Multiply-Charged Cation Attachment to Facilitate Mass Measurement in Negative Mode Native Mass Spectrometry

Abstract

Native mass spectrometry (MS) is usually conducted in the positive ion mode; however, in some cases, it is advantageous to use the negative ion polarity. Challenges associated with native MS using ensemble measurements (i.e., the measurement of many ions at a time as opposed to the measurement of the charge and mass-to-charge of individual ions) include narrow charge state distributions with the potential for overlap in neighboring charge states. These issues can either compromise or preclude confident charge state (and hence, mass) determination. Charge state determination in challenging instances can be enabled via the attachment of multiply-charged ions of opposite polarity. Multiply-charged ion attachment facilitates the resolution of charge states and generates mass-to-charge (m/z) information across a broad m/z range. In this work, we demonstrated the attachment of multiply-charged cations to anionic complexes generated under native MS conditions. To illustrate the flexibility available in selecting the mass and charge of the reagents, the 15+ and 20+ charge states of horse skeletal muscle apo-myoglobin and the 20+ and 30+ charge states of bovine carbonic anhydrase were demonstrated to attach to model complex anions derived from either β-galactosidase or GroEL. The exclusive attachment of reagent ions is observed with no evidence for proton transfer, which is key for the unambiguous interpretation of the post-ion/ion reaction product ion spectrum. To illustrate application to mixtures of complex ions, the 10+ charge state of bovine ubiquitin and was attached to mixtures of anions generated from the 30S and 50S particles of the E. coli ribosome. Six and five major components were revealed, respectively. In the case of the 50S anion population, it was shown that the attachment of two 30+ cations of carbonic anhydrase revealed the same information as the attachment of six 10+ cations of ubiquitin. In neither case was the intact 50S particle observed. Rather, particles with different combinations of missing components were observed. This work demonstrated the utility of multiply-charged cation attachment to facilitate charge state assignments in native MS ensemble measurements of heterogeneous mixtures

Graphical Abstract

Introduction

Among the remarkable characteristics of electrospray ionization (ESI) is its ability, under appropriate conditions, to liberate specific non-covalent complexes into the gas phase as multiply-charged ions^1,2,3,4. This ability underlies the field of native mass spectrometry^5,6,7 which is focused on the mass spectrometry and tandem mass spectrometry of biologically relevant complexes ranging in mass from tens of kilodaltons (kDa) to in excess of tens of megadaltons (MDa). The generation of multiply-charged ions facilitates detection and leads to ions with mass-to-charge (m/z) ratios significantly lower than the mass of the complex. Nevertheless, typical native MS conditions lead to ions of relative high m/z ratios^8,9, which generally requires mass analyzers capable of measuring ions of m/z ratios extending to the tens of thousands^10,11. While the generation of multiply-charged ions underlies native MS, it requires both the determination of m/z and charge state, z, in order to determine mass, m.

Several approaches can be used for charge state determination in ESI-MS. For example, provided the resolution of the mass analyzer is sufficiently high to resolve isotopes, the spacings of the isotope peaks on the m/z scale can be used to determine charge state. For native MS applications, however, it is usually not possible to resolve isotope spacings due, inter alia, to the high masses of the analytes and the small spacings due to multiple charging. The classical approach to charge state determination in ESI MS is via measurement of the m/z spacings between adjacent charge states¹². Provided the mass and charge differences between adjacent charge states is known, it is possible to determine the charge states of the ions¹³. Various algorithms have been developed to deconvolute ESI m/z spectra to generate “zero charge” mass spectra^14,15,16,17 and are often used in native MS. However, there are factors in play that can confound charge state determination via the measurement of the m/z values of adjacent charge states in native MS, particularly for highly heterogeneous systems. For example, high mass complexes, which are generally believed to be generated via the so-called charged residue mechanism^9,18,19,20, typically lead to relatively narrow charge state distributions⁷. The fact that there are relatively few peaks that fall over a relatively narrow m/z range can lead to significant uncertainty in charge state determination²¹. Furthermore, the use of native conditions that usually include salts and buffers to stabilize and preserve complexes usually results in significantly broadened peaks that can be difficult to resolve¹⁰.

Resolution of charge states is further complicated by the heterogeneity of the complex itself. These issues can be mitigated by adding charge reducing agents in solution^22,23,24,25, which increases spacings between charge states by reducing charge. Solution-phase charge reduction, however, is limited and still results in relatively few charge states that fall within a relatively narrow m/z range. Alternatively, the use of single proton transfer ion/molecule²⁶ or ion/ion reactions in the gas phase^21,27,28 generates lower charge states that have greater separation and are distributed over a wider m/z range. Nevertheless, it may still be challenging to resolve and identify overlapping charge state distributions resulting from heterogeneous mixtures. An alternative approach for charge state determination is the measurement of the m/z and z of individual ions, as is accomplished in the charge detection MS (CDMS) experiment^{29,30,31,32,33}. In CDMS, the charge state of an individual ion is determined on the basis of signal level while the m/z measurement is made on the basis of an ion frequency measurement. CDMS circumvents the requirement of resolved charge states and replaces it with the requirement for an absolute signal measurement on individual ions. A light scattering/reflectance approach, referred to as mass photometry^34,35 has recently been described that enables mass determination while avoiding ionization and mass spectrometry altogether.

We recently described another approach to charge state determination based on the measurement of adjacent charge states that differ by known differences in mass and charge that both greatly exceed 1 Da and one charge, respectively³⁶. This is accomplished by attaching one or more multiply-charged reagent ions to a multiply-charged analyte complex and measuring the resulting spectrum. Often, a segment of the analyte ion distribution is mass-selected in order to simplify the mixture of ions subjected to reaction, which is a strategy previously shown to be useful for single charge transfer reactions^37,38,39. We refer to this approach as “mass analysis of macromolecular analytes via multiply-charged ion attachment” (MAMA-MIA). Our original report described the attachment of anionic reagents to cationic complexes generated under native conditions. In this report, we describe cationic reagents for attachment to anionic complexes generated under native conditions. It can be advantageous to generate complex ions in the negative mode, at least for some systems. For example, it has been demonstrated that survival of membrane protein complexes is enhanced in the negative mode⁴⁰. There is merit, therefore, to developing high mass and high charge reagents for ion attachment to negatively-charged non-covalent complexes. In this work, we describe several cationic reagents for use with anionic complexes and apply them to mixtures of complexes derived from negative mode nano-ESI (nESI) of the E. coli ribosome as a demonstration the MAMA-MIA approach in the negative mode.

Experimental

Sample preparation for native mass spectrometry.

β-galactosidase from E. coli was purchased from Sigma Aldrich. The lyophilized solid was reconstituted in a 150 mM ammonium acetate (Sigma Aldrich) buffer to create a stock solution at a concentration of 10 μM (calculated using the mass of the tetramer). The sample underwent adduct removal, via centrifugation, a minimum of four times with the same buffer adjusted to pH 7 using a 10 kDa molecular weight cutoff Amicon Ultra 0.5 mL filter (Millipore Sigma). The recovered sample (17 μL) was diluted with the same buffer to achieve the same original concentration from the stock solution. GroEL (Sigma Aldrich) lyophilized powder preparation was described before in detail³⁶. E. coli 70S ribosome solution was purchased from New England Biolabs. The original sample, with an initial concentration of 13 μM, was constituted in a buffer containing 10 mM magnesium acetate, which is necessary for the 70S ribosome to be intact in the condensed phase. The sample preparation for the working solutions was described in detail previously⁴¹ and modified accordingly. Briefly, the stock solution was buffer exchanged 8 times with 150 mM ammonium acetate and 10 mM magnesium acetate (Sigma Aldrich) at pH 7.4 with the same filter mentioned above. The sample was then diluted with 150 mM ammonium acetate buffer at pH 7.4 to obtain a final concentration of 0.5 mM Mg²⁺ for the working solution.

Reagent preparation.

Horse skeletal muscle myoglobin, bovine ubiquitin, and bovine carbonic anhydrase were purchased from Sigma Aldrich. The lyophilized solid was reconstituted in 50:50 H₂O: Methanol with 1-5% glacial acetic acid at 10 μM concentration. Denaturing conditions were used to ensure higher charge state formation.

Mass spectrometry.

All experiments were performed on a TripleTOF 5600 hybrid QqTOF mass spectrometer (SCIEX, Concord, ON, Canada) which was previously modified for ion/ion reactions in a fashion similar to that described for another QqTOF platform in our laboratory⁴². Alternately-pulsed nESI allows for sequential injection of multiply deprotonated protein complexes and multiply protonated reagent proteins⁴³. Deprotonated analyte protein complexes were sprayed under native conditions and accumulated in q2. Ion isolation was performed in q2 using tailored waveforms built and downloaded to waveform generators using MS Devices software (SCIEX). Reagent proteins were isolated in Q1 using rf/dc apex isolation. The reagent proteins were sprayed under denatured conditions and were transferred to q2 for mutual storage with analyte protein complexes. The reagent ion number density was varied by altering the voltage and injection time to optimize these reactions. The ions were stored for 5-50 milliseconds in q2 for gas phase ion/ion reactions⁴⁴. Mass analysis was performed using time-of-flight (TOF). Masses were determined from the measured m/z values of the peak apexes using the relationship:

$$M=\left({\left(m/z\right)}_{(N+1)}\right)\left(n-N\text{Δ}z\right)+nx-N\text{Δ}m$$ (1)

where M is the mass of the neutral molecule/complex, N is the number of attached reagent cations, n is the absolute charge state of the initial analyte ion without any attachments, m/z_(N+1) is the m/z of the ion of interest ((m/z)₁ is the m/z value of the initial analyte ion without any attachments), Δz is the charge of the reagent cation, Δm is the mass of the reagent cation, and x is the average mass of the cations missing from the original analyte anion (usually protons from deprotonation of acidic sites). Post-ion/ion reaction spectra were simulated using an R shiny app developed in our lab (see Supporting Information for a description). Different m and z values can give rise to overlaps in m/z in narrow regions of the m/z scale but not across large ranges of m/z²¹. The simulated data are useful in establishing an accurate value of n by ensuring that each attachment product is observed at the predicted m/z value across a wide m/z scale.

The TOF m/z scale was calibrated in stages. Clusters of cesium iodide were used to calibrate the m/z scale up to m/z 10,000. A secondary calibrant, such as pyruvate kinase, was then measured using the CsI calibrated mass scale. The secondary calibrant ions were then charged reduced via proton transfer to yield ions across a m/z range extending beyond m/z 100,000. The mass measurement accuracy of this instrument after calibration is typically 5 ppm or less at low m/z. We find that the m/z scale remains remarkably linear and stable at high m/z as well. For m/z measurements of many of the complexes studied here, we find that imprecision in the mass measurement is dominated by variations in the masses of the analyte ions themselves, rather than from the TOF measurement. Variations associated with the use of different nESI tips, small changes in transmission conditions, room temperature and humidity, etc. can affect the extent of salt/solvent retention. We find that such variations can result in mass variations of +/− 100-200 Da under nominally fixed conditions. We therefore report m/z values to the tenths of kDa with the uncertainties arising primarily from challenges in generating consistently and efficiently desalted/desolvated ions.

Results and Discussion

Data are presented here using analyte anions derived from β-galactosidase and GroEL as standards to evaluate several candidate cationic reagents. Anions derived from the E. coli ribosome serve as examples of heterogeneous mixtures that can benefit from a MAMA-MIA experiment.

β-galactosidase anions.

β-galactosidase, an enzyme for lactose digestion, is a homo-tetrameric complex with a mass of ~465.4 kDa. Figure 1a shows the mass spectrum of an anionic β-galactosidase complex from a native solution. The spectrum shows the tetrameric complex with charge states from 42- to 35-. The inset on the upper left of Figure 1a shows the mass spectrum after an isolation step in q2 to select the 37- charge state. The anions derived from β-galactosidase, including the isolated 37- anion population, were subjected to reactions with several different charge states of apomyoglobin [myo+nH]ⁿ⁺ generated under denaturing conditions. The positive ion mass spectrum of bovine myoglobin, a commonly studied protein in ESI MS, is provided as Figure S1 and spectra obtained after Q1 rf/dc isolation of the +20 and +15 charge states are shown in Figure S2a and Figure S2b, respectively. Figure 1b shows the post-ion/ion reaction spectrum of the 37- charge state of β-galactosidase with [myo+15H]¹⁵⁺ while Figure 1c shows the result from the analogous experiment using [myo+20H]²⁰⁺ as the cationic reagent. (Note that the m/z scale in Figure 1a is far narrower than those of Figures 1b and 1c.) In both cases, ion attachment is, by far, the dominant mechanism. (The small signals for charge states above and below the dominant ion attachment product are likely due to incomplete ejection of adjacent charge states of β-galactosidase during isolation). Products from two successive attachments of the reagent cation are observed in the case of the reaction with [myo+15H]¹⁵⁺ whereas only one attachment is observed with [myo+20H]²⁰⁺. This is due to the fact that the addition of two [myo+20H]²⁰⁺ ions results in a net positively charged (i.e., 3+) complex. Likewise, the attachment of a third [myo+15H]¹⁵⁺ ion leads to a charge inverted complex. (Inversion of net charge upon multiply-charged ion attachment has been noted previously⁴⁵.) The post-ion/ion reaction spectrum of the entire β-galactosidase anion population in reaction with [myo+15H]¹⁵⁺ is provided in Figure S3. Collectively, these experiments with anions of β-galactosidase illustrate that ion attachment is the dominant process in reactions with protein cations, show the high degree of flexibility in selecting reactant ion charge states, and illustrate the possibility for charge inversion of the complex with multiple cation attachments.

Figure 1.

a.) Mass spectrum of β-galactosidase tetrameric complex from a negative nESI with isolation of the 37- charge state (insert). b.) Post-ion/ion reaction spectrum of the 37- charge state of β-galactosidase with [myo+15H]¹⁵⁺. c.) Post-ion/ion reaction spectrum of the 37- charge state of β-Galactosidase with [myo+20H]²⁰⁺.

GroEL anions.

GroEL is a tetradecamer chaperon protein complex that is commonly used in evaluating native MS methods^{46,47,48,49,50}. The monomer mass is ~57 kDa and the nominal mass of the complex is ~802 kDa⁵¹. Figure 2a shows a negative nESI mass spectrum of GroEL collected under moderate ion kinetic energy injection conditions into q2. The charge states are well-resolved and yield a mass of roughly 807 kDa. The excess mass is likely due to incomplete removal of residual salts (see Figure S4). The absolute charges of the GroEL anions are significantly higher than those of β-galactosidase described above. The higher GroEL charge states allow for higher charge reagent cations to be evaluated. In this case, carbonic anhydrase (CA) cations were evaluated as ion attachment reagents. The positive ion nESI mass spectrum of carbonic anhydrase derived under denaturing conditions is shown in Figure S5 along with spectra collected after Q1 rf/dc ion isolation of the [CA+20H]²⁰⁺ (Figure S5b) and [CA+30H]³⁰⁺ (Figure S5c) ions. Figure 2b and Figure 2c show the post-ion/ion reaction spectra after reaction of the anion population of Figure 2a with carbonic anhydrase charge states of 20+ and 30+, respectively. In both cases, ion attachment is, by far, the dominant process. Two attachments of the [CA+20H]²⁰⁺ reagent are possible prior to charge inversion whereas only one attachment of the [CA+30H]³⁰⁺ ion can take place while yielding a negatively charged complex, in this case. Furthermore, the ion attachment peaks yield the same 807 kDa mass as that derived from the unreacted GroEL charge states. There is no evidence that the cation attachment process leads to significant shedding of excess salts.

Figure 2.

a.) Negative ion nESI mass spectrum of GroEL. Post-ion/ion reaction spectra of the GroEL anions shown in a.) with b.) [CA+20H]²⁰⁺ and c.) [CA+30H]³⁰⁺.

E. coli ribosome 30S/50S anions.

The E. coli ribosome is a heterogeneous molecular machine comprised of two subunits (30S, 50S) that form the intact ribosome complex (70S). In total, the 70S complex contains three large nucleic acid strands and roughly sixty proteins⁵². Some of the protein components are very weakly associated with the complex and are often missing⁵². In other native MS studies conducted in positive ion mode, ions of the 30S, 50S and 70S complexes are all observed when high concentrations (e.g., 10 mM) of Mg²⁺ ions are present^53,54,42,34. At low Mg²⁺ concentrations (e.g., 0.5 mM), the 70S complex is usually absent. Both 30S and 50S components when analyzed in the positive mode using MAMA-MIA showed multiple components related to the presence or absence of several small proteins in the complex³⁶.

Here we illustrate the attachment of multiply-charged cations to anions derived from negative mode nESI of the E. coli ribosome obtained using a 0.5 mM solution concentration. Figure 3a shows the negative mode mass spectrum of the E. coli sample. Two major distributions of peaks are observed, each of which appears to show overlapping charge state distributions. The red box shows a region of the spectrum (i.e., m/z 16,000-17,000) selected for reaction with the 10+ ions derived from positive nESI of bovine ubiquitin ([ubi+10H]¹⁰⁺). (See Figure S6 for a zero-charge deconvolution of the ions in the lower m/z cluster of peaks in Figure 3a.) Figure 3b, which displays a much wider m/z range, shows the post-ion/ion reaction spectrum for the reaction of the ubiquitin cations with the selected anions indicated by the red box of Figure 3a. Products from up to four attachment reactions appear in the spectrum. The ions labelled with a ‘zero’ represent residual unreacted anions. The ions associated with four attachments of ubiquitin show clear separation of ions. A subsequent experiment using more ubiquitin cations was performed to increase the contribution from four attachments and a selected region of the post-ion/ion reaction spectrum showing the four cation attachment region is show in Figure 3c. The expanded region of this spectrum shows three major components that are consistent with the intact 30S particle (861.4 kDa), the 30S particle missing the S1 protein (799.8 kDa), and a smaller component corresponding to the 30S particle missing both the S1 and S2 proteins (772.8 kDa). Note that the expected mass of the 30S E. coli ribosome particle is 847.5 kDa (see Tables S1 and S2 for lists of the masses measured here and in the positive ion data reported earlier³⁶ for the 30S and 50S particles, respectively, and Table S3 for predicted masses of the relevant sub-set of particles and proteins taken from Supplementary Tables 1–3 of reference 42. The excess mass is attributed to additional salts, metal ions, etc. usually observed in native MS. Three components with similar masses were noted in the positive ion populations generated from a 0.5 mM Mg²⁺ solution in the previous MAMA-MIA study³⁶. Each of the ions corresponding to the three components mentioned above also showed a satellite peak greater in mass by 5.1 kDa. The satellite species could reflect bound stationary phase induced associated (SRA) protein⁵⁵, which was noted in the 30S population reported in the native MS study of the E. coli ribosome by van der Waterbeemd et al.⁴². The 5.1 kDa species may have been present in the positive ion MAMA-MIA data³⁶ but was not resolved. This MAMA-MIA experiment therefore shows at least six distinct complexes give rise to ions within the selected region (red box) of the mass spectrum of Figure 3a.

Figure 3.

a) Negative ion nESI mass spectrum of E. coli ribosome in a 0.5 mM Mg²⁺ solution. b) Post-ion/ion reaction spectrum derived from the reaction of ions in the red box of a) with [ubi+10H]¹⁰⁺ cations. c) Expanded region of the post-ion/ion reaction spectrum showing products from four cation attachments after using more [ubi+10H]¹⁰⁺ cations to increase the abundances of the products from four attachments.

The anions in the green box of Figure 3a (viz., m/z 19,000-20,000) were also subjected to ion/ion reactions with [ubi+10]¹⁰⁺ with up to 6 attachments being observed, as shown in the post-ion/ion reaction spectrum of Figure 4a. The separation of the individual components was greatest after six reagent cation attachments. An expansion of the region of the spectrum that highlights the most abundant products from six [ubi+10]¹⁰⁺ attachments is shown in Figure 4b. At least five major components are apparent in this region of the spectrum. The masses derived from the peaks in Figure 4b fall within a range of 34 kDa roughly centered around 1.4 MDa, which strongly suggests that they are all 50S-related species. Figure 4c shows the analogous region of the post-ion/ion reaction spectrum following two attachments of [CA+30H]³⁰⁺, which is equivalent in charge to the attachment of six [ubi+10]¹⁰⁺ but with slightly different mass (viz. 6×8575 Da vs. 2×29062 Da). Both experiments yielded the same masses for the 50S components (see Figure 4c and Scheme 1). All of the measured masses are lower than the expected mass of an intact 50S E. coli ribosome particle (1454.6 kDa, see Table S3). Furthermore, the measured masses likely incorporate excess salts, metals (e.g., Mg²⁺), etc., as was also noted above with the 30S particle. In the absence of ions that can be attributable to the intact 50S particle, assigning the masses measured here is challenging. However, the ions can be assigned on a provisional basis based on the masses of the 50S proteins expected to be relatively loosely-bound to the particle⁵⁶ and that the mass of the intact 50S ion population can be 2-3% higher than the expected mass. A plausible set of assignments is provided in Scheme 1 that is consistent with missing various combinations of the L1 (24,598 Da), L9 (15,769 Da), L10 (17,580 Da), L11(14,744 Da) proteins and the [L7/L12]₄ (48,740 Da) heterodimers (see Table S3). The arrows in the scheme are not intended to imply a reaction sequence. Rather, they are intended to convey the protein components missing from the ion. For example, the 1398.2 kDa component is consistent with [50S-L10[L7/L12]₄-L11-L9]. A mass of the intact 50S particle ions of 1494.8 kDa, which is roughly 2.8% higher than the expected mass and is shown in gray in Scheme 1, is consistent with the set of five masses associated with the mixture components observed here.

Figure 4.

a) Post-ion/ion reaction spectrum derived from the reaction of ions in the green box of Figure 4a with [ubi+10H]¹⁰⁺ cations. Expanded region of the post-ion/ion reaction spectrum showing products from b) six cation attachments of [ubi+10H]¹⁰⁺ cations and c) two [CA+30]³⁰⁺ attachments, respectively.

Scheme 1.

The five major components observed from the 50S population of ions in Figure 4 are assigned based on the assumption that the intact 50S particle would appear at a mass of 1494.8 kDa.

Conclusions

In general, native MS can be conducted in either ion polarity. The favored polarity is likely to be complex-dependent. In this work, we demonstrate several reagent cations for ion attachment to negatively charged complexes generated under native MS conditions. These include reagents over a mass range of 8,500-30,000 Da and a charge range of 10+-30+. In all cases, ion attachment is the strongly dominant process so that the MAMA-MIA strategy for facilitating the determination of the masses of components present in heterogeneous bio-complexes can be applied in negative mode native MS studies. The approach is illustrated here using anions derived from E. coli ribosome samples with 0.5 mM Mg²⁺ content under native conditions. Anions associated with 30S and 50S particles are observed. In the case of the 30S anions, six major components were revealed after multiple attachments of [ubi+10H]¹⁰⁺ ions. In the case of the 50S population anions, five major components were revealed after six attachments of [ubi+10H]¹⁰⁺ ions. In a separate experiment, equivalent information was obtained after two attachments of [CA+30H]³⁰⁺ ions. This comparison illustrates the flexibility in defining the shifts in mass and charge that derives from the attachment of multiply-charged ions generated via electrospray ionization to native MS complexes.