Crossing the phase boundary to study protein dynamics and function: combination of amide hydrogen exchange in solution and ion fragmentation in the gas phase

Abstract

Protein dynamics are the key to understanding their behavior. The static protein structure alone in most cases is insufficient to describe the vast array of complex functions they perform in vivo. Until recently there were relatively few techniques available to investigate the dynamic nature of these proteins. Mass spectrometry has recently emerged as a powerful biophysical method, capable of providing both structural and dynamic information. By utilizing the labile nature of amide hydrogens as a marker of the backbone dynamics in solution, combined with gas-phase dissociation techniques, we now have a high-resolution tool to locate these exchanging hydrogens within the sequence of the protein and to probe the functional importance of its structural elements. In this paper we describe several applications of these methodologies to illustrate the importance of dynamics to the biological functions of proteins. Copyright © 2002 John Wiley & Sons, Ltd.

INTRODUCTION: MASS SPECTROMETRY IN STRUCTURAL BIOLOGY

The last decade has witnessed explosive growth in both the number and scope of mass spectrometric approaches to various problems in biological sciences. Currently, mass spectrometry serves as a major catalyst in proteomics, one of the most visible and rapidly developing areas of modern biology. Unsurpassed sensitivity combined with an impressive (and ever expanding) arsenal of experimental techniques for structure elucidation make mass spectrometry a major player in the field. However, even complete characterization of the proteome will not necessarily be sufficient to answer the most intriguing question of modern biology, i.e. what makes this vast and very diverse collection of biopolymers a living organism.

Understanding the sophisticated mechanisms of this highly orchestrated macromolecular machinery will most certainly require detailed knowledge not only of the structures but also of the dynamic behavior of its constituents. At present, high-resolution NMR remains the only technique capable of providing structural information at a residue level as well as probing dynamics on time-scales ranging from sub-nanoseconds to days. Its disadvantages, however, include molecular mass limitations, problems with analyzing complexes containing paramagnetic ligands, difficulties with correlating site-specific dynamics with distinct conformers, etc. Most other biophysical techniques usually monitor either cumulative changes of the secondary structure (e.g. circular dichroism spectroscopy in the far-UV region, Fourier transform IR and Raman spectroscopy) or specific markers of dynamic events (e.g. fluorescence measurements to assess exposure of aromatic residues to solvent).

Mass spectrometry has become an increasingly popular technique to study protein dynamics, as it provides information often unavailable from other sources. Mass spectrometry is unique in that it is capable of detecting and characterizing individual conformational states that may co-exist in solution at equilibrium. Owing to its superior sensitivity, mass spectrometry, unlike NMR, uses only modest amounts of protein for analysis, which in many cases permits the study of protein behavior at, or even below, endogenous levels.

‘TRADITIONAL’ APPROACHES TO PROBING PROTEIN DYNAMICS WITH HYDROGEN EXCHANGE AND MASS SPECTROMETRY

The contributions of mass spectrometry towards the characterization and understanding protein dynamics and function have been particularly visible in the field of amide hydrogen exchange (HDX). The superior sensitivity and mass accuracy offered by mass spectrometry seem ideally suited for monitoring HDX reactions, as each individual ²H(D)¹H exchange reaction results in a 1.0063 u mass change. The pioneering work of Katta and Chait first demonstrated the great potential of the combination of HDX with electrospray ionization mass spectrometry (ESI-MS) as a tool to probe the conformational dynamics of small proteins.1 In the years since, the number and scope of the applications of HDX/ESI-MS (and, more recently, HDX combined with matrix-assisted laser desorption/ionization mass spectrometry (MALDI-MS)) methodology to probe both the architecture and dynamics of biomolecules has continued to expand at an increasing rate, catalyzed by continuous technological improvements in ‘soft’ ionization methods. One of the most significant advantages offered by ESI-MS is its ability to desorb biomolecular species directly from aqueous solutions, thus permitting ‘on-line’ HDX/MS measurements. Likewise, the ability of mass spectrometry to handle larger proteins and their complexes is particularly important when compared with high-field NMR, which still has limited application for proteins larger than ∼30 kDa. The practical upper mass limit of ESI-MS, on the other hand, has yet to be established, as the bar is being continuously raised following recent reports on the observation of intact antibody–antigen complexes,2 intact viral particles3 and even intact ribosomes.4 Finally, HDX/MS can be used to detect and characterize distinct protein intermediate states that can be either transiently populated during protein folding (kinetic intermediates) or induced by altering the protein's environment (equilibrium intermediates). The following brief discussion focuses on this unique capability of HDX/MS experiments.

Detection of distinct protein conformers by HDX/MS

Protein amide HDX involves two types of reaction: (i) reversible protein unfolding that disrupts the H-bonding network, thus exposing amides to ambient solvent, and (ii) isotope exchange at individual unprotected amides (Fig. 1). Therefore, the overall protein HDX kinetics are determined by both reactions. Two extreme situations, commonly referred to as EX₁ and EX₂ exchange mechanisms, are usually considered.5 If the protein refolding rate k₋₁ is low compared with the intrinsic exchange rate k₂, then all protons in the unfolded segment will exchange with the solvent, and the overall exchange follows pseudo-first-order kinetics. In this case, unfolding of the protein is a rate-limiting step, and the apparent (measured) exchange rate constant is k^HDX = k₁. Any unfolding event under these conditions will result in complete amide exchange within the affected region, an exchange regime commonly referred to as EX₁. If, however, protein refolding is much faster than the intrinsic exchange, the overall exchange will follow more complex kinetics (i.e. k^HDX = (k₁k₂)/k₋₁), a situation known as the EX₂ mechanism.

Figure 1. Schematic representation of protein amide HDX: EX₁ mechanism (A) and EX₂ mechanism (B). Deuterated segments of the protein are shown in light gray and the protiated segments are shown in black.

Protein amide HDX under native conditions almost always proceeds via an EX₂ mechanism.5 On the other hand, the EX₁ regime becomes favored when significant amounts of chaotropic agents are present in the system. HDX measurements carried out under such conditions provide the means to map and characterize distinct protein conformers, since any unfolding event will result in marking that region of the protein molecule with a characteristic ‘isotopic label.’ Mass spectrometry is unique in its ability to read such labels for the entire protein population (by measuring the mass difference), thus providing a means to detect distinct protein states differing by their exposure to the solvent.6 An example of such behavior is presented in Fig. 2(right panel). Mild acidification of the protein solution gives rise to transient partially unfolded states. Once a protein molecule samples such states, all amides in the destabilized segment undergo isotope exchange (i.e. ²H are substituted with ¹H), leading to a mass decrease which is equal to the number of amide groups exchanged in the destabilized segment. Obviously the protein molecule retains its isotopic label after returning to the ground (native) state. As a result, the entire population will become isotopically labeled after a certain period of time (which can be used to estimate k₁). ¹H NMR, on the other hand, generates residue-specific exchange data averaged across the entire ensemble of states, thus complicating detection of distinct conformations. If, however, HDX follows the EX₂ mechanism, both ESI-MS and NMR detection will produce a picture of protein dynamics averaged across the entire population (Fig. 2(left panel)). ¹H NMR has the advantage of providing dynamics data at the residue level, but in many cases mass spectrometry still may be the technique of choice, particularly if the protein system is too large (>30 kDa) or contains high-spin ligands, resulting in line broadening and/or severe resonance overlap.

Figure 2. Global backbone dynamics of CRABP I probed with HDX/ESI-MS under conditions favoring EX₁ exchange mechanism (right) and EX₂ mechanism (left). In each experiment a 100 µM solution of a fully deuterated protein in D₂O was diluted 1:50 (v/v) in exchange buffer solution (H₂O/CH₃CO₂NH₄, pH adjusted to 3.2 and 7.0 for experiments shown on the right and left panels, respectively). Na and K adducts are indicated with asterisks.

ESI-MS provides a simple and efficient way to monitor slow exchange kinetics in real time (k^HDX < 1 min⁻¹), as the solution of protein undergoing isotope exchange can be continuously infused to the ESI source. It is also possible to use HDX/MS to study significantly faster exchange kinetics using a quenched flow apparatus and operating the ion source under slow-exchange conditions (i.e. pH 2–3, 0° C).7, 8 Although this scheme was initially implemented using a fast atom bombardment (FAB) source, the use of ESI and MALDI9 ionization sources has become almost universal in the past several years.

Site-specific measurements of amide protection using enzymatic reactions in solution

The mass spectrometry-based experimental strategies outlined above characterize only the ‘global’ behavior of the proteins and polypeptides, as they provide information on the overall bulk exchange pattern, thus falling behind NMR methodologies in terms of the local structural detail obtainable. Such local information is necessary for producing a detailed picture of dynamic behavior of individual structural segments within a protein. Zhang and Smith resolved this difficulty by introducing a proteolytic step prior to mass analysis.10 Local exchange details can be maintained if the proteolytic degradation is performed under slow-exchange conditions (e.g. at low solution temperature and pH 2–3, when k₂ is minimal). Limitations of this method, however, include the necessary correction for loss of isotope information due to back-exchange during proteolysis and separation of the fragments prior to mass analysis, but this does not usually exceed 10% and may be accounted for by introducing a ‘back exchange’ correction factor.11 The spatial resolution offered by these measurements is usually limited only by the number of peptic fragments obtained. In general, a large number of fragments, particularly overlapping ones, would lead to greater spatial resolution, and hence more precise localization of the structural regions which have undergone exchange. As has been shown recently, this resolution can be improved further by introducing a gas-phase fragmentation step (collision-induced dissociation (CID)) during mass analysis.12

Site-specific measurements of amide protection using ion dissociation in the gas phase: initial experiments

An alternative method to probe amide HDX kinetics locally in smaller polypeptides with ESI-MS was introduced in 1994 by Anderegg et al.13 This method relies on the ability of mass spectrometers to produce a wealth of structural information in tandem (MS/MS or MSⁿ) experiments. In this scheme, proteolytic degradation of the polypeptide in solution is replaced by fragmentation in the gas phase using CID (Fig. 3). As a result, site-specific information on the peptide backbone dynamics in solution is obtained via a combination of solution-phase chemistry (HDX) and gas-phase ion–molecule reactions. This methodology was applied initially to study dynamics of rather small systems such as melittin (2.8 kDa) and a growth hormone releasing factor analog (3.7 kDa).13 More recent applications include the 23-residue N-terminal domain of HIV-1 glycoprotein gp41,14 model fibril-forming peptides15 and model trans-membrane peptides.16, 17

Figure 3. Schematic representation of an HDX/ESI-CID-MS experiment to probe the local deuterium content of a peptide undergoing amide hydrogen exchange in solution. In the absence of scrambling, the extent of exchange in solution is retained in individual fragments after dissociation.

These experimental schemes are very attractive, as they combine the simplicity of ‘on-line’ HDX/ESI-MS experiments, and yet produce detailed information on the amide protection at the residue level, whilst avoiding the need for ‘back exchange’ corrections. Extensive fragmentation of the polypeptide ions in the gas phase often allows the protection of every backbone amide to be measured; this advantage, however, diminishes dramatically as the polypeptide size increases. The practical upper mass limit of a polypeptide for which abundant structurally diagnostic CID fragment ions can be produced using most types of mass analyzer is about 5 kDa. Such a severe mass limitation (much less forgiving even than that imposed by ¹H NMR spectroscopy) has inevitably narrowed the scope of systems to which this new methodology has been applied. Short polypeptide systems also usually exhibit highly dynamic behavior even under native conditions. Therefore, the initial ‘on-line’ HDX/ESI-CID-MS experiments were typically performed under conditions strongly favoring secondary structure formation (e.g. in the presence of alcohols or using strongly amphiphilic systems).

Protein ion dissociation in the gas phase: use of FTICR-MS

In recent years we have witnessed spectacular progress in Fourier transform ion cyclotron resonance (FTICR) technology. Perhaps the most impressive gains have been made in the field of protein analysis. Following the interfacing of ESI and MALDI sources to FTICR analyzers and the development of several fragmentation techniques that target large macromolecular ions (proteins in particular), FTICR-MS has now become a practical tool to study protein structure, displaying an impressive array of unique analytical capabilities. Perhaps the most important ones within the context of the subject of this paper are unsurpassed mass resolution provided by FTICR analyzers and an expanded arsenal of ion fragmentation methods. Application of infrared multi-photon ionization (IRMPD)18 and electron-capture dissociation (ECD)19 for polypeptide and protein sequencing has greatly expanded the range of biopolymers for which structurally diagnostic fragment ions can be produced. In addition, even ‘classical’ ion fragmentation techniques (e.g. CID) result in significantly higher yields of fragment ions (compared with other types of mass analyzers) owing to the longer ion trapping times afforded by FTICR-MS. Furthermore, the versatility of FTICR-MS easily allows several ion activation techniques to be combined in one experiment, further increasing fragmentation yields and sequence coverage.20, 21 The interpretation of such fragmentation spectra can pose a significant challenge, however, owing to a plethora of possible fragment ion peaks, often overlapping within a relatively narrow m/z range (typically <2500 u). This task is greatly simplified when FTICR is used for mass analysis, as its superior resolving power provides a means to resolve not only the isotopic clusters of overlapping peaks of ions differing in their charge states, but also in many cases peaks corresponding to isobaric ions even in the broadband mode.22

These advances have had a major impact on the development of the HDX/ESI-CID-MS methodology. The protein ion fragmentation efficiency offered by commercial ICR-MS is sufficiently high to have allowed the extensive characterization of local dynamic events within the 18 kDa pseudo-wild type (wt*) cellular retinoic acid binding protein I (CRABP I) under mildly acidic conditions23, 24 and an 8 kDa polypeptide ubiquitin (Ub) under native conditions.25 Protein ion fragmentation in both cases was carried out directly in the ESI source using nozzle–skimmer collisional activation, i.e. without prior mass selection. This approach (fragmentation of the entire protein ion population) has also been adopted in other work, since it maximizes the yield of fragment ions without having any detrimental effect on the measurements. The mass evolution of each protein fragment ion provides a measure of the local deuterium content of the protein as a function of the exchange time in solution and, therefore, its backbone dynamics.

An example of such an experiment is shown in Fig. 4, where HDX/ESI-CID-MS was used to study the backbone dynamics of wt*-CRABP I, a 157 residue protein. Its sequence differs from that of the wild-type CRABP I in that it includes an N-terminal His-tag (22 residues long) and a point mutation R131Q. The protein consists of a 10-stranded β-sheet with a short helix–turn–helix motif inserted between the first and second β-strands.23 Unlike most proteins, it does not contain a tightly packed central hydrophobic core, but instead the β-barrel structure surrounds a cavity, which creates the binding site for the physiological ligand all-trans-retinoic acid (RA). The N-terminal His-tag is completely unstructured in solution, as judged by NMR. The protein exhibits a very complex global exchange pattern under acidic pH and low salt conditions, suggesting the presence of segments that follow an EX₁ exchange behavior, whereas others follow an EX₂ mechanism (Fig. 4(A)).

Figure 4. Backbone dynamics of wt*-CRABP I in acidic (pH 3) low salt conditions probed with HDX/ESI-CID-MS. Evolution of the isotope distribution within the intact protein ion (A) and fragment ions derived from the His-tag region (B), helix–turn–helix motif (C) and strand–turn–strand motif (D) as a function of the exchange time in solution. Sample work-up is the same as in Fig. 2. Adapted with permission from Eyles et al.24 Protein conformational stability probed by Fourier transform ion cyclotron resonance mass spectrometry. J. Am. Chem. Soc., 2000; 122: 495–500 Copyright 2000 American Chemical Society.

Measuring the deuterium content of CID fragment ions as a function of exchange time in solution provided a means to identify various regions within the protein that exhibit markedly different exchange kinetics (Fig. 4(B)–(D)). As expected, all fragment ions derived from the His-tag region and the adjacent four residues [Pro²² → Ala²⁵] become fully protiated following 2 min of exposure of the protein to exchange solution. A very different dynamic behavior, however, is exhibited by an N-terminal protein segment, comprising the first β-strand and the helix–turn–helix structural motif. Local HDX measurements within this segment indicate significant backbone protection and cooperative loss of structure upon local unfolding, as represented by bimodal intensity distributions within the isotopic clusters of b₃₈⁴⁺ and b₃₉⁴⁺. Thus, backbone dynamics within this segment clearly contribute to the partial EX₁ character of the global HDX pattern. This behavior is in sharp contrast with the dynamics of a C-terminal strand–turn–strand segment of the protein (as represented by y₁₁⁺) which apparently contributes to the partial EX₂ character of the global amide exchange. Although a detailed description of the local dynamics is beyond the scope of this paper, it is clear that a very complex and convoluted global HDX pattern can be successfully analyzed using the combined HDX/ESI-CID-MS technique.

The experiments discussed above were carried out on-line, i.e. by continuously infusing the protein solution from the exchange reactor to the ESI source. Off-line experiments (i.e. taking the aliquots from the exchange reactor after certain periods of time and quenching the exchange, followed by ESI-CID-MS analysis) present another choice.25, 26 This might be a preferred method when a longer data acquisition time is needed to improve the signal-to-noise ratio, e.g. in a situation when the protein ion fragmentation efficiency is low.

Protein ion fragmentation, proton mobility and hydrogen scrambling in the gas phase

A major concern in using protein ion fragmentation to measure local deuterium content is the possibility of hydrogen scrambling within the activated ions prior to fragmentation. Collisional activation of protein ions involves multiple collisions with molecules of nebulizing gas (nitrogen) and evaporated solvent. Furthermore, the average lifetime of activated ions inside the hexapole or r.f.-only quadrupole ion guide (or inside the ICR cell) may exceed tens or hundreds of milliseconds. Such metastable ions may potentially exhibit significant intra- and intermolecular H–D exchange (hydrogen scrambling) in the gas phase prior to dissociation. Furthermore, one of the widely accepted models of peptide ion fragmentation in the gas phase invokes the notion of the ‘mobile proton’ as a driving force of the dissociation process.27 Any significant redistribution of the protein ion isotope content would obviously compromise the quality of the HDX/ESI-CID-MS measurements and, in extreme cases, may render them useless.

A recent report by McLafferty et al. suggests that significant hydrogen scrambling takes place within the ions of cytochrome c activated by sustained off-resonance irradiation (SORI) in the ICR cell.28 On the other hand, the groups of Anderegg13 and Waring14 reported that scrambling was minimal for short helical peptides under typical CID conditions in the collision cell of a triple-quadrupole mass spectrometer. More recently, Kraus et al. reported on the use of ESI-MS/MS (ion trap) to monitor the deuterium content of short peptides in their studies of amyloid fibril formation in solution, and the results appeared to be unaffected by scrambling.15 Similarly, Demmers and co-workers saw no evidence of hydrogen scrambling upon fragmentation of trans-membrane peptides in the quadrupole collision cell of a hybrid (Q-TOF) mass spectrometer.16, 17 Deng et al. used low-energy CID in a quadrupole ion trap to probe deuterium content of peptic fragments of cytochrome c under slow exchange (quench) conditions.12 A careful comparison of the CID-MS data with those provided by ¹H NMR suggested that hydrogen scrambling does not affect the b-type fragment ions, but may introduce some alterations in the deuterium content of the y-type ions, a notion supported recently by another group.29

Such a broad range of observations clearly suggests that the extent of hydrogen scrambling depends on several experimental variables. In our opinion, it is the ion activation energy that largely determines the extent of scrambling. A substantial increase in the internal energy of peptide ions in the gas phase is known to decrease H–D exchange rates significantly,30 and the same would be expected to hold true also for intramolecular exchange. Thus, if the peptide ion is activated rapidly as a result of relatively few high-energy collisions, then the extent of hydrogen scrambling occurring prior to dissociation would be minimal. If, however, the ion activation proceeds slowly, one would expect that a significant amount of intramolecular rearrangement processes, including hydrogen scrambling, could occur prior to reaching a level of internal energy at which peptide bond cleavage becomes a favorable process (Fig. 5).

Figure 5. Hypothetical plot of ion dissociation constant as a function of its internal energy. Adapted with permission from Busch et al.45 Mass spectrometry/mass spectrometry: techniques and applications of tandem mass spectrometry, Copyright 1988 John Wiley & Sons Inc.

Therefore, very slow heating processes, such as SORI,31 would be expected to result in extensive hydrogen scrambling prior to fragmentation, in line with recent observations.28 Most other mass analyzers, however, employ significantly higher activation energy.31 Limited scrambling associated with the formation of y-ions (but not for b-ions)12, 29 in quadrupole ion traps most likely reflects the differences in the mechanisms of their formation.

Activation of larger protein ions in the ESI interface region of an FTICR mass spectrometer involves multiple collisions that may also lead to hydrogen exchange in the gas phase, both intermolecular (with evaporated solvent) and internal (hydrogen scrambling). Data obtained from exchange in wt*-CRABP I provide a convincing control to verify that such exchange is minimal under these conditions. If such a phenomenon had occurred in the gas phase, then one would expect significant incorporation of deuterons into the ‘fully protiated’ unstructured His-tag from elsewhere in the protein molecule prior to dissociation. This internal exchange has not been observed (Fig. 4(B)), as all fragments derived from the His-tag region of the protein remain fully protiated from the earliest time point. Likewise, fragments incorporating other more structured regions of the protein retain a significant deuterium content, and this proportion changes consistently over the time course of the experiment. These observations indicate that hydrogen scrambling in the gas phase is too slow to affect the measurements under the conditions employed.

Although the His-tag provides a convenient reporter of hydrogen scrambling, or a lack thereof, it is not always practical to require that it be part of a protein sequence. It is possible, however, in many cases to rule out the occurrence of scrambling based solely on the evolution of deuterium content of fragment ions that are part of a wild-type sequence. This task becomes particularly easy if the global exchange follows the EX₁ mechanism. Figure 6 illustrates the time evolution of isotopic clusters of b₁₆⁺ and y₁₆⁺ ions, which are ‘in-source’ CID products of a 15.5 kDa protein wt-CRABP I. The protein underwent HDX under low salt conditions at pH 3.0 in the presence of a mixture of retinoic acid stereoisomers, one of which (all-trans) was the natural ligand of CRABP I. The protein segment represented by b₁₆⁺ contains no ligand binding residues, and exhibits little backbone protection under these mildly denaturing conditions. This behavior contrasts sharply with y₁₆⁺, which contains two residues forming a salt bridge to retinoic acid (RA) in the protein–ligand complex.32 The presence of the ligand in solution induces significant stabilization within this region and hence a marked increase in amide protection. Importantly, although both b₁₆⁺ and y₁₆⁺ fragment ions exhibit EX₁ exchange behavior, the kinetics are quite different. If intramolecular hydrogen scrambling were to occur under these conditions, one would expect to see smeared isotopic profiles (EX₂-like) and more or less uniform exchange in all fragments. A very noticeable distinction between b₁₆⁺ and y₁₆⁺ indicates that hydrogen scrambling is negligible under these conditions.

Figure 6. Evolution of the isotopic content of b₁₆⁺ and y₁₆⁺ ions of a 15.5 kDa protein wt-CRABP I. The protein underwent HDX under low salt acidic conditions (pH 3.0) in the presence of a mixture of retinoic acid isomers. Protein sample work-up is as described in Fig. 2.

Further challenges: increasing the number of fragment ions to enhance the spatial resolution provided by HDX/ESI-CID-MS experiments

Local backbone dynamics can be mapped with high spatial resolution using the HDX/ESI-CID-MS methodology only within those protein segments whose sequence coverage with fragment ions is complete or near complete. An example of this is presented in Fig. 7. Collisional activation of ions produced upon desorption of a 37 kDa N-lobe of human serum transferrin (hTf/2N)33 leads to the appearance of a series of abundant fragment ions derived from cleavages of every amide bond within a relatively small segment of the protein (Ala⁵⁴ → Val⁸⁰). Monitoring the mass evolution of these fragment ions as a function of HDX time in solution (data not shown) provides a means to measure local dynamics with highest spatial resolution (i.e. protection of every amide within this segment can be determined). However, fragment ions representing the rest of the hTf/2N sequence are not abundant in the spectrum, thus making it impossible to analyze local backbone dynamics elsewhere in the protein. The remarkable selectivity in the formation of fragment ions within such a narrow segment of hTf/2N has a rather trivial explanation. This segment connects two sub-domains of the protein that have high degree of internal disulfide bonding. Even if amide bond cleavage does occur within these segments (which is very likely), the resulting complementary b- and y-ions will not physically separate, being kept together by the disulfide bridges.

Figure 7. Fragment ion spectrum of a 37 kDa N-lobe of human serum transferrin (10 µM protein solution in 10 mM CH₃CO₂NH₄) acquired with a 4.7 T FTICR-MS instrument (CID in the ESI interface region). The inset on the right shows an ESI mass spectrum of the protein acquired under mild conditions (low skimmer potential).

It appears that problems analogous to the one discussed above can be solved using alternative ion fragmentation methods. In particular, ECD seems very attractive in this respect, since it gives very specific fragmentation (only c- and z-ions) and induces cleavages specifically at disulfide bonds.34, 35 It is also important that ECD is a non-ergodic process,19 and therefore should lead to minimal rearrangement or hydrogen scrambling in the gas phase. Furthermore, ECD and nozzle skimmer CID appear to be complementary, thus making combined methods particularly promising.36, 37

Recently, the use of IRMPD as a means to measure deuterium content locally within a 21 kDa protein was reported.38 Although this ion fragmentation technique provided fairly extensive sequence coverage, it remains to be shown that the internal rearrangement processes during IRMPD do not result in extensive hydrogen scrambling.

FUTURE DIRECTIONS

A few reported successful examples of using methods of gas-phase ion chemistry to probe local deuterium content of the entire protein23–26, 38 certainly warrant further development of this technique. In this section we consider several modifications to the existing methodology that may result in further expansion of the range of biomolecules to which it can be applied, and also the amount and quality of information that can be obtained.

As already outlined, combined ion fragmentation methods appear to be very promising as far as obtaining full sequence coverage of protein ions in the 10–50 kDa mass range. It has been noted that multi-stage CID would probably be of limited help, since the peptide bonds surviving fragmentation during the first stage should have much higher dissociation energies, and would therefore be unlikely to undergo fragmentation during the following stage of an MSⁿ experiment.39 However, a tandem use of two different ion fragmentation techniques may well result in a dramatic increase in the spatial resolution in HDX/ESI-MS measurements (e.g. by ECD-induced reduction of disulfide bridges linking the CID-generated fragment ions).

The spatial resolution for probing the local deuterium content could be further improved by removing the background isotopic distributions arising from natural abundance of ¹³C and ¹⁵N. Depletion of these isotopes can be achieved by expressing proteins in cells grown on isotopically depleted media. In this manner FTICR spectra can be dramatically simplified,40 and permit more precise measurement of deuterium occupancies.

Finally, it remains to be seen if the local deuterium content can be measured by fragmenting mass-selected protein ions. The co-existence of several protein conformers in solution at equilibrium results in the appearance of multi-modal charge state distributions in the ESI spectra, with ions at high m/z (lower charge state) corresponding to more compact protein states, and those in the low m/z region corresponding to less structured conformers. Often the global HDX patterns exhibited by these high- and low-charge state protein ion peaks may be very different from each other, particularly if the solution conditions favor the EX₁ exchange regime.23 The ability to measure local proton occupancies within distinct conformers as a function of exchange time in solution would greatly simplify the procedure of assigning structural differences within the entire protein population.

A very important distinction between the HDX/ESI-CID-MS methodology presented in this paper and more conventional methods to probe local exchange using proteolysis is that the ‘exchange quench’ step is no longer required. This means that it might be possible in principle to monitor solvent accessibility not only for the backbone amides, but also for the side-chains possessing labile hydrogen atoms. Such information is inevitably lost during the quench step, since slow exchange conditions for amide protons correspond to relatively high exchange rates for labile side-chain protons,41 many of which exhibit significant protection under native conditions.42 Likewise, one might also think of extending the HDX/ESI-CID-MS methodology to probe the dynamics of oligonucleotides by measuring proton exchange at hydrogen-bonded imino groups.43

CONCLUSIONS

High-resolution structural information on proteins may be obtained by x-ray crystallography at the atomic level. However, crystallography by definition is a static measurement, thereby limiting us to snapshots, which can tell us little or nothing about kinetic processes. Dynamics, on the other hand, can be characterized by methods of time-resolved spectroscopy. However, these methods usually only provide limited information and involve placing the spectroscopic probe at many different sites within the protein to build up a detailed picture. While NMR hydrogen exchange methods have proven enlightening for small proteins,44 there are limitations, namely that spectral overlap and unfavorable relaxation properties preclude studies of large proteins, and the presence of paramagnetic cofactors such as iron is further detrimental to the quality of spectra owing to line broadening. Importantly also, NMR detects only the average proton occupancy at a single site over the entire sample, being in most cases unable to distinguish individual populations with varying extents of exchange.

In this paper we have illustrated that a combination of solution-phase chemistry (protein HDX) and ion chemistry in the gas phase (protein ion CID) can be an extremely powerful synergistic technique to study protein conformational dynamics in solution under a variety of conditions. It can provide a great deal of structural and dynamic information, making it a very attractive tool for studying the multiplicity of protein folding reactions. Continuing rapid progress in the field of protein ion fragmentation is almost certain to bring about further improvements in HDX/ESI-MS methodology that are needed in order to make this technique universally applicable to a wider range of biomolecules.

Acknowledgements

This work was supported by a grant from the National Institutes of Health (R01 GM61666). The authors thank H. Xiao for contributing data for this paper (Figure 2).