You are on page 1of 13

Archives of Biochemistry and Biophysics 602 (2016) 48e60

Contents lists available at ScienceDirect

Archives of Biochemistry and Biophysics

journal homepage:

Neutron protein crystallography: A complementary tool for locating

hydrogens in proteins
William B. O'Dell a, b, Annette M. Bodenheimer a, b, Flora Meilleur a, b, *
Department of Molecular and Structural Biochemistry, North Carolina State University, Raleigh, NC 27695, USA
Neutron Sciences Directorate, Oak Ridge National Laboratory, Oak Ridge, TN 37831, USA

a r t i c l e i n f o a b s t r a c t

Article history: Neutron protein crystallography is a powerful tool for investigating protein chemistry because it directly
Received 1 October 2015 locates hydrogen atom positions in a protein structure. The visibility of hydrogen and deuterium atoms
Received in revised form arises from the strong interaction of neutrons with the nuclei of these isotopes. Positions can be
12 November 2015
unambiguously assigned from diffraction at resolutions typical of protein crystals. Neutrons have the
Accepted 16 November 2015
Available online 22 November 2015
additional benet to structural biology of not inducing radiation damage in protein crystals. The same
crystal could be measured multiple times for parametric studies. Here, we review the basic principles of
neutron protein crystallography. The information that can be gained from a neutron structure is pre-
sented in balance with practical considerations. Methods to produce isotopically-substituted proteins
Crystallography and to grow large crystals are provided in the context of neutron structures reported in the literature.
Hydrogen Available instruments for data collection and software for data processing and structure renement are
Enzyme described along with technique-specic strategies including joint X-ray/neutron structure renement.
Protonation Examples are given to illustrate, ultimately, the unique scientic value of neutron protein crystal
Water structures.
2015 Elsevier Inc. All rights reserved.

1. Introduction Hydrogen atoms are central to enzyme chemistry, as ultimately,

reaction rates and chemistry are dependent upon the coordinated
Proteins, and more specically enzymes, are the lifeblood of changes in local electrostatics, hydrogen-bonding interactions, and
cellular processes. Enzymes catalyze a broad array of chemical re- protonation states of catalytic residues along the reaction coordi-
actions essential to life. No less than 20,000 distinctive enzyme- nate. Therefore, understanding enzyme chemistry at the atomic
catalyzed reactions are likely across all living organisms [1], con- level requires the visualization of hydrogen atoms on active site and
trolling processes as diverse as the capture and conversion of en- remote residues, co-factors, substrate, and water molecules. This
ergy from light, cell signaling, DNA translation and repair, and information can be challenging to obtain. While most structure-
metabolic pathways. Most of these catalytic functions exhibit an based knowledge arises from X-ray crystallography, hydrogen
exquisite chemical specicity, for both substrate and product, atoms are exceedingly difcult to visualize in X-ray structures.
which is encoded in the precise three-dimensional organization of When data are available to ultra-high resolution, few hydrogen
amino-acids at and leading to the active site of the enzyme. Enzyme atoms may be visible or their positions may be inferred from pre-
activity is tightly orchestrated and controlled by a variety of post- cise geometrical parameter analyses [6]. The combination of atomic
translational modications (phosphorylation, glycosylation, ubiq- resolution X-ray crystallographic data with quantum chemistry or
uitination, acetylation, lipidation) and allosteric regulation mech- charge density analysis can then provide a further level of detail on
anisms. Local environment, pH, and chemical ux also trigger the chemical prole of the enzyme [7e9]. However, even when
signicant structural reorganization contributing to activity mod- such ultra-high resolution data can be obtained, a signicant frac-
ulation [2e5]. tion (typically > 50%) of those more mobile or labile hydrogen
atoms remain difcult to discern, leaving specic questions con-
cerning catalytic mechanism unanswered.
* Corresponding author. Department of Molecular and Structural Biochemistry, The difculty of locating hydrogen atoms using X-ray crystal-
North Carolina State University, Raleigh, NC 27695, USA. lography can be circumvented by neutron protein crystallography.
E-mail address: (F. Meilleur).
0003-9861/ 2015 Elsevier Inc. All rights reserved.
W.B. O'Dell et al. / Archives of Biochemistry and Biophysics 602 (2016) 48e60 49

This is because the coherent scattering lengths of hydrogen (H) and length (3.99 fm) compared to H (25.27 fm) dramatically reduces the
the hydrogen isotope deuterium (D) for neutrons are similar in isotropic background intensity. The gain in signal and reduction in
magnitude to those of carbon, nitrogen and oxygen (Fig. 1) [10]. A noise allow datasets to be collected from smaller crystals
complication of neutron protein crystallography is that the scat- (<0.5 mm3) or with reduced total exposure times (<10 days)
tering length of hydrogen is negative while that of carbon (C), ni- [13,20,36e38]. Furthermore, the D coherent scattering length is
trogen (N) and oxygen (O) are positive, which gives rise to density positive as is that of common atoms found in proteins while the
cancelation in Fourier maps that can hamper interpretation and coherent scattering length of H is negative (Fig. 1). At the typical
analysis. In contrast, D has a neutron scattering length of the same resolutions (1.9e2.3 ) obtained for neutron protein crystallo-
sign as the heavier atoms and thus gives a clear positive peak in the graphic data, D substitution simplies the analysis of neutron
nuclear density maps. While visibility of hydrogen atoms requires density maps by minimizing the number of sites where the nega-
neutron crystallographic data at resolution of 2.0 or better, tive coherent scattering length of H atoms leads to density
deuterium atoms are readily visible, in crystallographic structures cancellation. Fig. 3 demonstrates the greater extent of signicant
identical to their hydrogenated counterparts [11e15], at typical neutron density in the 2FOFC map from a fully D-labeled, or per-
resolutions of 2.5 or better. deuterated, crystal of beta-lactamase as compared with the map
Despite a number of signicant technical challenges that we will from an H/D-exchanged crystal. Both datasets were measured on
discuss in this review, the ability to experimentally locate deute- the LADI III diffractometer at Institut Laue-Langevin and extend to
rium makes neutron diffraction a method of choice for visualizing 2.0 resolution [39]. However, data for perdeuterated beta lacta-
the positions of hydrogen in enzymes, whether used as a stand- mase were collected from a smaller crystal and in half the mea-
alone technique or in complementary with Nuclear Magnetic surement time than was required for the H/D-exchanged protein
Resonance (NMR) spectroscopy or high resolution X-ray crystal- [18,19]. In the case of aldose reductase, perdeuteration produced
lography. Neutron protein crystallography has, for example, been an improvement in diffraction resolution from 4.5 to 2.0 that
used to locate hydrogen atoms and determine the protonation enabled neutron data collection and structure determination
states at the active site of HIV-1 protease (HIV Pr) [16,17], beta- [13,40,41].
lactamase [18e20], dihydrofolate reductase (DHFR) [21,22], car-
bonic anhydrase II [23e26], and xylose isomerase [27e33]. These 2.1. H/D exchange
studies have been essential in identifying active site residues with
perturbed pKa's and contribute to the growing body of knowledge Bulk solvent, ordered water molecules, and hydrogen atoms
on pKa modulation at the active site of enzymes. bound to protein oxygen, nitrogen, or acidic carbon atoms such as
In this review, we focus on practical considerations to help guide histidine C1 [43] (i.e.: titratable hydrogen atoms) can all be
the successful development of neutron protein crystallography replaced with deuterium by preparing all reagents in deuterium
projects by newcomers. (The interested reader is encouraged to oxide (D2O) either during or after complete crystal growth. Of the
nd more historical perspectives on NPC in the works by Niimura neutron protein crystal structures currently deposited in the PDB
and Podjarny [34] and Blakeley [35].) A ow chart of the steps (Supplementary Material Table S1 and, approxi-
involved in neutron crystallographic structure determination is mately half were determined from proteins with titratable
presented in Fig. 2. We specically review H/D exchange protocols, hydrogen atoms exchanged during crystallization. In this approach,
crystal growth methods, data collection strategies, and structure the protein sample is exchanged by dialysis [44] or by repeated
renement options that have been used for the 85 neutron protein concentration and dilution into a D2O-prepared buffer solution.
structures currently deposited in the Protein Data Bank (October 1, Similarly, crystallization salts and/or precipitants are dissolved in
2015). We also give a perspective on the future of NPC illustrated D2O thus creating an H/D-exchanged mother liquor. The crystalli-
with recently determined neutron protein structures. zation experiment is then setup in the same way as that performed
with H2O solutions. Re-optimization of crystallization conditions
may be necessary for D2O solutions due to the reduced solubility of
2. Hydrogen/Deuterium isotopic substitution
proteins in D2O and the differences in physical properties of H2O
and D2O [45e47] (Table 1). The reduced solubility can be addressed
As stated previously, replacing H for D in a protein crystal pro-
by decreasing the crystallization agent concentration(s) [18,19,48],
vides signicant benets to the neutron protein crystallography
reducing the protein concentration [11,14,49], increasing the crys-
experiment. The greater magnitude of the D coherent scattering
tallization temperature [50,51], or by varying these parameters
length (6.671 fm) compared to H (3.741 fm) results in a larger
P ! simultaneously (see Crystal Growth).
F and, therefore, greater average reection intensities.
For the remaining structures deposited in the PDB, crystals were
Simultaneously, the six-fold decrease in the D incoherent scattering
grown to nal size in H2O mother liquor and then subjected to H/D
exchange either by vapor equilibration in capillary (Fig. 4) or
soaking in an articial H/D-exchanged mother liquor. Exchange
rates of individual hydrogen atoms in proteins in solution vary
greatly depending upon factors including hydrogen bonding, sol-
vent exclusion by protein folding, solvent pH, and incubation
temperature [52e55]. The crystal lattice further limits diffusion of
water to the network of pores formed by the packing of protein
molecules [45]. These factors make the choice of H/D equilibration
time arbitrary. In practice crystals are exchanged from the time the
Fig. 1. Incoherent neutron scattering cross sections and coherent neutron scattering crystal is fully grown until data collection. Reported times for H/D
lengths for selected elements. Relative incoherent scattering cross sections are rep- exchange after crystallization range from seven days for diisopropyl
resented by the left hemispheres, and relative coherent scattering lengths are repre- uorophosphatase [56,57] to ten years for a crystal of myoglobin
sented by the right hemispheres. The red hemisphere for hydrogen indicates the
negative sign of its scattering length while the others shown in green are positive.
[58]. By analyzing ten neutron structures deposited in the PDB,
Incoherent cross sections and coherent scattering lengths are not represented on the Bennett et al. [59] concluded that H/D exchange by crystal soaking
same scale. is likely to reach equilibrium after 30 days.
50 W.B. O'Dell et al. / Archives of Biochemistry and Biophysics 602 (2016) 48e60

Fig. 2. Neutron structure determination ow chart. The ow chart shows the steps in neutron protein crystallography. Most steps are similar to those of X-ray protein crystal-
lography. D-labeling or H/D exchange and growth of large crystals are specic steps for the neutron experiment.

Two structures of HIV Pr allow for an interesting comparison of 2.2. Perdeuteration

soaking and vapor diffusion exchange methods. Adachi et al. [16]
determined a structure of HIV Pr (PDB 2ZYE) from a crystal that Titratable hydrogen atoms represent approximatively 25% of the
was soaked for 14 days in articial D2O mother liquor with pD 5.0 total hydrogen atoms of any proteins. In order to exchange the
(pD pH 0.4) at 20  C and observed approximately 54% exchange remaining 75% attached to carbon atoms, proteins need to be per-
of backbone amide hydrogens. Weber et al. [17] also determined a deuterated during synthesis. Expression of perdeuterated protein
structure of HIV Pr (PDB 4JEC) in which approximately 71% of the in fully D-labeled growth media has yielded neutron protein crystal
backbone amide hydrogens exchanged after vapor equilibration structures currently deposited in the PDB for six proteins, namely
with D2O solution with pD 5.6 at 17  C for 28 days. Remarkably, as myoglobin [64,65], aldose reductase [13,41], transthyretin [38],
shown in Fig. 5, the patterns of backbone amide exchange in the type III antifreeze protein [66,67], beta lactamase [19,20], and
two structures are quite distinct despite very close agreement in rubredoxin [3,36,37,68,69].
atomic positions (<0.5 root mean square deviation). Multiple Each perdeuterated protein was expressed recombinantly in the
experimental distinctions may have contributed to these differ- host Escherichia coli. Cultures must be progressively adapted to
ences. The crystals were grown under signicantly different crys- grow in minimal media formulated with D2O, H/D-exchanged salts,
tallization conditions. The structures were also obtained with and uniformly D-labeled carbon source prior to induction of
different small molecule inhibitors bound at the HIV Pr active site, expression either in shake asks or in a bioreactor system. Methods
and it is known that different modes of ligand binding can inuence for perdeuterated protein expression specically for neutron pro-
backbone amide exchange patterns [60]. In addition, the 2ZYE tein crystallography have been described in detail by Meilleur et al.
model was rened using phenix.rene [61] while the 4JEC model [44,70] for cytochrome P40cam and Golden et al. [48] for choles-
was rened using CNSsolve [62]. terol oxidase. These studies describe two alternate approaches to
W.B. O'Dell et al. / Archives of Biochemistry and Biophysics 602 (2016) 48e60 51

experiments [71e74]. However, except for the case of HIV Pr, uni-
form partial deuterium labeling has not been applied in the eld of
neutron protein crystallography. Partial D-labeling does reduce
incoherent background in the diffraction experiment, but it re-
quires renement of the H/D-occupancies for all sites in the model
structure which considerably increases the number of parameters
to rene [17] (see Structure Renement). In addition, sites with H
and D occupancies of approximately 60% and 40% respectively
would also have null or insignicant density in Fourier maps due to
the negative H and positive D coherent scattering lengths further
complicating the structural analysis. More neutron diffraction ex-
periments using crystals of partial uniform D-labeled protein will
need to be conducted to fairly assess this approach.
Selective H-labeling in a perdeuterated background has been
applied in neutron diffraction studies of rubredoxin (PDB 3KYY)
[68,69] and type III antifreeze protein (PDB 4NY6) [4,75] with the
ultimate goal of using neutron data to solve protein structures by
direct methods [76]. Media supplementation with unlabeled a-
ketoisovalerate during perdeuterated protein expression resulted
in H-labeling of valine and leucine methyl groups producing (H-g
methyl)-valine and (H-d methyl)-leucine uniformly throughout the
proteins [77]. Fisher et al. [4] compared neutron diffraction datasets
for perdeuterated and CH3eH-labelled antifreeze protein to
demonstrate the negative effects of protein H atoms on the
observability of structured water molecules in neutron density
maps. Theoretically specic H-labeling in a perdeuterated back-
ground can enable direct determination of model phases for
structure solution; however, this application of neutron protein
crystallography with selective labeling has yet to be demonstrated

3. Crystal growth

The diffracted intensity in Bragg reections in a single crystal

Fig. 3. Neutron density maps from H/D-exchanged and perdeuterated beta-lactamase.
experiment can be written as:
Top: the 2FOFC map from the H/D-exchanged crystal (PDB 2WYX) is shown for res-
idue Arg 61. Bottom: the corresponding 2FOFC map from the perdeuterated crystal
(PDB 2XQZ) is shown. Both maps are contoured at 1.5 s. Structures were drawn using I0  jFj2  V  l2
If (1)
CCP4MG with coordinates taken from the PDB [42]. v20

where I, I0, jFj, V, l, and n0 are the diffracted intensity, incident

growing of E. coli in fully deuterated medium in a bioreactor: the beam intensity, structure factor magnitude, volume of the crystal,
former used glycerol-limited fed-batch cultivation while the latter incident wavelength and volume of the unit cell, respectively [78].
culture was grown until complete depletion of the culture broth Neutron beam uxes are inherently many orders of magnitude
initial carbon source content. Labeled proteins are generally iso- weaker than synchrotron X-ray beams [79]. Therefore, neutron
lated and puried using H2O-formulated buffers due to the cost of crystallography requires crystals that are at least three orders of
D2O and D-labeled reagents. The protein is then either exchanged magnitude larger than crystals suitable for X-ray crystallography
back to D2O buffer for crystallization or crystallized in H2O condi- because, as shown in Equation (1), the diffracted intensity is
tions and H/D-exchanged as discussed above. directly proportional to the incident beam intensity.
One structure of HIV Pr has been determined from 85% uniform The neutron crystallography experiment typically follows X-ray
D-labeled protein since it was observed that fully deuterated HIV Pr crystallography experiments that, despite providing high-
would not form crystals with sufcient volume for neutron resolution diffraction data, fell short of unambiguously answering
diffraction [17]. Uniformly D-labeled protein can be expressed with mechanistic questions due to the lack of information on hydrogen
95% D incorporation from E. coli, other prokaryotes, or even atom positions. All neutron crystallography projects have the
eukaryotic hosts using techniques largely developed for NMR starting advantage of known crystallization conditions, but these

Table 1
H2O and D2O physico-chemical properties.

Property H2O D2O

Bond energy, (kJ mol bond1, 0 K) 458.9 (HeOeH / O 2H ) 466.4 (DeOeD / O 2D )


Boiling point ( C) 100.0 101.42

Density (kg m3, 25  C) 997.05 1104.36
Dynamic viscosity (mPa s, 25 C) 0.8909 1.095
Melting point ( C) 0.00 3.82
Molar mass (g mol1) 18.015265 20.027508
pKW 13.995 14.870
52 W.B. O'Dell et al. / Archives of Biochemistry and Biophysics 602 (2016) 48e60

Fig. 4. H/D exchange of a fully grown crystal in the capillary mount by vapor diffusion. (1) Crystals are drawn into the capillary from the crystallization setup. (2) Excess mother
liquor is removed from the crystal, (3) exposing the crystal surface for exchange. (4) Plugs of articial mother liquor prepared in D2O are added on each side of the crystal to drive
the exchange in the closed environment provided by (5) the sealed capillary.

Fig. 5. Backbone amide H/D exchange observed in two neutron structures of HIV-1 protease. Left: A ribbon diagram of the structure obtained after crystal soaking for 14 days is
shown (PDB 2ZYE). Right: The corresponding diagram for the structure obtained after vapor diffusion H/D exchange for 28 days is shown (PDB 4JEC). In both structures, residues are
colored by the percent of D substitution observed (color bar) in the rened model. Structures were drawn using UCSD Chimera with coordinates from the PDB [63].

conditions most likely will need to be further optimized to sustain using this equation is provided as Supplementary Material.) In
the growth of large crystals suitable for neutron data collection. using this equation to plan crystallization experiments, one as-
Crystals typically need to be greater than 0.1 mm3 in volume sumes that only one crystal will grow per drop and that all of the
(Table S1 and reference herein) and must be well ordered to diffract protein in solution crystallizes. The estimation also relies on accu-
to sufcient resolution [80]. rate protein concentration determination.
It is important to note that while the beam ux is the dominant The readers are referred to the recent review by McPherson and
requirement for large crystals, the volume of the unit cell is also a Gavira [81] and to Gavira's review (this issue) for detailed protein
signicant parameter. Equation (1) shows that the diffracted in- crystallization principles. The most commonly utilized crystalliza-
tensity is inversely proportional to v20 . In cases where several tion method for neutron crystallography is vapor diffusion, with 35
crystal forms can be grown, crystallization conditions promoting out of the 85 neutron protein structures deposited in the PDB
smaller unit cells should be favored. Growing large crystals requires solved from crystals grown by vapor diffusion. Large 10e25 mL
that the crystallization setup contains sufcient amount of protein. hanging drops were used only for aldose reductase and diisopropyl
Equation (2) provides an estimation of the minimal volume of uorophosphatase [13,40,41,56,57]. Sitting drop crystallization al-
protein solution, Vmin, that should be used: lows for larger volume drops to be setup easily compared to
hanging drops. Sitting drop volumes reported range from 20 mL to
Vcrystal  Nasymmetric unit  Nprotein chain  MW 100's mL. Sandwich box setups (Hampton Research) consisting of a
Vmin  (2)
Vunit cell  NA  C siliconized nine-well glass plate placed inside a sealed plastic box,
are typically used to set up larger (>50 mL) volume sitting drops.
where Vcrystal is the target volume of the fully grown crystal, According to the manufacturer, each well is capable of holding drop
Nasymmetric unit is the number of asymmetric units for the expected sizes up to 800 mL, and the box, which serves as precipitant reser-
space group, Nprotein chain is the number of protein chains in the voir, can hold up to 25 mL of crystallization solution. The largest
asymmetric unit, MW is the molecular weight of the protein, Vunit reported sitting drops were setup with a total volume of 400 mL
cell is the volume of the unit cell, NA is the Avogadro number, and C is [17,82]. Neutron structures solved from crystals grown using the
the protein concentration. (A protein solution volume calculator sitting drop vapor diffusion method include carbonic anhydrase II,
W.B. O'Dell et al. / Archives of Biochemistry and Biophysics 602 (2016) 48e60 53

DHFR, and transthyretin [38,82,83]. Despite the fairly high success vary [89]. By changing the crystal growth temperature, one can
rate of the sitting drop vapor diffusion techniques, it should be drive the protein solution from a nucleating state to a metastable
noted that in about two thirds of the cases, the sitting drops are state or maintain the solution in the metastable state to promote
micro- or macroseeded as described below (also see Table S1). continued crystal growth. This can be performed using sophisti-
Batch crystallization is a useful alternative method with about cated equipment such as the temperature-controlled batch crys-
one third of the deposited structures reporting this technique tallization device developed and applied to urate oxidase by
(Table S1). Batch crystallization volumes for neutron crystallog- Budayova-Spano et al. [51,107]. On the other hand, simply trans-
raphy span from 10's of mL in the early days of protein crystalli- ferring crystals trays between incubators with different tempera-
zation [84e86] to volumes more compatible with modern protein ture set points or changing the incubator temperature by steps or
production of 100's of mL [20] and even down to 10 mL in a com- continuous gradients provides straightforward options to vary
bined batch/seeding crystallization approach such as that used by temperature, though in a less controlled fashion [16,87].
Matsumura et al. [87] to crystallize the HIV PreKNI-272 complex. As discussed above, titratable hydrogen atoms present in the
The batch method is usually applied to proteins available in large crystal should be exchanged to deuterium prior to data collection.
quantities such as xylose isomerase [33,46]. However, large batch Soaking, vapor diffusion, or crystallization in solution prepared
crystallizations are not a requirement during optimization for this with D2O can be used for H/D exchange. Crystallization in D2O
approach to be successful [88,89]. In fact, the batch method pre- solutions is a straightforward way to achieve H/D exchange, but this
sents the distinct advantage over other crystallization techniques approach often does not produce crystals suitable for neutron
that scaling to larger volumes does not affect the kinetic and diffraction or requires re-optimization of conditions. Soaking is the
thermodynamic factors that regulate the crystallization process. most efcient method for H/D exchange after crystal growth.
Crystallization conditions optimized using small batch volumes or However, crystals grown in H2O solution may suffer damage (e.g.:
even by high-throughput micro-batch screening can readily be cracks) from soaking in a fully deuterated solution. Gradual in-
scaled up. creases in D2O concentration of the soak solution may help prevent
Dialysis and counter diffusion have been used to a lesser extent such damage. A stepwise increase in D2O concentration was suc-
for neutron protein crystallography. Dialysis with protein solution cessfully applied to DHFR crystals [106]. A gentler approach to H/D
volumes of 200 mL and above has been performed for concanavalin exchange is vapor diffusion. Vapor diffusion H/D exchange can be
A using dialysis tubing [90e92], insulin using a micro-dialysis performed by replacing the hydrogenated mother liquor in the
chamber [93], and P450cam using dialysis buttons (Hampton reservoir with an articial deuterated mother liquor [108]. The
Research) [31]. Smaller volumes (50e100 mL) were dialyzed using well-to-drop ratio must be at least 10, and the D2O solution should
dialysis buttons for cytochrome c peroxidase [94,95] and insulin be replaced 3 times or more to maximize the exchange. If room
[96,97]. A customized counter diffusion apparatus was successfully temperature data collection is planned, H/D vapor exchange can
used to grow large inorganic pyrophosphatase crystals suitable for occur inside the capillary in which the crystal is mounted as shown
neutron data collection [98,99]. Dialysis, batch, and counter diffu- in Fig. 4. The largest possible volume of deuterated solution plug
sion setups can be supplemented with micro- and macro-seeding must be used for exchange in capillaries, and this plug must also be
and crystal feeding. In crystal feeding, fresh protein solution is repeatedly changed for new solution. For both the soak and vapor
added to a crystallization drop containing a crystal that appears to diffusion, longer exchange times permit greater exchange in the
have stopped growing. The crystal is typically fed several times. buried area of the protein and results in higher D incorporation.
This method was applied to Achromobacter protease I [100]. Special There is no systematic use of one exchange approach over
attention should be given to the crystal habit response to the another; the method used depends largely on the experimenter.
feeding process as the etching of the crystal surface that may occur However, data collection may restrict the choices. If data collection
can lead to further nucleation or growth of a multi-lattice crystal. at cryogenic temperature is planned, soaking in or vapor diffusion
Seeding techniques are powerful as they offer the opportunity to with a deuterated reservoir will be the safest approaches. For room
decouple nucleation from growth. Decoupling these two steps can temperature data collection, the crystals should be mounted and
be instrumental in growing large single crystals. Just under a third exchanged in a quartz capillary since borosilicate glass absorbs
of the crystals used to collect neutron data sets deposited in the neutrons (Fig. 4). Uses of thin and thick wall capillaries are both
PDB used micro- or macroseeding illustrating the usefulness of this reported in the literature. While thick wall capillaries generate
approach (Table S1). Seeding is used for both X-ray and neutron higher background, they are sturdier. The choice of the type of
crystallography when initial drops do not produce crystals with capillary is again an experimenter's preference.
sufcient volume or diffraction quality. For example, micro-seeding
was necessary to grow a large single crystal of aldose reductase 4. Instrumentation, data collection, and processing
[13]. Macroseeding was successfully applied to the thrombin-
bivalirudin complex [101], cholesterol oxidase [48], HIV Pr [16], Neutron crystallography is an inherently ux-limited technique.
rubredoxin [102], elastase [103], and the recently determined Available neutron uxes are several orders of magnitude less than
structure of farnesyl pyrophosphate synthase [104,105] (Table S1). the X-ray ux available from modern in-house generators and
In the last case, spontaneous nucleation and crystal growth was synchrotrons. This limitation is the primary impetus for increasing
conducted in large 400 mL sitting drops that produced crystals with the scattering power of the protein crystal through the isotopic
volumes up to 0.1 mm3. These crystals were then used to macro- labeling and crystal growth techniques discussed in the previous
seed 60 mL drops in which one crystal reached a nal volume of sections. Low available ux also results in the long exposure times
3.5 mm3. Interestingly, Bennett et al. [106] grew large crystals of required to measure a neutron diffraction dataset to sufcient
DHFRemethotrexate by vapor diffusion using large sitting drop resolution and completeness. While complete X-ray datasets can be
while Wan et al. [22] combined both micro- and macroseeding with measured in seconds or less at high-intensity X-ray sources,
vapor diffusion to grow DHFRefolateeNADP() crystals suitable neutron datasets typically require days to weeks for collection.
for neutron diffraction. However, through recent advances in instrumentation and labeling
While protein concentration, precipitant type and concentra- it has become possible to obtain useful neutron datasets in a single
tion, and pH are the most varied crystallization parameters during day or less in special cases [36].
optimization, temperature can also be an efcient parameter to Table 2 lists current single crystal diffractometers designed for
54 W.B. O'Dell et al. / Archives of Biochemistry and Biophysics 602 (2016) 48e60

Table 2
Diffractometers for neutron protein crystallography.

Instrument Source Mode of operation Wavelength () Status Refs.

BIX-3 JRR-3M, Japan monochromatic 2.35 operating [110]

BIX-4 JRR-3M, Japan monochromatic 2.6 operating [111]
Bio-C HANARO, South Korea monochromatic 1.95 commissioning [112]
BIODIFF FRM II, Germany monochromatic 2.4 operating [113]
LADI III ILL, France quasi-Laue 3.0e4.0 operating [39]
IMAGINE HFIR, USA quasi-Laue 2e3.0 operating [114]
PCS LANSCE, USA TOF-Laue 0.7e7.0 decommissioned [115]
MaNDi SNS, USA TOF-Laue 2.16e4.3 operating [116]
iBIX J-PARC, Japan TOF-Laue 0.7e3.85 operating [117]
NMX ESS, Sweden TOF-Laue design [118]

neutron protein crystallography at neutron scattering facilities provide neutrons in discrete pulses, and this time structure permits
worldwide. In addition to these instruments, other multi-purpose the use of a TOF-Laue mode for macromolecular diffractometers. A
single crystal diffractometers, such as the D19 instrument at ILL, relatively large wavelength range can be accepted by the instru-
have also been used for protein crystallography [5,28,37,38]. In- ment to maximize the number of stimulated reections. The de-
struments can be divided into three groups based upon mode of tectors record the positions, intensities, and time of incidence for
operation, namely monochromatic, quasi-Laue (also known as each scattered neutron from each pulse. The neutron time of ight
pink-Laue) and time-of-ight (TOF) Laue modes. All instrument can then be converted to the corresponding wavelength, and the
detector arrays are designed to maximize coverage of reciprocal observed quasi-Laue diffraction pattern can be binned or sepa-
space so that a large number of stimulated reections can be rated into reections and background for each incident wavelength.
recorded simultaneously whether the instrument uses mono- or TOF methods, therefore, provide both the increased number of
polychromatic incident radiation. Typically, neutron protein crys- reections of quasi-Laue mode and the reduced background of
tallography beam lines take advantage of longer wavelength neu- monochromatic mode.
trons to maximize the intensities of the diffraction peaks which are Data collection at two or more k angles is essential when using
proportional to the square of the incident wavelength, as shown in longer wavelength radiation because the curvature of the Ewald
Equation (1), and to reduce spatial overlap according to Bragg's law sphere is such that the blind region is signicant at any one k
(2d*sin(q) l). At the wavelength typically used for X-ray crystal- setting. In addition, instruments using a cylindrical detector ge-
lography (l  1.54 ), the diffracted beams lie on cones in the ometry (BIODIFF, BIO-C, BIX-3, BIX-4, IMAGINE, and LADI-III) have
forward scattering direction, and two-dimensional detectors are large, systematic vacancies in their coverage of reciprocal space
installed downstream of the crystal. At the longer wavelengths along the cylindrical long axis. Reections absent due to detector
(2.3 < l < 6.0 ) used by neutron diffractometers, the high res- coverage can be recovered by using more than one k angle.
olution diffraction also occurs in the backscattering direction. In- Sample handling for neutron diffraction experiments is gener-
strument detector arrays approximate cylindrical or spherical ally different from that for both in-house and synchrotron X-ray
geometries to record both forward- and backscattering reections. experiments due to the non-destructive nature of the neutron ra-
Existing monochromatic and quasi-Laue instruments are diation. Neutrons having wavelengths of 1.3e10 (<50 meV) do
installed at reactor neutron sources and take advantage of the high not induce ionization in biological samples. Therefore, cryogenic
time-integrated neutron ux provided by these facilities. Mono- sample temperatures are not necessary to protect against radiation
chromatic instruments operate analogously to standard X-ray damage in the protein crystal, and the majority of experiments are
macromolecular diffractometers with the exception of longer performed at ambient temperature. While room temperature data
incident wavelengths. Quasi-Laue instruments accept a dened, collection is typical, cryo-crystallography does, however, reduce
broad range of incident neutron wavelengths (Dl/lmid z 25e60%) thermal disorder in protein crystals and makes possible the study of
that greatly increases the number of stimulated reections at each reaction intermediates and temperature sensitive systems [119]. As
crystal setting relative to monochromatic mode. The broad wave- reviewed by Coates et al. [120] nitrogen cryostreams have been
length bandpass therefore is a more effective use of neutrons installed on a subset of neutron diffractometers to perform mea-
available from the reactor source, and the conguration also avoids surements at 100 K with sample mounting equipment standard for
the need for a monochromator crystal upstream of the sample that X-ray cryo-crystallography. For either temperature condition,
would inevitably have less than 100% reection efciency at any crystal mounting and alignment are often performed manually;
selected wavelength. Quasi-Laue represents a compromise on the however, the newest instruments such as MaNDi do incorporate
amount of observed incoherent background with it being more motorized sample handing and positioning under user control
than that observed with monochromatic radiation but less than [116]. Given that data collection time exceeds days, manual
that observed with truly white-beam radiation. While the centering is not a signicant overhead. Upon completion of data
monochromatic geometry produces data sets of higher quality collection, particularly at ambient temperature, the crystal is
compared to the quasi-Laue geometry (either for X-rays or neu- recovered. This same crystal can then be used for additional
trons), for neutron protein crystallography the quasi-Laue cong- neutron experiments or to collect an X-ray dataset for joint
uration has the unique advantage of allowing smaller crystals and/ renement of X-ray and neutron data (see Structure Renement).
or shorter exposure times to be used [109]. Diffraction data from reactor-based instruments can be indexed
Unlike reactor sources, accelerator-driven spallation sources and integrated using software packages developed for X-ray
W.B. O'Dell et al. / Archives of Biochemistry and Biophysics 602 (2016) 48e60 55

crystallography with appropriate modications to account for the produce lower quality X-ray data due to absorption, and radiation
instrument's detector geometry. Software packages used include damage from X-ray exposure would make the large crystal un-
Lauegen [121], d*TREK [122,123] and DENZO [124]. Handling the suitable for subsequent neutron parametric studies (Knihtila,
additional TOF information recorded by spallation-source in- Mattos, and Meilleur, unpublished results).
struments requires the development of dedicated data reduction Currently, there are three renement programs that can rene
packages [125,126]. Unlike monochromatic data, polychromatic neutron protein diffraction data: CNSsolve [62,139], phenix.rene
datasets (either for X-rays or neutrons) must be wavelength- [61,130,140], and SHELXL [141,142]. The phenix.rene module of the
normalized to account for the spectral shape of the incident PHENIX software package [140] is capable of rening X-ray,
beam. Software used to wavelength normalize data are also neutron, and joint X-ray/neutron data [61]. The familiarity of phe-
adapted from X-ray Laue crystallography software packages nix.rene to the crystallography community has made it the most
[121,127]. Data scaling is then performed with routine protein used package for neutron structure renement. Crystallography
crystallography software such as SCALEPACK [124] or SCALA [128]. and NMR system (CNS) was originally developed as an online
server to aid structure determination from solution NMR and X-ray
5. Structure renement crystallography data [139,143]. A neutron diffraction patch, termed
nCNS, can be added to CNS to perform renement of neutron or
Structure renement is carried out against neutron data alone or joint X-ray/neutron data with CNSsolve [62,130]. nCNS has been
simultaneously against both X-ray and neutron data. The latter used for 19 deposited neutron structures. SHELXL has been used to
protocol is referred to as joint renement or X/N renement [129]. rene neutron structures of crambin [144], xylose isomerase
The choice of one method of renement over the other may be [27,30], DHFR [21] and endothiapepsin [145]. Recently Gruene et al.
inuenced by the neutron dataset resolution and completeness but [142] modied SHELXL to include specic instructions for neutron
is ultimately the experimenter's choice. The number of parameters macromolecule structure renement.
to be rened for a neutron structure is considerably greater than for Renement against neutron data closely follows X-ray rene-
an X-ray structure: the number of atoms is nearly doubled, and ment protocols, and neutron density maps can be visualized using
~25% of the hydrogen positions require H/D-occupancy renement. the program COOT [146,147]. The starting model is a structure
However, typical neutron datasets contain fewer unique reections determined at the same temperature using X-ray data alone that
than their X-ray counterparts because of lower completeness and has been stripped of all ligands and water molecules. Deuterium or
resolution. Rening a model against both X-ray and neutron data hydrogen atoms are added at all non-titratable positions for per-
signicantly increases the data-to-parameter ratio and reduces the deuterated or hydrogenated proteins, respectively. Coordinate le
risk of over-tting the available data during renement [130]. editing can be performed manually or automatically by utilities
Joint renement is widely used for neutron structure renement such as phenix.ready_set [148]. Deuterium and hydrogen atoms,
against data collected from partially deuterated proteins at reso- each with half occupancy, are usually added to the titratable posi-
lutions below 2.2 . At these moderate resolutions, cancellations of tions when the crystal is H/D exchanged after crystallization.
neutron density for CH, CH2 and CH3 groups may obscure the direct Deuterium atoms with full occupancy can be added if the protein
modeling of side chains [61,130]. In joint renement the electron was crystallized from deuterated buffer [65,67,86,113,114,145,149].
density maps are used to rene the positions of heavy atoms while In this latter approach incompletely exchanged sites will show a
the corresponding neutron density maps are only used to rene negative density peak in the neutron 2FOFC map indicating that
hydrogen atom positions [129,130]. Examples of proteins for which the site is partially occupied by hydrogen [36]. At this initial state of
both renement against neutron data alone and joint renement renement, titratable groups are assumed to be in their expected
have been applied include photoactive yellow protein (PYP), protonation state based on the normal pKa value of the individual
transthyretin, and DHFR. Yamagushi et al. [131] rened the neutron amino acid in solution.
structure of PYP at 1.5 against neutron data alone while Fisher An initial rigid body renement and successive coordinate,
et al. [132] performed joint renement to determine the PYP atomic displacement parameter, and occupancy renements are
neutron structure at 2.5 . Both renement strategies have been performed in alternation with local rebuilding of the model
applied to neutron structures of transthyretin at equal resolution including the addition of small molecules. Water molecules can be
(2.0 ) [38,133] and to the neutron structures of DHFR at 2.0 [22] modeled as O, OD, OD, DOD, and DO3 depending on their degree
and 2.17 [21] resolutions. Protein perdeuteration prevents density of rotational anisotropy, neighboring groups, and ionic state
cancellation by hydrogen atoms, but joint renement is still broadly [32,37,150]. Water molecules can be oriented automatically by the
applied to data collected from perdeuterated crystals. The neutron renement software. However, visual inspection of the model and
structures of perdeuterated aldose reductase [41], transthyretin neutron density maps should be performed to conrm the auto-
[38], and type III antifreeze protein [67] were jointly rened. matically rened orientations. Experimenters should consider
Structures of perdeuterated beta-lactamase [19,20], myoglobin manual orientation of the water molecules to form hydrogen bonds
[65], and rubredoxin [36] were solved using neutron data alone. with accepted geometries when possible [151]. Map observations
Joint renement requires that the X-ray data set be collected at will also inform the modeling of titratable group protonation states.
the same temperature as the neutron data set. Collecting both data Careful interpretation of the maps is essential to model abnormal
sets at the same temperature is essential as temperature alters or perturbed ionization states. Here the neutron FOFC maps are
aspects of the protein structure or of the crystal itself [134e138]. X- used to model hydrogen and deuterium atoms. Positive map fea-
ray-induced radiation damage can be more severe in crystals tures directly indicate protonation of carboxylate, hydroxyl, amine,
measured at room temperature and can produce signicant struc- and imine groups. Proper orientation of these groups, guided by the
tural artifacts. Radiation damage can be minimized at room tem- neutron FOFC maps, often clearly denes hydrogen-bond net-
perature by using short exposure times and strong attenuation of works that may be ambiguous if only the electron density maps are
high-intensity incident beams. While the X-ray and neutron data considered. A negative FOFC feature at a titratable site indicates an
set would ideally be collected from the same crystal for joint overestimation of deuterium occupancy at this site. Deuterium
renement, the X-ray dataset is most often measured from a atoms are readily visible as positive features in neutron 2FOFC
smaller crystal grown in the same drop or under the same crys- maps at resolution 2.5 . It should be noted that while hydrogen
tallization conditions as the neutron crystal. Large crystals typically atoms included in the coordinate le will contribute to negative
56 W.B. O'Dell et al. / Archives of Biochemistry and Biophysics 602 (2016) 48e60

features in the neutron 2FOFC map, hydrogen atoms are usually [165,166]. The neutron structure solved by [133] evealed an
only visible at higher resolutions (2.0 ). extensive hydrogen bond network connecting six residues and
four water molecules located along both the monomer-
6. Perspective emonomer and dimeredimer interfaces. One of the residues
involved in the hydrogen bond network is the singly protonated
Atomic comprehension of protein chemistry requires knowing His88, which stabilizes the dimeredimer interface. When the
the simultaneous locations of numerous hydrogen atoms. Specif- protonation state of His88 is altered, the interfacial hydrogen
ically for enzymes, the interactions driving chemistry generally bond network is destabilized. Comparison of the neutron
cannot be unambiguously deduced solely from knowledge of the structure solved at pD 7.4 with a previously solved X-ray
protonation states of individual titratable side chains within the structure at pH 4.0 shows that the salt bridge between Lys76
active site [4]. It is widely accepted that remote amino acids and Glu89, the hydrogen bond between Asp74 and Ser77, and
[152e156] or water molecules [157e159], whether buried in the the hydrogen bond network including His88 are compromised
protein structure or bridging the active site to the hydration shell, at low pH. [38] lso described the extensive hydrogen bond
play essential roles in protein chemistry. At the active sites of en- network between monomeremonomer interface in their struc-
zymes, both the chemical identities of ligands, substrates and water tures leading to the conclusion that the dimer of TTR is the
molecules (H2O, OH, or H3O) and the hydrogen bond networks building block of amyloid brils. The neutron structures of TTR
formed must be known for a complete mechanistic understanding. increased the understanding of pH sensitivity and suggested a
Furthermore, the modulation of active site pKa's by the local elec- potential mechanism of amyloid formation.
trostatic environment must be understood. In the last year two neutron structures revealed unexpected
The neutron crystal structures of human farnesyl pyrophos- protonation in key states of established enzymatic mechanisms.
phate synthase (FPPS) in complex with the clinical inhibitor The neutron structure of the small GTPase RAS further illustrates
risedronate and of human transthyretin (TTR) illustrate these the uniquely rich information content provided by a single neutron
points. FPPS catalyzes condensation of dimethylallyl pyrophos- diffraction experiment [108]. This study suggests that the g-phos-
phate (DMAPP) with isopentyl pyrophosphate (IPP) yielding ger- phate group of the GTP substrate remains protonated at physio-
anyl pyrophosphate (GPP) which further condenses with a second logical pH when bound at the active site of RAS in the ground state.
IPP molecule yielding farnesyl pyrophosphate. Nitrogen-containing Efforts to elucidate the GTP hydrolysis mechanism of RAS have
bisphosphonates (N-BP), such as risedronate, inhibit FPPS by relied on the assumption that the g-phosphate group of GTP bound
occupying the DMAPP/GPP binding site [160]. N-BPs bind the FPPS at the active site of the enzyme would be deprotonated as is
active site with a high afnity that was proposed to result from a observed in solution. The neutron structure revealed for the rst
protonated nitrogen mimicking a carbocation intermediate of the time the modulation of the pKa of the g-phosphate group by the
FPPS condensation reaction [161,162]. The X-ray crystal structure of RAS active site microenvironment and sets a new stage for future
the FPPSerisedronate complex demonstrated that the risedronate investigation. As stated previously, neutron structures are usually
pyridine nitrogen was positioned within hydrogen bonding dis- determined at room temperature but neutron data collection at
tance of the Lys200 backbone carbonyl and the Thr201 hydroxyl cryogenic is necessary to characterize transient states. Intermediate
group. The neutron crystal structure of H/D-exchanged FPPSeri- freeze trapping was successfully applied to cytochrome c peroxi-
sedronate at pD 5.0 determined by Yokoyama et al. [104] con- dase to determine the structure of ferryl heme intermediates and
rms the existence of these hydrogen-bond interactions. A unambiguously determined the protonation state of the iron-oxo
deuterium atom binds to the pyridine nitrogen and participates in a species (also known as compound I) [94]. The study demonstrates
bifurcated hydrogen bond with Lys200 and Thr201. This proton- that compound I is deprotonated supporting earlier data. Com-
ation agrees with prior studies of pKa's for the risedronate titrat- parison of the room temperature neutron ferric cytochrome c
able groups in solution. However, the neutron structure revealed perodidase structure and of the cryo neutron structure of com-
that both risedronate phosphonate groups are fully deprotonated pound I critically revealed that the heme distal histidine becomes
when the drug is bound to FPPS despite the rst deprotonation of protonated. Hence both the neutron studies of Ras and cytochrome
the phosphonates in free risedronate having a pKa >10.5 [163,164]. c peroxidase prompt future calculations and modelling.
The phosphonate groups together coordinate three Mg2 ions and Readers are referred to the recent review by Golden and Vrielink
participate in hydrogen bonds with Arg112, Lys200 and Lys257 [167] for additional examples of neutron crystallographic struc-
sidechains and four ordered water molecules. While unexpected, tures providing insights into hydronium ions, oxyanion holes, low
phosphonate deprotonation agrees well with a FPPS mechanism in barrier hydrogen bonds, and proteineligand interactions.
which a basic pyrophosphate oxygen, stabilized by Arg112 and NMR spectroscopy, ultra-high resolution X-ray crystallography,
Lys200, abstracts the IPP 2R proton promoting nucleophilic attack and neutron crystallography contribute to a global understanding
of the carbocation intermediate [161,162]. These two mechanistic of protonation. Taken individually, each technique has specic
insights resulted directly from unambiguous delineation of the limitations. Acidebase titrations can be monitored by NMR spec-
residue, ligand, and water protonation states at the enzyme active troscopy to determine pKa values directly [168e170]. However, pH
site by the neutron crystallography experiment. titration methods can only be applied to proteins that are stable
The neutron structures of human transthyretin (TTR) reveal key over a wide range of pH, and the experiment does not simulta-
hydrogen bond networks and pH sensitivity leading to a new un- neously report on residue conformation. Residues surrounding
derstanding of substrate binding and amyloid formation. TTR is titratable sites also have chemical and structural behaviors sensi-
present in the blood, where it binds retinol and thyroxine (T4), and tive to pH that may complicate chemical shift assignments and pKa
in the cerebrospinal uid, where it binds T4. The TTR protein is a determination. In addition to probing pKa's, long residence-time
homotetramer, or a dimer of dimers, held together through an water molecules can be localized using NMR spectroscopy, but
extensive network of hydrogen bonds and hydrophobic in- water molecule orientations are difcult to determine [171]. For X-
teractions. Partial denaturation of TTR, which occurs in acidic ray crystallography, the weak X-ray scattering power of hydrogen
conditions, leads to aggregate structures that include amyloid bril atoms inherently limits the available experimental information.
formation. Previous TTR X-ray crystal structures did not display The protonation state of some titratable sites can be determined
signicant conformational changes when the pH was altered indirectly from X-ray structures by covalent bond geometry
W.B. O'Dell et al. / Archives of Biochemistry and Biophysics 602 (2016) 48e60 57

analysis. High resolution X-ray crystallography proves not to be the Best-Pratice Methods, Elsevier, 2010.
[2] S.K. Whittier, A.C. Hengge, J.P. Loria, Conformational motions regulate
most suitable technique to visualize hydrogens directly [40,69].
phosphoryl transfer in related protein tyrosine phosphatases, Science 341
Neutron protein crystallography directly identies H/D atoms, but (6148) (2013) 899e903.
is inherently limited by the ux of the available neutron beams. [3] P. Hanoian, et al., Perspectives on electrostatics and conformational motions
However, when large crystals can be grown, a single neutron in enzyme catalysis, Acc. Chem. Res. 48 (2) (2015) 482e489.
[4] T.K. Harris, G.J. Turner, Structural basis of perturbed pKa values of catalytic
structure informs simultaneously on all aspects of catalytic activity: groups in enzyme active sites, IUBMB Life 53 (2) (2002) 85e98.
structural conformation and protonation state at the active site, [5] A. Gutteridge, J.M. Thornton, Understanding nature's catalytic toolkit, Trends
conformation and protonation of remote residues, chemical iden- Biochem. Sci. 30 (11) (2005) 622e629.
[6] P. Neumann, K. Tittmann, Marvels of enzyme catalysis at true atomic reso-
tities and orientations of water molecules, and the organization of lution: distortions, bond elongations, hidden ips, protonation states and
hydrogen-bond networks. atom identities, Curr. Opin. Struct. Biol. 29 (2014) 122e133.
Neutron crystallography is a powerful complementary approach [7] C. Jelsch, et al., Accurate protein crystallography at ultra-high resolution:
valence electron distribution in crambin, Proc. Natl. Acad. Sci. U. S. A. 97 (7)
to NMR spectroscopy and X-ray crystallography. Neutron crystal- (2000) 3171e3176.
lography and NMR spectroscopic measurements following the [8] D. Liebschner, et al., Elucidation of the phosphate binding mode of DING
titration of tyrosine residues were elegantly combined in a study of proteins revealed by subangstrom X-ray crystallography, J. Am. Chem. Soc.
131 (22) (2009) 7879e7886.
carbonic anhydrase II by Michalczyk et al. [26]. NMR revealed the [9] B. Zarychta, et al., Cholesterol oxidase: ultrahigh-resolution crystal structure
perturbed pKa of Tyr7, and the neutron crystal structures provided and multipolar atom model-based analysis, Acta Crystallogr. D. Biol. Crys-
a structural context. Similarly, sub-atomic resolution X-ray and tallogr. 71 (Pt 4) (2015) 954e968.
[10] V.F. Sears, Neutron scattering lengths and cross sections, Neutron News 3 (3)
neutron crystallography were combined to describe in further
(1992) 26e37.
detail the catalytic mechanism of DHFR [22]. [11] L. Di Costanzo, et al., Expression, purication, assay, and crystal structure of
As we have discussed in this review, neutron protein crystal- perdeuterated human arginase I, Archives Biochem. Biophysics 465 (1)
lography does present a number of technical challenges. Deuterium (2007) 82e89.
[12] S.J. Fisher, J.R. Helliwell, An investigation into structural changes due to
incorporation is essential for neutron diffraction experiments, but deuteration, Acta Crystallogr. Sect. A 64 (2008) 359e367.
the studies reviewed here illustrate that the incorporation required [13] I. Hazemann, et al., High-resolution neutron protein crystallography with
is readily achievable. The exchange of titratable hydrogen atoms radically small crystal volumes: application of perdeuteration to human
aldose reductase, Acta Crystallogr. D. Biol. Crystallogr. 61 (Pt 10) (2005)
sufces when full perdeuteration is not practical. Many crystals will 1413e1417.
likely endure vapor diffusion H/D exchange. Data collection and [14] F. Meilleur, et al., Production and X-ray crystallographic analysis of fully
structure renement methods are now well established, and deuterated cytochrome P450cam, Acta Crystallogr. D. Biol. Crystallogr. 61 (Pt
5) (2005) 539e544.
various strategies can be selected to best suit the crystal system and [15] J.B. Artero, et al., A comparison of rened X-ray structures of hydrogenated
scientic question being addressed. and perdeuterated rat gE-crystallin in H2O and D2O, Acta Crystallogr. D. Biol.
Growing large protein crystals remains as the nal hurdle for the Crystallogr. 61 (Pt 11) (2005) 1541e1549.
[16] M. Adachi, et al., Structure of HIV-1 protease in complex with potent in-
widespread pursuit of neutron protein crystallography. Clearly, all hibitor KNI-272 determined by high-resolution X-ray and neutron crystal-
proteins are not amenable to the growth of large crystals. However, lography, Proc. Natl. Acad. Sci. U. S. A. 106 (12) (2009) 4641e4646.
a signicant fraction of proteins that produce crystals with X-ray [17] I.T. Weber, et al., Joint X-ray/neutron crystallographic study of HIV-1 pro-
tease with clinical inhibitor amprenavir: insights for drug design, J. Med.
diffraction limit 1.5 (currently, approximately 5000 proteins
Chem. 56 (13) (2013) 5631e5635.
meet this criteria) is likely to be amenable to neutron diffraction. [18] S.J. Tomanicek, et al., Neutron diffraction studies of a class A beta-lactamase
The foreseen brightness of neutron beams at facilities currently Toho-1 E166A/R274N/R276N triple mutant, J. Mol. Biol. 396 (4) (2010)
being built (European Spallation Source) or proposed (Second 1070e1080.
[19] S.J. Tomanicek, et al., The active site protonation states of perdeuterated
Target Station at the Spallation Neutron Source), combined with Toho-1 beta-lactamase determined by neutron diffraction support a role for
novel instrument concepts will contribute to further lowering the Glu166 as the general base in acylation, FEBS Lett. 585 (2) (2011) 364e368.
crystal size requirement. The application of polarization techniques [20] S.J. Tomanicek, et al., Neutron and X-ray Crystal structures of a perdeuterated
enzyme inhibitor complex reveal the catalytic proton network of the Toho-1
for neutron protein crystallography will open up new opportunities beta-lactamase for the acylation reaction, J. Biol. Chem. 288 (7) (2013)
in studying fully hydrogenated protein crystals [172,173]. These 4715e4722.
developments will undoubtedly expand the opportunities to [21] B. Bennett, et al., Neutron diffraction studies of Escherichia coli dihydrofolate
reductase complexed with methotrexate, Proc. Natl. Acad. Sci. U. S. A. 103
explore protein chemistry at the atomic level. (49) (2006) 18493e18498.
[22] Q. Wan, et al., Toward resolving the catalytic mechanism of dihydrofolate
Acknowledgments reductase using neutron and ultrahigh-resolution X-ray crystallography,
Proc. Natl. Acad. Sci. U. S. A. 111 (51) (2014) 18225e18230.
[23] S.Z. Fisher, et al., Neutron diffraction of acetazolamide-bound human car-
We thank Dean Myles for carefully reading our manuscript and bonic anhydrase II reveals atomic details of drug binding, J. Am. Chem. Soc.
providing input to improve the review. We also thank Steve Tom- 134 (36) (2012) 14726e14729.
[24] S.Z. Fisher, et al., Neutron structure of human carbonic anhydrase II: impli-
anicek and Leighton Coates for providing crystallographic data not cations for proton transfer, Biochemistry 49 (3) (2010) 415e421.
available in the PDB. WBO and AMB are supported by the NSF [25] Z. Fisher, et al., Neutron structure of human carbonic anhydrase II: a
foundation under Grant No. 1069091 and the US Department of hydrogen-bonded water network switch is observed between pH 7.8 and
10.0, Biochemistry 50 (44) (2011) 9421e9423.
Energy Ofce of Basic Energy Sciences (DOE BES) under the Grad- [26] R. Michalczyk, et al., Joint neutron crystallographic and NMR solution studies
uate Opportunities (GO!) program at Oak Ridge National Labora- of Tyr residue ionization and hydrogen bonding: Implications for enzyme-
tory. FM is supported by the College of Agriculture and Life Sciences mediated proton transfer, Proc. Natl. Acad. Sci. U. S. A. 112 (18) (2015)
of North Carolina State University and the DOE BES.
[27] A.K. Katz, et al., Locating active-site hydrogen atoms in D-xylose isomerase:
time-of-ight neutron diffraction, Proc. Natl. Acad. Sci. U. S. A. 103 (22)
Appendix A. Supplementary data (2006) 8342e8347.
[28] A. Kovalevsky, et al., Inhibition of D-xylose isomerase by polyols: atomic
details by joint X-ray/neutron crystallography, Acta Crystallogr. D. Biol.
Supplementary data related to this article can be found at http:// Crystallogr. 68 (Pt 9) (2012) 1201e1206. [29] A.Y. Kovalevsky, et al., Metal ion roles and the movement of hydrogen during
reaction catalyzed by D-xylose isomerase: a joint X-ray and neutron
diffraction study, Structure 18 (6) (2010) 688e699.
References [30] A.Y. Kovalevsky, et al., Hydrogen location in stages of an enzyme-catalyzed
reaction: time-of-ight neutron structure of D-xylose isomerase with
[1] D.L. Purich, Enzyme Kinetics. Catalysis & Control: A Reference of Theroy and bound D-xylulose, Biochemistry 47 (29) (2008) 7595e7597.
58 W.B. O'Dell et al. / Archives of Biochemistry and Biophysics 602 (2016) 48e60

[31] F. Meilleur, Etudes Cristallographiques Aux Rayons X et Aux Neutrons du [62] M. Mustyakimov, P. Langan, NM: Los Alamos National Security, 2007.
Cytochrome P450cam et de la D-xylose Isomerase, Ph. D. thesis (English), Copyright C-06, 104 Patch for CNS; nCNS an open source distribution patch
2004. for CNS macromolecular structure renement.
[32] F. Meilleur, et al., A quasi-Laue neutron crystallographic study of D-xylose [63] E.F. Pettersen, et al., UCSF chimera e a visualization system for exploratory
isomerase, Eur. Biophys. J. 35 (7) (2006) 601e609. research and analysis, J. Comput. Chem. 25 (13) (2004) 1605e1612.
[33] P. Munshi, et al., Neutron structure of the cyclic glucose-bound xylose [64] F. Shu, V. Ramakrishnan, B.P. Schoenborn, High-level expression and
isomerase E186Q mutant, Acta Crystallogr. D. Biol. Crystallogr. 70 (Pt 2) deuteration of sperm whale myoglobin. A study of its solvent structure by X-
(2014) 414e420. ray and neutron diffraction methods, Basic Life Sci. 64 (1996) 309e323.
[34] N. Niimura, A.D. Podjarny (Eds.), Neutron Protein Crystallography: Hydrogen, [65] F. Shu, V. Ramakrishnan, B.P. Schoenborn, Enhanced visibility of hydrogen
Protons, and Hydration in Bio-macromolecules, IUCr Monographs on Crys- atoms by neutron crystallography on fully deuterated myoglobin, Proc. Natl.
tallography, vol. 25, Oxford University Press, Oxford, 2011. Acad. Sci. U. S. A. 97 (8) (2000) 3872e3877.
[35] M.P. Blakeley, Neutron macromolecular crystallography, Crystallogr. Rev. 15 [66] I. Petit-Haertlein, et al., Perdeuteration, purication, crystallization and
(3) (2009) 157e218. preliminary neutron diffraction of an ocean pout type III antifreeze protein,
[36] P. Munshi, et al., Rapid visualization of hydrogen positions in protein neutron Acta Crystallogr. F. Struct. Biol. Cryst. Commun. 65 (Pt 4) (2009) 406e409.
crystallographic structures, Acta Crystallogr. D. Biol. Crystallogr. 68 (Pt 1) [67] E.I. Howard, et al., Neutron structure of type-III antifreeze protein allows the
(2012) 35e41. reconstruction of AFP-ice interface, J. Mol. Recognit. 24 (4) (2011) 724e732.
[37] M.G. Cuypers, et al., Near-atomic resolution neutron crystallography on [68] K.L. Weiss, et al., Preliminary neutron crystallographic analysis of selectively
perdeuterated pyrococcus furiosus rubredoxin: implication of hydronium CH3-protonated deuterated rubredoxin from Pyrococcus furiosus, Acta
ions and protonation state equilibria in redox changes, Angew. Chem. Int. Ed. Crystallogr. F. Struct. Biol. Cryst. Commun. 64 (2008) 537e540.
52 (3) (2013) 1022e1025. [69] A.S. Gardberg, et al., Unambiguous determination of H-atom positions:
[38] M. Haupt, et al., Binding site asymmetry in human transthyretin: insights comparing results from neutron and high-resolution X-ray crystallography,
from a joint neutron and X-ray crystallographic analysis using perdeuterated Acta Crystallogr. D. Biol. Crystallogr. 66 (Pt 5) (2010) 558e567.
protein, IUCrJ 1 (6) (2014) 429e438. [70] F. Meilleur, K.L. Weiss, D.A. Myles, Deuterium labeling for neutron structure-
[39] M.P. Blakeley, et al., Neutron macromolecular crystallography with LADI-III, function-dynamics analysis, Methods Mol. Biol. 544 (2009) 281e292.
Acta Crystallogr. D. Biol. Crystallogr. 66 (Pt 11) (2010) 1198e1205. [71] K.H. Gardner, L.E. Kay, The use of 2H, 13C, 15N multidimensional NMR to study
[40] M.P. Blakeley, et al., Comparison of hydrogen determination with X-ray and the structure and dynamics of proteins, Annu. Rev. Biophysics Biomol. Struct.
neutron crystallography in a human aldose reductase-inhibitor complex, Eur. 27 (1998) 357e406.
Biophys. J. 35 (7) (2006) 577e583. [72] W.D. Morgan, A. Kragt, J. Feeney, Expression of deuterium-isotope-labelled
[41] M.P. Blakeley, et al., Quantum model of catalysis based on a mobile proton protein in the yeast pichia pastoris for NMR studies, J. Biomol. NMR 17 (4)
revealed by subatomic X-ray and neutron diffraction studies of h-aldose (2000) 337e347.
reductase, Proc. Natl. Acad. Sci. U. S. A. 105 (6) (2008) 1844e1848. [73] T. Etezady-Esfarjani, et al., Cell-free protein synthesis of perdeuterated
[42] S. McNicholas, et al., Presenting your structures: the CCP4mg molecular- proteins for NMR studies, J. Biomol. NMR 39 (3) (2007) 229e238.
graphics software, Acta Crystallogr. D. Biol. Crystallogr. 67 (2011) 386e394. [74] M. Kainosho, P. Guntert, SAIL - stereo-array isotope labeling, Q. Rev. Biophys.
[43] N. Niimura, et al., High resolution neutron protein crystallography. Hydrogen 42 (4) (2009) 247e300.
and hydration in proteins, Z. Fur Kristallogr. 218 (2) (2003) 96e107. [75] I. Petit-Haertlein, et al., Incorporation of methyl-protonated valine and
[44] F. Meilleur, et al., Structural stability and dynamics of hydrogenated and leucine residues into deuterated ocean pout type III antifreeze protein:
perdeuterated cytochrome P450cam (CYP101), Biochemistry 43 (27) (2004) expression, crystallization and preliminary neutron diffraction studies, Acta
8744e8753. Crystallogr. F. Struct. Biol. Cryst. Commun. 66 (Pt 6) (2010) 665e669.
[45] N. Sukumar, et al., A preliminary time-of-ight neutron diffraction study on [76] H.A. Hauptman, D.A. Langs, The phase problem in neutron crystallography,
amicyanin from Paracoccus denitricans, Acta Crystallogr. D. Biol. Crys- Acta Crystallogr. A 59 (Pt 3) (2003) 250e254.
tallogr. 61 (Pt 5) (2005) 640e642. [77] N.K. Goto, et al., A robust and cost-effective method for the production of Val,
[46] E.H. Snell, et al., Optimizing crystal volume for neutron diffraction: D-xylose Leu, Ile (delta 1) methyl-protonated 15N-, 13C-, 2H-labeled proteins, J. Biomol.
isomerase, Eur. Biophys. J. 35 (7) (2006) 621e632. NMR 13 (4) (1999) 369e374.
[47] M. Budayova-Spano, et al., A preliminary neutron diffraction study of ras- [78] A.J. Schultz, et al., Conceptual design of a macromolecular neutron diffrac-
buricase, a recombinant urate oxidase enzyme, complexed with 8- tometer (MaNDi) for the SNS, J. Appl. Crystallogr. 38 (2005) 964e974.
azaxanthin, Acta Crystallogr. F. Struct. Biol. Cryst. Commun. 62 (2006) [79] F. Meilleur, D.A. Myles, M.P. Blakeley, Neutron Laue macromolecular crys-
306e309. tallography, Eur. Biophys. J. 35 (7) (2006) 611e620.
[48] E. Golden, et al., Production and characterization of recombinant perdeu- [80] A. McPherson, Y.G. Kuznetsov, Mechanisms, kinetics, impurities and defects:
terated cholesterol oxidase, Anal. Biochem. 485 (2015) 102e108. consequences in macromolecular crystallization, Acta Crystallogr. Sect. F
[49] I. Schlichting, C. Jung, H. Schulze, Crystal structure of cytochrome P-450cam Struct. Biol. Commun. 70 (2014) 384e403.
complexed with the (1S)-camphor enantiomer, FEBS Lett. 415 (3) (1997) [81] A. McPherson, J.A. Gavira, Introduction to protein crystallization, Acta Crys-
253e257. tallogr. F. Struct. Biol. Commun. 70 (Pt 1) (2014) 2e20.
[50] M. Budayova-Spano, et al., Comparison of solubility and interactions of [82] S.Z. Fisher, et al., Preliminary joint neutron and X-ray crystallographic study
aprotinin (BPTI) solutions in H2O and D2O, J. Cryst. Growth 217 (3) (2000) of human carbonic anhydrase II, Acta Crystallogr. F. Struct. Biol. Cryst.
311e319. Commun. 65 (Pt 5) (2009) 495e498.
[51] E. Oksanen, et al., Large crystal growth by thermal control allows combined [83] Q. Wan, et al., Preliminary joint X-ray and neutron protein crystallographic
X-ray and neutron crystallographic studies to elucidate the protonation studies of ecDHFR complexed with folate and NADP, Acta Crystallogr. F.
states in Aspergillus avus urate oxidase, J. R. Soc. Interface 6 (Suppl. 5) Struct. Biol. Cryst. Commun. 70 (Pt 6) (2014) 814e818.
(2009) S599eS610. [84] J.J. Abel, Crystalline insulin, Proc. Natl. Acad. Sci. U. S. A. 12 (2) (1926)
[52] A. Hvidt, K. Linderstromlang, Exchange of hydrogen atoms in insulin with 132e136.
deuterium atoms in aqueous solutions, Biochim. Biophys. Acta 14 (4) (1954) [85] J. Schlichtkrull, Insulin crystals .1. The minimum mole-fraction of metal in
574e575. insulin crystals prepared with Zn, Cd, Co, Ni, Cu, Mn, or
[53] A.A. Kossiakoff, Protein dynamics investigated by the neutron diffraction- Fe, Acta Chem. Scand. 10 (9) (1956) 1455e1458.
hydrogen exchange technique, Nature 296 (5859) (1982) 713e721. [86] A. Wlodawer, H. Savage, G. Dodson, Structure of insulin - results of joint
[54] S.W. Englander, N.R. Kallenbach, Hydrogen-exchange and structural dy- neutron and X-Ray renement, Acta Crystallogr. B Struct. Sci. 45 (1989)
namics of proteins and nucleic acids, Q. Rev. Biophys. 16 (4) (1983) 521e655. 99e107.
[55] S.W. Englander, et al., Hydrogen exchange: the modern legacy of Linder- [87] H. Matsumura, et al., Crystallization and preliminary neutron diffraction
strom-Lang, Protein Sci. 6 (5) (1997) 1101e1109. studies of HIV-1 protease cocrystallized with inhibitor KNI-272, Acta Crys-
[56] M.M. Blum, et al., Preliminary time-of-ight neutron diffraction study on tallogr. F. Struct. Biol. Cryst. Commun. 64 (Pt 11) (2008) 1003e1006.
diisopropyl uorophosphatase (DFPase) from Loligo vulgaris, Acta Crys- [88] J.R. Luft, J. Newman, E.H. Snell, Crystallization screening: the inuence of
tallogr. F. Struct. Biol. Cryst. Commun. 63 (Pt 1) (2007) 42e45. history on current practice, Acta Crystallogr. F. Struct. Biol. Cryst. Commun.
[57] M.M. Blum, et al., Rapid determination of hydrogen positions and proton- 70 (Pt 7) (2014) 835e853.
ation states of diisopropyl uorophosphatase by joint neutron and X-ray [89] J.R. Luft, et al., Efcient optimization of crystallization conditions by
diffraction renement, Proc. Natl. Acad. Sci. U. S. A. 106 (3) (2009) 713e718. manipulation of drop volume ratio and temperature, Protein Sci. 16 (4)
[58] A. Ostermann, et al., Hydrogen and deuterium in myoglobin as seen by a (2007) 715e722.
neutron structure determination at 1.5 angstrom resolution, Biophys. Chem. [90] M.P. Blakeley, et al., The 15-K neutron structure of saccharide-free conca-
95 (3) (2002) 183e193. navalin A, Proc. Natl. Acad. Sci. U. S. A. 101 (47) (2004) 16405e16410.
[59] B.C. Bennett, et al., On the determinants of amide backbone exchange in [91] J. Greer, H.W. Kaufman, A.J. Kalb, An X-ray crystallographic study of conca-
proteins: a neutron crystallographic comparative study, Acta Crystallogr. navalin A, J. Mol. Biol. 48 (2) (1970) 365e366.
Sect. D. Biol. Crystallogr. 64 (2008) 764e783. [92] A.J. Kalb, et al., The effect of metal-ion homogeneity on the diffraction limit of
[60] P. Langan, et al., L-Arabinose binding, isomerization, and epimerization by D- orthorhombic (I222) crystals of concanavalin-a, J. Cryst. Growth 88 (4)
xylose isomerase: X-ray/neutron crystallographic and molecular simulation (1988) 537e540.
study, Structure 22 (9) (2014) 1287e1300. [93] M. Maeda, et al., Crystallization of a large single crystal of cubic insulin for
[61] P.V. Afonine, et al., Joint X-ray and neutron renement with phenix.rene, neutron protein crystallography, J. Synchrotron Radiat. 11 (Pt 1) (2004)
Acta Crystallogr. Sect. D. Biol. Crystallogr. 66 (2010) 1153e1163. 41e44.
W.B. O'Dell et al. / Archives of Biochemistry and Biophysics 602 (2016) 48e60 59

[94] C.M. Casadei, et al., Neutron cryo-crystallography captures the protonation single crystal neutron diffractometer at J-PARC, Nucl. Instrum. Methods Phys.
state of ferryl heme in a peroxidase, Science 345 (6193) (2014) 193e197. Res. Sect. A Accel. Spectrom. Detect. Assoc. Equip. 600 (1) (2009) 195e197.
[95] C. Metcalfe, et al., The tuberculosis prodrug isoniazid bound to activating [126] A.J. Schultz, et al., Integration of neutron time-of-ight single-crystal Bragg
peroxidases, J. Biol. Chem. 283 (10) (2008) 6193e6200. peaks in reciprocal space, J. Appl. Crystallogr. 47 (2014) 915e921.
[96] T. Ishikawa, et al., An abnormal pKa value of internal histidine of the insulin [127] S. Arzt, et al., LSCALE - the new normalization, scaling and absorption
molecule revealed by neutron crystallographic analysis, Biochem. Biophys. correction program in the Daresbury Laue software suite, J. Appl. Crystallogr.
Res. Commun. 376 (1) (2008) 32e35. 32 (1999) 554e562.
[97] T. Ishikawa, et al., A neutron crystallographic analysis of a cubic porcine [128] M.D. Winn, et al., Overview of the CCP4 suite and current developments,
insulin at pD 6.6, Chem. Phys. 345 (2e3) (2008) 152e158. Acta Crystallogr. D Biol. Crystallogr. 67 (2011) 235e242.
[98] R.C. Hughes, et al., Inorganic pyrophosphatase crystals from Thermococcus [129] A. Wlodawer, W.A. Hendrickson, A procedure for joint renement of
thioreducens for X-ray and neutron diffraction, Acta Crystallogr. F. Struct. macromolecular structures with X-ray and neutron-diffraction data from
Biol. Cryst. Commun. 68 (Pt 12) (2012) 1482e1487. single-crystals, Acta Crystallogr. A 38 (Mar) (1982) 239e247.
[99] J.D. Ng, et al., Large-volume protein crystal growth for neutron macromo- [130] P.D. Adams, et al., Generalized X-ray and neutron crystallographic analysis:
lecular crystallography, Acta Crystallogr. F. Struct. Biol. Cryst. Commun. 71 (Pt more accurate and complete structures for biological macromolecules, Acta
4) (2015) 358e370. Crystallogr. D Biological Crystallogr. 65 (2009) 567e573.
[100] Y. Ohnishi, et al., Neutron and X-ray crystallographic analysis of Achromo- [131] S. Yamaguchi, et al., Low-barrier hydrogen bond in photoactive yellow
bacter protease I at pD 8.0: protonation states and hydration structure in the protein, Proc. Natl. Acad. Sci. U. S. A. 106 (2) (2009) 440e444.
free-form, Biochim. Biophys. Acta Proteins Proteomics 1834 (8) (2013) [132] S.Z. Fisher, et al., Neutron and X-ray structural studies of short hydrogen
1642e1647. bonds in photoactive yellow protein (PYP), Acta Crystallogr. D Biol. Crys-
[101] T. Yamada, et al., Neutron and X-ray crystallographic analysis of the human tallogr. 63 (2007) 1178e1184.
alpha-thrombin-bivalirudin complex at pD 5.0: protonation states and hy- [133] T. Yokoyama, et al., Hydrogen-bond network and pH sensitivity in trans-
dration structure of the enzyme-product complex, Biochim. Biophys. Acta thyretin: neutron crystal structure of human transthyretin, J. Struct. Biol. 177
1834 (8) (2013) 1532e1538. (2) (2012) 283e290.
[102] K.L. Weiss, et al., Preliminary neutron crystallographic analysis of selectively [134] R.F. Tilton Jr., J.C. Dewan, G.A. Petsko, Effects of temperature on protein
CH3-protonated deuterated rubredoxin from Pyrococcus furiosus, Acta structure and dynamics: X-ray crystallographic studies of the protein
Crystallogr. F. Struct. Biol. Cryst. Commun. 64 (Pt 6) (2008) 537e540. ribonuclease-A at nine different temperatures from 98 to 320 K, Biochem-
[103] T. Kinoshita, et al., Crystallization of porcine pancreatic elastase and a pre- istry 31 (9) (1992) 2469e2481.
liminary neutron diffraction experiment, Acta Crystallogr. F. Struct. Biol. [135] A. Deacon, et al., The structure of concanavalin A and its bound solvent
Cryst. Commun. 63 (Pt 4) (2007) 315e317. determined with small-molecule accuracy at 0.94 angstrom resolution,
[104] T. Yokoyama, et al., Protonation state and hydration of bisphosphonate J. Chem. Society Faraday Trans. 93 (24) (1997) 4305e4312.
bound to farnesyl pyrophosphate synthase, J. Med. Chem. 58 (18) (2015) [136] M. Weik, J.P. Colletier, Temperature-dependent macromolecular X-ray
7549e7556. crystallography, Acta Crystallogr. D Biol. Crystallogr. 66 (2010) 437e446.
[105] T. Yokoyama, et al., Crystallization and preliminary neutron diffraction [137] J.S. Fraser, et al., Accessing protein conformational ensembles using room-
experiment of human farnesyl pyrophosphate synthase complexed with temperature X-ray crystallography, Proc. Natl. Acad. Sci. U. S. A. 108 (39)
risedronate, Acta Crystallogr. F. Struct. Biol. Cryst. Commun. 70 (Pt 4) (2014) (2011) 16247e16252.
470e472. [138] Y. Fukuda, T. Inoue, High-temperature and high-resolution crystallography
[106] B.C. Bennett, et al., Preliminary neutron diffraction studies of Escherichia coli of thermostable copper nitrite reductase, Chem. Commun. (Camb) 51 (30)
dihydrofolate reductase bound to the anticancer drug methotrexate, Acta (2015) 6532e6535.
Crystallogr. D. Biol. Crystallogr. 61 (Pt 5) (2005) 574e579. [139] A.T. Brunger, P.D. Adams, L.M. Rice, New applications of simulated annealing
[107] M. Budayova-Spano, et al., A methodology and an instrument for the in crystallographic renement, Direct Methods Solving Macromol. Struct.
temperature-controlled optimization of crystal growth, Acta Crystallogr. D. 507 (1998) 143e157.
Biol. Crystallogr. 63 (Pt 3) (2007) 339e347. [140] P.D. Adams, et al., The Phenix software for automated determination of
[108] R. Knihtila, et al., Neutron crystal structure of RAS GTPase puts in question macromolecular structures, Methods 55 (1) (2011) 94e106.
the protonation state of the GTP gamma-phosphate, J. Biol. Chem. (2015) pii: [141] G.M. Sheldrick, Crystal structure renement with SHELXL, Acta Crystallogr. C
jbc.M115.679860. [Epub ahead of print]. Struct. Chem. 71 (2015) 3e8.
[109] J.R. Helliwell, Neutron Laue diffraction does it faster, Nat. Struct. Biol. 4 (11) [142] T. Gruene, et al., Renement of macromolecular structures against neutron
(1997) 874e876. data with SHELXL2013, J. Appl. Crystallogr. 47 (2014) 462e466.
[110] I. Tanaka, et al., A high-performance neutron diffractometer for biological [143] A.T. Brunger, Version 1.2 of the Crystallography and NMR system, Nat.
crystallography (BIX-3), J. Appl. Crystallogr. 35 (2002) 34e40. Protoc. 2 (11) (2007) 2728e2733.
[111] K. Kurihara, et al., A new neutron single-crystal diffractometer dedicated for [144] J.C. Chen, et al., Direct observation of hydrogen atom dynamics and in-
biological macromolecules (BIX-4), J. Synchrotron Radiat. 11 (Pt 1) (2004) teractions by ultrahigh resolution neutron protein crystallography, Proc.
68e71. Natl. Acad. Sci. U. S. A. 109 (38) (2012) 15301e15306.
[112] S.A.L. Kim, C.H. Lee, M.K. Moon, S.J. Cho, T.S. Yoon, Y. Noda, Y. Ishikawa, Bio- [145] L. Coates, et al., A neutron Laue diffraction study of endothiapepsin: impli-
diffractometers at HANARO, Neutron News 24 (2) (2013) 33e35. cations for the aspartic proteinase mechanism, Biochemistry 40 (44) (2001)
[113] L. Coates, et al., Cryogenic neutron protein crystallography: routine methods 13149e13157.
and potential benets, J. Appl. Crystallogr. 47 (4) (2014) 1431e1434. [146] P. Emsley, K. Cowtan, Coot: model-building tools for molecular graphics,
[114] F. Meilleur, et al., The IMAGINE instrument: rst neutron protein structure Acta Crystallogr. D. Biol. Crystallogr. 60 (Pt 12 Pt 1) (2004) 2126e2132.
and new capabilities for neutron macromolecular crystallography, Acta [147] P. Emsley, et al., Features and development of Coot, Acta Crystallogr. D. Biol.
Crystallogr. D. Biol. Crystallogr. 69 (Pt 10) (2013) 2157e2160. Crystallogr. 66 (Pt 4) (2010) 486e501.
[115] P. Langan, G. Greene, B.P. Schoenborn, Protein crystallography with spall- [148] P.V. Afonine, et al., Towards automated crystallographic structure rene-
ation neutrons: the user facility at Los Alamos Neutron Science Center, ment with phenix.rene, Acta Crystallogr. D Biol. Crystallogr. 68 (2012)
J. Appl. Crystallogr. 37 (2004) 24e31. 352e367.
[116] L. Coates, et al., The macromolecular neutron diffractometer MaNDi at the [149] N. Niimura, et al., Neutron Laue diffractometry with an imaging plate pro-
spallation neutron source, J. Appl. Crystallogr. 48 (2015) 1302e1306. vides an effective data collection regime for neutron protein crystallography,
[117] K. Kusaka, et al., Evaluation of performance for IBARAKI biological crystal Nat. Struct. Biol. 4 (11) (1997) 909e914.
diffractometer iBIX with new detectors, J. Synchrotron Radiat. 20 (Pt 6) [150] A.Y. Kovalevsky, et al., Identication of the elusive hydronium ion
(2013) 994e998. exchanging roles with a proton in an enzyme at lower pH values, Angew.
[118] NMX. Available from: Chem. Int. Ed. Engl. 50 (33) (2011) 7520e7523.
diffractometer-0. [151] I.K. Mcdonald, J.M. Thornton, Satisfying hydrogen-bonding potential in
[119] D.A.A. Myles, et al., Neutron protein crystallography at ultra-low (< 15 K) proteins, J. Mol. Biol. 238 (5) (1994) 777e793.
temperatures, J. Appl. Crystallogr. 45 (2012) 686e692. [152] H.R. Brodkin, et al., Evidence of the participation of remote residues in the
[120] C.N. Sundaresan, S. Dixit, P. Venugopalan, A supramolecular assembly catalytic activity of co-type Nitrile Hydratase from Pseudomonas putida,
dominated by NeH/S hydrogen bonds: structure of 2- Biochemistry 50 (22) (2011) 4923e4935.
thioureidobenzoxazole by single crystal X-ray diffraction, J. Mol. Struct. [153] H.R. Brodkin, et al., Prediction of distal residue participation in enzyme
693 (1e3) (2004) 205e209. catalysis, Protein Sci. 24 (5) (2015) 762e778.
[121] J.W. Campbell, et al., LAUEGEN version 6.0 and INTLDM, J. Appl. Crystallogr. [154] N.J. Ataie, et al., Zinc coordination geometry and ligand binding afnity: the
31 (1998) 496e502. structural and kinetic analysis of the second-shell serine 228 residue and the
[122] J.W. Pugrath, The ner things in X-ray diffraction data collection, Acta methionine 180 residue of the aminopeptidase from Vibrio proteolyticus,
Crystallogr. Sect. D Biol. Crystallogr. 55 (1999) 1718e1725. Biochemistry 47 (29) (2008) 7673e7683.
[123] P. Langan, G. Greene, Protein crystallography with spallation neutrons: col- [155] Q.Y. He, et al., Mutations at nonliganding residues Tyr-85 and Glu-83 in the
lecting and processing wavelength-resolved Laue protein data, J. Appl. N-lobe of human serum transferrin e Functional second shell effects, J. Biol.
Crystallogr. 37 (2004) 253e257. Chem. 273 (27) (1998) 17018e17024.
[124] Z. Otwinowski, W. Minor, Processing of X-ray diffraction data collected in [156] T. Dudev, et al., First-second shell interactions in metal binding sites in
oscillation mode, Macromol. Crystallogr. 276 (Pt A) (1997) 307e326. proteins: a PDB survey and DFT/CDM calculations, J. Am. Chem. Soc. 125 (10)
[125] T. Ohhara, et al., Development of data processing software for a new TOF (2003) 3168e3180.
60 W.B. O'Dell et al. / Archives of Biochemistry and Biophysics 602 (2016) 48e60

[157] Y. Pocker, Water in enzyme reactions: biophysical aspects of hydration- human transthyretin and the amyloidogenic Val-30-Met Variant to 1.7-a
dehydration processes, Cell. Mol. Life Sci. 57 (7) (2000) 1008e1017. resolution, J. Biol. Chem. 268 (4) (1993) 2416e2424.
[158] C.R. Sage, et al., An essential role for water in an enzyme reaction mecha- [166] A. Hornberg, et al., A comparative analysis of 23 structures of the amyloi-
nism: the crystal structure of the thymidylate synthase mutant E58Q, dogenic protein transthyretin, J. Mol. Biol. 302 (3) (2000) 649e669.
Biochemistry 35 (50) (1996) 16270e16281. [167] E.A. Golden, A. Vrielink, Looking for hydrogen atoms: neutron crystallog-
[159] A. Zaks, A.M. Klibanov, The effect of water on enzyme Action in Organic raphy provides novel insights into protein structure and function, Aust. J.
Media, J. Biol. Chem. 263 (17) (1988) 8017e8021. Chem. 67 (12) (2014) 1751e1762.
[160] K.L. Kavanagh, et al., The molecular mechanism of nitrogen-containing [168] J.L. Markley, W.M. Westler, Protonation-state dependence of hydrogen bond
bisphosphonates as antiosteoporosis drugs, Proc. Natl. Acad. Sci. U. S. A. strengths and exchange rates in a serine protease catalytic triad: Bovine
103 (20) (2006) 7829e7834. chymotrypsinogen A, Biochemistry 35 (34) (1996) 11092e11097.
[161] M.B. Martin, et al., Nitrogen-containing bisphosphonates as carbocation [169] B. Geierstanger, et al., Protonation behavior of histidine 24 and histidine 119
transition state analogs for isoprenoid biosynthesis, Biochem. Biophys. Res. in forming the pH 4 folding intermediate of apomyoglobin, Biochemistry 37
Commun. 263 (3) (1999) 754e758. (12) (1998) 4254e4265.
[162] D.J. Hoseld, et al., Structural basis for bisphosphonate-mediated inhibition [170] N. Shimba, et al., Quantitative identication of the protonation state of his-
of isoprenoid biosynthesis, J. Biol. Chem. 279 (10) (2004) 8526e8529. tidines in vitro and in vivo, Biochemistry 42 (30) (2003) 9227e9234.
[163] A.M. Hounslow, et al., Determination of the microscopic equilibrium disso- [171] I. Bertini, et al., Protein hydration and location of water molecules in
ciation constants for risedronate and its analogues reveals two distinct roles oxidized horse heart cytochrome c by H-1 NMR, J. Magnetic Reson. 147 (1)
for the nitrogen atom in nitrogen-containing bisphosphonate drugs, J. Med. (2000) 1e8.
Chem. 51 (14) (2008) 4170e4178. [172] I. Tanaka, et al., Fundamental studies for the proton polarization technique in
[164] M. Meloun, et al., Thermodynamic dissociation constants of risedronate us- neutron protein crystallography, J. Synchrotron Radiat. 20 (2013) 958e961.
ing spectrophotometric and potentiometric pH-titration, Central Eur. J. [173] J.K. Zhao, Dynamically polarized hydrogenous materials for neutron scat-
Chem. 10 (2) (2012) 338e353. tering, Phys. B Condens. Matter 356 (1e4) (2005) 168e173.
[165] J.A. Hamilton, et al., The X-ray crystal-structure renements of normal