Subunit Symmetry at the Extracellular Domain-Transmembrane Domain Interface in Acetylcholine Receptor Channel Gating*

Transmitter molecules bind to synaptic acetylcholine receptor channels (AChRs) to promote a global channel-opening conformational change. Although the detailed mechanism that links ligand binding and channel gating is uncertain, the energy changes caused by mutations appear to be more symmetrical between subunits in the transmembrane domain compared with the extracellular domain. The only covalent connection between these domains is the pre-M1 linker, a stretch of five amino acids that joins strand β10 with the M1 helix. In each subunit, this linker has a central Arg (Arg3′), which only in the non-α-subunits is flanked by positively charged residues. Previous studies showed that mutations of Arg3′ in the α-subunit alter the gating equilibrium constant and reduce channel expression. We recorded single-channel currents and estimated the gating rate and equilibrium constants of adult mouse AChRs with mutations at the pre-M1 linker and the nearby residue Glu45 in non-α-subunits. In all subunits, mutations of Arg3′ had similar effects as in the α-subunit. In the ϵ-subunit, mutations of the flanking residues and Glu45 had only small effects, and there was no energy coupling between ϵGlu45 and ϵArg3′. The non-α-subunit Arg3′ residues had Φ-values that were similar to those for the α-subunit. The results suggest that there is a general symmetry between the AChR subunits during gating isomerization in this linker and that the central Arg is involved in expression more so than gating. The energy transfer through the AChR during gating appears to mainly involve Glu45, but only in the α-subunits.

by which the agonist affinity change and channel opening/closing are linked is not yet available.
The adult neuromuscular AChR consists of five homologous subunits (two ␣-subunits and one each of the ␤-, ␦-, and ⑀-subunits) folded symmetrically around the central axis of the pore (3). Its structure is modular, as the extracellular N-terminal half of each subunit is a ␤-barrel and the transmembrane C-terminal half is a four-␣-helix bundle (M1-M4). The five sets of ␤-barrels form the ECD, and the five sets of ␣-helices form the TMD. Several structures have revealed important features that are relevant to understanding the mechanism of energy transfer through the protein in the gating isomerization, including the Torpedo AChR (4), two prokaryotic pentameric ligand-gated ion channels crystallized in either a non-conducting (ELIC) (5) or a presumably conducting conformation (GLIC) (6,7), an ECD fragment of the mouse AChR ␣-subunit (8), and the ECD homolog, the acetylcholine-binding protein (9). A comparison of the ELIC and GLIC x-ray structures suggests that the porelining M2 helices tilt tangentially and radially as part of the channel-opening process (6,7). Despite this structural information, we are still unsure of the molecular events that constitute the affinity change at the binding sites and the conductance change at the pore or the intermediate events that couple structural changes in these two widely separated domains (10).
The interface between the ECD and TMD is a complex region that has been studied in several members of the Cys loop receptor channel family (11)(12)(13)(14)(15)(16)(17)(18). This region has many charged residues and hence the potential for many non-bonded interactions. The only covalent link between the ECD and TMD is a stretch of five residues that join strand ␤10 with the M1 helix, known as the "pre-M1" linker ( Fig. 1). This linker has a central positively charged Arg that is conserved among all pentameric ligand-gated receptor channels (12). Here, we will call this position Arg 3Ј to mark it as the third position in this linker. Lee and Sine (12) proposed that the perturbation of a salt bridge between ␣Arg 3Ј and loop 2 residue ␣Glu 45 is the principal event that links the ECD and the TMD in gating, but other results suggest that ␣Arg 3Ј plays a smaller role in gating but is important for receptor expression (14,17).
The affinity change at the binding sites involves mainly ␣-subunit residues, but the opening and closing of the gate near the M2 equator involve the rearrangements of atoms in all five subunits. One goal of our experiments was to assay the symmetry of the gating energy changes at the pre-M1 linker, a location that is about halfway between the binding sites and the gate. In general, previous results indicate that large gating energy changes in the ECD are predominantly in the ␣-subunit but become more evenly spread among all subunits in the TMD (12, 19 -25). Although the mechanism of this spreading of the gating energy changes is not clear, an intersubunit energy transfer has been found between ␣Tyr 127 and ⑀Asn 39 or ␦Asn 41 (19).
So far, only one mutation at one pre-M1 linker position has been studied in all subunits using single-channel kinetic analysis. A Gln substitution at the fourth linker residue modestly increases diliganded gating (by ϳ3-fold) in the ␣-subunit but is without effect in the non-␣-subunits (26). Here, we report the effects of mutations of multiple pre-M1 residues in non-␣-subunits, estimated from the single-channel rate and equilibrium constants of the adult mouse AChR gating isomerization.

EXPERIMENTAL PROCEDURES
Mutagenesis and Expression-A detailed description of our methods is described by Jha et al. (27). Mutant AChR cDNAs were made by QuikChange TM site-directed mutagenesis (Stratagene) and confirmed by sequencing. We made 26 mutants of the ⑀-subunit, three double-mutant combinations of the ⑀-subunit, two mutants of the ␤-subunit, and two mutants of the ␦-subunit. HEK293 cells were transiently transfected by the calcium phosphate precipitation method. HEK cells were incubated with 2.5-5 g of mouse WT or mutant cDNAs in a 35-mm culture dish at a subunit ratio of 2:1:1:1 (␣/␤/␦/⑀). After ϳ16 h of incubation at 37°C, the transfected cells were washed with HEK culture medium. Electrophysiology recordings were performed 20 -40 h post-transfection.

Single-channel Recordings and Kinetic Analysis-Recordings
were carried out in the cell-attached patch configuration at room temperature (23°C). The pipette and bath solutions were both Dulbecco's PBS (137 mM NaCl, 0.9 mM CaCl 2 , 2.7 mM KCl, 1.5 mM KH 2 PO 4 , 0.5 mM MgCl 2 , and 8.1 mM Na 2 HPO 4 at pH 7.2). Pipettes were pulled from borosilicate capillaries to a resistance of ϳ10 megohms and coated with SYLGARD (Dow Corning Corp., Midland, MI). The pipette solution contained 0.5 mM acetylcholine, 20 mM choline, or 5 mM carbamylcholine. These agonist concentrations are approximately five times the corresponding equilibrium dissociation constants (K d ); thus, almost all currents arose from diliganded AChRs. Because the mutations were far from the binding site, we assumed that they did not change K d . The high concentration of agonist caused partial channel block, which decreases both the apparent single-channel current amplitude and the apparent closing rate constant. Although none of the mutations changed the degree of channel block, nonetheless we measured the closing rate constant using low concentrations of agonist (30 M acetylcholine, 200 M choline, and 200 M carbamylcholine), at which channel block is insignificant. The diliganded gating equilibrium constant (E 2 ) was calculated as the ratio of the opening/ closing rate constant. Choline was used to measure the diliganded opening rate constant for AChR mutants where E 2 was larger than or equal to the WT; acetylcholine was used for mutants where E 2 was less than the WT; and carbamylcholine was used to measure mutants where E 2 was approximately equal to the WT. It has been shown for many non-binding site mutations that different agonists support the same ⌽-values and fold-changes in E 2 (21).
Cells were held at a pipette potential of ϩ70 mV, which corresponds to a membrane potential of approximately Ϫ100 mV. Errors in the rate constants associated with the errors in the membrane voltage are small (in WT AChRs, an ϳ70-mV depolarization is necessary to decrease E 2 by e-fold) (28). Currents were filtered at 20 kHz and digitized at a sampling frequency of 50 kHz. Kinetic analyses were done using QUB software. Currents were idealized using the SKM (segmental K-Means) algorithm filtered at 12 kHz with a C 7 O (closed 7 open) model The conserved central Arg at position 3Ј is shown in boldface: ␣〈rg 209 , ␤〈rg 220 , ␦〈rg 223 , and ⑀〈rg 218 in mouse AChR numbering.

Protein
Positions 1-5 with starting rate constants of 100 s Ϫ1 . The diliganded opening (f 2 ) and closing (b 2 ) rate constants were estimated from idealized interval durations using a maximum interval likelihood algorithm after incorporating a dead time of 25 s (29). The diliganded rate constants were measured multiple times (n ϭ two to five patches) and then averaged. ⌽ was estimated as the slope of the linear fit to the log-log rateequilibrium free energy relationship (R/E analysis). The range energy (kcal/mol) ϭ Ϫ0.59 ln(E 2 max /E 2 min ), where the superscripts are the E 2 values for the side chains generating the largest and smallest E 2 values, respectively. The coupling free energy was calculated as ⌬⌬G (kcal/mol) ϭ Ϫ0.59 ln((E double mutant )/(E mutant 1 *E mutant 2 )).

RESULTS
⑀-Subunit Linker- Table 1 shows a sequence alignment of the pre-M1 linker and that Arg 3Ј is completely conserved. In the ␣-subunit linker, this is the only positively charged residue, but in the non-␣-subunits, it is flanked by two additional basic amino acids.
We estimated residue range energy (see "Experimental Procedures") and ⌽-values from AChRs with a mutation of one of the three positively charged amino acids in the ⑀-linker (⑀Arg 2Ј , ⑀Arg 3Ј , ⑀Lys 4Ј ). None of the side chain substitutions at ⑀Arg 2Ј (Ala, Cys, Asp, Asn, Trp, or Val) changed the diliganded gating equilibrium constant (E 2 ) by Ͼ2-fold ( Fig. 2 and Table 2). This set of substitutions was selected to include residues of different size, charge, and hydrophobicity. All mutations caused a slight reduction in E 2 , with the largest being for Asn (1.8-fold, which corresponds to a range energy of 0.3 kcal/mol). We conclude that like its homolog in the ␣-subunit, the ⑀2Ј-side chain is nearly isoenergetic between the ground state conformations, which suggests that this amino acid does not move with respect to its local environment during the gating isomerization.
Nine substitutions of the central position ⑀Arg 3Ј were examined, but only Ala, His, Lys, Asn, and Gln expressed functional channels. We were unable to observe single-channel currents from the Cys, Asp, Glu, and Val mutants (3-10 patches per mutant, recording for 15-25 min/ patch). In this respect, position ⑀3Ј behaves similarly to its homolog in the ␣-subunit (14) and to ␣ 1 Arg 220 in the GABA receptor channel (17), where mutations also reduce the expression of functional channels. Mutations of ⑀Arg 3Ј had substantial effects on E 2 ( Table 2). The Ala, His, Lys, Asn, and Gln substitutions all decreased E 2 compared with the  Table 2). The WT is boxed.

Pre-M1 Linker of the Acetylcholine Receptor
WT (Fig. 3). The Ala and Asn substitutions had the largest effect and decreased E 2 by ϳ180-fold. The range energy for ⑀Arg 3Ј was ϳ3.1 kcal/mol, which is about the median value for two ␣-subunit mutations (10,30). The R/E plot for ⑀Arg 3Ј had a slope (⌽-value) of 0.73 Ϯ 0.06 (Fig. 3B), which indicates that the change in E 2 was caused mainly by a reduction in the forward channel-opening rate constant (f 2 ). This ⌽-value is the same as for ␣Arg 3Ј (14), which suggests that the central Arg residues experience a change in energy at about the same time in the gating reaction in the ␣and ⑀-subunits.
All substitutions tested at ⑀Lys 4Ј expressed functional AChRs. Ala, Cys, Asp, and Trp mutations increased E 2 , but only modestly ( Fig. 4 and Table 2). The Trp mutant showed the largest energy change (ϳ1.1 kcal/mol). AChRs with an Asn or Val side chain here had WT gating properties, as did a Gln substitution (26). The ⌽-value of the ⑀Lys 4Ј series was 0.63 Ϯ 0.09, similar to that of its neighbor ⑀Arg 3Ј . However, in the ␣-subunit, the ⌽-value of ␣Leu 4Ј (⌽ ϭ 0.35) was distinctly lower than that of the adjacent residue ␣Arg 3Ј (⌽ ϭ 0.72) (14).
Energetic Coupling between ⑀Arg 3Ј and ⑀Glu 45 -In the cryo-EM structure of the Torpedo AChR (4), the positively charged side chain of ␣Arg 3Ј faces the negatively charged side chain of ␣Glu 45 , and swapping charges here restores functional gating (12). We investigated the effects of five side chain substitutions at ⑀Glu 45 and three ⑀Glu 45 / ⑀Arg 3Ј double-mutant combinations to explore the interactions between these ⑀-positions during gating. Fig. 5 and Table 3 show the effects of mutating ⑀Glu 45 . The ⑀Glu 45 substitutions Cys, Arg, Trp, and Val all decreased E 2 , but only slightly (Ͻ1 kcal/mol). The ⌽-value measured for the ⑀Glu 45 mutation series was 0.50 Ϯ 0.15. Because of the near-WT gating properties of the ⑀E45A mutant, we conclude that a negatively charged side chain at this position in the ⑀-subunit is not critical for efficient gating. The small range energy makes the ⑀Glu 45 ⌽-value estimate imprecise (31), but it may be that this residue changes its energy somewhat after its ␣-subunit homolog (⌽ ϭ 0.80) (14).  Table 2). F, acetylcholine-activated. The WT is boxed. The ⌽-value (lower right) was estimated as the linear slope of log f 2 versus log E 2 and gives the relative timing of the residue's gating energy change (1 to 0, start to end). The ⑀Arg 3Ј side chain changes its energy relatively early in the channel-opening process.  Table 2). The WT is boxed. The ⌽-value (lower right) was estimated as the linear slope of log f 2 versus log E 2 .  Table 3). The WT is boxed. The ⌽-value (lower right) was estimated as the linear slope of log f 2 versus log E 2 . DECEMBER 10, 2010 • VOLUME 285 • NUMBER 50

Pre-M1 Linker of the Acetylcholine Receptor
We created three ⑀Glu 45 /⑀Arg 3Ј double-mutant constructs: Arg/Glu, Ala/Gln, and Ala/Lys. In the ␣-subunit, the side chains of ␣Glu 45 influence the expression of ␣Arg 3Ј mutants (14). The ⑀R3ЈE construct alone did not express functional AChRs, and when we expressed this along with ⑀E45R (a charge swap), we still did not observe the expression of functional AChRs currents. The other two double-mutant pairs (Ala/Gln and Ala/Lys) did express functional channels and produced AChRs having slightly larger E 2 values than predicted assuming independence ( Fig. 6 and Table 3). The Ala/Gln and Ala/Lys double mutants exhibited very small coupling energies (Ϫ0.3 and Ϫ0.9 kcal/mol, respectively). The R/E plot for ⑀Arg 3Ј on the ⑀E45A background yielded a ⌽-value of 0.78 Ϯ 0.06 (Fig. 6B), which is similar to its value on the WT background (Fig. 3).
␤and ␦-Subunit Linkers-We also examined the kinetics of mutations at position 3Ј in the ␤and ␦-subunits ( Fig. 7 and Table 4). The mutation ␤R3ЈK had little effect on gating, whereas a Gln mutation here resulted in a 49-fold reduction of E 2 . The estimated ⌽-value of position ␤3Ј was 0.44 Ϯ 0.06 (Fig.  7B). In the ␦-subunit, Lys and Gln substitutions were assayed using two different agonists, acetylcholine and carbamylcholine. With both ␦Arg 3Ј mutations, the fold-changes in E 2 were similar for these ligands, even though acetylcholine is a full agonist and carbamylcholine is a partial agonist. The R/E plot for position ␦3Ј yielded a ⌽-value of 0.54 Ϯ 0.07 and a range energy of ϳ2.0 kcal/mol.

DISCUSSION
In the Torpedo AChR (4) structure, the ␣Arg 3Ј side chain forms a salt bridge with ␣Glu 45 in loop 2 (and possibly with ␣Glu 175 in loop 9). In the ECD fragment of the mouse nicotinic acetylcholine receptor ␣-subunit (8), ␣Arg 3Ј interacts indirectly with ␣Glu 45 via a structural water. An electrostatic contact is present between the Arg 3Ј and Glu 31 (in loop 2) side chains in the x-ray structure of GLIC (7). ELIC lacks this salt bridge because the loop 2 residue is a Thr that faces, but does not contact, Arg 3Ј (which does appear to form a salt bridge with Asp 122 in loop 7 and Glu 159 in loop 9). Thus, the conserved pre-M1 linker Arg 3Ј side chain is evidently involved in protein stability through electrostatic interactions with surrounding loops. However, analyses of function suggest that a perturbation of the salt bridge between ␣Arg 3Ј and ␣Glu 45 is not a critically important event in AChR gating (14).
One of our goals was to probe the degree of functional symmetry between subunits with regard to residues in the pre-M1 linker and loop 2. The non-␣-linker contains three positive residues, whereas the ␣-linker contains only one. Recall that in the ␣-subunit linker, (i) many mutations at position 3Ј reduce expression, (ii) only mutations at positions 3Ј and 4Ј affect E 2 , and (iii) loop 2 residue ␣Glu 45 experiences a very large range energy change (ϳ5 kcal/mol) early in the channel-opening process (⌽ ϳ 0.8) (14). We can compare this basic pattern in the ␣-subunit with that we have found in the ⑀-subunit and, to a lesser extent, in the ␤and ␦-subunits.
Position 2Ј is isoenergetic in both the ␣and ⑀-subunits. All mutants tested here in both subunits expressed functional AChRs that had WT-like currents. It appears that in the mouse AChR, this position in the pre-M1 linker in the ␣and ⑀-subunits is not essential for folding, expression, conductance, gating, or desensitization.
Position 3Ј is much more interesting. In both the ␣and ⑀-subunits, many substitutions here prevented the expression of functional AChRs. Single-channel currents were observed  (Table 4). F, acetylcholine (ACh); E, choline (Cho). The WT is boxed. The ⌽-value (lower right) was estimated as the linear slope of log f 2 versus log E 2 . only with the Arg, Gln, His, and Lys side chains in both the ␣and ⑀-subunits and also with Ala and Asn only in the ⑀-subunit. This difference in expression may simply reflect the fact that each AChR has two ␣-subunits but only one ⑀-subunit. The R3ЈQ substitution resulted in a reduction of E 2 in both ␣-subunits and all non-␣-subunits, but the gating energy change was only moderate. The range energy (per subunit) was slightly larger in the ⑀-subunit (ϳ3 kcal/mol) compared with the ␣-subunit (ϳ2 kcal/mol); this result was influenced by the inclusion of the Ala and Asn substitutions only in the ⑀-subunit. Limiting this estimate to the Arg, Gln, His, and Lys side chains, the range energies were 1.8 and 1.0 kcal/mol/subunit for the ␣and ⑀-subunits, respectively. The ⌽-value for position 3Ј was the same in the ␣and ⑀-subunits, but those in the ␤and ␦-subunits were somewhat smaller. Overall, this pattern for the central pre-M1 linker position suggests that this region is approximately symmetric between subunits with regard to protein expression and the magnitude and relative timing of the gating energy changes. However, a Lys substitution at Arg 3Ј resulted in an E 2 increase in the ␣-subunits but a decrease in the ␦and ⑀-subunits and no change in the ␤-subunit, suggesting that the ␣-subunit may have a unique chemical environment near the pre-M1 linker. Position 4Ј is not conserved between the ␣and ⑀-subunits (Leu versus Lys), and this residue behaved quite differently in these two subunits. The ␣L4ЈK mutation increased E 2 by ϳ50-fold, whereas the ⑀K4ЈV mutation (Leu was not tested) had no effect. Indeed, only small changes were observed for all tested mutations of ⑀Lys 4Ј , whereas all mutations in the ␣-subunit modestly increased E 2 (range energy of ϳ2.3 kcal/mol). An even more interesting difference is the distinct ⌽-values for position 4Ј in the ␣-subunit versus the ⑀-subunit (0.35 versus 0.63). This suggests that in the ␣-subunit, but not the ⑀-subunit, there is a boundary between pre-M1 linker positions 3Ј and 4Ј that defines the relative timing of the gating movements of the ECD and TMD. The gating behavior of loop 2 residue Glu 45 was also very different in the ␣-subunit compared with the ⑀-subunit. Substitutions in the ⑀subunit decreased E 2 very slightly, whereas those in the ␣-subunit either increased or decreased E 2 substantially. Indeed, the ␣Glu 45 range energy (ϳ5 kcal/mol, His to Ile) is the third largest measured so far in the ECD (after ␣Ala 96 and ␣Tyr 127 ) (30). An important and common feature shared by Glu 45 in FIGURE 7. Mutations of ␤Arg 3 (␤Arg 220 ) and ␦Arg 3 (␦Arg 223 ) decrease the diliganded gating equilibrium constant. A and C, example clusters of position ␤3Ј and ␦3Ј mutants activated by 0.5 mM acetylcholine (ACh), 5 mM carbamylcholine (CCh), or 20 mM choline (Cho). In C, the left clusters were activated by acetylcholine and the right clusters by carbamylcholine. B and D, R/E analyses of positions ␤3Ј and ␦3Ј mutants. Each point represents the average of three to four patches with its S.D. (Table 4). F, acetylcholine; E, choline; ‚, carbamylcholine. The WT is boxed. The ⌽-value (lower right) was estimated as the linear slope of log f 2 versus log E 2 .  DECEMBER 10, 2010 • VOLUME 285 • NUMBER 50 the ␣and ⑀-subunits is that double mutants of Arg 3Ј and Glu 45 in both subunits suggest weak energetic interactions between these side chains in gating.

Pre-M1 Linker of the Acetylcholine Receptor
In summary, the pre-M1 linker is mostly symmetrical between subunits at positions 2Ј and 3Ј with regard to expression, gating, and interactions with Glu 45 in loop 2. Asymmetry between subunits was apparent at pre-M1 linker position 4Ј and Glu 45 , where only the ␣-subunit plays a major role. It is of interest to identify the chemical details that define the distinct subunit environments at the ECD-TMD interface and to further explore how energy spreads between subunits in the AChR gating isomerization.