Characterizing Multidevice Capillary Vibrating Sharp-Edge Spray Ionization for In-Droplet Hydrogen/Deuterium Exchange to Enhance Compound Identification.

Multidevice capillary vibrating sharp-edge spray ionization (cVSSI) source parameters have been examined to determine their effects on conducting in-droplet hydrogen/deuterium exchange (HDX) experiments. Control experiments using select compounds indicate that the observed differences in mass spectral isotopic distributions obtained upon initiation of HDX result primarily from solution-phase reactions as opposed to gas-phase exchange. Preliminary studies have determined that robust HDX can only be achieved with the application of same-polarity voltage to both the analyte and the deuterium oxide reagent (D2O) cVSSI devices. Additionally, a similar HDX reactivity dependence on the voltage applied to the D2O device for various analytes is observed. Analyte and reagent flow experiments show that, for the multidevice cVSSI setup employed, there is a nonlinear dependence on the D2O reagent flow rate; increasing the D2O reagent flow by 100% results in only an ∼10-20% increase in deuterium incorporation for this setup. Instantaneous (subsecond) response times have been demonstrated in the initiation or termination of HDX, which is achieved by turning on or off the reagent cVSSI device piezoelectric transducer. The ability to distinguish isomeric species by in-droplet HDX is presented. Finally, a demonstration of a three-component cVSSI device setup to perform multiple (successive or in combination) in-droplet chemistries to enhance compound ionization and identification is presented and a hypothetical metabolomics workflow consisting of successive multidevice activation is briefly discussed.

Introduction

Metabolites are low-molecular-weight compounds that are found in metabolic processes within living organisms.[1,2] Playing roles in such vital, cellular processes, a cell’s full metabolite complement has been argued to reflect its phenotype.[3] Thus, it can be argued that the metabolite complement may be a harbinger of physiological state be that at the cellular, tissue, organ, or organism level.[4−6] Indicators of physiological state would come in the form of changes in the relative abundances of such molecular species within different biological samples. The field of metabolomics is dedicated to the characterization and comparison of the metabolite complement among different samples.[2,7]

Because metabolites are among the most diverse biomolecules,[8,9] full characterization of this molecular component from biological samples is extremely challenging. First, the sheer number of compounds exceeds the peak capacities of many analytical strategies.[10] Second, the wide range of physicochemical properties associated with the various compounds ensures that no one analytical strategy is highly suited for the study of all compounds. Finally, metabolites exist over a wide range of cellular concentrations which fluctuate in time.[11] These traits of metabolites have led to a variety of analytical strategies utilized in metabolomics experiments. The most often-used approaches are NMR spectroscopy[12−14] and mass spectrometry (MS) combined with gas or liquid chromatography (GC or LC).[15−19] Each of these approaches has advantages and disadvantages in the study of metabolites; however, all struggle to identify and quantify large numbers of compounds in complex mixtures due to the characteristics mentioned above. Indeed, the challenge is so acute that the National Institutes of Health recently issued a funding opportunity announcement with the goal of advancing new technologies that would provide fuller characterization.[20]

One way to improve compound identification capabilities is to incorporate measurements of different physicochemical properties of the same molecules within a sample. In a manner, the combination of GC and LC with MS accomplishes this to some degree as information related to volatility and polarity can be combined with mass. That said, the complexity of metabolomics samples[21] greatly exceeds the peak capacities of these hyphenated techniques prompting a desire for new measurements. This has led to the development of elegant (yet somewhat technically challenging) multidimensional chromatography approaches for metabolomics analyses.[22−24] More recently, ion mobility spectrometry (IMS) has gained favor as it provides new information in the form of a collision cross section (CCS) value;[25−29] however, CCS only provides some improvement as CCS is correlated with mass leading to a reduction in distinguishing power.[30−34] One could argue that a powerful approach would be combining measurements with the ability to elucidate organic functional groups (e.g., NMR) with relative volatility (e.g., GC) or polarity (e.g., LC) and mass would present tremendous advantages. One approach offering somewhat similar information that can be performed online with liquid chromatography–mass spectrometry (LC–MS) is hydrogen/deuterium exchange (HDX).[35−38] Here, the goal is to elucidate the number of heteroatoms within compounds based on the numbers of deuteriums that exchange for hydrogens resulting from the solution-phase reaction. Overall, remarkable enhancements to compound identification are reported.[35−38] A limitation of the online approaches (on-column and post-column HDX) is that they do not allow for alternating mass spectral measurements of unreacted and exchanged ions. Because it is not possible to obtain precursor ion exact mass and HDX reactivity within a single LC run, sequential separation runs are required.

Here, we investigate the ability to perform in-droplet HDX that can be initiated and terminated instantly and automated to allow for cycling between unreacted and exchanged ion mass spectral measurements. The approach takes advantage of the new ionization process termed vibrating sharp-edge spray ionization (VSSI),[39,40] which eliminates the need for a nebulization gas for ion generation such as that required by electrospray ionization (ESI). Removal of the nebulization gas requirement allows the utilization of multiple capillary vibrating sharp-edge spray ionization (cVSSI) emitter tips at the front of a single mass spectrometer, where each effectively nebulizes a different solvent system. In the studies described below, a multidevice cVSSI source is characterized for its ability to perform preionization, in-droplet HDX. In its simplest form, a single cVSSI device is used to nebulize and ionize the analyte while a second device is used to nebulize D2O reagent solvent. Droplet recombination occurs at the inlet region of a mass spectrometer, and the solution HDX for different analyte molecules is measured with the mass spectrometer.

The performance of in-droplet HDX using cVSSI benefits from groundbreaking experiments suggesting that reaction rates for different reactions could be increased by orders of magnitude when occurring in microdroplets as opposed to a bulk solution environment.[41−44] The most influential experiments to this work are those that utilized theta capillaries to monitor processes such as complex formation, HDX, protein unfolding, and ion supercharging using mass spectrometry.[45−47] In one study, the HDX rates of different amino hydrogens were monitored for a small molecule using mass spectrometry.[48] This study, in particular, prompted recent experiments using cVSSI in a manner similar to nano-electrospray ionization (nESI) with theta capillaries to obtain unique isotopic distributions for different low-molecular-weight compounds.[49] A problem with using the theta capillary approach to obtain exchangeable hydrogen information for metabolomics experiments is that HDX cannot be instantaneously turned on and off, making it difficult to effectively sample compounds eluting from LC columns.

A major motivation for investigating in-droplet HDX accomplished using separate cVSSI devices is that it should be possible to instantaneously turn HDX reactions on and off. That is, a difference between the theta capillary work is that here two, well-separated emitters are used for the analyte solution and deuterated reagent. Because droplet delivery of the deuterated reagent depends on emitter tip vibration rather than applied direct current (DC) voltage or solvent flow, it can be initiated or terminated instantaneously by turning on or off the mechanical vibration. Additionally, cVSSI is already shown to provide ionization efficiency gains for LC–MS analyses of metabolite standards.[50] Thus, the combination of multidevice cVSSI with LC–MS for metabolomics analyses has the potential to not only provide new information to aid compound identification but also provide access to a greater number of metabolites resulting from its increased sensitivity.

Finally, although the primary goal of LC-HDX-MS in metabolomics analyses is to determine the number of heteroatom sites on different molecules, it may also be possible to obtain additional distinguishing information. One example discussed in previous cVSSI work is the ability to generate unique isotopic distributions for compounds based on differences in exchange site reactivities.[49] In this scenario, it can be envisioned that such isotopic distributions could be matched to those in databases similar to searches performed with GC–MS data. Additionally, the development of computational techniques to predict HDX isotopic distributions for molecular structures could be very useful to identify compounds that do not exist in databases such as newly emerging drugs and their metabolites. This idea extends from seminal work that has shown the ability to obtain unique isotopic distributions for isobaric and isomeric compounds using gas- and solution-phase HDX.[51−54] This technical report describes investigations of factors influencing in-droplet HDX as well as its potential to aid compound identification in metabolomics analyses. A final analysis describes a hypothetical experimental sequence for conducting enhanced metabolomics analyses where the unique advantages of a multidevice cVSSI source are highlighted.

Results and Discussion

Achieving In-Droplet HDX

To initiate the in-droplet HDX experiments, the analyte solution is infused through the cVSSI device at a flow rate of 5 μL·min–1. The appropriate radio frequency (RF) voltage is applied to the analyte cVSSI device to initiate droplet plume formation using the actuator switch box (Figure ). The plume is visible to the naked eye. Next, the DC voltage (∼±1800 V) is applied to the droplet analyte cVSSI device. Figure shows the mass spectra that are generated for [M + H]+ serine, acetaminophen, and arginine ions when the analyte cVSSI device is activated in this manner. Upon demonstrating strong ion signals, the D2O is infused through the D2O reagent cVSSI device and the RF voltage for this device is turned on. Next, a DC voltage is applied to the D2O reagent (typically ±600 V for the source setup in Figure ). This DC voltage is found to be necessary as without its application, analyte droplets will condense on the D2O cVSSI device and effectively shut off vibration of the tip and reagent plume generation.

Figure 1

Photographs of the multidevice cVSSI ion source setup. (A) Expanded view of the dual device setup. Labeled are the component parts required to conduct the in-droplet HDX experiments (see the Experimental Section for details). (B) Close-up view of the dual cVSSI device arrangement. The devices (microscope slides with attached pulled-capillary tips) used to create analyte and D2O reagent plumes are labeled.

Figure 2

Representative mass spectra for cVSSI-MS and cVSSI-HDX-MS. Experimental isotopic distributions for the [M + H]+ ions of serine, acetaminophen, and arginine produced by cVSSI and recorded on the orbitrap mass spectrometer. Data for the respective analytes are shown in (a–c). In each panel, mass spectra on the left and right correspond with conditions under which the D2O reagent is turned off and on (see text for details). Analytes and reagent activation states are labeled.

Upon generating D2O droplets, new isotopologue peaks are observed for the analyte ions as shown in Figure . For these experiments, the D0, D1, D2, D3, and D4 isotopologue ions are observed for the [M + H]+ acetaminophen, serine, and arginine ions. Additionally, the D5 isotopologue is observed for the arginine ions. Using eq , the amount of deuterium incorporated for the respective ions is ∼1.9, ∼1.2, and ∼2.4, respectively. This calculation is shown for [M + H]+ acetaminophen ions in the Supporting Information. The [M + H]+ serine, acetaminophen, and arginine ions have 5, 3, and 8 exchangeable hydrogens, respectively (Table ). Considering this, the respective fractions of deuterium incorporation for the ions from these experiments are ∼0.4, ∼0.4, and ∼0.3. For an example calculation of fraction deuterium incorporated, see the Supporting Information.

Table 1

Molecules Used in cVSSI-HDX-MS Experiments

Name of the compound.

Chemical formula of the compound.

Molecular weight of the compound. Average masses are reported.

Chemical structure. Small-molecule structures were obtained from Scifinder at https://scifinder.cas.org. The bradykinin structure was obtained from: By Yikrazuul (talk)—Own work, Public Domain, https://commons.wikimedia.org/w/index.php?curid=15555663.

Number of exchangeable hydrogens for the observed ions. Note that bradykinin has two numbers for [M + H]+/[M + 2H]2+ ions.

Isotopic Patterns Primarily Reflect Solution Exchange

A question arises as to the degree that gas-phase HDX may be occurring for experiments such as those shown here. Several pieces of evidence suggest that solution-phase exchange is primarily revealed by the isotopic distribution. First, consider the data presented for [M + H]+ and [M + 2H]2+ bradykinin ions shown in Figure . As with the small molecules shown in Figure , upon activating the analyte and D2O cVSSI devices, the isotopic distributions for these ions shift to higher m/z values. For the singly-charged ions, the calculated deuterium incorporation value is about two deuteriums. Additionally, based on the highest observable isotopologue in the unexchanged and HDX datasets, it is noted that some ions incorporate up to three deuteriums. This is notable because singly-charged bradykinin does not react to any degree with D2O in the gas phase.[55,56] This most likely results from the formation of a salt-bridge structure[57,58] that does not allow the formation of a critical reaction intermediate associated with the proposed relay mechanism[59] for HDX. Briefly, such a structure is proposed to result from a deprotonated carboxy terminus bridging two charged arginine side-chain residues. Indeed, these charge interactions may impart a significant degree of structural rigidity to [M + H]+ bradykinin ions as ion mobility spectrometry experiments have shown similar collision cross section values (ion size) for such ions even at elevated (∼600 K) temperatures.[60] The relay mechanism requires simultaneous hydrogen bonding of the D2O reagent at a protonated site on the peptide ion and a less basic site.[59] Therefore, the conformational rigidity may prohibit access of the protonated side chains to multiple, less basic sites on the peptide where deuterium incorporation would occur. Indeed, multiple studies have argued that such access is required[58,61−64] and temperature-dependent gas-phase HDX studies with molecular dynamics simulations suggest that increased ion dynamics (conformational flexibility) allows greater access to incorporation sites by charge sites and thus increased levels of deuterium incorporation.[65]

Figure 3

Mass spectral evidence of solution-phase HDX. (a, b) Expanded mass spectral regions for the [M + 2H]2+ and [M + H]+, respectively, bradykinin ions recorded using the linear ion trap mass spectrometer. Solid and dashed line traces show the isotopic distributions during D2O inactivation and activation, respectively. Ions corresponding to the spectra are labeled. The bar graph in (c) shows the relative isotopologue intensities for [M + H]+ and [M + K]+ 6′-sialyllactose ions for each nominal m/z value. These data were obtained upon activation of the D2O reagent cVSSI device. The legend shows the bar color corresponding to the ion type.

Although the data for the [M + H]+ bradykinin ions show that solution-phase exchange must be occurring for these conditions, it does not show that gas-phase exchange could not occur for other ions. To examine whether or not gas-phase exchange affects the isotopic distributions, we can consider the [M + 2H]2+ bradykinin ions shown in Figure . This charge state is reported to undergo substantial HDX in the gas phase.[55,56,66] Remarkably, for these experiments, the [M + 2H]2+ ions show almost the exact same amount of deuterium uptake compared with the [M + H]+ ions. This would be highly unusual if these ions do undergo further gas-phase exchange. If that were to occur, it would be more likely that the incorporated deuterium would be different than that observed for the [M + H]+ ions.

In separate studies, the deuterium uptake was compared for [M + H]+ and [M + K]+ 6′-sialyllactose ions generated from the same analyte solution. Figure shows a bar graph representation of the two isotopic distributions that are obtained upon performing in-droplet HDX. These distributions are very similar in terms of their overall width and the relative intensities of the isotopologues. Indeed, the deuterium uptake for these two ion types is calculated to be ∼2.1 and ∼1.8 for the protonated and potassiated species, respectively. It is important to note that the potassiated ions are not expected to undergo gas-phase HDX as they cannot form the critical, hydrogen-bonded intermediate required by the relay mechanism[59] (see above). Admittedly, there are some small differences in relative isotopologue intensities on the sides of the overall distribution. Thus, a small degree of gas-phase exchange cannot be ruled out.

As a final note regarding gas-phase exchange, it is instructive to consider the HDX behavior of analyte ions as a function of applied voltage to the analyte cVSSI devices. Limited experiments show that the application of ∼±600 V to the D2O reagent creates a threshold voltage for the analyte that must be superseded to induce HDX. That is, below this voltage, analyte ions are observed in the mass spectra but no deuterium incorporation is observed. It is believed that a high voltage must be applied to the analyte to provide the necessary momentum for combination of droplets of like charge. However, the fact that the ions are not observed to undergo HDX prior to this onset voltage again suggests that the m/z shifts in the exchanged spectra reflect primarily a solution-phase process.

HDX Dependence on Applied Voltage and D2O Reagent Flow Rate

In considering the utility of performing in-droplet HDX in metabolomics studies, it is important that the approach is highly reproducible such that the same results can be obtained by different research groups as well as across different MS instrumentation. Thus, it is important to consider the cVSSI source parameters that may affect the overall deuterium incorporation levels. One factor that may influence HDX as it affects the field experienced by the charged droplets and thus their overall momentum is the voltage applied to the D2O reagent device. As mentioned above, droplet momentum is believed to affect droplet mixing and thus D2O incorporation. Other factors affecting D2O incorporation are likely the relative size of reagent droplets and their overall relative prevalence. Thus, the D2O reagent flow rate may also affect the degree of deuterium incorporation.

In a first set of parameter-testing experiments, the voltage applied to the D2O cVSSI device was stepped from +200 to +1200 V in 200 V increments. Figure shows the fraction of deuterium observed to be incorporated for the [M + H]+ serine, acetaminophen, and arginine ions at these separate voltage settings. Here, the voltage on the analyte cVSSI device was maintained at +1800 V. Overall, the same response to the different applied voltages was observed for the three analytes. The fraction of deuterium incorporated (see the Experimental Section) increases as the applied voltage is adjusted from +200 to +800 V and thereafter the incorporation level decreases. The change in fraction of deuterium incorporated over the range of +200 to +800 V is the largest for acetaminophen ions and the smallest for serine ions. The former analyte shows a ∼50% change, while the latter demonstrates an ∼30% change. Above +800 V, the arginine ions show the greatest decline (reduction of ∼30%) in deuterium incorporation. Overall, the fraction of deuterium incorporation is noticeably less for arginine, which is consistent with prior in-droplet HDX studies using a voltage-free theta capillary-like approach.[49]

Figure 4

HDX dependence on applied voltage to the D2O cVSSI device. The bar graph shows the fraction of deuterium incorporated for three small-molecule analytes as a function of the DC voltage applied to the D2O cVSSI device. The data were recorded on the linear ion trap mass spectrometer. The fraction of deuterium incorporated for the [M + H]+ serine, acetaminophen, and arginine ions as calculated using the weighted average m/z (eq ) of the exchanged ions and the average molecular weight (see the Supporting Information). Analyte labels are provided for each set of voltage data. Error bars represent one standard deviation about the mean for triplicate measurements. The legend shows the voltages associated with colored bars.

One question that arises from the applied voltage experiments is the effect of using opposite polarity for the voltage applied to the D2O reagent. Initially, such experiments were pursued as it was estimated that the interaction of the oppositely charged droplets might lead to higher levels of exchange. However, for this setup, the use of negative polarity for the D2O emitter tip resulted in immediate analyte droplet condensation on the D2O emitter tip, which resulted in the termination of its vibration and production of D2O droplets. Even with the application of 0 V to the D2O device, droplet buildup was observed on the D2O emitter, which resulted in uneven D2O droplet production. Therefore, for this setup, only the same-polarity voltage application conditions were employed for the remaining experiments.

As mentioned above, one of the means by which HDX can provide the identity of a compound is to reveal the number of heteroatom hydrogens that are present. An accurate number of such sites is obtained upon the molecule undergoing sufficient HDX such that a population of ions exhibiting exchange of all sites is observed. That is, in the mass spectrum, the observed isotopologue with the greatest m/z should correspond to the complete exchange of heteroatom hydrogens. From Figure , it is observed that only the [M + H]+ acetaminophen ions achieve this level of exchange. Indeed the [M + H]+ arginine ions fall significantly short of this HDX level as the isotopologue having the largest m/z value is the D4 peak.

One factor that may affect the degree of HDX is the amount of D2O droplets produced and their overall size. To investigate the effect of the relative number of D2O droplets produced, the flow rate of the D2O reagent was doubled such that it was twice that of the analyte solution (5:10 μL·min–1). Figure shows the results after the reagent flow change for the serine, acetaminophen, and arginine ions. With the increase in flow, the amount of deuterium increases for each of the analytes. The increase ranges from ∼17% for arginine to ∼24% for serine. Notably, none of the fraction of deuterium incorporated values significantly exceeds 0.5. However, the isotopic distribution for serine now shows a feature corresponding to the complete exchange of all heteroatom hydrogens as shown in Figure S1 in the Supporting Information. For arginine, the isotopic distribution reveals two new isotopologue features (D5 and D6). Although a population of ions exhibiting complete exchange is not evident, increased exchange levels such as this should be beneficial for obtaining compound identifications by eliminating candidate compounds with too few heteroatoms. Here, we note that prior studies have shown very limited deuterium incorporation levels for in-droplet HDX reactions for arginine in a field-free source region.[49]

Figure 5

Fraction of deuterium incorporated for two reagent flow rates. The bar graph shows the fraction of deuterium incorporated by [M + H]+ serine, acetaminophen, and arginine ions at two different D2O flow rates. Linear ion trap data were recorded for analyte and D2O reagent solution flow rates of 5:5 and 5:10 μL·min–1 are represented by the blue and orange bars, respectively. For these data, DC voltage values of +1500 and +800 V were applied to the analyte and reagent solutions, respectively. Molecular labels are provided for each set of flow rate data. Error bars represent one standard deviation about the mean for triplicate measurements. The legend shows the colors for the bars representing the different flow rates.

Response to D2O cVSSI Device Activation and Deactivation

One of the perceived advantages to the dual cVSSI device setup is that D2O droplet production can be initiated and terminated instantaneously while the analyte ions are produced continuously. The advantage of such an approach is that conditions can be periodically changed between those favoring HDX and those resulting in the production of precursor ions. That is, it should be possible to obtain the defining HDX isotopic distribution and precursor ion exact mass in alternating data collection time periods. Provided that no remnant of D2O reagent persists over a short timescale, it will be possible to alternate between such conditions on a timescale that allows adequate sampling of chromatographic features in LC–MS experiments.

To evaluate whether or not HDX could be initiated and shut off on a timescale commensurate with LC peak sampling, the RF voltage applied to the D2O cVSSI device was cycled manually using the actuation switch box (Figure ) for three on/off time periods of ∼2 s each. Data were collected on an orbitrap mass spectrometer to provide clear isotopologue resolution for calculating the fraction of deuterium incorporated in each molecule. Because D2O reagent builds up at the tip during the “off” time period for the cVSSI device, the first MS scans after activation can result in more extensive HDX. Similarly, upon turning off the D2O cVSSI device, residual droplet mixing can be observed for a few scans. To evaluate the reproducibility of the deuterium uptake in these successive activation periods, the first and last three scans of each period were not included. Figure S2 in the Supporting Information shows the mass spectra for the HDX replicates. Additionally, an internal standard (n-butylamine) was used. The fraction of deuterium uptake was also computed for the internal standard. By scaling the fraction of deuterium uptake of the analyte to that for a single replicate (multiplying by ratios of internal standards), remarkable reproducibility in deuterium uptake is obtained. Figure shows that the reproducibilities for HDX are ∼4.6, ∼6.2, and ∼8.0% RSD for serine, acetaminophen, and arginine, respectively. Figure also shows that there is a very minimal amount of HDX that occurs during the deactivation time periods. This most likely results from HDX occurring at the start of the time period (extending several scans beyond those discarded). However, the mass spectra (Figure S2) are nearly indistinguishable from those obtained in the absence of any D2O reagent. Thus, it can be argued that even at this early stage, in-droplet HDX can be initiated and terminated nearly instantaneously and in a manner that is sufficient to sample metabolite peaks in LC–MS separations.

Figure 6

HDX response to cVSSI device activation/deactivation. The bar graph shows the average fraction of deuterium incorporation for each of the compounds during three activation and three deactivation periods. Analyte and reagent solution flow rates of 5:5 μL·min–1 were used for these three on/off periods. The on/off time periods were set at ∼2 s each and initiated using manual activation of the RF actuation switch box (Figure ). Blue and orange bars correspond to average deuterium incorporation during activation and deactivation periods of the D2O cVSSI device. Molecular labels are provided for the pairs of fraction of deuterium incorporation values. Error bars represent one standard deviation about the mean for the three measurements after being scaled using the internal standard (see text for details). The legend shows the colors of the bars associated with each experimental period.

Examining HDX Isotopic Patterns to Distinguish Isomeric Compounds

As mentioned above, a number of studies have presented the idea that pre-MS gas- and solution-phase HDX can be performed to help distinguish different compounds including isomeric species.[49,51−54] Here, experiments are conducted to determine whether or not the dual cVSSI source can be used to distinguish different isomeric species. The basic idea is that although the confining droplets accelerate the HDX reaction, the relative rates of different hydrogens are sufficiently distinct for different isomers such that the very short timescale of the reaction can yield noticeable differences in deuterium uptake.

To demonstrate in-droplet HDX for isomer determination, sucrose and palatinose compounds were analyzed in negative-ion mode using the orbitrap mass spectrometer. As isomer distinction may rely on extremely precise experimental conditions, an internal standard (homoserine) was used. Here, the internal standard ensures that similar droplet mixing conditions are achieved through the positioning of the analyte and reagent devices. To accomplish this, the analyte cVSSI device is set and the D2O cVSSI device is adjusted slightly by hand while recording mass spectra. For the isomer experiments, the D2O cVSSI device position was adjusted until the D1 isotopologue of the internal standard (homoserine) was approximately half the intensity of the D0 isotopologue. Figure S3 in the Supporting Information shows the homoserine isotopologue intensities recorded during the collection of the isomer data.

Figure shows comparisons of the isotopic patterns for two pairs of isomeric compounds. First, the distributions for the [M – H]− disaccharides ions of palatinose (orange bars in Figure ) and sucrose (blue bars in Figure ) collected on an orbitrap mass spectrometer show noticeable differences. Data for the sucrose sample show that upon HDX, the most abundant ions are those belonging to the D1 isotopologue ion population. The other isotopologue features in order of abundance are the D2, D0, D3, D4, and D5 peaks. For palatinose, the D0 isotopologue peak is the most intense and the ordering of the other features is D1, D2, D3, and D4. Visual inspection of the data shows clear differences in the isotopic distributions for these two isomers. A question arises as to whether or not these results are reproducible across laboratories, cVSSI device, and even instrument platforms. Figure also shows the results for [M – H]− palatinose and sucrose ions collected on a different day using different cVSSI devices as well as a linear ion trap instrument. Nearly identical isotopic distributions are obtained for the disaccharide ions. One exception in terms of the relative intensities is observed for the D0 and D2 isotopologue features (day 2 compared to day 1 in Figure ) for the sucrose ions. That said, these changes are very small and do not significantly alter the level of deuterium incorporation.

Figure 7

Comparisons of isotopic HDX distributions for isomer pairs. (a) Isotopic distributions obtained for [M – H]− sucrose (blue bars) and palatinose (orange bars) ions on day 1 (orbitrap mass spectrometer) and day 2 (linear ion trap mass spectrometer), respectively. (b) Isotopic distributions obtained for [M + H]+−N-acetyllactosamine (LAcNAc) (green bars) and N-acetylglucosaminyl-β-1,2-mannose (red bars) ions on day 1 (orbitrap mass spectrometer) and day 2 (linear ion trap mass spectrometer), respectively. Data collected on the different days are labeled, and the D0 and D1 isotopologues are labeled for reference. The legend associates compounds with their data according to bar color. For these experiments, analyte and D2O reagent flow ratios of 5:5 μL·min–1 and DC voltages of /+1500 and −/+800 V, respectively, were used.

To demonstrate that the ability to distinguish isomer ions is not limited to the two disaccharides discussed above, separate orbitrap experiments have investigated the relative HDX reactivities of N-acetyllactosamine (LAcNAc) and N-acetylglucosaminyl-β-1,2-mannose. One purpose in conducting these studies is to show that the distinguishing capabilities can extend to compounds having more than one type of heteroatom hydrogen; with these compounds, amide hydrogens are included. Figure shows that greater deuterium incorporation is observed for [M + H]+N-acetylglucosaminyl-β-1,2-mannose ions having an ordering of isotopologue intensities (red bars in Figure ) that is similar to that observed for sucrose ions. The [M + H]+N-acetyllactosamine (LAcNAc) ions show an isotopic pattern (green bars in Figure ) that is similar to the palatinose ions with the exception that the D6 isotopologue is also observed at low abundance. Experiments for these compounds have also been conducted on different days and instruments using separate cVSSI devices. As before, remarkable reproducibility is achieved using the different instruments.

HDX of Favored Ions Using a Three-Component cVSSI Device Setup

Often metabolomics experiments employing LC–MS are conducted in both positive- and negative-ion modes due to preferential ionization of specific compounds. Although this provides increased metabolite coverage, it does come at a cost to increased throughput as well as the general expense associated with using different separation approaches (e.g., columns and solvent systems). For some experiments, an advantage can be envisioned in the ability to periodically favor specific ion types such that experiments can be conducted in a single ionization mode. One example could be the formation of cation adduction ions for species that do not readily form protonated species in positive-ion mode.

In separate experiments, the ability to periodically ionize and perform in-droplet HDX for sucrose was evaluated using a three-component cVSSI device setup. With this configuration, one device performs the charged droplet production of the analyte solution, while the second and third devices can provide charged droplets of KCl and D2O solutions. Figure shows the results that are obtained when the three devices are activated successively. With only activation of the analyte device, no sucrose ions are observed. That is, under these conditions, even with the application of voltage, no [M + H]+ ions are observed. With the activation of the KCl solution cVSSI device, the immediate production of [M + K]+ ions is observed. Finally, with the activation of the D2O cVSSI device, HDX is observed for these cation adduction ions. As with the HDX experiments described above, the production of [M + K]+ precursor and exchanged ions is instantaneous upon activation of the separate cVSSI devices. Additionally, the removal of these ions is instantaneous upon deactivation of the KCl cVSSI device.

Figure 8

Cation adduction and HDX using a three-component cVSSI setup. (a) Mass spectral region containing the m/z range that would be associated with [M + H]+ sucrose ions. (b) [M + K]+ ions produced after activation of the KCl cVSSI device (see text for details). (c) Isotopic distribution obtained for [M + K]+ ions that have undergone HDX; here, all three cVSSI devices have been activated. Insets show color-coded boxes indicating the status of each device associated with the dataset. Sucrose ions are labeled. For these experiments, all solution flow rates were set at 5 μL·min–1 and DC voltages of +1500, +800, and +800 V were applied for the analyte, KCl, and D2O cVSSI devices, respectively.

The ability to turn on and off ion production and reactions instantaneously is envisioned as a valuable characteristic of this multidevice cVSSI approach. The ability to alternate source conditions between those favoring protonated ions and those produced by cation adduction may significantly enhance LC–MS ‘Omics analyses. With this approach, a single LC run may now allow efficient ionization of basic and many acidic species (e.g., organic acids, carbohydrates, glycans, nucleotides, etc.). Furthermore, combining the periodic sampling of different ion types with periodic HDX can improve the breadth of compounds identified from single LC–MS experiments. Consider a six-period process that could be employed to allow the successive collection of protonated precursor, protonated precursor exchanged, cation adduct precursor, and cation adduct precursor exchanged ion data during the course of a single LC separation. For many time-limited experiments, such an approach could dramatically increase the information content of single LC–MS datasets. It should also be noted that the instantaneous activation/deactivation of the devices could also be incorporated into experiments that employ tandem mass spectrometry.

Final Considerations for In-Droplet HDX in Comparative Metabolomic Investigations

One important consideration for the use of in-droplet HDX to identify metabolite compounds is the elucidation of factors governing the individual HDX reactions. A goal would be to use such rules to predict the deuterium incorporation of specific compounds. Such a capability would be highly valuable for compound identification using an isotopic pattern matching approach especially for compounds that are not in databases (e.g., emerging drugs and their metabolites). A question arises as to whether or not the log of the acid dissociation constant (pKa) can provide any insight into the observed deuterium incorporation values. Consider the experiments for serine, acetaminophen, and arginine discussed above. These compounds have heteroatoms comprising guanidine, primary amine, primary alcohol, carboxylic acid, phenol, and amide functional groups. The pKa values for the respective acid (or conjugate acid depending on the equilibria in effect at pH ∼ 4.7) are approximately 12.5, 8.0, −2.4, 3.3, 10.0, and −0.5.[67] Because arginine is the compound that stands out with regard to deuterium incorporation (∼25% lower) and the pKa value is the greatest of the functional groups considered, it might be worthwhile to investigate the effect of pKa on deuterium incorporation. That said, to draw any statistically significant correlations would require the measurement of many different compounds followed by multiple regression analysis.

Previous experiments employing in-droplet HDX and voltage-free cVSSI examined a number of compounds and sought to correlate the frequency of occurrence of different functional groups within specific compounds and their overall deuterium incorporation level.[49] An improved ability to predict deuterium incorporation was obtained from regression analysis. Future work will investigate the utility of the present source setup (two separate cVSSI devices) for the development of a predictive HDX method. This will require major source modifications allowing for precise localization of emitter tips with micromanipulation as well as an encased design to control for air flow around the mass spectrometer inlet resulting from mechanical sources as well as convection.

Final Considerations for In-Droplet HDX in Structural ‘Omics Investigations

Although the focus of the discussion thus far has been on the usage of in-droplet HDX to aid compound identification, it is worthwhile to consider its applicability for the study of the three-dimensional structure of large biomolecules. Typical HDX studies of large biomolecules follow the general protocol demonstrated by Smith and co-workers in which protein HDX is conducted for a predetermined period of time; this is followed by reaction quenching, protein digestion, peptide separation, and MS analysis.[68,69] The experiment can be conducted online,[70] but such an approach requires highly sophisticated equipment providing intricate and precise solvent and reagent delivery and reaction quench conditions (low pH and temperature) must be carefully maintained during the digestion and peptide separation steps.

To determine the utility of in-droplet HDX for examining protein structure, it is useful to consider the exchange rates of amide backbone sites. Generally, for regions that are not buried in hydrophobic cores or involved in extensive hydrogen bonding (secondary structure), the rates of exchange for backbone amide sites are ∼1000 min–1.[71] Thus, to achieve full labeling of such sites, exchange times of seconds to minutes are used.[72] As the droplet lifetimes of these charged microdroplets are orders of magnitude shorter, it may not be possible to fully label unprotected backbone amide hydrogens. It can be argued that this is the reason for the relatively low level of deuterium incorporation in the bradykinin peptide. It should also be noted that in traditional experiments, the D2O:H2O ratio is carefully controlled during the reaction such that the amount of D2O is much greater (≥10×) than H2O.[72] This is done to favor exchange processes in which deuterium is incorporated into the peptide. With the in-droplet HDX approach described here, it is likely that the ratio is much closer to being equal. Understanding how to control this ratio will require significant methods development work in the future. That said, experiments that may be pursued in the near future can be envisioned. For example, rapid labeling of accessible sites may provide clues about the locations of interface regions in the study of protein complexes. Thus, in-droplet HDX combined with top-down proteomics experiments may find some utility in biomolecular structure studies.

Conclusions

The characterization and proof-of-principle studies described here present the new ionization technique termed capillary vibrating sharp-edge spray ionization (cVSSI) as a suitable approach to conduct in-droplet HDX reactions with the goal of aiding compound identification by MS. One salient feature of the approach is that each (analyte and reagent) cVSSI device can be controlled independently. This allows the instantaneous production and removal of specific ion types and the periodic initiation and termination of HDX reactions for such ions on a timescale that is commensurate with LC–MS and LC–MS/MS strategies employed in metabolomics studies. In the future, such techniques may be further developed to aid compound identification using information about the probable numbers of heteroatom sites and/or unique isotopic distributions that can be matched to those in databases. One attractive feature of the approach is that parameters such as DC voltage application and solution flow rates can be changed to provide highly reproducible, unique mass spectra; this suggests an ability to tune HDX reactivity for specific studies. Finally, another benefit of the approach is the performance of the extremely efficient reaction on a very short timescale. This allows for distinguishing intractable species such as isomers due to reaction quenching that captures subtle differences in site reaction rates.

Future work will focus on demonstrating the approach for real-world metabolomics samples as well as for the performance of biomolecular structure studies. Next, work will be directed toward developing an integrated interface for the multidevice CVSSI setup to couple in a straightforward manner with existing LC–MS systems. Overall, this technical report effectively introduces a new measurement platform in the form of a simple, versatile ionization source that can be used for efficient comparative ‘Omics analyses.

Experimental Section

Chemicals and Solvents

Metabolite standards and solvents were purchased from various sources. Multiple compounds were purchased from Sigma-Aldrich (St. Louis, MO), including acetaminophen, l-serine, l-arginine, bradykinin (reagent grade, >95%), N-acetylglucosaminyl-β-1,2-mannose, N-acetyllactosamine (LAcNAc), sucrose (>98%), palatinose (>99% hydrate), homoserine, and n-butylamine. The solvents deuterium oxide (D2O) (99.9%), acetic acid (98%), methanol (Optima LC–MS grad), water (Optima LC–MS grade), and ammonium acetate (98.6%) were purchased from Fisher Chemical (New Jersey, NJ). All standards were utilized without further purification.

For positive-ion-mode experiments, standard solutions of concentration 0.01 mg·mL–1 in 100 mM ammonium acetate buffer solution were prepared. The buffer solution was reduced to a pH of 4.75 by the addition of acetic acid. For negative-ion-mode experiments, standard solutions of concentration 0.01 mg·mL–1 in pure water were prepared. For experiments employing internal standards of homoserine and n-butylamine, concentrations of standards were also set at 0.01 mg·mL–1.

MS Instrumentation

A Q-Exactive Hybrid Quadruple-Orbitrap mass spectrometer (Thermo Fisher Scientific, San Jose, CA) was utilized for mass analysis. The instrument features high-performance quadrupole precursor ion selection with high-resolution, accurate-mass (HR/AM) Orbitrap detection. Experiments were conducted in both positive- and negative-ion modes using an inlet capillary temperature of 250 °C. A full mass scan resolving power of 70 000 was used as well as an AGC setting of 1 × 106. In separate experiments designed to confirm those obtained on the Orbitrap instrument, a linear ion trap (LTQ-XL, Thermo Fisher Scientific, San Jose, CA) was utilized. The same inlet capillary temperature and AGC settings were employed.

cVSSI Device Fabrication and Source Setup

A detailed description of cVSSI device fabrication has been described elsewhere.[40,49,73]Figure shows photographs of one configuration of a multidevice cVSSI source used to ionize the analytes and perform in-droplet HDX. In this setup, two cVSSI emitter tips are employed to separately deliver analyte and HDX reagent solutions. Briefly, each cVSSI device was constructed by attaching a piezoelectric transducer (7BB-27-4L0, diameter = 27 mm, Murata) to one end of a no. 1 glass slide coverslip (w × l = 25 × 60 mm) (VWR, Radnor, PA) using epoxy glue. A fused-silica capillary, approximately 5 cm in length (360 μm OD × 100 μm ID), which was pulled (Sutter Instruments) to a final tip diameter of approximately 25 μm, was attached at a 60° angle relative to the distal edge of the glass slide. One end of a piece (∼40 cm long) of thin-wall poly(tetrafluoroethylene) (PTFE) tubing was slip fit over the blunt end of the attached pulled-tip capillary. Near this end, the PTFE tubing was punctured with a short platinum wire, which was glued to the outside wall of the tubing to prevent leakage. This wire was used to provide DC voltage to the infused solutions. The other end of the platinum tubing was glued to a needle (BD 0.3 mm × 13 mm) luer lock. The luer locks for each device were press-fit onto syringes (BD 1 mL) to allow for direct infusion of samples using a syringe pump (see below). For in-droplet HDX experiments, two cVSSI devices were arranged ∼90° relative to each other and were placed ∼0.5–1 cm from the inlet of the mass spectrometer as shown in Figure .

Dual cVSSI Source Operation

For experiments using the multidevice cVSSI setup shown in Figure , direct infusion of analyte sample and deuterating reagent solutions was achieved using a Chemyx Fusion 4000 multichannel syringe pump (Chemyx, Stafford, TX) equipped with two independently controlled precision syringe pump channels with programmable step-rate functionality. The dual drive system allowed for two independent pumping channels to be controlled separately. Constant flow through the vibrating fused-silica capillaries produced microdroplet plumes from each device that ultimately underwent droplet mixing of the two solutions prior to entering the mass spectrometer inlet ion transfer tube. The RF signal (10 Vpp at ∼95 kHz) applied to the piezoelectric devices was generated using a function generator (Tektronix AFG-1062, Beaverton, OR), which was connected to an amplifier (Krohn-Hite 7500, Brockton, MA) as shown in Figure . The RF voltage on the two cVSSI devices was applied independently using an actuator switch box (Figure ).

DC voltage (variable magnitude; see the Results and Discussion section) was applied to both the analyte and D2O reagent solutions near the emitter tip. Previous studies have shown the advantage of combining DC voltage with cVSSI to enhance analyte ionization efficiency compared with electrospray ionization (ESI)[74] and nano-electrospray ionization (nESI).[75] Ionization efficiency is especially enhanced for negatively charged ions where vibration disrupts corona discharge formation;[58,76] however, efficiency gains have also been reported for positively charged ions.[50,76]

Deuterium Uptake

The fraction of deuterium uptake for each sample was calculated as follows. Upon actuating both cVSSI devices (analyte solution and D2O reagent), the HDX mass spectra were recorded. The mass-to-charge ratio (m/z) shift associated with deuterium uptake was calculated from the isotopic distribution of the reacted ions as described previously;[77] this provides essentially the same deuterium uptake as the centroiding approach pioneered by Engen etal.[78] First, the weighted average m/z values (Aw) for the exchange ions were computed according to eq .In eq , (m/z) and I are, respectively, the m/z value and the corresponding intensity (I) of the ith isotopologue in the isotopic distribution consisting of n total isotopologue features. Next, the amount of deuterium incorporated into the analyte ion was computed by subtracting the average molecular weight of the unreacted ion from Aw obtained from eq . Finally, the fraction of deuterium incorporation was computed by dividing the amount of deuterium incorporated by the total number of exchangeable hydrogens (those attached to heteroatoms).