An open-pore structure of the mechanosensitive channel MscL derived by determining transmembrane domain interactions upon gating

Abstract

Mechanosensation, the ability to detect mechanical forces, underlies the senses of hearing, balance, touch, and pain, as well as renal and cardiovascular regulation. Although the sensors are thought to be channels, relatively little is known about eukaryotic mechanosensitive channels or their molecular mechanisms. Thus, because of its tractable nature, a bacterial mechanosensitive channel that serves as an in vivo osmotic "emergency release valve," MscL, has become a paradigm of how a mechanosensitive channel can sense and respond to membrane tension. Here, we have determined the structural rearrangements and interactions between transmembrane domains of MscL that occur upon gating. We utilize an electrostatic repulsion test: If two residues approach upon gating we predicted that substituting like-charges at those sites would inhibit gating. The in vivo growth and viability and in vitro vesicular flux and electrophysiological data all support the hypothesis that residues G26 and I92 directly interact upon gating. The resulting model predicted other interacting residues. One of these sets, V23 and I96, was confirmed to truly interact upon gating by disulfide trapping as well as the electrostatic repulsion test. Together, the data strongly suggest a model for structural transitions and residue-residue proximities that occur upon MscL gating.

Figure 1.

Schematic representations of the MscL channel from the M. tuberculosis crystal structure. Left: side view of an individual MscL subunit with the positions of the two transmembrane helices, TM1 and TM2, labeled. Middle: side view of the MscL pentamer. Right: top view from the periplasmic side.

Figure 2.

A positive charge at residue 92 suppresses the severe in vivo GOF phenotype effected by a positive charge at residue 26. A) Growth of cells expressing WT, G26H, G26H/I92R, G26H/I92K, and G26H/I92H at pH 5.5 (black bars) and pH 8.5 (gray bars) was monitored by OD₆₀₀. Right panel: fold increase after induction. Left panel: typical growth curve of G26H/I92R at pH 5.5 (solid diamonds) and pH 8.5 (open squares); arrow indicates addition of IPTG. B) Right panel: percentage viability (treated with MTSET vs. no treatment) of cells expressing WT, G26C, and G26C/I92H MscL grown at pH 5.5 (black bars) and pH 7 (gray bars). Left panel: raw data for G26C/I92H. After a culture was grown at pH 5.5 (top) or 7.0 (bottom), it was treated (+) or not treated (−) with MTSET, then 5-μl drops from 10-fold dilutions (10⁻⁴ to 10⁻⁶) were plated and grown overnight. C) Right panel: percentage viability after treatment with MTSET (black bars) or MTSES (gray bars) of cells expressing WT, G26C G26C/I92R, and G26C/I92K MscL. Left panel: raw data for G26C and G26C/I92R, as above.

Figure 3.

A positive charge at residue 92 suppresses the severe in vitro GOF phenotype effected by a positive charge at residue 26. A) Percentage calcein release from vesicles reconstituted with G26C (black line), G26C/I92R (dashed line), WT MscL (dotted line), and no protein (gray line). Arrow indicates addition of MTSET, followed by Triton X-100 detergent to measure the 100% calcein content. B) Effect of MTSET on single-channel recording of MscL activity of G26C (top) and G26C/I92R (bottom). Each recording shows the current (top trace) and pressure (bottom trace). MscS (∗) and MscL (▿) activities are labeled. After MTSET was added into the bath (arrow), the spontaneous locked-open MscL channel activities (▾) were consistently observed for G26C but not G26C/I92R.

Figure 4.

Evidence for the interaction between V23 and I96 residues of MscL. A) Schematic illustration of the relative position of 26–92 and 23–96 pairs on adjacent TM1 and TM2 helices. B) Western blot showing disulfide bridge trapping of V23C/I96C. Cells were grown in high osmolarity (0.5 M NaCl); pentamer (5×) formation is seen on in vivo osmotic downshock in the presence of 150 μM copper phenanthroline [Cu(II)]. The multimer was reduced to control mock-shocked levels (M) by treatment with 2% βME. C) Right panel: percentage viability of cells expressing WT, V23C, V23C/I96R, and V23C/I96K. Left panel: raw data for V23C and V23C/I96R when downshocked in the presence (+) or absence (−) of MTSET.

Figure 5.

A positive charge at residue 96 suppresses the severe in vitro GOF phenotype effected by a positive charge at residue 23. A) Percentage calcein release from vesicles reconstituted with V23C (black line), V23C/I96R (dashed line), WT MscL (dotted line), and no protein (gray line). B) Effect of MTSET on single-channel recording of MscL activity of V23C (top) and V23C/I96R (bottom). Each recording shows current (top trace) and pressure (bottom trace). After MTSET reacted, spontaneous MscL channel activities were observed for both V23C and V23C/I96R but disappeared only for the V23C/I96R MscL within 30 min, where even a subsequent stimulus of ∼140 mmHg failed to effect any channel activity (normal pressure to gate this mutant is 77±15 mmHg; n=6).

Figure 6.

Model of the interaction of TM1 and TM2 of MscL upon gating. Left: TM1, TM2, and locations of 23, 26, 92, and 96 in the closed state of MscL. Middle: interaction of the 26/92 residue pair and 23/96 pair while TMs tilt and rotate (indicated by arrow) during channel opening. Right: enlarged view of the interactions.