H/D exchange mass spectrometry and 
statistical coupling analysis reveal a role for 
allostery in a ferredoxin-dependent 
bifurcating transhydrogenase catalytic cycle
Authors: Luke Berry, Saroj Poudel, Monika Tokmina-Lukaszewska, 
Daniel R. Colman, Diep M. N. Nguyen, Gerrit J. Schut, Michael 
W.W. Adams, John W. Peters, Eric S. Boyd, & Brian Bothner
NOTICE: this is the author’s version of a work that was accepted for publication in BBA - General 
Subjects. Changes resulting from the publishing process, such as peer review, editing, 
corrections, structural formatting, and other quality control mechanisms may not be reflected in 
this document. Changes may have been made to this work since it was submitted for publication. 
A definitive version was subsequently published in BBA - General Subjects, VOL# 1862, ISSUE# 
1, (January 2018). DOI# 10.1016/j.bbagen.2017.10.002
Berry, Luke, S Poudel, M Tokmina-Lukaszewska, D.R. Colman, D.M.N. Nguyen, G.J. Schut, 
M.W.W. Adams, J.W. Peters, Eric S. Boyd, and Brian Bothner. "H/D exchange mass 
spectrometry and statistical coupling analysis reveal a role for allostery in a ferredoxin-dependent 
bifurcating transhydrogenase catalytic cycle." BBA - Biochimica et Biophysica Acta 1862, no. 1 
(January 2018): 9-17. DOI: 10.1016/j.bbagen.2017.10.002.
Made available through Montana State University’s ScholarWorks 
scholarworks.montana.edu 
H/D exchange mass spectrometry and statistical coupling analysis reveal a
role for allostery in a ferredoxin-dependent bifurcating transhydrogenase
catalytic cycle
Luke Berrya, Saroj Poudela, Monika Tokmina-Lukaszewskaa, Daniel R. Colmana,
Diep M.N. Nguyenb, Gerrit J. Schutb, Michael W.W. Adamsb, John W. Petersc, Eric S. Boyda,
Brian Bothnera,⁎
a Montana State University, Bozeman, MT 59717, United States
b University of Georgia, Athens, GA 30602, United States
c Washington State University, Pullman, WA, United States
Recent investigations into ferredoxin-dependent transhydrogenases, a class of enzymes responsible for electron
transport, have highlighted the biological importance of flavin-based electron bifurcation (FBEB). FBEB gen-
erates biomolecules with very low reduction potential by coupling the oxidation of an electron donor with
intermediate potential to the reduction of high and low potential molecules. Bifurcating systems can generate
biomolecules with very low reduction potentials, such as reduced ferredoxin (Fd), from species such as NADPH.
Metabolic systems that use bifurcation are more efficient and confer a competitive advantage for the organisms
that harbor them. Structural models are now available for two NADH-dependent ferredoxin-NADP+ oxidor-
eductase (Nfn) complexes. These models, together with spectroscopic studies, have provided considerable in-
sight into the catalytic process of FBEB. However, much about the mechanism and regulation of these multi-
subunit proteins remains unclear. Using hydrogen/deuterium exchange mass spectrometry (HDX-MS) and sta-
tistical coupling analysis (SCA), we identified specific pathways of communication within the model FBEB
system, Nfn from Pyrococus furiosus, under conditions at each step of the catalytic cycle. HDX-MS revealed
evidence for allosteric coupling across protein subunits upon nucleotide and ferredoxin binding. SCA uncovered
a network of co-evolving residues that can provide connectivity across the complex. Together, the HDX-MS and
SCA data show that protein allostery occurs across the ensemble of iron‑sulfur cofactors and ligand binding sites
using specific pathways that connect domains allowing them to function as dynamically coordinated units.
1. Introduction
NADH-dependent ferredoxin-NADP+ oxidoreductase (Nfn) catalyzes the
formation of reduced ferredoxin (FdRed), from the less energetic donor, ni-
cotinamide adenine dinucleotide phosphate (NADP+/NADPH), by coupling
this reaction to the thermodynamically favorable reduction of nicotinamide
adenine dinucleotide (NAD+/NADH), through a process termed flavin-based
electron bifurcation (FBEB) [1,2]. FBEB modulates the energy metabolism
within a cell by coupling the oxidation of an electron donor with an
intermediate reduction potential to promote the reduction of two electron
acceptors, one with a high reduction potential and one with a low potential.
Doing so allows the formation of low potential, high-energy compounds, such
as hydrogen gas and reduced Fd. Due to the relatively recent discovery of
FBEB, the mechanism behind catalysis remains largely unresolved. However,
crystal structures of the bifurcating enzyme Nfn were successfully obtained
from Thermotoga maritima (Tm) [3] and Pyrococcus furiousus (Pf) [4] which
enhances the ability to study the FBEB mechanism using this enzyme.
Nfn is a heterodimeric protein comprising a small (Nfn-S) and large
Abbreviations: FBEB, flavin-based electron bifurcation; HDX-MS, H/D exchange mass spectrometry; #D, absolute number of deuterium incorporated; %D, percentage of deuterium
incorporated; SCA, Statistical Coupling Analysis; Nfn, NADH-dependent ferredoxin-NADP+ oxidoreductase I; Nfn-S, Nfn Small Subunit; Nfn-L, Nfn Large Subunit; FdOx, oxidized fer-
redoxin; FdRed, reduced ferredoxin; FAD, flavin adenine dinucleotide; NAD+/NADH, nicotinamide adenine dinucleotide; NADP+/NADPH, nicotinamide adenine dinucleotide phosphate;
Tm, Thermotoga maritima; Pf, Pyrococcus furiosus
MARK
Abstract
(Nfn-L) subunit (Fig. 1A) [4,5]. Each of these subunits have cofactors
involved in the bifurcation reaction, including two flavin adenine di-
nucleotides (FAD), one in each subunit, a proximal [4Fe-4S] cluster (p
[4Fe-4S]) in Nfn-L, a distal [4Fe-4S] cluster (d[4Fe-4S]) in Nfn-L, and
one [2Fe-2S] cluster in Nfn-S. These cofactors allow Nfn to catalyze the
formation of reduced Fd, from NADPH and NAD+ (Eq. (1), Fig. 1B). The
enzyme can also catalyze the reverse confurcating reaction to form
NADPH by oxidizing Fd and NADH (Fig. 1C); this is thought to be the
physiological role of Nfn.
+ + ⇌ + + +
+ + − +2 NADPH NAD Fd 2 NADP NADH Fd HOx Red2 (1)
Studies of Nfn purified from Pf and Tm have revealed that NADH/
NAD+ binds to Nfn-S, and NADPH/NADP+ binds to Nfn-L, and Fd is
predicted to bind to Nfn-L near the distal [4Fe-4S] cluster [4,6]. Recent
studies have revealed that the bifurcation reaction proceeds by the
formation of a highly unstable flavin anionic semiquinone (ASQ) that
forms through the oxidation of NADPH, and this ASQ has sufficient
energy to reduce Fd [4]. A key feature of bifurcating enzymes, speci-
fically Nfn from Pf, that remains unknown is how energy is conserved
during the exergonic reduction reaction, how this energy is then used to
drive the endergonic reaction, and whether gating of electrons might be
involved in either of these processes. Studying the effects of substrate
binding on the structure and dynamics of Nfn may provide critical new
information regarding the role of protein structure in modulating FBEB.
Fig. 1. Schematic representation of Nfn and the electron bifurcation reaction: (A). Surface rendered model of Nfn from P. furiosus (PDB ID: 5JCA) with the small subunit, Nfn-S, shown in
green, and the large subunit, Nfn-L, shown in blue. Cofactors are spheres and their location within the complex is shown. Nfn-S contains a FAD (S-FAD) and a [2Fe-2S] cluster while Nfn-L
contains a FAD (L-FAD) and two [4Fe-4S] clusters, one proximal (p[4Fe-4S]) and one distal (d[4Fe-4S]) to the FAD. (B). Schematic representation of the bifurcating reaction. During the
bifurcation reaction, two NADPH bind to the large subunit and transfer electrons to L-FAD. One electron is then sent to the high potential branch and another to the low potential branch.
These individual electrons then reduce NAD+ and Fdox to NADH and Fdred. (C). The reverse reaction of electron bifurcation is termed electron confurcation and involves the oxidation of
Fdred and NADH to reduce NADP+. The electrons follow the path indicated by black arrows (B and C). (For interpretation of the references to colour in this figure legend, the reader is
referred to the web version of this article.)
While previous studies have described the structure and catalytic
properties associated with Nfn [7,8], the finer mechanistic details of
FBEB in this enzyme largely remain unknown. In a recent study of Tm
Nfn using H/D exchange mass spectrometry (HDX-MS), it was shown
that binding of the nucleotide NADPH/NADP+ to Nfn-L resulted in a
subtle conformational change near the proposed Fd binding site [3],
which suggested the possibility of long distance communication be-
tween the nucleotide and Fd binding sites. This observation was made
possible through the application of the powerful technique, HDX-MS,
that tracks exchange of amide hydrogens in the protein backbone with
water via a base-catalyzed reaction [9]. By replacing water (H2O) with
deuterated water (D2O), the amide hydrogens that exchange will be
replaced with deuterium which can be tracked using mass spectro-
metry. The tertiary and quaternary structure of a protein, as well as the
stability of the hydrogen bonds within the secondary structural ele-
ments, influence the rate by which H/D exchange occurs [10,11]. This
allows HDX-MS to reveal substrate/ligand binding sites, points of
contact between subunits in a complex, and changes in dynamics even
in peptides that participate in allosteric communication within a single
protein, or an entire complex [12,13].
Conformational changes caused by ligand binding are increasingly
being recognized as an intrinsic feature involved in the functionality of
many proteins [14]. The ability of ligands to induce conformational
changes at a site remote from the site of binding has been termed al-
lostery [15]. However, the pathways of allosteric communication in a
protein complex, or even a single protein are rarely known. One method
that has proven useful for mapping allosteric pathways of connectivity
and can provide mechanistic insights on protein regulation is Statistical
Coupling Analysis (SCA). SCA is a method that identifies co-varying (co-
evolving) suites of amino acid residues that exhibit evidence for en-
ergetic connectivity, which in turn has been suggested to be evidence
for possible allosteric communication within the protein [15,16,17].
SCA quantifies how changes in the distribution of amino acids at one
position influence changes in the amino acid distribution at another
position. Energetic connectivity is then computed using a curated se-
quence alignment of homologs, within the context of a combined sta-
tistical and protein structural framework.
To investigate the potential role of allosteric communication in
FBEB by Pf Nfn, we used SCA to identify potential pathways of com-
munication within the enzyme complex and HDX-MS to study changes
in the stability and dynamics of the protein complex when different
substrates are present. Analyses were conducted on enzymes in the
presence and absence of nucleotides and FdOx. Our HDX-MS results
reveal that ligand binding sites display protection from exchange when
substrates are bound to a distant site, suggesting allosteric commu-
nication. SCA analyses identified 137 residues in Nfn that form an al-
losteric communication pathway, providing strong evidence supporting
the role of ligand induced allosteric communication during FBEB.
Additionally, changes in H/D suggests a role for protein dynamics and
communication pathways in tuning electron transfer by influencing the
protein microenvironment around the cofactors.
2. Materials and methods
2.1. Protein purification
The Pf Nfn complex (Nfn-SL) and Fd were purified from P. furiosus,
as previously described [4].
2.2. Native mass spectrometry
Non-covalent mass spectrometry experiments were conducted on a
SYNAPT G2-Si instrument (Waters) in a similar fashion as described
previously [18]. Briefly, the Nfn complex sample was buffer exchanged
to 200 mM ammonium acetate and pH 7 (Sigma) using 3 kDa molecular
weight cutoff spin filters (Pall Corporation). Nfn was then infused from
in-house prepared gold-coated borosilicate glass capillaries to the
electrospray source at a protein concentration of 5 μM and a rate of
roughly 90 nL/min. The instrument was tuned to enhance performance
in the high mass-to-charge range. Settings were as follows: Source
temperature 30 °C, capillary voltage of 1.7 kV, trap bias voltage of 16 V,
and argon flow in the collision cell (trap) of 7 mL/min. Transfer colli-
sion energy was held at 10 V while trap energy varied between 10 and
200 V. Data analysis was performed in the MassLynx software version
4.1 (Waters).
2.3. HDX-MS analysis
Nfn was subjected to HDX-MS analysis under eight conditions: 1) as-
purified, 2) NADH bound, 3) NAD+ bound, 4) NADP+ bound, 5)
NADPH bound, 6) NADH+ NADP+ bound, 7) NADPH+ NAD+
bound, and 8) FdOx bound. Pyridine nucleotides were incubated with Pf
Nfn at a final concentration of 1 mM. Nfn:FdOx complex was formed by
first making a 1:1 M ratio of Nfn to FdOx using 10 μL of Nfn (16.58 mg/
mL) followed by incubation under anoxic conditions for 1 h at room
temperature (~21 °C) before initiating the reaction. The exchange re-
action was initiated by a 10-fold dilution of Nfn with reaction buffer at
60 °C. This temperature was chosen due to temperature limitations of
the equipment available. Reaction buffer contained 100 mM Tris-HCl
and 150 mM NaCl in D2O (pD 7.0) with 1 mM of pyridine nucleotides
(NAD+, NADH, NADP+, NADPH, NADPH+ NAD+, or NADH
+ NADP+). Samples were removed and quenched to stop exchange
after 1 min., 3 min., 15 min., 60 min., 3 h, and 24 h. At each time point,
a 10 μL subsample was withdrawn from each reaction vial and placed
into quenching/digestion solution containing 1% formic acid (FA,
Sigma) and porcine pepsin (Sigma, 0.2 mg/mL final concentration) on
ice. After a two min. incubation, the reaction mixture was frozen in
liquid N2 and stored at −80 °C until it was subjected to liquid chro-
matography-mass spectrometry (LC-MS) analysis.
LC-MS analysis of Nfn peptide fragments was completed on a 1290
UPLC series chromatography stack (Agilent Technologies) coupled di-
rectly to a 6538 UHD Accurate-Mass Q-TOF LC/MS mass spectrometer
(Agilent Technologies). Before electrospray-time of flight (ESI-TOF)
analysis, peptides were separated on a reverse phase (RP) column
(Phenomenex Onyx Monolithic C18 column, 100 × 2 mm) at 1 °C using
a flow rate of 500 μL/min. Under the following conditions:
0.0–1.0 min., 5% B; 1.0–9.0 min., 5–45% B; 9.1–9.8 min., 95% B;
9.8–9.9 min., 5% B. Solvent A contained 0.1% FA in water
(ThermoFisher) while solvent B contained 0.1% FA in acetonitrile
(ACN, ThermoFisher). Data was acquired at 2 Hz over a scan range
50–1700 m/z in positive mode. Electrospray settings were: nebulizer at
3.7 bar, drying gas at 8.0 L/min., drying temperature at 350 °C, and
capillary voltage at 3.5 kV. Data dependent MS/MS, specifying a se-
lection window of 4 m/z, was used to generate peptide sequence tags.
Data processing was carried out in Agilent MassHunter Qualitative
Analysis version 6.0. Peptide identification was performed using the
peptide analysis worksheet (PAWs, ProteoMetrics LLC.). Deuterium
uptake was determined by monitoring shifts of the centroid peptide
isotopic distribution by using the program HDExaminer (Sierra
Analytics Inc.). Measured values were then used to generate uptake
curves to compare relative deuterium exchange in all conditions tested.
2.4. HDX-MS clustering
The percent deuterium (%D) incorporated at a given time point was
calculated by dividing the number of deuterium atoms incorporated
(#D) by the number of deuterium atoms incorporated after 24 h. The
percent difference between the deuterium uptake in the as-purified
condition and the percent deuterium incorporated during experimental
manipulation was calculated as published by Chalmers et al. [19].
Briefly, the %D of the as-purified condition (no substrate bound) was
then subtracted from the %D of the nucleotide bound and FdOx bound
conditions. The difference was then plotted for peptides with sufficient
deuterium incorporation at the early (60 s.), middle (900 s.), and late
(10,800 s.) time points. A hierarchical clustering script in R (R version
3.0.0) was used to determine the number of trends present within the
deuterium uptake profiles based on the slope of the treadline. For in-
stance, linear treadlines with a positive slope were clustered together,
whereas profiles with a polynomial treadline were clustered depending
on if the treadline continually increased, continually decreased, if the
parabola formed is positive or negative, or if there is no change
throughout the time course (Supp. Figs. 1–7). The clustered profiles
were then implemented in a network analysis (described below).
2.5. Generation of Nfn and related paralog database
A database of Nfn-S and Nfn-L homologs, compiled as part of our
previous study [5], was used here. Briefly, the database was constructed
by subjecting Nfn-S and Nfn-L from Pf (AAD36706 and AAD36707,
respectively) to separate BLASTp queries against the National Center
for Biotechnology Information (NCBI) complete genome database
(n = 4586 genomes as of March 2016) specifying thresholds of> 30%
amino acid identity and> 60% coverage. Only Nfn-S and Nfn-L se-
quences that were co-localized in the genome were retained for further
analysis. Compiled Nfn-S and Nfn-L homologs were aligned with Clustal
Omega specifying default settings and were then manually screened to
filter out non-Nfn sequences using residues previously suggested to be
essential to protein function, as outlined in Demmer et al. [6,20]. These
steps resulted in a curated database of 467 Nfn-SL homologous se-
quences.
2.6. Statistical coupling analysis (SCA)
Aligned homologous sequences of both Nfn-S and Nfn-L (n = 467)
were concatenated using a custom python script (python version 2.7).
The concatenated sequences were then curated to include only posi-
tions that contained a maximum gap frequency of 20%, resulting in a
total of 750 positions (out of 1955 positions including gaps). The cu-
rated Nfn-SL sequences along with the Pf Nfn structure (PDB: 5JFC)
were subjected to SCA analysis in MATLAB (MATLAB version 9.1) as
previously described in Halabi et al. 2009 [17]. Chimera (version 3.2.0)
was used to map and visualize the spatial location of positions that
exhibited evidence for co-variance (co-evolution) in the Pf Nfn structure
[21].
2.7. Network analysis
The matrix obtained by clustering the HDX-MS data was subjected
to Pearson correlation using the correlation package in Cytoscape
(version 3.2.0) [22]. The network was visualized with the force directed
organic layout [23]. Each unique peptide was denoted as a node and
the edges between nodes represent the degree of correlation between
HDX exchange between a given set of peptides (nodes). The network
was further analyzed using the Network Analyzer plugin in Cytoscape
to map the correlation information onto the network [24]. Network
nodes were colored based on results from SCA analysis.
3. Results
3.1. Evidence of long distance communication in Pf Nfn
To investigate the role of structural dynamics in the mechanism of
FBEB in Nfn, we applied HDX-MS to Pf Nfn poised at different steps in
the catalytic cycle. Individual steps were simulated by adding pyridine
nucleotides and FdOx to oxidized Nfn in seven different combinations.
Deuterium uptake over time was then compared in the as-purified Nfn
enzyme with that observed when provided with pyridine nucleotides or
FdOx. Before beginning experiments, the as-purified Nfn complex from
Pf was analyzed by native MS to confirm its composition. Native MS
showed that Pf Nfn contains two FAD molecules, two [4Fe-4S] clusters,
and one [2Fe-2S] cluster (Fig. 2). Based on native MS data collected at
low collision energy, cofactor occupancy is estimated to be 100%. This
analysis also provided assurance that pyridine nucleotides and FdOx did
not co-purify with the Nfn enzyme. A pepsin digestion of Nfn pro-
duced> 200 peptides that could each be uniquely mapped to the
amino acid sequence of Pf Nfn. From these 200 peptides, 97 and 125
were used as input to generate deuterium uptake curves for Nfn-S and
Nfn-L, respectively. Sequence coverage was 100% for Nfn-S and 94%
for Nfn-L for the peptides chosen (Supp. Fig. 8). These peptides were
used to map specific regions of deuterium incorporation during HDX-
MS.
Of the identified peptides, a subset from each subunit displayed
significantly different levels of deuterium incorporation dependent on
nucleotide binding. We reasoned that peptides displaying changes in
dynamic behavior upon substrate binding are important in the me-
chanism underpinning FBEB and catalytic activity. Also, since the
deuterium uptake in a number of peptides did not change under any of
the conditions tested, we are confident that the observed changes are
not an artifact of ligand binding [25]. Heat maps of the selected pep-
tides show a global view of exchange in Nfn-S (Supp. Fig. 9) and Nfn-L
(Supp. Fig. 10) as a function of incubation time. The heat maps show
that the as-purified state of Nfn had the most deuterium incorporation
while the NADPH+ NAD+ bound form had the lowest deuterium in-
corporation. Uptake curves for each of the peptides provide a local view
of the deuterium incorporation in each condition during the course of
the reaction (Supp. Figs. 11 and 12).
To investigate the possible role of protein dynamics in the me-
chanism of FBEB, we focused additional analyses on peptides in close
proximity to the cofactors and nucleotide binding sites. Six key peptides
(two in Nfn-S and four in Nfn-L) were selected for a detailed analysis.
All peptide residue numbers are relative to Nfn from Pf. The two pep-
tides from Nfn-S (residues 131–139 and 210–226) are involved in
NADH/NAD+ binding and S-FAD/[2Fe-2S] cluster coordination, re-
spectively [4,6]. The other four peptides are located in Nfn-L; residues
43–64 ([4Fe-4S] cluster binding/near putative Fd docking site), 70–85
(control region), 245–262 (L-FAD binding domain), and 370–392
(stabilizing loop upon NADPH/NADP+ binding).
To visualize changes in the exchange profile of Nfn upon nucleotide
and Fd binding, we created a histogram plot comparing HDX data for
the 3 h time point. Nfn-S residues 131–139, involved in NADH/NAD+
binding, showed a 30% decrease in deuterium incorporation when
Fig. 2. Native mass spectrum of Pf Nfn complex (as-purified) in the gas phase. The intact
complex (magenta diamonds represent charge envelope centered around 16+ charge)
contains two subunits (Nfn-S, blue and Nfn-L, green), two FAD, two [4Fe-4S] clusters and
one [2Fe-2S] cluster. The mass of the complex based on deconvoluted signal is 86,407 Da
(calculated MW= 86,425 Da). (For interpretation of the references to colour in this
figure legend, the reader is referred to the web version of this article.)
NADH or NAD+ are bound, when compared to the as-purified condition
(Fig. 3A). When the nucleotides NADP+ or NADPH bind to Nfn-L, a
decrease in deuterium incorporation of 10 and 20%, respectively, oc-
curs at residues 131–139 of Nfn-S. This suggests that the binding of
NADP+ or NADPH to Nfn-L induces a change in the structure or sta-
bility of Nfn-S. When Fd is bound to Nfn-L, residues 131–139 of Nfn-S
incorporate 10% less deuterium than the as-purified condition. This
indicates a change in the secondary structure stability in this region.
The remaining conditions tested, NADH + NADP+, and NADPH
+ NAD+, both exchange less deuterium than the as-purified condition,
with the NADH+ NADP+ condition exhibiting less exchange than the
NADPH+ NAD+ condition. Residues 210–226 in Nfn-S exhibited a
20% and< 10% decrease in deuterium incorporation when NADH
+ NADP+ are both bound or when the nucleotides are individually
bound to Nfn, respectively (Fig. 3B). Similarly, when NADPH+ NAD+
are both bound, there is a 30% decrease in exchange at residues
210–226 in Nfn-S, indicating that this condition has the greatest effect
on the structure or stability of this peptide.
Of the four peptides that were closely examined in Nfn-L, residues
70–85 exhibited small changes in deuterium incorporation in all con-
ditions (Fig. 3E) and thus can be considered as one of the control re-
gions for the HDX experiments. Other control regions include residues
128–141 of Nfn-S, and residues 302–323 of Nfn-L (Supp. Fig. 11E and
Supp. Fig. 12M). The remaining three peptides in Nfn-L (43–64,
245–262, and 370–392) displayed evidence of synergistic exchange
dynamics during nucleotide binding (Fig. 3C, D, and F). Nfn-L residues
245–262, which are in close proximity to the L-FAD, exhibit high
amounts of exchange in the as-purified, NADH bound, NADP+ bound,
NAD+ bound, and NADPH bound conditions. Among these conditions
the NADH and NADP+ conditions exchanged 15% and 10% less deu-
terium relative to the as-purified condition, respectively, whereas
NAD+ and NADPH bound exchanged 20% and 25% less deuterium,
respectively (Fig. 3C). The least amount of exchange was observed in
this peptide when FdOx, NADH + NADP+, or NADPH+ NAD+, are
bound. Relative to as-purified Nfn, a 45%, 30%, and 35% decrease in
deuterium exchange was observed in the NADPH+ NAD+, NADH
+ NADP+, and Fd bound states, respectively (Fig. 3C).
Residues 370–392 of Nfn-L are involved in a loop that becomes less
flexible, and therefore less accessible to exchange, upon NADPH/
NADP+ binding because of the interactions formed between the loop
and the NADPH/NADP+ binding pocket [4]. Importantly, FdOx and
NADPH+ NAD+ binding results in a decrease of 40% in the amount of
deuterium incorporated in this peptide. This indicates that residues
370–392 are the least accessible to exchange under these two condi-
tions. Additionally when NADP+, NADPH, and NADH+ NADP+ are
bound to Nfn, a decrease in deuterium incorporation (20%, 30%, and
30%, respectively) is observed at residues 370–392, albeit not as much
of a decrease when compared to FdOx, or NADPH+ NAD+ bound
conditions (40% decrease in both conditions). The remaining two
conditions, with NADH or NAD+ bound, display a 15% decrease in
exchange at residues 370–392 relative to the as-purified condition. This
is the opposite of the results for Nfn-S where a decrease of 10% and
Fig. 3. Histograms of HDX-MS at 3 h in key regions of the protein structure that are putatively involved in the bifurcation of electrons. Ribbon diagram depicts peptides of interest (A–F),
as mapped in red, in the Nfn structure from P. furiosus (PDB ID: 5JCA). Each histogram plot shows exchange for the indicated peptide after 3 h of incubation in deuterated buffer. Each bar
represents a different condition: Blue (as-purified Nfn), brown (Fd bound), orange (NADH bound), grey (NADP+ bound), gold (NADH+ NADP+ bound), green (NAD+ bound), red
(NADPH bound), and purple (NADPH + NAD+ bound). The peptides indicated are key regions that are thought to be involved in the electron bifurcation reaction. (A). Nfn-S (green
ribbon structure) residues 131–139 involved in NAD(H) binding. (B). Residues 210–226 involved in coordinating the S-FAD and the [2Fe-2S] of Nfn-S are shown. (C). Nfn-L (blue ribbon
structure) residues 245–262 are near the L-FAD. (D). Residues 370–392 are involved in NADP(H) binding. (E). Residues 70–85 act as a HDX-MS control region. (F). Residues 43–64 are
involved in coordinating the two [4Fe-4S] clusters of Nfn-L. Error bars indicate the standard deviation of three replicates. (For interpretation of the references to colour in this figure
legend, the reader is referred to the web version of this article.)
20% is observed in residues 131–139 when NADP+ or NADPH are
bound to Nfn-L.
The fourth peptide of interest in Nfn-L (residues 43–64) is re-
sponsible for coordinating both [4Fe-4S] clusters, and is believed to be
near the site of Fd binding to the large subunit [4]. Fd is expected to
bind near the distal FeS cluster on the large subunit, close to the 43–64
region [3,4]. It was observed that FdOx binding to Nfn caused increased
protection from exchange in residues 43–64 of Nfn-L. This observation
is consistent with previous studies of protein-protein and protein-ligand
interactions, in which a large decrease in exchange indicates a site of
possible interaction between the two subjects of interest [26,27].
3.2. SCA analysis of Nfn
Data generated from HDX-MS revealed evidence of long distance
communication within the complex. To elucidate specific details about
potential pathways of communication across Nfn, we utilized SCA. SCA
can reveal energetically linked, or co-evolving, amino acid residues. A
structure-based sequence alignment was constructed of 467 Nfn-like
proteins. From this, a total of 137 co-evolving positions were identified
of which 52 positions were in Nfn-S and 85 were in Nfn-L (Supp.
Table 1). Of the 52 co-evolving residues identified in Nfn-S, 18 residues
were within 10 Å of the S-FAD and another two were within 10 Å the
[2Fe-2S] cluster in Nfn-S. Of these 20 residues, six that were in close
proximity to S-FAD were highly conserved (i.e., > 80%) and one re-
sidue near the [2Fe-2S] cluster was also highly conserved (Supp.
Table 1). In Nfn-L, 16 of the 85 residues were within 10 Å of the L-FAD,
two residues were located within 10 Å of the proximal [4Fe-4S] cluster,
and six residues were within 10 Å of the distal [4Fe-4S] cluster. Of the
16 residues that were near the L-FAD, two were highly conserved
(Supp. Table 1).
Residues that were 100% conserved (a total of 50 residues) were not
included in the SCA due to the lack of variation among homologous
proteins used in the multiple sequence alignment. Twelve of the 100%
conserved residues were identified in Nfn-S and 38 were identified in
Nfn-L (Supp. Table 2). Of the 12 identified in Nfn-S, five are within 10 Å
of the [2Fe-2S] cluster and six are within 10 Å of the S-FAD. In Nfn-L,
five 100% conserved residues were located near the proximal [4Fe-4S]
cluster, five were located near the distal [4Fe-4S] cluster, and 21 were
located near the L-FAD.
To visualize and identify a putative communication pathway in Nfn,
the combination of residues that were 100% conserved in our sequence
dataset and those that exhibited evidence for co-evolution from SCA
were mapped onto the Pf Nfn structure. A total of 14 residues were
identified that were within 6 Å from one another or from the cofactors
involved in electron transfer (Fig. 4); all of the residues identified
were> 70% conserved. Of the 14 residues, six were identified in Nfn-S,
connecting the NAD(H) binding site to L-FAD. Three of the six residues
in Nfn-S are present in peptides identified via HDX-MS analysis: Glycine
at position 138 within the 131–139 region peptide that interacts with
NADH/NAD+, G112 in peptide 95–113 that interacts with the S-FAD,
and G226 that interacts with the [2Fe-2S] cluster. Likewise, eight of the
14 residues were identified in Nfn-L that connect L-FAD to the NADPH/
NADP+ and Fd binding sites. Two of these residues were detected by
HDX-MS as being involved in deuterium exchange: Proline at position
252 (P252) in peptide 242–262 that interacts with L-FAD and cysteine
at position 45 (C45) in peptide 43–64 that interacts with the distal [4Fe-
4S] cluster and Fd.
3.3. Network analysis of HDX-MS and SCA data
Protein function is an emergent property arising from the integra-
tion of functional units. We reasoned that the biophysical properties
and evolutionary fingerprint could be combined to elucidate underlying
modules that give rise to protein functionality. Peptides were clustered
based on the percent difference in deuterium uptake between the as-
purified and different pyridine nucleotide/FdOx bound conditions.
Thirty six of the mapped peptides from Nfn were used. These peptides
all had significant deuterium incorporation (#D > 1) and were prox-
imal to key regions of the protein. We also selected several peptides in
which little to no difference in exchange was observed between the
tested conditions. The resulting matrix was analyzed by Pearson cor-
relation and converted to a network map for visualization. This network
was then overlayed with the SCA data which identified evidence for co-
evolving residues within the designated peptides. Two networks were
generated by separating the bifurcating and confurcating reactions in
order to reveal pathways of connectivity (Fig. 5A and Supp. Fig. 13A).
The bifurcating and confurcating reactions were separated based on the
pyridine nucleotides needed to perform the reaction in a specified di-
rection (see Eq. (1)). For instance, the bifurcating reaction uses NAD+,
NADPH, and FdOx, whereas the confurcating reaction uses NADH,
NADP+, and FdRed. The HDX conditions involving one or both of the
nucleotides that are required to drive reaction directionality were
grouped together to create the network plots.
The bifurcation network revealed two distinct clusters of peptides
based on HDX (Fig. 5A). One cluster corresponds to peptides that ex-
hibit high levels of HDX exchange under conditions (NADPH, NAD+,
NADPH+ NAD+, and FdOx bound) that promote bifurcation, while a
second corresponds to peptides that exhibit high levels of correlation
under conditions (NADP+, NADH, and NADH+ NADP+ bound) that
promote confurcation (Supp. Fig. 13). In each of these conditions,
peptides (i.e., network nodes) that contained at least one residue that
exhibited evidence for co-evolution with other residues within the
protein are highlighted in red whereas network nodes that did not ex-
hibit evidence for co-evolution are highlighted in green. Peptides in-
volved in the bifurcation direction are separated into the red or purple
clusters and mapped onto the Pf Nfn structure (Fig. 5B and C, respec-
tively). Mapping of the peptides that comprise the red cluster on the Pf
Nfn structure reveals that they are heavily centered around the cofac-
tors of Nfn-S and Nfn-L. In contrast, mapping of the peptides that
comprise the purple cluster on the Pf Nfn structure reveals peptides that
are more spread across the enzyme complex, including residues near
the cofactors and the NADH/NAD+ binding site.
4. Discussion
The use of FBEB to improve metabolic efficiency in anaerobic or-
ganisms is of great interest because of the advantage provided in energy
poor environments and potential application for biotechnology [1,4].
The key step in FBEB is the transfer of electrons from a molecule of
medium reduction potential to one of lower potential (thermo-
dynamically unfavorable) by coupling this with a transfer to a high
potential acceptor (thermodynamically favorable). Our interest is in
understanding the mechanism behind the splitting of electron pairs and
how enzymes ensure that both electrons do not travel down the en-
ergetically favorable pathway. Current hypotheses include a “gating-
step” in which a change in the reduction potential or distance between
cofactors is involved. Two recent studies that used HDX to investigate
Nfn homologs identified evidence in support of a role for conforma-
tional change during FBEB [4,6]. Here we have significantly deepened
our understanding of this mechanism using native mass spectrometry,
HDX, in combination with SCA of Pf Nfn at each step of the catalytic
cycle.
The first step in our analysis was to establish the subunit and co-
factor stoichiometry of purified Nfn. Using native MS, it was de-
termined that Nfn purified from Pf contained the predicted cofactors
(two [4Fe-4S] clusters, one [2Fe-2S], and two FADs) for the holo-en-
zyme complex (Fig. 2). This is the first native MS of an oxygen sensitive
multisubunit protein complex and opens the door for investigating
other systems under strictly anaerobic conditions using this powerful
technique.
Many proteins utilize long distance communication between
binding sites and subunits to ensure efficient function [e.g., [14,15]].
HDX-MS of Pf Nfn in the presence of different substrates and products
revealed behavior consistent with allosteric regulation (Fig. 3A–F). The
goal of this experiment was to track the conformational dynamics of
Nfn as it progresses through the catalytic cycle to elucidate the func-
tional role of protein motion. We hypothesized that these effects are
involved in tuning the affinity of Nfn for substrate, or gating the elec-
trons by affecting the protein microenvironment.
The transfer of electrons within proteins is highly sensitive to dis-
tance, with very few examples in which there is> 14 Å between sites
[28]. Most of the cofactors within Nfn are within efficient electron
transfer distance of one another (< 14 Å), with the exception of the L-
FAD and the [2Fe-2S] cluster of Nfn-S, which are separated by a dis-
tance of 14.1 Å, as revealed by the crystal structure of Pf Nfn [4]. It was
previously hypothesized that Nfn undergoes a structural rearrangement
that brings the L-FAD and the [2Fe-2S] cluster of Nfn-S together during
catalysis [6,29]. HDX-MS of Pf Nfn shows that residues 210–226 of Nfn-
S are the most protected from deuterium exchange when NAD+ and
NADPH are simultaneously bound to the complex (Fig. 3B). This in-
crease in protection is consistent with a structural rearrangement that
Fig. 4. Co-evolving residues putatively involved in connecting the nucleotide and Fd binding sites through allostery. (A). Structure of Nfn (PDB ID: 5JCA) with residues represented by red
sticks that are potentially involved in allosteric communication. All the identified residues along the putative pathway were within 6 Å of each other. Here, the small subunit of Nfn (i.e
Nfn-S) is represented in green and the large subunit of Nfn (i.e. Nfn-L) is represented in cyan. Here, the first letter abbreviation is used to denote each residue while the number indicates
the position in the structure of P. furiosus Nfn (PDB ID: 5JCA). The number in parentheses represents the percent conservation of this residue in our Nfn sequence database comprising 467
Nfn-like homologous sequences. (B). Potential pathway of electron transfer between the nucleotide binding and Fd binding sites. Here, the red box indicates the nucleotide or Fd binding
site while the red dash lines indicates the electron transfer pathway. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this
article.)
Fig. 5. Network analysis of HDX-MS data and co-evolving residues. (A). Network plot generated from clustered HDX-MS data and co-evolving residues for the bifurcating conditions (NAD+,
NADPH, NADPH + NAD+, and FdOx). Each node represents a peptide analyzed in this study with HDX-MS. The node label indicates whether it is located on the large (Nfn-L) or small
(Nfn-S) subunit and includes the residue number of the first amino acid in the sequence (residue numbering is defined by the sequence position in P. furiosus Nfn (PDB ID: 5JCA)). Red
nodes indicate the presence of a residue that exhibits strong evidence for co-evolution, relative to other residues within the protein (see materials and methods). Edges connecting each
node indicate the correlation between the deuterium uptake profiles of each peptide. A high correlation is indicated by red edges and a low correlation is indicated by blue edges, as noted
by the scale in the upper right. Nodes were grouped together qualitatively based on the degree of correlation between peptides and can be separated into two groups, indicated by the red
and purple outlines. (B). The residues depicted by the red outline and which exhibit evidence for co-evolution were mapped in red on the structure of Nfn from P. furiosus. They are
heavily centered around the cofactors of Nfn-S and Nfn-L.(C). The same has been done for the residues with the purple outline. These peptides connect the cofactors of Nfn-S and Nfn-L
and the NADH/NAD+ binding site. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
brings the two cofactors closer together, improving the efficiency of
electron transfer while decreasing the accessibility of several amide
hydrogens to exchange. A similar effect is seen at residues 245–262 of
Nfn-L, which are near the L-FAD binding site (Fig. 3C). This additional
protection from deuterium exchange when NAD+ and NADPH are
present may act as a way of orienting the L-FAD closer to the [2Fe-2S]
cluster of Nfn-S. Demmer et al. performed HDX-MS experiments on Nfn
purified from Tm and hypothesized that a rigid body movement in the
small subunit brings the FAD of the large subunit and [2Fe-2S] cluster
of the small subunit closer together in Tm Nfn [6]. These two cofactors
are 15 Å apart in the crystal structure, however, upon binding of
NADPH, the distance becomes ~13 Å [6]. Our analysis suggests that
rather than a rigid body movement, a more sophisticated means of
communication likely occurs within Nfn during catalysis.
To further test our hypothesis and to supplement our understanding
of how long distance communication is accomplished in Nfn, SCA was
used to identify co-evolving amino acids for use in predicting a pathway
of communication (Fig. 4). We identified the presence of a pathway of
amino acids that connects cofactors and ligand binding sites within and
between subunits, providing a potential physical conduit for informa-
tion transfer through Nfn. To test whether the pathway of co-evolving
residues facilitate allostery, we combined our HDX data on protein
dynamics with our SCA data on co-selection (co-evolution) of residues).
The percent change in deuterium uptake for each condition was
used to cluster peptides based on HDX profiles generated using early,
middle, and late time points. (Supp. Figs. 1–7). By grouping peptides
with similar HDX profiles under bifurcating and confurcating reaction
conditions, networks were generated. These networks were then over-
laid with the identity of the co-evolving residues and the extent that
they are undergoing co-evolution. In the bifurcating conditions, pep-
tides with co-evolving residues formed a sub-network together and are
all highly correlated (Fig. 5A). When mapped onto the structure of Nfn,
many of these peptides are localized near the cofactors of both Nfn-S
and Nfn-L (Fig. 5B). The high correlation of the deuterium uptake
profiles and the presence of co-evolving residues pinpoints a scaffold for
communication around the cofactors involving these peptides. These
subtle structural rearrangements in Nfn involve conformational
switching at the Nfn-S and Nfn-L interface controlling the flow of
electrons. A second network of peptides with correlated deuterium
uptake and co-evolving residues is more distributed across Nfn
(Fig. 5C). These peptides are located near the NADH/NAD+, [2Fe-2S]
cluster, and S-FAD binding site of Nfn-S, and the L-FAD binding site, as
well as the interface between Nfn-L and Nfn-S. This pathway suggests
the presence of a signaling network throughout the complex that could
function as a sensor for pyridine nucleotides.
In this study we have made progress toward revealing the finer
mechanistic details of FBEB by using HDX-MS and SCA techniques to
reveal multiple pathways of communication within Nfn purified from P.
furiosus. Our analysis of Pf Nfn suggests that Nfn is a precisely tuned
catalyst that undergoes coordinated motions using networks of amino
acids. We hypothesize that the connectivity illustrated by these net-
works allows Nfn to integrate input from multiple substrate and product
concentrations making it an important factor in the gating of electrons.
It was previously hypothesized that electron gating in Nfn occurred
through a rigid body movement [6]. Our in-depth analysis on the
conformational dynamics of Nfn revealed physical networks that could
utilize allosteric communication to trigger the conformational gating of
electrons. Additionally the HDX-MS data collected on Nfn here suggests
that a network of peptides displaying similar exchange behavior, that
also contain co-evolving residues, are responsible for coordinating so-
phisticated movements within the complex upon nucleotide substrate
and Fd binding. This suggests that the physical connectivity of the
peptides plays a role in allostery, and helps the complex prepare to send
electrons in either a bifurcating or confurcating direction. The con-
nectivity of peptides also influences the gating of electrons, ensuring
that an electron transfers down both the high and low potential
pathways. Additionally, it could also play a role in the transfer of
protons, which occurs simultaneously with electron transfer during
bifurcation [4]. With the identification of key amino acids involved in
communication within Nfn, we can now focus our efforts on de-
termining the specific role of each amino acid residue in the FBEB
mechanism.
Conflict of interest
The authors declare that they have no conflicts of interest with the
contents of this article.
Transparency document
The http://dx.doi.org/10.1016/j.bbagen.2017.10.002 associated
with this article can be found, in online version.
Acknowledgements
This work is supported as part of the Biological and Electron
Transfer and Catalysis (BETCy) EFRC, an Energy Frontier Research
Center funded by the U.S. Department of Energy, Office of Science (DE-
SC0012518). We would like to thank Montana State University
Microfabrication Facility for help in preparation of gold-coated bor-
osilica capillaries for non-covalent mass spectrometry. The Mass
Spectrometry Facility at MSU is supported in part by the Murdock
Charitable Trust and an NIH IDEA program grant P20GM103474.
Appendix A. Supplementary data
Supplementary data to this article can be found online at https://
doi.org/10.1016/j.bbagen.2017.10.002.
References
[1] W. Buckel, R.K. Thauer, Energy conservation via electron bifurcating ferredoxin
reduction and proton/Na+ translocating ferredoxin oxidation, BBA-Bioenergetics
1827 (2013) 94–113.
[2] G. Herrmann, E. Jayamani, G. Mai, W. Buckel, Energy conservation via electron-
transferring flavoprotein in anaerobic bacteria, J. Bateriol. 190 (2008) 784–791.
[3] J.K. Demmer, H. Huang, S. Wang, U. Demmer, R.K. Thauer, U. Ermler, Insights into
Flavin-based FBEB via the NADH-dependent reduced ferredoxin: NADP oxidor-
eductase structure, J. Biol. Chem. 290 (2015) 21985–21995.
[4] C.E. Lubner, D.P. Jennings, D.W. Mulder, G.J. Schut, O.A. Zadvornyy, J.P. Hoben,
M. Tokmina-Lukaszewska, L. Berry, D.M. Nguyen, G.L. Lipscomb, B. Bothner,
A.K. Jones, A. Miller, P.W. King, M.W.W. Adams, J.W. Peters, Mechanistic insights
into enegy conservation by flavin-based FBEB, Nat. Chem. Biol. 13 (2017) 655–659.
[5] D.M.N. Nguyen, G.J. Schut, O.A. Zadvornyy, M. Tokmina-Lukaszewska, S. Poudel,
G.L. Lipscomb, L.A. Adams, J.T. Dinsmore, W.J. Nixon, E. Boyd, B. Bothner,
J.W. Peters, M.W.W. Adams, Two functionally distinct bifurcating NADP-dependent
ferredoxin oxidoreductases maintain the primary redox balance of Pyrococcus fur-
iosus, J. Biol. Chem. 292 (2017) 14603–14616.
[6] J.K. Demmer, F.A. Rupprecht, M.L. Eisinger, U. Ermler, J.D. Langer, Ligand binding
and conformational dynamics in a flavin-based electron-bifurcating enzyme com-
plex revealed by hydrogen-deuterium exchange mass spectrometry, FEBS Lett. 590
(2016) 4472–4479.
[7] J.K. Hurley, R. Morales, M. Martinez-Julvez, T.B. Brodie, M. Medina, C. Gomex-
Morena, G. Tollin, Structure-function relationships in Anabaena ferredoxin/
ferredoxin:NADP+ reductase electron transfer: insights from site-directed muta-
genesis, transient absorption spectroscopy and X-ray crystallography, BBA 1554
(2002) 5–21.
[8] N.P. Chowdhury, A.M. Mowafy, J.K. Demmer, V. Upadhyay, S. Koelzer,
E. Jayamani, J. Kahnt, M. Hornung, U. Demmer, U. Ermler, W. Buckel, Studies on
the mechanism of FBEB catalyzed by electron transferring Flavoprotein (Etf) and
Butyryl-CoA dehydrogenase (Bcd) of Acidaminococcus fermentans, J. Biol. Chem.
289 (2014) 5145–5157.
[9] V. Katta, B.T. Chait, Conformation changes in proteins probed by hydrogen-ex-
change electrospray-ionization mass spectrometry, Rapid Commun. Mass Spectrom.
5 (1991) 214–217.
[10] L. Konermann, J. Pan, Y. Liu, Hydrogen exchange mass spectrometry for studying
protein structure and dynamics, Chem. Soc. Rev. 40 (2011) 1224–1234.
[11] S.R. Marcsisin, J.R. Engen, Hydrogen exchange mass spectrometry: what is it and
what can it tell us? Anal. Bioanal. Chem. 397 (2010) 967–972.
[12] J. Rumi-Masante, F.I. Rusinga, T.E. Lester, T.B. Dunlap, T.D. Williams, A.K. Dunker,
D.D. Weis, T.P. Creamer, Structural basis for activation of Calcineurin by
Calmdoulin, J. Mol. Biol. 415 (2012) 307–317.
[13] J.R. Engen, T.E. Wales, S. Chen, E.M. Marzluff, K.M. Hassell, D.D. Weis,
T.E. Smithgall, Partial cooperative unfolding in proteins as observed by hydrogen
exchange mass spectrometry, Int. Rev. Phys. Chem. 32 (2013) 96–127.
[14] N.M. Goodey, S.J. Benkovic, Allosteric regulation and catalysis emerge via a
common route, Nat. Chem. Biol. 4 (2008) 474–482.
[15] J.F. Swain, L.M. Gierasch, The changing landscape of protein allostery, Cur. Op,
Struc. Biol, vol. 16, 2006, pp. 102–108.
[16] S.W. Lockless, R. Ranganathan, Evolutionary conserved pathways of energetic
connectivity in protein families, Science 286 (1999) 295–299.
[17] N. Halabi, O. Rivoire, S. Leibler, R. Ranganathan, Protein sectors: evolutionary units
of three-dimensional structure, Cell 138 (2009) 774–786.
[18] M. Luo, R. Jackson, S. Denny, M. Tokmina-Lukaszewska, K. Maksimchuk, W. Lin,
B. Bothner, B. Wiedenheft, C. Beisel, The CRISPR RNA-guided surveillance complex
in Escherichia coli accommodates extended RNA spacers, Nucleic Acids Res. 44 (15)
(2016) 7385–7394.
[19] M.J. Chalmers, B.D. Pascal, S. Willis, J. Zhang, S.J. Iturria, J.A. Dodge, P.R. Griffin,
Methods for the analysis of high precision differential hydrogen-deuterium ex-
change data, Int. J. Mass Spectrom. 302 (2011) 59–68.
[20] F. Sievers, A. Wilm, D. Dineen, T.J. Gibson, K. Karplus, W.Z. Li, R. Lopez,
H. McWilliam, M. Remmert, J. Soding, J.D. Thompson, D.G. Higgins, Fast, scalable
generation of high-quality protein multiple sequence alignments using Clustal
Omega, Mol. Syst. Biol. 7 (2011).
[21] E.F. Petterson, T.D. Goddard, C.C. Huang, G.S. Couch, D.M. Greenblatt, E.C. Meng,
T.E. Ferrin, UCSF Chimera—a visualization system for exploratory research and
analysis, J. Comput. Chem. 25 (2004) 1605–1612.
[22] P. Shannon, A. Markiel, O. Ozier, N.S. Baliga, J.T. Wang, D. Ramage, N. Amin,
B. Schwikowski, T. Ideker, Cytoscape: a software environment for integreated
models of biomolecular interaction networks, Genome Res. 13 (2003) 2498–2504.
[23] M.E. Smoot, K. Ono, J. Ruscheinski, P. Wang, T. Ideker, Cytoscape 2.8: new features
for data integration and network visualization, Bioinformatics 27 (2011) 431–432.
[24] Y. Assenov, F. Ramirez, S. SChelhorn, T. Lengauer, M. Albrecht, Computing topo-
logical parameters of biological networks, Bioinformatics 24 (2008) 282–284.
[25] J.K. Hilmer, A. Zlotnick, B. Bothner, Conformational equilibria and rates of loca-
lized motion within Hepatitis B virus capsids, J. Mol. Biol. 375 (2008) 581–594.
[26] S.S. Jaswal, Biological insights from hydrogen exchange mass spectrometry, BBA
Proteins Proteomics 1834 (2013) 118–1201.
[27] J.R. Engen, Analysis of protein conformation and dynamics by hydrogen/deuterium
exchange MS, Anal. Chem. 81 (2009) 7870–7875.
[28] C.C. Page, C.C. Moser, P.L. Dutton, Mechanism for electron transfer within and
between proteins, Curr. Op, Chem. Biol, vol. 7, 2003, pp. 551–556.
[29] J.W. Peters, A. Miller, A.K. Jones, P.W. King, M.W.W. Adams, FBEB, Curr. Op. in
Chem. Biol. 31 (146-152) (2016) 6–152.