Full Terms & Conditions of access and use can be found athttp://www.tandfonline.com/action/journalInformation?journalCode=iedc20

Expert Opinion on Drug Discovery

ISSN: 1746-0441 (Print) 1746-045X (Online) Journal homepage: http://www.tandfonline.com/loi/iedc20

An overview of hydrogen deuterium exchangemass spectrometry (HDX-MS) in drug discovery

Glenn R. Masson, Meredith L. Jenkins & John E. Burke

To cite this article: Glenn R. Masson, Meredith L. Jenkins & John E. Burke (2017) An overview ofhydrogen deuterium exchange mass spectrometry (HDX-MS) in drug discovery, Expert Opinion onDrug Discovery, 12:10, 981-994, DOI: 10.1080/17460441.2017.1363734

To link to this article: https://doi.org/10.1080/17460441.2017.1363734

Accepted author version posted online: 03Aug 2017.Published online: 17 Aug 2017.

Submit your article to this journal

Article views: 750

View related articles

View Crossmark data

Citing articles: 2 View citing articles

REVIEW

An overview of hydrogen deuterium exchange mass spectrometry (HDX-MS) in drugdiscoveryGlenn R. Masson a, Meredith L. Jenkinsb and John E. Burke b

aProtein and Nucleic Acid Chemistry Division, MRC Laboratory of Molecular Biology, Cambridge, UK; bDepartment of Biochemistry andMicrobiology, University of Victoria, Victoria, Canada

ABSTRACTIntroduction: Hydrogen deuterium exchange mass spectrometry (HDX-MS) is a powerful methodologyto study protein dynamics, protein folding, protein-protein interactions, and protein small moleculeinteractions. The development of novel methodologies and technical advancements in mass spectro-meters has greatly expanded the accessibility and acceptance of this technique within both academiaand industry.Areas covered: This review examines the theoretical basis of how amide exchange occurs, howdifferent mass spectrometer approaches can be used for HDX-MS experiments, as well as the use ofHDX-MS in drug development, specifically focusing on how HDX-MS is used to characterize bio-therapeutics, and its use in examining protein-protein and protein small molecule interactions.Expert opinion: HDX-MS has been widely accepted within the pharmaceutical industry for thecharacterization of bio-therapeutics as well as in the mapping of antibody drug epitopes. However,there is room for this technique to be more widely used in the drug discovery process. This isparticularly true in the use of HDX-MS as a complement to other high-resolution structural approaches,as well as in the development of small molecule therapeutics that can target both active-site andallosteric binding sites.

ARTICLE HISTORYReceived 14 March 2017Accepted 1 August 2017

KEYWORDSHydrogen-deuteriumexchange massspectrometry; HDX-MS;hydrogen deuteriumexchange; HDX; proteindynamics; epitope mapping;bio-therapeutics; drugdevelopment; allostery

1. Introduction

Hydrogen deuterium exchange mass spectrometry (HDX-MS)is rapidly becoming an important part of the biochemist’stoolkit for exploring protein structure, dynamics, and function.HDX-MS has been applied to study protein folding [1], proteinoligomerization [2,3], protein:protein [4], protein:DNA [5], pro-tein:small molecule [6], and protein:membrane [7–9] interac-tions, as well as being applied to complement orthogonalbiophysical techniques such as X-ray crystallography [10,11],nuclear magnetic resonance [12], and electron microscopy[13]. Central to this diverse array of applications is the mea-surement of deuterium uptake along a protein’s backboneamide groups. The rate of uptake of deuterated solvent ofan amide hydrogen is primarily dependent on its involvementin hydrogen bonds present in secondary structure elements[14,15]. The technique is particularly well suited to examinechanges in conformation in disordered regions of proteins, asdisorder-order transitions will cause extremely large changesin the exchange rates of amide hydrogens [16]. Furthermore,this rate of exchange is also sensitive to changes in proteinconformation/solvent accessibility that occur on the additionof binding partners. Thus, through the observation of amideexchange rates under various conditions, such as the presenceand absence of a ligand, structural and dynamic properties ofthe protein and the binding interface can be examined.

The utilization of the hydrogen/deuterium exchange phe-nomenon to study protein structure is not a new concept,with much of the theoretical basis being determined in the1950s [17]. Exchange rate measurements were initially con-ducted using tritium and density gradient columns, and thenreplaced with deuterium and nuclear magnetic resonance(NMR) experiments in the 1980s and 1990s (for a review onthe early history of H/D exchange please consult [18]). NMR isstill used in H/D exchange measurements and can providehighly accurate single-residue exchange data; however, thereare disadvantages in the amount of protein required and thesize of proteins that can be studied. H/D exchange massspectrometry experiments rely on the coupling of low-tem-perature ultra-high-pressure liquid chromatography with thehigh sensitivity and resolution associated with modern massspectrometry instruments to determine the locations and ratesof amide hydrogens deuterium uptake. Additionally, advancesin the automation of both sample preparation and software-driven early stage data analysis has allowed for an increase inpopularity of HDX-MS studies, especially in the field of proteinbiopharmaceuticals where it is now routinely employed inepitope determination and quality control (for a review see[19]). Examples of the use of HDX-MS in drug discovery pro-cesses are highlighted in Figure 1. HDX-MS has long beendescribed as an ‘emerging’ technology, but its use is nowfairly commonplace – it can now be more accurately described

CONTACT John E. Burke jeburke@uvic.ca Biochemistry and Microbiology, University of Victoria, 3800 Finnerty Road 270 Petch Building Victoria, BritishColumbia, Canada

EXPERT OPINION ON DRUG DISCOVERY, 2017VOL. 12, NO. 10, 981–994https://doi.org/10.1080/17460441.2017.1363734

© 2017 Informa UK Limited, trading as Taylor & Francis Group

as an ‘emerged’ technology. This review hopes to highlightthe theoretical basis of amide exchange, how technologicaladvancements in mass spectrometry have expanded the cap-abilities of this technique, and focus on specific examples thatdemonstrate the power of this technique in drug discovery.

1.1. Theoretical underpinnings of amide deuteriumexchange

Backbone amides engaged in secondary structure exchangewith solvent via the following mechanism (see Equation 1).

N�Hcl�!kop �kcl

N�Hop �!kch N�Dop; (1)

where the equilibrium between ‘open’ and ‘closed’ forms ofthe backbone amide is determined by the rate constant kop forthe opening of the amide hydrogen bond and kcl, the rateconstant for the closing of the amide hydrogen bond. In thisscheme, solvent exchange can only occur while the amidegroup is in the ‘open’ conformation – when there is a localizedminor unfolding event. The rate of solvent exchange thatoccurs while the amide group is in the open conformation,kch, is governed by the steric and electrostatic effects ofsurrounding residues, the residue itself, the pH (or pD) of thesolution, and the temperature [28]. This is also frequentlycalled the intrinsic rate constant (frequently referred to as kior kint). Overall, the observed exchange rate under the steadystate approximation is given by Equation 2.

kHX ¼ kopkch = ðkop þ kcl þ kchÞ (2)

This scheme assumes that the protein is exposed to 100%D2O, and only the deuteration exchange reaction proceeds,that is, there is no H2O present, and no incorporated deuter-ium could be replaced with a proton (known as back-exchange). The intrinsic exchange rate constant for theamide hydrogen of any residue can be theoretically calculatedfor a specific pH and temperature, as the parameters govern-ing this exchange have been carefully measured experimen-tally [28]. Within a folded protein the observed rate ofhydrogen deuterium exchange of an amide (kHX) can bedecreased relative to the intrinsic rate (kch) by 1010, and thisdecrease in exchange can be represented as a protectionfactor ‘P’, which is the ratio of the intrinsic rate kch over theexperimentally measured exchange rate kex.

Article highlights

● HDX-MS has established itself as a well-validated tool for the char-acterization of bio-therapeutics.

● Improvements in instrumentation and data analysis have greatlyexpanded the complexity of samples that can be interrogated byHDX-MS, as well as increasing the spatial resolution at which they canbe measured.

● HDX-MS is well suited as a tool for determining altered proteindynamics upon binding to both protein and small molecule ligands

● HDX-MS provides major advantages in the rapid determination ofboth protein and small molecule binding sites on proteins

● HDX-MS is also well suited upstream in the drug discovery process asit can be used for the development of optimal constructs for highresolution structural approaches that can be used for SAR studies

This box summarizes key points contained in the article.

276-287

365-373

399-410

Compound

Protein-Ligand Binding:RORg binding T0901315

Epitope MappingInterleukin 23/FAB complex

IL-23FAB

a b

c d

Guiding X-ray Construct Design: PI4K/Rab11 Complexes

Crystallisation

i%D Full Length PI4K %D Xtal PI4k

0 200 400 600 8000

50

100

% d

eute

rium

inc.

Biopharmaceutical Characterisation: Oxidation of IFN

121-134

62-67

88-101

8-15

154-162

Figure 1. Utility of HDX-MS in drug discovery. (a.) HDX-MS to probe the dynamics of protein-ligand interactions. The difference in HDX-MS induced upon theinteraction of the Retinoic Acid Receptor gamma (RORγ) with an inverse agonist, T0901315 is shown. Three peptides (399–410 in green, 365–373 in violet, 276–287in yellow) all exhibited alterations in H/D exchange on binding to the compound (shown in pink, stick form) [20,21]. (b.) HDX-MS to inform construct design for highresolution structural approaches [22,23]. HDX-MS exchange rates for a full length PI4KIIIβ construct (HDX incorporation shown in red on the uptake graph) was usedto identify disordered regions, which were subsequently removed to create an optimized crystal construct (HDX incorporation shown in blue). This constructmaintained the same overall fold, and allowed for the high resolution structure determination of PI4KB. (c.) HDX-MS for characterizing biopharmaceuticals. HDX-MSwas used to determine the structural deformations of Interferon β 1a (IFN) that occur due to oxidation of various cysteine and methionine residues (shown in green).Five peptides, shown in orange, exhibited a large increase in HDX, most likely caused by large-scale unfolding [24,25]. (d.) HDX-MS for use in Epitope Mapping. HDX-MS was used in conjunction with a variety of other techniques, to determine the epitope of Interleukin-23 [26,27].

982 G. R. MASSON ET AL.

Amide exchange MS profiles have been classified into twodifferent regimes depending on the relative rates between kchand kcl. In EX2 kinetics, where kch ≪ kcl, the amide bond mustundergo numerous unfolding events before the amide willexchange with solvent. EX2 kinetics can be identified from thecharacteristic gradual increase in a peptide’s isotopic distributionas the heavier deuterium is incorporated into backbone amides.Alternatively, EX1 kinetics, less common under native conditions,occurs when kcl ≪ kch, leading to a situation where the openingand closing rates are slow compared to the intrinsic exchange rate.This leads to every opening event leading to exchange withdeuterium. Peptides exhibiting EX1 kinetics produce a signaturebimodal isotopic distribution that can present a challenge to dataanalysis of deuterium incorporation [29].

While the equations governing H/D exchange rates with sol-vent have been known for decades, the exact molecular under-pinnings determining why amide hydrogens exchange at a givenrate are still not exactly clear. Themost commonly cited reasons forwhy amide hydrogens are protected from exchange is due to theirinvolvement in secondary structure, as well as their protectionfrom bulk solvent [14,15]. However, it must be taken into accountthat these same studies also identified a number of amide hydro-gens in unstructured regions that still showed significant protec-tion factors. One of the contributing factors of decreasedexchange rates for these exposed regions was the presence ofstructured water molecules [14]. However, this is unlikely to fullyexplain why these regions are protected as experiments usingubiquitin as a model protein found that exposed unstructuredregions showed large differences in protection factors, with noclear relationship to nearby structured water networks [30]. Initialattempts have been made to examine the open amide state asdescribed in equation 1 using long (~1 ms) molecular dynamicsimulations, with potential open amide state intermediates beingidentified as very short-lived (~100 ps) species [31]. In summary,there is still much we do not understand concerning why amidehydrogens in a protein will be protected from exchange, andcaution must be applied in the use of overly simplistic modelsthat define protection from exchange as only being driven byeither involvement in secondary structure or accessibility tosolvent.

1.2. HDX-MS methodology

A HDX-MS experiment can be split into numerous discrete steps.This section will introduce these different methodological steps,and specifically highlight the most frequently used approaches.The important steps in any HDX approach are: (1) deuteriumincorporation, (2) spatial localization of deuterium incorporation,and (3) data analysis of deuterium incorporation. Readers areadvised to consult excellent recent reviews for more comprehen-sive methodological details on best HDX practice [32–34].

1.2.1. Deuterium incorporationDeuterium labeling can be carried out as an on-exchange reaction,where deuterium is incorporated through incubation of the pro-tein in a deuterated buffer, or as an off-exchange experimentwhere loss of deuterium is measured from a fully deuteratedprotein when it is diluted in a non-deuterated buffer [35]. Mostexperiments are carried out using an on-exchange labeling

approach, as complications can exist in the generation of a fullydeuterated protein required for the off-exchange approach, parti-cularly for large and unstable protein complexes. Proper executionof deuterium labeling requires the examination of multiple time-points of deuterium incorporation, so differences in the kinetics ofH/D exchange for amides with varying exchange rates can bemeasured. This is particularly important for users solely usingautomated sample handling systems, as these frequently under-represent very short deuterium incorporation time points, and thiscan lead to an underestimation of changes in protein dynamics inareas ofweak/transient secondary structure. This can be addressedby varying both the pH, the temperature, and the length ofdeuterium exposure, which enables the measurement of amideexchange rates that range over 8 orders of magnitude [36,37].However, caution must be applied when altering pH as thisrequires that the protein be stable across the pH ranges examined.The use of other biophysical measurements (differential scanningfluorimetry, multi angle light scattering, etc.) is suggested to verifythat protein conformation is stable over the desired pH range.

The protein, typically in a high nanomolar to low micro-molar concentration, is labeled in an aqueous semi-‘native’environment both in the presence and absence of any bindingpartners. Dependent on the mass spectrometer used in theanalysis, roughly 5–100 pmol of sample are required.Specialized nano-spray systems have allowed for the use ofsub-picomole levels of protein per sample [38]. Of importantnote is that HDX measurements are dependent on the occu-pancy of the protein of interest with the binding partner, andtherefore careful consideration must be given in the relativeconcentrations of the protein of interest and the bindingpartner to maximize occupancy. D2O exposure is carried outfor a variety of time points ranging typically from seconds tohours, although this can be extended to days for extremelystable proteins. The use of pulsed label and stop flow appa-ratus is particularly useful for the analysis of extremely rapidlyexchanging amide hydrogens with rates on the millisecondtimescale [39,40]. Amides with extremely rapid exchange ratescan also be labeled in the gas phase of the mass spectrometer,with a 0.1–10 ms labeling pulse of deuterium using ND3 in theion guide of the MS [41]. Carrying out experiments at veryshort deuterium incorporation time courses is exceptionallyimportant in the analysis of intrinsically disordered proteinsand is necessary to properly identify disordered regions thatcan be removed for optimization of constructs for other struc-tural approaches.

1.2.2. Spatial localization of deuterium incorporationDifferent techniques to spatially resolve H/D exchangeinformation are summarized in Figure 2. The simplestHDX-MS experiment is carried out by examining globaldifferences in H/D exchange. It is carried out by directlyinjecting the intact protein onto an LC-MS system, and theoverall mass of the entire protein is measured. The proteinof interest can be incubated with different ligands (pro-teins, inhibitors, etc.), and any differences in H/D exchangecan be visualized. This provides no information on thelocation of the altered deuterium exchange, and any con-ditions where there are an equal amount of increases and

EXPERT OPINION ON DRUG DISCOVERY 983

decreases in exchange in multiple different regions willshow no net change in global deuterium incorporation.

Spatial localization of amide exchange rates using HDX-MS can be carried out using both top-down and bottom-up approaches. In the bottom-up approach, deuteriumincorporation is localized in solution using protease-

generated peptide fragments, and in the top- downapproach deuterium incorporation is localized within themass spectrometer through the fragmentation of the pro-tein. These techniques can be combined in a so-calledmiddle-down approach where the location of deuteriumincorporation within protease-generated peptides can be

+D2O

H D

Labeling

Acid Proteolysis

Peptide Separation & IdentificationNon-Deuterated

UnboundDeuterated

Ligand BoundDeuterated

Data Analysis & Interpretation

Peptide 35-42

Injection & Ionisaiton

with ESI source

Mapping onto

protein structures

Bottom-Up HDX-MS

Pseudo Amino-Acid Resolution

Achieve very high degree of peptide redundancy

MESPLEEEAGFSWRRNSA1 10

Mathematical modeling of most-likely exchange

pattern with back-exchangecorrection

ETD Amino-Acid Resolution

Identification ofpertinent peptides

Creation of c & z fragment seriesusing ETD in non-scrambling conditions

Intact Mass Spectrometry

+D2O

+88 Da

+D2O & Ligand

+82 Da

Top-Down Approaches

+D2O + Ligand

Single- Amino Acid Level Resolution Data Interpretation

Fragmentation

Domain-level resolutionHDX-MS

Single amide resolutiondepending on c/z series

Middle-Down Approaches

Injection into

Fluidics system

Figure 2. The various experimental approaches for HDX-MS. All HDX-MS (with the exception of gas-phase HDX-MS) starts with an aqueous labelling step, where aprotein of interest is exposed to deuterium for a pre-determined amount of time, and then quenched. This protein can then either be injected onto a fluidics systemand digested by an acid-functional protease (the ‘bottom-up’ approach), or injected onto the mass spectrometer for an intact mass measurement. This intact proteinmay then be partially fragmented whilst inside the mass spectrometer to yield either intact domains or single peptides for comparison between states (the Top-Down approach). The bottom-up approach may be furthered by either the a) mathematical determination of individual exchange rates (to achieve pseudo amino-acid resolution) or b) targeted ETD fragmentation of individual peptides (middle-down approaches). Data adapted from [42].

984 G. R. MASSON ET AL.

further refined through fragmentation within the massspectrometer [43].

1.2.3. Bottom-up HDX-MS approachesDue to the simple requirements for MS infrastructure androbust experimental workflows, most published HDX-MSexperiments have been performed using a bottom-upapproach. The key steps in a bottom-up H/D exchange experi-ment downstream of labeling are: (1) Quenching, (2)Digestion, (3) Separation, and (4) Mass analysis. Top-downexperiments exclude the digestion step, as fragmentation ofthe protein occurs in the mass analysis step.

After deuterium labeling has occurred, the exchange reac-tion can be quenched. Quenching the exchange reaction canbe achieved by simultaneously reducing the temperature to 0°C and the pH to 2.5. The exchange of amide hydrogens is anacid and base catalyzed process with a global minimum at apH of ~2.5–2.8, thus the addition of a quench buffer minimizesback-exchange of amide hydrogens in all subsequent steps.All steps from this point on must be performed as rapidly aspossible at low temperatures, and held at pH 2.5, to preventback-exchange. Generally, while quenching the exchangereaction the labeled protein is also denatured through theinclusion of a denaturant (e.g. guanidinium chloride) and anacid-functional reductant (e.g. TCEP) in the quench solution.TCEP is particularly important for proteins containing multipledisulfide bonds [9,44], as peptides containing linked cysteinespresent a challenge for peptide identification. Denaturing andreducing the protein (either through the use of an acid func-tional reductant or an electrochemical cell [45]), facilitates theproteolysis of the protein using an acid-functional proteaseeither in solution or through an immobilized protease columnwithin a fluidics system. The primary protease used for theseexperiments is pepsin, with other acid functionalized pro-teases such as fungal XIII, XVI, and nepenthesin also beingused [46–49]. The proteolysis reaction may be conducted atslightly elevated temperatures (~10–20°C) to increase diges-tion efficiency for proteins that are resistant to protease diges-tion. Caution must be taken when digesting at highertemperatures, as this can lead to increased back-exchangerates with protiated solvent. The use of pepsin coupled tosmall particles that operate at high back pressures (~10,000psi) also leads to enhanced protease efficiency [50,51].

The deuterium-labeled peptides are then separated on achilled HPLC/UPLC system. This system is kept as cool aspossible to limit the loss of the deuterium signal throughback-exchange. Separation has to occur as fast as possible,and the use of UPLC backpressures (>15,000 psi) allows forexcellent peptide separation in 5–20 min [52]. The use ofdifferent buffer systems allows for the use of subzero tem-peratures during peptide separation, further reducing back-exchange levels [53,54]. Robotic liquid handling systems areavailable that automate the labeling, digestion, and separationsteps; however, caution must be applied in the use of thisapproach depending on the stability of the protein understudy. Peptides are then injected into the mass spectrometryinstrument and ionized, typically using an electrospray ioniza-tion source.

An important note on the methodology of bottom-updeuterium exchange experiments is that users need to payclose attention to back-exchange that occurs during the pro-teolysis and LC separation steps. For proteins that requirelonger digestion and reduction times before injection on thefluidics system, considerations about spurious non physiologi-cal on exchange that occurs in the quench buffer also need tobe taken into account. Some of the different parameters thataffect back-exchange have been measured [55], and carefulanalysis of the back-exchange levels of any given system issuggested through the generation of fully deuterated proteinsamples (for more discussion of these considerations consult[33]). However, for experiments examining only the relativehydrogen exchange between conditions, the generation offully deuterated samples is not necessary, although it is impor-tant to understand that all H/D exchange incorporation curveswill be a relative measurement of deuterium incorporation.One of the most common mistakes new users experience isnot properly addressing peptide carryover, which can mas-sively distort H/D exchange profiles [56,57].

The isotopic profile of the charged deuterated peptides arethen determined using the mass spectrometer. The exchangerates of individual peptides are then compared in the absenceor presence of a binding partner. Information on protein con-formation is contained both in the mass centroid of the deut-erated peptide, as well as in the shape of the isotopic profile.Isotope profiles are useful in determining EX1 and EX2 kinetics[29], and can also be useful in determining residue-specificexchange information for regions with multiple peptides [58].Through comparing the profiles of identical peptides thatwere deuterated under differing conditions – with and with-out binding partners, for example – it is possible to identifypeptides that have altered deuteration rates, and thus deter-mine the location of structural changes that accompany bind-ing events.

Data analysis and interpretation are significant steps inHDX-MS, particularly with larger proteins that may producethousands of peptides. The discussion of this topic is outsidethe scope of this review, and readers are recommended toconsult excellent recent reviews on different H/D exchangesoftware analysis packages [32,59]. Although software hasautomated much of the data analysis, the manual inspectionof all peptides and their deuterium incorporation is still essen-tial to minimize any complications from overlapping peptides/noisy data.

1.2.4. Top down HDX-MS approachesSome inherent difficulties with the bottom-up approachare that the spatial localization of any change in exchangerate is dependent on the length of protease generatedpeptide fragments, as well as the required generation ofa number of overlapping peptide fragments.Fragmentation within the mass spectrometer, specificallythe use of electron transfer dissociation (ETD) [60–62] orelectron capture dissociation (ECD) [63,64], can be used togenerate localized spatial information. Collision-induceddissociation (CID) is not compatible with HDX-MS experi-ments due to the propensity for CID to randomize the

EXPERT OPINION ON DRUG DISCOVERY 985

location of the amide hydrogens/deuteriums in the result-ing fragments, a phenomenon known as scrambling[65,66]. ETD and ECD can used for MS/MS experiments,but in order to minimize scrambling it is crucial to usestandard peptides to determine the instrument parameterswhich result in minimal scrambling [67]. The top-downapproach can be used on intact proteins, but the coverageand spatial localization will be dependent on the genera-tion of the full spectrum of c/z fragment ions. This can bechallenging for large (>40 kDa) proteins; however, this canbe surmounted through the use of a middle-down methodwhere protease generated peptides are fragmented usingeither ETD or ECD to gain site specific exchange informa-tion [60,61].

1.3. HDX-MS in drug discovery

HDX-MS has been widely used in the academic communitysince the early 1990s. It has proven itself to be a valuablebiophysical tool to study many biological processes, includ-ing pioneering studies examining protein–protein and pro-tein–small molecule interactions. A growing area of interesthas been in the study of membrane proteins, including bothintegral membrane proteins [68–72], as well as the coordi-nated recruitment of peripheral membrane proteins [7–9,42,69,73–83]. The analysis of membrane proteins hasbecome amenable to HDX-MS studies, including the largestfamily of cell surface receptors (G-protein coupled recep-tors) [84]. For technical details on the use of HDX-MS foranalysis of membrane proteins readers are advised to con-sult [33]. We will highlight the use of HDX-MS in drugdiscovery on three areas: the analysis of protein–small mole-cule interactions, characterization of bio-therapeutics/biosi-milars, and epitope mapping of bio-therapeutics. We willalso discuss emerging areas of how HDX-MS can be usedin drug discovery approaches, including as an aid for thegeneration of novel constructs for high-resolution structuralapproaches.

1.3.1. HDX-MS to examine protein–small moleculeinteractionsAlthough the administration of protein therapeutics is rapidlyexpanding, small molecules – typically in the order of less than800 Da, are still the most common therapeutics on the market.There are a variety of analytical tools that are able to detectprotein–ligand interactions –NMR, isothermal titration calori-metry, microscale thermophoresis, and surface plasmon reso-nance can all be used to detect and characterize protein–ligand interactions. HDX-MS provides an opportunity to notonly observe ligand binding and determine the location of aligand binding site, but also provides an insight into differ-ences in conformational dynamics that accompany binding –information that is crucial in developing novelpharmaceuticals.

The low protein requirements and sensitivity of HDX-MSmakes it well suited to probing the dynamics of ligand bind-ing, especially with low-affinity compounds [85] and whenworking with proteins that are problematic for crystallogra-phy, such as those with extensive unstructured regions [86,87].

Additionally, there is the potential for both the automation ofsample preparation using fluid handling robotics and, withincertain limits, automated data processing [88]. Despite this,the use of HDX-MS is still not widespread in the screening ofcompounds. This is primarily due to two issues: the resolutionof the technique and the interpretation of binding results.

The interpretation of hydrogen/deuterium exchange datafor small molecule binding events requires careful considera-tion. Amide hydrogen exchange is dependent not only onsteric shielding caused by a ligand interacting with the back-bone amide groups of a peptide, but also allosteric conforma-tional changes in the surrounding structure and dynamics ofthe protein [89]. Additionally ligand binding can, possiblycounter intuitively, cause both increases in exchange rateand no net change in hydrogen deuterium exchange rate[90] – suggesting that it is an oversimplification to simplydesignate peptides with the largest decreases in exchangerate as the locus of ligand binding. This problem is com-pounded by the peptide-level resolution of HDX-MS, as theobserved exchange rates are a net change in deuteration for asingle peptide, which is the sum of each individual amide’schange in deuteration rate. As recently demonstrated, ligandinteractions that result in a mixture of increases and decreasesin exchange rate on different amides within the same peptidemay ‘cancel out’, making the changes in solvent exchange‘invisible’ at peptide-level resolution [91]. While peptide-levelresolution may be sufficient for many applications, such asprotein–protein interactions where there is often a largeextended binding surfaces, small molecule-binding eventsmay be the result of very few interactions in noncontiguousamino acids that constitute the binding pocket. This problemis exacerbated in fragment-based lead generation (FBLG)methods, where initial screens may be of compounds assmall as 200 Da, and an even smaller binding footprint couldbe expected. Increasing the resolution of HDX-MS to thesingle-amino acid level can be achieved through two meth-ods: using tandem mass-spectrometry with ETD/ECD [64], orcomputationally using overlapping peptides [58,92–94]. Theprimary means of enhanced coverage for most H/D exchangeexperiments has been through the use of overlapping pep-tides. However, there can be difficulties in the use of over-lapping peptides for single amide resolution, particularlydepending on variable levels of back exchange and the effi-ciency of protein digestion in producing a sufficient number ofoverlapping peptides, as well as destructive interference thatmay occur with FT-MS instruments [95].

However, even with these caveats of data interpretationand analysis, HDX-MS has played a key role in numerouslaboratories for the characterization of protein–small moleculeinteractions. The following is not an exhaustive summary of allprotein–small molecule HDX-MS experiments, and selectivelyhighlights a number of studies that verify the usefulness ofusing HDX-MS as a tool to study protein–small moleculeinteractions.

The use of HDX-MS is particularly well suited for thediscovery of allosteric inhibitors, as well as how inhibitorsmediate conformational changes distant from their bindingsite. HDX-MS carried out on inhibitors of the Hepatitis CViral RNA dependent RNA polymerase revealed long range

986 G. R. MASSON ET AL.

allosteric conformational changes driven by inhibitor bind-ing, revealing the inhibitor mode of action [96]. HDX-MShas revealed how alternative ligand binding sites on thenuclear receptor transcription-factor PPAR-gamma causeconformational changes potentiating the receptor forhyperactivation [97]. HDX-MS on the retinoid X receptor αligand binding domain revealed how it is differentiallyeffected by the presence of co-activating peptides andantineoplastic therapeutic agents [98]. Combined crystal-lographic and HDX-MS studies on ligand binding of theestrogen receptor alpha [99], provided useful mechanisticinsight into differences between the inhibitors, and con-firmed crystallographic analysis. HDX-MS was critical forthe identification of a novel allosteric inhibitor bindingsite on antiapoptotic protein MCL-1, providing a novelstrategy to target this protein in disease [100]. HDX-MSwas used to determine the ligand binding site on IL-17A[101] and revealed that an additional α-helical region ofthe cytokine that was destabilized upon compound bind-ing, providing a hypothesis to why they were unable tocrystallize the ligand-protein complex. Combined crystal-lographic and HDX-MS analysis of the excitatory aminoacid transporter 1 revealed the structure of allosteric inhi-bitors bound to this membrane protein, and how differ-ences in protein dynamics in the presence of the allostericinhibitor mediate inhibition determining [102].

HDX-MS can also provide insight into distant structural rear-rangements that occur upon ligand binding in fragment screen-ing approaches. A recent study investigated both high-affinityligands (KD approx. 20 nM) and low-affinity fragments (500 μM)binding to the N-terminal ATPase domain of Hsp90 using HDX-MS [85], concluding that both the ligands and fragments sharedthe same interaction loci. Through overlaying HDX-MS dataonto structures produced from X-ray crystallographic studies,not only was the primary binding site determined, but orthos-teric sites were also identified, along with structurally distaldifferences in conformational dynamics of the protein whichaccompanied ligand binding, structural insights that would nothave been observed using crystallography alone. Significantdifferences in exchange were determined between two low-affinity phenolic fragments that were of similar size and affinity– highlighting the possible utility of HDX-MS in structurallyscreening low-affinity compound series.

HDX-MS has been particularly useful in the targeting ofprotein kinases using small molecule inhibitors. The bindingof both allosteric and active site inhibitors has been exam-ined for the oncogenic Bcr-Abl protein [103], which revealedhow allosteric inhibitors modify the conformation of theactive site. Further HDX-MS studies revealed novel allostericcommunication between the active site and myristate bind-ing site [104] on both WT and the T315I gatekeeper mutant,revealing the advantages of targeting non-active site inhibi-tor binding sites. Study of the protein kinase Hck and itsactivation downstream of the HIV-1 viral protein Nef revealedthat Nef led to subtle dynamic conformational changes thatactivate the kinase, and that the presence of Hck antiretro-viral inhibitors reversed these changes [105].

HDX-MS has also shown promise in the design and char-acterization of small molecule inhibitors for lipid-modifying

proteins, as many times these inhibitors can be extremelyhydrophobic and unsuitable for crystallographic approaches.Inhibitors of the phospholipase A2 family of enzymes havebeen designed using a combined HDX-MS and moleculardynamics approach [6,106–108] that were retractable toother structural methods. HDX-MS analysis of different inhibi-tors of the monoacyl glycerol lipase critical in cannabinoidsignaling revealed distinct stabilization of different conforma-tional ensembles by different active site inhibitors [109]. HDX-MS was critical for the identification and mutation of a novelallosteric inhibitor binding site in phosphatidylinositol 5-phos-phate 4-kinase gamma, allowing for the study of the role ofthis kinase in epithelial cell polarity [110].

1.3.2. HDX-MS to examine bio-therapeuticsThe bio-therapeutic market has grown explosively, and has ledto a number of challenges in how to characterize these drugs(the following review provides an excellent summary of someof these challenges [111]). One of the major growth areas ofbio-therapeutics has been in the development of humanizedmonoclonal antibodies. With this growth, HDX-MS has takenan important role in mapping epitopes of antibody–antigeninteractions. Another major area of growth for HDX-MS hasbeen in the characterization of posttranslational modificationsin protein therapeutics, specifically how these posttransla-tional modifications modify protein conformation, as well asdetermining similarity guidelines for biosimilar therapeutics.We will highlight a subset of studies that highlight the useful-ness of HDX-MS for bio-therapeutic characterization anddiscovery.

1.3.3. Characterization of bio-therapeutics by HDX-MSThe development of bio-therapeutics has fundamentally chan-ged the drug development landscape. Almost all bio-thera-peutics are posttranslationally modified, and there areimportant aspects of their production that require novel bioa-nalytical techniques to carry out batch-to-batch quality con-trol. HDX-MS has emerged as an important technique to carryout this analysis. Houde et al. were one of the first to use HDX-MS to probe antibody structure, using both global and localHDX-MS on IgG1 antibodies to localize structural changeswhich accompanied glycosylation [112]. HDX-MS was alsoused to examine how posttranslational modifications of IgG1antibodies, specifically methionine oxidation, fucosylation, andgalactosylation, result in changes in protein conformationaldynamics, impacting on their ability to binding to theFcγRIIIA receptor [113]. The most common posttranslationalmodification is glycosylation, and this is a particularly challen-ging modification to deal with, as glycosylation sites are fre-quently heterogeneous. HDX-MS of glycosylated proteins canbe carried out on the modified peptides [114]; however, thiscan be challenging for very heterogeneous samples. HDX-MSworkflows have been developed to analyze these highly het-erogeneous samples, specifically using deglycosylatingenzymes to remove all PTMs [115]. HDX-MS is also beingused as a tool to define how different biosimilars are fromthe original bio-therapeutic [111]. One of the most commonmodifications of protein therapeutics is the addition of cova-lently attached PEG molecules, and HDX-MS has been used

EXPERT OPINION ON DRUG DISCOVERY 987

successfully to examine how nonnative PTMs change overallprotein conformation [116].

1.3.4. HDX-MS to map epitope binding sitesOne of the most frequent uses of HDX-MS has been for use ofcharacterizing protein–protein binding sites. The developmentof bio-therapeutic drugs has led to a massive demand for themapping of antibody epitope binding sites on proteins. HDX-MS has proved to be a robust methodology for this approachwith many epitope mapping experiments published using thisapproach, with many more unpublished studies within thepharmaceutical industry. The epitope mapping approach suf-fers from some of the same disadvantages described in thesmall molecule section, as changes in deuterium exchange canbe driven by both direct interactions, as well as long-rangeallosteric conformational changes. Even with this caveat, HDX-MS has been proven to be exceptionally well suited to deter-mining antibody–antigen epitopes [26,117–119]. The HDX-MSapproach is particularly useful for epitopes that span multipleprotein binding sites, as these are particularly challenging tostudy by other structural approaches [117]. Another majoradvantage of the HDX-MS approach is that these interactionscan be determined within their native environment, which isparticularly important for studying antibody interactions withintegral membrane proteins [120]. The use of HDX-MS hasbeen particularly useful for the development of novel vac-cines, as it allows for the determination of epitope bindingsites for neutralizing antibodies [121–127]. HDX-MS has pro-ven itself as a well validated approach for the epitope map-ping of antibody–antigen complexes, and the furtherdevelopment of methodologies that allow for a faster, morecomprehensive, and higher resolution analysis will only lead tofurther adaptation in both academia and industriallaboratories.

1.4. Other possible uses of HDX-MS in drug development

HDX-MS is a powerful bioanalytical technique, but it will mostlikely have the most impact on questions of major biologicalsignificance if it is used in a synergistic manner with otherbiophysical and structural techniques. The most commonlyapplied approach for HDX-MS experiments is mapping thedynamic differences in amide exchange between conditionson previously determined structures determined by NMR orX-ray crystallography. This is particularly useful in examiningconformational changes that are unamenable to high-resolu-tion approaches, especially in examining protein membranecomplexes. The recent explosion of Cryo-electron microscopy,and the development of novel infrastructure and data analysispipelines for high-resolution EM structures, provides an excit-ing opportunity for synergistic application with HDX-MS. SinceCryo-EM is also able to identify multiple different proteinconformations, it is particularly well suited to be combinedwith the dynamic information provided by hydrogenexchange. Further, there are many opportunities for HDX-MSto be used as a key tool for optimizing approaches for high-resolution approaches.

HDX-MS was first reported as a potential tool to optimizecrystallization constructs in 2004, with the use of short

deuterium pulses (10 s) to identify disordered N- and C-termini that could be truncated in further crystallographicconstructs [10,128]. The use of this approach has also beenapplied to the crystallization of large multidomain proteincomplexes, as an example the complex of phosphatidylino-sitol 4-kinase IIIβ and its complex with the G protein Rab11,where intrinsically disordered regions were identified andtruncated internally as well as at the N- and C-termini[22,23]. This led to constructs that were amenable to high-resolution structures with potent small molecule inhibitors[129]. HDX-MS was used in both construct design andcharacterization of small molecule inhibitors in a studytargeting the NAD+-dependent lysine deacylase human sir-tuin 1 (hSIRT1), a therapeutic target implicated in a widerange of diseases related to aging [130]. Unstructuredregions in hSIRT1 were determined, which allowed themto identify the boundaries of constructs used in X-ray crys-tallographic studies. This approach has not obtained wide-spread acceptance within the crystallographic community,and there is most certainly a potential for growth in thisarea. This is particularly important for large multidomainmultiprotein complexes containing many intrinsically disor-dered regions.

Another important advance in protein structural biologyhas been the use of conformational selective nanobodies astools for crystal generation, and this has been particularlyimportant in high resolution structures of GPCRs. HDX-MS iswell suited for the screening of conformational specific nano-bodies, particularly in the selection of nanobodies that aremost likely to promote crystallization [75]. HDX-MS is alsouseful in the generation of constructs for NMR structuralapproaches [12,131]. Finally HDX-MS is well primed to helpinterpret structures generated by Cryo-EM, as protein–proteininterfaces identified by EM can be verified by HDX-MS [13,69],with HDX-MS also playing key roles in defining dynamic pro-cesses invisible to both Cryo-EM and X-ray crystallography.The use of HDX-MS as a tool for optimizing and synergizingwith high-resolution approaches provides another opportunityto enhance drug discovery efforts. For systems that are recal-citrant to high resolution approaches the combined approachof using HDX-MS with molecular dynamic simulations andcomputational docking provides an alternative approach toexamine protein structure–function relationships. In theabsence of high-resolution information, careful interpretationof HDX-MS data is required, with site-directed mutagenesisdriven by HDX-MS as a suggested technique to verify struc-tural hypotheses.

2. Conclusions

HDX-MS has become a widely accepted bioanalytical tool tostudy protein structure and dynamics. It has obtained wideacceptance in the biotech and pharma industry for the char-acterization of bio-therapeutics. The continued developmentof dedicated infrastructure at all stages of the HDX-MS experi-ment will lead to the ability to examine larger and morecomplex proteins, at higher spatial resolution. The use ofHDX-MS is primed to grow, both as a technique to be used

988 G. R. MASSON ET AL.

in tandem with other medium- to high-resolution structural

approaches, as well as in the drug development pipeline forscreening of both fragments and small molecules. Continuedrefinement of the theoretical basis for why amides exchangeat given rates will allow for a more detailed understanding ofthe molecular basis for differences in exchange among differ-ent protein states. The continued growth of bio-therapeuticsprovides an exciting application for HDX-MS and will certainlylead to further developments in both infrastructure and dataanalysis.

3. Expert opinion

Until recently, the majority of HDX-MS publications haveexamined protein dynamics by applying bottom-up HDX-MSapproaches. This was conducted using home-built tempera-ture-controlled fluidics devices, and analytical options werelimited. Despite these shortcomings, this particular utilizationof HDX-MS had immediate relevance for commercial andindustrial scientists working in biopharmaceutical sectorswho could immediately apply this to epitope mapping andbio-therapeutic characterization. With an increased interest inHDX-MS from these sectors came further advances in instru-mentation, automation, and software that have also greatlybenefited academia. In parallel to these improvements hasbeen the concurrent increases in sensitivity and resolutionseen in all classes of mass spectrometers.

This has led to HDX-MS experiments that are faster, moreaccurate, and more reproducible than ever before. Withincreased automation in both data collection and processing,the amount of time spent on any single experiment hasdecreased dramatically, and this has allowed for the investiga-tion of larger protein complexes. In addition to this, the pos-sible increases in spatial resolution achievable through the useof middle-down approaches have opened new avenues forresearch. HDX-MS has been fully utilized in the developmentof bio-pharmaceuticals and is widely used in various aspects ofbiopharmaceutical development and manufacture, from epi-tope mapping to monitoring batch-to-batch variation.However, there are areas where HDX-MS could be morewidely applied, particularly as a tool for construct design forhigh-resolution structural approaches, as well as for moreroutine use in small-molecule drug development.

HDX-MS is uniquely well placed as a tool for medium-throughput small molecule binding studies in tandem withother biophysical approaches. HDX-MS places very few con-straints on the protein being investigated, intrinsically disor-dered segments are as amenable to analysis as folded proteindomains. Membrane proteins can be screened for compoundbinding effects while held within a lipid or detergent mem-brane – more accurately mimicking their in vivo environment.The complexity of protein systems able to be studied, as wellas the ability to probe protein dynamics, will allow for a muchdeeper level of investigation into the molecular basis of howsmall molecules act as agonists, partial agonists, antagonists,and inverse agonists on important therapeutic targets. Thesmall amounts of protein required for HDX-MS (when com-pared to NMR or X-ray crystallography), when paired with the

ability to automate sample preparation, readily allows for thescreening of small molecules. Furthermore, given the sensitiv-ity of H/D exchange, and its capacity to tolerate high concen-trations of compound present in the labeling experiment,HDX-MS is well suited to screen weakly binding compounds,or early stage fragments. Importantly, the time required forHDX-MS experiments is on the order of days compared toweeks or months in other structural approaches. HDX-MScould also be more widely adapted as a strategy to generateconstructs amenable for high-resolution approaches, espe-cially for protein complexes with extensive intrinsically disor-dered regions. A synergistic application of H/D exchange as atool for optimization of other structural approaches combinedwith its ability to probe conformations unamenable to otherapproaches will greatly improve the ability to answer complexbiological questions.

However, despite the increase in the HDX-MS user base,and despite the numerous possible advantages to screeningsmall compounds with HDX-MS, HDX-MS is not yet routinelyused in small-molecule drug discovery. This may be due to thefact that many of the same questions that dogged HDX-MSdata 10 or 20 years ago still remain, and are exacerbated whenworking with small molecules. What is a ‘significant’ shift inexchange rate? Why do certain binding events counterintui-tively result in increases in amide exchange rates? How can wedistinguish between changes in amide exchange that are dueto direct ligand interactions, and the allosteric shifts in distantparts of protein structure? How can H/D exchange informationbe better integrated with high-resolution structural informa-tion? Can HDX-MS experiments be streamlined and standar-dized into a repeatable approach where multiple labsexamining the same sample will obtain the same result?

Many of these issues are surmountable, but they requiremore fundamental research. The application of the under-lying theoretical framework surrounding amide-exchangehas advanced little from Linderstrøm-Lang – we still workroutinely at peptide-level resolution and expect exchangerates to rigidly follow overly simplistic amide exchangemodels. Higher resolution mapping of amide exchangerates for more and more proteins combined with high-reso-lution structural information, may help bring some clarity tothese questions, with increasingly detailed models for thetheoretical basis of amide exchange can be rigorouslytested. Extensive work has been done in the use of a varietyof computational methods to predict amide exchange ratesof a given protein structure, and compare these to experi-mentally determined rates [132–134]. However, there arestill limitations for our understanding of why amide hydro-gens exchange with solvent at a given rate [30]. Furthercomputational and theoretical work combined with rigorousexperimental analysis will be essential to elucidate whyamides exchange at a given rate.

Broadly, many current commercial software analysispackages attempt to brute-force peptides into EX2 kineticprofiles and carry out all analysis based on mass centroids,largely failing to take into account these nuances. Othersoftware packages are engaging with more detailed spec-tral analysis however to discern deuterium incorporationpatterns across peptides [135,136], and further

EXPERT OPINION ON DRUG DISCOVERY 989

development in this approach will be useful to fully capi-talize on all of the information present in HDX massspectra.

HDX-MS has the potential to play a fundamental role in thefuture development of therapeutics, both bio-therapeuticsand small molecules, and it is particularly well suited to actas complementary technique to other biophysical and struc-tural techniques. The most likely place for hydrogen deuter-ium exchange to make a major impact is as a powerful toolalongside other structural techniques, as the combination ofprotein structure and dynamics will allow us to answer manyfundamental questions about how proteins function. Thetechnique’s strengths are numerous: an aqueous measure-ment that is highly sensitive, requiring only minor amountsof protein, utilizes only an isotope as a label, can operate inthe presence of lipids, small molecules, and nucleic acids, canuse semi-automated sample preparation and data collection,and has the potential for single amino acid resolution onmega-Dalton protein complexes. HDX-MS has far more poten-tial for exploitation in drug discovery, but further research intothe theoretical underpinnings of why amides exchange withsolvent, as well as further developments in the MS infrastruc-ture and data analysis will be essential for its continuedprogress.

Funding

JE Burke is supported by an open operating grant from the Institute ofInfection and Immunity (Grant No. FRN 142393) of the Canadian Institutesof Health Research and a discovery grant from the Natural Sciences andEngineering Research Council of Canada (NSERC-2014-05218). GR Massonis funded by the Medical Research Council (MC_U105184308) and StCatharine’s College, Cambridge.

Declaration of interest

The authors have no relevant affiliations or financial involvement with anyorganization or entity with a financial interest in or financial conflict withthe subject matter or materials discussed in the manuscript. This includesemployment, consultancies, honoraria, stock ownership or options, experttestimony, grants or patents received or pending, or royalties.

ORCID

Glenn R. Masson http://orcid.org/0000-0002-1386-4719John E. Burke http://orcid.org/0000-0001-7904-9859

References

Papers of special note have been highlighted as either of interest (•) or ofconsiderable interest (••) to readers.

1. Walters BT, Mayne L, Hinshaw JR, et al. Folding of a large protein at highstructural resolution. Proc Natl Acad Sci U S A. 2013 Nov;110:18898–18903.

2. Mysling S, Betzer C, Jensen PH, et al. Characterizing the dynamicsof α-synuclein oligomers using hydrogen/deuterium exchangemonitored by mass spectrometry. Biochem Am Chem Soc. 2013Dec;52:9097–9103.

3. Dautant A, Meyer P, Georgescauld F. Hydrogen/deuteriumexchange mass spectrometry reveals mechanistic details of activa-tion of nucleoside diphosphate kinases by oligomerization.Biochemistry. 2017 May;56:2886–2896. acs.biochem.7b00282.

4. AnandGS, LawD,Mandell JG, et al. Identification of the protein kinase Aregulatory RIalpha-catalytic subunit interface by amide H/2H exchangeand protein docking. Proc Natl Acad Sci USA National Acad Sciences.2003 Nov;100:13264–13269.

5. Graham BW, Tao Y, Dodge KL, et al. Probed by hydrogen-deuter-ium exchange (HDX) Fourier transform ion cyclotron resonancemass spectrometry confirm external binding sites on the minichro-mosomal maintenance (MCM) helicase. J Biol Chem. 2016Jun;291:12467–12480.

6. Mouchlis VD, Bucher D, McCammon JA, et al. Membranes serve asallosteric activators of phospholipase A2, enabling it to extract,bind, and hydrolyze phospholipid substrates. Proc Natl Acad SciU S A. 2015 Jan;112:E516-E525.

7. Man P, Montagner C, Vernier G, et al. Defining the interactingregions between apomyoglobin and lipid membrane by hydro-gen/deuterium exchange coupled to mass spectrometry. J MolBiol. 2007 Apr;368:464–472.

8. Burke JE, Hsu Y-H, Deems RA, et al. A phospholipid substratemolecule residing in the membrane surface mediates opening ofthe lid region in group IVA cytosolic phospholipase A2. J Biol ChemAm Soc Biochemistry Mol Biol. 2008 Nov;283:31227–31236.

9. Burke JE, Karbarz MJ, Deems RA, et al. Interaction of group IAphospholipase A2 with metal ions and phospholipid vesiclesprobed with deuterium exchange mass spectrometry. BiochemAm Chem Soc. 2008 Jun;47:6451–6459.

10. Pantazatos D, Kim JS, Klock HE, et al. Rapid refinement of crystal-lographic protein construct definition employing enhanced hydro-gen/deuterium exchange MS. Proc Natl Acad Sci U.S.A. 2004Jan;101:751–756.

•• This article provides an excellent summary of how HDX-MS canbe used for the design of crystallisation constructs through anoptimised generation of N and C terminal deletions.

11. McPhail JA, Ottosen EH, Jenkins ML, et al. The molecular basis ofAichi virus 3A protein activation of phosphatidylinositol 4 kinaseIIIβ, PI4KB, through ACBD3. Structure. 2017 Jan;25:121–131.

12. Sharma S, Zheng H, Huang YJ, et al. Construct optimization forprotein NMR structure analysis using amide hydrogen/deuteriumexchange mass spectrometry. Proteins. 2009 Sep;76:882–894.

13. Noble AJ, Zhang Q, O’Donnell J, et al. A pseudoatomic model ofthe COPII cage obtained from cryo-electron microscopy and massspectrometry. Nat Struct Mol Biol. 2013 Feb;20:167–173.

14. Skinner JJ, Lim WK, Bédard S, et al. Protein dynamics viewed byhydrogen exchange. Protein Sci. 2012 Jul;21:996–1005.

• This article provides in depth examination for why amidehydrogens in proteins exchange with solvent.

15. Skinner JJ, Lim WK, Bédard S, et al. Protein hydrogen exchange:testing current models. Protein Sci. 2012 Jul;21:987–995.

• This article provides in depth examination for why amidehydrogens in proteins exchange with solvent.

16. Balasubramaniam D, Komives EA. Hydrogen-exchange mass spec-trometry for the study of intrinsic disorder in proteins. BiochimBiophys Acta. 2013 Jun;1834:1202–1209.

17. Hvidt A, Linderstrøm-Lang K. Exchange of hydrogen atoms ininsulin with deuterium atoms in aqueous solutions. Biochimica EtBiophysica Acta. 1954;14:574–575.

18. Englander SW. Hydrogen exchange and mass spectrometry: a his-torical perspective. J Am Soc Mass Spectrom. 2006 Nov;17:1481–1489.

19. Deng B, Lento C, Wilson DJ. Hydrogen deuterium exchange massspectrometry in biopharmaceutical discovery and development - areview. Anal Chim Acta. 2016 Oct;940:8–20.

20. Saltzberg DJ, Broughton HB, Pellarin R, et al. Bayesian approachto quantitative interpretation of hydrogen deuterium exchangefrom mass spectrometry: application to characterizing protein-ligand interactions. J B Am Chem Soc. 2016 Nov;121:acs.jpcb.6b09358–3501.

21. Fauber BP, De Leon Boenig G, Burton B, et al. Structure-baseddesign of substituted hexafluoroisopropanol-arylsulfonamides asmodulators of RORc. Bioorg Med Chem Lett. 2013 Dec;23:6604–6609.

990 G. R. MASSON ET AL.

22. Fowler ML, McPhail JA, Jenkins ML, et al. Using hydrogen deuter-ium exchange mass spectrometry to engineer optimized constructsfor crystallization of protein complexes: case study of PI4KIIIβ withRab11. Protein Sci. 2016 Apr;25:826–839.

23. Burke JE, Inglis AJ, Perisic O, et al. Structures of PI4KIIIβ com-plexes show simultaneous recruitment of Rab11 and its effec-tors. Science. 2014 May;344:1035–1038.

• This article demonstrates the power of HDX-MS to generateconstructs for high resolution structural approaches, particu-larly for large multidomain, multiprotein complexes.

24. Houde D, Berkowitz SA, Engen JR. The utility of hydrogen/deuterium exchange mass spectrometry in biopharmaceuticalcomparability studies. J Pharm Sci. 2011 Jun;100:2071–2086.

25. Karpusas M, Nolte M, Benton CB, et al. The crystal structure ofhuman interferon beta at 2.2-A resolution. Proc Natl Acad Sci USANational Academy Sciences. 1997 Oct;94:11813–11818.

26. Li J, Wei H, Krystek SR, et al. Mapping the energetic epitope of anantibody/interleukin-23 interaction with hydrogen/deuteriumexchange, fast photochemical oxidation of proteins mass spectro-metry, and alanine shave mutagenesis. Anal Chem. AmericanChemical Society. 2017 Feb;89:2250–2258.

27. Beyer BM, Ingram R, Ramanathan L, et al. Crystal structures of thepro-inflammatory cytokine interleukin-23 and its complex with ahigh-affinity neutralizing antibody. J Mol Biol. 2008 Oct;382:942–955.

28. Bai Y, Milne JS, Mayne L, et al. Primary structure effects on peptidegroup hydrogen exchange. Proteins. 1993 Sep;17:75–86.

• This paper is an essential resource for all laboratories carryingout HDX-MS experiments, as it allows for the calculation of theintrinsic exchange rate for any amide hydrogen depending onconditions and primary sequence.

29. Weis DD, Wales TE, Engen JR, et al. Identification and characteriza-tion of EX1 kinetics in H/D exchange mass spectrometry by peakwidth analysis. J Am Soc Mass Spectrom. Springer-Verlag. 2006Nov;17:1498–1509.

30. McAllister RG, Konermann L. Challenges in the interpretation ofprotein h/d exchange data: a molecular dynamics simulation per-spective. Biochemistry. 2015 Apr;54:2683–2692.

• This is an excellent article that addresses some of the limita-tions that still exist in modeling why amide hydrogensexchange with solvent.

31. Persson F, Halle B. How amide hydrogens exchange in nativeproteins. Proc Natl Acad Sci U S A. 2015 Aug;112:10383–10388.

32. Gallagher ES, Hudgens JW. Mapping protein-ligand interactionswith proteolytic fragmentation, hydrogen/deuterium exchange-mass spectrometry. Methods Enzymol. 2016;566:357–404.

33. Vadas O, Jenkins ML, Dornan GL, et al. Using hydrogen-deuteriumexchange mass spectrometry to examine protein-membrane inter-actions. Meth Enzymol. Elsevier. 2017;583: 143–172.

34. Guttman M, Lee KK. Isotope labeling of biomolecules: structuralanalysis of viruses by HDX-MS. Meth Enzymol. 2016;566:405–426.

35. Woods VL, Hamuro Y. High resolution, high-throughput amidedeuterium exchange-mass spectrometry (DXMS) determinationof protein binding site structure and dynamics: utility in phar-maceutical design. J Cell Biochem Suppl. 2001;84:89–98.

36. Goswami D, Devarakonda S, Chalmers MJ, et al. Time windowexpansion for HDX analysis of an intrinsically disordered protein.J Am Soc Mass Spectrom. 2013 Oct;24:1584–1592.

37. Coales SJ, E SY, Lee JE, et al. Expansion of time window for massspectrometric measurement of amide hydrogen/deuterium exchangereactions. Rapid Commun Mass Spectrom. 2010 Dec;24:3585–3592.

38. Sheff JG, Hepburn M, Yu Y, et al. Nanospray HX-MS configuration forstructural interrogation of large protein systems. Analyst Royal SocChem. 2017 Mar;142:904–910.

39. Rist W, Rodriguez F, Tjd J, et al. Analysis of subsecond proteindynamics by amide hydrogen exchange and mass spectrometryusing a quenched-flow setup. Protein Sci. Cold Spring HarborLaboratory Press. 2005 Mar;14:626–632.

40. Simmons DA, Konermann L. Characterization of transient proteinfolding intermediates during myoglobin reconstitution by time-

resolved electrospray mass spectrometry with on-line isotopicpulse labeling. Biochemistry. 2002 Feb;41:1906–1914.

41. Rand KD, Pringle SD, Murphy JP, et al. Gas-phase hydrogen/deuteriumexchange in a traveling wave ion guide for the examination of proteinconformations. Anal Chem. American Chemical Society. 2009Dec;81:10019–10028.

42. Masson GR, Perisic O, Burke JE, et al. The intrinsically disordered tails ofPTEN and PTEN-L have distinct roles in regulating substrate specificityand membrane activity. Biochem J. 2016 Jan;473:135–144.

43. Kaltashov IA, Bobst CE, Abzalimov RR. H/D exchange and massspectrometry in the studies of protein conformation and dynamics:is there a need for a top-down approach? Anal Chem. 2009Oct;81:7892–7899.

44. Zhang H-M, McLoughlin SM, Frausto SD, et al. Simultaneous reduc-tion and digestion of proteins with disulfide bonds for hydrogen/deuterium exchange monitored by mass spectrometry. Anal Chem.2010 Feb;82:1450–1454.

45. Trabjerg E, Jakobsen RU, Mysling S, et al. Conformational analysis oflarge and highly disulfide-stabilized proteins by integrating onlineelectrochemical reduction into an optimized H/D exchange massspectrometry workflow. Anal Chem. American Chemical Society.2015 Sep;87:8880–8888.

46. Kadek A, Mrazek H, Halada P, et al. Aspartic protease nepenthesin-1as a tool for digestion in hydrogen/deuterium exchange massspectrometry. Anal Chem. 2014 May;86:4287–4294.

47. Mayne L, Kan Z-Y, Chetty PS, et al. Many overlapping peptides forprotein hydrogen exchange experiments by the fragment separa-tion-mass spectrometry method. J Am Soc Mass Spectrom. 2011Nov;22:1898–1905.

48. Rey M, Yang M, Burns KM, et al. Nepenthesin from monkey cups forhydrogen/deuterium exchange mass spectrometry. Mol CellProteomics. 2013 Feb;12:464–472.

49. Yang M, Hoeppner M, Rey M, et al. Recombinant nepenthesin II forhydrogen/deuterium exchange mass spectrometry. Anal Chem.American Chemical Society. 2015 Jul;87:6681–6687.

50. Jones LM, Zhang H, Vidavsky I, et al. Online, high-pressure digestionsystem for protein characterization by hydrogen/deuteriumexchangeand mass spectrometry. Anal Chem. 2010 Feb;82:1171–1174.

51. Ahn J, Jung MC, Wyndham K, et al. Pepsin immobilized on high-strength hybrid particles for continuous flow online digestion at10,000 psi. Anal Chem. 2012 Aug;84:7256–7262.

52. Wales TE, Fadgen KE, Gerhardt GC, et al. High-speed and high-resolution UPLC separation at zero degrees celsius. Anal Chem.2008 Sep;80:6815–6820.

53. Pan J, Zhang S, Parker CE, et al. Subzero temperature chroma-tography and top-down mass spectrometry for protein higher-order structure characterization: method validation and appli-cation to therapeutic antibodies. J Am Chem Soc. 2014Sep;136:13065–13071.

54. Venable JD, Okach L, Agarwalla S, et al. Subzero temperature chroma-tography for reduced back-exchange and improved dynamic range inamide hydrogen/deuterium exchange mass spectrometry. Anal ChemAm Chem Soc. 2012 Nov;84:9601–9608.

55. Walters BT, Ricciuti A, Mayne L, et al. Minimizing back exchangein the hydrogen exchange-mass spectrometry experiment. J AmSoc Mass Spectrom. 2012 Dec;23:2132–2139.

56. Majumdar R, Manikwar P, Hickey JM, et al. Minimizing carry-over in an online pepsin digestion system used for the H/Dexchange mass spectrometric analysis of an IgG1 monoclonalantibody. J Am Soc Mass Spectrom. 2012 Dec;23:2140–2148.

57. Fang J, Rand KD, Beuning PJ, et al. False EX1 signatures caused bysample carryover during HX MS analyses. Int J Mass Spectrom.2011 Apr;302:19–25.

58. Kan Z-Y, Walters BT, Mayne L, et al. Protein hydrogen exchange atresidue resolution by proteolytic fragmentation mass spectrometryanalysis. Proc Natl Acad Sci U S A. 2013 Oct;110:16438–16443.

59. Wales TE, Eggertson MJ, Engen JR. Considerations in the analysis ofhydrogen exchange mass spectrometry data. Methods Mol Biol.2013;1007:263–288.

EXPERT OPINION ON DRUG DISCOVERY 991

60. Landgraf RR, Chalmers MJ, Griffin PR. Automated hydrogen/deuter-ium exchange electron transfer dissociation high resolution massspectrometry measured at single-amide resolution. J Am Soc MassSpectrom. 2012 Feb;23:301–309.

61. Rand KD, Pringle SD, Morris M, et al. ETD in a traveling wave ionguide at tuned Z-spray ion source conditions allows for site-specifichydrogen/deuterium exchange measurements. J Am Soc MassSpectrom. 2011 Oct;22:1784–1793.

62. Rand KD, Zehl M, Jensen ON, et al. Protein hydrogen exchangemeasured at single-residue resolution by electron transfer dissocia-tion mass spectrometry. Anal Chem. 2009 Jul;81:5577–5584.

63. Pan J, Han J, Borchers CH, et al. Hydrogen/deuterium exchangemass spectrometry with top-down electron capture dissociation forcharacterizing structural transitions of a 17 kDa protein. J Am ChemSoc. 2009 Sep;131:12801–12808.

64. Abzalimov RR, Bobst CE, Kaltashov IA. A new approach to measur-ing protein backbone protection with high spatial resolution usingH/D exchange and electron capture dissociation. Anal Chem. 2013Oct;85:9173–9180.

65. Tjd J, Gårdsvoll H, Ploug M, et al. Intramolecular migration ofamide hydrogens in protonated peptides upon collisional acti-vation. J Am Chem Soc. 2005 Mar;127:2785–2793.

66. Demmers JAA, Rijkers DTS, Haverkamp J, et al. Factors affectinggas-phase deuterium scrambling in peptide ions and their implica-tions for protein structure determination. J Am Chem Soc. 2002Sep;124:11191–11198.

67. Rand KD, Jørgensen TJD. Development of a peptide probe for theoccurrence of hydrogen (1H/2H) scrambling upon gas-phase frag-mentation. Anal Chem. 2007 Nov;79:8686–8693.

68. Adhikary S, Deredge DJ, Nagarajan A, et al. Conformationaldynamics of a neurotransmitter: sodiumsymporter in a lipid bilayer.Proc Natl Acad Sci U S A. 2017 Mar;114:E1786–E1795.

69. Shukla AK, Westfield GH, Xiao K, et al. Visualization of arrestinrecruitment by a G-protein-coupled receptor. Nature. 2014Jun;512:218–222.

•• This is an outstanding paper showing the power of combiningCryo-electron microscopy with HDX-MS to examine challen-ging protein complexes of important therapeutic relevance.

70. Marcoux J, Wang SC, Politis A, et al. Mass spectrometry revealssynergistic effects of nucleotides, lipids, and drugs binding to amultidrug resistance efflux pump. Proc Natl Acad Sci U S A. 2013Jun;110:9704–9709.

71. Rey M, Forest E, Pelosi L. Exploring the conformational dynamics ofthe bovine ADP/ATP carrier in mitochondria. Biochemistry. 2012Dec;51:9727–9735.

72. Chung KY, Rasmussen SGF, Liu T, et al. Conformational changes inthe G protein Gs induced by the β2 adrenergic receptor. Nature.2011 Sep;477:611–615.

• This paper describes the first characterization of a GPCR het-erotrimeric complex using HDX-MS.

73. Dornan GL, Siempelkamp BD, Jenkins ML, et al. Conformationaldisruption of PI3Kδ regulation by immunodeficiency mutationsin PIK3CD and PIK3R1. Proc Natl Acad Sci U S A. 2017Feb;114:1982–1987. 201617244.

74. Srinivasan S, Dharmarajan V, Reed DK, et al. Identification andfunction of conformational dynamics in the multidomain GTPasedynamin. Embo J. 2016 Feb;35:443–457.

75. Rostislavleva K, Soler N, Ohashi Y, et al. Structure and flexibility ofthe endosomal Vps34 complex reveals the basis of its function onmembranes. Science. 2015 Oct;350:aac7365.

76. Vadas O, Burke JE. Probing the dynamic regulation of peripheralmembrane proteins using hydrogen deuterium exchange-MS(HDX-MS). Biochem Soc Trans. 2015 Oct;43:773–786.

77. Pirrone GF, Emert-Sedlak LA, Wales TE, et al. Membrane-associatedconformation of HIV-1 Nef investigated with hydrogen exchangemass spectrometry at a langmuir monolayer. Anal Chem. 2015Jul;87:7030–7035.

78. Vadas O, Dbouk HA, Shymanets A, et al. Molecular determinants ofPI3Kγ-mediated activation downstream of G-protein-coupled

receptors (GPCRs). Proc Natl Acad Sci USA National AcadSciences. 2013 Nov;110:18862–18867.

79. Dbouk HA, Vadas O, Shymanets A, et al. G protein-coupledreceptor-mediated activation of p110β by Gβγ is required forcellular transformation and invasiveness. Sci Signal. 2012 Dec;5:ra89.

80. Burke JE, Perisic O, Masson GR, et al. Oncogenic mutations mimicand enhance dynamic events in the natural activation of phosphoi-nositide 3-kinase p110α (PIK3CA). Proc Natl Acad Sci U S A. 2012Sep;109:15259–15264.

81. Burke JE, Vadas O, Berndt A, et al. Dynamics of the phosphoinosi-tide 3-kinase p110δ interaction with p85α and membranes revealsaspects of regulation distinct from p110α. Structure. 2011Aug;19:1127–1137.

82. Hsu Y-H, Burke JE, Li S, et al. Localizing the membrane bindingregion of Group VIA Ca2+-independent phospholipase A2 usingpeptide amide hydrogen/deuterium exchange mass spectrometry.J Biol Chem. 2009 Aug;284:23652–23661.

83. Siempelkamp BD, Rathinaswamy MK, Jenkins ML, et al. Molecularmechanism of activation of class IA phosphoinositide 3-kinases(PI3Ks) by membrane-localized HRas. J Biol Chem. AmericanSociety for Biochemistry and Molecular Biology. 2017 May. jbc.M117.789263;292:12256–12266.

84. Li S, Lee SY, Chung KY. Conformational analysis of g protein-coupled receptor signaling by hydrogen/deuterium exchangemass spectrometry. Meth Enzymol. Elsevier. 2015;557: 261–278.

85. Chandramohan A, Krishnamurthy S, Larsson A, et al., Predictingallosteric effects from orthosteric binding in Hsp90-ligand inter-actions: implications for fragment-based drug design. Liu J,editor. PLoS Comput Biol Public Library Science. 2016 Jun;12:e1004840.

•• This is an excellent article that describes the effectiveness ofHDX-MS as a screening platform for both low and high affinityinhibitors, and its effectiveness for identifying conformationalchanges at both orthosteric and allosteric binding sites.

86. Mondal T, Wang H, DeKoster GT, et al. ApoE: in vitro studies of asmall molecule effector. Biochemistry. American Chemical Society.2016 May;55:2613–2621.

87. Dickinson ER, Jurneczko E, Nicholson J, et al. The use of ionmobility mass spectrometry to probe modulation of the structureof p53 and of MDM2 by small molecule inhibitors. Front Mol Biosci.Frontiers. 2015;2: 39.

88. Chalmers MJ, Busby SA, Pascal BD, et al. Probing protein ligandinteractions by automated hydrogen/deuterium exchange massspectrometry. Anal Chem. 2006 Feb;78:1005–1014.

89. Konermann L, Rodriguez AD, Sowole MA. Type 1 and type 2scenarios in hydrogen exchange mass spectrometry studies onprotein-ligand complexes. Analyst. 2014 Dec;139:6078–6087.

•• This review highlights many of the complications in analysingHDX-MS data for protein small molecule interactions.

90. Sowole MA, Konermann L. Effects of protein-ligand interactions onhydrogen/deuterium exchange kinetics: canonical and noncanoni-cal scenarios. Anal Chem. 2014 Jul;86:6715–6722.

91. Masson GR, Maslen SL, Williams RL. Analysis of phosphoinositide 3-kinase inhibitors by bottom-up electron-transfer dissociationhydrogen/deuterium exchange mass spectrometry. Biochem J.2017 May;474:1867–1877.

92. Keppel TR, Weis DD. Mapping residual structure in intrinsically disor-dered proteins at residue resolution using millisecond hydrogen/deu-teriumexchange and residue averaging. J AmSocMass Spectrom. 2015Apr;26:547–554.

93. Babić D, Smith DM. Localization improvement of deuterium uptakein hydrogen/deuterium exchange in proteins. J Chemom. 2017Jan;31:e2876.

94. Walters BT. Empirical method to accurately determine peptide-averaged protection factors from hydrogen exchange MS data.Anal Chem. American Chemical Society. 2017 Jan;89:1049–1053.

95. Burns KM, ReyM, Baker CAH, et al. Platformdependencies in bottom-uphydrogen/deuterium exchange mass spectrometry. Mol Cell

992 G. R. MASSON ET AL.

Proteomics. American Society for Biochemistry and Molecular Biology.2013 Feb;12:539–548.

96. Deredge D, Li J, Johnson KA, et al. Hydrogen/deuterium exchangekinetics demonstrate long range allosteric effects of thumb site 2inhibitors of hepatitis C viral RNA-dependent RNA polymerase. JBiol Chem. American Society for Biochemistry and MolecularBiology. 2016 May;291:10078–10088.

97. Hughes TS, Giri PK, Ims DV, et al. An alternate binding site forPPARγ ligands. Nat Commun. 2014 Apr;5:3571.

98. Boerma LJ, Xia G, Qui C, et al. Defining the communicationbetween agonist and coactivator binding in the retinoid X receptorα ligand binding domain. J Biol Chem. 2014 Jan;289:814–826.

99. Srinivasan S, Nwachukwu JC, Bruno NE, et al. Full antagonism ofthe estrogen receptor without a prototypical ligand side chain. NatChem Biol. Nature Research. 2017 Jan;13:111–118.

100. Lee S, Wales TE, Escudero S, et al. Allosteric inhibition of antiapop-totic MCL-1. Nat Struct Mol Biol. 2016 Jun;23:600–607.

101. Espada A, Broughton H, Jones S, et al. A binding site on IL-17A forinhibitory macrocycles revealed by hydrogen/deuterium exchangemass spectrometry. J Med Chem. American Chemical Society. 2016Mar;59:2255–2260.

102. Canul-Tec JC, Assal R, Cirri E, et al. Structure and allosteric inhibitionof excitatory amino acid transporter 1. Nature. Nature Research.2017 Apr;544:446–451.

103. Zhang J, Adrián FJ, Jahnke W, et al. Targeting Bcr-Abl by combiningallosteric with ATP-binding-site inhibitors. Nature. NaturePublishing Group. 2010 Jan;463:501–506.

104. Iacob RE, Zhang J, Gray NS, et al., Allosteric interactions betweenthe myristate- and ATP-site of the Abl kinase. Rodrigues-Lima F,editor. PLoS ONE. 2011 Jan;6:e15929.

• This paper along with reference 95 describes the usefulness ofHDX-MS to probe the dynamic effects of allosteric inhibitorsbinding to the critical drug target Bcr-Abl.

105. Wales TE, Hochrein JM, Morgan CR, et al. Subtle dynamic changesaccompany Hck activation by HIV-1 Nef and are reversed by anantiretroviral kinase inhibitor. Biochemistry. 2015 Oct;54:6382–6391.

106. Hsu Y-H, Bucher D, Cao J, et al. Fluoroketone inhibition of Ca(2+)-inde-pendent phospholipase A2 through binding pocket associationdefined by hydrogen/deuterium exchange and molecular dynamics. JAm Chem Soc. 2013 Jan;135:1330–1337.

107. Cao J, Burke JE, Dennis EA. Using hydrogen/deuterium exchange massspectrometry to define the specific interactions of the phospholipaseA2 superfamily with lipid substrates, inhibitors, and membranes. J BiolChem. 2013 Jan;288:1806–1813.

108. Burke JE, Babakhani A, Gorfe AA, et al. Location of inhibitors bound togroup IVA phospholipase A2 determined by molecular dynamics anddeuterium exchange mass spectrometry. J Am Chem Soc. 2009Jun;131:8083–8091.

109. Karageorgos I, Wales TE, Janero DR, et al. Active-site inhibitorsmodulate the dynamic properties of human monoacylglycerollipase: a hydrogen exchange mass spectrometry study.Biochemistry. 2013 Jul;52:5016–5026.

110. Clarke JH, Giudici ML, Burke JE, et al. The function of phosphatidylino-sitol 5-phosphate 4-kinase γ (PI5P4Kγ) explored using a specific inhibi-tor that targets the PI5P-binding site. Biochem J. 2015Mar;466:359–367.

111. Berkowitz SA, Engen JR, Mazzeo JR, et al. Analytical tools forcharacterizing biopharmaceuticals and the implications for biosi-milars. Nat Rev Drug Discov. 2012 Jun;11:527–540.

• This review is an excellent summary of the challenges in char-acterizing bio-therapeutics and outlines the role HDX-MS canplay in bio-therapeutic drug development.

112. Houde D, Arndt J, Domeier W, et al. Characterization of IgG1conformation and conformational dynamics by hydrogen/deuter-ium exchange mass spectrometry. Anal Chem. American ChemicalSociety. 2009 Apr;81:2644–2651.

113. Houde D, Peng Y, Berkowitz SA, et al. Post-translational modifica-tions differentially affect IgG1 conformation and receptor binding.Mol Cell Proteomics. 2010 Aug;9:1716–1728.

114. Guttman M, Scian M, Lee KK. Tracking hydrogen/deuteriumexchange at glycan sites in glycoproteins by mass spectrometry.Anal Chem. American Chemical Society. 2011 Oct;83:7492–7499.

115. Jensen PF, Comamala G, Trelle MB, et al. Removal of N-linked glycosyla-tions at acidic pH by PNGase A facilitates hydrogen/deuteriumexchange mass spectrometry analysis of N-linked glycoproteins. AnalChem. American Chemical Society. 2016 Dec;88:12479–12488.

116. Wei H, Ahn J, Yu YQ, et al. Using hydrogen/deuterium exchangemass spectrometry to study conformational changes in granulo-cyte colony stimulating factor upon PEGylation. J Am Soc MassSpectrom. 2012 Mar;23:498–504.

117. Adams R, Burnley RJ, Valenzano CR, et al. Discovery of a junctionalepitope antibody that stabilizes IL-6 and gp80 protein:proteininteraction and modulates its downstream signaling. Sci Rep.Nature Publishing Group. 2017 Jan;7:37716.

• This article demonstrates the power of HDX-MS to probe pro-tein–antibody interactions, particularly for antibodies thatbind to multiple epitopes on different proteins

118. Iversen R, Mysling S, Hnida K, et al. Activity-regulating structuralchanges and autoantibody epitopes in transglutaminase 2assessed by hydrogen/deuterium exchange. Proc Natl Acad SciUSA National Acad Sciences. 2014 Dec;111:17146–17151.

119. Zhang Q, Willison LN, Tripathi P, et al. Epitope mapping of a 95 kDaantigen in complex with antibody by solution-phase amide back-bone hydrogen/deuterium exchange monitored by Fourier trans-form ion cyclotron resonance mass spectrometry. Anal Chem.American Chemical Society. 2011 Aug;83:7129–7136.

120. Kim M, Sun Z-YJ, Rand KD, et al. Antibody mechanics on a mem-brane-bound HIV segment essential for GP41-targeted viral neu-tralization. Nat Struct Mol Biol. 2011 Nov;18:1235–1243.

•• This article highlights one of the major advantages of theHDX-MS techniques, as protein–protein interactions, in thiscase an antibody bound to its target membrane protein, canbe interrogated in its native membrane environment.

121. Kong L, Lee DE, Kadam RU, et al. Structural flexibility at a major con-served antibody target on hepatitis C virus E2 antigen. Proc Natl AcadSci USA National Acad Sciences. 2016 Oct;113:12768–12773.

122. Liang Y, Guttman M, Williams JA, et al. Changes in structure andantigenicity of HIV-1 Env trimers resulting from removal of a con-served CD4 binding site-proximal glycan. Silvestri G, editor. J Virol.American Society for Microbiology. 2016 Oct;90:9224–9236.

123. Gribenko AV, Parris K, Mosyak L, et al. High resolution mappingof bactericidal monoclonal antibody binding epitopes onStaphylococcus aureus antigen MntC. Zhang G, editor. PLoSPathog. 2016 Sep;12:e1005908

124. Chen E, Salinas ND, Huang Y, et al. Broadly neutralizing epitopes inthe Plasmodium vivax vaccine candidate Duffy Binding Protein.Proc Natl Acad Sci U S A. 2016 May;113:6277–6282.

125. Ciferri C, Chandramouli S, Leitner A, et al. Antigenic characterization ofthe HCMV gH/gL/gO and pentamer cell entry complexes reveals bind-ing sites for potently neutralizing human antibodies. Heldwein EE,editor. PLoS Pathog. Public Library of Science. 2015 Oct;11:e1005230.

126. Guttman M, Cupo A, Julien J-P, et al. Antibody potency relatesto the ability to recognize the closed, pre-fusion form of HIVEnv. Nat Commun. Nature Publishing Group. 2015 Feb;6:6144.

127. Malito E, Faleri A, Lo P S, et al. Defining a protective epitope onfactor H binding protein, a key meningococcal virulence factor andvaccine antigen. Proc Natl Acad Sci USA National Acad Sciences.2013 Feb;110:3304–3309.

128. Spraggon G, Pantazatos D, Klock HE, et al. On the use of DXMS toproduce more crystallizable proteins: structures of the T. maritimaproteins TM0160 and TM1171. Protein Sci. 2004 Dec;13:3187–3199.

129. Rutaganira FU, Fowler ML, McPhail JA, et al. Design and structuralcharacterization of potent and selective inhibitors of phosphatidy-linositol 4 kinase IIIβ. J Med Chem. 2016 Mar;59:1830–1839.

130. Dai H, Case AW, Riera TV, et al. Crystallographic structure of a smallmolecule SIRT1 activator-enzyme complex. Nat Commun. 2015Jul;6:7645.

EXPERT OPINION ON DRUG DISCOVERY 993

131. Khan AG,Whidby J, Miller MT, et al. Structure of the core ectodomain ofthe hepatitis C virus envelope glycoprotein 2. Nature. 2014May;509:381–384.

132. Liu T, Pantazatos D, Li S, et al. Quantitative assessment ofprotein structural models by comparison of H/D exchange MSdata with exchange behavior accurately predicted byDXCOREX. J Am Soc Mass Spectrom. Springer-Verlag. 2012Jan;23:43–56.

133. Park I-H, Venable JD, Steckler C, et al. Estimation of hydrogen-exchange protection factors from MD simulation based on amide

hydrogen bonding analysis. J Chem Inf Model. 2015 Sep;55:1914–1925.

134. Craig PO, Lätzer J, Weinkam P, et al. Prediction of native-statehydrogen exchange from perfectly funneled energy landscapes.J Am Chem Soc. 2011 Nov;133:17463–17472.

135. Rey M, Sarpe V, Burns KM, et al. Mass spec studio for integrativestructural biology. Structure. 2014 Oct;22:1538–1548.

136. Pascal BD, Willis S, Lauer JL, et al. HDX workbench: software for theanalysis of H/D exchange MS data. J Am Soc Mass Spectrom. 2012Sep;23:1512–1521.

994 G. R. MASSON ET AL.