In-droplet hydrogen-deuterium exchange (HDX) to examine protein/peptide solution conformer heterogeneity

Abstract

Rationale:

Many different structure analysis techniques are not capable of probing heterogeneity of solution conformations. Here, we examine the ability of an in-droplet hydrogen/deuterium exchange (HDX) to directly probe solution conformer heterogeneity of a protein with mass spectrometry detection.

Methods:

Two vibrating cVSSI devices have been arranged such that they generate microdroplet plumes of the analyte and D₂O reagent which coalesce to form reaction droplets where HDX takes place in the solution environment. The native HDX-MS setup has been first explored for two model peptides having distinct structural compositions in solution. The effectiveness of the multidevice cVSSI-HDX in illustrating structural details has been further exploited to investigate coexisting solution-phase conformations of the protein ubiquitin.

Results:

In-droplet HDX reveals decreased backbone exchange for a model peptide having greater helix-forming propensity. Differences in intrinsic rates of the Alanine and Serine residues may account for much of the observed protection. The data allow the first estimates of backbone exchange rates for peptides undergoing in-droplet HDX. That said, the approach may hold greater potential for investigating tertiary structure and structural transitions of proteins. For ubiquitin protein, HDX reactivity differences suggest that multiple conformers are present in native solutions. The addition of methanol to buffered aqueous solutions of ubiquitin results in increased populations of solution conformers of higher reactivity. Data analysis suggests that partially-folded conformers such as the A-state of ubiquitin increase with methanol content; the native state may be preserved to a limited degree even under stronger denaturation conditions.

Conclusion:

The deuterium uptake after in-droplet HDX has been observed to correspond to some degree with peptide backbone hydrogen protection based on differences in intrinsic rates of exchange. The presence of co-existing protein solution structures under native and denaturing solution conditions have been distinguished by the isotopic distributions of deuterated ubiquitin ions.

Introduction

The advent of soft ionization techniques such as ESI¹ and MALDI² placed mass spectrometry (MS) as a suitable tool for examining the structures of biomolecules. The combination of ESI and MS for studying higher-order protein structure has gained particular attention after the discovery that non-covalent interactions could be preserved into the gas-phase and ultimately vacuum environment of mass spectrometers^3,4. It is well understood that, under physiological conditions, a protein’s function is largely dependent on its 3-dimensional structure. This largely consists of different interactions/arrangements of tertiary and secondary structural components. It was discovered that the composite solution-phase structures (comprising quaternary, tertiary, and secondary elements) of protein molecules could be retained to some extent if the proteins undergo a gentle transition from the solution- to the gas-phase environment during ionization^3,5,6. Based on this observation, a new branch of biological mass spectrometry was ultimately developed known as native mass spectrometry (native MS)⁷.

The exact working conditions and even definition of native MS can be hard to specify, but in general, this technique aims to produce and characterize the most native-like conformations using MS by optimizing various solvent and instrument conditions⁸. Early attempts in native MS utilized a buffered solvent system to control the pH and ionic strength of the droplets undergoing fissioning during the ESI process^3,5,6,9. Traditionally, a capillary emitter tip having a relatively smaller orifice diameter compared to that in regular ESI is used for native MS to improve ion signal levels⁸; this implementation also provides the added benefit of using a lower flow rate and thus sample consumption¹⁰. This modified ionization approach is known as nanoelectrospray ionization (nESI) and it is beneficial to native MS due to its increased sensitivity, lower bias voltage requirement and overall desalting effect^10,11. The other parameters that also play a crucial role in performing native MS are the inlet/source temperature^12,13, pressure¹⁴ and acceleration voltage used in ion optics¹².

During a similar time period as the aforementioned developments for native MS, another technique that witnessed continued development in the study of solution structures of proteins is hydrogen/deuterium exchange mass spectrometry (HDX-MS)^15,16. The labeling of proteins with hydrogen isotopes dates back to the 1950s¹⁷ where the rate of labeling was shown to reflect the folding arrangement of the protein conformation. Before the use of MS, Nuclear Magnetic Resonance (NMR) spectroscopy was more readily combined with HDX to study the structures for proteins as demonstrated in a range of experiments¹⁸. Initial attempts at coupling HDX with MS commenced in the early 1980s by Sethi and Verma where fast atom bombardment (FAB) ionization techniques were utilized^19,20. The later development of ESI would extend the application of HDX-MS to large biomolecules. For example, over the years HDX-MS has become more accessible for tracking conformational changes and structural heterogeneity in solution^21,22. Here, the shift in mass upon HDX can easily be measured with MS and used to elucidate aspects of a protein’s structure and dynamics.

One of the widely used approaches in this field has been ‘bottom-up’ HDX-MS where the localization of accessible hydrogens is revealed in enzymatically cleaved peptides which ultimately indicates the presence and possibly arrangement of secondary structural components of a protein in solution as well as tertiary and quaternary interactions^23–25. Such experiments take advantage of the fact that the time frames required for labeling accessible sidechain and backbone sites in a protein are very different than those for protected sites^26–31. Solution-phase HDX has also been coupled with tandem MS performed in the form of ‘top-down’ approaches to assess the level of protection in different segments of a protein^32–34. Although highly informative, conventional HDX-MS approaches have some accompanying challenges including those associated with sample preparation, obtaining high reproducibility, as well as isotope scrambling, etc¹⁶. For bottom-up approaches a complicating factor can be the cost of high-end instrumentation for performing the on-line method³⁵. In such experiments, generally the side chain deuteration is lost in the back exchange arising from the use of the water-containing mobile phase²⁴. A concern is that some backbone exchange can be susceptible to back exchange as well.

Recently, Zare and coworkers have demonstrated the use of a theta capillary to achieve in-droplet, solution-phase HDX³⁶. It has been proposed that some reactions inside a microdroplet occur at a much faster rate compared to that in bulk solution.^37–40 Indeed, this phenomenon may be exploited to achieve the rapid, in-droplet deuterium labeling reported here. Microdroplet HDX has further been utilized to successfully differentiate isobaric molecules based on their shift in the isotopic envelopes of the exchanged ions⁴¹. One remarkable study performed by Gallagher and coworkers even demonstrated the ability to distinguish carbohydrate isomers using in-electrospray HDX⁴². A related study combined theta capillary techniques with the new ionization method termed vibrating sharp-edge spray ionization (VSSI),⁴³ to examine the ability to distinguish the contributions by different organic moieties to rapid HDX for small molecules⁴⁴. Here, it was suggested that establishing the contributions to HDX by different functional groups could be used to begin predicting the exchange behavior of small molecules.

A recent addition to the field of in-droplet HDX, is the application of a multidevice capillary VSSI (or cVSSI)⁴⁵ method. Briefly, the technique known as VSSI achieves bulk solution nebulization through the vibration (~96 kHz) of a brittle substrate. A requisite property of the substrate is that it contains a sharp edge from which the mircrodroplet plume emerges. Notably, cVSSI can function with and without the application of a bias voltage^46,47. In the dual emitter tip setup, two cVSSI devices are positioned before the MS inlet; one nebulizes the analyte solution and the other the D₂O reagent under the same polarity bias to achieve in-droplet HDX. Furthermore, additional experiments with this setup confirm that the majority of the exchange occurs in the solution phase⁴⁵. One of the interesting aspects of this setup is that the HDX reaction can be terminated at any time point by switching off the device used for D₂O reagent nebulization as the DC bias is not sufficient to initiate plume generation. This distinguishes the approach from theta capillary based nESI³⁶ or dual glass capillary based cVSSI⁴⁴ experiments.

cVSSI possesses key attributes that make it highly suited for native MS studies. The first is that of improved sensitivity. Recently, field-enabled cVSSI was shown to provide ~10 to 100 fold improvements in ion signal production compared to state-of-the-art ESI for biomolecules where the effect is more pronounced for negatively charged ions⁴⁷. This was accomplished while preserving biomolecular structure such as that demonstrated by the mass spectra for G-Quadruplex DNA ions which were shown to be similar to those acquired using ESI under native conditions. The second advantage of cVSSI has been demonstrated recently, where the new ionization approach is shown to be a remarkably gentle technique allowing the preservation of solution structures for fragile proteins that may unravel during the ESI process⁴⁸. A third attribute of cVSSI is its relative simplicity in that a droplet plume is generated in a straightforward fashion without the need for an auxiliary nebulization gas which would not allow facile in-droplet HDX measurements. Additionally, the technique decouples droplet production from a coulombically-driven process such as encountered with nanoelectrospray ionization (nESI). Recently it has been shown that this allows improved preservation of structure for fragile proteins relative to nESI^48,49. These unique attributes (softness, ionization efficiency, and simplicity/accessibility) of cVSSI suggest it may be possible and indeed useful to couple native MS and in-droplet HDX for examining challenging biomolecular solution structures including coexisting conformers and structural transformations. In seeking to apply new techniques to the study of biomolecular structure, this work builds upon the seminal experiments of Derrick and Williams and their coworkers who studied the nature of in-droplet processes for biomolecules using theta capillaries^50,51.

In this work, a multidevice cVSSI setup has been utilized for in-droplet HDX to demonstrate structural characterization of two model peptides (acetyl-PAAAAKAAAAKAAAAKAAAAK and acetyl-PSSSSKSSSSKSSSSKSSSSK) with contrasting helical propensity. The results from the in-droplet HDX experiment show that the serine-rich model peptide (S-peptide) possessing more random coil structure has a greater number of backbone hydrogens that exchange for deuteriums compared to the alanine-rich peptide (A-peptide) with a greater propensity to form a helix⁵². The small but statistically significant difference in backbone exchange for the two peptides raises important questions regarding the evolution of biomolecule structure as it transitions from the solution to the gas phase. One unanswered question is whether or not the ~12 fold difference in the peptide backbone reactivity arises from protection due to the presence of secondary structure or to differences in intrinsic rates of the amino acids comprising the two peptides.

The application of this in-droplet HDX setup has been further extended to the analysis of the globular protein ubiquitin. From such experiments, it may be argued that the multidevice cVSSI-HDX approach is more effective in distinguishing protein conformers having different tertiary structures. The evolution of various low-charge-state mass spectral features of ubiquitin ion solutions containing differing amounts of methanol has been revealed by cVSSI-HDX-MS experiments. Methanol has previously been reported to disrupt the native structure of ubiquitin in solution and a range of partially-folded to unfolded structures are reported⁵³. Here, multimodal isotopic distributions are observed after deuterium exchange occurs in the fused microdroplets confirming the presence of a heterogeneous mix of solution-phase conformers. Several lower intensity features with higher exchange levels become more pronounced at higher methanol content showcasing the capability of dual-cVSSI for rapid solution-phase deuterium labeling. This is the first attempt to date to distinguish such coexisting conformations of a protein using the cVSSI-based in-droplet solution-phase labeling approach.

Experimental Section:

Chemicals and Solvents.

Ubiquitin and Deuterium Oxide (D₂O) reagent were purchased from Sigma-Aldrich (St. Louis, MO). Standard amino acids Alanine and Lysine were also purchased from Sigma-Aldrich (St. Louis, MO). The other standard amino acid Serine was purchased from Alfa Aesar (Haverhill, MA). The alanine-rich model peptide (A-peptide) and the serine-rich model peptide (S-peptide) were both purchased from Genscript Biotech (Piscataway, NJ). Ammonium acetate (AmAc), LC-MS grade methanol and LC-MS grade water were purchased from Fisher Chemical (New Jersey, NJ). All the chemicals and reagents were used without further purification. Single amino acid standard solutions were prepared in 50 mM AmAc at 0.1 mgˑmL⁻¹ concentration. The ubiquitin and model peptides were also dissolved in 50 mM AmAc buffer solution to prepare stock solutions each at 1 mgˑmL⁻¹ concentration. The pH of the 50 mM AmAc peptide and protein solutions was determined to be ~6.5. For the in-droplet HDX of the model peptides, stock solutions were diluted (10×) in concentration and Lysine, Serine, and Alanine (0.1 mg·mL⁻¹) were used as internal standards. For the denaturation experiments, the A- and S-peptide stock solutions were diluted 5 fold with methanol. Ubiquitin stock solution was also diluted by adding buffer (50 mM AmAc) and methanol to achieve a final concentration of 0.1 mg·mL⁻¹ in five different solvent compositions (addition of 0%, 10%, 20%, 40% and 80% methanol). To prepare the internal standard stock solution for the protein experiments, A-peptide was dissolved in 50 mM AmAc to provide a concentration of 40 μM. 50 μL of A-peptide was incorporated into 1 mL of each ubiquitin solution so that the final concentration of the internal standard was 2 μM.

Multidevice cVSSI.

As mentioned above, two different cVSSI devices were constructed; one nebulized the analyte solution while the other was used to generate a plume of D₂O to achieve in-droplet HDX. Figure 1 shows the schematic diagram of a dual cVSSI setup. The cVSSI devices are constructed upon uniformly thin, rectangular-shaped glass substrates (VWR, Radnor, PA) to which a standard 4.6 KHz piezoelectric buzzer (Murata Electronics, Smyrna, GA) is attached using epoxy glue. The buzzers are 1.063 inches in diameter and have 300 Ohm impedance. Emitter tips, having an orifice diameter of ~10 to 20 μm, were pulled using a P2000 micropipette puller (Sutter Instrument, Novato, CA) from standard fused-silica capillaries (360 μm OD / 100 μm ID). A single capillary emitter tip was then glued to the glass slide in a manner such that the tip maintained a ~60° angle with the furthest edge of the slide. A poly-tetrafluoroethylene (PTFE) tube was pulled over the blunt end of the tip to deliver the analyte solution to the device. A small piece of platinum (Pt) wire was inserted into the PTFE tube not far from the emitter tip. An alligator clip holding the Pt-wire was connected to the DC-voltage source to apply the bias voltage.

Figure 1.

Illustration of the multidevice cVSSI setup with mass spectrometer inlet. For in-droplet HDX, both cVSSI devices are actuated and the DC potentials are turned on. For standard cVSSI-MS (without HDX), the reagent-spraying device is turned off and removed from the mass spectrometer inlet.

To achieve in-droplet HDX, two of the cVSSI devices were vibrated using an arrangement of a waveform generator (Tektronix AFG-1062, Beaverton, OR) and an amplifier (Krohn-Hite 7500, Brockton, MA) described previously and shown in Figure 1. The amplifier was connected to a homebuilt actuator switchbox with a cable. The switch box contains three different output sockets and two were used to power the separate cVSSI devices via two different cables. Separate syringe pumps were used to infuse analyte and reagent solution from BD 1 ml-size syringes to the respective devices. The devices were finally arranged in a way such that their emitter tips were located at a ~90° angle (see Figure 1). Additionally, the analyte emitter tip was positioned in head-on fashion with the mass spectrometer inlet and was situated relatively more removed (~4 mm) compared to the D₂O-reagent emitter tip (~1 mm). The bias voltages applied to the analyte solution and reagent solution were +1400 and +300 volts, respectively.

MS Settings.

A linear ion trap mass spectrometer (LTQ-XL, Thermo Fisher Scientific, San Jose, CA) was used to collect the data for the peptides. All experiments were conducted in the positive ion mode setting using a scan range of 100 to 1200 m/z and the capillary inlet temperature was set at 350°C. For the ubiquitin protein, the data were collected using an orbitrap mass spectrometer (QExactive Hybrid Quadrupole / Orbitrap, Thermo Fisher Scientific, San Jose, CA). The protein mass spectra were collected in positive ion mode with a scan range of 800 to 3000 m/z. The capillary inlet temperature was set at 275°C. The default heated ESI (HESI) source was removed to place the cVSSI devices on two lab jack stand platforms. The removal of the HESI probe triggers a relay switch that forces the instrument into the offline mode. Therefore, an external switch was used to inactivate the relay while using the cVSSI source.

Deuterium uptake determination.

For data analysis, the xCalibur Qual Browser software (Thermo Scientific, San Jose, CA) was used and the data points of each mass spectrum were extracted and imported to an Excel (Microsoft, Redmond, WA) worksheet for further analysis and plotting.

The deuterium uptake experimental protocols were as follows. The cVSSI device infusing the D₂O reagent plume was switched off first and only the cVSSI device containing analyte solution was actuated. Mass spectra were then recorded for non-deuterated analyte ions. Next, both of the devices were activated to initiate in-droplet HDX. For the peptide ions, the total amount of deuterium exchange was determined by measuring the shift in intensity-weighted average m/z values after HDX reactions and multiplying the obtained shift by the respective charge state. The weighted m/z-values (M_w) of an isotopic distribution were calculated according to Eq 1.

$$M_w=\frac{{\sum }_i^n{\left(m/z\right)}_i{I}_i}{{\sum }_i^n{I}_i}$$ (1)

Here, (m/z)_i and I_i are the m/z-value and intensity of the ith isotopic peak, respectively, from an isotopic distribution having total number n isotopolgues. The deuterium uptake level of the internal standard (Lysine) was used to scale the deuterium incorporation calculated for the analyte ions. The deuterated peptide data were recorded in triplicate and average values were calculated to report the total deuterium uptake.

Estimating the relative backbone amide protection for the model peptides.

To estimate the number of side chain hydrogens that undergo exchange, moiety-dependent HDX parameters were calculated from the analyses of the standard amino acids. From 48 measurements, multiple regression was employed as described previously⁴⁴ to determine the relative contribution by each side-chain moiety to the overall exchange level of the S- and A-peptides. Multiple regression was performed using MATLAB (MathWorks, Natick, MA) and confirmed with an online calculator (https://www.statskingdom.com/410multi_linear_regression.html) to determine the HDX parameters and the associated significance (Table S2). The side chain hydrogen exchange levels for each peptide were then estimated using the HDX parameters for the associated side chain moieties. Finally, the backbone hydrogen exchange value was calculated by subtracting the estimated side chain exchange levels from the total deuterium uptake.

Conformer population fitting.

The Origin Pro (Origin Lab Corp, Northampton, MA) software suite was used to perform the multiple peak fitting of the isotopic distributions of the deuterated ubiquitin ions. The base point x,y coordinates of each isotopic peak were exported to the Origin Pro worksheet and plotted as a scatter graph. The Gaussian function was selected to fit this cumulative curve with a minimal number of Gaussian features. The selection criteria for the peak parameters are explained in the Results and Discussion section.

Solvent accessibility determinations.

The initial structure of ubiquitin was obtained from the RCSB Protein Data Bank (PDB ID: 1UBQ)⁵⁴. The structure was refined by generating a protein-only PDB file after removing water from the ubiquitin crystal structure. The Psfgen plugin in VMD was utilized to approximate the hydrogen atom coordinates. The top_all36_prot_lipid.inp topology file was used to generate the PSF and PDB files. Next, the Solvate1.6 package was used to create a water box with periodic boundary conditions where the protein molecule is surrounded by 10 Å of water. 150 mM NaCl (ions) was added to this water box using the ‘Add Ions’ plugin in VMD⁵⁵. The CHARMM36 force field parameter file was employed for the simulations, which were performed for 1000 steps for energy minimization followed by equilibration for 1 ns in NPT ensemble at a pressure of 1.01325 and temperature of 310 K, using NAMD2.14^56,57. The entire trajectory was wrapped using PBC tools, and the last frame was used to generate a water-containing, well-equilibrated PDB structure. Subsequently, a TCL script was used to compute all water-contact distances (≤5 Å) for side chain exchangeable hydrogens as shown in Figure S1 in the Supporting Information. The total number of side chain accessible hydrogens was then computed based on the summation of water accessible sites.

Results and Discussion:

Examining peptides with different solution structures using in-droplet HDX.

The solution structures of the two model peptides (S-peptide and A-peptide) selected for this experiment have been extensively characterized by ion mobility spectrometry (IMS)-MS experiments^52,58,59. Additionally, CD data for these solutions suggest that the peptides possess very different helical characteristic in solution such that the Alanine-rich peptide (A-peptide) exhibits helical propensity in solution whereas the Serine-rich peptide (S-peptide) is mostly random coil⁵². Although the side chain hydrogens are expected to exchange at a very rapid rate in bulk solution, the backbone amide hydrogens should exchange more slowly especially those of the A-peptide where they are involved in hydrogen bonding. A question arises as to whether or not rapid, in-droplet HDX can distinguish the backbone protection of hydrogens by the A-peptide. This question arises because the droplet environment may cause the HDX reaction to proceed at a much higher rate as observed for other reactions. That is, there may be a potential to “trap” even slow exchanging backbone hydrogens. Based on prior in-droplet HDX studies that showed different organic moieties could contribute differently to the overall molecular exchange⁴⁴, it was considered that the differences in hydrogen bonding status for backbone amides should be investigated using a multidevice cVSSI HDX approach.

The mass spectral features observed for the A-peptide and S-peptide are the [M+2H]²⁺ and [M+3H]³⁺ ions. Figure 2 shows a comparison of the isotopic distributions obtained before and after deuteration of the A- and S-peptides. The S-peptide exhibits a greater mass shift after HDX reactions demonstrating an average hydrogen exchange for deuterium of 17.5±1.2. This is ~2.5-times the deuterium incorporation recorded for the A-peptide. The difference in total deuterium uptake may be due to two conditions, namely: 1) the two peptides have unequal numbers of exchangeable side chain hydrogens (as shown in Table 1); and 2) they have different numbers of backbone hydrogens accessible for exchange depending on their local environment.

Figure 2.

Expanded mass spectral regions for the [M+H]²⁺ ions of A-peptide and S-peptide respectively. The top panels (A, C) show spectral features without in-droplet HDX, and the bottom panels (B, D) show spectral features with in-droplet HDX.

Table 1.

Total deuterium uptake values and estimation of backbone deuterium incorporation from cVSSI-HDX-MS

Peptide	[M+zH]z+	Total exchangeable hydrogen	Total average D-uptake (D)	Estimated sidechain-H exchange (SH)	Estimated backbone-H exchange (D-SH)
S-peptide	[M+2H]2+	47	17.5±1.2	15.3	2.2±1.2
	[M+3H]3+	48
A-peptide	[M+2H]2+	31	6.4±.4	6.2	0.2±.4
	[M+3H]3+	32

Comparison of backbone hydrogen exchange for the A- and S-peptides.

The in-droplet HDX with a dual cVSSI setup causes the formation of coalesced droplets having approximately a 50:50 H₂O:D₂O ratio. Previously, a limit of ~45% deuteration efficiency has been reported for similar experiments with small molecules⁴⁵. Consistent with the prior work, the highest levels of exchange observed for the internal standard Lysine are approximately similar; roughly 50% deuterium labeling efficiency is typically observed as shown in Figure S2 and Table S1 in the Supporting Information . Notably, for the S- and A-peptide, the %deuterium labeling efficiencies are ~37% and 20%, respectively. A question arises as to whether or not the decreased efficiency for the A-peptide can be attributed to the presence of secondary structure especially considering that molecular dynamics (MD) simulations of this peptide suggest considerable unfolding during the ESI process⁶⁰.

To estimate differences in backbone reactivity, it is necessary to distinguish side chain contributions to HDX. It can be argued that the side chains should not significantly exceed the upper bound of ~50%. That is, because the coalesced droplets also contain water, the side chains would not significantly surpass this exchange level. However, the potential for the Serine and Lysine side chain groups to contribute differently to the overall exchange should be considered. To account for such differences, a multiple regression analysis (see above) obtained from 48 measurements of the HDX levels of mixtures of Alanine, Serine, and Lysine suggested slightly different contributions to HDX for the Serine and Lysine side chains (see Table S2). For the S-peptide, there are 26 exchangeable side chain hydrogens for the [M+2H]²⁺ ions. These are comprised of 10 Lysine hydrogens (including the two charges). These ions would also contain 16 exchangeable side chain hydrogens from the Serine residues. Based on the internal standard measurements (Table S2), the contribution from Lysine side chains to the total exchange would therefore be limited to ~5.7 deuteriums (10×0.571) for the doubly-charged ions. Using again the relative exchange parameters (Table S2), the exchange by the Serine side chains would be slightly higher at ~9.1 (16×0.567). Also, the relative contribution by the C-terminus hydrogen would be 0.215. Therefore, the total exchange occurring on side chain sites and C-terminus would not be expected to exceed 15 on average for doubly-charged ions of the S-peptide. The rest of the deuterium uptake (~2.5 hydrogens on average) would occur for backbone exchangeable sites. Table 1 shows the average estimated backbone hydrogen exchange for all observed charge states of the two model peptides. In general, the S-peptide is expected to show a greater overall exchange level given the larger number of side chain hydrogens available. However, the computed backbone hydrogen exchange depicts an interesting scenario. The S-peptide actually exhibits a lower protection level (increased deuterium incorporation) for the backbone hydrogens compared to the A-peptide even though they have the same number of backbone hydrogens. From this analysis, the in-droplet HDX experiment results in a decrease in deuterium incorporation for the A-peptide (Table 1) of ~2 on average.

It is instructive to consider the origin of the increased accessibility to exchange for the backbone sites of the S-peptide. Although tempting to ascribe such a difference to the random coil nature and helix forming propensity of the S- and A-peptides, respectively, such an argument is somewhat cursory. Indeed, exchange data for model dipeptides suggest that the HDX rate (k_ex) of the backbone hydrogen for an Alanine model compared to a Serine model is decreased by ~2 to 3 fold over a pD range of 4 to 6⁶¹. For the experiments described here, a ratio of estimated k_ex values (R(k_ex1/k_ex2) for backbone amides of the S- and A-peptides can be obtained from Equation 2:

$$R\left(k_{ex1}/k_{ex2}\right)=\left(\frac{\ln \left(\left(T_{BB}-S_{BB}\right)/T_{BB}\right)}{t}\right)/\left(\frac{\ln \left(\left(T_{BB}-A_{BB}\right)/T_{BB}\right)}{t}\right)$$ (2)

In Equation 2, k_ex1, k_ex2, T_BB, S_BB, A_BB, and t represent the k_ex for S-peptide, k_ex for A-peptide, total backbone hydrogens available for exchange (~1/2 of all backbone hydrogens), estimated backbone exchange for S-peptide, estimated backbone exchange for A-peptide, and the time of the reaction. Using a value of 10 for T_BB as well as the average A_BB and S_BB values (average of 2+ and 3+ ions) from Table 1, an exchange rate ratio of ~12 is obtained. This is considerably larger than the value (~2 to 3) estimated from the literature for dipeptide models⁶¹. However, caution should be used in suggesting that this arises from exchange protection (decreased deuterium incorporation) associated with helical structure for the A-peptide. Table 1 shows a sizeable error associated with the estimated backbone exchange numbers. Admittedly any protection that may be obtained from secondary structure is not easily discerned from these limited in-droplet experiments. That said, the disproportionate HDX of the side chain and backbone hydrogens for the two peptides suggests that further experiments are warranted to determine the degree to which secondary structure may inhibit HDX. Coupling in-droplet HDX with non-ergodic ion fragmentation techniques in the future may reveal the role (if any) of secondary structure in small peptides.

Assuming a t value of ~100 μs (roughly droplet lifetime)³⁶, k_ex values can be estimated for backbone HDX for the S- and A-peptides. Using numerator and denominator portions of Equation 2, values of 2.5×10³ and 2.0×10² s⁻¹ are obtained for k_ex1 and k_ex2, respectively. This analysis suggests an enhancement in k_ex of 10⁵ to 10⁶-fold for the in-droplet experiments compared with the bulk solution measurements for the dipeptides. This is quite remarkable but consistent with the findings of reaction acceleration factors in the literature⁶². Notably, the results presented here represent the first estimates of k_ex and acceleration factors for peptide backbone amide hydrogens upon undergoing HDX in microdroplets.

Enhanced bulk solution structure differentiation of the two model peptides by in-droplet HDX is perhaps compromised due to the ionization process. It has been argued that the ionization of peptides having low molecular weight are believed to proceed through the ion evaporation model (IEM)⁶³ where a high electric field at the surface of the nanodroplets causes the ejection of small, solvated ions. Such a process could render backbone hydrogens of small structured species more accessible due to structural disruption. This could be especially pronounced at the droplet surface where D₂O concentration may be higher as a result of incomplete mixing. Indeed, as mentioned above, prior MD simulations studies of the A-peptide suggest significant disruption of the helical structure within nanodroplets as they dry during the ESI process⁶⁰.

To determine whether or not increased backbone exchange can be obtained for the A-peptide upon disruption of the solution helix, experiments were conducted in which the peptide was incubated in solution containing 80% methanol. In-droplet HDX was then conducted and the degree of backbone exchange was estimated as described above. On average, for the A-peptide the number of incorporated deuteriums is 8.7±0.7. This represents an ~36% increase compared with the experiments utilizing the native solution conditions. For the S-peptide the number of incorporated deuteriums increases to 25.2±2.5 (~44% increase). Using the total exchange for the A-peptide as well as the increased exchange for the internal standard, the number of backbone hydrogens that exchange is estimated to be 0.5±0.7. This represents an ~2.5-fold increase relative to the A-peptide examined under native solution conditions. In comparison, the estimated backbone exchange number for the S-peptide was 5.0±2.5. This represents an ~2.3-fold increase relative to that determined from experiments employing native solution conditions. The similar factors would seem to suggest that the increased backbone exchange for the A-peptide may result simply from increased deuterium content in the microdroplets rather than helical disruption. That is, having only one hydrogen, the greater methanol content in the microdroplets would result in a decreased ratio of hydrogen:deuterium. Again, these results are consistent with the idea that the helix may already be disrupted during the ESI process for the A-peptide ions originating from native solution conditions.

In comparison to the ionization process for small peptides, relatively larger globular proteins are argued to experience ionization via the charged residue model (CRM) where the nanodroplets reach complete dryness to produce the ions⁶³. In principle, ion production via CRM would require more time than IEM to generate ions. For protein tertiary structure however, the timescale may not be sufficient to allow reaction of interior hydrogens which would only become accessible upon large conformer-transforming structural changes. Therefore, the application of multidevice cVSSI based in-droplet HDX has been further extended to globular protein analysis to gain insight into the effectiveness of in-droplet HDX to monitor protein structure transformations in solution. Here, structural transformation of tertiary structure is achieved using incubation in solutions with increasing methanol content followed by in-droplet HDX as conducted for the model peptides.

Isotopic distributions for deuterated [M+5H]⁵⁺ ubiquitin ions from different solutions.

Ubiquitin is a well-studied globular protein known to have a native structure (N-state) with high-density packing of different secondary structural elements like alpha-helices, 3₁₀-helices and beta sheets⁵⁴. An earlier investigation into the dynamics of ubiquitin (in solution) has revealed that it exercises a range of high-frequency motions on a nanosecond to microsecond timescale resulting in an ensemble of crystal structures (more than 30)⁶⁴. Ubiquitin has also been reported to have multiple discernible solution structures depending on the solvent conditions employed^53,65. Especially notable is the presence of a stable intermediate structure (A-state) that has been recorded for ubiquitin when dissolved in a solvent composed of 60% methanol in acidified water^66,67. The A-state is a partially-folded conformer that precedes the unfolded state (U-state) and has traces of secondary structure from the N-state^66–71. Ion mobility spectrometry (IMS) studies demonstrated the ability to sample potentially similar coexisting solution structures of the [M+8H]⁸⁺ ubiquitin ions for a series of water: methanol solvent compositions⁵³. For such solution conformer heterogeneity, it is instructive to evaluate the utility of in-droplet HDX characterization achieved by multidevice cVSSI considering its efficiency in rapid deuterium labeling.

In this work, ubiquitin dissolved in different solvent systems has been subjected to HDX using the above-mentioned experimental setup. Methanol has been added to ammonium acetate buffer at five different concentrations (0%, 10%, 20%, 40%, and 80%) to induce differing degrees of protein denaturation. With an increasing amount of methanol, more of the compact, globular structures are expected to form partially-folded species and as a result more exchangeable hydrogens become accessible for HDX reactions. Thus, the deuterium uptake level from the in-droplet (solution-phase) HDX would reflect the increased diffuse nature of the protein structures. When sprayed from only ammonium acetate without any organic solvent, the [M+5H]⁵⁺ ions are the dominant mass spectral feature of ubiquitin as shown in Figure S3 in the Supporting Information section. Indeed, until a solvent system containing 80% methanol is used, the low charge states ([M+4H]⁴⁺, [M+5H]⁵⁺, and [M+6H]⁶⁺ ions of ubiquitin) dominate the mass spectrum. Figure 3A-E show the evolution of the deuterated [M+5H]⁵⁺ ubiquitin ions as a function of added methanol content. The isotopic distribution of such ions begins to resolve into multimodal isotopic envelopes after deuterium exchange with the increasing presence of methanol in solution. It is here suggested that these observed features result from the distinctive deuterium uptake of different coexisting conformation populations in solution. The ions that exhibit higher m/z shifts (or deuterium uptake) after in-droplet HDX reactions are argued to consist of solution structures that are unfolded to some degree from the denaturation effects of added methanol.

Figure 3.

Evolution of the expanded mass spectral feature for [M+5H]⁵⁺ ubiquitin ions after in-droplet HDX as a function of methanol content in solution (A to E). Gaussian peak fitting model for the deuterated conformer curve (black circles, see text for details) of [M+5H]⁵⁺ ubiquitin ions ionized from five different solution compositions (F to J). The purple line trace shows the cumulative fit curve. Red (P1), green (P2), blue (P3), cyan (P4), pink (P5), yellow (P6), khaki (P7), and dark blue (P8) lines are fitted Gaussian peaks for each isotopologue grouping (see text for details) in an increasing order of average m/z value. Solution compositions have been labeled for each expanded mass spectrum (A to J) as volume percent methanol (% MeOH) in 50 mM ammonium acetate solution.

Representing isotopologue features having different m/z shifts as distinct conformer populations.

To evaluate the relative abundances of the proposed coexisting conformers, the deuterated isotopologue base points (troughs) of the low charge state ions have been further plotted (conformer curve) and these traces have been peak fitted using gaussian functions. In this analysis, the conformer curve for [M+5H]⁵⁺ ions from ubiquitin in 80% methanol has the greatest number of peak attributes with discernable centers and magnitudes (Figure 3E). Such centers have been manually selected by visual inspection of the conformer curves and an optimized peak width has been used to obtain a cumulative fit curve that comprises a minimum number of gaussian peaks. In this treatment, each of these peaks would represent a hypothetical conformation population with distinct deuterium labeling. The same peak analysis parameters have been used to individuate all of the low charge state ([M+4H]⁴⁺, [M+5H]⁵⁺, and [M+6H]⁶⁺ ions) mass spectral features from different solvent compositions into conformational sub-features as shown in Figure 3F-J for [M+5H]⁵⁺ ions.

Duplicate experiments provide a sense of reproducibility for the spectral fitting approach. Figure S4 in the Supporting Information shows the isotopic distributions for [M+5H]⁵⁺ ubiquitin ions that have undergone HDX for samples having methanol content of 40% and 80%. Here the isotopic distributions have also been fit with the Gaussian distributions (Figure S4). On average, the intensities differ by ~20±40% between experimental runs (Figure 3 and Figure S4). The large standard deviation results from greater disagreement for the lower intensity Gaussian peaks. That said, an inspection of the data shown in Figure 3 and Figure S4 suggests that, similar to the peptide studies, relatively high reproducibility for these experiments is achieved.

Conformer population conversion as a function of methanol.

The relative abundances of fitted peaks in Figure 3E-J represent differences in populations of coexisting solution structures as distinguished by the m/z of the peak center. From the HDX-MS data for ubiquitin in 0% methanol, it appears that multiple conformers may already be present in solution (Figure 3F). Four characteristic peaks with the lowest m/z values represent most of the conformer curve. The isotopologue features in the conformer curve having higher hydrogen exchange for deuterium are relatively low though which may be somewhat anticipated as compact native solution structures might be expected to dominate the overall protein population. As methanol is introduced to the solution, the conformer population peaks that are located in the medium to high m/z range of the conformer curve, become more intense (Figure 3G-J). This may be interpreted as the solution structures of ubiquitin become more diffuse due to methanol-induced denaturation and more hydrogens become available for HDX. The greatest number of conformer features are observed for the 80% methanol solvent system where those with relatively high m/z are significantly more pronounced compared to the other solutions. To better understand how the proposed conformational populations evolve with the increasing amounts of methanol, a comparison of peak area has been conducted for all of the low charge state (4+, 5+, and 6+) ions as shown in Figure 4.

Figure 4.

Normalized peak areas as a function of methanol content. The charge states utilized in these summed intensities include the 4+ to 6+ for all different solvent compositions. Red squares, green spheres, blue triangles, cyan triangles, pink rhombuses, yellow triangles, khaki triangles, and dark blue hexagons represent normalized peak areas of the P1, P2, P3, P4, P5, P6, P7, and P8 features (see text for more details) in Figure 3, respectively.

The Gaussian fitting that has been used for the data analysis of individual charge states utilizes the same width for all of the selected peaks as a first approximation of similar exchange behavior (peak broadening) by different conformers; different conformational populations are suggested to encompass approximately the same range of isotopologues (i.e., the same mass range). The peaks obtained from the Gaussian fitting are labeled in increasing order of their peak center m/z value as P1, P2, P3, P4, P5, P6, P7 and P8 respectively. The deuterium incorporation values that would be represented by the m/z shifts associated with P1 to P8 of the [M+5H]⁵⁺ ions are presented in Table S3 in the Supporting Information section. For the low charge state ions, the Gaussian fitting has been performed with adjusted peak widths and peak centers according to the respective charge states. The peak widths are scaled by charge state to maintain the overall mass range of each conformer type. The peak centers are shifted to represent the same deuterium incorporation levels for the different charge states. Conformer peak areas for those with the same amount of deuterium incorporation from different charge states have been combined in order to represent the same conformer population in this treatment. Normalized peak areas are plotted against the added methanol content as shown in Figure 4. The contributions from different peaks to the conformer curve are presented as a function of different methanol compositions in solution. In one consideration of the data, P1 and P2 are assumed to represent native-like solution structures of ubiquitin that differ little in their deuterium incorporation (Table S3). Overall, P1 and P2 decrease with increasing methanol content. The normalized areas of remaining peaks rise with the amount of methanol (from 10% to 80%) possibly indicating the emerging/increasing presence of partially-unfolded structures. With 80% methanol in solution, the normalized area of the peaks with higher deuterium incorporation increases. This may be attributed to a greater conversion of the low-exchanging compact solution structures to more diffuse, higher-exchanging structures.

The impact of higher methanol on protein structure becomes more evident when higher charge states are included in the analysis. Up to the addition of 40% methanol, no charge state greater than the 6+ is observed under HDX conditions. Higher charge states including the 7+ to 12+ species are only observed for samples incubated in 80% methanol. Figure S5 shows the normalized peak area as a function of methanol content when the high charge states observed for the 80% methanol solvent composition are considered. The conformer populations after the Gaussian peak fitting of the high charge states are labeled separately as P9, P10, P11, P12, P13, and P14 in Figure S5. These non-native conformers comprise more than 70% of the total protein ion population. Notably, this indicates that the conformer peaks representing native structure are overemphasized in Figure 4 and are merely shown in this manner to clearly indicate a decrease in intensity.

One consideration of this treatment of the data is the effect (if any) that adducts might have. Figure S6 in the Supporting Information shows the adducts associated with the A-peptide prior to producing D₂O reagent microdroplets. Evident are ions having Na⁺ and K⁺ adducts with relative intensities of ~5% and 3%, respectively. With the addition of D₂O reagent, these ions display relatively unchanged intensities of ~5% and 3.5%, respectively. This behavior is essentially the same for the S-peptide. Figure S6 in the Supporting Information also shows the adducts observed for the [M+5H]⁵⁺ ions of ubiquitin. Here the intensities of the ions having Na+ and K+ adducts are ~9% and 3%, respectively. Upon addition of D₂O, the isotopic distributions broaden as indicated in Figure 3. At this point it is not possible to resolve the Na⁺ and K⁺ adducted species. For example, in Figure 3A and 3F, the maximum of the isotopic distribution occurs at ~m/z 1717.3. The Na⁺ adduct ion would then be expected to have a maximum at ~m/z 1721.7. At this higher m/z, the isotopologue features are nearly 20% of the most intense ions indicating that the tailing observed in Figure 3A cannot be assigned to adduct peaks alone. That is, many protonated species must have shifted to higher m/z to account for this increase. Additionally, if a portion of the protonated species has shifted to higher m/z, so too, it may be expected, have adduct ions thereby further decreasing their presence at ~m/z 1721.7. Clearly, the dramatic growth of the features at ~m/z 1721 and 1722.5 (arrows in Figure 3D-E and Figure S4) cannot correspond to adduct ions as the addition of D₂O does not lead to appreciably greater production of such ions (Figure S6C in the Supporting Information and Figure 3A). Having said that, because of the overlap in isotopic distributions, it is not possible to deconvolute the minor contribution by adduct species. It is here recognized that small contributions from broad isotopic distributions of the adduct ions will contribute a small portion to the intensities derived from the Gaussian peak fitting.

Comparison with IMS data.

Prior IMS-MS studies have revealed a number of co-existing solution structures of ubiquitin in acidified aqueous solutions containing varying amounts of methanol. Remarkably, conformations yielding largely extended ions having collision cross sections consistent with the intermediate (A-state conformation) of ubiquitin were observed even in the absence of methanol. As the methanol content increased it was observed that [M+8H]⁸⁺ ion conformers arising from the A-state and unfolded (U) state are favored over those from the native solution structure (N-state). Features corresponding with the N-state are demonstrated to disappear completely in the gas phase for a 30% methanol in acidified water solution. In contrast, the data reported here for the HDX solution scans present some notable differences. First, low levels of HDX are retained even at the highest methanol content for the [M+5H]⁵⁺ ions. It would be somewhat remarkable that a denatured protein would protect a large number of hydrogens. That said, the proposed A-state structure exhibits a relatively high degree of secondary structure character which could prevent HDX to some degree as could any related state. We also cannot rule out the possibility that a non-native compact structural form that provides for a degree of protection (decreased deuterium incorporation) persists at high methanol content. Indeed, the IMS studies suggest the presence of compact structures even in acidified solutions containing methanol.

It is also possible that some native structures are preserved even at high methanol content. Here, it is instructive to consider whether or not the peaks representing near-native HDX decrease in proportion to the overall charge state distribution for ubiquitin. For example, for ions from pure aqueous samples with no methanol, the [M+5H]⁵⁺ ions comprise nearly 60% of the mass spectral features for ubiquitin. This number decreases to ~10% with the addition of 80% methanol. Thus, under the 80% methanol conditions, the [M+5H]⁵⁺ ion features possibly representing HDX reactivities of near native species would comprise only ~5% of the entire ion distribution (See Figure S4). Such low levels of native or near native conformer types would be easily missed by standard structure characterization techniques. In a manner this is similar to the ability of IMS to detect the limited A-state structure in pure aqueous solutions although the latter represents a post-ionization measurement.

Establishing solution-phase HDX behavior.

In evaluating the utility of in-droplet HDX for distinguishing solution reactivities, it is instructive to consider the influence of any gas-phase HDX that may occur in the ion source and ion transfer optics regions of the mass spectrometer. Consideration of the former is important as carefully crafted experiments have shown that appreciable solvated-ion exchange occurs from residual D₂O vapor in an ion source region⁷². The question arises as to whether or not gas-phase exchange could result from any such legacy source of reagent in the dual cVSSI device setup. Prior experiments have shown that upon stopping the vibration of the cVSSI device for D₂O droplet plume generation, the HDX reactions for standard compounds are essentially eliminated indicating that no delayed reaction occurs in the ion source or ion transfer optics region for these experiments⁴⁵. Presumably the open nature of the cVSSI setup removes any other source of D₂O compared with an enclosure that may have surface release of reagent.

A related concern arises as to whether or not gas-phase HDX could occur in subsequent instrument regions due to the overall solvent makeup of the droplets being introduced during the experiment as well as the ambient water content. Such a phenomenon has been demonstrated with well-designed experiments for traditional HDX-MS studies⁷³. In the prior work, it was shown that a source of infused solvent could alter isotopic distributions and that these could be eliminated with careful attention to instrument parameters. The prior cVSSI-HDX-MS work again suggests that the observed HDX reactivity is reflective of solvated molecules. For example, prior studies conducted in an identical manner have shown HDX behavior that is similar for [M+H]⁺ and [M+2H]²⁺ bradykinin ions⁴⁵. As singly-charged bradykinin essentially does not undergo gas-phase HDX^74,75, it can be argued that the observed m/z shift results from primarily solution exchange. Additionally, that the isotopic distribution for the doubly-charged ions is not significantly altered with regard to HDX levels displaying a very similar level of exchange compared with the singly-charged ions also suggests that the measurements primarily reflect solution exchange reactions. Separate experiments show that sodiated glycans undergo a very similar level of exchange as compared to their protonated counterparts⁴⁵. Because the sodiated, gas-phase ions should not undergo HDX via the generally accepted relay mechanism⁷⁶, it can again be argued that the observed exchange occurs primarily in the droplet environment.

Another argument for solution-phase exchange results from the fact that the magnitudes of the mass spectral shifts for the [M+5H]⁵⁺ ubiquitin ions are very different than typically observed for gas-phase HDX reactions of proteins. Previous studies by McLafferty and coworkers have suggested an ~100% increase in the number of reactive hydrogens from compact (native-like) to unfolded states during gas-phase HDX of cytochrome c⁷⁷. Separate HDX-MS experiments conducted for cytochrome c in which the reaction occurs in a drift tube suggest at most an ~45% increase in the number of reactive hydrogens between compact and unfolded states^78,79. Also, separate conformational investigations of ubiquitin anions using IMS-HDX have shown an increase in deuterium incorporation level between the elongated and compact conformations species which is at best ~70%⁸⁰. For the data presented here however, the higher exchanging features in this study exhibit exchange level differences (~400%) that well exceed the gas-phase values for compact and elongated ubiquitin ions under saturation conditions^80,81. Taken together, these results suggest that the observed shifts primarily reflect solution exchange.

A final concern with regard to solution HDX is whether or not the observed changes in the isotopic distributions for ions such as the [M+5H]⁵⁺ species result simply from differences in the change in the number of available hydrogens from the solution. That is, as the methanol content increases, the overall hydrogen:deuterium ratio changes resulting in a greater amount of available deuterium. Two observations suggest that this does not affect the ability to distinguish different conformer populations for proteins. The first observation is that the leading edge (lower m/z) of the isotopic distributions remains relatively unchanged occurring at ~m/z 1715 in every case (see Figure 3). Indeed, from the Gaussian fitting analysis, Peak 1 would represent an exchange of ~12.5 hydrogens. From the MDS analysis, there are ~64 side chain hydrogens that are within 5Å of water molecules for the native conformation. Thus, only ~20% of these are accessible for exchange for the low-exchanging ions (leading edge of the distributions in Figure 3). An analysis of expected exchange for side chain hydrogens similar to that described for the model peptides above, suggests that ~27 hydrogens (~42%) should exchange. As not all side chains could be estimated by the parameters in Table S2, parameters for other side chain moieties such as those found on Arginine residues are obtained from the relative contributions calculated previously⁴⁴. The decreased level of experimental deuterium incorporation suggests some tertiary protection. As this isotopic distribution leading edge does not shift at higher methanol content (still commencing at m/z ~1715 in Figure 3), this protection (decreased deuterium incorporation) remains for a portion of the molecules despite higher solution deuterium content suggesting continued tertiary protection.

The second observation indicating a minimal role for the increased droplet deuterium content is the observation of dramatic changes in the overall shape of the isotopic distributions. For example, the abundances of isotopologues centered at m/z 1721 and 1723 (arrows in Figure 3) increase relative to those at the leading edge resulting in a dramatic change in the overall isotopic distribution shape. If the solution HDX simply increased due to more available deuterium, the entire distribution at 0% methanol should simply shift to higher m/z values as is observed for the peptides (see discussion above). Therefore, the systematic change in the overall shape of the isotopic distribution with increased methanol content is another indicator of capturing different solution behavior of various conformers. Notably, at higher methanol content, Peaks 5, 6, 7, and 8 increase in relative abundance and represent deuterium incorporation values that well exceed 27 suggesting that many initially protected side chain hydrogens become accessible with increasing methanol content for a portion of the molecules. Note also that this shift away from the leading edge to these peaks represent a >3 fold change in deuterium incorporation. This is well above the ~35% to 45% increase for the peptides which is suggested to result from changes in solvent composition (see discussion above).

Admittedly, some gas-phase exchange, or even increased exchange near the droplet surface cannot be ruled out. Despite this possibility, the peptide data presented here suggest that it is possible to distinguish species based on differences in deuterium uptake associated with structural differences (primary sequence and possibly secondary structure to a limited extent) and the protein data provides the same capability based on tertiary structure arrangement differences.

Conclusions

The solution deuterium uptake of two structurally distinct model peptides has been reported for in-droplet HDX reactions achieved by utilizing two field-enabled cVSSI devices. The unstructured S-peptide displays increased exchange for backbone hydrogens. However, the decreased exchange of backbone hydrogens for the A-peptide may result in large part from differences in intrinsic exchange rates (k_ex) for the Alanine and Serine residues. That said, one interesting outcome of the in-droplet HDX results for the model peptides is the potential to determine intrinsic k_ex estimates for backbone hydrogens that become accessible in in-droplet reactions of different peptides. The HDX experiments have been further implemented to obtain isotopic distributions for deuterated ions of ubiquitin originating from five different solvent compositions. The expanded mass spectra of [M+4H]⁴⁺, [M+5H]⁵⁺, and [M+6H]⁶⁺ ubiquitin ions show an increase in intensity for isotopologue features having high m/z shifts as the methanol content of the solution is increased. A peak-fitting method that employs Gaussian functions to model the entire isotopic distribution has been used to estimate the minimum number of solution-phase conformation types formed as a function of the denaturing methanol content. Notably, this model reveals a number of co-existing solution structures existing under native solution conditions. Additionally, the model indicates that a small portion of native structure may persist even at high methanol content. This analysis highlights the ability of multidevice cVSSI-HDX-MS to characterize coexisting conformer populations and detect structural transformations of a protein in solution. It should be noted that, as performed here, in-droplet HDX-MS allows the characterization of solution conformations of proteins and any structural transformation of the ions in the gas phase cannot be observed. In that regard, in-droplet HDX may prove invaluable for the study of unstable gas-phase structures that should not be interrogated for solution structure details post ionization.

In the future, the utility of rapid, in-droplet HDX may be extended by demonstrating the ability to locate sites of deuterium incorporation. Several studies have shown that ion fragmentation by non-ergodic methods such as electron transfer dissociation can mitigate against H/D scrambling thereby preserving the sites of deuterium incorporation^82–90. Additionally, further experiments to delineate the roles of solution flow rate, applied voltage, emitter tip distance (reaction time), and solvent composition on the overall HDX levels would provide valuable information that can be used to tailor experiments for the study of specific biomolecular systems. In the future it may be possible to extend HDX characterizations to challenging systems such as proteins containing intrinsically disordered regions, protein conformational changes, as well as conformation formation for functional oligonucleotides.