PROTEINS Structure, Function, and Genetics 8:386-397 (1990)

Modeling of Agonist Binding to the Ligand-Gated Ion Channel Superfamily of Receptors Victor B. Co~kcroft,'.~ David J. Osguthorpe,' Eric A. Barnard: and George G. Lune 'Molecular Graphics Unit, Bath University, Bath BA2 7AY, UK, 2Molecular Neurobiology Unit, MRC Centre, Cambridge CB2 2QH, UK, and 3Biochemist~ Department, Bath University, Bath BA2 7AY, UK

ABSTRACT A generalized model is presented of agonist binding to ligand-gated ion channels (LGICs).Broad similarity in the structure of agonists suggests that the binding sites of LGICs may have evolved from a protobinding site. Aligned sequence data identified as a candidate for such a site a highly conserved 15 residue stretch of primary structure in the Nterminal extracellular region of all known LGIC subunits. We modeled this subregion, termed the cys-loop, as a rigid, amphiphilic phairpin and propose that it may form a major determinant of a conserved structural binding cleft. In the model of the binding complex (1) an invariant aspartate residue at position 11of the cys-loop is the anionic site interacting with the positively charged amine group of agonists, (2) a local dipole within the =-electron system of agonists is favorably oriented in the electrostatic field of the invariant aspartate, (3) the E ring-proton of a conserved aromatic residue at the turn of the cys-loop interacts orthogonally with the agonist =-electron density at its electronegative center, and (4) selective recognition is partly a result of the type of amino acid residue at position 6 of the cys-loop. Additionally, the formation of a hydrogen bond between the electronegative atom of the n-electron system of agonist and a complementary group in the receptor may be important in the high-affinity binding of agonists. Key words: LGIC superfamily, proto-binding site, cys-loopmotif, docking model, anionic site, specificity residue INTRODUCTION Ligand-gated ion channels (LGICs), a superfamily of membrane proteins, mediate fast neural signaling. In response to the binding of agonist they permit a flux of ions across the membrane through an ion channel intrinsic to their structure. On the basis of analysis of primary structure the superfamily presently encompasses the receptors for y-aminobutyric acid (of the GABA, type),' glycine? and acetylcholine (with a nicotinic pharmacology)? In addition, electrophysiological evidence suggests that 0

1990 WILEY-LISS, INC.

receptor subtypes for glutamate: the 5-HT3 serotonin receptor,' and a histamine receptor of the arthropod visual system6 may also be members of the LGIC superfamily. Recently, a subunit of the kainate type of glutamate receptor in rat was cloned,' as well as kainate-binding proteins from chicken and Analysis of these sequences showed them to be clearly homologous to each other but not sufficiently similar to the LGIC sequences to establish a definite evolutionary link. The nicotinic acetylcholine (nACh) receptor from Torpedo electric organ was the first LGIC to be studied a t the molecular l e ~ e lStoichiometric .~~~ analysis by N-terminal sequencing indicates there to be two a-subunits and one p-, y-, and &subunit per oligomer." These subunits are homologous and are arranged to give a barrel shape with a cylindrical extracellular funnel and a centrally located ion channel." Studies to localize sites involved in ligand binding have focused attention on two regions of the extracellular portion of the Torpedo nACh receptor. The first is the region around the paired cysteines 192-193 of the a-subunit. Various affinity ligands, including (4-N-ma1eimide)benzyltrimethylammonium and bromoacetylcholine," label these cysteine residues after reduction of the disulfide bridge between them. In addition, the surrounding sequence from position 185 to 196 has been shown to bind a-bungar0to~in.l~ The second region is from position 125 to 147 of the a-subunit. A synthetic peptide to this region was shown to interact with acetylcholine and a-b~ngarot0xin.l~ This peptide contains a highly conserved 15 residue stretch of sequence termed the cys-loop, so called because a disulfide bridge links cysteine residues at positions 128 and 142.l' More recently, Madhok et al.I5 have reported that for the brain nACh receptor high-affinity binding of nicotine is specifically inhibited by antibodies raised to a peptide covering positions 312 of the cys-loop of neuronal a-subunits. We have used a comparative molecular modeling

Received October 2, 1989;revision accepted June 6,1990. Address reprint requests to Dr. George G. Lunt, Biochemistry Department, The University of Bath, Chaverton Down, Bath, BA2 YAY, England.

AGONIST BINDING TO LIGAND-GATED ION CHANNELS

approach making use of aligned amino acid sequences of LGIC subunits, of which over 40 are now known, to identify residues that may be of importance in ligand binding. The region on which we have concentrated is the cys-loop, which is highly conserved in all LGIC subunit sequences characterized to date. Some earlier proposals have been made16*17about the structure of this region in the a-subunit of the Torpedo nACh receptor as a possible agonist binding site, but these did not include the mutual pairing of cysteines 128-142. With the establishment of the concept of a LGIC superfamily we have been able to look beyond the nACh receptor and take into account the extensive biochemical, pharmacological, and biophysical information on all known members of the receptor superfamily.

MATERIALS AND METHODS The programs MOLEDT and INSIGHT from Biosym were used to build and view structures, respectively. The valence force field with the potential parameters of Dauber-Osguthorpe et al." was used in molecular mechanics calculations. Energy minimization was performed to a final maximum derivative of 0.05 kcal mol-' k' using conjugate gradient optimization afler initial minimization to 2.0 kcal mol-' k' using steepest descent optimization. Conformational searching using molecular dynamics was performed using the program DISCOVER from Biosym using a step size of 1 fsec and sampling of atomic coordinates of the trajectory every 4 fsec. Analysis of the displacement of atomic coordinates from the average coordinate set of the trajectory histories was done using the program FOCUS." The Birkbeck Integrated Protein Engineering Database (BIPED)20 implemented a t the Daresbury Laboratories, Warrington UK was used interactively to list details of the protein structures of the Brookhaven Database.

RESULTS AND DISCUSSION Structural Similarity of LGIC Agonists The structural and conformational requirements for agonist binding to the receptors for acetylcholineF1 GABA," and glycineZ3based on QSAR studies of each of them separately have been extensively covered in the literature. Here we propose a more unified pharmacophore model for LGIC receptor agonists as a class. Examples of agonists of LGICs and a schematic representation of proposed common features are shown in Figure l a and b, respectively. From an analysis of these structures, we propose the following basic structural requirements for their agonist activity: (1)a positively charged center (termed the "positive pole")-this group is essential, (2) a n-electron system containing an sp2hybridized electronegative center that produces a local dipole in the nelectron system. The distance between the nitrogen

387

atom of the positive pole and the electronegative atom is 4.5-5.5 A for acetylcholine and GABA ligands, whereas for glycine the distance is around 3.5 A. This broad similarity leads us t o suggest that the agonist binding sites of individual LGIC receptor types may have evolved from a proto-binding site. Both of the above features occur in glutamate, histamine, and serotonin, which is consistent with the tentative assignment of at least some receptor subtypes for these neurotransmitter^^^^,^ to the LGIC superfamily. To assess the significance of the proposed similarities, agonists recognized by receptors that link to second-messenger systems were also considered. The endogenous ligands of this receptor superfamily are numerous and are structurally diverse compared with those of the LGIC superfamily. They include CAMP:* retin01,'~ and substance K,26 as well as a~etylcholine,2~ the c a t e ~ h o l a r n i n e s , ~and '~~~ serot ~ n i n . ~It' can be noted that the absence of a nelectron system in muscarine is an indication that the requirements of agonist binding to second-messenger linked receptors are not identical to those of LGIC receptors. However, the catecholamine neurotransmitters do contain a n-electron system, but the local dipole therein has a different orientation from that present in LGIC agonists. This dissimilarity may be sufficient to have prevented the catecholamines, unlike transmitters such as acetylcholine, from crossing-over between receptor superfamilies through the course of evolution.

Construction and Structural Features of the Cys-Loop Model An alignment of all available cys-loop sequences and a motif to represent the conservation of amino acid residues within it is given in Figure 2a and b, respectively. To define the main chain conformation of the cys-loop the method of Gaboriaud et al.48was used in the prediction of a-helix and P-strand. Only a subset of the aligned cys-loop sequences was used in this analysis (see Fig. 3) in order to avoid biasing by over-representation. A marked two residue periodicity in the average hydrophobicity predicted a pstrand to occur over positions 1 to 7 and over positions 10 to 14. A chain reversal to allow for the formation of the disulfide bridge between the cysteine residues at positions 1and 15 is provided by a type VIa p-turn starting at position 7, and was selected by examination of similar sequence turns occurring in known protein structure^.^^ This turntype assignment is consistent with the invariance of the proline residue at position 9 of the cys-loop. In addition, energy calculations also suggested that a type VIa turn was favored over each of the alternative defined p-turns (see Table I). The preference values of MacGregor et al.,50 obtained from an analysis of known protein structures, were used to define the conformations of the side

388

V.B.COCKCROFT ET AL.

0

0

%

0

'X/\

GABA

Acetylcholine

0

%/>

0

Glycine

0

N

Muscimol

Nicotine

0

L-Alanine

0

-5:-

a

THlP

Cytisine

\\

3.5 to 5.5

Angstrom

b Fig. 1. (a) Structures of agonists of LGlC receptors. The ligands in columns from left to right are for the nACh receptor, the GABA, receptor, and the Glycine receptor. In rows from top to bottom are the neurotransmitters, semirigid agonist analogues, and almost totally rigid agonist analogues. (b) A schematic representation to show proposed common structural features of LGlC

agonists. The circle containing a plus symbol represents the positive role. The dashed lines represent the T-electron system. The local dipole within the T-system is represented by the symbols 6 + and 8-. The latter represents the electronegative center of the local dipole.

I

I

NEURAL

1 1 1 1 1 1 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 Alpha2 (r) C S I D V T P F P F D Q Q & C (c) y - - -

-

----

----

-

referenca 39 33

C S I D V T - F P P D Q Q N C Alpha3 (r) (c)

C K I D V T Y F P P D Y Q B C

_ _ _ _ - - -- -- - - - - -

39 33

C K I D V T Y F P F D Y Q Z C

-- -- -- -- -- --

E p s i l o n (b) C A V E V T Y P P I D W Q B C (=)

-

-

------

(r) - - -

ARD ( d )

C T I D V T Y F P F D Q Q T C

41

ALS(d1

C E I D V E Y P P P D E Q T C

41

44

35 38

---

- a -

C A V E V T Y P P P D W Q & C

e

Glr R

E

2

(

h

)

8 kD (r)

1 1 1 1 1 1 1 2 3 4 5 6 7 8 9 0 1 2 3 I 5

reference

C Q L Q L H N P P N D E H S C

46

C P U D L K N P P W D V Q T C

2

a

1 1 1 1 1 1 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5

Sequence Motif

C

-

h

-

r

h

*-F Y t t

(-1 P h D

t

-

Q H

t

-

C

+ + + + t + + + + + t + + + Fig, 2. (a) Alignment of amino acid sequences of the cys-loop. The numbering of residue positions within the cys-loop are equivalent to ositions 128-142 of the a-subunit of Torpedo nACh receptor. Abbreviations of species names are given as lower case letters in brackets: (h) = humans; (r) = rat; (m) = mouse: (b) = bovine; (c) = chicken: (x) = Xenopw (9)= goldfish: (t) =

e

b

Torpedo; (d) = Drosophila. (b) The cys-loop sequence motif. Invariant or strongly conserved residues are in upper case letters. h = conserved hydrophobic; t = binding surface residue; (-) = anionic site; * = specificity residue. The assigned X, conformations ( + = gauche+; t = trans) used in the construction of cys-loop models are indicated in the bottom line.

390

V.B. COCKCROFT ET AL. 21

I

A

A

-3 0

5

10

15

residue position Fig. 3. Hydrophobicity plot of the cys-loop.The hydrophobicity scale of Eisenberg4' was used. The average (circle),minimum (triangle),and maximum (invertedtriangle)values are shown and are based on an analysis of the a- and p-subunits of the bovine muscle nACh receptor, the a2- and p2-subunits of rat neuronal nACh receptor, the a-1, pl-%and y2-subunits of the GABA, receptor, and the 48 kDa subunit of the Glycine receptor.

TABLE 1. Minimized Energies for Different Turn Types for the Sequence Acetyl-Tyr-Phe-ProPhe-N-methyl Turn type I I' I1 11' I11 111' IVa IVb VIa VIb

Energy (kcal/mol) 163.9 167.5 160.5 167.5 163.9 162.2 160.6 160.5 154.2 157.8

chains. The most favored x1 torsion angle, either g + , t, or g - , was deduced using the alignment of cys-loop sequences and the assigned secondary structure for each residue position. With the x1 torsion angles defined (see Fig. 2b) the remaining side chain torsion angles used were as given by Sutcliffe et aL5' An initial model was constructed using the cysloop sequence of the a2-subunit of chick brain nACh receptor, as this sequence does not have an N-glycosylation site a t position 14. Energy minimization was then performed to produce an energetically reasonable structure (see Fig. 4). An accessible conformation search using molecular dynamics was performed to determine whether this cys-loop had other plausible structures. A high temperature simulation (600 K) of 5 psec was calculated. Analysis of the trajectory revealed no tendency for any residue to undergo major conformational change-only fluctuations around the orig-

inal structure occurred. However, because the side chains of residues of P-hairpins interlock and could limit the conformations searched, structurally modified forms of the cys-loopwere also analyzed (see Fig. 5). Surprisingly, when each of the residues except the two cysteines was substituted for alanine, or when the disulfide bridge of the original cys-loop was reduced, there was little change in the main chain conformation. Only when the disulfide bridge was reduced in the alanine substituted cys-loop did major conformational changes occur to the cys-loop structure. This analysis indicated that the residue side chains as well as the disulfide bridge of the cys-loop act to constrain its main chain flexibility. In addition, a high degree of rigidity for the cys-loop is suggested by the absence of glycine residues in any of the known cys-loop sequences and the common occurrence a t positions 3 and 5 of P-branched residues. A model for the different cys-loop types was constructed by residue substitution of the side chains of the initial model, followed by energy minimization. The final derived structure in each case was a p-hair pin with a type VIa turn and a disulfide bridge between positions 1and 15. Each of the derived structures clearly had a hydrophobic and a hydrophilic face; the latter is presumed to be exposed to the solvent. The comparison of the different cys-loop models revealed a marked conservation in the amino acid groups surrounding the invariant aspartate residue a t position 11, on the hydrophilic face. An invariant proline a t position 9 appears to have a structural role in maintaining the stereochemistry of the type VIa turn. A phenylalanine residue a t position 8 occurs in all LGIC sequences, except the p-subunit of the GABA, receptor, in which case a tyrosine residue is present (see Fig. 2a and discussion below). Only glutamine or histidine occurs a t position 13, which suggests that the residue a t this position acts as a hydrogen bond donor. This residue could possibly form a conserved hydrogen bond in a network involving the invariant aspartate residue a t position 11. As discussed below the residue at position 6 is proposed as conferring selectivity in the recognition of different LGIC receptor agonists. Although the residue a t this position does vary between members of the superfamily, it is highly conserved for a given LGIC subunit type (see Fig. 2a). On the hydrophobic face the residues at positions 3,5, and 10 are invariably hydrophobic and a patch is formed that could contribute to the inner core of the folded protein. An asparagine residue often occurs at position 14 of this same face as part of an N-glycosylation consensus sequence. Experimental evidence indicates that in the Torpedo nACh receptor this site is g l y c o ~ y l a t e d It .~~ is,~ ~ therefore, ~ likely that this site is a t the protein surface. In accord with this, the plot of the average hydrophobicity (see Fig. 3) shows that the strand containing this

391

Fig. 4. Energy minimized model of the qs-loop structure of the 1x2-subunit of chick brain nACh receptor. Numbering refers to the position of residues within the cys-loop. Side-chains are colored: magenta = serine, threonine, asparagine, and glutamine; red = aspartic acid: green = valine, isoleucine, phenylalanine, and tyrosine; white = proline: yellow = cysteine. The atoms of the main chain are colored white.

Fig. 6. Energy minimized model of (-)nicotine docked onto the cys-loop of the a2-subunit of chick brain nACh receptor. Atom colors are carbon = green for the cys-loop, and yellow for (-)nicotine; nitrogen = blue; oxygen = red; hydrogen = white.

392

V.B. COCKCROFT ET AL

I

I 0;

5

10

15

residue position Fig. 5. Standard deviation of C-u atom coordinates from the average coordinate set in the dynamic trajectories of the cys-loop of the a2-subunit of chick brain nACh receptor and modified forms of it. Symbols are: circle = the cys-loop; triangle = alanine substituted cys-loop; diamond = the cys-loop with the disulfide bridge reduced; square = alanine substituted cys-loop with the disulfide bridge reduced.

site is more hydrophilic than the oppositely facing strand. A disulfide bridge between the strands of a phairpin rarely occurs in known protein structures because the distance between cysteine residues of a disulfide bridge (C" . . . C" average distance = 5.5 A) opposes the formation of anti-parallel p-strands (C" . . . C" average distance = 4.9 A)."" An example of a disulfide-bridged p-hairpin has been reported for n e ~ r a m i n i d a s e , "in ~ which case a distortion of the main chain was found to accommodate the disulfide bridge. Likewise, a local main chain distortion (+ = -80; $ = 100) a t position 2 of the cys-loop was clearly evident when proline occurred at this position. Interestingly, in the cys-loop models with serine and threonine a t this position the same main chain distortion was stabilized by the formation of a side chain to main chain hydrogen bond.

Model of the Binding Complex Our main approach in modeling the interactions of ligand docking was to complement the common features of LGIC agonists (see discussion above) with the conservation of the groups surrounding the invariant aspartate residue in the different cys-loop sequences. The energy minimized model of (-)nicotine in its pharmacophore conformation docked onto the cys-loop model of the a2-subunit of chick brain nACh receptor is given in Figure 6. The following interactions are proposed as common features of recognition of LGIC agonists by their receptors. The first is the formation of an ion-pair interaction between the positive pole of agonist and the invariant aspartate residue a t position 11. The possibility that the positive charge of the agonist is stabilized by hydrogen bonding was considered un-

likely, as acetylcholine contains a quaternary ammonium group. The second is the interaction of the t ring-proton of the conserved aromatic residue a t position 8 with the r-electron density over the electronegative atom of the r-electron system of the agonist. The interaction is one in which the plane of the aromatic ring is orthogonal to that of the agonist r-electron system (this type of interaction is documented in the l i t e r a t ~ r e " ~ - ~In " ) addition, . the conserved local dipole of the agonist .rr-electron system is favorably oriented in the electrostatic field of the invariant aspartate group. We propose that the residue a t position 6 of the cys-loop is a key determinant of selective recognition. of GABA, glycine, and acetylcholine a t their receptors. An arginine residue occurs a t this position in the P-subunit of the GABA, receptor, along with a tyrosine residue at position 8. It is noteworthy that this is the only LGIC sequence having a tyrosine residue at this position of the cys-loop; a phenylalanine residue occurs in all other LGIC sequences known to date. It is proposed that a hydrogen bond formed between the tyrosine and the arginine residue maintains the arginine residue a t the right distance from the invariant aspartate residue to make an interaction with the carboxylate group of GABA. That a hydrogen bond can form between these two residues is supported by an analysis of known protein structures, which indicates that hydrogen bonds can occur between the side chains of these residue types when separated in primary structure by two residues (see Table 11). In contrast to this, a lysine residue occurs a t position 6 in the 48 kDa subunit of the Glycine receptor. The proposal is that the flexibility of the lysine side chain and the small size of its primary amine group allow it to bend back and interact with the carboxylate group of glycine. Thus, the shorter methylene chain length separating the amino and the carboxylate moieties of glycine as compared to that of GABA can be accommodated. In the acetylcholine cys-loop a threonine residue occurs a t position 6. It is proposed in this case that the hydroxyl group of threonine forms a hydrogen bond interaction with the ether oxygen of the ester bond of acetylcholine. A feature of semirigid LGIC agonists is that they can accommodate a chiral center between their positive pole and the electronegative atom of their T electron system, even when the chiral atom is within a cyclic ring structure. Thus, (R)-nicotine,61 (R) -dihydromuscimol,"" and (R) -a-amino-3-hydroxy5-methyl-4-isoxazolopropionate"" are less potent than their S-isomeric forms but, nevertheless, have greater than expected potency. Such weak stereoselectivity, on first analysis, may suggest a two rather than a three attachment-site model for ligand binding. In contrast, the weak stereoselectivity is accountable in our model by point-surface interactions, rather than discrete point-point interactions.

AGONIST BINDING TO LIGAND-GATED ION CHANNELS

TABLE 11. Analysis of Tyrosine and Arginine Side Chain to Side Chain Hydrogen Bond Interactions Database uuerv* (1)Number of occurrences of side chain to side chain hydrogen bonds between arginine and the residue at position (i + 2). (2) Number of occurrences of side chain to side chain hydrogen bonds between tyrosine and the residue at position ( i + 2).

Result Glutamate (191, tyrosine (111, aspartate (91, serine (71, histidine (11, glutamine (11, asparagine (1), total (49)

Arginine (111, lysine (lo), serine (91, glutamine (81, asparagine (81, threonine (6), glutamate (51, aspartate (31, methionine (21, histidine (11, total (63)

(3) Number of

112

occurrences of side chain to side chain hydrogen bonds between tyrosine and arninine *The BIPED relational database of protein structural information was queried using Standard Query Language CSQL).

The formation of a counterion pair can account for 5-10 kcal mol-' of binding energyF5 whereas the weaker interaction between the aromatic ring proton of the cys-loop and the ir-electron system of agonist could account for a further 1 kcal m 0 1 - l . ~The ~ sum of these two energies of interaction reasonably accounts for the low-affinity binding of agonists to LGICs, which for potent agonists of the well studied Torpedo nACh receptor and the GABAA receptor is around the micromolar concentration range.63p64It is noted that a property common to the nACh receptor and the GABA, receptor is that they convert to a desensitized state, and the receptors in this state bind agonist with a n affinity that is typically several orders of magnitude higher than the lowaffinity We note that this change in affinity equates well with the provision of 3-5 kcal mol-1 of binding energy resulting from the formation of a hydrogen bond with the electronegative atom within the .rr-electron system of agonist. From the docking model a recognition pathway is proposed. (1) When the agonist is within 12 A from the invariant aspartate residue, long-range electrostatic interaction between the negative charge of this residue and the positive pole of agonist is sufficient to cause the agonist to be attracted toward it. ( 2 ) The local dipole of the agonist becomes oriented by the electrostatic field of the invariant aspartate when the agonist is within about 6 A of it. The re-

393

orientation of the ligand at this step may assist subsequent binding. (3) On close approach, the size of the local dipole of the agonist is increased in the electrostatic field of the invariant aspartate, causing a shift of electron density over the electronegative atom of the a-electron system. This in turn favors the interaction of this electronegative center with the E ring-protan of the aromatic residue a t position 8 of the cys-loop.

Experimental Support for the Docking Model The specific residues that are spatial neighbors of the invariant aspartate residue, particularly that at position 6 of the cys-loop proposed above as being important in selective recognition of agonist, in the case of the GABA, receptor and the Glycine receptor account for several experimental findings: 1. Agonist binding to the GABA, receptor is abolished by chemical modification of arginine residues with either 2,3-butadione or phenylglyoxal.6' It is the @-subunitof the GABA, receptor that is the site of photoaffinity labeling by the agonist muscim01.'~,~*Thus, the presence of an arginine residue at position 6 of the cys-loop of the P subunit of the GABA, receptor agrees with these observations. 2. Chemical modification of tyrosine residues with p-diazobenzenesulphonic acid, tetranitromethane, or N-acetylimidazole also causes disruption of agonist binding t o the GABAA receptor?' This is explained by the unique occurrence of a tyrosine residue at position 8 of the cys-loop of the @-subunitof the GABA, receptor, as this residue is proposed a s forming a crucial hydrogen bond interaction with the arginine residue at position 6 (see discussion above). 3. The modification of histidine residues with diethylpyrocarbonate specifically disrupts benzodiazepine binding with no marked effect on GABA agonist binding.69.70 The y2-subunit of the GABA, receptor, tentatively assumed to contain determinants of the high-affinity site of benzodiazepines as based on recent cloning and functional expression data:' has a histidine residue at position 6 of its cys-loop, whereas the cys-loop of the P-subunit, the site of GABA agonist labeling, contains no histidine residues. 4. For the Glycine receptor, Gomez et al.71 have recently shown that chemical modification of lysine residues with fluorescein isothiocyanate affects the interaction of glycine at its binding site. Chemical cleavage at tryptophan residues revealed that an 8.5 and a 13.9 kDa fragment of the 48 kDa subunit is labeled. These observations are in accord with the occurrence of a lysine residue at position 6 of the cys-loop of the 48 kDa subunit, as this residue is in a predicted cleavage fragment of 9.3 kDa and a predicted partial cleavage fragment of 13.3 kDa.

394

V.B. COCKCROFT ET AL.

General Discussion The cys-loop is the most conserved stretch of amino acid sequence in the diverse set of sequences of LGIC subunits known to date; 4 of its 15 residue positions are invariant. In contrast, 14 residues are invariant in the N-terminal extracellular region of LGIC subunits, which is >200 residues long. Additionally, a t position 11 of the cys-loop an invariant aspartic acid residue occurs, which is one of only two invariant acidic residue positions present in the extracellular region of LGIC subunits and is, therefore, a good candidate for the anionic site. In none of the cys-loop sequences does an insertion or deletion of amino acid residues occur. For this reason and because the first and last position are disulfide linked, it can be considered to be a coherent structural motif of LGIC receptors. The major strength of the modeling is that, first, a single type of structure accommodates all of the sequence variations of the cys-loop in the >40 known polypeptides of the LGIC superfamily. Second, specific chemical modifications of the GABA, receptor and Glycine receptor can be accounted for by the residues that are spatial neighbors of the invariant aspartate group on the hydrophilic face of the cysloop, and in particular the residue a t position 6 of the cys-loop. There is evidence to suggest that for the Torpedo nACh receptor there are multiple low-affinity binding sites involved in receptor activation, which may be on other subunits besides the a - s u b ~ n i t . 6It~ is recognized that the photoactivatable ligand p-(dimethy1amino)benzenediazoniumf luoroborate (DDF) labels not only the a-subunit of the Torpedo nACh receptor but also the y-subunit in this receptor.72 Moreover, labeling is inhibited by the agonist carbamoylcholine, suggesting that the y-subunit may indeed have its own agonist site. This is also suggested by the cys-loop of this subunit which shares with the a-subunit the feature of a threonine a t position 6, whereas the (3 and the 6 subunit have methionine and leucine, respectively. From studies using reagents that alkylate cysteines 192-193 of the Torpedo nACh receptor it is often suggested that the region around 192-193 plays a dominant role in agonist binding. For such reagents the active part of the molecule is in a position equivalent to the bromomethyl group of bromoacetylcholine and is probing for residues in a limited region of the whole binding site. It is noteworthy that the docking model in no way precludes labeling outside of the cys-loop, and indeed in an extended model of the binding cavity (Cockcroft et al., in presss3) cysteines 192-193 can be accommodated such that the reactive moieties of afinity ligands docked onto the cys-loop may interact specifically with the reactive thiol groups of these two residues. Furthermore, the heavy labeling of a-sub-

units may not be a reflection of normal agonist activation but of receptor desensitization involving the two high-affinity binding sites per oligomer associated with the a - s ~ b u n i t s . ~ ~ Interestingly, cysteines 192-193 are absent in all other LGIC subunits. Also, the surrounding region even between a-subunits of the muscle/electric organ type31 and the neuronal subtypes of nACh receptor3’ can be seen to have accepted residue substitutions and insertionsldeletions. Moreover, cysteines 192-193 have been experimentally mutated to serines without complete loss of agonist binding.73 This noted lack of structural conservation is in our view inconsistent with the overlap in nicotinoid pharmacology seen for members of the nicotinic family of receptors.61 We, therefore, favor the view that this region may be close to the agonist binding site as part of a binding cleft and that it contains important determinants of toxin binding. Recently, the primary structures of subunits of the kainate receptor7 and kainate binding prot e i n ~ from ~ ~ a* variety ~ ~ of vertebrates have been derived from their cloned gene sequences. Although these polypeptides show homology to each other, none of the invariant residues in the LGIC set of related sequences is invariant in the kainate set, which indicates that subunits of the kainate receptor are not homologous to those of LGICs. There is low similarity over the transmembrane regions of the kainate and the nACh receptor but this may simply be because membrane spanning domains use a limited set of amino acids. Interestingly, although there are several ligands that block the ion channels of both the nACh receptor and the NMDA type of glutamate r e c e p t ~ r ,no ~ ~such ?~~ effect has been reported for the kainate receptor. We favor the view that certain subunits in an intact receptor may have within their structure all the determinants for the recognition of a given agonist. The contribution of other subunits could then be in stabilizing the tertiary and quaternary structure of the receptor. This view is partly supported by experiment since binding of nicotine to an approximate 80 kDa polypeptide species can be detected after treatment of rat brain membranes with SDS.” Also, single subunits of the GABAAre~eptor,~’ and the Glycine re~eptor,~’ in the absence of the other subunits that comprise these receptors, give electrophysiological responses to applied agonist on expression in the Xenopus oocyte. Similarly, when the Torpedo nACh receptor devoid of either the p-, y-, or &subunit is expressed in the Xenopus oocyte weak channel activity can be observed.” It is noteworthy that for the GABA, receptor addition of the y2-subunit introduces benzodiazepine-like modulation to the heterologously expressed form of receptor.46If each subunit brings to a receptor its own recognition site, this would facilitate the evolution of “mixed” receptor forms, such as the NMDA receptor, and the

AGONIST BINDING TO LIGAND-GATED ION CHANNELS

GABA, receptor. Indeed, invertebrates are suggested to have receptors that display combinations of the pharmacological properties of the classical vertebrate LGIC receptors.81z82 The molecular modeling is presented as a first step in the understanding of the structure-activity relationships of receptor ligands, while taking into consideration the receptor sites themselves. We are currently employing the techniques of site-directed mutagenesis, functional expression, and chemical analysis of receptor proteins to test the premises of our modeling. Finally, the comparative approach to receptor structural modeling we have adopted has much potential value in the rational design of novel ligands with subtype and species specificity.

ACKNOWLEDGMENTS We thank Drs. Owen Jones, Bill Turnell, Anne Stephenson, Mike Duggan, and Pnina DauberOsguthorpe for helpful discussion and Dr. Pnina Dauber-Osguthorpe for assistance in manuscript preparation. We thank the SERC and Shell Research (UK) Ltd. for financial support.

14.

15.

16.

17.

18.

19.

20. 21.

REFERENCES 1. Schofield, P. R., Darlison, M. G., Fujita, N., Burt, D.,

Stephenson, F. A,, Rodriguez, H., Rhee, L. M., Ramachandran, J., Reale, V., Glencorse, T. A., Seeburg, P. H., Barnard, E. A. Sequence and functional expression of the GABA, receptor shows a ligand gated receptor superfamily. Nature (London) 328:221-227, 1987. 2. Grenningloh, G., Rienitz, A., Schmitt, B., Methfessel, C., Beyrether, K., Gundelfinger, E. D., Betz, H. The strychnine binding subunit of the glycine receptor shows homology with the nicotninc acetylcholine receptor. Nature (London) 328:215-220, 1987. 3. Noda, M., Takahashi, H., Tanabe, T., Toyosato, M., Furutani, Y., Hirose, T., Asai, M., Inayama, S., Miyata, T., Numa, S. Primary structure of a-subunit precursor of Torpedo californicu acetylcholine receptor deduced from cDNA sequence. Nature (London) 299:793-797, 1982. 4. Cull-Candy, S., Usowicz, M. M. Patch-clamp recording from single glutamate receptor channels. Trends Pharmacol. 8:218-224, 1987. 5. Derkach, V., Surprenant, A,, North, R. A. 5-HT3 receptors are membrane ion channels. Nature (London) 339:706709, 1989. 6. Hardie, R. C. A histamine-activated chloride channel in neurotransmission at a photoreceptor synapse. Nature (London) 339:704-706, 1989. Shea-Greenfield, A,, Rogers, S. W., Heine7. Hollman, M. 0.. mann, S. Cloning by functional expression of a member of the glutamate receptor family. Nature (London)342:643648, 1989. 8. Popot, J. L., Changeux, J.-P. Nicotinic receptor of acetylcholine: structure of a n oligomeric integral membrane protein. Physiol. Rev. 64:1162-1239, 1984. 9. Stroud, R. M., Finer-Moore, J. Acetylcholine receptor structure, function, and evolution. Annu. Rev. Cell Biol. 1:317-351, 1985. 10. Raftery, M. A., Hunkapiller, M. W., Strader, C. D., Hood, L. E. Acetylcholine receptor: Complex of homologous subunits. Science 208:1454-1457, 1980. 11. Toyoshima, C., Unwin, N. Ion channel of acetylcholine receptor reconstructed from images of postsynaptic membranes. Nature (London) 336:247-250, 1988. 12. Kao, P. N., Karlin, A. Acetylcholine receptor binding site contains a disulphide cross-link between adjacent halfcystinyl residues. J. Biol. Chem. 261:8085-8088, 1986. 13. Neumann, D., Barchan, D., Fridkin, M., Fuchs, S. Analysis of ligand binding to the synthetic dodecapeptide 185-

22.

23. 24.

25. 26.

27.

28. 29.

30.

31.

32.

395

196 in the acetylcholine receptor a subunit. Proc. Natl. Acad. Sci. U.S.A. 83:9250-9253, 1986. McCormick, D. J., Atassi, M. Z. Localisation and synthesis of the acetylcholine-binding site in the a-chain of the Torpedo californica acetylcholine receptor. Biochem. J. 224: 995-1000, 1984. Madhok, T. C., Chao, C. C., Matta, S., Hong, A,, Sharp, B. M. Monospecific antibodies against a synthetic peptide predicted from the alpha-3 nicotinic receptor cDNA inhibit binding of [3H]nicotine to rat brain nicotinic cholinergic receptor. Biochem. Biophys. Res. Commun. 165:151-157, 1989. Luyten, W. H. M. L. A model for the acetylcholine binding site of the nicotinic acetylcholine receptor. J. Neurosci. Res. 165-73, 1986. Smart, L., Meyers, H., Hilgenfeld, R., Saenger, W., Maelicke, A. A structural model for the ligand-binding sites a t the nicotinic acetylcholine receptor. FEBS Lett. 178:64-68, 1984. Dauber-Osguthorpe, P., Roberts, V. A,, Osguthorpe, D. J., Wolff, J., Genest, M., Hagler, A. T. Structure and energetics of ligand binding to proteins: Escherichia coli dihydrofolate reductase-trimethoprim, a drug-receptor system. PROTEINS: Struc. Func. Genet. 4:31-47, 1988. Sessions, R. B., Dauber-Osguthorpe, P., Osguthorpe, D. J. Filtering molecular dynamics trajectories to reveal low frequency collective motions: phospholipase A,. J. Mol. Biol. 209:617-633, 1989. Islam, S. A,, Sternberg, M. J. E. A relational database of protein structures designed for flexible enquiries about conformation. Prot. Eng. 2:431-442, 1989. Sheridan, R. P., Nilakantan, R., Dixon, J. S., Venkataraghavan, R. The ensemble approach to distance geometry: Application to the nicotinic pharmacophore. J. Med. Chem. 29:899-906, 1986. Krogsgaard-Larsen, P., Nielsen, L., Falch, E., Curtis, D. R. GABA agonists. Resolution, absolute stereochemistry, and - 1-dihydromuscienantioselectivity of (SM+ 1- and (R)-( mol. J. Med. Chem. 28:1612-1617, 1985. Tokutomi, N., Kaneda, M., Akaike, N. What confers specificity on glycine for its receptor site? Br. J. Pharmacol. 97:3553-360, 1989. Pupillo, M., Klein, P., Vaughan, R., Pitt, G., Lilly, P., Slim, T., Devreotes, P., Kumagal, A,, Firtel, R. CAMPreceptor and G-protein interactions control development of Dictyostelium. Cold Spring Harbor Symp. 53:657-665, 1988. Nathans, J., Thomas, D., Hogness, D. S. Molecular genetics of human colour vision: The genes encoding blue, green, and red pigments. Science 232:193-202, 1986. Masu, Y., Nakayama, K., Tamaki, H., Harada, Y., Kuno, M., Nakanishi, S. cDNA cloning of bovine substance-K receptor through oocyte expression system. Nature (London) 329:836-838, 1987. Kubo, T., Fukuda, K., Mikami, A,, Maeda, A,, Takahashi, H., Mishima, M., Haga, T., Ichiyama, A,, Kangawa, K., Matsuo, H., Hirose, T., Numa, S. Cloning, sequencing and expression of complementary DNA encoding the muscarinic acetylcholine receptor. Nature (London) 323:411-416, 1986. Lefkowitz, R. J., Caron, M. G. Models for the study of receptors coupled to guanine nucleotide regulatory proteins. J. Biol. Chem. 263:4993-4996, 1988. Bunzow, J. R., Tol, H. H. Van, Grandy, D. K., Alpert, P., Salon, J.,Christie, M., Machida, C. A., Neve, K. A,, Civelli, 0. Cloning and expression of a rat D, receptor cDNA. Nature (London) 336783-787, 1988. Kobilka, B. K., Frielle, T., Collins, S., Yang-Feng, T., Kobilka, T. S., Francke, U., Lefkowitz, R. J., Caron, M. G. An intronless gene encoding a potential member of the family of receptors coupled to guanine nucleotide regulatory proteins. Nature (London) 329:75-79, 1987. Kubo, T., Noda, M., Takai, T., Tanabe, T., Kayano, T., Shimizu, S., Tanaka, K., Takahashi, H., Hirose, T., Inayama, s., Kikuno, R., Miyata, T., Numa, s. Primary structure of 6 subunit precursor of calf muscle acetylcholine receptor deduced from cDNA sequence. Eur. J. Biochem. 1495-13, 1985. Boulter, J., Luyten, W., Evans, K., Mason, P., Ballivet, M., Goldman, D., Stengelin, S., Martin, G., Heinemann, S., Patrick, J. Isolation of a clone coding for the a-subunit of a

396

33.

34.

35.

36.

37.

38. 39.

40.

41. 42.

43.

44.

45.

46.

47. 48.

49. 50.

51.

52.

V.B. COCKCROFT ET AL.

mouse acetylcholine receptor. J. Neurosci. 5:2545-2552, 1985. Nef, P., Oneyser, C., Alliod, C., Couturier, S., Ballivet, M. Genes expressed in the brain define three distinct neuronal nicotinic acetylcholine receptors. EMBO J. 7595601,1988. Baldwin, T. J., Yoshihara, C. M., Blackner, K., Kinter, C. R., and Burden, S. J. Regulation of acetylcholine receptor transcript expression during development in Xenopus laeuis. J. Cell Biol. 106:469-478, 1988. Buonanno, A,, Mudd, J., Merlie, J. P.Isolation and characterisation of the p and E subunit genes of mouse muscle acetylcholine receptor. J. Biol. Chem. 264:7611-7616, 1989. Nef, P., Mauron, A,, Stalder, R., Alliod, C., Ballivet, M. Structure, linkage, and sequence of the two genes encoding the 6 and y subunits of the nicotinic acetylcholine receptor. Proc. Natl. Acad. Sci. U.S.A. 81:7975-7979, 1984. LaPolla, R. J., Mayne, K. M., Davidson, N. Isolation and characterisation of a cDNA clone for the complete protein coding region of the 6 subunit of the mouse acetylcholine receptor. Proc. Natl. Acad. Sci. U.S.A. 81:7970-7974, 1984. Criado, M., Witzemann, V., Koenen, M., Sakmann, B. Nucleotide sequence of the rat muscle acetylcholine receptor r-subunit. Nucl. Acids Res. 16:10920, 1988. Wada, K., Ballivet, M., Boulter, J., Connolly, J.,Wada, E., Deneris, E. S., Swanson, L. W., Heinemann, S. Functional expression of a new pharmacological subtype of brain nicotinic acetylcholine receptor. Science 240:330-334, 1988. Deneris, E., Boulter, J., Swanson, L. W., Patrick, J.,Heinemann, S. p3: A new member of nicotinic acetylcholine receptor gene family is expressed in brain. J. Biol. Chem. 264:6268-6272, 1989. Bossy, B., Ballivet, M., Spierer, P. Conservation of neural nicotinic acetylcholine receptors from Drosophila to vertebrate central nervous systems. EMBO J. 7:611-618, 1988. Cauley, K., Arganoff, B. W., Goldman, D. Identification of a novel nicotinic acetylcholine receptor structural subunit expressed in goldfish retina. J. Cell. Biol. 108636-645, 1989. Levitan, E. S., Schofield, P. R., Burt, D. R., Rhee, L. M., Wisden, W., Kohler, M., Fujita, N., Rodriguez, H., Stephenson, A,, Darlison, M. G., Barnard, E. A., Seeburg, P. H. Structural and functional basis for GABA, receptor heterogeneity. Nature (London) 33536-79, 1988. Ymer, S., Schofield, P. R., Draguhn, A., Werner, P., Kohler, M., Seeburg, P. H. GABA, receptor beta subunit heterogeneity: Functional expression of cloned cDNAs. EMBO J. 81665-1670,1989. Takai, T., Noda, M., Mishina, M., Shimizu, S., Furutani, Y., Kayano, T., Ikeda, T., Kubo, T., Takahashi, H., Takahashi, T., Kumo, M., Numa, S.Cloning, sequencing and expression of cDNA for a novel subunit of acetylcholine receptor from calf muscle. Nature (London) 315:761-764, 1985. Pritchett, D. B., Sontheimer, H., Shivers, B. D., Ymer, S., Kettenmann, H., Schofield, P. R., Seeburg, P. H. Importance of a novel GABA, receptor subunit for benzodiazepine pharmacology. Nature (London) 338:582-585, 1989. Eisenberg, D., Schwarz, E., Komaromy, M., Wall, R. Analysis of membrane and surface protein sequences. J. Mol. Biol. 179:125-142, 1984. Gaboriaud, C., Bissery, V., Benchetrit, T., Mornon, J. P. Hydrophobic cluster analysis: a n efficient new way to compare and analyse amino acid sequence. FEBS Lett 224: 149-155, 1987. Chou, P. Y., Fasman, G. D. p-turns in proteins. J. Mol. Biol. 115:135-175, 1977. MacGregor, M. J., Islam, S. A., Sternberg, M. J. E. Analysis of the relationship between side-chain conformation and secondary structure in globular proteins. J. Mol. Biol. 198:295-310,1987, Sutcliffe, M. J., Hayes, F. R. F., Blundell, T. L. Knowledge based modelling of homologous proteins, part 11: rules for the conformations of substituted sidechains. Prot. Eng. 1(5):385-392, 1987. Conti-Tronconi, B. M., Hunkapiller, M. W., Raftery, M. A. Molecular weight and structural nonequivalence of the

53.

54. 55. 56.

57. 58. 59. 60.

61. 62.

63. 64. 65.

66. 67.

68.

69.

70.

71.

72.

73.

mature a subunits of Torpedo californica acetylcholine receptor. Proc. Natl. Acad. Sci. U.S.A. 81:2631-2634, 1984. Nomoto, H., Takahashi, N., Nagaki, Y., Endo, S., Arata, Y., Hayashi, K. Carbohydrate structures of acetylcholine receptor from Torpedo californica and distribution of oligosaccharides among the subunits. Eur. J. Biochem. 57: 233-242,1986. Varghese, J. N., Laver, W. G., Colman, P. M.Structure of the influenza virus glycoprotein antigen neuraminidase at 2.9 & . resolution. Nature (London) 303:35-44, 1983. Andrews, P. R., Craik, D. J., Martin, J. L. Functional group contributions to drug-receptor interactions. J. Med. Chem. 27:1648-1657, 1984. Thomas, K. A., Smith, G. M., Thomas, T. B., Feldmann, R. J. Electronic distributions within protein phenylalanine aromatic rings reflected by the three-dimensional oxygen atom environments. Proc. Natl. Acad. Sci. U.S.A. 79:48434847,1982. Czupinski, O., Malarski, Z., Sobczyk, L. Spectroscopic and dielectric studies of n-electron hydrogen bonded systems. Bull. Acad. Pol. Sci. Sev. Sci. Chim. 28:661-667, 1980. Burnley, S. K., Petsko, G. A. Weakly polar interactions in proteins. Adv. Protein Chem. 39:125-189, 1988. Pettersen, I., Liljefors, T., Benzene-benzene (phenyl-phenyl) interactions in MM2/MMP2 molecular mechanics calculations. J. Comp. Chem. 8:1139-1145, 1987. Matthews, D. A,, Alden, R., Bolen, J. T., Freer, S. T., Hamlin, R., Xuong, N., Kraut, J., Poe, M., Williams, M., Hoogstein, K. Dihydrofolate reductase: X-ray structure of the binary complex with methotrexate. Science 197:452455, 1977. Wonnacott, S. a-Bungarotoxin binds to low-affinity nicotine binding sites in rat brain. J. Neurochem. 47:17061712, 1986. Laurisden, J., Honore, T., Krogsgaard-Larsen, P. Ibotenic acid anologues. Synthesis, molecular flexibility, and in vitro activity of agonists at central glutamic acid receptors. J. Med. Chem. 28:668-672, 1985. Dunn, S. M. J., Raftery, M. A. Multiple binding sites for agonists on Torpedo californica acetylcholine receptor. Biochemistry 21:6264-6272, 1982. Agey, M. W., Dunn, S. M. J. Kinetics of [3Hlmuscimol binding to the GABA, receptor in bovine membranes. Biochemistry 28:4200-4208, 1989. Higgins, L. S., Berg, D. K. A densensitised form of neuronal acetylcholine receptor detected by 3H-nicotine binding on bovine adrenal chromaffin cells. J. Neurosci. 8: 1436-1446, 1988. Widdows, K. B., Kirkness, E. F., Turner, A. J. Modification of the GABAlbenzodiazepine receptor with the arginine reagent, 2,3-butadione. FEBS Lett. 222:125-128, 1987. Casalotti, S., Stephenson, F. A., Barnard, E. A. Separate subunits for agonist and benzodiazepine binding in the y aminobutyric acid, receptor oligomer. J. Biol. Chem. 261: 15013-15016,1986. Deng, L., Ransom, R. W., Olsen, R. W. [3H1Muscimolphotolabels the y-aminobutyric acid receptor binding site on a peptide subunit distinct from that labelled with benzodiazepine. Biochem. Biophys. Res. Commun. 138:13081314, 1986. Maksay, G., Ticku, M. K. Characterisation of y-aminobutyric acid-benzodiazepine receptor complexes by protection against inactivation by group-specific reagents. J. Neurochem. 42:1715-1727,1984. Lambolez, B., Rossier, J. Benzodiazepine agonists protect a histidine residue from modification by diethyl pyrocarbonate whereas propyl p-carboline does not. FEBS Lett. 219: 301-305, 1987. Gomez, I. V., Fernandez-Shaw, C., Valdivieso, F., Mayor, F. Chemical modification of the glycine receptor with fluorescein isothiocyanate specifically affects the interaction of glycine with its binding site. Biochem. Biophys. Res. Commun. 160:374, 1989. Langenbuch-Cachat, J., Bon, C., Mulle, C., Goeldner, M., Hirth, C., Changeux, J. Photoaffinity labeling of the acetylcholine binding on the nicotinic receptor by a n aryldiazonium derivative. Biochemistry 27:2337-2345, 1988. Mishina, M., Tobimatsu, T., Imoto, K., Tanaka, K., Fulita, Y., Fukuda, Y., Hirose, T., Inayama, S., Takahashi, T., Kuno, M., Numa, S. Location of functional regions of ace-

397

AGONIST BINDING TO LIGAND-GATED ION CHANNELS

74.

75.

76.

77. 78.

tylcholine receptor a-subunit by site-directed mutagenesis. Nature (London) 313:364-369, 1985. Gregor, P., Mano, I., Maoz, I., McKeown, M., Teichberg, V. I. Molecular structure of the chick cerebellar kainatebinding subunit of a putative glutamate receptor. Nature (London) 342:689-692, 1989. Wada, K., Dechesne, C. J., Shimasaki, S., King, R. G., Kusano, K., Buonanno, A., Hampson, D. R., Banner, C., Wenthold, R. J., Nakatani, Y. Sequence and expression of a frog brain complementary DNA encoding a kainatebinding protein. Nature (London) 342:684-689, 1989. Kavanaugh, M. P., Tester, B. A. C., Weber, E. Interaction of MK-801 with the nicotinic acetylcholine receptor-associated ion channel from electroplax. Eur. J. Pharmacol. 164:397-398,1989. Ramoa, A. S., Albuquerque, E. X. Phencyclidine and some of its analogues have distinct effects on NMDA receptors of rat hippocampal neurons. FEBS Lett. 212:292-296, 1987. Blair, L. A. C., Levitan, E. S., Marshall, J., Dionne, V. E., Barnard, E. A. Single subunits of the GABA, receptor form ion channels with properties of the native receptor Science 242:577-581, 1988.

79. Schmieden, V., Grenningloh, G., P. R. Schofield, Betz, H. Functional expression in Xenopus oocytes of the strychnine binding 48 kD subunit of the glycine receptor. EMBO J. 8:695-699, 1989. 80. Kurosaki, T., Fukuda, K., Mori, Y., Tanaka, K., M. Mishina. Numa. S. Functional DroDerties of nicotinic acetvlcholine receptor subunits exireised in various combin&ions. FEBS Lett. 2:253-258, 1987. 81. Ono, J. K., Salvaterra, P. M. Snake a-toxin effects on cholinergic and non-cholinergic responses of Aplysia californica neurons. J. Neurosci. 1:259-270, 1980. 82. King, W. M., Carpenter, D. 0. Distinct GABA and glutamate receptors may share a common channel in Aplysia neurons. Neurosci. Lett. 82:343-348. 1987. 83. Cockcroft, V.B., Lunt, G. G., Osguthorpe, D. J. Modelling of binding sites of the nicotinic acetylcholine receptor and their relation to models of the whole receptor. In "Protein Structure, Prediction and Design." Kay, J., Lunt, G. G., Oseuthome. D. J.. eds. Biochemical Societv SvmDosium No. 57. Lbndon: Portland Press (in press). "

I

_