A release of local subunit conformational heterogeneity underlies gating in a muscle nicotinic acetylcholine receptor

Synaptic receptors respond to neurotransmitters by opening an ion channel across the post-synaptic membrane to elicit a cellular response. Here we use recent Torpedo acetylcholine receptor structures and functional measurements to delineate a key feature underlying allosteric communication between the agonist-binding extracellular and channel-gating transmembrane domains. Extensive mutagenesis at this inter-domain interface re-affirms a critical energetically coupled role for the principal α subunit β1-β2 and M2-M3 loops, with agonist binding re-positioning a key β1-β2 glutamate/valine to facilitate the outward motions of a conserved M2-M3 proline to open the channel gate. Notably, the analogous structures in non-α subunits adopt a locally active-like conformation in the apo state even though each L9’ hydrophobic gate residue in each pore-lining M2 α-helix is closed. Agonist binding releases local conformational heterogeneity transitioning all five subunits into a conformationally symmetric open state. A release of conformational heterogeneity provides a framework for understanding allosteric communication in pentameric ligand-gated ion channels.


Tables of contents
Table S1.S4.Non-specific bumping of residues at the αγ/αδ M2-αM3 -γ/δ β8-β9 loops on nAChR.5 Table S5.Subunit specific effects of mutating residues contributing to conserved tripartite salt bridges at the ECD -TMD interfaces..  S1.Functional effects of mutations to the ECD triad and residues in the M2-M3 loop of the Torpedo versus human adult muscle nAChR.

Figure S1. Relative sensitivities of EC50 versus  to changes in channel function.
Measured changes in EC50 values relative to wild type for mutations in the muscle nAChR are typically of lesser magnitude than the changes observed in the di-liganded gating equilibrium constant, , for the same mutation.Measurements of EC 50 are thus less sensitive to changes in channel gating.Such differences are expected, however, because EC50 values, which reflect both agonist binding (Kd) and channel gating (), are mathematically weighted towards the Kd, as illustrated by equation 3 which was determined elsewhere for the human adult muscle nAChR 3 .A) Assuming a constant Kd,R, we illustrate the non-linearity of the relationship between EC50 and  by plotting the expected change in  as a function of a change in EC50 value.The black dot represents WT values for the Torpedo nAChR.The nonlinearity of the relationship is further illustrated in B), which compares how the fold change in  relative to WT (red curve) manifests as a fold change in EC50 relative to WT (blue curve) as  deviates from WT (black dot).
The two curves illustrate that the greater the functional effect of the mutant on channel gating (i.e., the greater the change in ), the larger the discrepancy between the measured fold change in EC50 versus .
For example, using these approximations, 2-, 5-and 10-fold changes in EC50 in the Torpedo nAChR should correspond to roughly 3, 15 and 70-fold changes in , respectively.
C) The table compares the experimentally measured EC50 values for select mutants in both the Torpedo and the adult human muscle nAChR, along with the fold change in each EC50 relative to WT (this study), to the changes in the di-liganded gating equilibrium constant, , calculated for each mutation in the adult human muscle nAChR from kinetic fitting of single channel measurements 1 .Considering the expected differences in the sensitivities of EC50 versus  values to changes in channel function, the functional data obtained using both whole cell and single channel electrophysiological recordings present a consistent picture regarding the importance of residues at the intra- subunit ECD-TMD domain interface.Despite structures showing that 1-2 moves roughly orthogonally to M2-M3 upon agonist binding (Fig. 1b and Movie 1, panel a), the mutagenesis data reaffirms that the ECD triad plays an important role in channel gating.In contrast, unless the diffuse density observed in the pore is modeled with bound lipids (see 4 ), the open pore of the nicotine bound structure closes within ~40 ns to a conformation with a diameter slightly larger than the apo state (top right, orange traces).On the bottom right, the pore diameter of the quadruple Asn mutant (orange) after 100 ns simulation is compared to the apo WT (green) after 100 ns simulation.The average distance between αV13' residues from the three repeats show that the quadruple Asn mutation in M2-M3 accelerates the collapse of the pore all the way back to the apo WT state.To assess whether the local conformational asymmetry observed at the interface between the 1-2 and M2-M3 loops in different subunits extends to the entire ECD of each subunit, as originally proposed by Unwin et al. 5 , we used several parameters including those developed by Lev et al. 6 and Calimet et al. 7 .
To assess the capping of loop C, we first analyzed the apparent counterclockwise rotation of each subunit's ECD that occurs during activation.To capture these changes, we aligned the TMDs of apo and bound states and calculated the ECD twist angle using the metric documented in ref 6.Although GLIC undergoes an ~2° decrease in twist angle between the closed and open states, there is essentially no change in twist angle for the  subunits of the nAChR (Figure S5, panel B).The non- subunits also have little change in twist angle but appear to increase rather than decrease during gating.It appears that there are differences in the twisting motions between homomeric and heteromeric pLGICs.
We quantified the ECD motions by measuring both the solvent accessible surface area (SASA) at the different ECD-ECD interfaces and the binding site contraction 7 .The SASA of the subunit-subunit interfaces in the ECD decreased in all subunits (Fig. S5, panel C), but the largest decreases occurred in both  subunits primarily due to a larger binding site contraction (Fig. S5, panel D).Neither metric seem to provide any evidence for pre-activation elsewhere in the ECD of the non- subunits.The loop C capping of the non- subunits, which should be captured in the binding site contraction metric, did not provide evidence for pre-existence in either apo or bound states.This is likely because the non- subunit C loops adopt unique conformations that are morphologically dissimilar to the  subunits (Fig. S5, panel A) and are therefor difficult to directly compare.
We next used three previously defined metrics of activation, upper and lower ECD spreading and expansion, to assess whether the non- subunits adopt a pre-active conformation.ECD spreading was calculated as the distance between the center of mass (COM) of the entire ECD at both the extracellular (upper) and TMD adjacent (lower) ends, to the COM of each subunit at the upper and lower ends.To find the COMs at the upper and lower ends of the entire ECD and the individual subunit ECD, an ellipsoid around the targets were created using the "measure inertia" command on ChimeraX using just the C's of conserved -strands in each subunit.The COM of the ellipsoid, along with the vector of the longest principal axis (corresponding to the height axis of the domain), were used to find COMs at two z positions at the top and bottom of each ECD (see panel A of Fig. S6).The distance between the full ECD COM and the subunit COM in both apo and bound states is defined as the ECD spread (ref.6) and is plotted on the right section of panel A of Fig. S6.We found that there is a small inward motion of the upper ECD from each subunit, with no obvious change between a and non-a subunits.At the lower ECD, there does seem to be a subtle difference in spreading in a versus non-a subunits with a larger spreading occurring in the non-a subunits.This is a product of the scissor-like motion that occurs during activation in the  subunits that is described in Zarkadas et al. 4 .The magnitudes of the ECD spreads do not provide evidence for a preactive ECD conformation of the non- subunits.
We also used -expansion, a parameter used for both GLIC 6 and the homomeric ρ1 GABAAR 8 to measure activation.We used the same protocol as previously described taking the COM of the C's from two five residues stretches (one in the 1-2 loop and one in the 10-M1 linker) and measuring the distance in both apo and agonist bound state for each subunit (see panel B of the Fig. S6).The expansion was originally described from the open to closed state of GLIC and thus is a contraction in the activation transition.We also see a subtle contraction in the  subunits, but the non- subunits instead expand.It is notable however that for both homomeric receptors, the magnitude of the contraction are ~2Å , while here the magnitudes of the  subunit contractions are <0.2Å and the non- subunit expansions are <0.8Å.Furthermore, neither the contractions in the  subunits nor the expansions in non- subunits suggest a pre-activate like conformation of the non- subunits in the apo state.
Finally, we examined the changes in tilt angles that occur in each subunit ECD upon agonist binding.This was done in two ways: first by assessing how much each ECD changes its tilt angle along the z plane (relative to the pore) during activation and second by determining the relative change in tilt angle between the ECD and TMD in each subunit during activation.The former was measured by calculating the change in angle between the vector derived from longest principal axis used to create the ellipsoids around each ECD and the z-axis (or pore axis).During activation, each subunit (other than the δ subunit) undergoes a straightening motion that decreases the angle relative to the pore axis (panels A and B in Fig. S7).This motion is conserved, and of a similar magnitude, in  and non- subunits leaving no evidence for preactivation of non- subunits.It is notable however that while the non- subunits seem to pivot with their membrane juxtaposed ends staying in the same position, the  subunits seem to translate toward the pore axis while also tilting (Fig. S7 panel A).Because each subunit was aligned by their TMD, the relative translation of the  subunit ECD suggests the angle between ECD and TMD in the  subunits would differ from the non- subunits.To measure this, we calculated the change in angle between the vectors of the longest principal axis of ellipsoids around the ECD and TMD of the same subunit for both apo and bound states.Panel C shows that the interdomain tilt angle in the non- subunits increase during activation but decreases in the  subunits due to their scissor like interdomain motions.
Altogether, we were not able to identify any evidence for the pre-active like conformation of the ECD -TMD interfaces of the non- subunit extending more globally to the remainder of the ECD.A COM and vector corresponding to the principal axis through the COM were defined for each subunit ECD using only the Cα carbons from the conserved β stands in each subunit.The ECD tilt was defined as the angle between the principal ECD axis and the pore axis for each subunit.A COM and vector corresponding to the principal axis for each TMD was then defined and the angle between the vectors for the principal ECD and TMD axis was defined as the ECD-TMD tilt.

ca
No response (NR).No significant agonist induced current observed up to 4 days after cRNA injection.b  values from 1 .

Figure S2 .
Figure S2.Raw TEVC traces used to plot dose-response curves.Representative traces from select mutants are presented in the following seven pages as a change in current over time in response to increasing concentrations of ACh.Horizontal bars above each trace represent the time the oocyte was exposed to the indicated concentration of ACh.

Figure S3 .
Figure S3.The interdependence between γGly182/δGly188 and the αγ/αδ M2-M3 loops is mediated by a backbone hydrogen bond with αSer268.Apo (pink) and agonist-bound (green) states of the nAChR are shown in panel a with a dashed line highlighting the backbone hydrogen bond between γGly182 and αSer268.The αS268P mutant has almost no effect on channel function alone but eliminates the effect of the γG182A/δG188A double mutant.

Figure S4 .
Figure S4.The quadruple Asn αM2-M3 mutant stabilizes the resting state.The M2 α-helices and M2-M3 loops of both α subunits are shown on the left for the WT apo state after 100 ns of simulation (pink) and the WT nicotine state after minimization but prior to the simulation (green).Residues in the M2-M3 loop are shown as sticks and the α carbons of αVal255 (V13') are shown as spheres with the distance between the two depicted by dashed lines to highlight the difference in pore diameter.Three repeat all atom simulations of the apo nAChR (top right, green traces) show that the pore is stably closed.In contrast, unless the diffuse density observed in the pore is modeled with bound lipids (see4 ), the open pore of the nicotine bound structure closes within ~40 ns to a conformation with a diameter slightly larger than the apo state (top right, orange traces).On the bottom right, the pore diameter of the quadruple Asn mutant (orange) after 100 ns simulation is compared to the apo WT (green) after 100 ns simulation.The average distance between αV13' residues from the three repeats show that the quadruple Asn mutation in M2-M3 accelerates the collapse of the pore all the way back to the apo WT state.
the conformation adopted by the ECD of each subunit in apo and agonist bound forms of the Torpedo nAChR.

Figure S5 .
Figure S5.ECD twist does not provide evidence for pre-activation.Panel A shows top-down views of each subunit in the nAChR aligned by their TMDs in apo (pink) and agonist bound (green) states to show the relative motion of the ECD.Panel B shows change in ECD twist angle as defined in ref. 6. Panel C shows changes in solvent accessible surface area at each ECD -ECD interface.Panel D shows binding site contraction at each subunit interface as defined in ref. 7.

Figure S6 .
Figure S6.ECD spreading and β-expansion.Upper and lower ECD spreading (defined in ref. 6) is quantified for each subunit in panel A. The left shows the ECD of the apo (pink) and agonist bound (green) states aligned by their entire TMDs.Center of mass (COM) for the ECD of each subunit as well as the COM for the entire ECD are shown as spheres and the distances are depicted as dashed lines and plotted in the graph on the right.Panel B shows β-expansion in each subunit (also defined in ref. 6) with the minimal changes shown in both the models (left) and the graph (right).

Figure S7 .
Figure S7.ECD-tilt angles.A COM and vector corresponding to the principal axis through the COM were defined for each subunit ECD using only the Cα carbons from the conserved β stands in each subunit.The ECD tilt was defined as the angle between the principal ECD axis and the pore axis for each subunit.A COM and vector corresponding to the principal axis for each TMD was then defined and the angle between the vectors for the principal ECD and TMD axis was defined as the ECD-TMD tilt.

Table S2 .
8. Human adult FHill=6.736.Exact p values can be found in a supplemental file.c No response (NR).No significant agonist induced current observed up to 4 day after cRNA injection d No expression (NE).No significant expression or agonist induced current observed up to 4 day after cRNA injection Functional effects of mutations to 8-9/10-M1 and M2-M3 of the Torpedo nAChR.
a Measurements performed 2-4 days after cRNA injection (Vhold = -60 mV).Error values represented as standard deviation b Predicted fold change if both individual mutants influence function independently c p < 0.001 relative to WT via one-way ANOVA followed by Dunnet's post hoc test.DF=237.FEC50=180.6.FHill=3.562.Exact p values can be found in a supplemental file.dAnenergetic coupling is calculated from EC50 values to facilitate comparisons.Energy values provided are not quantitative (see methods and FigureS1).
a Measurements performed 2-4 days after cRNA injection (Vhold = -60 mV).Error values represented as standard deviation b Measurements performed 2 days after cRNA injection on a minimum of 10 oocyes.Error values represented as standard deviation c No response (NR).No significant agonist induced current observed up to 4 day after cRNA injection d No response (NE).No significant agonist induced current observed up to 4 day after cRNA injection e p < 0.001 relative to mock injected oocytes via one-way ANOVA followed by Dunnet's post hoc test f p < 0.001 relative to WT via one-way ANOVA followed by Dunnet's post hoc test.DF=183.FEC50=209.9.FHill=2.339.Exact p values can be found in a supplemental file.

Table S5 .
Subunit specific effects of mutating residues contributing to conserved tripartite salt bridges at the ECD -TMD interfaces.
cNo response (NE).No significant agonist induced current observed up to 4 day after cRNA injection d None of the mutants tested were significantly (p < 0.001) more expressed than mock injected oocytes via one-way ANOVA followed by Dunnet's post hoc test e p < 0.001 relative to WT via one-way ANOVA followed by Dunnet's post hoc test.DF=313.FEC50=112.5.FHill=3.667.Exact p values can be found in a supplemental file.