HDX probes the exchange kinetics of the main-chain amide (NH) hydrogens for deuterium in samples exposed to D₂O (Fig. 2A). The advantage is that, in principle, it provides site-specific probes along the whole polypeptide chain (except prolines and the N-terminal amino group). Under extreme pH conditions (pH<1 or pH>14) a free amide group can be either protonated or ionized 45. Amide hydrogens can exchange via those states albeit slowly under physiologically relevant conditions. The reaction is greatly accelerated by an acid or a base. In aqueous solutions the hydronium ion (H₃O⁺) or the hydroxide anion (OH^-) plays the role of the acid or the base, respectively 45. The acid- or base-catalyzed substitution of the unprotected (free) amide hydrogen, i.e. the chemical step of the exchange reaction, is strongly pH- and temperature-dependent (Fig. 2A right panel). This means that an exposed amide group would exchange on millisecond time scale at room temperature and neutral pH. Conversely, it would take about an hour to exchange the same exposed NH at low pH = 2.5 and 0°C 46. The strong pH dependence of the chemical step allows preserving the pattern of labeling that was attained under native, neutral pH conditions by shifting the sample to pH 2.5 and low temperature. This, in turn provides a window of opportunity for the analysis of isotope incorporation by mass spectrometry 46.

Steric hindrance by the neighboring side chains (R in Fig. 2A) also affects the rate of the chemical step. This dependence was calibrated using short, unfolded peptides as model compounds and constitutes relatively small correction to the intrinsic exchange rates of unprotected peptide groups (further designated as k₂ rate) 474849. This correction becomes important when quantitative interpretation for site-specific rates is being sought e.g. determination of the energetics for individual peptide bonds during folding.

In folded proteins most of the amide hydrogens participate in hydrogen bonding within secondary structure elements, e.g. α-helices and β-sheets (Fig. 2B). This considerably slows down the exchange since it requires temporal breaking of hydrogen bonds. In the folded state secondary structure is further stabilized by tertiary contacts and thus exchange requires partial or complete unfolding of the protein and exposure of the buried segments to solvent. This process is sometimes called structural opening and is a prerequisite for the chemical step of exchange 45. The open states in Fig. 2B correspond to higher-energy, partially unfolded intermediates on the energy landscape in Fig. 1. Quaternary, subunit contacts provide additional stabilization and protect NH sites on or close to the subunit interfaces 5051. In a typical well-folded protein the exchange rates may differ by many orders of magnitude reflecting local and global stability 46.

The exchange rates are governed by the folding energy landscape. However, the correspondence is not straightforward and it is necessary to consider all factors contributing to the overall exchange rate. In order to simplify such analysis several reasonable assumptions and approximations are made. For amides within protected regions (secondary structure, hydrophobic core, subunit interfaces) structural opening is the rate-limiting step and the exchange process may be considered as a two step reaction (Fig. 2C): (1) structural opening which is governed by rate constants k₁ and k_-1. (2) chemical exchange step which is governed by rate k₂ and considered irreversible under the condition of excess D₂O. Then the overall HDX rate can be described by:

k_ex = k₁k₂/(k₂+k_-1)

There are two limiting cases that are considered when interpreting HDX exchange rates:

(EX1) k₂ >>k_-1, i.e. the chemical exchange is fast compared to the structural opening step:

k_ex = k₁

The EX1 limit is typical for exchange via global unfolding under strongly denaturing conditions or for subunit dissociation in large assemblies. The rate does not depend on pH. The temperature dependence of the rate yields the activation energy for the unfolding event, i.e. provides information about the heights of barriers (roughness) on the energy landscape.

(EX2) k_-1 >> k₂, i.e. the open state is short lived compared to the rate of the chemical step:

k_ex = k₁k₂/k_-1 = K_opk₂

where K_op is the apparent equilibrium constant for structural opening. EX2 is typical for exchange from the folded state under native or mildly denaturing conditions and is strongly pH dependent (through k₂). EX2 is a product of the local unfolding equilibrium constant and the chemical step rate constant. When site-specific exchange rates in unfolded state are available it is possible to factor them out and to obtain the protection factor:

p = k₂/k_ex

This factor is then related to the local stability, i.e. depths of "wells" on the energy landscape:

ΔG_local = -R·T·ln(p)

where R is the universal gas constant and T is absolute temperature.

Because different mechanisms of local opening usually operate simultaneously, neighboring amides within the same element of secondary structure may exchange with different rates under native conditions. Upon destabilization under progressively more denaturing conditions, these rates converge to a value that is characteristic of the cooperative unfolding unit to which these sites belong 3652. This approach identifies structural characteristics of partially unfolded states and enables to map the energy landscape in structural terms 10.

Several experimental techniques for measurement of HDX are available. NMR, which determines HDX through disappearance/appearance of amide proton resonances in D₂O/H₂O solutions, is the most widely used and provides site-specific probe of protection. Presently, NMR is not applicable to large protein complexes (>200 kD) and requires high concentrations (>10 mg/ml). HDX dynamics of large macromolecular complexes can be readily examined by Fourier transform infrared (FTIR) 555657585960 and Raman spectroscopy 29326162. These techniques probe incorporation of deuterium by monitoring shifts of frequencies arising from collective vibrations of the peptide bond and the amide NH group (so called amide bands, deuterium incorporation increases the local mass and hence these bands shift to lower frequencies). These techniques differentiate the exchange rates only by the type of secondary structure and do not provide mapping onto the protein sequence.

Mass spectrometry measures deuterium incorporation as an increase of apparent mass, more specifically as a change in the isotopic composition. Advances in protein MS enabled routine resolution of isotopic composition for oligopeptides with masses up to several kDa. Similarly, shifts due to deuterium incorporation are readily discernible in mass spectra of intact proteins. This in turn allowed to probe HDX kinetics for large proteins 6364 and macromolecular complexes 3365666768 and map subunit-subunit interactions in assemblies 5169707172. Clear advantages of MS detection over NMR are the lower protein concentrations and sample quantities needed. Although, in principle, MS allows for residue-specific resolution of HDX kinetics this is seldom achieved and only region-specific information is obtained 537374.

The extent and rate of HDX is measured from mass increases of peptide fragments after enzymatic cleavage of a protein (Fig. 3). The exchange is usually initiated by diluting the protein into D₂O exchange buffer (Fig. 3A). The sample is incubated under the exchange conditions for the desired exchange period (usually 30 s to 10 h) and then quenched by rapid acidification to pH 2.5 on ice (usually done with formic acid). This effectively slows down any further exchange (so called exchange-in) and minimizes pickup of hydrogens during subsequent handling in H₂O solutions (so called back-exchange). The quenching solution may be supplemented with a denaturant (e.g. guanidine hydrochloride or urea) to facilitate dissociation of stable complexes prior to proteolytic digestion. The quenched sample may be analyzed immediately or flash-frozen and stored for up to several weeks in liquid nitrogen for off-line analysis.

Prior to MS analysis the sample is digested with a non-specific acid protease (e.g. pepsin, Aspergillus type XIII, Rhizhopus type XVIII, or their mixture 75) on ice. The resulting digest is loaded onto a reverse-phase column (usually short C-8 or peptide trap) and rapidly separated by a steep gradient. As shown in Fig. 3B it is essential that the HPLC column and the injector are cooled and kept at constant temperature (e.g. in an ice-water bath) to minimize the back-exchange and maintain reproducible conditions 76. With the advent of high-resolution MS, good HPLC separation is not essential. The chromatography step effectively provides rapid desalting and removal of undigested protein before sample introduction into the mass spectrometer. This is usually done through a standard or micro-capillary electrospray ionization interface (ESI) 7778. Any mass spectrometer with resolving power 5000 or better (e.g. a quadrupole-time of flight (Q-TOF), an ion trap) is in principle suitable for detecting HDX. Higher resolution enables a more accurate measurement of multiply charged peptides and is essential for resolving overlapping isotopic envelopes that are commonly found in MS spectra of large proteins and their assemblies. High mass accuracy and reproducibility, usually provided by calibration with a suitable standard, are essential. The highly repetitive nature of HDX-MS experiment (i.e. collection of many time points) makes it suitable for automation and several setups were reported 5379.

Note that the exchangeable side-chain hydrogens (e.g. tyrosine or serine OH groups, the amino group of lysine etc.) exchange rapidly back during digestion and HPLC separation performed in H₂O 80. On the other hand, deuterium at main-chain amide sites persists much longer (see Fig. 2A). However, back-exchange of main-chain amides, which inevitably occurs during sample handling after quenching (i.e. peptic digestion and reverse-phase separation), interferes with HDX data analysis by diminishing the differences between exchanged populations and needs to be kept as low as practically possible. Recently, several attempts were made to alleviate the back-exchange problems during separation. In principle ultra performance liquid chromatography (UPLC) provides faster and superior resolution. Comparison between UPLC and conventional HPLC in HDX/MS experiments using cytochrome c as the benchmark protein failed to provide convincing evidence in favor of UPLC mainly due to higher back-exchange. This was caused by inadequate cooling of the injector in the commercial UPLC system 81. A radically different approach using a supercritical fluid chromatography (SFC) that employs non-aqueous CO₂ mobile phase and fast flow rates yielded substantial reduction of back-exchange compared to the HDX-optimized HPLC separation. The slightly impaired peptide separation in SFC was offset by the superior resolving power of the Fourier transform ion cyclotrone resonance mass spectrometer (FTICR-MS) which was used in this study 82.

Digestion with pepsin or other acidic proteases is rather non-specific and under given conditions (e.g. digestion time, denaturant concentration, protease-to-protein ratio) yields overlapping fragments with lengths ranging from 5 to 15 amino acids. The same set of peptide fragments is obtained for a given protein and digestion conditions 75. This assures that consistent and reproducible data sets are obtained in independent runs. Each of the fragments acquires one or more positive charges during ionization and produces a typical isotopic envelope in the mass spectrum (Fig 4A). The relative abundance of different ionization states and fragments strongly depends on the particular ionization interface and mass spectrometer design. Hence, it is advisable to use one MS instrument throughout the whole study.

Each peak in the envelope corresponds to an increasing number of heavier isotopes (naturally occurring ¹³C, ³⁴S and the HDX introduced ²H) and spacing between them is inversely related to the acquired charge. Even the non-exchanged control (t = 0 s) contains series of peaks corresponding, in succession, to the monoisotopic fragment mass (the lowest m/z, all atoms are of the lightest isotope) followed by the peaks from species containing one or more of the heavier, naturally occurring isotopes. Note, that due to 1% abundance of ¹³C and other naturally occurring isotopes the monoisotopic peak is usually weak for longer fragments (>10 residues).

The fragments in non-deuterated sample are assigned to the amino acid sequence by tandem MS (MS²; i.e. fragmentation of the selected peptide in the mass spectrometer, usually performed by collision with gas or application of strong electric field, followed by analysis of the resulting ions; see 77 for details) 83. In favorable cases, especially when using ultra-high resolution mass spectrometers (e.g. FT-ICR), assignments may be based on the exact mass alone. However, a confirmation by MS² is still desirable 84.

As expected, deuteration shifts the isotopic pattern to higher m/z values (Fig. 4A). As seen in the second spectrum from top the monoisotopic peak may not be resolved in the shifting isotopic envelopes and hence the patterns are usually identified by the expected progression of m/z peaks for the given charge state. One of the main advantages of MS detection is that MS spectra exhibit distinctly different patterns for EX1 and EX2 mechanisms. This provides basis for distinguishing between the two limits without measuring the tedious and often experimentally inaccessible pH dependence.

In the EX1 regime amides remain either all unexchanged or become all exchanged at once and thus the mass spectra show a bimodal pattern (c.f. t = 0 and 40 sec in Fig. 4B). The apparent rate constant, k_ex = k1 in Eq. 2, corresponds to the rate of structural opening or a conformational change, and it is determined from the time-dependence of the integrated intensity belonging to the deuterated envelope.

In the EX2 regime exchange causes progressive shift and binomial broadening of the isotopic envelope (Fig. 4A). In this case the centroid of the isotopic envelope or the average of the peak positions weighted by their intensities are both reasonable and frequently used measures of the degree of deuteration 4651648586. In this case the HDX rate constant, k_ex in Eq. 3, is obtained from the dependence of the centroid/average position on exchange time. This analysis provides kinetics averaged over the whole fragment and consequently corrections for the effects of primary structure (sequence) on k₂ in the EX2 limit are impossible. Such correction is not necessary when comparing two states of the same protein, e.g. for mapping subunit interfaces in protein assemblies.

In certain cases (e.g. high resolution spectra like those shown in Fig. 4) the true deuteration distribution may be obtained from the measured envelope by deconvolution with the naturally occurring isotope distribution 87. Such procedure allows analyzing exchange under conditions when part of the fragment is exchanging via EX1 mechanism while the rest belongs to an EX2 limit 88.

Site-specific resolution of exchange will be essential for quantitative study of enzymes and molecular motors and for direct comparison with theoretical results 899091. Several attempts to improve the resolution have been made. One approach was to analyze simultaneously the exchange kinetics of overlapping fragments and separate the individual contributions computationally. In favorable cases this may produce site-specific resolution for few residues within the entire sequence. An attempt at more reliable experimental solution was made by using MS² for sequencing of deuterated fragments. However, it was found that the collision-induced fragmentation step led to significant scrambling of deuterium among the amide sites and hence this approach proved impractical 74. Recent experiments using gentler fragmentation methods demonstrated significant preservation of the deuteration pattern 7392.

In all cases the measured degree of exchange is always lower than the actual value because some of the label is lost during sample handling (e.g. digestion and HPLC) due to the back-exchange. The true extent of exchange is an important parameter for interpretation. Take the following example: In the absence of site-specific resolution the exchange is averaged over the whole peptide fragment. Hence, situations, in which half of the fragment residues are exposed and exhibits fast exchange while the rest is buried and does not exchange at all, are common. Ignoring the true extent of the exchange, which amounts to only half of the available sites in this example, and taking into account only the fast kinetics would lead to an erroneous conclusion that the whole peptide is exposed.

In order to obtain the true extent of exchange one must correct for the back-exchange. The degree of back-exchange depends on the experimental setup and sample handling and varies from as little as 10% for ESI-MS to about 40% in MALDI-TOF MS detection. There is also variation between different fragments, presumably due to sequence effects on the back-exchange. Therefore, the best way to correct for back-exchange is to measure a fully deuterated control sample under identical conditions, e.g. a protein which was first denatured and then re-folded in D₂O buffer. For some proteins refolding in D₂O is impossible and one must resort to approximations. For example, it is fair to qualitatively compare the relative extent of exchange for two states of the same protein (e.g. assembled and free subunit of a virus) if the two data sets were collected under identical conditions 70.

Mass spectrometry is venerable for generating large data sets and HDX-MS is no exception. A typical 35 kD protein may yield about 100 assigned fragments which contribute to the mass spectra. In addition, the spectra contain plethora of unassigned peaks e.g. from oxidized protein or from pepsin. In HDX-MS the useful MS data is spread over several (4–10) scans within the LC-MS run. Usually, the full kinetics contains at least 15 time-points collected in three independent replicas. Taken together one needs to extract and process at least 4000 isotopic envelopes from about 200 spectra per each kinetic curve. Such a task calls for a considerable automation and deployment of data-mining tools.

MS² peptide assignments can be done using either commercial proteomics add-on packages for automated database searches and subsequently validated by spectral prediction tools when necessary 93. Despite the apparent popularity of HDX-MS, specific tools are not to be found in any of the commercial software packages that are currently supplied with instruments. This spurred in-house development of various HDX-MS-specific software packages 547994 and eventually led to applications ranging from semi-automated to fully-automated processing that are now freely available 889596. The utility of some packages is still hindered by the plethora of proprietary data formats used by different manufacturers but could be overcome by adopting the mzXML standard for data interfacing 9697. Recently, Pascal and colleagues launched a web application ("Deuterator") compatible with multiple file formats for automated HDX-MS data analysis 9698. Although promising, the utility of the web-based approach is somewhat compromised by the need to supply large datasets to a remote server.

Multi-state and two-state protein folding cannot be often distinguished due to instability of partially folded intermediates. Meta-stable intermediates can be detected via HDX and play a crucial role in the determination of folding kinetic and native state dynamics 85. HDX experiments are also considered most effective in teasing out the structural details of protein folding intermediates, often with amino-acid resolution 99.

The three-state unfolding of ubiquitin (highly structured, partially unstructured A-state, and fully unstructured state) was examined by HDX-MS 100. The highly dynamic A-state consisted of flexible, rapidly exchanging C-terminal region while the N-terminal domain adopted less dynamic, native-like β-strand configuration.

Pulse-labeling HDX-MS techniques enable detection and characterization of transient folding intermediates which are not readily populated under equilibrium conditions 101. An on-line HDX pulse-labeling ESI MS apparatus was developed and tested on myoglobin folding and heme incorporation 86102 (Fig. 5). The method is based on tandem mixing chambers: the first one serves to initiate folding/assembly reaction while the second one is used to stop the folding reaction and transiently expose the products to deuterium label. Subsequently, the exchange is quenched and the incorporated label is quantified by MS. This provides a "snapshot" of protection at a particular stage of the reaction. Usually, this is applied to intact proteins but there is no principal obstacle in carrying out pepsin digestion and LC-MS analysis and obtain region-specific exchange kinetics.

The method enabled millisecond resolution of the folding dynamics for a well-studied model protein ubiquitin and led to the discovery of a previously uncharacterized intermediate 103. The method was then used to follow folding and association of dimeric protein, S100A11, and revealed three different kinetic species – a relatively unfolded monomer, a more compact folded monomer and the native dimer 104.

Pulse labeling can also reveal the interplay between folding and assembly in multidomain, multimeric proteins. Rabbit muscle aldolase is a homotetramer encompassing subunits with three domains each. A locally unfolded aldolase was pulse-labeled with deuterium after destabilization in urea for a defined period of time (1 min to 48 h). Isotope patterns revealed four populations corresponding to distinct conformations: one native and three distinct, partially unfolded intermediates. The intermediates, which were further characterized using pepsin digestion, corresponded to aldolase with one, two or all three domains unfolded, respectively. Remarkably, the completely unprotected intermediate, which also lacked secondary structure, still remained tetrameric, suggesting strong coupling between folding and assembly of the tetramer. The observed HDX kinetics supported sequential and cooperative mechanism of aldolase unfolding 5180105.

When there is no high resolution structure available, HDX-MS experiment can provide valuable information about the domain structure and may be used to validate structural models which are built solely on the basis of limited sequence homology. This approach was applied to a protein kinase scaffolding protein, D-AKAP2, which augments interactions of signaling molecules inside cells 106. HDX-MS revealed two regions of low exchange located in the midst of fast exchanging regions and identified two distinctly folded regions (Fig. 6). Fig. 6A illustrates a plausible representation of region-specific HDX kinetics in the absence of three-dimensional structure. The sequence, which was found homologous to regulator of G-protein signaling (RGS), mapped within the first folded region. The second folded region encompassed a highly protected protein kinase A (PKA) binding site and a less protected PDZ-binding motif (PDZ domain is a potential target of D-AKAP2). HDX-MS thus confirmed the multi-domain architecture of D-AKAP2. HDX-MS was also used to validate a homology-based structural model of the RGS domain (Fig. 6B–C) 106.

An interesting application of HDX-MS has been developed to follow protein folding during in vitro, cell-free, transcription-translation 107. This is a fast and easy way to use method based on MALDI-TOF and yields global degree of folding using minute protein quantities without purification. In addition, it affords rapid identification of the expressed protein. HDX-MS was also used to refine protein constructs for crystallization trials 108109. A somewhat similar MALDI-MS approach was used to characterize stability of an expressed protein in the cell extracts i.e. under conditions close to those found in the cytoplasm 38. The method, abbreviated by the authors as SUPREX (stability of unpurified proteins from rates of H/D exchange) revealed that the model protein, bacteriophage λ repressor, exhibited identical stability in the cell lysate and in the dilute solution of pure protein, respectively. This study was later extended to perform and detect HDX of the λ repressor in the cytoplasm of intact E. coli cells, i.e. in vivo 37. This method exploited permeability of lipid membranes for small molecules like water and urea which allowed for rapid equilibration of D₂O between the cytoplasm and the deuterated medium containing increasing amounts of denaturant. As in their previous study the authors found no difference between repressor stability under in vivo and in vitro conditions, respectively. However, the in vivo stability was significantly enhanced by administering hyperosmotic shock to the cells prior to the exchange experiment.

The first step towards functional macromolecular complex is assembly of subunits guided by subunit-subunit interactions. In many instances only the high-resolution structures of subunits are available while a medium resolution electron density of the whole assembly is readily obtained by cryo-EM. The fitting of the subunit structures into the EM density is greatly facilitated by knowing the subunit interfaces within the complex. This strategy was adopted for bacteriophage ϕ12 which is a dsRNA virus belonging to the Cystoviridae family and is structurally related to members of the Reoviridae family 110. Although the high-resolution structure of the virus-associated packaging ATPase, a hexameric molecular motor P4, had been known 111 and the hexameric structure had been resolved by cryo-EM 112113 it was not possible to determine which way the hexamer interacted with the viral procapsid (Fig. 7A). HDX-MS was used to map the subunit interfaces 94. Fig. 7B shows comparison of the isotopic patterns for the C-terminal helix in the free hexamer (red) and the PC-bound motor (blue). A substantial increase in protection (lower final plateau in Fig. 7C) and slower kinetic (smaller initial slope) is clearly apparent for the assembled state. Mapping of the average exchange rates for all resolved fragments onto the surface representation of the hexamer confirmed that P4 associates with the procapsid using its C-terminal facet (Fig. 7D). This information was subsequently used to fit the atomic model of the hexamer into a refined asymmetric cryo-EM reconstruction of a related virus ϕ6 113 (Fig. 7E).

HDX-MS also revealed that the association of the motor with the viral capsid stabilized subunit interfaces (c.f. exchange of helices IH1 and IH2 in Fig. 7F) and prevented spontaneous opening of the hexameric ring (Fig. 7G). Ring opening had been detected by HDX-MS and associated with loading of ssRNA into the central channel (Fig. 7G) 114. The exchange dynamics of the PC-associated hexamer closely followed temperature B-factors derived from the crystal structure while the free, solution state was more dynamic 94 (Fig. 7F).

HDX-MS and crosslinking was used to identify new contacts between subunits in the capsid of human immunodeficiency virus (HIV) 70. The subunit of HIV capsid (CA) encompasses N-terminal domain connected via a linker to C-terminal domain. The capsid assembly process was thought to be driven by association (dimerization) of C-terminal domains and hexamer formation by six N-terminal domains. While the C-terminal domain was known to form stable dimers in solution there was no evidence for the oligomerization of the N-terminal domain. Hence, the assembly mechanism was far from being certain. By comparing the exchange of the full-length CA in the monomeric and the assembled state Lanman and colleagues were able to delineate a new interface between the N-terminal and C-terminal domains of neighboring subunits 70115. This contact was shown to play essential role in stabilizing hexamers during assembly.

HDX-MS can also provide information about subunit association kinetics in functional assemblies. Small heat shock proteins (sHSP) belong to a family of molecular chaperones, which transiently bind partially unfolded proteins and prevent their aggregation. One member of this family, HSP16.9, assembles into a dodecamer at room temperature. Surprisingly, subunit interfaces exhibited no protection against HDX after incubation for 5 s in D₂O 50. Note, that under these conditions most oligomeric proteins show significant protection of their subunit interfaces. Almost complete exchange suggested that large conformational motions were taking place within the assembly, leading to the disruption of subunit interfaces. When pulse-labeled for just 10 ms, the subunit interfaces showed significantly lower exchange comparing to the 5 s experiment. Thus, the HSP16.9 oligomer underwent fast association-dissociation dynamics on a sub-second time scale. At 42°C, HSP16.9 forms a dimer with the same exchange pattern and kinetics as the dodecamer at room temperature. It is believed that HSP16.9 is heat-activated by shifting the equilibrium between the two forms 50 and the fast association-dissociation dynamics plays essential role in the process. This example also shows that it is important to consider HDX kinetics on millisecond time scale when dealing with metastable, highly dynamic complexes.

EM and X-ray diffraction yield structures of well-defined, stable conformations. However, even highly symmetric, icosahedral viral capsids are dynamic entities. Essential biological processes, like delivery of viral genomes, are mediated by structural transformations. HDX-MS enables monitoring of these dynamic events.

Many viruses require a maturation step in which a capsid precursor undergoes large scale structural rearrangements in response to e.g. protease processing of the viral polyprotein. HIV maturation is triggered by a virion-associated protease which specifically cleaves the assembled Gag polyprotein to releases the individual structural proteins (NC, nucleocapsid; CA, capsid; MA matrix; ENV, envelope glycoprotein). During the process the assembled CA collapses from a spherical form into a conical core. This step is essential for infectivity and hence the protease has been successfully targeted by several antiviral drugs. However, in structural terms, the maturation process has been elusive 115. HDX-MS was used to compare CA in the mature and immature form of virus-like particles (VLP). The intact CA protein in the mature form exhibited significant protection with respect to the immature state (Fig. 8A) 116. A bimodal pattern in the m/z spectra of the intact, mature CA indicated two populations of CA conformers. The first exhibited protection similar to the immature CA while the second was significantly more protected. Similar bimodal HDX pattern was observed for a peptide fragment from the interface between N- and C-terminal domain (Fig. 8B). This demonstrated that only half of the subunits in HIV-1 capsid matured into the conical cores and that the formation of the heterotypic N-C interface is a key feature of the maturation process. On the other hand, the formation of new interfaces between N-terminal domains, which had been proposed on the basis of EM to mediate maturation, was not detected 116.

Unlike HIV dsDNA and dsRNA bacteriophages (e.g. Cystoviridae discussed above) mature via packaging of their genomes into empty procapsids which simultaneously expand and undergo substantial stabilization 117118. HDX-MS application to viral maturation was pioneered using dsDNA bacteriophages P22 and Raman spectroscopy for detection 29. This initial investigation was subsequently extended by mass spectrometry 33 and demonstrated that the observed stabilization of the viral lattice was due to refolding of coat protein domains. Further study of P22 capsids by FT-ICR HDX-MS identified the N-terminal domain as being substantially stabilized upon maturation while the C-terminal dimerization domain remained unaffected. Together with crosslinking experiments HDX results assisted mapping of the refolded domains into the EM-derived electron density 119120.

Flexibility of virions has been demonstrated as an important factor in the delivery of genomes by plus-sense ssRNA viruses 121. Capsid dynamics of human rhinovirus was studied by HDX-MS 122. A fast and intermediate exchange pattern was observed for the interfaces around the five-fold axis. These sites may serve as the initiation site for uncoating and release of the viral RNA.

Native-state HDX of whole protein subunits within the intact large and small ribosomal subunit was detected by MALDI-MS 66. While this method did not resolve subunit interfaces it provided first glimpse at the overall dynamics for almost all ribosomal proteins. The protection correlated well with the assembly pathway i.e. subunits that were incorporated early during ribosomal assembly exhibited significantly higher protection. On the other hand, faster HDX was observed for subunits implicated in tRNA translocation between sites or those involved in pivoting of large and small ribosomal subunits. These functions presumably require higher flexibility.