Identification of Residues in the Adult Nicotinic Acetylcholine Receptor That Confer Selectivity for Curariform Antagonists*

We identify residues in the ε and δ subunits of the adult nicotinic acetylcholine receptor that give the αε and αδ binding sites different affinities for the curariform antagonist dimethyl d-tubocurarine (DMT). By constructing ε-δ subunit chimeras, coexpressing them with complementary subunits, and measuring DMT binding, we identify two pairs of residues, Ileε58/Hisδ60 and Aspε59/Alaδ61, responsible for DMT site selectivity in the adult receptor. The two determinants contribute approximately equally to the binding site and interact in contributing to the site. Exchange of these residues from one subunit to the other exchanges the affinities of the resulting binding sites. These determinants in the adult receptor are far from those that confer site selectivity in the fetal receptor; determinants in the fetal receptor are Ileγ116/Valδ118, Tyrγ117/Thrδ119, and Serγ161/Lysδ163. Thus, alternative residues confer DMT selectivity in fetal and adult acetylcholine receptors.

Acetylcholine receptors (AChRs) 1 from vertebrate skeletal muscle are pentamers of homologous subunits with the compositions ␣ 2 ␤␥␦ in fetal muscle and ␣ 2 ␤⑀␦ in adult muscle (2). The ligand-binding sites are generated by pairs of subunits, ␣␦ and either ␣␥ or ␣⑀. The two sites in each receptor selectively bind competitive antagonists of the curare family (3,4), with the ␣␥ and ␣⑀ sites binding with high affinity and the ␣␦ site binding with low affinity. Because the ␣ subunit is common to each binding site, differences in affinity are attributed to different contributions of the non-␣ subunits (5)(6)(7)(8).
We previously identified three pairs of residues at equivalent positions in the ␥ and ␦ subunits that confer DMT selectivity in the fetal receptor (1). The primary source of selectivity is the pair of equivalent residues Tyr ␥117 /Thr ␦119 , of which Tyr ␥117 stabilizes one of two quaternary nitrogens in DMT through a -cation interaction (9). Paradoxically, the ⑀ subunit of the adult receptor, which also confers high affinity for DMT, contains serine at this key position (10). We therefore reasoned that alternative residues confer high affinity for DMT at the ␣⑀ site. To identify these alternative residues, we constructed ⑀-␦ subunit chimeras, coexpressed them with complementary subunits, and measured binding of DMT. Our results reveal alternative residues in the ⑀ and ␦ subunits that confer site selectivity for DMT in the adult receptor.

EXPERIMENTAL PROCEDURES
Materials-Dimethyl d-tubocurarine was generously provided by Lilly. 125 I-Labeled ␣-bungarotoxin was purchased from NEN Life Science Products, d-tubocurarine chloride from ICN Pharmaceuticals, Inc., and the 293 human embryonic kidney cell line from the American Type Culture Collection.
Plasmids and Mutagenesis-Mouse subunit AChR cDNAs were generously provided by Drs. Norman Davidson and John Merlie and were subcloned into the cytomegalovirus-based expression vector pRBG4 as described (1). Chimeric subunit cDNAs were constructed by bridging naturally occurring or mutagenically installed restriction sites with synthetic double-stranded oligonucleotides (1). The chimeras are designated as follows. The first letter gives the subunit from which Nterminal sequence is taken, the following number gives the position of the chimeric junction, and the final letter gives the subunit from which C-terminal sequence is taken. The chimera ⑀63␦ was constructed by bridging a 35-base pair (bp) synthetic double-stranded oligonucleotide from a PflMI restriction site in the ⑀ subunit to a mutagenically installed SalI site in the ␦ subunit. The chimera ⑀43␦63⑀ was constructed by bridging the BstEII site in the ⑀ subunit and the SalI site in ⑀63␦ with an 85-bp synthetic oligonucleotide. The chimera ⑀43␦56⑀ was constructed by bridging a 60-bp oligonucleotide from the BstEII site to the PflMI site in the ⑀ subunit and ligating a 1030-bp PflMI-PflMI fragment prepared by digestion of the ⑀ subunit. To construct the chimera ⑀43␦59⑀, a 3426-bp PflMI-DraIII fragment prepared by digestion of ⑀43␦56⑀ was ligated with a 1900-bp AflIII-DraIII fragment from the ⑀ subunit and a 90-bp oligonucleotide bridging the PflMI and AflIII sites. The chimeras ⑀57␦63⑀, ⑀58␦63⑀, and ⑀59␦63⑀ were constructed from the same cassette using a 52-bp oligonucleotide that bridges the PflMI site and a mutagenically installed HgaI site. The point mutations ⑀I58H, ⑀D59A, and ⑀I58H/⑀D59A were constructed by bridging a 90-bp oligonucleotide from the PflMI site to the AflIII site. The double mutant ␦H60I/␦A61D was constructed by bridging a 50-bp oligonucleotide from PflMI to HgaI in the ␦ subunit and ligating with a 300-bp HgaI-PflMI fragment. The double mutant ␥M58H/␥Q59A was constructed by bridging a 98-bp oligonucleotide from the PflMI site to a mutagenically installed EagI site. The triple mutant ␥I116V/␥Y117T/␥S161K was constructed as described (1). This triple mutant was further modified to include mutations of determinants identified in this study to ␥M58I/ ␥Q59D/␥I116V/␥Y117T/␥S161K, which was constructed by bridging a 106-bp oligonucleotide from the PflMI site to the EcoRV site. All constructs were confirmed by dideoxy sequencing.

Expression of Mutant Receptors and Ligand
Binding Measurements-Human embryonic kidney cells were transfected with mutant or wild-type AChR subunit cDNAs using calcium phosphate precipitation as described (1). Three days after transfection, intact human embryonic kidney cells were harvested by gentle agitation in phosphatebuffered saline plus 5 mM EDTA. DMT binding to intact cells was measured by competition against the initial rate of 125 I-␣-bungarotoxin binding (11). After harvesting, the cells were briefly centrifuged, resuspended in potassium Ringer high solution, and divided into aliquots for DMT binding measurements. Potassium Ringer solution contains 140 mM KCl, 5.4 mM NaCl, 1.8 mM CaCl 2 , 1.7 mM MgCl 2 , 25 mM HEPES, and 30 mg/liter bovine serum albumin adjusted to pH 7.4 with 10 -11 mM NaOH. Specified concentrations of DMT were added 30 min prior to the addition of 125 I-␣-bungarotoxin, which was allowed to bind for 30 min to occupy approximately half of the surface receptors. Binding was terminated by the addition of 2 ml of potassium Ringer solution containing 300 M d-tubocurarine chloride. Cells were then harvested by filtration through Whatman GF/B filters using a Brandel Cell Harvester and washed four times with 3 ml of potassium Ringer solution.
Prior to use, filters were soaked in potassium Ringer solution containing 4% skim milk for a minimum of 2 h. Nonspecific binding was determined in the presence of 10 mM carbamylcholine. The total number of ␣-bungarotoxin sites was determined by incubation with toxin for 120 min. The initial rate of ␣-bungarotoxin binding was calculated as described (11) to yield fractional occupancy by DMT. Binding measurements were analyzed according to either the monophasic Hill equation (Equation 1) or the sum of two distinct binding sites (Equation 2), where Y is fractional occupancy by DMT, n is the Hill coefficient, K app is an apparent dissociation constant for a monophasic binding profile, K A and K B are intrinsic dissociation constants for two binding sites, fract A is the fraction of sites with dissociation constant K A , and ACh is acetylcholine. For binding determinations from single experiments, fitted parameters and standard errors were obtained using the program UltraFit (BIOSOFT). For multiple experiments, means Ϯ S.D. of the individual fitted parameters are presented (Table I).

RESULTS
⑀-␦ Subunit Chimeras-Previous work showed that the pair of equivalent residues Tyr ␥117 /Thr ␦119 are major determinants of DMT selectivity in the fetal AChR (1). Tyr ␥117 contributes to high affinity of the ␣␥ site, whereas Thr ␦119 contributes to low affinity of the ␣␦ site. Because serine occupies the equivalent position in the ⑀ subunit and high affinity requires an aromatic side chain, the ␣⑀ site is not expected to bind DMT with high affinity. To determine the origin of high affinity conferred by the ⑀ subunit, we constructed a series of ⑀-␦ subunit chimeras (Fig. 1, A and B), coexpressed them with complementary subunits, and measured DMT binding by competition against the initial rate of ␣-125 I-bungarotoxin binding.
Selectivity of the adult AChR for DMT is illustrated in Fig.  1C; distinct affinities of the ␣⑀ and ␣␦ sites are clearly resolved, with the two-site fit disclosing dissociation constants differing by 70-fold (Table I). Substituting ⑀ sequence into the N-terminal 63 positions of the ␦ subunit increases DMT affinity to approach that conferred by the pure ⑀ subunit (Fig. 1C, ⑀63␦). Conversely, substituting ␦ sequence between positions 43 and 63 of the ⑀ subunit decreases affinity to approach that of the pure ␦ subunit (Fig. 1C, ⑀43␦63⑀). Thus, determinants of DMT selectivity in the adult AChR are located in the major extracellular domain between residues 43 and 63 of the ⑀ subunit and equivalent residues of the ␦ subunit.
Expanding the Window of Selectivity with ␦L121K-Because we anticipated multiple contributions to DMT selectivity, we sought to increase the window of selectivity to help discern small affinity changes. Recent studies in our laboratory revealed that the mutation ␦L121K decreases DMT affinity for the ␣␦ site while retaining good levels of expression (19). When incorporated into the adult AChR, ␦L121K markedly expands the selectivity window from 70-to ϳ2000-fold (Fig. 2B). The broad plateau allows determination of the fraction of each site and clear resolution of the dissociation constants for the two sites. The apparent fraction of sites with high affinity is ϳ0.6 in the AChR containing ␦L121K. Because the receptors lacking the ␦ subunit (i.e. ␣ 2 ␤⑀ 2 ) used in this study do not express significantly at 37°C, the binding profile should arise entirely from ␣ 2 ␤⑀␦L121K receptors, and the fraction of each site should equal 0.5. Thus, the apparent fraction of 0.6 suggests that the intrinsic rate of ␣-toxin binding is somewhat slower at the ␣␦L121K site compared with the native ␣⑀ or ␣␦ sites.
We further localized selectivity determinants by constructing ⑀-␦-⑀ subunit chimeras that maintained one junction at position 43, but shifted the other junction from position 63 toward the N terminus ( Fig. 2A). When coexpressed with ␣, ␤, and ␦L121K subunits, the chimera ⑀43␦59⑀ confers low, ␦-like affinity (Fig. 2B), as observed for ⑀43␦63⑀ (Fig. 1C). The result-ing binding profile is the sum of contributions of the site formed by ␣ and ␦L121K and that formed by ␣ and ⑀43␦59⑀. We determined the dissociation constant of the site formed by ␣ and ⑀43␦59⑀ by fitting a two-site equation (Equation 2) to the data; the fit reveals a dissociation constant of 16 M for the site formed by ␣ and ⑀43␦59⑀, indistinguishable from that of the pure ␣␦ site (Table I). The dissociation constant of 0.39 mM for the ␣␦L121K site agrees with that obtained for the same site in receptors containing the wild-type ⑀ subunit (Table I). These results narrow the region containing selectivity determinants to between residues 43 and 60 of the ⑀ subunit.
Moving the chimera junction three more positions toward the N terminus, yielding ⑀43␦56⑀ ( Fig. 2A), increases DMT affinity to that conferred by the pure ⑀ subunit (Fig. 2B). Observation of pure ⑀-like affinity suggests that the region from residues 44 to 56 contains no additional selectivity determinants. These results narrow the region containing selectivity determinants to between residues 56 and 60 of the ⑀ subunit; the corresponding ⑀ sequence is GID, whereas the equivalent ␦ sequence is DHA.
Point Mutations of Selectivity Determinants-To determine whether the selectivity determinants identified using chimeras TABLE I DMT binding parameters for receptors containing wild-type or mutant ⑀ subunits K A and K R are dissociation constants for each site obtained from a two-site fit. fract A is the fraction A of sites with dissociation constant K A and is set to 0.5 for wild-type or mutant ⑀ coexpressed with wild-type ␦. fract A is a free parameter for wild-type or mutant ⑀ coexpressed with ␦L121K. The number of experiments is given in parentheses. . For the wild-type ⑀ subunit and the chimera ⑀43␦56⑀, the curve through the data points is a least-squares fit to the two-site equation with the fitted parameters K A ϭ 0.20 M, K B ϭ 0.37 mM, and fract A ϭ 0.62. For clarity, a single averaged curve is drawn, and results of the individual fits are given in Table I. For the chimera ⑀43␦59⑀, the curve is the two-site fit with K A ϭ 16 M, K B ϭ 0.39 mM, and fract A ϭ 0.62. Means Ϯ S.E. of the fitted parameters are given in Table I.  Table I. are solely responsible for DMT selectivity in the adult receptor, we constructed point mutations in the ⑀ subunit, coexpressed them with ␣, ␤, and ␦L121K subunits, and measured DMT binding. As observed with the chimera ⑀58␦63⑀, the single point mutations ⑀I58H and ⑀D59A partially decrease the affinity of the ␣⑀ site to approach that of the ␣␦ site (Fig. 4A). Moreover, the double mutation ⑀I58H/⑀D59A fully decreases the affinity to that conferred by both the chimera ⑀57␦63⑀ and the pure ␦ subunit (Fig. 4A). We determined the dissociation constant of the site formed by ␣ and ⑀I58H/⑀D59A by fitting a two-site equation (Equation 2) to the data; the fit revealed a dissociation constant of 20 M for the site formed by ␣ and ⑀I58H/⑀D59A, close to that of the native ␣␦ site (Table I). Thus, the pair of residues Ile ⑀58 and Asp ⑀59 fully account for DMT site selectivity in the adult receptor.
Having defined selectivity determinants using ␦L121K to increase the window of selectivity, we sought to confirm the contributions of Ile ⑀58 and Asp ⑀59 in the presence of complementary wild-type subunits. Again, the ⑀ single mutations partially decrease affinity, and the double mutation fully decreases affinity to that of the pure ␦ subunit (Fig. 4B). Fitting the Hill equation to the data for the ⑀ double mutant reveals a single class of sites with a dissociation constant of 14 M (Table I).
Exchange of Selectivity Determinants between the ⑀ and ␦ Subunits-To further confirm that the pairs of equivalent residues Ile ⑀58 /His ␦60 and Asp ⑀59 /Ala ␦61 are solely responsible for DMT selectivity in the adult receptor, we expressed the double mutants ⑀I58H/⑀D59A and ␦H60I/␦A61D, alone or together, and measured DMT binding. Receptors containing ␦H60I/ ␦A61D bind DMT with a single high affinity dissociation constant, approaching that conferred by the ⑀ subunit ( Fig. 4C and Table I). As just described, receptors containing ⑀I58H/⑀D59A bind DMT with a single low affinity dissociation constant, approaching that conferred by the ␦ subunit. Moreover, incorporating both the ⑀ and ␦ double mutants into a single receptor mimics the selective binding of DMT to the wild-type adult receptor (Fig. 4C). Thus, the pair of equivalent residues Ile ⑀58 / His ␦60 and Asp ⑀59 /Ala ␦61 account entirely for DMT selectivity in the adult receptor.
Point Mutations of Ser ⑀117 -The experiments using ⑀-␦ subunit chimeras were motivated by the presence of serine at position 117 of the ⑀ subunit and the observation that tyrosine at the equivalent position of the ␥ subunit is associated with high affinity for curariform antagonists (1,9). To determine whether Ser ⑀117 is nevertheless close enough to affect DMT binding, we mutated Ser ⑀117 to threonine, valine, tyrosine, phenylalanine, tryptophan, and arginine. The phenylalanine mutation slightly enhances the affinity of DMT for the ␣⑀ site, whereas the threonine and tyrosine mutations are without effect (Table II). The remaining aromatic mutation, tryptophan, slightly decreases affinity to about the same extent observed with the valine mutation. The arginine mutation shows the greatest decrease in affinity. Thus, four of the six mutations at Ser ⑀117 produce small changes in the affinity of the ␣⑀ site, with little effect on the affinity of the ␣␦ site. The side chain specificity of position 117 of the ⑀ subunit, however, differs markedly from that observed with point mutations of Tyr ␥117 , where, for example, ␥Y117W enhances affinity 30-fold compared with ␥Y117S (9). The relatively small effects of these  Table I.  ␥58 and Gln ␥59 -Because the determinants identified in this study differ from those that confer high affinity in the homologous ␥ subunit, we mutated the equivalent residues in the ␥ subunit, Met ␥58 and Gln ␥59 , to their counterparts in the low affinity ␦ subunit. We coexpressed the resulting double mutant (␥M58H/␥Q59A) with ␣, ␤, and ␦ subunits and measured DMT binding. The site formed by ␣ and ␥M58H/␥Q59A maintains high affinity for DMT, with the dissociation constant of 0.23 M close to that of the native ␣␥ site (Table III). Thus, the determinants identified in this study do not affect the contribution of the ␥ subunit to DMT binding.
We considered the possibility that ␥M58H/␥Q59A does not decrease DMT affinity because the previously identified determinants of high affinity (Ile ␥116 , Tyr ␥117 , and Ser ␥161 ) are still present. Thus, we constructed ␥I116V/␥Y117T/␥S161K to mimic the low affinity ␦ subunit (1) and then added the high affinity determinants from the ⑀ subunit to form ␥M58I/␥Q59D/ ␥I116V/␥Y117T/␥S161K. The triple mutation fully decreases DMT affinity to that of the pure ␦ subunit (Table III), as described previously (1). However, the affinity conferred by the quintuple mutation coincides with that of the triple mutation alone, confirming that Met ␥58 and Gln ␥59 do not contribute to DMT selectivity in the ␥ subunit. Thus, distinct sets of residues confer high affinity to the ␥ and ⑀ subunits. DISCUSSION These experiments identify residues in the ⑀ and ␦ subunits that give the two binding sites of the adult mouse AChR different affinities for the curariform antagonist DMT. Previous work identified a different set of residues in the ␥ and ␦ subunits that confer site selectivity in the fetal mouse AChR (1). Two sets of determinants were identified, with each set flanking the disulfide loop common to all members of the AChR superfamily; the pre-disulfide set is Ile ␥116 /Val ␦118 and Tyr ␥117 / Thr ␦119 , and the post-disulfide set is Ser ␥161 /Lys ␦163 . By contrast, selectivity determinants in the adult receptor are far from these in the linear sequence and comprise Ile ⑀58 /His ␦60 and Asp ⑀59 /Ala ␦61 , showing that alternative residues confer DMT selectivity in fetal and adult receptors. The results support a basic scaffold hypothesis because selectivity can be exchanged between the ⑀ and ␦ subunits by exchanging a small number of residues at equivalent positions of the primary sequence.
Of the three pairs of selectivity determinants identified in the fetal receptor, the pair Tyr ␥117 /Thr ␦119 makes the greatest contribution to DMT selectivity (1). Studies of side chain specificity indicate that Tyr ␥117 stabilizes one of two quaternary nitrogens in DMT through a -cation interaction (9). Because serine is present at position 117 of the high affinity ⑀ subunit, we reasoned that the source of high affinity conferred by the ⑀ subunit should be elsewhere. Our results of point mutations of Ser ⑀117 reveal relatively small changes in the affinity of the ␣⑀ site, indicating that although Ser ⑀117 is not the origin of high affinity for DMT, it may nevertheless be close to the site of binding. In contrast to the -cation stabilization observed in the fetal receptor, DMT selectivity appears to owe to electrostatic forces in the adult receptor. High affinity of the ␣⑀ site results from an isoleucine-aspartic acid pair, whereas low affinity of the ␣␦ site results from a histidine-alanine pair.
Mutation of one of the two determinants, ⑀I58H or ⑀D59A, partially decreases affinity, whereas mutation of both determinants fully decreases affinity to that conferred by the native ␦ subunit. The sum of the contributions of the single mutations slightly exceeds that of the double mutation (Fig. 4B), pointing to some sort of interaction between these determinants. High affinity of the ␣⑀ site may owe to electrostatic forces between Asp ⑀61 and one of the two quaternary nitrogens in DMT. If the local pH renders His ␦61 positively charged, low affinity of the ␣␦ site may owe to electrostatic repulsion of a positive charge in DMT. Preliminary measurements with the mutation ⑀I58K reveal decreased affinity similar to that of ⑀I58H. Alternatively, other nearby residues, perhaps with aromatic side chains, may directly stabilize DMT, and the determinants that we have identified may affect their interaction with DMT.
The selectivity determinants in the adult receptor belong to FIG. 5. Sequence alignments of ⑀, ␦, and ␥ subunits in the region containing DMT selectivity determinants. Selectivity determinants are highlighted by enlarged letters. one of the four loops identified to contribute to the non-␣ subunit portion of the ligand-binding interface (12,20). Local sequences of this loop are compared across species for the ⑀, ␦, and ␥ subunits (Fig. 5). The two determinants are highly conserved among ⑀ and ␥ subunits, with the first determinant an invariant isoleucine among ⑀ subunits and methionine in all ␥ subunits except Torpedo. The second determinant is an acidic group in ⑀ subunits and a neutral glutamine in ␥ subunits. By contrast, the determinants are not as well conserved among ␦ subunits, with the proton acceptor histidine or polar glutamine at the first position and small but variable residues at the second.
The selectivity determinants are near other residues known to contribute to the ligand-binding site. The invariant pair Trp ␥55 /Trp ␦57 was first identified by photoaffinity labeling with 3 H-labeled d-tubocurarine (13) and has been shown to contribute to ligand affinity by mutagenesis (14,15). Two residues carboxyl-terminal to Trp ␥55 /Trp ␦57 is the pair Glu ␥57 /Asp ␦59 , which contributes to agonist selectivity in a state-specific manner, preferentially affecting the desensitized state (12). Equivalent to this pair in neuronal subunits is Thr ␤259 /Lys ␤259 , which contributes to higher affinity of dihydro-␤-erythroidine and neuronal bungarotoxin for ␣ 3 ␤ 2 compared with ␣ 3 ␤ 4 receptors (16). Thus, a stretch of five residues, beginning with Trp ␥55 / Trp ␦57 , contributes to the non-␣ subunit portion of the ligandbinding site.
Previous studies showed that tyrosines in the juxtaposed ␣ and ␥ subunits, Tyr ␣198 and Tyr ␥117 , stabilize DMT through symmetrical -cation interactions (9). By studying a series of mutations in both subunits, the contributions of each residue were found to be approximately equal and additive, indicating that DMT bridges the ␣␥ subunit interface. DMT may also bridge the ␣⑀ subunit interface in the adult receptor, but with a different point of attachment in the ⑀ subunit. The determinants we identify may also affect attachment of DMT to the ␣ subunit portion of the binding site.
The high degree of homology among AChR subunits suggests that the polypeptide chains of each subunit fold into similar basic scaffolds. Support for this basic scaffold hypothesis comes from the observation that site selectivity of agonists and antagonists can be exchanged between the ␥ and ␦ subunits by exchanging a small number of residues at equivalent positions of the primary sequence (1,12,17). Nearly interchangeable alterations in affinity are observed for the two pairs of selectivity determinants in the ⑀ and ␦ subunits identified in this study; this would not be expected for a rigid ligand such as DMT unless the determinants are in the same positions at both the ␣⑀ and ␣␦ subunit interfaces. Thus, the identified determinants probably occupy equivalent positions within the ⑀ and ␦ scaffolds, supporting the basic scaffold hypothesis.
In addition to showing that different residues confer high affinity to the ⑀ and ␥ subunits, our experiments reveal an additional difference between these subunits. In particular, neither ␥M58I/␥Q59D nor ␥M58H/␥Q59A affects the contribution of the ␥ subunit to the DMT-binding site, whereas the corresponding mutations in the ⑀ subunit elicit profound changes in DMT affinity. Thus, while the polypeptide scaffolds in the ␥ and ⑀ subunits are likely to be similar, particularly since they are the most homologous pair of AChR subunits, a structure unique to the ␥ subunit prevents residues placed at positions 58 and 59 of the ␥ subunit from contributing to DMT binding.
One can reasonably ask how two completely different sets of residues confer selectivity for the same ligand. Considering first the low affinity ␣␦ site, DMT binds with a dissociation constant of 10 M, which of course requires some stabilizing interactions. Perhaps these include stabilization of one of the two quaternary nitrogens in DMT by Tyr 198 in the juxtaposed ␣ subunit (14,18) plus stabilization of the hydrophobic and hydrophilic faces of DMT by as yet unidentified residues within the ␣␦ subunit interface. Considering the high affinity ␣␥ site, DMT is still stabilized by structures common to the ␣␦ and ␣␥ sites, but its second quaternary nitrogen is within reach of Tyr ␥117 , which provides additional stabilization and high affinity. The high affinity ␣⑀ site, on the other hand, contains serine at position 117, which is within reach of the second quaternary nitrogen of DMT, but provides no further stabilization. Instead, DMT appears to position its second quaternary nitrogen close to Ile ⑀58 and Asp ⑀59 to bind with high affinity and a dissociation constant of ϳ0.1 M. Thus, at the various ␣␥, ␣␦, and ␣⑀ binding sites, DMT chooses among possible sources of stabilization within the binding cleft and associates with the most favorable stabilizing residues.