Competitive Antagonists Bridge the ex-y Subunit Interface of the ...

T H E JOURNALOF BIOLOGICAL CHEMISTRY 0 1994 by The American Society for Biochemistry and Molecular Biology, Inc

Vol 269, No. 42, Issue of October 21, pp. 26152-26157,1994 Printed in U.S.A.

Competitive Antagonists Bridge theex-y Subunit Interface of the Acetylcholine Receptor through Quaternary Ammonium-Aromatic Interactions* (Received for publication, May 20, 1994, and in revised form, August 8,1994)

Da-Xiong Fu and StevenM. Sine$ From the Receptor Biology Laboratory, Department of Physiology and Biophysics, Mayo Foundation, Rochester, Minnesota 55905

We recently demonstrated that conserved tyrosines with the a-y site showing high affinity and the a-8site showing of the a subunit andTyr117 ofthe y subunit of the lowaffinity. One of these residues, a conservedtyrosine at acetylcholine receptor stabilize binding of the curari- position 117 of y, accounts in part forhigh affinity of the a-y site form antagonist dimethyl-&tubocurarine (DMT).To test (lo),and therefore may stabilizethe second quaternary group the hypothesis that DMT interacts directly with these in DMT. tyrosines, and therefore bridges the a - y subunit interUsing cryoelectron microscopy, Unwin (11) recently obtained face, we introduced point mutationsinto these key po- electron density mapsof the Torpedo AChR toa resolution of 9 sitions and expressed one or both mutant subunits in A. In the density maps, the two ligand binding sites appear as a2@'y2 acetylcholine receptors in 293 HEK cells. Binding gorges centered primarily within the (Y subunit, but within of DMT, measuredby competition againstthe initial rate 10-15 A of the neighboring y or 6 subunit. Because DMT is a of '2SI-a-bungarotoxin binding,shows high affinity for conformationally restric!ed ligand with two quaternary nitroaromatic mutations, reduced affinity for polar mutagens separated by 10.8 A (121, demonstrating tyrosine interactions, and lowest affinity for arginine mutations. Similar tions with these charged moeities would establish points of side chain dependences were observed for bothTyFnlS8 close approach between defined positionsin the two subunits. and Tyry'", indicating interaction of these residues with Thus, in the present study, we examined the side chain dependtwo symmetrical chemical groups in DMT. T w o more bis- ence of DMT affinity for positions198 of t h e a subunit and 117 quaternaryantagonists,pancuroniumandgallamine, of the y subunit. We also examined side chain dependences for show side chain dependences similar to that of DMT, pancuronium and gallamine, ligands that maintain the same indicating that the primary stabilizing interactions are spatial separationof quaternary groups but contain essentially aromatic-quaternary in both subunits. For the rigid li- different scaffolds. Finally, we co-expressed mutant a a n d y gands DMT and pancuronium, co-expressingmutant a subunits to test for independence of the contributions by the and y subunits revealed independent contributions by binding determinant in each subunit. each determinant, but strict independence was not obEXPERIMENTALPROCEDURES served forthe flexible ligand gallamine.The free energy Materials-DMT was generously provided by the Eli Lilly Co. Galcontributed by each aromatic-quaternaryinteraction was estimatedto be 2-4 kcaYmo1, as determined fromthe lamine and pancuronium were purchased from Sigma; '251-labeled a-bungarotoxin was purchased from DuPont NEN. The 293 human free energy difference between aromatic and alkyl hy- embryonic kidneyfibroblast cell line (293 HEK) was obtained from the droxyl mutations. Our results suggest that bis-quater- American Type Culture Collection. Mouse AChR subunit cDNAs were nary competitiveantagonists bridge the a-ysubunit in- generously provided by Drs. John Merlie and Norman Davidson (for terface by fitting into a pocket bounded by tyrosines at cDNA sequences, see Ref. 10). positions 198 of the a subunit and117 of the y subunit. Mutagenesis-Mouse AChR subunit cDNAs were subclonedinto the cytomegalovirus-basedexpressionvector, pRBG4, as described (10). Mutations in the a and y subunits were constructed by bridging naturally occurring or mutagenically installed restriction sites with synThe acetylcholine receptor (AChR)l from vertebrate skeletal thetic double-stranded oligonucleotides(9, 10).All constructs were conmuscle contains twoligand binding sites withinan a,py8 pen- firmed by restriction mapping and dideoxy sequencing. Expression of a2pyz Surface AChRs Containing Mutations in the a tamer. Affinity labeling ( 1 4 ) and expression of different subthat the two bindingsites and y Subunits-HEK cells at about 75%confluencewere co-transunit combinations (7,s) have shown fected with wild type or mutant a, p, and y subunit cDNAs, each in are formed by a-y, the a-8 subunit pairs. In the (Y subunit, pRBG4, by calcium phosphate precipitation as described (9, 10).Cells conserved tyrosines at positions 190 and 198 stabilize binding were incubated in the calcium phosphate DNA medium for 5-10 h, of the curariform antagonist dimethyl-d-tubocurarine (DMT) changed backto normal growth medium, and incubated at 37 "C for 24 h and at 31"C for another 48 h. (9). Because these tyrosines stabilize the elementary quaterLigand Binding Measurements-Following incubation at 31"C, cells nary ammonium ligand TMA (91, they may also stabilize DMT through interaction with one of its two quaternary ammonium were harvested by gentle agitation in phosphate-buffered saline containing 5mM EDTA, briefly centrifuged, resuspended in high potassium groups. In t h e y a n d 8 subunits, differences in three homolo- Ringer's solution, and divided into aliquots for ligand binding measureat each site (lo), ments (9). Specified concentrations of ligands were added 30 min prior gous residues produce different DMT affinity to the addition of '251-labeled a-bungarotoxin. To measure the initial rate of binding, toxin was allowed to bind for 30 min to occupy at most * This work was supported in part by National Institutes of Health half of the surface receptors. The total number of binding sites was Grant NS31744 (to S. M. S.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must determined by incubating with toxin for 120 min. Unbound toxin was therefore be hereby marked "advertisement" in accordance with 18 removed by washing twice with potassium Ringer's solution containing 600 p d-tubocurarine followed by centrifugation, and theradioactivity U.S.C. Section 1734 solely to indicate this fact. bound to the cell pellet was measured with a y counter. Nonspecific $. To whom correspondence should be addressed. The abbreviations used are: AChR, acetylcholine receptor; DMT, binding was determined in thepresence of saturating concentrations of the ligand under study. dimethyl-&tubocurarine.

26152

Antagonists Bridge

Rates of toxin binding in the presence and absence of competing ligand were calculated from binding measured at 30 and 120 min as described (13). These rates are related to ligand occupancy by, kabs = fa,,,=.(l - Y),where k,, is the rateof toxin binding in the presence of a specified concentration of competing ligand, k, is the rate in the absence of competing ligand, and Y is the occupancy function for the competing ligand (in these studies given by the Hill equation). Differences in the intrinsic rate of toxin binding ( i e . fa,,,,) for the various mutant AChRs donot affect Y because it depends on the ratio of the two rates. The Hill equation was fitted by least squares to the competition data to yield the apparentdissociationconstant (K,)and Hill coefficient (13). The standard errorof the Kd was calculated from the least squares fit using SigmaPlot (Jandel Scientific, Inc.) and is the error estimate given in Tables I-Iv. RESULTS

Side Chain Dependences of DMT Binding Determinants in the y Subunit-Differences in threehomologous residues in the y and 6 subunits confer selectivity of DMT for the two binding sites in the AChR (10). One of these residues is a conserved tyrosine at position 117 ofthe y subunit. Because DMT binds to the a-y site with high affinity, T y r n 7 is a good candidate for interaction with one of the two quaternary nitrogens in DMT. We therefore examined the chemical nature of the interaction between DMT and TyrYll’l by constructing point mutations at this position. Mutant subunits were co-expressedwith nonmutant a and p subunits in293 HEKcells, and DMT binding was measured to cell surface a2py2AChRs (8).Fig. lA shows that mutations at position 117 ofthe y subunit alterDMT affinity by

A

1,2

1.o

1.o

0.8

0.8

-

0.6

>;

7

0.4

0.6

>; 0.4

0.2

0.2

0.0

0.0

0.2

le-08 l e 4 7 le-06 l e 4 5 le-04 l e 4 3 le-02 le-10

( DMT I

C le44

-

le-05

-

lea

-

le07

-

le48

-

26153

up to 1000-fold. Aromaticmutations produce high affinity, alkyl hydroxyls produce low affinity, and the positively charged arginine produces the lowest affinity (Fig. 1C). The increased affinity produced by the phenylalanine and tryptophan mutations is not purely hydrophobicin naturebecause the isoleucine mutation produces low affinity. Introducing a negative charge, as in the glutamate mutation, still yields micromolar affinity, but the affinity is lower than with any of the aromatic mutations. The low affinity produced by arginine is consistent with repulsion between the positively charged guanidinium side chain and a quaternary nitrogen group in DMT. Thus aromaticity at position 117 of the y subunit produces high affinity DMT binding. The overall side chain dependences are consistent with quaternary-aromatic (14), aromatic-aromatic (151, or both types of interactions with DMT. We also examined side chain dependences of the two remaining DMT determinants in the y subunit, Ile116and SerlZ6.We previously demonstrated that placing lysine at position 126 of the y subunit decreased DMT affinity, consistent with the low affinity of the 6 subunit, which contains lysine in the homologous position, and with possible repulsion of a quaternary group in DMT (10). However, in contrast to the greater than 1000-foldchanges in affinity seen at position 117, only1.1-5.6fold changes in affinity were seen among mutations at position 126 (TableI). We also examinedmutations at position 116,and again found onlysmall changes in DMT affinity (Table I). Thus, Ile116and Se? of the y subunit appear to affect DMT affinity

B

1.2

4.2 le-10 le-09

AChR

a-y Subunit Interface of

le49 le-@ le-07

le% 1 6 0l5e 4 4

le-03 le02

( DMT I

W F Y E S I T R F Y H C E S T R FIG.1. Side chain dependence of DM” binding to surface azP-yzAChRa. Panels A and C show results for the indicated substitutions at position 117 of the y subunit, and panels B and D show results for substitutions at position 198 of the a subunit. a2py2AChRs containing the indicated mutations were expressed in 293 HEK cells,and DMT binding was measured as described under “Experimental Procedures.” The curves are least squares fits to the Hill equation with the dissociation constants plotted in panels C and D to illustrate side chain dependences. Hill coeffkients ranged between 0.85 and 1.05 among the various mutations.

a-y Subunit Interface of AChR

Antagonists Bridge

26154

through indirect interactions, whereas T y r 1 1 7 is a good candiFig. 2A shows how mutations a t position 117 of the y subunit date for direct interaction with DMT. affect binding of pancuronium and gallamine. For both ligands, Side Chain Dependences of DMT Binding Determinants in the side chain dependences are very similar to those of DMT; the a Subunit-We previously demonstrated that aromatic mu-aromatic mutations produce high affinity, alkyl hydroxyls protations at position 198 of the a subunit increase DMT affinity duce low affinity, and thepositively charged arginine produces for azpy6AChRs (9). To further examine the chemical nature of the lowest affinity. Mutations at position 198 of the a subunit interaction between DMT and TYY”~,we studied DMT binding show remarkably similar side chain dependences t o position to symmetrical AChRs containing mutant a subunits 117 of the y subunit (Fig. 2 B ) , as was observed for DMT.To (Fig. 1B).Similar toposition 117 of the y subunit, mutationsat quantitatively compare the effects of the mutations on the position 198 of the a subunit alterDMT affinity by greater than three ligands, we plot the dissociation constant of DMT against 1000-fold. Aromatic mutations produce high DMT affinity, that of pancuronium or gallamine for both the a and the y alkyl hydroxyls produce low affinity, and thepositively charged subunit mutants (Fig. 3). For mutations in either subunit, the arginine produces the lowest affinity (Fig. lD). Thus in both dissociation constant of DMT is correlated with that of either the a and y subunits, thekey DMT binding determinants show pancuronium or gallamine.Thesecorrelations indicate that similar side chain dependences, implyinginteraction with sym- within these structurallydifferent ligands, the essential pharmetrical counterparts inDMT. macophore is a pair of quaternary ammonium groups. Side Chain Dependences for Pancuronium and GallamineAlthough the overall panel of mutations affects the three Two observations suggest that the symmetrical counterparts in ligands approximately equally, the aromatic mutations reveal DMT are its two quaternary nitrogens. First, ’ Q d g 8 of the a detectable differences among the ligands. The phenylalanine subunit stabilizes binding of the elementary quaternary am- and tryptophan mutationsdo not enhancepancuronium or galmonium ligand TMA (9). Second, the side chain dependences of lamine binding, in contrast to their enhancement of DMT bindpositions y 117 and a 198 are consistentwith aromatic-quater- ing (Fig. 3). Also, a Y198H confers high DMT affinity but low nary interactions (14). However, the side chain dependences gallamine and pancuronium affinity. Becausepancuronium are also consistent with aromatic-aromatic interactions (151, lacks aromaticity and thesingle benzene ring of gallamine lies which is possible because DMT contains four symmetrically between and distantfrom the quaternarygroups, these results disposed aromatic groups. suggest thataromatic-aromatic interactions produce some staTo determine thechemical group inDMT that interacts with bilization of DMT binding. the conserved tyrosines, we studied two more bis-quaternary Independence of the Binding Determinants-Although our ligands,pancuroniumand gallamine. Both ligands contain mutations alter free energy of ligand binding by an amount quaternary nitrogens separated by 10.8 A but have molecular consistent with direct quaternary-aromatic interactions (see scaffolds distinct from that of DMT (12); pancuronium has a “Discussion”), the results to this point cannot exclude two alsaturated steroidframework, whereas gallamine has a central ternative mechanisms. First, the conserved tyrosines may be aromatic ringwith quaternary groups attached toflexible alkyl remote from the binding site but allosterically affect the site. side chains. Second, the mutationsmay affect binding by allowing reorientation of the ligand in the binding pocket. In both of these TABLEI alternatives, combining mutant a and y subunits intoa single Effect of mutations at positions y126 and y116 on DMT binding AChR would not be expected to result in independent contriKd mutJK, wt Mutant Kd butions to binding energy for a wide range of double subunit PM mutant AChRs. In contrast, a direct interaction mechanism 0.2 f 0.004 Wild type 1.0 would be expected to result in independent contributions in 1.1 f 0.027 5.6 S126K double mutant AChRs. To distinguish among these possibili2.8 S126D 0.57 f 0.075 ties, we co-expressed mutant ct and y subunits and compared 0.33 f 0.011 1.6 S126Y the resulting change in binding free energy to the sumof free 0.23 f 0.016 S126I 1.1 0.7 I116V 0.14 f 0.022 energy changesfor single subunit mutants (Tables 11-IV). 0.28 I116A & 0.019 1.4 Fig. 4A shows freeenergy of DMT binding for double subunit

n

pancuroniurn

0 Gallamine

I

nfl

Y

W

F

I

E

S

T

R

Y

F

C

E

S

H

T

R

FIG.2. Side chain dependence of pancuronium and gallamine binding to surface AChRs. Panel A shows apparent dissociation constants (K,) for the indicated substitutions at position 117 of the y subunit, andpanel B shows dissociationconstants for substitutions at position 198 of the (Y subunit.

Antagonists Bridge a-y~ u ~ u nInterface it of AChR

FIG.3. Correlationbetween DMT bindingandpancuroniumandgallamine bindingfor mutations at position 117 of the y subunit (panel A) and position 198 of the (Y subunit (panel B ) . Dissociation constants for the three ligands are taken from Figs. 1 and 2. The straight lines were obtained by linear regression analysis, and indicate statistically significant correlations for all three ligands for the mutations at these two positions. R values for y 117 are 0.89 for ~ a l l ~ i n e l ? ) M Tand 0.72 for ~ancu~nium/DMT. R values for cy 198 are 0.91 forgaIlaminel?)MT and 0.84 for pancuronium/DMT.

0

le03

26155 0 Gallamine

Gallamine

/

Pancuronium

Pancuronium

1004

x

l

a

9 le06 lei17

lea

1aOP

1-08

le-07

lea

1%05

le03

I&

KdDMT

B

A

Pancuronium

DMT

Gollomlne

5 4

3 P -

2 1 0 I

2 3

o -

0

0

1

o

4

4

.

2

0

2

4

6

MG,

( kcal/rnole )

B

mutants plotted against the sum of free energy changes for single subunit mutants. In these experiments, we co-expressed as many mutants aspossible with a Y198F, a Y198S, y Y117F, and y Y117S. Over a range of 7 kcal/mol of DMT binding free energy, the relationship between observed and summed free energies is linear with a slope of 0.95 and an intercept close to zero, indicating independent contributions by the a and y binding determinants to DMT affinity We tested independence with the other rigid ligand pancuronium and again found a linear relationship with a slope of 0.85 (Fig. 4B). By contrast, the flexible ligand gallamine, although showing a roughly linear relationship, does not show strictly independent contributions by the two binding determinants; the best fit line has a slope of 0.67 (Fig. 4C). The gallamine data also show considerable scatter about the fitted line, with a correlation coefficient of 0.79, in contrast to values of 0.97 for DMT and pancuronium. Gallamine, owing to its flexibility, appears to seek out slightly different orientations in the various mutant AChRs. For the rigid ligands DMT and pancuronium, determinants in the two subunits contribute independently, supporting direct aromaticquaternary cont~butionsto ligand afEnity. E#'& of a and y Subunit M u ~ Q on ~ ~ a-Bungarotoxin s Binding-In the course of these studies, we obtained intrinsic association rates (k,,,=) of a-bungarotoxin binding for the various point mutants, as well as total number of binding sites (RmJ.Although k,, and R,, vaned among the mutants, neither parameter depended systematically on side chain chemistry of the mutation. Expressed relative to wild type subunits,

2

3

4

5

6

AAGsu,,, ( kcoil mole)

1

FIG.4. Comparison of free energy changes for double subunit mutant AChRa with the sum of free energy changes for single subunit mutants. Mutant cy and y subunits wereco-expressed with the p subunit, and apparent dissociation constants of the resulting surface AChRs were measured as in Fig. 1. The calculated free energy of binding fordouble subunitmutantsis plotted against the sum of free energy changes for the corresponding singlesubunit mutants given in Figs. 1 and 2. The solid lines are linear regression fits to the points. See Tables 11-IV for definitions of symbols. The fitted slopes and intercepts are 0.95 and -0.1, 0.85, and 0.3, and 0.67 and 0.3 for DMT, pancuronium, and gallamine, respectively.

C

k,,, for the F, E, and R mutations was 1.0,1.2, and 1.1for a198 and 0.8,1.2, and 0.9 for y117. R,,, for the F, E, and R mutations was 1.0, 1.1, and 1.9 for a198 and 1.9, 1.1,and 0.8 for y117. Thus the F, E, and R mutations, which as a set span the entire range of antagonist affinities, do not similarly affect parameters of a-bungarotoxin binding. Because binding of a-bungarotoxin likely depends on multiple points of contact at the binding interface, the point mutations studied here do not appear to affect overall structure of the binding pocket. DISCUSSION

The experiments described herein characterize potential docking sites for bis-quaternary antagonists on the a and y subunits of the AChR. The conserved tyrosines, T~TY''~ and T Y f a 1 % , satisfy several criteria for direct contributions to his-quaternary antagonist binding. The first isside chain dependences in both subunits consistent with aromatic-quaternary ammonium interactions; aromatic mutations produce high affinity, polarmutations decrease aEnity, and the positively charged arginine markedly decreases affinity. The second is similar side chain dependences for ligands that have different molecular scaffolds but maintain the same spatial separation of quaternary nitrogens. The third is independent contributions t o ligand affinity by the key binding determinant in each subunit. The two conformationally restricted ligands DNT and pancuronium clearly show independent contributions for a wide range of mutations, whereas the flexible ligand gallamine does not show strict independence. The overall results strongly support the idea that

26156

Antagonists Bridge a-y S Ixbunit Interface of AChR TABLEI1

Binding free energy of DMT for double subunit mutant AChRs

TABLEIV Binding free energy ofgallamine indouble subunit mutant AChRs

UM

aY198F

0

w F E I S T

R aY198S

w E

I

0

S

T R

C V

E T R

vYll7S

v

0.0010 f 0.0005 0.0006 f 0.0007 0.0640f 0.0044 0.0200f 0.0019 0.0086 f 0.0026 0.0800& 0.0059 0.2500& 0.0058

-3.1 -3.4 -0.67 -1.4 -1.9 -0.54 0.13

-3.1 -2.9 -0.89 -1.5 -1.4 -0.46 0.99

0.66 f 0.096 3.40 f 0.52 13.0f 0.69 16.0f 1.5 8.40f 0.90 24.0 2 1.2 24046

0.69 1.7 2.5 2.6 2.2 2.8 4.2

0.95 1.2 2.5 3.2 2.6 3.6 5.0

0.56f 0.04 0.45f 0.04 14.0 2 0.29 1.50f 0.09 88.06.7

0.61 0.48 2.5 1.2 3.6

0.42 0.53 1.4 1.3 3.0

0.86 f 0.04 2.90f 0.25 12.0 f 1.1 6.502 0.65 180 f 14

0.86 1.6 2.4 2.1 4.0

0.97 1.9 2.8 2.7 4.3

a-Mutant €I C

E T R

TABLE I11 Binding free energy of Dancuronium in double subunit mutant AChRs

PM

a198C

0 a198F 0

a198T V

?-Mutant

I R

F E I y-Mutant F

w E R

a198S

v

15 2 1.0 4.2 250 6.1 5.9 f 21

4.2

y-Mutant

w

0.047 f 0.013 0.78 0.80 0.077 2 0.027 2.4 1.20 f 0.22 0.240f 0.10 4.920 4.4 f 1.7 23.2 .8 * 0.12 6.1 360 f 36 880 f 64

1.1 2.7 1.8

0.99

1.3

4.7 6.3 7.1 6.6

y-Mutant

T

a198F

y-Mutant F

0

6.1 5.3 88 f 14

bis-quaternary antagonists bridge the a-y subunit interface through aromatic-quaternary ammonium interactions. Further support for direct aromatic-quaternary interactions comes from estimates of binding free energy contributed by each binding determinant. For DMT, the free energy difference between the best aromatic substitution and the arginine substitution is 4.1 and 5.8 kcal/mol for the y and a determinants, respectively. These estimates are upper bounds because they include repulsive forces due to the arginine side chain. A suitable neutral substitution might be threonine, which lacks aromaticity but retains the polar hydroxyl.Free energy differences between threonine and aromatic substitutions are 2.8 and 4.1 kcal/mol forDMT and 2.6 and 3.9 kcal/mol forpancuronium for the y and a determinants, respectively. These values are in the range of values derived theoretically for benzene-TMAinterac-

I R

a198C

a-Mutant

H

0

y-Mutant F R

a198T

y-Mutant

F

yY117F

UM

y-Mutant

V

y-Mutant W S

T

v

?-Mutant R

a198R 0

y-Mutant

a198E

R

160 -c 153.4 1800 f 3035.5

2.8 4.2

1.5 f 0.4 0.0 1.3f 0.14 1.5 -0.08 115 2 173.0 2.6

0.9

4.2 f 0.53 2.7 0.61 50 f 3.7 3.4 2.1 23 f 1.44.4 1.6 5800 f 601 4.3

4.9

8000 f 1073 6.6

5.1

tions (1.8 kcal/mol) (16) and for amino-aromatic interactions (3.6 kcal/mol) (17, 18). We can estimate the fraction of binding free energy contributed by the two tyrosines at the wild type a-yinterface from the free energy difference betweenthe threonine and tyrosine substitutions, summed over both subunits. This estimate is tenative, however, because the appropriate neutral substitution is uncertain. The estimated tyrosine contributions are 3.8 kcal/ mol for DMT, 6.5 kcal/mol for pancuronium, and 5 kcal/mol for 65%of the total binding gallamine, correspondingto 41,59, and free energy, respectively.These percentages are inrough accord with hydrophobicity of the ligand, suggesting that hydrophobic stabilization of the ligand scaffold providesan additional source of binding free energy. This hydrophobic contribution may prevent reorientation of DMT and pancuronium in our double subunit mutant AChRs, which exhibit sumable free energies over a wide range. Nearby aromatic side chains are another possible source of binding free energy; T y r l g o of the a subunit contributes to DMT affinity, although its contribution differs from that of position 198 in that phenylalanine and tryptophan substitutions do not enhance affinity (9).Aconserved tryptophan is present at position 118of the y subunit, which in the proper orientation could also contribute to DMT binding. Binding of bis-quaternary ligands is mechanistically similar in theAChR and acetylcholine esterase. In acetylcholine esterase, the pathway to the active site is a long aromatic gorge penetrating deep into the enzyme (19). Radic et al. (20) have demonstrated thatthe bis-quaternary ligand BW284C51 bridges two clusters of aromatic residues in the active site gorge; onecluster is near thecatalytic site at thebottom of the gorge, and the other is near themouth of the gorge. The outer cluster contains tryptophan 279 and tyrosines 70 and 124. Replacing any oneof these with polar or charged residues decreases binding free energy by 1.0-2.7 kcal/mol, similar to the changes we observe at either docking site in the AChR. Also, the crystal structure of decamethonium bound to acetylcholine esterase shows onequaternary group apposed to tryptophan 84 at the bottom of the gorge, and the other apposed totryptophan 279 (21).The binding domains in these two proteins, nevertheless, show fundamental differences, as they are confined t o a single protein subunit inacetylcholine esterase butare present on different subunits in theAChR. Aromatic-aromatic interactions also appear important for DMT binding because at both positions a 198 and y 117, phenylalanine and tryptophan substitutions enhance binding of

Antagonists Bridge a-y Subunit Interface of AChR DMT, which contains aromatic rings, but do not enhance binding of pancuronium, which lacks aromaticity. In these mutants, the aromatic side chain may interact with both quaternary and aromatic groups in DMT. These observations are consistent with those of Filatov et al. (22) who showed that a Y198Fenhanced block of acetylcholine-induced currents by curare, but not block by pancuronium or gallamine. In the wild type AChR, aromatic-aromatic contributions appear small because alkyl hydroxyl substitutions decrease pancuronium and DMT binding affinities by similar amounts. The electron density profiles of the Torpedo AChR show that each a subunit contains three rods of density typical of a helices (11);these rods definethe innerboundaries of a cavity proposed to contain the ligand binding pocket. Our results indicate that bis-quaternary ligands associate with residues in the cavity through three roughly equivalent but distinct interactions. The first is interaction between one quaternary group of the antagonist and Tyr198 of the (Y subunit; we hypothesize that T y r l g 8 is close to or within the triadof helices. The second is interaction between the second quaternary group, 10.8 A away from the first, and Tyr117 of the y subunit. Jncluding optimal aromaticquaternaryseparations of 3-5 A, the two intersubunittyrosines are predicted to be about 18 A apart when ligand is bound. This intertyrosine distance is consistent with the apparent intersubunit distances in the electron density profiles (11)and thus implies that T y r Y l l ' l lies outside of the helical triad. The third site of interaction remains unknown but is

26157

expected to be hydrophobic in natureand to lie between the two aromatic docking sites. REFERENCES 1. Kao, P., Dwork, A., Kaldany, R., Silver, M., Wideman, J., Stein, S. & Karlin,A. (1986)J. Biol. Chem. 269,286-287 2. Abramson, S., Li, Y., Culver, P. & Taylor, P. (1989)J. Biol. Chem. 264, 1266612672 3. Galzi, J.-L., Revah, F., Black, D., Goeldner, M., Hirth, C. & Changew, J.-P. (1990)J. Biol. Chem. 266, 10430-10437 4. Cohen, J. B., Sharp, S. D. & Liu, W. S. (1991)J.B i d . Chem. 266,23354-23364 5. Middleton, R. E. & Cohen, J. B. (1991)Biochemistry 30, 6987-6997 6. Pedersen, S. & Cohen, J. (1990)Proc. Natl. Acad. Sci. U. S. A . 87, 2785-2789 7. Blount, P. & Merlie, J. (1990)Neuron 3, 349-357 8. Sine, S. M. & Claudio, T. (1991)J. B i d . Chem. 266, 19369-19377 9. Sine, S. M., Quiram,P., Papanikolaou, F., Kreienkamp, H.-J. & Taylor, P. (1994)J . Biol. Chem. 269,8808-8816 10. Sine, S. M. (1993)Proc. Natl. Acud. Sei. U. S. A. 90, 9436-9440 11. Unwin, N.(1993)J. Mol. Biol. 229, 1101-1124 12. Pauling, P. & Petcher, T. J. (1073)Chem. Bid. Interact. 6, 351-365 13. Sine, S. M. & Taylor, P. (1981)J. Biol. Chem. 266, 6692-6699 14. Dougherty, D.A. & Stauffer, D. A. (1990)Science 250, 1558-1560 15. Burley, S. K. & Petsko, G. A. (1988)Adu. Protein Chem. 39, 125-189 16. Gao, J., Chou, L. W.& Auerbach, A. (1993)Biophys. J . 66, 43-47 17. Levitt, M. & Perutz, M. F. (1988)J. Mol. Biol. 201, 751-754 18. Lau, F. & Karplua, M. (1994)J. Mol. B i d . 236, 1049-1066 19. Sussman, J. L., Harel, M., Frolow, F., Oefner, C., Goldman, A,, lbker, L. & Silman, I. (1991)Science 263, 872-879 20. Radic, Z., Pickering, N., Vellom, D., Camp, S. & Taylor, P. (1993)Biochemistry 32, 12074-12084 21. Harel, M., Schalk, I., Ehret-Sabatier, L., Bouet, E, Goeldner, M., Hirtb, C., Axelsen, P. H., Silman, I. & Sussman, J. L. (1993)Proc. Natl. Acad. Sci. L'. S. A. 90, 9031-9035 Aylwin, M. L. & White, M. M. (1993)Mol. Pharmucol. 44, 22. Filatov, G. N., 237-241