Mechanosensitive Channels in Bacteria: Functional Domains and Mechanisms of Gating

Today, we are only approaching a minimal level of understanding of how cells may sense and respond to mechanical stimuli. There is a drastic mismatch between the 'functional' phenomenology collected at the organismal and cellular levels and the molecular description of mechanosensory mechanisms. MscL, a mechanosensitive channel of large conductance, was the first isolated molecule shown to respond to membrane stretch by opening a large aqueous pore. It resides in the inner membrane of Escherichia coli and serves as protective safety valve in the event of osmotic shock. The channel is a multimeric complex made of 15 kDa subunits and responds directly to the tension in the lipid bilayer. During the previous funding period, we have been able to determine the overall structure of MscL, its topology, secondary structure and multimerism, and to evaluate energetic and spatial parameters that characterize conformational transitions in native MscL. The analysis of mutations now allows us to narrow the search for functionally important regions to the short N-terminal helix, two transmembrane domains of the protein, and the loop between them. The proposed study will be focused on identification of individual or groups of residues that determine the energy of closed-to-open transitions and the dwell time in each state. The methodology will include site-directed mutagenesis, covalent protein modifications, patch- clamp measurements, thermodynamic and kinetic analysis of dose-response curves and rates of transitions, videomicroscopy, and molecular modeling. The study of MscL, a highly convenient model system, will give us basic understanding and the first example of what type of intramolecular interactions, sequence motifs, and conformations make membrane proteins sensitive to a physiologically relevant mechanical stimulation. The recently solved crystal structure of MscL from Mycobacterium tuberculosis (Chang et al., 1998, Science 282:2220-2226) provided a strong framework for the evaluation of conformations that permit opening of a large MscL pore by membrane tension. Previous kinetic and thermodynamic analyses (Sukharev et al., 1999, J. Gen. Physiol. 113:525-539) suggested that expansion of the channel protein precedes the opening. Following these data, in collaboration with Dr. H. R. Guy (NIH), we built molecular models for the E. coli MscL in the closed, closed-expanded and open conformations and found that the channel can not be gated solely with gate that is placed within the transmembrane domain as was previously proposed (Spencer et al., 1999, Curr.Opin. Struct. Biol. 9:448; Batiza et al., Structure Fold. Des. 7:R99). Instead, we propose that the gating is accomplished by unresolved N-terminal domains (S1) connected to the transmembrane barrel via flexible linkers. We also propose that stretch in the membrane increases the tilt of the M1 and M2 helices, making the barrel wider. The pore remains occluded from the cytoplasmic side by the bundle of S1 helices until stress in the linkers connecting S1 and M1 pulls the bundle apart, thus opening the channel. We tested the following modeling predictions: (1) the short N-terminal domains (S1) are amphipathic and form a bundle that occludes the pore from the cytoplasmic side; (2) when membrane tension stretches the transmembrane (TM) region, the tilt of the TM helices increases and M1s move away from the axis of the pore in an iris- like manner; (3) when the S1 bundle breaks apart leading to the fully open conformation, S1 helices may dock to a specific site on the inner surface of the pore. The pairs of residues on S1 and S2 predicted to be proximal in either closed or open conformations were mutated to Cys, and the effects of coupling were tested with patch-clamp and biochemically using Western blots. Spontaneous bridging of cysteines at positions 7 or 10 linked pairs of subunits and prevented the channel from opening until the bonds were reduced. The docked position of S1 in the open state was supported by an I3C to I96C bridge, which formed under oxidizing conditions and locked the channel in one of the subconducting states, preventing complete closures. The data support the hypothesis that in the closed state S1 domains form the bundle, which must separate when the channel opens, consistent with the proposed gate function.