Essentially all previous structural studies on MscL, as with most membrane proteins, have been performed on channels that have been solubilized, under the assumption that detergents do not alter the subunit stoichiometry. Our data examining the effect of different detergents on the oligomeric state of SaMscL have far-reaching implications for membrane protein research, as we now clearly demonstrate that changes in oligomeric state can and do occur upon detergent solubilization, and these alterations are not easily detected by SEC (see and ). To circumvent these artifacts associated with solubilization, we have developed an in vivo cross-linking assay. Although the pentamer is by far the major product observed in , very small amounts of smaller oligomers can be resolved upon overloading of protein gels and long Western blot exposures. The small amount of tetramers and smaller complexes observed upon long exposure could be incompletely assembled protein or may simply be due to the lack of disulfide bridging, which is rarely observed to be this efficient (e.g. see 
). In addition, we cannot completely rule out the possibility that oligomerization is a dynamic and even reversible process in vivo. Regardless, it seems unlikely that oligomeric species smaller than pentamers have a physiological role given that there are tens of channels per cell 
, and the smaller species compose only a few percent of the total channels. Our data definitively show that the vast majority of the SaMscL channel adopts a pentameric structure in the cell membrane, consistent with the MtMscL crystal structure 
, but incompatible with the SaMscL crystal structure 
A close inspection of the recent tetrameric SaMscL structure, which was speculated to be in an expanded gating intermediate state, reveals structural interactions between residues in TM1 and TM2 residues of a neighboring subunit similar to those in the closed MtMscL structure. This result was surprising because several studies using various techniques have suggested TM1 undergoes a significant clockwise rotation (as viewed from the periplasmic side) upon channel gating. The approaches supporting this interpretation include: spin labeling combined with electron paramagnetic resonance (EPR) studies 
, accessibility of sulfhydryl reagents to cysteine mutants in TM1 upon gating 
, accessibility of heavy metals to engineered binding sites in the pore 
, suppression mutagenesis demonstrating interactions between TM1 and TM2 upon gating 
, and confirmation of multiple interaction sites between TM1 and TM2 upon gating by using an electrostatic repulsion approach as well as disulfide trapping 
. Both the EPR studies as well as the heavy-metal binding in the pore strongly suggest that this rotation occurs quite early in the gating process, before ion permeation. By this model, V21 and G24 (V23 and G26 in EcoMscL) should rotate away from the pore constriction, allowing V22 and the positively charged K29 (I24 and K31 in EcoMscL) to line the open pore. The observation that this rearrangement has not occurred strongly suggests that the crystal SaMscL structure actually reflects a strained closed state due to improper oligomeriazation, rather than an expanded intermediate state.
The observed stoichiometry changes in detergents also evoke questions regarding how membrane protein subunits assemble into physiological multimeric complexes. A previous report on EcoMscL demonstrated that when the protein is translated in vitro in the absence of any lipid or detergent and is subsequently purified in the presence of Triton X-100, the channel spontaneously assembles into pentamers 
. A separate study demonstrated that even synthetically synthesized EcoMscL will oligomerize into functional channels when incorporated into lipid membranes 
. These studies demonstrate that the EcoMscL channel has the ability to self-assemble into the functional oligomer in vitro. Similarly, it appears that C8
and Triton X-100 allow SaMscL to properly fold and assemble into a pentameric structure, which correlates with the observed oligomeric state in vivo. In contrast, LDAO reorganizes SaMscL into tetramers that do not appear to exist in significant quantities in vivo, suggesting that they are not physiologically relevant. The observed self-assembly properties of EcoMscL under many conditions, combined with the reversible nature of the detergent-induced oligomeric state changes for SaMscL, may explain why Liu et al. observed channel function for the LDAO solubilized SaMscL tetramers 
; presumably, once the tetrameric LDAO-solubilized SaMscL is reconstituted into lipids, it rearranges back into functional pentameric structures.
The crystal structure of the SaMscL was derived from a truncation mutant in which several amino acids at the C-terminus had been removed 
. One previous study suggested that at least some C-terminus deletions of EcoMscL could lead to the formation of larger aggregates when translated in a cell-free system 
. While we cannot rule out the possibility that the SaMscL deletion contributed to the stabilization of the tetramer in the crystal, our data do demonstrate that detergents alone can stabilize alternative oligomeric states of the full-length SaMscL in solution.
Our observations make it tempting to speculate that larger, more hydrated head groups (i.e. the head groups of C8E5 and Triton X-100) promote pentamer formation, while smaller head groups (i.e. LDAO) promote tetramer formation. However, with such a small number of detergents tested thus far, additional experiments will be required to identify what chemical properties of the detergents cause the oligomeric state change and what effect, if any, lipids may have on the stabilization of specific oligomeric states.
In sum, our data demonstrate that the physiologically active SaMscL oligomer in vivo is a pentamer and that purification in detergents can cause a reorganization of the SaMscL channel's oligomeric state. Our findings also show that SEC alone is insufficient to evaluate the usefulness of a particular detergent for the study of membrane proteins; ideally, SEC-MALS and/or analytical ultracentrifugation should be correlated with in vivo experiments that can more confidently determine the functional and physiologically relevant oligomeric state of membrane proteins.