OCAM: A new tool for studying the oligomeric diversity of MscL channels

Abstract

We have developed a new technique to study the oligomeric state of proteins in solution. OCAM or Oligomer Characterization by Addition of Mass counts protein subunits by selectively shaving a protein mass tag added to a protein subunit via a short peptide linker. Cleavage of each mass tag reduces the total mass of the protein complex by a fixed amount. By performing limited proteolysis and separating the reaction products by size on a blue native PAGE gel, a ladder of reaction products corresponding to the number of subunits can be resolved. The pattern of bands may be used to distinguish the presence of a single homo-oligomer from a mixture of oligomeric states. We have applied OCAM to study the mechanosensitive channel of large conductance (MscL) and find that these proteins can exist in multiple oligomeric states ranging from tetramers up to possible hexamers. Our results demonstrate the existence of oligomeric forms of MscL not yet observed by X-ray crystallography or other techniques and that in some cases a single type of MscL subunit can assemble as a mixture of oligomeric states.

Introduction

The mechanosensitive channel of large conductance (MscL) is a homo-oligomeric, stretch-activated membrane protein responsible for regulating osmotic pressure in bacteria and archaea.^1–6 The MscL channel acts as an emergency release valve that activates when cells are placed under osmotic stress. Increasing membrane tension stabilizes the channel's open form leading to the formation of a nonselective pore with an open channel conductance orders of magnitude greater than ion-selective channels. When the gene product for the Escherichia coli MscL (EcMscL) was identified as a protein of 136 amino acids, it was recognized that MscL must form a homo-oligomer to create such a high conductance channel.⁷ Early crosslinking and electron microscopy studies indicated that EcMscL formed hexamers,^8–9 but this view was subsequently revised when the crystal structure of the Mycobacterium tuberculosis (MtMscL) channel revealed a pentamer.^10–11

The view that all MscL channels are pentamers was recently challenged by the crystal structure of a C-terminal truncation of Staphylococcus aureus MscL (SaMscL(CΔ26)), which revealed a tetramer [Fig. 1(A)].¹² Reflecting their high sequence similarity, EcMscL, MtMscL, and SaMscL(CΔ26) all form functional tension-gated channels with similar open channel conductances.^12,13 Additionally, all three homologs and wild-type SaMscL are able to rescue cells in osmotic down shock assays.^12,14 Given their high sequence relatedness and functional similarity, the structural results raise intriguing questions as to the assembly and activation of MscL channels. As a starting point, we attempted to characterize the oligomeric state of different MscL constructs to address whether the crystal structures accurately reflected the oligomeric state of these proteins in solution. In the course of implementing a variety of methods, including protein crosslinking, multiangle light scattering, single molecule photobleaching, and crystallography, we noted inconsistent interpretations concerning the oligomeric state, prompting us to develop an alternative technique.

Figure. 1.

Related MscL channels demonstrate oligomeric diversity despite high sequence identity. (A) Crystal structures of pentameric MtMscL (PDB: 2OAR) and tetrameric SaMscL(CΔ26; PDB: 3HZQ). The structures have different dimensions and oligomeric states. The MtMscL pore is less expanded than SaMscl when viewed from the extracellular side of the membrane (top panels) and shorter when viewed from the side (bottom panels). Each MscL subunit is composed of two membrane spanning helices (TM1 and TM2). TM1 lines the channel pore and packs against TM2 from an adjacent subunit (TM2′). (B) Sequence alignment of MscL proteins used in this study. Dark and light blue residues correspond to identity and similarity, respectively. The N-helix, TM1, TM2, and C-helix of Mt, Sa, and EcMscL are 46, 32, 22, and 71% identical, respectively. The TM helices have a combined identity and similarity of 63%. Deletion constructs were truncated at the conserved C-terminus glutamate residue.

Membrane proteins pose special problems for determination of oligomeric states due to complications arising from the presence of lipid and detergent molecules. In addition, techniques such as crosslinking,¹⁵ single molecule subunit counting,¹⁶ or mass spectroscopy¹⁷ involve difficulties in interpretation or require specialized equipment; as such there is no gold standard for membrane protein oligomer characterization. To meet this need, we have developed an alternative technique for counting protein subunits based on the addition and removal of mass. Our method is designed to be fast, reproducible with modest amounts of protein, generalizable to membrane and soluble proteins, and experimentally accessible to most protein laboratories. We have applied this technique, which we call Oligomer Characterization by the Addition of Mass (OCAM), to investigate the oligomeric diversity of several MscL homologs. The OCAM-assigned oligomeric states are consistent with the commonly used technique of size exclusion chromatography coupled to multiangle light scattering (SEC-MALS) and existing crystal structures. More importantly, OCAM is able to identify the existence of oligomeric states not detected by SEC-MALS or crystallography. For two cases OCAM detected an unexpected mixture of oligomeric states.

Results

OCAM reaction scheme

We begin by developing the principles underlying the OCAM technique and validating the method using pentameric MtMscL as a test case. OCAM counts protein subunits by the addition and removal of mass from a protein oligomer through selective partial proteolysis. Proteolysis is quenched at several time points, and the cleavage products are separated via blue native polyacrylamide gel electrophoresis (BN-PAGE). By using the detergent mimic Coomassie blue dye to maintain membrane protein solubility and confer an overall negative charge, BN-PAGE has the important property of separating native membrane protein complexes on the basis of size.^18–20 The number of reaction products observed is dependent on the oligomeric state of the protein.

The reaction is described in terms of a protein of interest (target) fused with a protein partner (mass partner; Fig. 2). The mass partner is added via a short linker containing a TEV protease recognition site so that it is expressed as a fusion with the target protein. The initial mass of the oligomer will then be composed of the mass of the wild-type oligomer plus the added masses of the mass partners. For a homo-oligomer of n subunits, the total mass of the fusion protein oligomer is equal to the sum of the masses of n target subunits and n mass partners. Partial enzymatic cleavage by TEV protease removes mass partners generating a series of reaction products separated by the mass of a single mass partner. A total of n + 1 reaction products (corresponding to the loss of 0, 1, 2, …n mass partners) are expected. In the case of MtMscL, the crystallographically observed pentameric assembly is expected to generate 5 + 1 unique bands.

Figure. 2.

The OCAM reaction counts subunits of oligomeric proteins by selectively removing mass from protein complexes. (A) A 15 kDa subunit of a hypothetical pentameric membrane protein is fused to a 27 kDa sGFP mass tag via a cleavable TEV protease recognition linker to create an individual fusion subunit of 42 kDa. The assembled pentamer fusion protein is 210 kDa. Treatment with TEV protease selectively removes an sGFP mass partner from each subunit. (B) The expected OCAM reaction products after TEV protease treatment of (A). For a pentameric protein, six reaction products corresponding to the loss of 0–5 mass tags are produced. The reaction products have masses ranging from a fully tagged oligomer of 210 kDa to a fully cleaved oligomer of 75 kDa. By controlling the TEV protease exposure, time points containing all reaction products are produced, and their products are separated and counted via BN-PAGE.

In principle, any combination of target protein and mass partner may be used that satisfies the following requirements: (1) the mass partner must be sufficiently large to generate resolvable shifts on BN-PAGE gels, and (2) the mass partner must be monomeric so as not to induce or influence oligomerization. The distribution of reaction products as a function of TEV protease activity can be observed by comparing several time points.

MtMscL is a pentamer by OCAM

We added a mass partner to the C-terminus of MtMscL by fusing a monomeric, superfolder variant of GFP (sGFP) via a short linker containing a TEV protease recognition site (…SASGENFLYQ…) [Supporting Information Fig. S1(A)]. The sGFP variant contains an A206V mutation, which removes GFP's tendency to form weak dimers at high concentrations by disrupting the dimerization interface.²¹ The sGFP mass partner adds ∼27 kDa to each MscL subunit [Fig. 2(A)]. Overexpression, membrane solubilization in the detergent n-dodecyl-β-d-maltopyranoside (DDM), and purification by metal affinity chromatography yielded a single band on an SDS gel and a single size exclusion chromatography peak (not shown). The purified fusion protein was partially proteolyzed with TEV protease removing some but not all of the sGFPs fused to MtMscL. A complete proteolysis was also performed to ensure total removal of sGFP. The reactions were quenched and their products separated via BN-PAGE to preserve noncovalent interactions in the oligomer and separate the reaction products by size. The gel was then stained and examined visually or analyzed in detail by densitometry measurements.

Figure 3(B, C) shows the evolution of reaction products as a function of time. At the initial time point, >60% of the protein runs as a single band corresponding to the mass of the entire fusion protein complex. Below the major band are two smaller bands representing the loss of 1–2 sGFP mass tags. The relative proportion of these bands may be interpreted in terms of a binomial distribution consistent with <10% loss of sGFP, presumably due to proteolytic activity in the host cell (Supporting Information Fig. S1). Increasing TEV protease exposure results in a ladder of six bands consistent with the pentameric structure observed in the crystal structure [Fig. 3(B)]. The bands correspond to the full mass complex (top band, 0 sGFPs lost), the serial loss of 27 kDa resulting from the proteolytic cleavage of increasing numbers of sGFP (middle bands, 1–4 sGFPs lost), and the completely cleaved MtMscL target protein (bottom band, 5 sGFPs lost). A low molecular weight band corresponding to free sGFP is also seen in the later and final time points. Gel densitometry of the reaction time course reveals that all reaction bands are well separated and identifiable. Nearly all cleavage products are observed by 10 min, and cleavage is complete before 3 h [Fig. 3(B) and 3(C)]. As an independent check on the oligomeric state, we performed SEC-MALS measurements on the nonfusion form of MtMscL. The measured mass is in good agreement with OCAM measurements [Fig. 3(A) and Table I].

Figure. 3.

The OCAM measurement of MtMscL counts five protein subunits. (A) Elution profile of purified MtMscL examined by SEC-MALS. The red and blue lines correspond to SEC-MALS calculated protein mass and modifier (lipid and detergent) mass respectively. (B) BN-PAGE separation of OCAM reaction products. MtMscL sGFP fusion starting material was run alongside TEV protease reactions quenched at 10 and 180 min. A family of six reaction products corresponding to the loss of 0–5 sGFP mass tags is observed. The cleaved sGFP is observed at a lower gel migration distance. (C) Densitometry traces of (B). Cartoons marking each peak show the expected OCAM reaction product and its theoretical mass.

Table I.

Oligomeric States of MscL Proteins by Different Methods

			SEC-MALS
	X-ray crystallography	OCAM	Subunits	Protein mass (kDa) ± stdev	Modifier mass (kDa) ± stdev
MtMscL	5	5	4.7	87.4 ± 1.0	89.4 ± 2.0
MtMscL(CΔ49)		5
SaMscL		5	4.6	72.9 ± 0.5	92.1 ± 0.8
SaMscL(CΔ26)	4	5/4 mixture	3.6	46.5 ± 1.6	80.4 ± 1.9
EcMscL		6/5 mixture	6.0	103.0 ± 3.1	91.5 ± 7.1

Oligomeric states of MscL proteins by different methods. OCAM identifies mixtures of oligomeric species not observed by crystallography or SEC-MALS. SEC-MALS measurements are the result of three injections of each non-fusion protein. The oligomeric state is measured from the theoretical mass of a single subunit. EcMscL and SaMscL(C 26), proteins that have mixed oligomeric states by OCAM, have larger standard deviations than single oligomeric state proteins when measured by SEC-MALS. Calculated protein and modifier (detergent and lipid) masses are shown for the SEC-MALS measurements.

SaMscL is a pentamer by OCAM

Having developed and tested the OCAM procedure on MtMscL, we turned our attention to the SaMscL homolog. SaMscL shares 34% identity to MtMscL [Fig. 1(B)]. Like its Mt counterpart, SaMscL forms a tension activated nonselective channel. The structure of wild-type SaMscL has not been determined by crystallography; however, a C-terminal deletion (SaMscL(CΔ26)) has been solved in a tetrameric form, and previous crosslinking experiments suggested that wild-type SaMscL may also be tetrameric.¹² We used OCAM to determine the oligomeric state of wild-type SaMscL in solution. As with MtMscL, SaMscL was fused to sGFP via a linker containing a TEV protease recognition site. The fusion protein was overexpressed and extracted in DDM. Purification by metal affinity chromatography yielded a single band on an SDS gel and a single size exclusion chromatography peak (not shown). The purified protein was subject to partial and complete proteolysis by TEV protease. Figure 4(B,C) shows the OCAM reaction products separated by BN-PAGE and identified by densitometry. A ladder of six reaction products and free sGFP was observed, consistent with a pentameric assembly [Fig. 4(B)]. SEC-MALS measurement of the nonfusion protein is in good agreement with OCAM's pentameric assignment [Fig. 4(A) and Table I]

Figure. 4.

The OCAM measurement of SaMscL counts five protein subunits. (A) Elution profile of purified SaMscL protein examined by SEC-MALS. The red and blue lines correspond to SEC-MALS calculated protein mass and modifier (lipid and detergent) mass, respectively. (B) BN-PAGE separation of OCAM reaction products. SaMscL sGFP fusion starting material was run alongside TEV protease reactions quenched at 10 and 180 min. A family of six reaction products corresponding to the loss of 0–5 sGFP mass tags is observed. The cleaved sGFP is observed at a lower gel migration distance. (C) Densitometry traces of (B). Cartoons marking each peak show the expected OCAM reaction product and its theoretical mass.

A C-terminal deletion of SaMscL exists as a mixture of oligomeric species

We used OCAM to investigate the oligomeric state of a C-terminal truncation mutant of SaMscL (SaMscL(CΔ26)) in solution. The fusion protein was overexpressed, purified in DDM, and subjected to partial and complete proteolysis by TEV protease. The untreated sample runs as a single band on an SDS gel and as a single peak by size exclusion chromatography (not shown); however, when the purified protein was analyzed with BN-PAGE, two distinct major bands are present [Fig. 5(B)]. This behavior is also observed for the nonfusion SaMscL(CΔ26) [Supporting Information Fig. S1(A)]. When the sGFP fusion is subject to a partial TEV protease digestion and separated by BN-PAGE, we were not able to observe the n + 1 ladder bands expected for a single tetrameric or pentameric species despite numerous purification attempts. In contrast to wild-type Mt and SaMscL, we observed a complex mixture of greater than six reaction products [Fig. 5(B)]. Several intermediate peaks in the gel densitometry are present and offset from major peaks by small gel shifts [Fig. 5(C)]. The final reaction product results in two distinct bands of different apparent masses. Densitometry identifies 11 distinct bands across all time points, excluding free sGFP [Fig. 5(C)].

Figure. 5.

OCAM measurement of SaMscL(CΔ26) reveals the presence of mixture of two oligomers. (A) Elution profile of purified SaMscL(CΔ26) protein examined by SEC-MALS. The red and blue lines correspond to SEC-MALS calculated protein mass and modifier (lipid and detergent) mass respectively. (B) BN-PAGE separation of OCAM reaction products. TEV protease reactions were quenched at several time points. The number of reaction products is inconsistent with a single oligomeric species, and the final time point indicates the presence of high molecular weight (oligomer a) and low molecular weight (oligomer b) species. (C) Densitometry traces of (B) at several time points. A total of 11 unique MscL reaction products are observed representing a mixture of six pentameric (oligomer a) and five tetrameric (oligomer b) products. (D) Migration distances of OCAM reaction products from (B) plotted against the masses corresponding to a pentamer reaction family (open pentagons) and tetramer reaction family (close squares).

Given that the wild-type SaMscL was purified as a pentamer and the C-terminal deletion was crystallized as a tetramer, we asked if the observed OCAM band pattern was consistent with a mixture of pentamers and tetramers. A mixed population of pentameric and tetrameric channels should produce exactly 11 products (six products for the pentamer and five products for the tetramer). Assuming the BN-PAGE conditions separate the reaction products predominately by size, we would predict two ladders of digestion products with the ladder of pentamer products offset from the tetramer ladder by the mass of a single MscL subunit. Indeed, when we measure the peak position for each reaction product, we find that the product ladders are offset by an average of 26 pixels, which is consistent with the 13.8 kDA mass of a single SaMscL(CΔ26) subunit. Additionally, the predicted tetramer peaks migrate ∼50% of the distance between the predicted pentamer peaks consistent with 13 and 13.8 kDa difference between these products [Supporting Information Fig. S1(C)]. Finally, when the peak positions of the pentamer ladder are plotted against their predicted masses, the masses of the tetrameric ladder lie on the same curve [Fig. 5(D)]. These results suggest that the truncation mutant, at least under the reported expression and purification conditions, exists as both tetramers and pentamers in solution.

The sGFP mass tag does not alter the oligomeric states of SaMscL(CΔ26)

We investigated if the observed SaMscL(CΔ26) oligomeric mixture may have been caused by the presence of the sGFP mass tag. Free sGFP is monomeric and should not induce oligomerization on its own. However, it is possible that the MscL-sGFP fusion may alter the assembly of the fusion protein complex. To test this possibility, we ran BN-PAGE gels of purified SaMscL(CΔ26) lacking a sGFP mass tag. The sGFP fusion version of SaMscL(CΔ26) runs as two distinct final reaction products [Fig. 5(B)]. If sGFP was the cause of mixed oligomerization, we expect that the untagged protein would run as single band with a migration distance consistent with either a pentamer as suggested by the wild-type protein or the tetrameric form as suggested by the crystal structure of the deletion mutant. We find no evidence that sGFP alters the oligomeric forms of SaMscL(CΔ26). The untagged version of the protein, like fully cleaved SaMscL(CΔ26)-sGFP, migrates as two bands (Supporting Information Fig. S1(A)]. In addition, the intensity ratio of the lower to upper band is identical for both SaMscL(CΔ26) and SaMscL(CΔ26)-sGFP, indicating that the distribution between pentameric and tetrameric forms is ∼1:1 and equivalent for both proteins (Supporting Information Fig. S1). However, there are variations between preparations, which skew the ratio by as much as 10% in either direction.

A C-terminal deletion of MtMscL does not alter oligomerization

The OCAM results observed for SaMscL(CΔ26) immediately raised the question as to whether an identical truncation of MtMscL would cause it to behave in a similar manner. We created a truncation of MtMscL at position E102 (MtΔ49 MscL). This position is conserved between Mt, Sa, and EcMscL [Fig. 1(B)] and is the equivalent position of the SaMscL E95 truncation (SaMscL(CΔ26)). MtMscL(CΔ49) was fused to an sGFP mass partner, expressed, and purified in the detergent DDM. Like the wild-type MtMscL fusion, the purified MtMscL(CΔ49) fusion runs as a single band on an SDS gel and a single peak by size exclusion chromatography (not shown). Figure 6(A) shows the results of an OCAM digestion. On a BN-PAGE gel, incomplete proteolytic cleavage of sGFP results in a ladder of six bands followed by free sGFP. Analysis of the densitometry shows that the ladder bands occur at a regular interval [Fig. 6(B)], and their positions are well described by a fit to the predicted masses of a family of reaction products resulting from the digestion of an initial pentamer (not shown). In addition, the untagged versions of Mt and MtMscL(CΔ49) both migrate as single bands by BN-PAGE [Supporting Information Fig. S1(A)]. Taken together, these results indicate both wild-type and MtMscL(CΔ49) are pentamers and that the mixed oligomeric forms is not a general property of MscL C-terminal deletions.

Figure. 6.

The OCAM measurement of MtΔ49 MscL counts five protein subunits. (A) BN-PAGE separation of OCAM reaction products. MtΔ49 MscL sGFP fusion starting material was run alongside TEV protease reactions quenched at 10 and 180 min. A family of six reaction products corresponding to the loss of 0–5 sGFP mass tags is observed. A high-molecular weight band corresponding to a dimer of pentamers is also observed. The cleaved sGFP is observed at a lower gel migration distance. (B) Densitometry traces of (A). Cartoons marking each peak show the expected OCAM reaction product and its theoretical mass.

The wild-type EcMscL channel exists as a mixture of oligomers

We extended our study to include the E. coli MscL homolog (EcMscL). EcMscL is the most functionally characterized member of the MscL family. Although there is presently no crystal structure determined for EcMscL, electron microscopy and a study using tandem MscL subunits suggested that EcMscL may be hexameric,^8,9 although more recent studies of tandem MscL fusions were interpreted to be pentameric.^22,23 Crosslinking studies have suggested both hexameric and pentameric forms.^10,22–24

Purified EcMscL-sGFP runs as a single band on an SDS gel and a single peak by size exclusion chromatography (not shown). Figure 7(A,B) shows the results of a typical OCAM measurement. The family of reaction products is complex and qualitatively resembles the products observed for SaMscL(CΔ26). Like SaMscL(CΔ26), complete removal of the sGFP mass tag from EcMscL results in two bands of different apparent molecular weights, suggesting a mixture of oligomeric forms. However, unlike SaMscL(CΔ26), these bands migrate at distances inconsistent with a pentamer and tetramer, as the uppermost band has a migration distance above either full length pentameric Sa and MtMscL [Supporting Information Fig. S1(B)] suggesting the presence of a higher order oligomer. Densitometry reveals that the ladder of reaction products is formed by two different oligomeric species offset by the mass of a single MscL subunit.

Figure. 7.

Wild-type EcMscL is a mixture of two oligomeric species. (A) BN-PAGE separation of OCAM reaction products of EcMscL-sGFP. TEV protease reactions were quenched at the indicated time points. Similar to SaMscL(CΔ26), the number of reaction products is inconsistent with a single oligomeric species, and the final time point indicates the presence of highmolecular weight (oligomer a) and low molecular weight (oligomer b) species. (B) Densitometry traces of (A) at several time points. A total of nine unique MscL reaction products are observed representing a mixture of two oligomers. A mixture of hexamers and pentamers produces 13 reaction products depicted in the cartoon; however, their masses are sufficiently similar (<9 kDa difference, Supporting Information Fig. S1) to prevent resolution via BN-PAGE and intermediate products migrate as a single collapsed band. (C) Linear fit of the predicted EcMscL pentameric and tetrameric masses together with the masses of OCAM reaction products for Sa and MtMscL against their measured migration distance. Inclusion of the EcMscL pentamer and tetramer masses results in a relatively poor fit (R² = 0.63). (D) Same as (C) except with hexameric and pentameric EcMscL. The resulting linear fit (R² = 0.93) is more consistent with a hexamer and pentamer mixture.

Given the earlier reports of EcMscL hexamers, we asked if the reaction ladder can be explained by a mixture of hexamers and pentamers. A fully uncleaved hexamer would have a theoretical mass of 270 kDa while a fully uncleaved pentamer would have a mass of 225 kDa. Cleavage of a hexamer/pentamer mixture would create two ladders of reaction products whose masses are offset by only 9 kDa; however, this falls below the resolution limit of BN-PAGE, which prevents a full count of the expected 13 reaction products [Supporting Information Fig. S1(D)].

Figure 7(A) shows the OCAM digestion time course. The corresponding densitometry clearly shows the presence of two unique high molecular weight bands and three unique low molecular weight bands flanking four bands of intermediate weight. The two lowermost bands correspond to fully cleaved products consistent with the two bands observed for EcMscL expressed without the sGFP mass tag [Supporting Information Fig. S1(A)] suggesting that wild-type EcMscL exists in at least two oligomeric forms. The intermediate bands are likely a combination of unresolvable products from the two oligomeric forms [Fig. 7(B)].

We estimated the apparent molecular weight of the fully cleaved OCAM EcMscL reaction products by comparing their migration distances to the migration distances of the SaMscL and MtMscL OCAM reaction products using their assigned molecular weights as standards. The migration distances of OCAM reaction products can be empirically described by a log-linear relationship; however, over short migration distances the molecular weight of a reaction product can be estimated by a linear regression between products of known masses. By performing a linear fit of the migration distances of final SaMscL and MtMscL OCAM reaction products and comparing it with the migration distance of the final EcMscL OCAM products, we estimate an apparent molecular weight of 109–116 kDa and 92–95 kDa for the final products. These weights are in good agreement with the theoretical hexamer (110 kDa) and pentamer (91 kDa) masses [Supporting Information Fig. S1(D)]. The migration distance of the assigned hexamer and pentamer bands have a corresponding R² value of 0.93 when fit to the Sa and Mt MscL migration curves[Fig. 7(D)], whereas an assigned pentamer and tetramer have an R² value of 0.63 [Fig. 7(C)]. SEC-MALS measurements for the nonfusion EcMscL are consistent with a hexameric species (Table I).

Discussion

The oligomeric state of a membrane protein can be critical for biological function; however, it is not uncommon for the oligomeric state to be determined for one member of a protein family and then assigned to other members without direct verification. This situation reflects the experimental difficulties associated with unambiguous quantitation of the number of subunits in an oligomer, particularly if multiple species are present. We have developed a new technique Oligomer Characterization by the Addition of Mass or OCAM as a readily applicable method to study the oligomeric states of detergent-solubilized membrane proteins. As with all detergent based techniques OCAM does not measure oligomeric states within a native lipid bilayer; however, it does offer several experimental advantages. First, the method is experimentally accessible to most protein laboratories as OCAM simply requires common molecular biology reagents, a gel electrophoresis apparatus, and a protein subunit expressed with either an N- or C-terminal mass tag. Second, the results of an OCAM measurement are interpretable without extensive analysis as the oligomeric state is simply a count of proteolytic cleavage reaction products visualized by BN-PAGE. Third, although we have used purified protein for our OCAM study, in principle an unpurified sample can be used in combination with Western Blots.

OCAM compares favorably and offers distinct advantages over complementary techniques such as protein crosslinking, mass spectrometry, single subunit counting, and SEC-MALS. In the case of protein crosslinking, a method that counts up from monomer to a completely crosslinked oligomer by denaturing SDS gels, the crosslinking reaction can often be difficult to control. Crosslinking studies are typically interpreted qualitatively in terms of a ladder of gel separated reaction products where conditions are empirically adjusted to maximize the extent of intramolecular crosslinking, while minimizing intermolecular reactions. The assumption is generally made that distinct bands correspond to different numbers of crosslinked subunits, although oligomers of the same stoichiometry can crosslink in multiple configurations (such as circularly closed as opposed to open forms), resulting in distinct mobilities that complicate analysis of the gel ladder.¹⁵ As one example, our previous crosslinking study of full length SaMscL revealed a ladder of four bands indicating a tetrameric species,¹² yet the current studies clearly establish it as a pentamer. In comparison, the OCAM reaction is easier to interpret as the technique counts down from full oligomer to completely cleaved reaction products by native gels. Excessive treatment with TEV protease simply leads to an accumulation of fully cleaved product; as such there is no possibility of over treating the sample.

Mass spectrometry, single molecule subunit counting, and SEC-MALS are techniques not routinely accessible to many laboratories due to their specialized equipment requirements. Additionally, in the case of single molecule subunit counting and SEC-MALS, it is unclear if these techniques can unambiguously measure oligomeric heterogeneity, if the different species cannot first be separated. Single molecule counting requires a fluorescent marker, which is photo-destroyed. Like OCAM, it usually uses a fusion of GFP or another variant. Unfortunately, because not all of the expected GFPs are fluorescent the technique often requires a calibration to correct for nonfluorescent GFP proteins.¹⁶ In such cases, a mixture of oligomeric states may be mistaken as an unusually high number of nonfluorescent GFPs.

Similarly, in the case of SEC-MALS, in the absence of other evidence, mixed oligomers may be misidentified as an unexpectedly broad peak from a single oligomer. Our experience with this technique is instructive as neither a Superdex sizing column nor a Shodex HPLC sizing column was able to separate mixtures of SaMscL(CΔ26) pentamer and tetramer, although SEC-MALS measurements of SaMscL(CΔ26) and EcMscL have higher standard deviations consistent with a poorly separated mixture of oligomeric states. The individual oligomer protein–detergent complexes would likely have very similar hydrodynamic radii, which impedes separation by this technique. An anion exchange column was able to enrich for the SaMscL(CΔ26) tetramer species, but even then complete separation was never achieved. Additionally, the accuracy of SEC-MALS is dependent on the specific refractive index increment for the protein detergent complex (dn/dc) and the protein extinction coefficient, which may be difficult to experimentally determine. These factors may have contributed to the subunit undercount for most of our samples.

When performing an OCAM measurement, care must be taken in the choice of mass tags and detergents. In general, mass tags must be at least 13 kDa to ensure detectable separation by BN PAGE. This is the minimum mass difference we were able to resolve on gradient gels. In addition, detergents used for purification must be compatible with protein stability, BN PAGE migration, and TEV protease activity²⁵ (Supporting Information Table S1). These concerns are protein specific and must be empirically determined. In the case of SaMscL(CΔ26), the protein was measured by OCAM in several detergents (Supplementary Table S1) as a mixture of tetramers and pentamers (not shown). However, for EcMscL, the choice of detergent affected the measured oligomeric state indicating that detergent extraction or detergent interaction with Coomassie favors one oligomeric form (Supporting Information Fig. S1). It is also assumed that Coomassie interaction with protein does not perturb the oligomeric state.²⁶

In addition to MscL, other examples of membrane protein homologs with differing oligomeric states have been described, including various ATPases, such as the k ring from E. hirae V_o ATPase (decamer), the c ring from I. tartaricus F_o ATPase (undecamer), and the c ring from spinach chloroplast CF_o ATPase (tetradecamer).²⁷ The light harvesting complex 2 (LH2) has been crystallized as different oligomers from R. molischianum (octamer) and R. acidophilum (nonamer).²⁸ Full-length GLIC from G. violaceus crystallizes in a pentameric form^29,30 while its free extracellular domain crystallizes as a hexamer.³¹ These observations indicate that variability in the assembly of homologous oligomeric membrane proteins does occur, although the biological significance of this variation is not always clear.

What is determining the oligomeric state of these proteins? One possibility is that the oligomeric forms observed are a byproduct of protein overexpression in E. coli and/or detergent extraction and purification. Overexpression may overwhelm normal checks on protein assembly leading to the mixed oligomers we observe for SaMscL(CΔ26) and EcMscL. It would be instructive to test this hypothesis using cells producing normal levels of MscL protein. However, as many studies of MscL and other membrane proteins use purified protein from similar overexpression systems, our findings indicate that checking the oligomeric state of an uncharacterized protein, whether by OCAM or another technique, should be a routine method added to the protein toolbox.

Another possibility is that the ability to adopt multiple oligomeric states is an intrinsic property of the protein sequence, particularly for a protein such as MscL that undergoes substantial conformational rearrangements. For a symmetrical protein packed around a pore, the inner pore lining helices can accommodate either multiple conformations or multiple oligomeric forms through changes in the helix packing angle and helix-membrane tilt angle.³² For example, the packing angle between TM1 pore lining helices of tetrameric SaMscL(CΔ26) is −64°, which is close to the angle predicted by the pentameric Sukharev and Guy open state model (−67°).³³ The conformational change from closed to open is predicted to involve a reduction in the pore helix packing angle, which requires a concomitant decrease in the helix-membrane tilt angle. Conversely, proteins with different oligomeric states can have the same packing angle if the helix-membrane tilt angle varies. As an example, the helix packing angle for the pore lining helices of the tetrameric postassium channel KcsA (−46°) is close to pentameric MtMscL (−41°) despite their different oligomeric states. Consequently, when there is flexibility in helix-helix crossing and helix-membrane tilt angles, it should be anticipated that multiple oligomeric forms could be generated. Related considerations apply to the cytoplasmic bundle of C-terminal coiled-coil helices (present in the full length proteins but deleted in the truncation mutants), since coiled-coil sequences can exhibit oligomeric states ranging from dimers to heptamers that are sensitive to apparently modest sequence variations.³⁴ For different MscLs, the observed oligomeric state behavior may then reflect the competition between the conformational preferences governing the packing of TM1 helices and of the C-terminal helices, as influenced by modification of the protein sequence (mutation and truncation) or changes in environmental conditions (membrane properties, detergent, tension, etc.).

The presence of mixed oligomeric forms raises the possibility that the functional properties of MscL channels may not be solely due to a single oligomeric form. Indeed, some soluble proteins switch between oligomeric forms with different activities.^35,36 It is possible that the single channel conductances, open channel dwell times, and pressure sensitivity normally attributed to pentameric channels may be due to a mixture of oligomers or to nonpentameric channels. SaMscL(CΔ26) and EcMscL, proteins with mixed oligomeric states, functionally reconstitute into giant vesicles. In our hands, mixtures of tetramers and pentamers (SaMscl(CΔ26)) and predicted pentamers and hexamers (EcMscL) give single channel conductances around 2.6–2.9 nS, respectively. This difference may be due to variations in protein sequence and/or oligomeric states; however, the question remains open until the two forms can be separated from one another and characterized individually.

Our results suggest that oligomeric state variability may be a general feature of the MscL family, where some members (MtMscL) may only have a single form while other members (EcMscL and SaMscL(CΔ26)) have multiple forms. The underlying structural determinants of this oligomeric diversity are currently being investigated.

Materials and Methods

Molecular biology

MscL-TEV linker-sGFP fusions were created using standard PIPE cloning and in-fusion reactions. pET expression vectors carrying either Mt, Sa, or EcMscL were linearized by PCR to create a single PCR product opened between the last coding base pair at the 3′ end of MscL and the adjacent stop codon. The 5′ end of the PCR primers contained 18 base pairs of homology to an sGFP insert. All PCR products were treated with DpnI and gel purified. PIPE and in-fusion reactions were performed as described.^37,38

Protein expression

Expression plasmids were transformed into BL21-Gold (DE3) cells (Stratagene) or Rosetta (DE3) cells (EMD). A single colony was used to inoculate 20 mL of lysogeny broth-Miller (LB)³⁹ or Terrific broth media with 1% glucose (TB glucose), and the culture was grown overnight at 37°C with shaking at 225 rpm. One liter of TB glucose media was inoculated with 20 mL of the overnight culture and shaken at 37°C. TB glucose cultures were grown until the absorbance at 600 nm reached 2.0. Expression was induced with 1 mM IPTG at 37°C for 2 h. Cells were pelleted at 6,000g and stored at −20°C.

Protein purification

Five grams of cell pellet were resuspended in 50 mL lysis buffer (50 mM Tris-HCl, 200 mM NaCl, 1% DDM, pH 8.0), passed four times through a microfluidizer, and centrifuged at 17,000 rpm (JA-17, Beckman) for 40 min at 4°C. The supernatant was gravity loaded onto a 2 mL Ni-NTA (Qiagen), washed with 1 column volume (CV) equilibration buffer (20 mM Tris-HCl, 150 mM NaCl, 10 mM imidazole, 0.05% DDM, pH 7.5) followed by 5 CV high salt buffer (20 mM Tris-HCl, 500 mM NaCl, 25 mM imidazole, 0.05% DDM, pH 7.5), 5 CV low imidazole wash buffer (20 mM Tris-HCl, 150 mM NaCl, 75 mM imidazole, 0.05% DDM, pH 7.5), and eluted in 3 CV elution buffer (20 mM Tris-HCl, 150 mM NaCl, 300 mM imidazole, 0.05% DDM, pH 7.5). The eluate was concentrated to ∼10 mg/mL in 100 kDa cutoff concentrators (Amicon Ultra-4, Millipore) and loaded onto a Superdex 200 10/30 HR column (GE Healthcare) equilibrated in 20 mM Tris-HCl, 150 mM NaCl, 0.02% DDM, pH 7.5. The major peak fraction was collected and used directly for OCAM proteolytic digestion or SEC-MALS.

OCAM proteolytic digestion

Samples for OCAM experiments were suspended in reaction buffer (20 mM Tris pH7.5, 50 mM NaCl, 0.5 mM EDTA, 5 mM DTT) at 0.3 mg/mL. AcTEV protease (Invitrogen) was added at 1 U/μg substrate, and reactions were incubated at 34°C. Reactions were quenched at multiple time points by adding 10x stop solution (0.1M Iodoacetamide, 0.1% Ponceau-S, and 50% Glycerol), and ∼6 μg of substrate was loaded onto BN-PAGE gels for analysis. TEV protease is crucial to OCAM as other proteases such as thrombin, enterokinase, and Factor Xa may yield nonspecific cleavage products.⁴⁰

Blue native PAGE

BN-PAGE was performed as described with modifications.^19,20 The anode buffer was 50 mM Bis-Tris pH 7.0. Two cathode buffers were used. Buffer A was 50 mM Tricine and 15 mM Bis-Tris pH 7.0, with 0.02% Coomassie brilliant blue G-250. Buffer B was the same as A but without Coomassie. Bio-Rad precast 4–15% Tris-HCl gradient gels were run with Buffer A at 150 V for 1 h followed by 4 h at 150 V using buffer B. In some cases, Invitrogen Native PAGE 4–16% gradient gels were used. The anode buffer was constant throughout the run. All buffers were made at 4°C and gels were run at 4°C. Gels were stained in 40% methanol/10% acetic acid with 0.2% Coomassie Brilliant Blue R and destained in 40% methanol/10% acetic acid followed by rehydration in water before imaging and densitometry analysis (Alpha Innotech). Traces were smoothed with a rolling ball algorithm implemented in AlphaEaseFC 6.0.0 for display purposes.

Size exclusion chromatography and multiangle light scattering

Size exclusion chromatography coupled to multiangle light scattering (SEC-MALS) was carried out using a Varian HPLC with three in-line detectors. Detectors were a Varian Prostar 345 UV detector followed by a DAWN HELEOS and an Optilab rEX (Wyatt Technologies). Samples were run over a Shodex KW803 column at 0.5 mL/min in 20 mM Tris pH 7.5, 150 mM NaCl, 0.02% DDM. Data was processed using the protein conjugate analysis implemented in Astra 5.3.4.16. UV extinction coefficients were calculated using Expasy Protparam, and detergent dn/dc values were obtained from Anatrace. The protein dn/dc value of 0.185 mL/g was used for all analysis. Protein molecular weight values are the average of three injections. Peak data was calculated at one half of the peak height. A single tryptophan was inserted into the N-terminal His tag of MtMscL and EcMscL to increase their UV extinction coefficients.

Note added in proof

A paper by Dorwart et al has appeared characterizing the variable stoichiometry of SaMscL; Dorwart MR et al, PLoS Biol 8(12): e1000555. doi:10.1371/journal.pbio.1000555