Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2021 May 5.
Published in final edited form as: Anal Chem. 2020 Apr 14;92(9):6622–6630. doi: 10.1021/acs.analchem.0c00454

Charge Reduction of Membrane Proteins in Native Mass Spectrometry Using Alkali Metal Acetate Salts

John T Petroff II 1, Ailing Tong 1, Lawrence J Chen 1, Gregory T Dekoster 2, Farha Khan 3, Jeff Abramson 3, Carl Frieden 4, Wayland W L Cheng 5
PMCID: PMC7275249  NIHMSID: NIHMS1593372  PMID: 32250604

Abstract

Native mass spectrometry (MS) provides the capacity to monitor membrane protein complexes and noncovalent binding of ligands and lipids to membrane proteins. The charge states produced by native MS of membrane proteins often result in gas-phase protein unfolding or loss of noncovalent interactions. In an effort to reduce the charge of membrane proteins, we examined the utility of alkali metal salts as a charge-reducing agent. Low concentrations of alkali metal salts caused marked charge reduction in the membrane protein, Erwinia ligand-gated ion channel (ELIC). The charge-reducing effect only occurred for membrane proteins and was detergent-dependent, being most pronounced in long polyethylene glycol (PEG)-based detergents such as C10E5 and C12E8. On the basis of these results, we propose a mechanism for alkali metal charge reduction of membrane proteins. Addition of low concentrations of alkali metals may provide an advantageous approach for charge reduction of detergent-solubilized membrane proteins by native MS.

Graphical Abstract

graphic file with name nihms-1593372-f0001.jpg

Native mass spectrometry (MS) allows detection of protein–protein and ligand–protein complexes and is often combined with ion-mobility measurements (native ion-mobility MS or ion-mobility spectrometry–mass spectrometry (IM-MS)) to simultaneously assess protein structure.1–5 Native MS or IM-MS has been utilized to probe a variety of biological macromolecules including membrane proteins.6–8 To study membrane proteins by native MS or IM-MS, the protein is usually solubilized in detergent and introduced to the instrument in a volatile salt solution by nanoelectrospray ionization (ESI).5,9 Positive-ion-mode ESI is commonly used and leads to positive ions at multiple charge states for a single protein species. This distribution of charge states affords insight into protein structure and conformation in solution.10,11 However, the charge states obtained are often of a magnitude that, in the setting of activating conditions necessary to minimize spectral noise, may result in gas-phase unfolding of membrane proteins by Coulombic repulsion.12–14 In addition, higher charge states may limit the number of bound small molecules that can be detected due to overlapping peaks.15

To overcome these drawbacks, several means to reduce membrane protein charge have been reported. These include instrument modification, solubilization with different detergents, gas-phase ion transfer, and addition of charge-reducing small molecules.15–19 Instrument modification is effective, although it is prohibitive for many groups.16 Altering the detergent to polyethylene glycol (PEG)-based or amine N-oxide amphiphiles affords meaningful charge reduction; even so, the degree of charge reduction may be insufficient to prevent the deleterious effects of excessive Coulombic repulsion.13 To further reduce charge, small molecules are added to the protein solution including triethylamine (TEA), imidazole, and trimethylamine N-oxide (TMAO).15,18,20,21 Recently, Townsend et al. reported a new host of imidazole analogues that aid in charge reduction and afford peptide– nanodisc stability.22 However, these small amine-containing molecules may interact with the protein and result in undesired effects. Thus, small-molecule additives that achieve effective charge reduction and are native to most physiologic environments may be advantageous.

Charge reduction of synthetic polymers has been achieved by the stripping of sodium and other alkali metal cations complexed to cyclic ethers.23 The complexation agent used was 1,4,7,10,13-pentaoxacyclopentadecane (15-crown-5),23 which possesses as many ethylene oxide units as the PEG-based detergent, C10E5, which is used for native MS of membrane protein.24,25 In light of this work, it was posited that the addition of alkali metal cations, through alkali metal acetate salts, to proteins used for native MS may lead to charge reduction.23 Herein, we explore the influence of alkali metals in reducing the charge states of membrane proteins solubilized in PEG-based detergents and compare the effects to other known charge-reducing additives. We also probe the means by which this alkali metal charge reduction occurs and propose a mechanism for this process based on our findings.

EXPERIMENTAL SECTION

Expression and Purification of ELIC.

The pET26-MBP-ELIC was provided by Raimund Dutzler (Addgene plasmid no. 39239), and ELIC was expressed as previously described26,27 in OverExpressTM C43 (DE3) E. coli (Lucigen, Middleton, WI). Briefly, cultures were grown in Terrific Broth (Sigma, St. Louis, MO) and induced with 0.1 mM isopropyl-d-thiogalactopyranoside (IPTG) for ∼16 h at 18 °C. Cell membranes were solubilized in 1% n-dodecyl-β-d-maltopyranoside (DDM) (Anatrace, Maumee, OH) and purified with amylose resin (New England Biolabs, Ipswich, MA). After overnight digestion with HRV-3C protease (Thermo Fisher, Waltham, MA) (10 units per mg ELIC), the protein was further purified on a Sephadex 200 10/300 (GE Healthcare Life Sciences, Pittsburgh, PA) size-exclusion column.

Expression and Purification of KirBac1.1.

The KirBac1.1 plasmid was a gift from Colin Nichols of Washington University in St. Louis and was used for KirBac1.1 expression. KirBac1.1 was expressed and purified as previously described with some modifications.28 KirBac1.1 was expressed in OverExpressTM C43 (DE3) E. coli. Cultures were grown in Terrific Broth to an o.d. of 1.0 and induced with 0.1 mM IPTG for 3 h at 37 °C. The bacteria were lysed by freeze–thaw, solubilized in 30 mM decylmaltoside, and purified with Ni-NTA resin. Eluted protein was further purified by size-exclusion chromatography (20 mM Tris pH 7.5, 100 mM KCl, 0.02% DDM).

Expression and Purification of SaMscL(CΔ26).

The method used was previously reported.29 SaMscL(CΔ26) (plasmid no. 79027, Addgene) in C43 (DE3) E. coli cells was grown in Terrific Broth to an o.d. of 2.0 and induced with 1 mM IPTG at 30 °C for 2 h. Cells in 50 mM Tris-HCl pH 8, 200 mM NaCl, and 44 mM n-dodecyl-N,N-dimethylamine-N-oxide (LDAO) were lysed and solubilized by sonication and purified with Ni-NTA resin. Eluted protein was purified by size-exclusion chromatography (10 mM Tris-HCl pH 7.5, 150 mM NaCl, 2 mM LDAO).

Expression and Purification of mVDAC1.

Expression and purification of mVDAC1 was carried out as described previously.30 In short, mVDAC1 was expressed in M15 E. coli cells. The expressed protein was solubilized from inclusion bodies and purified on a Talon affinity purification column. The protein was refolded by removing guanidine hydrochloride through dialysis and purified by size-exclusion chromatography (20 mM Tris-HCl pH 8.0, 150 mM NaCl, 0.1% LDAO).

Emitter Tip Preparation.

Utilizing a previously reported procedure, capillary emitter tips were prepared in-house.31 A Sutter Instrument (model P-1000) was used to pull fire-polished borosilicate glass capillary tubes (o.d. 1.5 mm, i.d. 0.86 mm, 10 cm length, Sutter Instrument Item no. BF150–86-10) using the following settings: heat, 570; pull, 125; vel., 125; delay, 50; and pressure, 500. Ramp was 533. This afforded capillaries with tips having inner diameters of slightly above 1 µm (Supporting Information, Figure 1). The tips were coated with gold using a Leica EM ACE600 Sputter Coater under 9.0 E−3 mbar of argon at a rate of 0.23 nm/s for 120 s.

Preparation of Protein Solutions.

l-Glutamic dehydrogenase from bovine liver and pyruvate kinase from rabbit muscle were obtained from Sigma-Aldrich (St. Louis, MO). Ammonium acetate (AA), ammonium hydroxide, cysteamine, imidazole, trimethylamine (TEA), trimethylamine N-oxide (TMAO), lithium acetate (LiAc), sodium acetate (NaAc), potassium acetate (KAc), and cesium acetate (CsAc) were purchased from Sigma-Aldrich. Rubidium acetate (RbAc) was obtained from Stream Chemicals, Inc. (Newburyport, MA). All detergents were obtained from Anatrace (Maumee, OH). All membrane proteins, when analyzed by native MS or IM-MSin a detergent other than that in which it was purified, were exchanged on a Sephadex 200 10/300 size-exclusion column in Buffer A with the respective detergent at 2× the critical micelle concentration (CMC). Prior to analysis, all proteins were exchanged into 200 mM AA solution with 2× the CMC of the respective detergent, pH 7.5, using Biospin gel filtration spin columns (Bio-Rad, Hercules, CA). Soluble proteins were exchanged at concentrations previously reported.32 Membrane proteins were exchanged at ∼10 µM (monomer). Proteins were diluted by 50% with either the addition of equal volumes (2–4 µL) of 200 mM AA with 2× the CMC of detergent at pH 7.5 or a 200 mM AA solution with 2× the CMC of detergent at pH 7.5 containing a 2× concentration of the charge-reducing additive.

Ion-Mobility–Mass Spectrometry Experiments.

All native MS or IM-MS experiments were performed on a quadrupole ion-mobility time-of-flight mass spectrometer (Synapt G1 HDMS, Water Corp., Manchester, U.K.) equipped with a 32k m/z quadrupole filter and a Z-spray source. Gold-coated capillary tips prepared in-house were used (description earlier). Each condition was measured a minimum of three times with separate emitter tips, yielding reproducible results. All measurements utilized a source temperature of 50 °C, a backing pressure of 5.5 mbar, positive ion mode, reflectron mode (V-mode for the Synapt G1), and argon as the collision gas. Unless otherwise specified the instrument settings were optimized as previously reported.5 For soluble proteins, instrument settings were as follows: capillary voltage, 1.55 V; cone voltage, 20 V; extraction voltage, 4 V; trap collision voltage, 10 V; transfer collision voltage, 5 V; trap gas flow, 7.9 mL min−1; and IMS gas flow, 31 mL min−1. Because the soluble proteins were also used for collision cross section (CCS) calibration, the settings for the back end of the instrument (after the IMS) were kept the same as was used for membrane proteins in C10E5. For membrane proteins in C10E5, instrument settings were as follows: capillary voltage, 1.55 V; cone voltage, 200 V; extraction voltage, 10 V; trap collision voltage, 100 V (unless denoted otherwise); transfer collision voltage, 20 V; trap gas flow, 7.9 mL min−1; and IMS gas flow, 31 mL min−1. Ion-mobility settings were wave velocity 360 m/s and wave height 22 V in the IMS cell. Transfer wave velocity was 1000 m/s, and wave height was 0.5 V. For membrane proteins in C8E4, the instrument settings were identical to those described for C10E5 except the transfer collision voltage was 150 V. ELIC in C12E8 was analyzed using the identical tuning parameters as in C10E5. For membrane proteins in LDAO, the instrument settings were as follows: capillary voltage, 1.7 V; cone voltage, 75 V; extraction voltage, 5 V; trap collision voltage, 50 V (unless denoted otherwise); and transfer collision voltage, 180 V; trap gas flow, IMS gas flow, and TriWave settings were maintained the same as with C10E5. For membrane proteins in DDM, the instrument settings were as follows: capillary voltage, 1.8 V; cone voltage, 200 V; extraction voltage, 10 V; trap collision voltage, 165 V; and transfer collision voltage, 20 V; the trap gas flow, IMS gas flow, and TriWave settings were maintained the same as with C10E5. Quadrupole isolation was accomplished by selecting a particular m/z (7246 for PyrK and 6904 for ELIC) and adjusting both LM and HM to 0. The approximate window of transmission was determined by examining the isolated PyrK signal (Supporting Information, Figure 2).

Traveling Wave Ion-Mobility Spectrometry Calibration for Collision Cross Section Measurements and MS Spectra Deconvolution.

IM-MS measurements were made of l-glutamic dehydrogenase and pyruvate kinase utilizing published protocols.20,32 The front end of the Synapt G1 prior to the IMS cell was tuned in such a way as to minimize activation of the protein as previously described.32 However, the back end (i.e., from the IMS cell and onward) was kept the same as the tune settings used to analyze membrane proteins in C10E5. These data were utilized for a calibration in PULSAR so as to estimate CCS values of ELIC.32 Deconvolution of ELIC MS spectra at 200 V trap energy was conducted between 5 000 and 10 000 m/z to determine adduction of small-molecule charge reducers.33

Tryptophan Fluorescence.

Tryptophan fluorescence emission spectra of ELIC and a small tryptophan-containing peptide were collected using a PTI spectrofluorometer (Photon Technology International, now Horiba Scientific) equipped with a Peltier-controlled, four-position cell changer (Quantum Northwest). The excitation and emission slit widths were set to 4 and 5 nm, respectively. The excitation wavelength was set to 295 nm, and the emission spectra was collected from 310 to 450 nm. All ELIC samples were 3 mL at 0.06 mg/mL and were contained in a 10 mm × 10 mm quartz cuvette stirred at 25 °C. In addition to ELIC, measurements were obtained for a small control peptide (H2N-AAAWGGFL-OH, New England Peptide) at 4.28 µM.

RESULTS

Charge Reduction of ELIC in C10E5 Due To Alkali Metal Cations.

To characterize the effect of alkali metal cations on the charge-state distribution of a membrane protein, we examined a pentameric ligand-gated ion channel, Erwinia ligand-gated ion channel (ELIC). ELIC is optimized for native MS measurements when solubilized with C10E5 in a 200 mM ammonium acetate (AA) solution at pH 7.5.24 Native MS of ELIC yielded charge states centered at 23+ with a distribution that approximates a single Gaussian (Figure 1). The addition of sodium acetate (NaAc) reduced the lowest observed charge state from 20+ to 15+ (Figure 1). This magnitude of charge reduction is greater than those in previous reports that sought charge reduction in ELIC either in the negative-ion mode or by using small-molecule additives.7 In the presence of NaAc, higher charge states also became more prominent (Figure 1). Higher concentrations of NaAc, up to 10 mM, resulted in greater charge reduction (Figure 1). However, addition of 5 and 10 mM NaAc resulted in a significant increase in adduction (Supporting Information, Figure 3). In addition, the spectra of ELIC with NaAc showed multiple distinct distributions, in contrast to the NaAc-free condition, which approximated a single Gaussian often seen in native MS spectra (Figure 1).7,34

Figure 1.

Figure 1.

Open in a new tab

(Left) Mass spectra of ELIC in 200 mM AA at pH 7.5 with no NaAc additive (A), 500 µM NaAc (B), 1 mM NaAc (C), 5 mM NaAc (D), and 10 mM NaAc (E). The gray areas indicate the most intense charge state in the additive-free spectrum (23+) and the lowest observed charge state in the NaAc spectra (15+). (Right) Mass spectra of ELIC in 200 mM AA at pH 7.5 with 1 mM LiAc (F), 1 mM NaAc (G), 1 mM KAc (H), 1 mM RbAc (I), and 1 mM CsAc (J). The gray areas indicate the most intense charge state in the LiAc spectrum (25+) and the lowest observed charge state in the other spectra (15+).

In light of the charge-reducing effect of NaAc, additional metal cations were tested (in the form of their acetate salts), including lithium (LiAc), potassium (KAc), rubidium (RbAc), cesium (CsAc), magnesium (MgAc2), and calcium (CaAc2). The other alkali metals, at 1 mM, all influenced charge similarly to NaAc (Figure 1), with lithium, the smallest cation, displaying the weakest effect (Figure 1) and cesium, the largest cation, being the most profound (Figure 1). The charge-reducing effect of these alkali metal cations also increased with concentration (data not shown). The alkaline earth metals, magnesium and calcium, were also tested and did not yield charge reduction (Supporting Information, Figure 4). In fact, magnesium increased the charge state of the most intense ion by one. The increase of charge states with metal ions has previously been described using trivalent (3+) metal cations; the results with ELIC suggest a more modest supercharging effect by divalent cations.35

Having recognized the capacity of alkali metals to reduce the charge state of the membrane protein, ELIC, we next examined soluble proteins. l-Glutamic dehydrogenase (GluD) and pyruvate kinase (PyrK) were analyzed by native MS in the presence and absence of 1 mM NaAc. The soluble proteins were prepared for MS as previously reported; the only deviation was the addition of NaAc and/or C10E5 to the respective sample.32 Neither protein showed charge reduction with the addition of 1 mM NaAc without (Supporting Information, Figures 5 and 6) or with 0.06% C10E5 (data not shown). We suggest that alkali metal additives at the concentrations examined do not significantly alter the charge state of these soluble proteins. In summary, alkali metal cations cause marked charge reduction in the membrane protein ELIC solubilized in C10E5. This effect is concentration-dependent, is greater with larger alkali cations, and is not present for soluble proteins under the same experimental conditions. Moreover, the valence of the metal cation plays a significant role as divalent metal cations increase the charge state slightly.

Comparison of Other Charge-Reducing Additives.

To elucidate the mechanism of charge reduction by alkali metals in ELIC, we analyzed the effect of other charge-reducing agents, including triethylamine (TEA), imidazole, and trimethylamine N-oxide (TMAO).14,15,20,21 Additionally, we examined the possibility that charge reduction is occurring through a specific interaction with the protein by comparing the aforementioned charge-reducing agents with the effect of a known agonist of ELIC, cysteamine.36 Imidazole, TEA, TMAO, and cysteamine were separately introduced, and the respective samples were analyzed by native MS. The addition of imidazole or TEA resulted in a small charge-reducing effect, while TMAO afforded the largest effect, consistent with a prior report15 (Figure 2). Addition of cysteamine induced charge reduction with a shift of the most intense central signal from 23+ to 21+ and a bimodal distribution (Figure 2F). Alkali metal cations also induce a multimodal distribution of charge states, while the other charge-reducing additives examined earlier exhibit a unimodal distribution with overall less charge reduction. This dissimilarity, both in the degree of charge reduction and the shape of the charge state distributions, is suggestive that alkali metals and possibly cysteamine do not induce their effect by the same mechanism as the known charge-reducing agents. The fact that the known charge-reducing agents consistently reduce charge of soluble proteins37 while alkali metals do not also strongly suggests that they act through a different mechanism.

Figure 2.

Figure 2.

Open in a new tab

Mass spectra of C10E5-solubilized ELIC in 200 mM AA at pH 7.5 (A) with 1 mM NaAc (B), 30 mM TMAO (C), 30 mM TEA (D), 30 mM imidazole (E), and 30 mM cysteamine (F). The gray area shows the most intense charge state of the additive-free spectrum (23+).

It is possible that charge-reducing additives may bind to the protein and induce a structural change, altering the solvent-accessible surface area and thus the observed charge states.10,11,38 This possibility was examined by assessing binding of the charge-reducing agent to ELIC, the collision cross section (CCS) of ELIC, and the tryptophan fluorescence of ELIC in the absence and presence of the additive. To identify additives that bind with relatively high affinity, spectra of ELIC were collected at 200 V trap energy with and without the additives, and charge states within 5 000–10 000 m/z were deconvoluted in Unidec.33 Addition of imidazole, TEA, and TMAO did not show a mass shift of the peak intensity compared to the additive-free spectrum (Supporting Information, Figure 7). In contrast, addition of NaAc resulted in an approximately +90 Da shift of the ELIC mass, possibly corresponding to the retention of up to approximately four sodium adducts (Supporting Information, Figure 7). While the individual bound species could not be resolved, the addition of cysteamine resulted in a broad adducted peak centered at approximately +140 Da and dropping off at approximately +300 Da, likely corresponding to up to four bound cysteamine moieties (Supporting Information, Figure 7). Assigning the mass change to agonist binding seems reasonable because ELIC possesses five agonist binding sites.36 Because the sodium and cysteamine spectra showed evidence of binding to ELIC and induced multimodal charge-state distributions,10 we explored the possibility that these charge state distributions are reflective of different protein conformations in solution.

The estimated CCS of ELIC across multiple charge states was examined for the additive-free condition, 1 mM NaAc, and 30 mM cysteamine. The CCS values ranging from charge states 19+ to 27+ (additive-free), 16+ to 28+ (NaAc), and 18+ to 27+ (cysteamine) were nearly indistinguishable between these samples (Supporting Information, Figure 7). The increase in CCS at higher charge states likely stems from the rise in Coulombic repulsion.12 Thus, the CCS data indicate that the observed charge reduction with NaAc and cysteamine is unlikely to arise from major changes in the ELIC structure in solution.39

To further evaluate ELIC structure in solution, we measured the intrinsic tryptophan fluorescence emission of ELIC solubilized in 0.06% C10E5 (Supporting Information, Figure 9) or 0.02% n-dodecyl-β-d-maltopyranoside (DDM) (Supporting Information, Figure 10) in the presence of the various additives. Addition of any of the known charge reducers and NaAc did not alter the fluorescence emission in C10E5 or DDM (Supporting Information, Figures 9 and 10), except for a small increase with 100 mM NaAc and 20 mM Tris in C10E5 (Supporting Information, Figure 9, condition 3). Conversely, the addition of cysteamine reduced the fluorescence emission of ELIC by ∼25% in C10E5 and DDM (Supporting Information, Figures 9 and 10). A small peptide containing a single tryptophan residue (AAAWGGFL) was utilized as a control and showed a small reduction in fluorescence emission with cysteamine (Supporting Information, Figures 11–13), which does not account for the fluorescence change seen in the protein. These data suggest that all charge-reducing additives except cysteamine do not significantly alter the ELIC structure in solution and are consistent with the CCS measurements. Whether cysteamine induces conformational changes not resolved by ion mobility that alter the charge state distribution is less clear.

Detergent-Specific Charge Reduction of ELIC.

Because charge reduction by alkali metals is present in the C10E5-solubilized membrane protein, ELIC, and not in soluble proteins, the role of other detergents commonly used for native MS was examined. The detergents studied included n-dodecyl-β-d-maltopyranoside (DDM), n-dodecyl-N,N-dimethylamine-N-oxide (LDAO), and other PEG-based detergents: tetra-ethylene glycol monooctyl ether (C8E4) and octaethylene glycol monododecyl ether (C12E8). The addition of 1 mM NaAc to ELIC in DDM resulted in modest charge reduction (the peak charge state was reduced from 28+ to 26+), as well as a reduction in the tetramer signal (Figure 3A). As seen with C10E5, nearly the entire series of alkali metals caused charge reduction and reduced tetramer signal, an effect that scaled with the size of the alkali metal (Supporting Information, Figure 14). Lithium failed to reduce the tetramer signal despite reducing the charge. LDAO showed the lowest charge states, consistent with previous reports for other membrane proteins,6,13 but the addition of 1 mM NaAc to LDAO-solubilized ELIC had no influence on the charge states (Figure 3B). Interestingly, there appeared to be multiple distributions of charge states in both LDAO spectra, which are similar to the multiple distributions seen in alkali metal-doped spectra of ELIC in C10E5 (Figure 1).

Figure 3.

Figure 3.

Open in a new tab

Mass spectra of ELIC in 200 mM AA at pH 7.5 solubilized in 0.02% DDM (A), 0.05% LDAO (B), 0.06% C10E5 (C), 0.5% C8E4 (D), and 0.09% C12E8 (E). For each detergent, the bottom spectrum is additive-free and the top is with 1 mM NaAc. The narrow gray box with a black outline highlights the most intense charge state of the NaAc-free spectrum. The larger gray gradient box highlights the ELIC tetramer when it is clearly present.

C10E5 belongs to the polyethylene glycol (PEG) family of detergents, which also includes C8E4 and C12E8; these other PEG-based detergents were also examined. ELIC in C8E4 displayed a 24+ centered charge state distribution in the NaAc-free spectrum and a broader distribution centered at 22+ in the presence of 1 mM NaAc (Figures 3D and 4). The addition of NaAc to ELIC in C8E4 showed less charge reduction than that in C10E5 and did not produce multiple charge state distributions as in C10E5. The other alkali metals were also examined, which showed increasing charge reduction from lithium to potassium and a regression of this effect through rubidium and cesium (Figure 4). This trend contrasts with that observed for C10E5 (Figure 1). In C12E8, the most abundant charge state was 20+, which is the lowest of the PEG-based detergents (Figures 3E and 4). There was also a modest tetramer signal that was absent in C8E4 and weak in C10E5. Addition of 1 mM NaAc resulted in marked charge reduction and a multimodal distribution similar to that in C10E5 (Figure 4). Unlike C10E5 or C8E4, all alkali metals caused a similar degree of charge reduction in C12E8 (Figure 4).

Figure 4.

Figure 4.

Open in a new tab

(Left) Mass spectra of C8E4-solubilized ELIC in 200 mM AA at pH 7.5 (A) with 1 mM LiAc (B), 1 mM NaAc (C), 1 mM KAc (D), 1 mM RbAc (E), and 1 mM CsAc (F). The gray area shows the most intense charge state of the alkali metal-free spectrum (24+). The black dotted box area shows the most intense charge state (19+) of the most charge-reduced spectrum (KAc). (Right) Mass spectra of C12E8-solubilized ELIC in 200 mM AA at pH 7.5 (G) with 1 mM LiAc (H), 1 mM NaAc (I), 1 mM KAc (J), 1 mM RbAc (K), and 1 mM CsAc (L). The gray area shows the most intense charge state of the alkali metal-free spectrum (20+). The larger gray gradient box highlights the ELIC tetramer when it is readily present.

Charge Reduction in Other Membrane Proteins.

Alkali metal charge reduction of ELIC is dependent on detergent; to determine if other membrane proteins are influenced in a similar fashion, we examined additional proteins, including mouse voltage-dependent anion channel (mVDAC1), the inward rectifier potassium channel (KirBac1.1), and the mechanosensitive channel of large conductance [SaMscL-(CΔ26)]. While the resolution of spectra varied due to challenges of protein desolvation and lipid adduction, the addition of alkali metal cations to mVDAC1 (Figure 5) and KirBac1.1 (Supporting Information, Figure 15) in C10E5 afforded charge reduction. In contrast, mVDAC1 and SaMscL in LDAO showed no significant change in charge states with 1 mM NaAc (Supporting Information, Figures 16 and 17). Interestingly, mVDAC1 in C10E5 with NaAc showed reduced charge of both monomer and dimer while also displaying the evolution of a trimer. The trimer was not sampled in the NaAc-free spectrum. These results are consistent with the detergent-specificity of alkali metal cation charge reduction in ELIC.

Figure 5.

Figure 5.

Open in a new tab

Mass spectrum of C10E5-solubilized mVDAC1 in 200 mM AA at pH 7.5 (top) with 1 mM NaAc (bottom). The green triangle is the monomer, the orange pentagon is the dimer, and the blue trapezoid is the trimer.

Post-Quad Charge Reduction.

It is counterintuitive that the addition of charged species to the protein solution would lead to charge reduction. However, this could occur from the loss of alkali metal cations such as sodium during the ESI process. On the basis of the combined charged residue field-emission model (CCRFEM), the charge states of a protein are determined by its solvent-accessible surface area39 (surface area equivalent diameter) and the identity of charge carriers present in the droplet.40 For charge reduction to occur by the loss of sodium, sodium must be lost late in the ESI process, when excesses of other charge carriers are not present.40 To test the possibility that charge reduction is occurring late in the ESI process, the 26+ and 27+ charge states of ELIC with or without 1 mM NaAc were isolated by the quadrupole. The quadrupole acts as a minimum charge filter for ELIC; only ELICs with 26+ charge or higher (i.e., higher charge than 26+ and correspondingly higher mass due to adduction) are transmitted. The spectrum with 1 mM NaAc showed three lower charge states (25+, 24+, and 23+) that are significantly less apparent in the NaAc-free spectrum (Figure 6). The presence of charge states below 26+ indicates that a proportion of ELIC, in which charge has already been deposited, is undergoing further loss of charge, presumably through the loss of sodium cations. Given the low intensity of these charge-reduced species, it is likely that some proportion of this process is occurring prior to the quadrupole.

Figure 6.

Figure 6.

Open in a new tab

Mass spectra of C10E5-solubilized ELIC in 200 mM AA at pH 7.5 (top), with 1 mM NaAc (bottom). The gray area indicates the approximate window of transmission from the quadrupole as determined experimentally (see Experimental Section). Green dots highlight the charge states present with 1 mM NaAc. The magnified mass spectrum (right) is an overlap of both the no-additive (red) and 1 mM spectra (black).

DISCUSSION

Utility of Alkali Metal-Induced Charge Reduction.

Reducing the charge state of membrane proteins is advantageous in native MS or IM-MS studies to preserve protein structure and small-molecule binding in the gas phase.15 Charge reduction has previously been achieved through modification of the mass spectrometer or the addition of small-molecule additives to the sample solution.11,15,17 Instrument modification is prohibitive for many, and the addition of exogenous small molecules to a solution may result in undesirable interactions. Both of these issues are potentially avoided through the addition of low concentrations of alkali metal cations when examining membrane proteins solubilized in DDM or PEG-based detergents. The latter shortcoming is especially overcome when using sodium or potassium, as both ions are ubiquitous in biology, and should produce minimal perturbation of higher-order structure. Moreover, small-molecule charge-reducing additives are typically added to samples at concentrations of 10–250 mM, which may alter sample viscosity and inhibit passive flow through the emitter tip. By contrast, alkali metals at low concentrations perfuse with ease while achieving marked charge reduction. These low concentrations of alkali metals are nearly the most effective charge-reducing additives, second only to TMAO at high concentrations. A notable caveat of the alkali metal charge reduction is peak broadening seen when the higher concentrations of alkali metal acetate salts were used. However, significant charge reduction with minimal increased adduction could be achieved with 1 mM NaAc. Thus, in the presence of certain detergents, alkali metal cations provide a practical and efficacious means of charge reduction for native MS of membrane proteins.

Protein Conformation, Protein Identity, and Detergent.

To understand how alkali metal cations induce charge reduction of membrane proteins, we examined multiple possibilities, including a change in protein conformation. This possibility was initially bolstered by the understanding that multiple charge distributions, as seen with alkali metals in ELIC, often reflect multiple protein conformations in solution.10 Through the examination of CCS values and tryptophan fluorescence, we determined that ELIC structure was not dramatically altered in the presence of sodium. Thus, changes in protein conformation are unlikely to account for the observed charge reduction. However, it may be possible that ELIC conformational changes not resolved in this study underlie the observed multimodal charge state distribution, while an additional ESI-related mechanism causes the reduction in charge states.

The determinants of charge reduction by alkali metal cations were further examined by testing different proteins and detergents. We determined that the charge states of soluble proteins were unaltered by 1 mM NaAc, while other membrane proteins including KirBac1.1 and mVDAC1 showed charge reduction with NaAc in C10E5 but not LDAO. Similarly, the effect of alkali metals on ELIC was detergent-specific with no charge reduction in LDAO, modest charge reduction in DDM, and greater charge reduction in the PEG-based detergents. Among the PEG-based detergents, the magnitude of charge reduction was lowest with C8E4 and greatest with C12E8. The effect of alkali metal size on charge reduction by the various detergents also differed. Thus, the charge reduction by alkali metals is specific to membrane proteins and certain detergents, particularly PEG-based detergents. The varying effects of alkali metals with different PEG-based detergents show the significance of the PEG unit length in the mechanism of charge reduction.

Putative Mechanism of Alkali Metal Charge Reduction.

On the basis of the findings of this study, we posit an ESI-based mechanism for alkali metal cation charge reduction of membrane proteins. In CCRFEM, charge is determined by the solvent-accessible surface area of the macromolecule (charged-residue model) and the emission of charge carriers (field-emission model).40,41 On the basis of this model, we propose a mechanism where metal cations serve as charge carriers that either remain bound to the protein, increasing protein charge, or are lost after charge deposition, decreasing protein charge. For alkali metal cations to be effective charge carriers, they must be present at sufficient abundance around the protein. In an ELIC sample consisting of 200 mM AA and 1 mM alkali metal acetate, <0.5% of the cations are alkali metals. If 20+ is the lowest charge state of ELIC in the absence of alkali metals, then the 15+ charge state observed in the presence of alkali metals requires emission of 5 alkali metal cations on average. This means that ∼25% of the charge carriers should be alkali metal cations. This enrichment of alkali metal cations in the droplet could be achieved through specific interactions with the detergent, especially PEG-based detergents. Computational and experimental studies indicate that various PEG molecules form strong interactions with alkali metal cations.42–44 Additionally, cyclic ethers have been used to abstract sodium from macromolecules in MS experiments to reduce charge.23 Thus, binding of alkali metal cations to PEG-based detergents may produce a higher local concentration of metals, forcing other charge-contributing species, such as ammonium, to the exterior of the droplet where they are emitted.45 The proposed mechanism requires that PEG-based detergents bind alkali metal cations preferentially over ammonium, which is consistent with the reported capacity of PEG molecules to discriminate between these two ions despite similar properties.46–48 Small alkali metal cations such as sodium and lithium interact more strongly with proteins than ammonium, and so it is possible that this will enrich these cations in the droplet. However, because the charge-reducing effect is not present in soluble proteins, it is unlikely that cation–protein interactions contribute significantly to this effect.

Within the scope of this mechanism, the magnitude of charge reduction is proportional to the capability of the detergent micelle to bind and thus enrich alkali metal cations. Because DDM and LDAO display modest and absent charge reduction, respectively, it may be concluded that they fail to interact with the alkali metal cations in the same manner as PEG detergents do. The greater charge reduction of C12E8 and C10E5 over C8E4 is consistent with the longer PEG-based detergents having higher affinity, and therefore enrichment, of alkali metal cations in the droplet. The different effects of lithium through cesium on charge reduction in C8E4, C10E5, and C12E8 may also reflect the relative affinity of these detergents for the different alkali metal cations. C12E8 does not show an alkali metal size dependence on charge reduction, suggesting that the eight ethylene glycol unit is long enough to complex all metals with similar affinity. Conversely, C8E4 shows increasing charge reduction from lithium to potassium, which then decreases from rubidium to cesium, suggesting that the four ethylene glycol units form weaker complexes with the larger rubidium and cesium cations.

With local enrichment of alkali metals cations in the droplet about the micelle (Figure 7(1)), more poorly solvated charge carriers are emitted as the droplet shrinks (Figure 7(2)).40 When the macromolecule–detergent complex is mostly desolvated, it has adopted the maximum charge according to its solvent-accessible surface area and the emission of solvated charge carriers (Figure 7(3)).39 At this point, when other charge carriers have been emitted, we propose that further charge reduction occurs by the loss of alkali metal cations (Figure 7(4, 5)). It is plausible that this loss of cations occurs by two means: the alkali metal cation is emitted (Figure 7(4)) or it is carried away by the detergent in a complex (Figure 7(5)). Computational modeling indicates a high affinity of PEG molecules for alkali metal cations in aqueous solution, which is 10-fold lower in the gas phase.42 Conversely, PEG5 units can form strong multipoint complexes with sodium ions in the gas phase, suggesting that PEG–cation complexes could be emitted, stripping charge from the macromolecule.49 Ultimately, the protein has lost much of the alkali metal cation and remaining detergent (Figure 7(5)),5,50 resulting in a detergent-free, charge-reduced protein (Figure 7(6)).

Figure 7.

Figure 7.

Open in a new tab

Proposed mechanism of charge reduction by the addition of alkali metals. The black circle indicates the solvent droplet surrounding the protein. Gray circles with tails indicate detergent molecules. White circles denoted with “+” indicate metal cations. BH+ indicates more poorly solvated charge carriers, and AH+ indicates better-solvated charge carriers.

Along with charge reduction, higher charge states are also observed in ELIC with the addition of alkali metal cations. It is unlikely that these charge states reflect denaturation of ELIC in solution because major structural changes were not appreciated in the tryptophan fluorescence measurements, and it seems implausible that the addition of low millimolar concentrations of sodium to 200 mM ammonium should significantly alter ELIC structure, an ion channel that is nonselectively permeable to cations. Increased alkali metal cation deposition on the protein may cause this increased charge. Another possibility is that alkali metal cations via a ternary complex with PEG detergents may bridge the transient formation of dimers of ELIC pentamers. These dimers of pentamers may then dissociate with asymmetric charge partitioning, resulting in both higher and lower charge states with a bimodal charge state distribution.51

CONCLUSION

The addition of sub-to low-millimolar concentrations of alkali metal acetate salts to detergent-solubilized membrane proteins for native MS affords marked charge reduction. The role of detergents, particularly PEG-based detergents, in this process suggests a mechanism whereby alkali metal cations bound to detergent are enriched in the droplet during ESI, and charge reduction occurs by loss of these cations. The use of alkali metals, especially sodium and potassium, for charge reduction of membrane proteins may be advantageous compared to other charge-reducing agents because of their efficacy and because they are biologically ubiquitous and less likely to have undesired interactions with the protein.

Supplementary Material

Supplementary Material

ACKNOWLEDGMENTS

We thank Arthur Laganowsky and Yang Liu for advice on native MS analysis of ELIC and Michael Gross for reviewing the manuscript. Funding was provided by the NIH (K08GM126336) and the Center for the Investigation of Membrane Excitability Diseases.

Footnotes

ASSOCIATED CONTENT

Supporting Information

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.analchem.0c00454.

MS and fluorescence data (PDF)

The authors declare no competing financial interest.

Contributor Information

Gregory T. Dekoster, Biochemistry and Molecular Biophysics, Washington University in St. Louis, St. Louis, Missouri 63110, United States

Carl Frieden, Biochemistry and Molecular Biophysics, Washington University in St. Louis, St. Louis, Missouri 63110, United States.

Wayland W. L. Cheng, Departments of Anesthesiology, Washington University in St. Louis, St. Louis, Missouri 63110, United States.

REFERENCES

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary Material
NIHMS1593372-supplement-Supplementary_Material.pdf (702.3KB, pdf)

RESOURCES