tetracosane

187
0.0987 1,1-diphenyl-2-cyano-2-carbo-octoxy-ethylene (t)
0.10% pentacosane
 0.59% 2,6,10,15,19,23-hexamethyl tetracosane 16.689 2,6,10,15,19,23-hexamethyltetracosane 7.1334 2,6,10,15,19,23-hexamethyltetracosane 3.235 2,6,10,15,19,23-hexamethyltetracosane
10.71% squalene 5.3017 squalene 11.76 squalene 15.465 squalene
0.54% nonacosane 0.1764 nonacosane 0.4316 nonacosane
0.8837 hentriacontane 0.3428 hentriacontane (c31)
0.19% heptatriacontane
1.0566 dotriacontane (c32 hydrocarbon)
1.2846 tritriacontane (c33 hydrocarbon)
1.142 tetratriacontane (c34 hydrocarbon)
1.0093 pentatriacontane (c35 hydrocarbon)

misc 0.48% sulfur dioxide 0.0974 sulfur dioxide 0.2235 sulfur dioxide
0.09% methylene chloride 0.0166 methylene chloride 0.3321 methylene chloride 0.0275 methylene chloride
0.0304 tetrahydrofuran
0.1115 pyrrolidine
0.3683 dimethyl disulfide
0.0115 2,6-dimethylpyrazine
0.0216 dimethyl trisulfide
0.06% n-methyl-2-pyrrolidone
0.32% benzothiazole 0.3814 benzothiazole 0.239 benzothiazole 0.1675 benzothiazole
0.0425 pyrrole, 2-carboxamide
0.039 dicyclohexyl
0.37% indole 0.0125 indole
0.00% diphenoxide 0.0167 diphenoxide 0.0192 diphenoxide
0.01% 2,4-dimethylquinoline 0.0101 2,4-dimethylquinoline 0.028 2,4-dimethylquinoline
0.023 p-tertiarybutyl-acetonitrile
0.0655 n-butylbenzene sulfonamide 0.2933 n-butylbenzene sulfonamide
0.5037 cyclo-(proline-proline)
0.0522 9-octadecenamide 0.0368 9-octadecenamide
Table S2 continued from previous page

0.241 neobee oil component 0.243 neobee oil component 2.37 neobee component
0.0193 octadecanamide
0.1108 13-docoseneamide 0.245 13-docosamide 0.8055 13-docosenamide

188