Analyst
View Article Online
TUTORIAL REVIEW
Cite this: Analyst, 2017, 142, 2874
Published on 13 July 2017. Downloaded by University of Toronto on 05/11/2017 03:27:58.
Received 25th April 2017,
Accepted 7th July 2017
DOI: 10.1039/c7an00662d
rsc.li/analyst
a
Department of Chemistry, York University, Toronto, ON, M3J 1P3, Canada
b
Center for Research in Mass Spectrometry, York University, Toronto, ON, M3J 1P3,
Canada. E-mail: dkwilson@yorku.ca
Kerene A. Brown
2874 | Analyst, 2017, 142, 2874–2886
View Journal | View Issue
Bottom-up hydrogen deuterium exchange mass
spectrometry: data analysis and interpretation
a,b
Kerene A. Brown and Derek J. Wilson *a,b
Hydrogen Deuterium Exchange (HDX) Mass Spectrometry (MS) is a sensitive analytical technique that pro-
vides information on protein conformation and dynamics in solution. It is commonly used in the study of
protein–ligand and protein–protein interactions and more recently in the pharmaceutical industry for
epitope mapping, screening drug candidates and in the comparison of biopharmaceuticals to biosimilars.
HDX-MS monitors the exchange of protein backbone hydrogen atoms with deuterium in solution. Recent
advancements in HDX automation and data analysis, have taken the emphasis off developing a funda-
mental understanding of HDX, which is still lacking. This tutorial review will cover the different
mechanisms of exchange and how the exchange reaction is affected by various factors. We also explore
the basis of data analysis and the difficulties that often arise in the interpretation of site-specific and
segment-averaged HDX data, such as overlapping isotopic distributions and correct identification of
peptides. Initial data analysis generates a list of peptides and the deuterium incorporation of each peptide
at each labeling time point, i.e., a set of deuterium uptake profiles. Data interpretation and error analysis is
subsequently required to ensure that deuterium uptake profiles accurately reflect conformational
dynamics in solution. Finally, this review will also discuss the different ways in which HDX data can be
represented and how the data can be interpreted.
1. Introduction
Hydrogen Deuterium Exchange (HDX) is becoming an increas-
ingly commonly used technique in the biological sciences. It
allows for the study of solution phase protein conformations
Kerene Brown graduated from Derek Wilson completed his PhD
Carleton University (Ottawa, at Western University (Ontario,
ON) with a BSc in Biochemistry Canada) where he trained in
and Biotechnology and she is bioanalytical mass spectrometry.
currently a PhD student in This was followed by a post-doc
Dr Derek Wilson’s lab at York in 2005 at Cambridge
University (Toronto, ON). Her (Cambridgeshire, UK) in bioana-
research is focused on studying lytical NMR with a focus on
the dynamics of RNA chaperones protein folding and interactions.
and biosimilar characterization Shortly thereafter, he was hired
using Hydrogen Deuterium as an assistant professor at York
Exchange Mass Spectrometry University (Ontario, Canada),
and Ion Mobility. Derek J. Wilson taking up the position in 2007.
He is currently director of the
Center for Research in Mass Spectrometry (CRMS), manager of the
Mass Spectrometry Enabled Science and Engineering (MS-ESE)
training program, and lead investigator in the Technology-
enhanced Biopharmaceuticals Development and Manufacturing
(TBio-DM) initiative.
This journal is © The Royal Society of Chemistry 2017

Analyst
by monitoring the exchange of backbone amide hydrogens
with deuterium.1,2 The coupling of HDX to mass spectrometry
(MS) is beneficial as MS allows for the analysis of large and
complex samples and eliminates the need for the incorpor-
ation of NMR active nuclei (15N and 13C).3 More recently, the
development of time-resolved HDX-MS has allowed for the
structural analysis of intrinsically disordered proteins which is
difficult to accomplish using other structural techniques.4,5
The growth of HDX-MS can be attributed to its many
different applications. One of the more recent applications is
in the development of biopharmaceuticals, which comprises
Published on 13 July 2017. Downloaded by University of Toronto on 05/11/2017 03:27:58.
mostly monoclonal antibodies. HDX-MS has been used in the
development of monoclonal antibodies, biosimilar antibodies,
antibody drug conjugates and also in epitope mapping.3,6–8
Another growing area of application for HDX-MS is the study
of protein–membrane interactions. MS has the ability to differ-
entiate between proteins and lipids and the areas of the
protein that interacts with the membrane can be deter-
mined.3,9 More common areas of application include, protein–
ligand binding, homology-guided structural characterization
of proteins, protein folding and protein–protein interactions.
There are three approaches to local HDX-MS experiments:
top-down, middle-down and the most common approach:
bottom-up. Top-down analyses, first introduced by Anderegg
et al. 1994, involved deuteration of the intact protein followed
by gas phase fragmentation through collision induced dis-
sociation (CID).10 However, CID results in a high degree of
hydrogen/deuterium scrambling prior to fragmentation which
can significantly alter the results.11–13 ‘Non-ergodic’ fragmen-
tation techniques such as electron capture dissociation
(ECD)14,15 and electron transfer dissociation (ETD)16 result in
exchange
less or no hydrogen scrambling and are now widely used for
top-down HDX-MS. Bottom-up HDX involves deuteration of
the intact protein followed by acid quenching and acid pro-
tease digestion prior to ionization (usually by electrospray).
The deuterium level of each peptide is determined by monitor-
ing the change in the mass.17 There is no issue with hydrogen
scrambling as there is no gas phase fragmentation involved,
however, the spatial resolution is low as peptides are typically
4–10 amino acids long. On the other hand, unlike top-down
analysis there is no size barrier and large proteins and protein
complexes can be studied. To overcome the size barrier of
top-down HDX, a middle-down HDX workflow has been devel-
oped, which involves a combination of the top-down and
bottom-up approach, in which the intact protein is labelled,
quenched and partially digested followed by gas phase
ETD/ECD fragmentation of the resulting peptides.18
With the growth of HDX-MS there have been many develop-
ments in automated data analysis. However, in using HDX-MS
as a bioanalytical technique, it is important to understand the
physical basis upon which the method relies. The growth of
automated data analysis systems has greatly reduced the
perceived importance of understanding the fundamentals of
HDX. This tutorial review will cover the physical basis of the
HDX reaction, data analysis, data interpretation and represen-
tation of results.
This journal is © The Royal Society of Chemistry 2017
View Article Online
Tutorial Review
2. Factors affecting intrinsic
exchange
HX (hydrogen exchange) rates of backbone amide hydrogens
vary substantially depending on the presence of hydrogen
bonding, solvent properties ( pH, temperature), primary
sequence and most importantly protein structure and
dynamics.1,19–21 In order to quantitatively asses HX in back-
bone amides, the observed HX rate should be referenced to
the rate of the unprotected amide to determine its protection
factor PF = kint/kobs, where kobs is the experimentally deter-
belonging naturally
mined rate constant and kint is the intrinsic rate constant for a
given backbone amide hydrogen.21–24
2.1 pH
The intrinsic exchange rate of backbone amide hydrogens is
highly sensitive to pH. The mechanism by which the backbone
amide exchanges hydrogen with deuterium is primarily cata-
lyzed by acid (H3O+) or base (OH−) and in some cases there
could be a small contribution by water (H2O) catalysis. In base
catalysis, which dominates in the pH range above 2.5, the
amide group is deprotonated by the OH− ion producing an
amidate anion which in turn is reprotonated by H2O (or D2O)
to regenerate OH−. Acid catalysis can occur via N- or
O-protonation. In the case of N-protonation, the mechanism is
the reverse of base catalysis. The amide N is protonated by
H3O+ (or D3O+), after which H2O (or D2O) extracts a H+ from
the protonated intermediate.25–27 On the other hand,
O-protonation involves the protonation of the peptide bond
oxygen which is the most basic site of the peptide bond
leading to acidification of the NH proton. The NH proton is
then removed by water producing an imidic acid which is later
protonated by D3O+.28 Eqn (1) below, represents the pseudo
first order chemical exchange rate constant for unprotected
amides (kch):
kch ¼ kint;A ½Hþ  þ kint;B ½OH  þ kint;W ½H2 O ð1Þ
in terms of the intrinsic second order rate constants for the
acid catalyzed (A), base catalyzed (B) and water catalyzed (W)
reactions. The contribution by water catalysis is exceptionally
small in comparison to the base and acid catalysis and it is
often omitted from eqn (1).29
A plot of amino-acid-averaged log(kch) versus pH results in a
V-shaped curve (Fig. 1) in which the minimum pH is 2.5–3.0.
At the pH minimum, the rate of acid and base catalysis is
equal and the rate of exchange is at its lowest.20,30 Deuterium
labelling is often carried out under ‘native-like’ conditions at
∼pH 7, where the reaction is primarily base catalyzed (kch =
kint,OH[OH−]), followed by quenching by lowering the pH to
∼2.5, to provide sufficient time for Liquid Chromatography
(LC) separation and MS analysis.20
2.2 Temperature
For the base-catalyzed reaction, HX has an activation energy of
approximately 17 kcal mol−1, resulting in an increase in the
Analyst, 2017, 142, 2874–2886 | 2875

Tutorial Review
Published on 13 July 2017. Downloaded by University of Toronto on 05/11/2017 03:27:58.
Fig. 1 The dependence of intrinsic rate of exchange (kch) on the solu-
tion pH in which the rate of exchange reaches a minimum at pH 2.5–3.
Acid and base catalysis is dominant below and above the pH minimum
respectively.
As the temperature is increased, the HX rate is increased.
29,31
HX rate by a factor of 3 for each 10 °C increment. In order
to predict HX rate constants at other temperatures, each rate
constant in eqn (1) should be modified according to eqn (2).
The theoretical HX rate constant (kch) as a function of tempera-
ture can be determined by:


Ea 1 1
kch ðTÞ ¼ k293 exp   : ð2Þ
R T 293
In which k293 is the rate constant kint, H, kint, OH, or kint, H2O
at 293 K; Ea is the activation energy corresponding to 14, 17
and 19 kcal mol−1 for the acid, base and water-catalyzed
exchange respectively; and R is the gas constant 0.001986
kcal mol−1 K−1. During the quench step in a bottom up
HDX-MS experiment, along with decreasing the pH, the temp-
erature is also typically decreased to 0 °C to minimize back-
exchange.27
2.3 Primary sequence
Another factor affecting amide HX rates is the primary
sequence of the protein in which neighboring side chains can
produce an inductive or steric effect. Inductive effects are
caused by polar side chains, such as serine, which withdraw
electron density from neighboring peptide groups resulting in
an increase in their acidity. This increases the rate of deproto-
nation of the amide H by (OH− or OD−) in the base catalyzed
reaction whereas the overall rate of the acid-catalyzed reaction
decreases. Therefore, at pH > pHmin, an amide H alongside
serine residues will exchange 2.3 times faster than an amide H
alongside a non-polar residue (e.g. alanine).21,24 Bulky side
chains such as that of isoleucine decreases the amide HX rate
by blocking solvent accessibility. Depending on pH this can
decrease the rate of exchange by a factor of 10 to 20 if an
amide is flanked by 2 isoleucines.21 Both steric and inductive
effects are additive and depend on the side chains of the
nearest amino acid. In an unstructured peptide, the exchange
2876 | Analyst, 2017, 142, 2874–2886
View Article Online
Analyst
rate constant of an amide H can be calculated based on
eqn (3) in which Aleft, Aright, and Bleft, Bright are side chain acid
and base factors respectively.21
kch ¼ kHþ ðAleft  Aright Þ½Hþ  þ kOH ðBleft  Bright Þ½OH 
ð3Þ
þ kH2 O ðBleft  Bright Þ:
3. Exchange regimes
Although pH, temperature and primary sequence has a high
impact on the HX rate in random coil polypeptides, in a folded
protein the higher order structure has a greater impact on the
HX rate. In order for a labile H to undergo exchange with D2O,
it needs to be exposed to the solvent (i.e. not buried in the
core of the protein) and not be involved in any H bonding
network.1,32 The basic model for HX in folded proteins is
depicted as:
kop kch kcl
NHcl Ð NHop ! NDop Ð NDcl : ð4Þ
kcl kop
Due to the dynamic nature of proteins, buried or H-bonded
backbone amide hydrogens do not always remain buried or
H-bonded and can often come into an open state in which
exchange can occur. In the mechanism depicted above, kop
and kcl represents the rate constant of the opening and closing
motions respectively. When in the open state, HDX occurs
with the rate constant kch. As a result, the observed rate con-
stant of exchange in folded proteins is a product of the rate
constants of unfolding and refolding (i.e., ‘opening’ and
‘closing’) and the intrinsic rate constant.32–34 The observed
rate constant for exchange kHX, in a folded protein under
native-like conditions can then be expressed as:35
kop  kch
kHX ¼ : ð5Þ
kcl þ kch
3.1 EX1 kinetics
The conventional framework for analyzing HDX data is orga-
nized into two ‘regimes’ based on the relative magnitudes of
kch and kcl. In the EX1 regime, during the opening event, all
exposed labile hydrogens undergo exchange prior to the
exchangeable sites returning to their ‘closed’ state (i.e., kcl ≪
kch) and the observed HX rate constant is equivalent to the rate
constant of the opening reaction. Therefore eqn (5) is simpli-
fied to: since kcl is very small, it's ignored
kHX ¼ kop : ð6Þ
This type of exchange is not typical of proteins under
physiological conditions (though it does occur), and can be
induced through the addition of denaturants or a change in
pH (usually an increase; pH > 9.0). Although this is not com-
monly representative of physiological protein dynamics, it can
provide insights into protein folding.34
This journal is © The Royal Society of Chemistry 2017

Analyst
Published on 13 July 2017. Downloaded by University of Toronto on 05/11/2017 03:27:58.
Fig. 2 Cooperative unfolding representative of EX1 kinetics in selected
SH3 domains. ESI spectra of peptic peptides of (a) Lyn SH3 99–116 and
(b) α-spectrin 996–1014. Reproduced with permission from ref. 36.
Copyright 2006, Elsevier.
Fig. 3 Deuterium uptake of 12 intact SH3 domains from the centroid(s) of HDX isotopic distributions, in which the dotted line at the top represent
the maximum possible deuterium level. Bimodal distributions corresponding to EX1 exchange are observed in panels a, b and j in which multiple
centroids are identified. Traces with a sigmoidal shape exhibit distortions in their isotopic distributions that may correspond to EX1 exchange.
Reproduced with permission from ref. 36. Copyright, 2006 Elsevier.
This journal is © The Royal Society of Chemistry 2017
View Article Online
Tutorial Review
Although somewhat rare, cases of EX1 kinetics have been
observed in proteins under physiological conditions such
as the Hck, Lyn and α-spectrin SH3 domains which undergo
a slow unfolding event under physiological conditions.36,37
One important benefit of coupling HDX with MS is the fact
that EX1 kinetics can be easily identified in a mass
spectrum through its characteristic bimodal distribution
(Fig. 2) representing 2 populations. The low mass popu-
lation represent protein that has never undergone the EX1
conformational transition whereas the high mass popu-
lation has undergone this conformational transition at least
once.34
EX1 hydrogen deuterium exchange kinetics can be useful
in the determination of slow unfolding processes in proteins.
The effectiveness of this method was demonstrated by Engen
et al. in 1997 when they published an account of intact HcK
SH3 demonstrating EX1 kinetics.37 This work was later
continued by Wales and Engen in 2006 in which they studied
11 other SH3 domains using HcK SH3 as a reference.36 Their
results revealed that HcK, Lyn and α-spectrin SH3 domains
exhibited clear EX1 kinetics that were evident from the
characteristic bimodal distributions, in 3 out of the 12
domains (Fig. 3).36
Analyst, 2017, 142, 2874–2886 | 2877

Tutorial Review
3.2 EX2 kinetics
When the rate constant of the closing reaction is faster that
the intrinsic rate of exchange (kcl ≫ kch) EX2 kinetics are
observed, and eqn (5) becomes:32
kop
kHX ¼  kch : ð7Þ
kcl
EX2 HDX results in a gradual increase in the mass over
time in a single (binomial) distribution. As most proteins are
structurally stable at physiological pH, transient unfolding
Published on 13 July 2017. Downloaded by University of Toronto on 05/11/2017 03:27:58.
events in structured regions occur faster than kch, so EX2 is the
most common exchange regime of a protein under native con-
ditions.34,36 EX2 is also the more informative of the two domi-
nant exchange regimes, as it allows for the determination of the
‘open’/‘closed’ equilibrium, which, when measured specific
sites or segments, can be viewed as a local measurement of
local thermodynamic structural stability.38 It is important to be
able to distinguish between EX1 and EX2 because there are
some cases in which EX1 is observed under native conditions as
discussed previously, and this can confound automated data
analysis software. Experimentally, HDX-MS provides the advan-
tage of easily differentiating between EX1 and EX2, which is
more difficult to do with NMR experiments.37
3.3 EXX kinetics
EXX kinetics refer to an intermediate state where kcl is neither
faster nor slower than the kch (kcl ≈ kch). This phenomenon
has been explored by Xiao et al., in which they observed a
bimodal distribution similar to that in the EX1 regime
however each population drifted as observed in EX2 kinetics.39
EXX kinetics should not be confused with mixed EX1 and EX2
exchange, which is what is commonly observed in whole pro-
teins that undergo an EX1 conformational transition (e.g.,
some of the SH3 domains in Fig. 3).
4. Data analysis
HDX-MS data analysis involves the correct identification of
peptides (in the case of local HDX), back exchange correction,
determination of deuterium uptake and rates of exchange.
Many software programs (e.g. MS Studio,40 DynamX, HDX
Workbench,41 HX-Express v2,42 Hexicon 2,43 HDXFinder,44
HDX-Analyzer,45 ExMS,46 MassAnalyzer,47 and more) are avail-
able for automated data analysis but it remains crucial to
understand the process, as this can aid in the interpretation of
the results. In some cases, there may be a need to double-
check results obtained from automated software packages.
4.1 Global HDX
A typical workflow for global HDX involves the protein of inter-
est incubated in an excess of D2O for a period of time, after
which the reaction is quenched by decreasing the pH to 2.5
and the temperature to 0 °C in order to minimize the exchange
rate in what is (from quenching onwards) mainly protic
2878 | Analyst, 2017, 142, 2874–2886
View Article Online
Analyst
solvent.17 Separation is not typically required for global ana-
lysis so a trap column is often used in the LC step to desalt the
protein sample prior to MS analysis. The LC step also elimi-
nates deuterium that was incorporated into side chains (all
ultra-fast exchangers) through complete back-exchange. As a
result, by the time the sample is subjected to ESI, deuterium
will have equilibrated with solvent for all but the backbone
amides, which significantly simplifies data analysis and
interpretation.17,48
As there is no protease digestion or gas phase fragment-
ation involved in global HDX analysis, it provides an overall
picture of the conformational dynamics of a protein of inter-
est. To determine the amount of deuterium incorporated in
the protein, the mass of the non-deuterated protein is sub-
tracted from the mass of the deuterated protein at each time
point being studied. This can be done for individual charge
states of the protein or by using deconvoluted masses. The
data are then plotted as a change in mass or relative deuterium
level (y axis) vs. time (x axis).48 The information that is gained
from global HDX is limited and so local HDX is more common.
However, global HDX can be useful to determine if the protein
of interest is stable under HDX-MS conditions and serve as an
initial guide as to how dynamic the protein is. The relative
extent of disordered regions in a protein can be estimated from
the number of hydrogens that exchanged at early time points
(10 s or less). Likewise, the number of hydrogens involved in
more rigid or buried residues can be determined by subtracting
the maximum observed exchange (after several hours of label-
ling) from the total number of exchangeable backbone hydro-
gens.49 To ensure the accuracy of these estimates, an internal
calibrant (such as an unstructured peptide) should be used to
account for backbone amide back-exchange. An example of the
utility of global HDX is provided in the comparison of biosimi-
lar antibodies published by Pan and coworkers in 2016,7 in
which the deuterium uptake was calculated for the intact heavy
and light chains for each antibody being studied (Fig. 4). Their
results showed that the antibodies had the same solution
dynamics which was later verified by local HDX.7
4.2 Local HDX
The experimental workflow for local HDX is essentially
identical to global HDX with one additional step to digest the
deuterated protein sample. The term “local” refers to the type
of information one gets from the experiment, in which it is
possible to measure the conformational or dynamic changes
in localized regions of the protein, allowing for instance, for
the determination of ligand binding sites.7,50 There are two
main types of local HDX workflows: bottom up and top down.
Bottom-up local HDX is the more common of the two,
which involves digestion of the deuterated sample with an acid
resistant protease prior to ESI. This approach is in some ways
more straightforward than top-down HDX and does not require
specialized equipment, but provides peptide-level resolution
(typically 4–10 residues). The top down workflow involves gas
phase fragmentation of the deuterated protein. Although top-
down analysis can provide amino acid level resolution, there is
This journal is © The Royal Society of Chemistry 2017

Analyst
Published on 13 July 2017. Downloaded by University of Toronto on 05/11/2017 03:27:58.
Fig. 4 Global HDX of the intact light chain and heavy chain. (A) BEV vs.
BEV1; (B) BEV vs. BEV2. Each data point represents an average of two
experiments. Reproduced with permission from ref. 7. Copyright 2016,
Elsevier.
an analyte size barrier and specialized equipment is required to
prevent deuterium scrambling in the gas phase.7,50,51
In the bottom-up workflow, because the reaction is
quenched with acid, an acid resistant protease must be used.
Pepsin is the most commonly used protease in HDX experi-
ments,52 but others can be employed as well such as: aspergil-
lopepsin ( protease type XIII),53 rhizopuspepsin ( protease type
XVIII)53 and plasmepsin.54 Non-specific proteases often result
in multiple overlapping peptides and with careful analysis
information on single residue amide hydrogens can theoreti-
cally be obtained but this should be done with caution as this
could lead to over-interpretation of the results. A major dis-
advantage of protease digestion is incomplete sequence cover-
age or peptides that are too large in some cases where the
protein does not get digested very well. This can sometimes be
overcome with the use of a mix of different proteases or by
adding a denaturant in the quench buffer to improve the
digestion.53 After the protein is digested, the peptides are
separated using reversed-phase LC prior to MS analysis.
With the addition of lengthy digestion and LC separation
steps, back exchange may occur, but can usually be accounted
for in data analysis.17 To account for back exchange, a comple-
tely deuterated sample (M100%) is used as a control and the
total deuterium uptake level in each peptide, D, is calculated
at each time point using the equation:
Mt  MUND
D¼ N
M100%  MUND
where Mt and MUND is the mass of the deuterated peptide and
mass of the un-deuterated peptide respectively and N is the
maximum number of deuterium atoms that can be incorpo-
rated.17 There has been debate over the years as to whether a
back-exchange correction is useful as most HDX experiments
This journal is © The Royal Society of Chemistry 2017
View Article Online
Tutorial Review
measure relative rather than absolute deuterium uptake and
studies have shown that back exchange correction does not
affect the final results in relative measurement experiments
(i.e., where one state is being compared to another state of the
same protein).50,51 In addition, back exchange correction is
more of an estimate, as it relies heavily on preparing a fully
deuterated form of the sample of interest, which can be
difficult to prepare and maintain.51
The 1st step in data analysis is the determination of the
peptide sequences based on their m/z and LC retention time.
Accurate mass is often not sufficient to determine the peptide
sequences but it can be used to determine the possibilities.55
To ensure accuracy in peptide peak assignments, tandem MS
can be used to conduct classical MS/MS peptide sequencing.56
This step is crucial as miss-identifying the peptides can lead to
fundamental misinterpretation of the results. There are
2 main methods for peptide sequencing by LC MS/MS: Data
Dependent Acquisition (DDA) and Data Independent
Acquisition (DIA). DDA performs the MS/MS analysis on pre-
cursor m/z selected from the MS survey scans. There is a selec-
tion criteria and specific m/z values can be included or
excluded from the MS/MS analysis.57 Conversely, DIA acquires
the MS/MS scans using a wide isolation window with no
specific precursor ion selected.58
After the peptide list is generated, deuterium incorporation
can be quantified. There are 2 main ways in which this is
done: through the change in the centroid mass or through the
change in the isotopic distribution of the peptide. Through
the former method, the centroid mass of the non-deuterated
peptide (MUND) and deuterated peptide (Mt) is calculated and
the relative deuterium level (Dt) of each peptide is determined
by the equation:
Dt ¼ M t  M UND : ð9Þ
This method is not a direct measure of uptake but it can be
sensitive enough in detecting changes in conformation.
However, this method is not ideal for detecting subtle changes
in deuterated peptide isotopic distributions associated with
small levels of uptake and it may miss exchanges in the EX1 or
EXX regime.59,60
Another method used to quantify deuterium uptake levels
is to model and then fit the isotopic distribution of the deute-
rated peptide, which is often more accurate than using the cen-
troid. This is particularly true at low deuteration levels or where
there is overlap for some peaks in the distribution. In a non-
deuterated peptide, the isotopic distribution observed is mostly
as a result of 12C and 13C. However, as deuterium is incorpor-
ated the distribution becomes a convolution of (mainly) the
12 13
C/ C distribution and 1H/2H distribution.55 The isotopic dis-
ð8Þ
tribution after labelling can be modelled as a convolution of the
natural isotope distribution given by the equation:
D¼NB ð10Þ
where D is the isotope distribution after deuterium
labeling, N is the natural isotope distribution and B is the
Analyst, 2017, 142, 2874–2886 | 2879
 View Article Online

Tutorial Review Analyst

binomial distribution. This binomial distribution is rep-
resented by:
n!
Bðp; n; kÞ ¼ pk ð1  pÞnk :
k!ðn  kÞ!
In which n is the total number of exchangeable hydrogens
in the peptide of interest and k is the number of hydrogens
that have undergone exchange (0,1,…n). The segment averaged
level of deuteration (0–100%) is represented by p. The actual
Published on 13 July 2017. Downloaded by University of Toronto on 05/11/2017 03:27:58.
4.3 Automated data analysis
The introduction of automated data analysis has made
HDX-MS more widely used, as analyzing HDX-MS data can
become extremely time-consuming if it is done manually.
ð11Þ
Recent software developed by Zhang et al. 2012 can be used to
automatically derive protection factors for each peptide,
making data analysis and interpretation more straight-
forward.47 The most common approach in quantifying deuter-
ium levels is the use of centroid masses, which is the basis of
many automated HDX analysis software packages. As men-

tioned earlier, if only the centroid masses are used and isotopic
deuteration level of each peptide can then be determined by
distribution is ignored, a degree of accuracy could be lost and it
fitting a simulated distribution B, to the experimental data.61
is not possible to directly identify EX1 exchange.60 To correct
It is important to emphasize that the best-fit value of p rep-
this, some recently developed software packages automatically
resents the average uptake for the peptide and does not
identify bimodal distributions characteristic of EX1 kinetics
provide information on the distribution of uptakes at individ-
using alternative methods.46,65 Another downside with many
ual sites on the peptide. For instance, in an 8 amino acid
automated systems is the inability to analyze overlapping pep-
peptide, a highly skewed distribution (e.g., 4 sites exchanging
tides and peptides with low signal intensity. Recently developed,
at 100% and 4 sites at 0%) would produce the same value for p
is a semi-automated platform (HX-Express v2) which has tools
as a flat distribution (e.g., 8 sites exchanging at 50%), both of
for bimodal deconvolution and binomial fitting, and the ability
which are correct (i.e., in both cases, a perfect fit to the data
to analyze overlapping peptides.42 The use of automated
would yield p = 50% and the peptide is in fact 50%
systems is becoming more common as it significantly reduces
exchanged). There are several methods to fit theoretical distri-
the time spent on data analysis and increases the amount of
butions to the data. The most common and straightforward of
results being processed. With that being said, it is important to
these is the least-squares approach which minimizes the
understand the data analysis process as this will aid in the
square of the difference between the normalized peak intensi-
interpretation of the results obtained and often times manual
ties of the theoretical and observed distributions (Fig. 5).59,62
checks are required to ensure the data was analyzed correctly.
A major hurdle in the quantitation of deuterium levels, is
the presence of over lapping peptides. Although the LC step
after digestion separates the peptides, overlap still occurs often
in samples from large proteins. One way in which this 5. Data interpretation
problem could be overcome is using a 2nd dimension of separ-
ation, such as ion mobility, that separates species based on After the data is analyzed, a list of peptides with the amount of
deuterium incorporated into each peptide at each time point
their charge and conformation. This could essentially be used
to separate peptides with the same LC retention time.63 It is studied is generated. However, after this step, there is still more
also noted that the ‘distribution fitting’ method described work to be done. There are several post data analysis steps
which include but are not limited to: extracting kinetic infor-
above requires only three overlap-free peaks within a distri-
bution to accurately determine the uptake level. In reporting mation, determination of exchange amplitudes, determining
the relative deuterium level, it is also important to provide an the difference between binding interactions and allosteric
effects and finding the best way in which to represent the data.
error in these measurements. Errors are commonly estimated
through standard deviations which are usually obtained from
at least three technical replicates of the same labelling time 5.1 Exchange kinetics vs. exchange amplitudes
point. It is equally important to use biological replicates where HDX experiments are often carried out at multiple labelling
possible to test the reproducibility of the results.55,64 times, most commonly from 10 s to 5 h or more. As a result,

Fig. 5 Representation of how the theoretical distribution D (red dots) is fitted to the spectrum of a deuterated peptide to determine the percent
deuterium incorporation.

2880 | Analyst, 2017, 142, 2874–2886 This journal is © The Royal Society of Chemistry 2017
 View Article Online

Analyst Tutorial Review

the data can be presented as a graph of deuterium uptake vs.
HDX reaction time. Exchange amplitude refers to the
maximum amount of deuterium incorporated into a peptide
or whole protein at infinite time, represented by level at which
uptake reaches an apparent steady state in the kinetic data.
When the HDX profiles of a native protein are compared to a
ligand-bound protein a difference in the exchange amplitude
(usually a decrease) is often observed in peptides that are
within the binding region due to the formation of new strong
H-bonds.55 Changes in solvent accessibility may affect uptake
kinetics, but are unlikely to change the exchange amplitude
Published on 13 July 2017. Downloaded by University of Toronto on 05/11/2017 03:27:58.
ommended to group the backbone amides into fast, intermedi-
ate and slow exchangers and fit the data using the equation:
X
N
D¼N expðki tÞ: ð12Þ
i¼1
In which N is the number of exchangeable backbone
amides, ki is the exchange rate and t is the time allowed for
exchange. Depending on the complexity of the data collected,
the kinetic curves can be fit to between one and three exponen-
tials.17 Another approach is to use a ‘stretching’ term β, which

since that would require a subset of exchangeable sites to
become completely solvent inaccessible upon complexation.
If enough time points are obtained in a bottom-up HDX
experiment, the uptake rate can be determined for each
peptide in the protein sample. These rates can then be com-
pared between the native protein and protein bound state to
determine any changes in rate of the HDX reaction. In order to
determine rates, the absolute number of deuterium should be
known and therefore a back-exchange correction must be per-
formed. While each backbone amide will have a different rate
of exchange, the observed rate is effectively the average of
those rates, with the possibility of multi-phasic behavior due
to a multi-modal distribution of rates. It is therefore rec-
takes into account the multimodal nature of the distribution
of intrinsic rates.66
Here we will briefly discuss how kinetics can be used to
interpret HDX data using an ATPase packaging motor,
P4 hexamer as an example. HDX kinetics of the free hexamer
was compared to a PC associated P4. In any given peptide,
each amide hydrogen has different exchange rates which leads
to the kinetics becoming rather complex. Therefore, the kine-
tics obtained can be used to generate exchange rate distri-
butions shown in Fig. 6.67,68 A maximum entropy method
(MEM) algorithm developed by Zhang et al. 1997 can be used
to derive hydrogen exchange rate distributions from kinetic
curves obtained from HDX-MS experiments.67 With the rate

Fig. 6 HDX-MS kinetic curves of selected peptides of the P4 hexamer (red) and P4–PC complex (blue): residues 69–83 (A), 131–152 (B), 215–224
(C), 230–244 (D), 294–320 (E), 316–324 (F). Panels from left to right: Localization of the region within the structure of the P4 subunit, HDX kinetic
curves plotted as deuterium content with respect to labelling time, rate distribution, number of sites exchanging with slow (<0.1 h−1, blue), inter-
mediate (0.1–100 h−1, green) and fast (>100 h−1, red) rates.68 Reproduced with permission from ref. 68. Copyright 2006, Elsevier.

This journal is © The Royal Society of Chemistry 2017 Analyst, 2017, 142, 2874–2886 | 2881
 View Article Online

Tutorial Review Analyst

distributions obtained, the number of protected sites, sites
with intermediate exchange and sites with fast exchange rates
can be determined.68,69
Peptides on the surface of the isolated hexamer were
revealed to have fast or intermediate exchange rates (Fig. 6A, E
and F) in which 2 of these peptides (Fig. 6E and F) became pro-
tected upon binding to PC which was evident by the decrease in
the number of fast and intermediate exchange sites and an
increase in the number of slow exchange sites. However, the
peptide 69–83 revealed no evidence of protection in the P4–PC
complex as depicted in Fig. 6A. Peptides that are located within
Published on 13 July 2017. Downloaded by University of Toronto on 05/11/2017 03:27:58.
mational change. However, if ligand binding results in
decrease deuterium uptake in an allosteric region, it becomes
difficult to differentiate between the binding interactions and
allosteric conformational changes.70,71 If a crystal structure of
the protein being studied is available that shows the protein in
complex with its ligand, then HDX can be used to monitor
conformational changes upon ligand binding to determine the
allosteric hotspots.72 A recent publication by Chandramohan
et al. 2016, used HDX-MS to distinguish allosteric effects from
ligand binding in Hsp90. The ligand interactions are known
based on published high resolution structures of Hsp90 with

the core of the hexamer (Fig. 6B and C) showed significantly
slower kinetics than the other peptides and when bound to PC
there was an even greater decrease in the HDX kinetics. This is
likely as a result of global stabilization as it is not expected that
these buried residues would come into contact with PC. In the
isolated hexamer, the C-terminal peptides (Fig. 6E and F) have
almost no protected sites but upon binding to PC both peptides
gained a significant amount of protected sites.68
5.2 Binding vs. allosteric effects
A major limitation of HDX-MS is the inability to distinguish
between the binding interactions and allosteric conformational
changes. If ligand binding results in an increase in deuterium
uptake in a certain region of the protein, this change can
be often (but not always) be attributed to an allosteric confor-
2 high affinity ligands (PDB ID:4EGK and 1YET). HDX-MS
revealed significant decreases in deuterium uptake in the
ligand binding regions. In addition to testing high affinity
ligands, low affinity ligands (fragment 1 and 2) were tested
and found to have more significant allosteric responses to
ligand binding. The same 4 ligand binding sites were observed
when using the high or low affinity ligands but the high
affinity ligands had a much more significant decrease in deu-
terium uptake. The fragments are smaller and are expected to
have fewer interactions in the binding pocket.70
Both fragments had similar changes in the binding pocket
but had significantly different allosteric responses. This was
unexpected as both fragments have similar molecular weight
and affinities and they belong to the same phenolic class. It is
proposed that changes due to protection from HX as a result

Fig. 7 Conformational changes due to ligand binding and allosteric effects. (A) a difference plot showing the absolute difference in deuterium
uptake between the free and ligand bound state for each peptic peptide of Hsp90 at (t = 0.5, 2, 5, 10 min). Positive shifts represent decreases in deu-
terium uptake and negative shifts represents increases in deuterium uptake in comparison to the apo-Hsp90. Orthosteric sites are highlighted in the
blue boxes while allosteric sites are highlighted in the red boxes. (B, C) Orthosteric and allosteric regions are mapped onto the 3D structure of
Hsp90 (PDB ID:4EGK). Reproduced with permission from ref. 70. Copyright 2016.

2882 | Analyst, 2017, 142, 2874–2886 This journal is © The Royal Society of Chemistry 2017

Analyst
of H-bonding would show up in earlier reaction times whereas
changes due to allosteric conformational changes would
appear at longer times. Additional experiments would be
Published on 13 July 2017. Downloaded by University of Toronto on 05/11/2017 03:27:58.
Fig. 8 Difference in deuterium uptake mapped to CjDHDPS crystal
structures. Blue represents the regions with lower deuteration levels
after (a) Changes in deuterium uptake upon Lys binding. (b) Same as
panel a, viewed from the central cavity for subunit C in space-filling
representation. (c and d) Corresponding deuteration changes after
bisLys binding. Reproduced with permission from ref. 73. Copyright
2016, ACS publications.
Fig. 9 HDX exchange on Cdc42. (A) Deuterium level of Cdc42 measured at 6 time points (10 s to 3000 s) shown per the color scheme. The HDX
results at (B) 10 s and (C) 3000 s are mapped onto the crystal structure of Cdc42 (PDB ID: 2QRZ). Reproduced with permission from ref. 75.
Copyright 2016, Elsevier.
This journal is © The Royal Society of Chemistry 2017
View Article Online
Tutorial Review
needed in order to thoroughly test this hypothesis.
Nonetheless, when used together, HDX-MS and X-ray crystallo-
graphy can be used to distinguish between binding inter-
actions and long range allosteric changes.
Another important application of HDX-MS is in the study of
allosteric regulation in enzymes. This has been recently impli-
cated by Sowole et al., 2016 using Dihydrodipicolinate Synthase.
X-ray crystallography reveals only very subtle changes in the struc-
ture of inhibitor free and the bis Lys-inhibited enzyme which
does not explain the allosteric inhibition of the enzyme. HDX-MS
revealed increased dynamics in the inhibitor free enzyme which
allows binding of the substrate. On the other hand, there is
decreased dynamics in the bis Lys-inhibited state which prevents
the substrate from binding. This study shows how valuable
HDX-MS can be in the study of allosteric regulation.73
5.3 Data representation
There are many ways in which HDX-MS data can be presented
and it sometimes challenging to choose a method that best
represents the data. After the initial data analysis, the result is
a list of peptides with the amount of deuterium incorporated
at each reaction time studied. Arguably the most common way
of representing HDX data is by using a butterfly plot.74 This
type of plot allows for easy comparison, as multiple time
points can be plotted on the same graph which allows for the
observations of trends that may develop over the HDX reaction
times being studied. Different protein states are also plotted
together for easy comparisons. An example of this can be seen
in Fig. 7A. To compare the uptake rates of individual peptides,
deuterium uptake curves are the most direct method (Fig. 3, 4
Analyst, 2017, 142, 2874–2886 | 2883

Tutorial Review
and 6). If the data are fitted to eqn (12), the HX rate can be
determined for each peptide. This is an easy way to compare
peptides within the same protein or the same peptide in
different protein states (e.g. free vs. bound protein).
When two or more states of a protein are studied, the differ-
ence in uptake can be mapped onto the ground-state 3D struc-
ture if one is available from X-ray crystallography or NMR data.
In the simplest case, decreases in deuterium uptake are often
represented on the structure with blue while increases are
represented as red (Fig. 8). This provides an easy way to diplay
HDX-MS results in the context of localized changes in
Published on 13 July 2017. Downloaded by University of Toronto on 05/11/2017 03:27:58.
dynamics. For a more nuanced view of the dynamics, a color
scheme that typically ranges from blue to red with the lowest
deuterium uptake represented as blue and the highest uptake
represented as red can be used. An example is shown in Fig. 9,
in which HDX experiments were carried out on Cdc42 at
6 HDX reaction time points from 10 s to 3000 s. The results
of 1st and last time points were mapped onto the crystal struc-
ture and a very high deuterium uptake can be observed at the
3000 s vs. 10 s that show very little uptake in comparison.75
6. Conclusions and outlook
HDX-MS is a fast-growing technique that continues to evolve
with new applications but the core process remains the same.
This review has outlined some of the basic concepts of
HDX-MS and the process of data analysis and interpretation.
With understanding the basic concepts of HDX it becomes
easier to interpret the data correctly. HDX can be used to deter-
mine exchange kinetics of peptides and intact proteins and
the effect the ligand binding has on the kinetics.34,36,68
Although most applications of HDX-MS involve the determi-
nation of binding sites, it can be used to determine allosteric
changes under varying conditions although it remains difficult
to distinguish binding sites from allosteric sites if the binding
sites are not previously known.70,73 There may be a time
dependent observable change that is affected differently by
ligand binding and allosteric effects.70 This would be an inter-
esting development if HDX-MS could be used to differentiate
allostery from binding without the use of other techniques but
this is an area that needs to be explored further. The use of
computational modelling, is another fast growing field and
when used along with HDX-MS, it could potentially aid in the
interpretation of results to distinguish binding and allosteric
responses.76 HDX-MS has come a long way but continues to
improve rapidly, so a firm understanding of the process is
necessary for anyone who wishes to develop (rather than
simply use) this technique.
References
1 C. Woodward, I. Simon and E. Tüchsen, Hydrogen
exchange and the dynamic structure of proteins, Mol. Cell.
Biochem., 1982, 48, 135–160.
2884 | Analyst, 2017, 142, 2874–2886
View Article Online
Analyst
2 J. R. Engen, Analysis of Protein Conformation and
Dynamics by Hydrogen/Deuterium Exchange MS, Anal.
Chem., 2009, 81, 7870–7875.
3 G. F. Pirrone, R. E. Iacob and J. R. Engen, Applications of
Hydrogen/Deuterium Exchange MS from 2012 to 2014,
Anal. Chem., 2015, 87, 99–118.
4 C. Lento, G. F. Audette and D. J. Wilson, Time-resolved
electrospray mass spectrometry — a brief history,
Can. J. Chem., 2014, 93, 7–12.
5 S. Zhu, et al., Hyperphosphorylation of intrinsically dis-
ordered tau protein induces an amyloidogenic shift in its
conformational ensemble, PLoS One, 2015, 10, e0120416.
6 B. Deng, C. Lento and D. J. Wilson, Hydrogen deuterium
exchange mass spectrometry in biopharmaceutical discov-
ery and development – A review, Anal. Chim. Acta, 2016,
940, 8–20.
7 J. Pan, S. Zhang and C. H. Borchers, Comparative higher-
order structure analysis of antibody biosimilars using com-
bined bottom-up and top-down hydrogen-deuterium
exchange mass spectrometry, Biochim. Biophys. Acta,
Proteins Proteomics, 2016, 1864, 1801–1808.
8 M. Prądzińska, et al., Application of amide hydrogen/deu-
terium exchange mass spectrometry for epitope mapping
in human cystatin C, Amino Acids, 2016, 48, 2809–2820.
9 O. Vadas, M. L. Jenkins, G. L. Dornan and J. E. Burke,
Chapter Seven - Using Hydrogen–Deuterium Exchange
Mass Spectrometry to Examine Protein–Membrane
Interactions, in Methods in Enzymology, ed. M. H. Gelb,
Academic Press, 2017, vol. 583, pp. 143–172.
10 R. J. Anderegg, D. S. Wagner and C. L. Stevenson, The mass
spectrometry of helical unfolding in peptides, J. Am. Soc.
Mass Spectrom., 1994, 5, 425–433.
11 A. G. Harrison and T. Yalcin, Proton mobility in protonated
amino acids and peptides, Int. J. Mass Spectrom. Ion
Processes, 1997, 165, 339–347.
12 R. S. Johnson, D. Krylov and K. A. Walsh, Proton mobility
within electrosprayed peptide ions, J. Mass Spectrom., 1995,
30, 386–387.
13 P. T. M. Kenny, K. Nomoto and R. Orlando, Fragmentation
studies of peptides: The formation of Y ions, Rapid
Commun. Mass Spectrom., 1992, 6, 95–97.
14 K. Breuker, H. Oh, C. Lin, B. K. Carpenter and
F. W. McLafferty, Nonergodic and conformational control
of the electron capture dissociation of protein cations, Proc.
Natl. Acad. Sci. U. S. A., 2004, 101, 14011–14016.
15 R. A. Zubarev, N. L. Kelleher and F. W. McLafferty, Electron
Capture Dissociation of Multiply Charged Protein Cations.
A Nonergodic Process, J. Am. Chem. Soc., 1998, 120,
3265–3266.
16 J. E. P. Syka, J. J. Coon, M. J. Schroeder, J. Shabanowitz and
D. F. Hunt, Peptide and protein sequence analysis by elec-
tron transfer dissociation mass spectrometry, Proc. Natl.
Acad. Sci. U. S. A., 2004, 101, 9528–9533.
17 Z. Zhang and D. L. Smith, Determination of amide hydro-
gen exchange by mass spectrometry: a new tool for protein
structure elucidation, Protein Sci., 1993, 2, 522–531.
This journal is © The Royal Society of Chemistry 2017

Analyst
18 J. Pan, S. Zhang, A. H. Chou and C. Borchers, Higher-order
structural interrogation of antibodies using middle-down
hydrogen/deuterium exchange mass spectrometry,
Chem. Sci., 2016, 7, 1480–1486.
19 P. A. McBride-Warren and D. D. Mueller, Hydrogen-
Deuterium Exchange of Lysozyme. I. Rate Constants and
pH Dependence, Biochemistry, 1972, 11, 1785–1792.
20 J. B. Matthew and F. M. Richards, The pH Dependence of
Hydrogen Exchange in Protein, J. Biol. Chem., 1983, 258,
3039–3044.
21 Y. Bai, J. S. Milne, L. Mayne and S. W. Englander, Primary
Published on 13 July 2017. Downloaded by University of Toronto on 05/11/2017 03:27:58.
structure effects on peptide group hydrogen exchange,
Proteins: Struct., Funct. Bioinf., 1993, 17, 75–86.
22 J. J. Englander, D. B. Calhoun and S. W. Englander,
Measurement and calibration of peptide group hydrogen-
deuterium exchange by ultraviolet spectrophotometry,
Anal. Biochem., 1979, 92, 517–524.
23 G. P. Connelly, Y. Bai, M.-F. Jeng and S. W. Englander,
Isotope effects in peptide group hydrogen exchange,
Proteins: Struct., Funct. Bioinf., 1993, 17, 87–92.
24 R. S. Molday, S. W. Englander and R. G. Kallen, Primary
structure effects on peptide group hydrogen exchange,
Biochemistry, 1972, 11, 150–158.
25 B. H. Leichtling and I. M. Klotz, Catalysis of Hydrogen-
Deuterium Exchange in Polypeptides, Biochemistry, 1966, 5,
4026–4037.
26 A. Berger, A. Loewenstein and S. Meiboom, Nuclear
Magnetic Resonance Study of the Protolysis and Ionization
of N-Methylacetamide1, J. Am. Chem. Soc., 1959, 81,
62–67.
27 I. M. Klotz and B. H. Frank, Deuterium-Hydrogen
Exchange in Amide N—H Groups1, J. Am. Chem. Soc., 1965,
87, 2721–2728.
28 C. L. Perrin, Proton exchange in amides: Surprises from
simple systems, Acc. Chem. Res., 1989, 22, 268–275.
29 S. W. Englander, Hydrogen Exchange and Mass
Spectrometry: A Historical Perspective, J. Am. Soc. Mass
Spectrom., 2006, 17, 1481–1489.
30 S. O. Nielsen, W. P. Bryan and K. Mikkelsen, Hydrogen-
deuterium exchange of small peptides in aqueous solution,
Biochim. Biophys. Acta, 1960, 42, 550–552.
31 S. W. Englander, N. W. Downer and H. Teitelbaum,
Hydrogen Exchange, Annu. Rev. Biochem., 1972, 41, 903–
924.
32 D. M. Ferraro, N. D. Lazo and A. D. Robertson, EX1
Hydrogen Exchange and Protein Folding, Biochemistry,
2004, 43, 587–594.
33 J. Witten, et al., Mapping Protein Conformational
Landscapes under Strongly Native Conditions with
Hydrogen Exchange Mass Spectrometry, J. Phys. Chem. B,
2015, 119, 10016–10024.
34 D. D. Weis, T. E. Wales, J. R. Engen, M. Hotchko and
L. F. Ten Eyck, Identification and Characterization of
EX1 Kinetics in H/D Exchange Mass Spectrometry by Peak
Width Analysis, J. Am. Soc. Mass Spectrom., 2006, 17, 1498–
1509.
This journal is © The Royal Society of Chemistry 2017
View Article Online
Tutorial Review
35 S. W. Englander and N. R. Kallenbach, Hydrogen exchange
and structural dynamics of proteins and nucleic acids,
Q. Rev. Biophys., 1983, 16, 521–655.
36 T. E. Wales and J. R. Engen, Partial Unfolding of Diverse
SH3 Domains on a Wide Timescale, J. Mol. Biol., 2006, 357,
1592–1604.
37 J. R. Engen, T. E. Smithgall, W. H. Gmeiner and
D. L. Smith, Identification and Localization of Slow,
Natural, Cooperative Unfolding in the Hematopoietic Cell
Kinase SH3 Domain by Amide Hydrogen Exchange and
Mass Spectrometry, Biochemistry, 1997, 36, 14384–14391.
38 J. Hollien and S. Marqusee, Structural distribution of stabi-
lity in a thermophilic enzyme, Proc. Natl. Acad. Sci. U. S. A.,
1999, 96, 13674–13678.
39 H. Xiao, et al., Mapping protein energy landscapes with
amide hydrogen exchange and mass spectrometry: I. A gen-
eralized model for a two-state protein and comparison with
experiment, Protein Sci., 2005, 14, 543–557.
40 M. Rey, et al., Mass Spec Studio for Integrative Structural
Biology, Structure, 2014, 22, 1538–1548.
41 B. D. Pascal, et al., HDX Workbench: Software for the
Analysis of H/D Exchange MS Data, J. Am. Soc. Mass
Spectrom., 2012, 23(9), 1512–1521.
42 M. Guttman, D. D. Weis, J. R. Engen and K. K. Lee,
Analysis of Overlapped and Noisy Hydrogen/Deuterium
Exchange Mass Spectra, J. Am. Soc. Mass Spectrom., 2013,
24, 1906–1912.
43 R. Lindner, et al., Hexicon 2: Automated Processing of
Hydrogen-Deuterium Exchange Mass Spectrometry Data
with Improved Deuteration Distribution Estimation, J. Am.
Soc. Mass Spectrom., 2014, 25, 1018–1028.
44 D. E. Miller, C. B. Prasannan, M. T. Villar, A. W. Fenton
and A. Artigues, HDXFinder: Automated analysis and
data reporting of deuterium/hydrogen exchange mass
spectrometry, J. Am. Soc. Mass Spectrom., 2012, 23(2),
425–429.
45 S. Liu, et al., HDX-Analyzer: a novel package for statistical
analysis of protein structure dynamics, BMC Bioinf., 2011,
12, S43.
46 Z.-Y. Kan, L. Mayne, P. S. Chetty and S. W. Englander,
ExMS: Data Analysis for HX-MS Experiments, J. Am. Soc.
Mass Spectrom., 2011, 22, 1906.
47 Z. Zhang, A. Zhang and G. Xiao, Improved Protein
Hydrogen/Deuterium Exchange Mass Spectrometry
Platform with Fully Automated Data Processing, Anal.
Chem., 2012, 84, 4942–4949.
48 V. Katta, B. T. Chait and S. Carr, Conformational changes
in proteins probed by hydrogen-exchange electrospray-
ionization mass spectrometry, Rapid Commun. Mass
Spectrom., 1991, 5, 214–217.
49 C. R. Morgan and J. R. Engen, Investigating solution-phase
protein structure and dynamics by hydrogen exchange
mass spectrometry, Curr. Protoc. Protein Sci., 2009, ch. 17,
Unit 17.6, pp. 1–17.
50 I. A. Kaltashov, C. E. Bobst and R. R. Abzalimov, H/D
exchange and mass spectrometry in the studies of protein
Analyst, 2017, 142, 2874–2886 | 2885

Tutorial Review
conformation and dynamics: Is there a need for a top-down
approach?, Anal. Chem., 2009, 81, 7892–7899.
51 T. E. Wales and J. R. Engen, Hydrogen exchange mass spec-
trometry for the analysis of protein dynamics, Mass
Spectrom. Rev., 2006, 25, 158–170.
52 J. J. Rosa and F. M. Richards, An experimental procedure
for increasing the structural resolution of chemical hydro-
gen-exchange measurements on proteins: Application to
ribonuclease S peptide, J. Mol. Biol., 1979, 133, 399–416.
53 L. Cravello, D. Lascoux and E. Forest, Use of different pro-
teases working in acidic conditions to improve sequence
Published on 13 July 2017. Downloaded by University of Toronto on 05/11/2017 03:27:58.
coverage and resolution in hydrogen/deuterium exchange
of large proteins, Rapid Commun. Mass Spectrom., 2003, 17,
2387–2393.
54 J. Marcoux, et al., Investigating Alternative Acidic Proteases
for H/D Exchange Coupled to Mass Spectrometry: Plasmepsin
2 but not Plasmepsin 4 Is Active Under Quenching
Conditions, J. Am. Soc. Mass Spectrom., 2010, 21, 76–79.
55 T. Wales, M. Eggertson and J. Engen, Considerations in the
Analysis of Hydrogen Exchange Mass Spectrometry Data, in
Mass Spectrometry Data Analysis in Proteomics, ed.
R. Matthiesen, Humana Press, 2013, pp. 263–288.
56 D. F. Hunt, J. R. Yates, J. Shabanowitz, S. Winston and
C. R. Hauer, Protein sequencing by tandem mass spec-
trometry, Proc. Natl. Acad. Sci. U. S. A., 1986, 83, 6233–6237.
57 J. D. Venable, M.-Q. Dong, J. Wohlschlegel, A. Dillin and
J. R. Yates, Automated approach for quantitative analysis of
complex peptide mixtures from tandem mass spectra, Nat.
Methods, 2004, 1, 39–45.
58 R. S. Plumb, et al., UPLC/MSE; a new approach for generat-
ing molecular fragment information for biomarker struc-
ture elucidation, Rapid Commun, Mass Spectrom., 2006, 20,
1989–1994.
59 J. K. Chik, J. L. Vande Graaf and D. C. Schriemer,
Quantitating the Statistical Distribution of Deuterium
Incorporation To Extend the Utility of H/D Exchange MS
Data, Anal. Chem., 2006, 78, 207–214.
60 J. Zhang, P. Ramachandran, R. Kumar and M. L. Gross,
H/D Exchange Centroid Monitoring is Insufficient to Show
Differences in the Behavior of Protein States, J. Am. Soc.
Mass Spectrom., 2013, 24, 450–453.
61 P. L. Ferguson, et al., Hydrogen/deuterium scrambling
during quadrupole time-of-flight MS/MS analysis of a zinc-
binding protein domain, Anal. Chem., 2007, 79, 153–160.
62 Z. Zhang, S. Guan and A. G. Marshall, Enhancement of the
effective resolution of mass spectra of high-mass bio-
molecules by maximum entropy-based deconvolution to
eliminate the isotopic natural abundance distribution,
J. Am. Soc. Mass Spectrom., 1997, 8, 659–670.
63 R. E. Iacob, J. P. Murphy and J. R. Engen, Ion mobility adds
an additional dimension to mass spectrometric analysis of
solution-phase hydrogen/deuterium exchange, Rapid
Commun. Mass Spectrom., 2008, 22, 2898–2904.
2886 | Analyst, 2017, 142, 2874–2886
View Article Online
Analyst
64 J. R. Engen and T. E. Wales, Analytical Aspects of Hydrogen
Exchange Mass Spectrometry, Annu. Rev. Anal. Chem., 2015,
8, 127–148.
65 A. Kreshuk, et al., Automated detection and analysis of
bimodal isotope peak distributions in H/D exchange mass
spectrometry using HeXicon, Int. J. Mass Spectrom., 2011,
302, 125–131.
66 T. R. Keppel, B. A. Howard and D. D. Weis, Mapping
Unstructured Regions and Synergistic Folding in
Intrinsically Disordered Proteins with Amide H/D
Exchange Mass Spectrometry, Biochemistry, 2011, 50, 8722–
8732.
67 Z. Zhang, W. Li, T. M. Logan, M. Li and A. G. Marshall,
Human recombinant [C22A] FK506-binding protein amide
hydrogen exchange rates from mass spectrometry match
and extend those from NMR, Protein Sci., 1997, 6,
2203–2217.
68 J. Lísal, et al., Interaction of packaging motor with the poly-
merase complex of dsRNA bacteriophage, Virology, 2006,
351, 73–79.
69 J. Lísal, et al., Functional visualization of viral molecular
motor by hydrogen-deuterium exchange reveals transient
states, Nat. Struct. Mol. Biol., 2005, 12, 460–466.
70 A. Chandramohan, et al., Predicting Allosteric Effects from
Orthosteric Binding in Hsp90-Ligand Interactions:
Implications for Fragment-Based Drug Design, PLoS
Comput. Biol., 2016, 12, e1004840.
71 S. Krishnamurthy, et al., Distinguishing direct binding
interactions from allosteric effects in the protease–HK97
prohead I δ domain complex by amide H/D exchange mass
spectrometry, Bacteriophage, 2014, 4, e959816.
72 T. Rob, P. K. Gill, D. Golemi-Kotra and D. J. Wilson, An
electrospray ms-coupled microfluidic device for sub-second
hydrogen/deuterium exchange pulse-labelling reveals allo-
steric effects in enzyme inhibition, Lab Chip, 2013, 13,
2528–2532.
73 M. A. Sowole, S. Simpson, Y. V. Skovpen, D. R. J. Palmer and
L. Konermann, Evidence of Allosteric Enzyme Regulation via
Changes in Conformational Dynamics: A Hydrogen/
Deuterium Exchange Investigation of Dihydrodipicolinate
Synthase, Biochemistry, 2016, 55, 5413–5422.
74 D. Houde, S. A. Berkowitz and J. R. Engen, The utility of
hydrogen/deuterium exchange mass spectrometry in bio-
pharmaceutical comparability studies, J. Pharm. Sci., 2011,
100, 2071–2086.
75 S.-W. Yang, et al., Guanine nucleotide induced confor-
mational change of Cdc42 revealed by hydrogen/deuterium
exchange mass spectrometry, Biochim. Biophys. Acta,
Proteins Proteomics, 2016, 1864, 42–51.
76 R. G. McAllister and L. Konermann, Challenges in the
Interpretation of Protein H/D Exchange Data: A Molecular
Dynamics Simulation Perspective, Biochemistry, 2015, 54,
2683–2692.
This journal is © The Royal Society of Chemistry 2017