Ion mobility-mass spectrometry reveals that α-hemolysin from Staphylococcus aureus simultaneously forms hexameric and heptameric complexes in detergent micelle solutions

Abstract

Many soluble and membrane proteins form symmetrical homooligomeric complexes. However, determining the oligomeric state of protein complexes can be difficult. Alpha-hemolysin (αHL) from Staphylococcus aureus is a symmetrical homooligomeric protein toxin that forms transmembrane β-barrel pores in host cell membranes. The stable pore structure of αHL has also been exploited in vitro as a nanopore tool. Early structural experiments suggested αHL forms a hexameric pore, while more recent x-ray crystal structure and solution studies have identified a heptameric pore structure. Here, using native ion mobility-mass spectrometry (IM-MS) we find that αHL simultaneously forms hexameric and heptameric oligomers in both tetraethylene glycol monooctyl ether (C₈E₄) and tetradecylphosphocholine (FOS-14) detergent solutions. We also analyze intact detergent micelle-embedded αHL pore-like complexes by native IM-MS without the need to fully strip the detergent micelle, which can cause significant gas-phase unfolding. The highly congested native mass spectra are deconvolved using Fourier- and Gábor-transform (FT and GT) methods to determine charge states and detergent stoichiometry distributions. The intact αHL micelle complexes are found to contain oligomeric state-proportional numbers of detergent molecules. This evidence, combined with IM data and results from vacuum molecular dynamics simulations, is consistent with both the hexamer and the heptamer forming pore-like complexes. The ability of αHL to form both oligomeric states simultaneously has implications for its use as a nanopore tool and its pore formation mechanism in vivo. This study also demonstrates more generally the power of FT and GT to deconvolve the charge state and stoichiometry distributions of polydisperse ions.

Graphical Abstract

Introduction

At least half of all known soluble and membrane-associated proteins form multimeric complexes, with many of these complexes forming as homooligomers^1–3. Most homooligomers display some form of symmetry, and evolution favors larger complexes due to the increased stability afforded by minimizing solvent exposure of hydrophobic regions and functionality constraints that require large structures^1–5. However, determining the functional oligomeric form of protein complexes can be difficult, especially for transmembrane proteins that require suitable environments not easily amenable to traditional structural elucidation techniques (e.g. x-ray crystallography, electron microscopy, analytical ultracentrifugation, and electrophoretic techniques). Both oligomeric state heterogeneity^6–8 and oligomeric state dependence on solution conditions^9,10 have been reported and can further increase the difficulty of structure characterization.

An example of a large homooligomer protein complex with more than one observed oligomeric state is alpha-hemolysin (αHL) from Staphylococcus aureus (S. aureus)¹¹. S. aureus is a common human pathogen that can cause severe skin and respiratory tract infections leading to extensive disease burdens in the US and internationally¹². αHL is a key virulence factor for S. aureus which forms a transmembrane β-barrel pore complex. This pore structure permeabilizes cell membranes in many types of host cells as well as causes a broad range of other toxic cellular insults^13,14. Beyond the medical relevance of studying the role of αHL pores in S. aureus infections, the ability of αHL to form stable transmembrane pore complexes in vitro has led to its development as a nanopore tool¹⁵ for molecular sensing of small molecules¹⁶, nucleotide sequencing^17–19, and directed movement of nanometer-sized cargo within αHL pores²⁰.

Early experiments using electron microscopy (EM)^21–23, atomic force microscopy²⁴, electrophysiology²⁵, and solution-based size-exclusion chromatography and analytical ultracentrifugation²⁶ indicated αHL can form a hexameric complex. However, the first high-resolution x-ray crystal structure of αHL was heptameric^27,28, and several other crystal structures of αHL pore complexes solved since then are heptameric^29–31. The heptameric state is also supported more indirectly by other solution studies, such as photobleaching of fluorescently-labeled αHL subunits in pore complexes³², pore conductivity measurements made using electrophysiology²⁵, and experiments with covalently-linked αHL subunit dimers³³. Over the course of the last couple decades, with these combined studies, the consensus view has been that αHL forms only functional heptameric pore complexes^11,33,34. It has been proposed that the identification of a hexameric complex may have been due to the image processing techniques used in EM and that previous size-exclusion chromatography and analytical ultracentrifugation studies may have lacked sufficient resolution to distinguish between hexamer and heptamer forms²⁶.

Native ion mobility mass spectrometry (native IM-MS) using nano-electrospray ionization (nESI) has proven a useful tool in structural biology for identifying oligomeric states of biological complexes due to its ability to maintain native non-covalent interactions upon ionization of protein complexes³⁵. For example, native IM-MS was used to determine the oligomeric states populated by anthrax toxin prepore^9,10 at different solution pH and by the lysenin pore³⁶, both of which form β-barrel pore complexes similar to αHL. Native IM-MS has also been shown to be a powerful tool for studying small-molecule association with high chemical specificity and without the need for crystallization of membrane protein complexes^37–40. Here, we use native IM-MS of αHL pore-like complexes formed in two different detergent solutions (tetraethylene glycol monooctyl ether (C₈E₄) as an ether-like detergent, and n-tetradecylphosphocholine (FOS-14) as a lipid-like detergent), to show that αHL forms both hexameric and heptameric complexes simultaneously in both detergent solutions. Under the tested solution conditions, the heptameric complex is the dominant species, but a sizable population of hexameric complexes is detected, and this result was verified on two different mass spectrometer platforms (an IM-time-of-flight instrument and an Orbitrap instrument without IM). Using the phospholipid-like detergent FOS-14, αHL complexes embedded in nearly-intact detergent micelles are resolved enough in the native IM-MS data to characterize both their stoichiometry and collision cross section (CCS). These native mass spectra are highly congested due to hundreds of overlapped mass spectral peaks, but Fourier transform (FT) and Gábor transform (GT) are used to deconvolve both the charge state and stoichiometry distributions of associated detergent molecules^41–43. The ability of αHL to form hexameric and heptameric pore-like complexes has ramifications for the mechanism of αHL pore formation and the use of αHL as a nanopore tool.

Experimental Section

Expanded method and experimental details can be found in the Supporting Information. Briefly, lyophilized monomers of αHL from S. aureus were purchased from Millipore Sigma (St. Louis, MO, USA) and were resuspended in deionized water to a concentration of 0.5 mg/mL. 150 μL of this 0.5 mg/mL αHL monomer solution was centrifugally concentrated to ~4x in the presence of either C₈E₄ (32 mM, CMC = 8 mM) or FOS-14 (2mM, CMC = 0.12 mM) detergent micelles in 200 mM ammonium acetate pH 7.5 to induce oligomerization and pore formation²⁶. A portion of this sample (10 μL diluted to 30 μL with more detergent solution) was then used to buffer exchange into 200 mM ammonium acetate at pH 7.5 with 2x the CMC of the appropriate detergent for mass analysis. Oligomer formation was also checked using SDS-PAGE (Figure S-1). As part of a detergent screen, the detergents n-dodecyl-β-D-maltopyranoside (DDM) and n-dodecyl-N,N-dimethylamine-N-oxide (LDAO) were additionally tested, but under similar instrumental conditions no oligomers of αHL were detected. Mass spectra were acquired on either a Waters Synapt G2-Si Quadrupole–Ion-Mobility–Time-of-Flight (University of Oregon, Eugene, OR) or Thermo Scientific Exactive Plus extended mass range Orbitrap (University of California, San Francisco, CA) mass spectrometer, and all ion mobility-mass spectra were acquired on a Waters Synapt G2-Si. Both instruments were equipped with a nESI source, and tuning parameters can be found in the Supporting Information. Gas-phase compaction of model structures for native-like pores were simulated using GROMACS v. 2016.4, and theoretical collision cross sections⁴⁴ were computed using Collidoscope. These experiments pose no unusual safety hazards.

Results and Discussion

αHL forms both hexameric and heptameric pore-like complexes in C₈E₄ detergent micelles.

In order to characterize effects of detergent on αHL oligomerization, we initially obtained a mass spectrum of αHL in detergent-free solutions. A typical native mass spectrum of αHL monomers (~5 μM) formed by nESI from detergent-free solutions using a Waters Synapt Q-IMS-ToF mass spectrometer is shown in Figure 1A. αHL monomer ions form a narrow charge state distribution from 10–12+ indicating that the monomer ions are compact. For each charge state there are two peaks of similar abundance attributed to the presence of the αHL monomers (33,259 ± 1 Da) and an unknown protein present in the commercial αHL sample (34,126 ± 1 Da; see Figure S-2 for more information). The measured αHL monomer mass matches well with the theoretical sequence mass of 33,248 Da for the mature 293 amino acid protein. At higher m/z (~4500–5000) there is a low-abundance distribution attributed to another contaminant protein with a mass that is inconsistent with any oligomeric state of the αHL monomers. No oligomers of αHL monomers are detected at this concentration and under these detergent-free solution conditions.

Figure 1.

Native mass spectra from the Synapt Q-IMS-ToF instrument of αHL monomers in detergent-free solutions and αHL hexamer and heptamer complexes in C₈E₄ detergent micelle solutions. (A) Mass spectrum of αHL monomers in detergent-free solutions of 200 mM ammonium acetate. The presence of αHL monomers and an unidentified co-purified protein are seen while no oligomers are present. Inset shows the small abundance of a second contaminant. (B) Mass spectrum of αHL oligomerized complexes formed in C₈E₄ detergent solutions with 200 mM ammonium acetate and under the instrumental conditions of sample cone at 50 V and trap at 75 V. (C) IM-MS spectrum under the same instrumental conditions as in (B) showing multiple unfolding states.

nESI of αHL monomers concentrated in the presence of C₈E₄ detergent micelles at a concentration above the critical micelle concentration (CMC) (32 mM for concentration step, CMC = 8 mM) yields two higher-order oligomeric states with masses of 199,553 ± 15 Da and 232,806 ± 16 Da and charge state distributions of 25–33+ and 25–37+, respectively (Figure 1B). C₈E₄ was chosen initially for its compatibility with native nESI and ease of removal in the gas phase of the mass spectrometer instrument^39,40,45. These measured masses match the expected masses for the hexameric (199,554 Da) and heptameric (232,813 Da) oligomeric states based on the measured monomer mass. By contrast, there is no evidence for the incorporation of the heavier (34 kDa) unidentified protein in the observed hexamers or heptamers.²⁸

Native mass spectra of αHL complexes embedded in FOS-14 detergent micelles confirms solution hexameric and heptameric pore-like complexes.

In order to clearly resolve the two αHL oligomeric distributions in the above experiments, moderately activating instrumental conditions (sampling cone 50 V, trap 75 V, transfer 5 V) were used to strip off nearly all the detergent, as is done in the majority of native IM-MS studies to date of transmembrane proteins embedded in detergent micelles. Figure 1C shows the IM-MS data under the same instrumental conditions used to obtain the mass spectrum in Figure 1B. Under these instrumental conditions, most of the charge states for both the hexamer and heptamer have multiple unfolded conformational states. Non-native monomers are detected in low abundance (Figure S-3) that, in principle, could have been ejected from activated heptameric complexes, resulting in the hexameric distribution. To eliminate this possibility, 37+ heptamer ions were first isolated under conditions where the ions remained compact and folded (sampling cone 50 V, trap 25 V) and then these ions were activated in the trap (125 V) to dissociate them into high-charge monomers (14–25+) and stripped hexamers (15–24+) (Figure S-4). The drastically different drift time and charge state distributions for the collision-induced hexamers show that the hexameric series in Figure 1B does not arise from gas-phase activation of the heptamers. We also tested whether the the formation of hexamer is a result of early activation in the electrospray process as protein ions are transferred to the gas phase. Increasing the sample capillary voltage does not increase the abundance of hexamer ions relative to heptamer ions but does significantly diminish signal quality (Figure S-5). These combined experiments demonstrate that the hexameric complex is indeed an oligomeric state formed in solution.

To more directly confirm the presence of native heptamers and hexamers in detergent solution, we acquired mass spectra under conditions where the detergent micelles surrounding the ions are largely preserved. However, obtaining resolved mass spectra of micelle-embedded αHL complexes in C₈E₄ was difficult. Detergent resolution was only obtained on protein complex ions under relatively high-activation conditions for which only a small number of detergent molecules remained adducted to protein complex ions (Figure 1B). αHL has been co-crystallized with each subunit bound to glycerol phosphocholine in a groove between the rim and stem domains of the complex²⁹. This evidence for phosphocholine binding has been used to reason why αHL appears to bind and form pores preferentially to membranes containing phosphocholine lipids⁴⁶. The detergent FOS-14 is a lipid-like detergent with a phosphocholine headgroup, and thus it might be expected to form strong interactions with αHL pores that encourage more native-like oligomer formation than C₈E₄. We reasoned that detecting the hexameric and heptameric oligomeric states in a more phospholipid-like detergent would remove doubt about the hexameric state being artefactual due to the ether-based C₈E₄ detergent. FOS-14 and other phospholipid-like detergents have not been reported previously as a vehicle for transmembrane protein native IM-MS studies.

Following a similar procedure for oligomer formation in C₈E₄, αHL monomers were concentrated in FOS-14 detergent solutions at a FOS-14 concentration (2mM) well above the CMC (~0.12 mM). Under the same gentle nESI conditions, native mass spectra of these samples indicate oligomerized complexes associated with large FOS-14 micelles and much less stripping of detergent than for C₈E₄ (Figure 2A–D). At lower m/z a large distribution of protein-free detergent micelles is present at much higher abundance than that of the αHL micelle-embedded ions (Figure S-6). Due to the polydispersity of detergent stoichiometry in the micelles, the αHL ions embedded in detergent micelles have complicated distributions of peaks in the mass spectrum (Figure 2A–B) with overlapping charge state and detergent distributions that are difficult to assign. However, these overlapping distributions lend themselves well to Fourier transform (FT) and Gábor transform (GT) based analysis developed by our laboratory (iFAMS software) to deconvolve the charge state and detergent distributions^41–43.

Figure 2.

Native mass spectrum from the Synapt Q-IMS-ToF instrument of αHL micelle-embedded complexes in FOS-14 detergent with 200 mM ammonium acetate with the sample cone at 150 V and trap at 50 V. (A) GT spectrogram is shown with the IM-MS mass spectrum cutout across the top and FT spectrum down the right. Each individual point in the spectrogram corresponds to a charge state that is then reconstructed on the mass spectrum above (colored traces). For the heptameric series, secondary harmonics were resolved and included in the reconstruction, resulting in higher resolution than for the hexameric series. (B) IM-MS spectrum of αHL micelle-embedded complexes showing drift time overlap of hexamer and heptamer distributions that remain compact and folded. (C) Detailed stoichiometry analysis of mass spectrum shown in (A). Inset tables provide the detergent stoichiometry distributions for each individual charge state from the GT with the ± representing the standard deviation in the detergent stoichiometry. The inset shows the repeating peaks from detergent association. Colors in table match with their respective detergent distribution for each charge state. (D) Zero-charge spectrum of the combined charge state data from the GT. Dashed vertical lines correspond to the masses calculated for detergent-stripped bare hexamer and heptamer oligomers based on the measured monomer mass.

Under the moderately-activating instrumental conditions used to collect the mass spectrum in Figure 2A–D, both hexameric and heptameric oligomeric states of αHL are detected and separated in the GT spectrogram, which have highly overlapped distributions that are not easily separated or characterized using IM-MS without deconvolution (Figure 2B). The GT also allows for the analysis of detergent stoichiometry distributions for each charge state. The charge state distributions are plotted with the mass spectrum in Figure 2C and as a combined “zero-charge” spectrum in Figure 2D. The nearly Gaussian total mass distributions in the zero-charge spectrum indicate that the αHL hexamer associates with 103 ± 24 FOS-14 detergent molecules while the heptameric complex associates with 111 ± 19 molecules of FOS-14 in these mass spectra, consistent with a roughly oligomer size-proportional micelle stoichiometry. The IM-MS spectrum in Figure 2B indicates that the αHL pore-like complexes are compact under these conditions, demonstrating that GT can be used to deconvolve and characterize these overlapped distributions without the need to strip the ions of detergent as in Figure 1C where concomitant protein unfolding is observed.

αHL pore-like ion stoichiometry is consistent across mass spectrometer platforms.

To demonstrate that the observed αHL oligomeric states are reproducible across mass spectrometer platforms, these samples were studied with nESI on an Orbitrap EMR instrument, which uses a heated ESI capillary to transfer ions into the low-pressure region of the instrument rather than a gentler “StepWave” ion guide as in the Synapt platform (Figure 3A–C). In contrast to the above-described Synapt instrument, mass spectra acquired on the Orbitrap instrument do not exhibit a large population of FOS-14 only micelles on the Orbitrap instrument, which is consistent with the Orbitrap having a harsher source. Also, on the Orbitrap more αHL complexes completely stripped of FOS-14 are detected under these conditions with peaks that are about half as wide in full-width at half-maximum compared to C₈E₄ detergent-stripped pore-like complexes on the Synapt instrument at ~7500 m/z (Figure S-7A inset)⁴². The signal for stripped complexes clearly indicates both hexamer and heptamer oligomeric states for αHL pore-like complexes with masses that closely match the detergent stripped complexes in C₈E₄ (hexamer: 199,577 ± 4 Da, heptamer: 232,845 ± 6 Da).

Figure 3.

Native mass spectrum from the Orbitrap instrument of αHL micelle-embedded complexes in FOS-14 detergent with 200 mM ammonium acetate and under the instrumental conditions of source CID 100 V and HCD at 50 V. (A) GT spectrogram is shown with the mass spectrum across the top and the FT down the right. (B) Detailed stoichiometry analysis of mass spectrum shown in (A). Tables provide the detergent stoichiometry distributions for each individual charge state pulled from the GT with the ± representing the standard deviation in the detergent stoichiometry. (C) Zero-charge spectrum of the combined charge state data from the GT. Dashed vertical lines correspond to the masses calculated for detergent-stripped bare hexamer and heptamer based on the measured mass of the monomer. For the middle overlapped distribution the number of FOS-14 units associating with the micelle-embedded hexamers and the nearly detergent-stripped heptamers is given.

Under the least activating conditions we used on the Orbitrap instrument, detecting signal for the hexameric micelle-embedded complexes is difficult in comparison to the heptamer (Figure S-7A–C). When the mass spectrum in Figure S-8A is processed with FT in iFAMS using only charge states 23–25+, which are well separated in the frequency domain, signal of hexameric complexes in detergent micelles is more clearly observed. Under these instrumental conditions, the hexameric ions are determined to contain 112 ± 32 detergent molecules while the heptameric ions associate with 121 ± 30, which is consistent with the number of detergent molecules associating with the heptamer on the Synapt instrument under the least activating instrumental conditions used (Figure S-9A–D).

Increasing the degree of in-source activation removes a small number of detergent molecules (~10) and significantly increases peak resolution yielding the mass spectrum and GT spectrogram seen in Figure 3A–C. The GT spectrogram contains three distinct distributions of αHL pore-like complexes that are hard to separate using FT alone and would be extremely difficult to analyze by conventional methods. In the GT, the most abundant distribution with the highest overall frequency values represents the heptameric αHL complexes embedded in detergent micelles that associate on average with 111 ± 17 FOS-14 molecules. The middle distribution corresponds to two strongly-overlapped distributions of nearly detergent-stripped heptamers and micelle-embedded hexamers with overlapped charge states and very similar mass distributions that are difficult to distinguish from the GT spectrogram. (Coincidentally, the 25+ micelle-embedded hexamer and 29+ micelle-embedded heptamer distributions have nearly identical abundance and m/z ranges and would be exceptionally difficult to deconvolve with other methods.) Based on the mass distribution width and average for each charge state in the middle series of Figure 3A, charge states 23–26+ are predominantly micelle-embedded hexamers, while charge states 27–29+ are attributed mostly to nearly detergent-stripped heptamer complexes. Under these conditions the micelle-embedded hexamers associate with 94 ± 15 FOS-14 molecules. The third, lowest-frequency distribution is attributed to nearly detergent-stripped hexameric complexes that on average have ~5 remaining detergent molecule adducts. Overall, these values for the masses of the stripped αHL pore-like complexes and the detergent stoichiometries for the micelle-embedded complexes are highly consistent with data acquired on the Synapt instrument and further corroborate the existence of both hexameric and heptameric pore-like complexes in solution.

Due to the size of these complexes and the similarity in molecular identity, it is unlikely the nESI ionization efficiencies of the hexamer and heptamer are drastically different. Thus, the abundance ratios seen from the mass spectrum likely reflect the abundance ratios of the hexameric and heptameric complexes in solution. The heptameric state is clearly favored at a measured ratio of detergent-stripped heptamer to hexamer of ~5:1 in both FOS-14 and C₈E₄ detergents as determined by fitting Gaussian detergent stoichiometry distributions to each charge state in Figure 1B and Figure S-7A and totaling the abundances of each oligomer, or by using Unidec (a Bayesian deconvolution algorithm) to estimate the abundances of each oligomer population^47–49. Determining the abundance ratios for each oligomeric state for the micelle-embedded αHL complexes on the Synapt and Orbitrap instruments is more complicated. From the zero-charge spectrum acquired on the Synapt (Figure 2D) the ratio of micelle-embedded heptamer to hexamer is ~10:1 in comparison to that of the Orbitrap (Figure 3C) of ~2:1 (including signal for detergent-stripped heptamers). This difference is likely due to the significantly better resolution of the mass spectra acquired on the Orbitrap instrument, which should result in more reliable reconstructed relative abundances. Based on this we conclude the abundance ratios of heptamer to hexamer in the micelle-embedded complexes are likely closer to that of the detergent-stripped complexes, i.e., ~5:1 heptamer : hexamer. Therefore, as has been previously reported by multiple techniques, the heptameric oligomer is the predominant species under these conditions, but αHL also forms a large population of hexameric complexes in detergent solutions.

IM-MS collision cross section measurements and MD-simulated collision cross section calculations of stripped heptameric and hexameric complexes reveal compact native state.

The native mass spectra of FOS-14 micelle-embedded αHL pore-like complexes show that both the hexameric and heptameric oligomers are native oligomeric states for αHL complexes in detergent solutions, but these data alone do not reveal much about the structure or conformation of the oligomeric complexes⁵⁰.

To determine whether the αHL oligomers survive in pore-like structures upon transfer to the gas-phase, triplicate IM-MS measurements of αHL hexamer and heptamer complexes formed with C₈E₄ were collected under activation conditions for which each charge state for both oligomers remained compact. Under these instrumental conditions more C₈E₄ molecules remain attached to each oligomer and the hexameric and heptameric ion distributions partially overlap both in m/z and drift time (Figure 4A, Figure S-10). αHL hexameric complexes were determined to have CCS values ranging from 90–96 nm² for charge states 29–33+, while the heptamers had CCS values ranging from 100–109 nm² for charge states 30–37+ indicating that the hexamer is ~6/7 the size of the heptamer. Together these results suggest that these detergent-stripped hexameric and heptameric complexes have globally similar compact structures.

Figure 4.

Comparison between collision cross section measurements in both detergents and computationally derived CCSs. (A) IM-MS spectrum of nearly-detergent-stripped αHL pore-like complexes in C₈E₄ detergent micelle solutions under instrumental conditions (sampling cone 50 V, trap 25 V) where no unfolding is seen. (B) Same as in (A) except using FOS-14 as the detergent (sampling cone 25 V, trap 50 V, transfer 25 V). (C) αHL heptameric pore crystal structure (orange mesh, PDB: 7AHL) and vacuum MD simulated structure showing gas-phase compaction (solid blue surface). (D) Comparison between measured CCSs for αHL heptamer and hexamer ions in C₈E₄ and FOS-14 detergent and for CCSs predicted from MD structures. The measured CCS value in parentheses for FOS-14 is the 36+ charge state which was the highest charge state observed.

IM-MS experiments on compact αHL pore-like complexes formed with FOS-14 resulted in similar CCS values to complexes formed in the detergent C₈E₄. Using IM-MS, signal for stripped heptameric complexes could be detected and separated from FOS-14 clusters that overlap in m/z, but have different drift time distributions (Figure 4B). For the heptameric series with charge states 30–36+ the measured CCS is 99–106 nm², indicating there is no significant difference in the size of the stripped heptamer ions formed using either of the two detergents tested.

It is well-known that native-like protein and protein complex ions often compact (by as much as 22%) in native IM-MS during the nESI process in comparison to their condensed-phase structures⁵⁰ but that much tertiary and even secondary structure can be preserved⁵¹. We recently showed that performing vacuum MD simulations using the GROMOS96 43a2 force field results in ion structures having calculated CCSs within 4% on average of experimental IM-MS data for a set of globular and transmembrane proteins and protein complexes commonly used as IM-MS calibration standards⁵⁰. To enable more confident comparison of experimental IM-MS data presented here for the putative hexameric and heptameric αHL pore-like complexes, vacuum MD simulations were performed at 300 K with the GROMOS96 43a2 force field for both the crystal structure of the heptameric pore (PDB: 7AHL) and a model of the hexameric pore produced by Furini et al.²⁵.⁵²Figure 4C shows the crystal structure 7AHL as a mesh surface with the aligned vacuum MD simulated structure as a solid surface. These vacuum MD simulated structures were then used to calculate CCS values in N₂ gas using the Trajectory Method in Collidoscope. After MD relaxation in vacuum, a small degree of compaction (~12% for the heptamer and ~18% for the hexamer) is predicted. Figure S-11 shows models of the hexameric pore and vacuum MD-simulated structures.

Figure 4D compares the measured and computed CCS values. The calculated CCS for the hexamer averaged 96 nm² for charge states 29+, 31+, and 33+ and 107 nm² for the 30+, 34+, and 37+ heptamer. These simulated CCSs fall inside the range of experimentally measured CCSs for both the bare hexameric and heptameric pore-like complexes and are within the expected range of error (± 4%) of the charge-state-averaged experimental CCSs for both oligomers (93/105 nm² for the hexamer/heptamer). Figure 5 summarizes all the native IM-MS CCS measurements and the computationally-derived CCS values for the uncompacted and MD compacted hexamer and heptamer structures. The detergent-stripped bare pore-like complexes and the micelle-embedded complexes overall have linear CCS trends as a function of charge state with similar slopes and differ by only a few nm², which we attribute to the presence or absence of the detergent micelles. Although CCS measurements do not provide direct evidence of a “pore” in the physiological sense (i.e., a channel capable of permeabilizing lipid bilayers), these results indicate that the native IM-MS conditions used here preserve not only the stoichiometry but also structure consistent with the crystal and model structures of the heptameric and hexameric pores from solution into the gas phase.

Figure 5.

Compilation of all measured and computationally predicted CCS values for αHL hexamers and heptamers. Vacuum MD CCS for the 33+ hexamer and the measured CCS for the 33+ hexamer are slightly offset because they are nearly identical. CCSs for αHL micelle-embedded complexes in FOS-14 determined from Fig. 2B and S-7. CCSs for αHL bare complexes formed in C₈E₄ determined from Fig. 4A.

Conclusions

Here, αHL from S. aureus is observed to adopt two oligomeric states, a hexamer and a heptamer in solution that are preserved upon transfer to the gas phase, using two different types of detergent and two different types of mass spectrometer platforms. For both the ether-based and phospholipid-like detergent used, both hexameric and heptameric detergent-stripped complexes are detected at a ratio of ~5:1 heptamer to hexamer. Based on native IM-MS results, these detergent-stripped complexes have CCSs within 4% of their respective vacuum MD simulated structure. All these observations and measurements point to the coexistence of hexameric and heptameric pore-like complexes of αHL in the condensed phase.

The native mass spectra of micelle-embedded pore-like complexes reported here illustrate the powerful capabilities of FT and GT to deconvolve charge state and mass information to allow for interpretation of these types of challenging samples without requiring detergent removal. These results also demonstrate the utility of FOS-14 as a detergent in native mass spectrometry, which has the same phosphocholine headgroup as many of the most common physiological lipids. With FOS-14, intact membrane protein micelle complexes could be transferred to the gas phase with high enough resolution for analysis with FT and GT to determine the charge state, mass, and stoichiometries of associated detergent and protein oligomeric state. The results also illustrate advantages of FT- and GT-based deconvolution methods for CCS and structure determination based on IM data, for which accurate charge states and mass determination are prerequisites. These same methods could be used to aid oligomeric state determination of other membrane protein complexes for which resolving oligomeric states may be difficult. FT and GT analysis can as well be extended to other applications that inherently produce complicated mass spectra with repeating subunits. In this case, detergent is the repeating subunit, but this type of analysis could be extended more generally, for example, to lipid-containing complexes, proteins with multiple bound isobaric glycans, and polymers.

The propensity for intact FOS-14 micelle-embedded membrane protein complexes to be transferred to the gas phase of mass spectrometer instruments highlights the significant differences in gas phase behavior between FOS-14 and C₈E₄. As observed here, C₈E₄ has been demonstrated to be readily removable from protein complexes in the gas phase, in contrast to other detergent groups such as maltosides (e.g. DDM)⁴⁰. Reading et al. argued that the ease of detergent removal may relate to the protein stability within the detergent micelle and may suggest that the ease of a detergent’s release is inversely correlated with its ability to substitute for lipid association. Here, with FOS-14 being a phospholipid-like detergent, once the micelle-embedded pore-like complexes reach the gas phase of the mass spectrometer instrument, it is more difficult to remove all the detergent with tuning conditions available on the Synapt instrument. Only minimal detergent loss and charge stripping are seen under a wide range of activation conditions (Figure S-12A–C), suggesting these micelle-embedded pore-like complexes are indeed highly stable.

In these experiments, αHL pore-like complexes are made through direct association with detergent micelles. In vivo, αHL monomers have specific cell surface binding interactions with the protein ADAM10 at nanomolar concentrations of toxin^52,53. At higher concentrations (~1 μM) αHL monomers have been shown to associate with phosphocholine lipids and to oligomerize and form pores in an analogous fashion to how pore-like complexes were formed in detergent in these experiments^46,52,54. An exciting future direction of research is to use native IM-MS techniques similar to those described here to investigate effects of protein receptor models or lipid headgroups on the oligomeric state distribution of αHL complexes or other pore forming toxins (such as anthrax toxin). Additionally, different oligomeric states of αHL pores almost certainly have different pore diameters and thus different channel conductance properties^25,33. Tailoring the oligomeric state of αHL pores by manipulating solution conditions or membrane environment, informed by IM-MS studies, could therefore enable broader control of pore diameter and conductance in applications.