Stabilization of Acetylcholine Receptor Channels by Lipids in Cholate Solution and during Reconstitution in Vesicles*

Acetylcholine receptors were solubilized from electric organ membranes of Torpedo californica in mixed micelles of sodium cholate and soybean lipids. Sodium cholate, when supplemented with relatively low amounts of soybean lipids (cholate:lipid, 20:1, molar ratio), was effective in solubilizing receptors without denaturing their agonist-regulated cation channels. Another dialyzable detergent, octylglucoside, denatured the ion channel even in the presence of excess lipids. Reassembly of receptors and lipids into vesicles was achieved by cholate dialysis. About 70% of the receptors were oriented with their toxin binding sites on the external surface of the vesicles. Evidence suggests that all of the receptors in a single vesicle were oriented either right side out or inside out.

tisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
fi To whom reprint requests should be addressed at The Salk Institute for Biological Studies, P.O. Box 85800, San Diego, CA 92158. pentameric complex of partially homologous subunits in the mole ratio (~$ 3~8 (1-3) which in the native membrane exists as a dimer formed by a disulfide bond between the 6 subunits (4, 5). The a subunits can be affinity labeled by probes for the acetylcholine binding sites (6, 71, and the agonist-regulated cation channel is contained within the a2fiy8 monomer (8). However, it is still not known which of the subunits contribute to the structure of the cation channel or how opening and closing of this channel is regulated.
The fist reproducible reconstitution of AChRs' into lipid vesicles was reported by Epstein and Racker (9), who used crude AChR-containing membrane preparations and soybean lipids. This work led to the realization that the continuous presence of lipids during solubilization of the AChR in cholate is essential for preserving the functional integrity of the channel (Refs. 10-12, except see Ref. 13). This notion permitted purscation of AChRs with intact cation channels and the reconstitution of these AChRs into lipid vesicles (10, 11) and planar lipid bilayers (14). Reconstitution of functional AChRs into vesicles (13,15,16) or planar bilayers (17) has also been reported by several other laboratories. The role of specific lipid subclasses for the performance of the agonist-induced cation translocation has been investigated, but no absolute requirement for a specific lipid has been demonstrated (18).
In this paper we determine the conditions required for the stabilization of AChR channels in cholate-lipid solution and during the process of reincorporation into a reconstituted membrane. Surprisingly, more lipid is required to protect the AChR channel during reconstitution than in cholate solution. Another unexpected observation is that AChRs and lipids reassemble into vesicles at a fixed protein/lipid ratio which is much higher than the protein/lipid ratio in cholate-lipid solution. Addition of a large excess of lipid does not result in a reduced amount of AChRs/vesicle. These results provide insights in the mechanisms by which AChRs and lipids interact during the formation of reconstituted vesicles.
Preparation of AChR-rich Membranes-AChR-rich membranes from the electric organ of T. californica (Pacific Biomarine, Venice, CA) were prepared by sucrose density centrifugation, following dif-ferential centrifugation of the initial homogenate as previously described (11). AChR concentration was determined by radioimmunoassay (11). Protein was determined according to the method of Lowry et al. (22) using bovine serum albumin as standard.
Solubilization ofAChR-rich Membranes-AChR-rich membranes were dissolved in 2% sodium cholate, unless otherwise indicated, in the presence of the desired concentration of crude soybean lipids in 10 mM Na phosphate buffer, pH 7.4, 100 mM NaCl. Crude soybean lipids were prepared as a 150 m g / d stock dispersion in distilled water by sonication under argon in a Bransonic bath type sonicator. The solubilized membranes were gently shaken for the indicated periods of time at 4 "C. Undissolved material was removed either by centrifugation at 165,000 X g for 30 min or, for small volumes, by centrifugation for 15 min in an Eppendorf microfuge at 4 "C. Both procedures gave comparable AChR concentrations in the supernatants.
Reconstitution of AChRs into Lipid Vesicles-Lipid concentrations in the extract were adjusted using a sonicated dispersion of soybean lipids (150 mg/ml) in distilled water. Di[l-'4C]palmitoyl-~a-phosphatidylcholine was sometimes included as tracer by sonication (10-15 min) of the lipid stock dispersion added to a dry film of the radioactive lipid after evaporation of the solvent under argon. Final concentrations of the other components in the reconstitution mixture were 100 mM NaCI, 2% sodium cholate, and 10 mM Na phosphate buffer, pH 7.4. The reconstitution mixture was dialyzed for 16-18 h against 500 volumes of 100 mM NaCl, 10 mM NaN3, 10 mM Na phosphate buffer, pH 7.4, and then for the same period against 500 volumes of 145 mM sucrose, 10 mM NaN3,lO mM Na phosphate buffer, pH 7.4.
Measurement of Carbamylcholine-induced Uptake of 22Na'-lnflux of "Na+ into vesicles was assayed at room temperature using the basic method of Gasco et al. (23) as previously described (11). All assays were done in triplicate. At time zero, 40 pl of reconstituted vesicles were added to 5 pl of '*Na' (200 pCi/ml of stock solution, New England Nuclear) plus 5 pl of distilled water (background control) or 5 pl of carbamylcholine ( W 3 M). After 10 s, 40 pl of mixture were applied to 2-ml columns of Dowex 50W-8X (purchased from Sigma, converted to the Tris form as described in (11). and washed with 3 ml of 170 mM sucrose, 3 mg/ml of bovine serum albumin immediately before use). The resin was immediately rinsed carefully with 3 ml of 175 m M sucrose, and the eluate was counted in a Nuclear Chicago model 1185 y-counter with background correction at 38% efficiency.
The apparent equilibrium volume of the vesicles was measured in cpm/mg of lipid after incubation of a regular background control sample with 22Na+ for 48 h at 4 "C prior to passage through Dowex.
Isopycnic Sucrose Gradient Centrifugation of Reconstituted Vesicles-Reconstituted vesicles were diluted with an equal volume of 100 mM NaC1, 10 mM NaN3, 10 mM Na phosphate buffer, pH 7.4, containing a trace label of '251-aBGT (3.0 X lo-' M) and incubated for 1-2 h at 4 "C. Samples of 0.1 ml were applied to 4.8 ml of 5-25% (w/ w) linear sucrose gradients and centrifuged to equilibrium in a Beckman S W 50.1 rotor at 200,000 X g for 18 h. Fractions of 0.23 ml were collected after puncturing the tubes and counted in a y-counter. The sucrose concentration of each fraction was determined at 20 "C with a refractometer.

RESULTS
Stability of AChR Channels during Solubilization of the Native Membrane-It has previously been shown that the presence of supplementary soybean lipids in mixed micelles with cholate at a 1:10 (w/w) ratio can protect AChR channels from denaturation (10, 11). These observations have recently been confirmed by other investigators (12). Under these conditions extraction of native membranes is maximal at 2% cholate and amounts to a recovery of about 68% of the initial AChRs in the extract (Fig. 1A). Half-maximal extraction is obtained at a Concentration of about 0.8% cholate. Increasing the cholate concentration above 2% does not increase the extraction yield further, although variable amounts of additional ['251]a-BGT binding activity can be obtained from the undissolved residue by treatment with 2% Triton X-100 (data not shown). When detergent concentrations above 3% cholate are used for solubilization of the membranes, a strong irreversible denaturing effect of the detergent on the channel *Or FIG. I. Solubilization of AChRs from AChR-rich membranes (A) and detergent inactivation of AChR function by mixed micelles of sodium cholate and soybean lipids (B). Aliquots of AChR-enriched membranes were centrifuged for 15 min in an Eppendorf microfuge. The pellets were suspended at a concentration of 4 mg/ml of protein in 100 mM NaCI, 10 mM Na phosphate buffer, pH 7.4, supplemented with different concentrations of sodium cholate and soybean lipids at a constant weight ratio of 1O:l. After 18-24 h of shaking, the cholate and lipid concentrations were in all samples adjusted to 2% and 25 mg/ml, respectively. After an additional incubation period of 30 min of slow shaking, the samples were centrifuged in an Eppendorf microfuge as before; the supernatants were collected, assayed for 1251-a( BGT binding activity, and dialyzed in order to form reconstituted vesicles. The resulting vesicles were then assayed for carbamylcholine-dependent uptake of *'Na+ at M carbamylcholine. Data were obtained from three sets of triplicate measurements. becomes evident, even though the same 1O:l (w/w) cholate/ lipid ratio is maintained (Fig. 1B).
Inclusion of low concentrations of supplementary lipids in the micellar solution does not significantly affect the extraction yield, which is constant up to protein concentrations of 6-8 mg/ml (Fig. 2). In this concentration range endogenous components of the membranes can, in the absence of supple-

Acetylcholine Receptor Reconstitution
immune precipitation of the reconstituted AChRs. The results of these experiments are presented in Fig. 4 and 5. Precipitation of AChRs was complete in all samples, as measured by a > 85% precipitation of a ['251](x-BGT trace label. Immune precipitation of the reconstituted membranes does not, however, result in coprecipitation of all the lipid with the AChRs (Fig. 4). Previously we observed in freeze-fracture electron micrographs that reconstituted preparations contained a mixture of large (-0.4 pm) vesicles containing intramembranous particles, i.e. AChRs, and many small (mostly 0.02-0.08 pm) vesicles lacking such particles (11). It is evident from all these results that in the presence of excess lipid, AChRs assemble in the reconstituted membranes at a constant protein/lipid ratio, forming a vesicle population distinct from a population of non-AChR-containing liposomes. The apparent association of lipid with the AChR, according to Fig. 4, amounts to 4.3 k 1.3 mg of lipid/mg of protein. This value, however, is most likely an underestimate, because the trapping of non-AChRcontaining I4C-labeled vesicles in immune precipitates of AChR-containing vesicles is probably smaller than the trapping of these vesicles in immune precipitates in the absence of AChR used as background control in these experiments.
A quantitatively more reliable estimate of the composition of the AChR-containing reconstituted membranes was obtained by isopycnic sucrose gradient centrifugation. ['"Ila-BGT trace-labeled vesicles, when centrifuged to equilibrium on a sucrose gradient, formed symmetrical peaks indicating an apparently homogeneous population of vesicles in terms of density. No AChR was pelleted in these tubes, indicating that aggregation of AChRs without lipid association was not detectable (Fig. 6). The apparent weight fraction of AChRs in the reconstituted membranes was calculated from the density of the peaks (Fig. 7). The results of these experiments are consistent with those involving immune precipitations of radioactive lipids in that they demonstrate an AChR-lipid association at a constant protein/lipid ratio, which at lipid concentrations > 15 mg/ml is independent of the lipid concentration. Under these conditions AChRs assemble with soybean lipids into vesicles containing an apparent weight fraction of 7% AChRs. This is a 5-fold lower packing density suspended in 100 mM NaCI, 2% sodium cholate, 5 mg/ml of soybean lipid, 10 mM Na phosphate buffer, pH 7.4, and incubated for 1 h at 4 "C under gentle shaking. After centrifugation for 15 min in an Eppendorf microfuge, aliquots of the extract were reconstituted at 1.25 mg/ ml of protein in the presence of different concentrations of soybean lipid containing a ['4C]dipalmitoylphosphatidylcholine tracer. The samples were reconstituted by cholate dialysis, and AChR-associated lipid was measured as described in the legend to Fig. 4. than in the native membrane, which contains approximately 35% AChR protein by weight (Fig. 7). When AChRs are reconstituted a t 1. 25  weight percentage of AChRs ( Fig. 6 and 7). Centrifugation of vesicles, formed at 25 mg/ml of soybean lipid in the presence of ['4C-]dipalmitoylphosphatidylcholine, yields a measurement of about 9% for the apparent weight percentage of AChRs in the reconstituted vesicles (data not shown). This is AChR-containing vesicles, reconstituted at 1.25 mg/ml of protein and at different soybean lipid concentrations, were subjected to isopycnic sucrose gradient centrifugation. The apparent weight fraction of AChRs in the reconstituted vesicles was calculated from the observed densities in the peaks (Fig. 6 ) , using the partial specific volume for AChR according to Reynolds and Karlin (1) and the observed isopycnic density for liposomes without AChR formed by cholate dialysis at 25 mg/ml of lipid in the presence of [' ' CC]dipalmitoylphosphatidylcholine. The isopycnic density of native AChR-rich membranes was measured during the isolation of these membranes on a 32-38% (w/w) linear sucrose gradient. Values for reconstituted vesicles containing 5 2 mg/ml of soybean lipid were obtained by isopycnic centrifugation on a 5-3576 (w/w) linear sucrose gradient in a Beckman SW 40 rotor at 255,000 X g for 18 h. AChRs were in each instance solubilized from the native membrane in a 101 weight ratio of cholate/lipid.

TABLE I1
Immune precipitation of reconstituted vesicles by monoclonal antibodies AChRs (1.25 m g / d of protein) were incubated for 1 h at 4 "C with a trace label (lo-' M) of ['251]-a BGT after solubilization of the native membranes in 10 mM Na phosphate buffer, pH 7.4, 100 mM NaCI, 2% sodium cholate, 5 m g / d of soybean lipid. The lipid concentration was subsequently adjusted to 25 mg/ml and reconstituted vesicles were formed by cholate dialysis. Immune precipitation was performed by 2-h incubation at room temperature with a IO-fold excess of antibody in 10 mM Na phosphate buffer, pH 7.4,100 mM NaCI, in the presence or absence of 0.5% Triton X-100. This was followed by 30 min of incubation at room temperature with an excess of goat-antirat serum. Pellets and supernatants were counted in a y-counter. Monoclonal antibody No. 6 is directed against the main immunogenic region on the extracellularsurface of Torpedo AChR (21). Monoclonal antibody No. 32 was raised against AChR from Electrophorus electricus and does not cross-react with Torpedo AChR." The overall fraction of outward-oriented AChRs was determined by radioimmunoassay in the presence and absence of detergent as described in Ref. 11 and was in this set of experiments 72 f 8%. The data are compiled from a duplicate set of triplicate measurements. Triton X-100-treated vesicles in fair agreement with the value calculated from the isopycnic density (7%). We investigated whether AChRs during the reassembly process were incorporated in the reconstituted membranes in such a way that in the same vesicle AChRs are oriented with their toxin-binding sites facing either all outward or all inward, or whether outward-and inwardfacing AChRs are both present in the same vesicles. Table I1 shows that a monoclonal antibody, which binds with high affinity to the main immunogenic region of the Torpedo AChR on the external surface of the membrane (21), precipitated only 69 f 8% of the total AChRs in intact reconstituted vesicles. This percentage corresponded closely with the fraction of AChRs in the preparation which was oriented toward the external surface as measured by its ability to bind ['251]-a-BGT (Table 11, Ref. 11). These data suggest therefore that all AChRs which are assembled in the same vesicle during reconstitution are incorporated in the same orientation.
Stability of AChR Channels during Incorporation into Reconstituted Membranes- Fig. 8A demonstrates the dependence of AChR function in the reconstituted vesicles on the lipid concentration at which reconstitution is p2rformed.
If during reconstitution at 1.25 mg of AChR protein/ml the lipid concentration is decreased below 20 mg/ml, a decrease in carbamylcholine-induced 22Na+ flux/AChR results. This effect is observed in the same range of lipid concentrations at which the packing density of AChRs in the reconstituted vesicles increases (Fig. 6 and 7 ) . An optimal condition for the

Acetylcholine Receptor Reconstitution
retention of channel activity during reconstitution appears to be attained at a lipid/protein ratio of -16 (w/w). Further addition of lipids during reconstitution has no effect, and results only in the formation of increased numbers of non-AChR-containing liposomes, as is evident from a continuous increase in total internal volume. The equilibrium volume/mg of lipid remains constant at high lipid concentrations, but increases over the range of lipid concentrations at which AChRs pack with greater density in the membranes. This most likely reflects an increase in overall internal volume due to the incorporation of a greater fraction of the lipid in larger AChR-containing vesicles rather than in the smaller liposomes ( Fig. SA).
When the lipid/protein ratio is maintained > 16 (w/w), full channel activity is retained during reconstitution at soybean lipid concentrations between 2.5-15 mg/ml, and the equilibrium volume/mg of lipid remains essentially constant (Fig.  SB). Under these conditions AChRs are inserted in the recon-  stituted membranes at a 7% apparent weight fraction even at lipid concentrations below 15 mg/ml (Fig. 9).
We considered the possibility that the decrease in carbamylcholine-dependent "Na+ influx/AChR at suboptimal lipid/protein ratios was due to equilibration of the vesicles with "Na+ during the response as a result of the greater packing density of AChRs rather than due to actual channel inactivation. In order to test this we titrated the carbamylcholine-dependent "Na+ flux response at a suboptimal lipid/ protein ratio with increasing amounts of toxin. In native vesicles the packing density of AChRs is so great that equilibration limits the response, and more than 70% of the AChRs must be blocked by toxin before carbamylcholine-induced Na+ uptake decreases in direct proportion with the addition of toxin (11, 25). Here we found, however, that carbamylcholine-induced "Na+ uptake decreased in direct proportion to blockage of toxin binding sites (Fig. lo), as is observed with reconstituted vesicles formed under optimal conditions (8,ll). This indicates that equilibration of the internal volume of the vesicles with "Na+ did not limit the response, but that the response was proportional to the amount of active AChR present. Extrapolation to the intercept on the abscissa in Fig.  10 yields a measurement for the fraction of the AChRs oriented with their toxin-binding sites on the external surface of the reconstituted vesicles. Both vesicles formed at 10 mg/ml of lipid, and at 25 m g / d of lipid in the presence of 1.25 mg/ ml of protein contain the same proportion of outward-oriented ['251]a-BGT binding sites, namely 70 +. 4%. It is clear from these observations that the decrease in the carbamylcholineregulated "Na+ flux response at suboptimal lipid/protein ratios is indeed due to inactivation of AChR channels during the reconstitution process rather than due to equilibration of the vesicles during the response or to an inverse orientation of AChRs caused by reconstitution at suboptimal lipid concentrations.
We considered the possibility that the inactivation of channels might be reversible. In order to test this we solubilized membranes formed at a suboptimal lipid/protein ratio, in 2% cholate in the presence of sufficient supplementary soybean 22 lipids to allow reconstitution for a second time under optimal conditions. Table I11 shows that only the channel activity which survived the first reconstitution under suboptimal conditions can be recovered after a subsequent reconstitution under optimal conditions. The decrease in the carbamylcho-

TABLE 111
Irreversible inactivation of AChR channels by reconstitution a t a suboptimal soybean lipid concentration Reconstituted vesicles containing 1.25 mg/ml of protein were formed at soybean lipid concentrations of 5 mg/ml and 20 mg/ml, respectively. Uptake of "Na' at IO" M carbamylcholine was measured, and aliquots of each batch of vesicles were dissolved in 2% sodium cholate, 100 m~ NaC1, IO mM Na phosphate buffer, pH 7.4, supplemented with soybean lipids to give a final lipid concentration of 20 m g / d in both aliquots. The aliquots were then reconstituted for a second time by cholate dialysis, and the resulting vesicles were again assayed for carbamylcholine-induced uptake of "Na+. Measured concentrations of ['251]o(-BGT binding sites were identical for the two samples both after the first reconstitution (7.3 X IO" M) as well as after the second reconstitution (4.8 X 10" M). Isopycnic sucrose gradient centrifugation of vesicles reconstituted at a suboptimal soybean lipid concentration and containing irreversibly inactivated AChRs. Reconstituted vesicles containing 1.25 mg/ml of protein were formed at soybean lipid concentrations 5 m g / d and 20 mg/ml, respectively. After assay for carbamylcholine-regulated 22Na+ uptake (Table 111), aliquots of each batch of vesicles were dissolved in 2% sodium cholate, 100 mM NaCI, 10 mM Na phosphate buffer, pH 7.4, supplemented with soybean lipids to give a final lipid concentration of 20 m g / d in both aliquots. Both samples were then reconstituted by cholate dialysis for a second time and analyzed by isopycnic sucrose gradient centrifugation. A, vesicles formed at 5 m g / d of soybean lipid during the first reconstitution; and B, vesicles formed at 20 mg/ml of soybean lipid during the first reconstitution. and dissolved in 1 0 0 mM NaC1,2% sodium cholate, 5 mg/ml of soybean lipid in 10 mM Na phosphate buffer, pH 7.4. Aliquots of 50 pl were applied to 4.8 ml of 5-20% (w/w) linear sucrose gradients in 10 mM Na phosphate buffer, pH 7.4.100 m~ NaCl, 2% sodium cholate, 5 mg/ ml of soybean lipids. Centrifugation was performed in a Beckman SW50.1 rotor at 300,000 X g for 5 h, fractions of 0.21 ml were collected after puncturing the tubes, and they were counted in a y-counter. A, vesicles reconstituted at 5 m g / d of soybean lipid; and B, vesicles reconstituted at 20 mg/ml of soybean lipid. line-induced 22Na+ flux response at suboptimal lipid/protein ratios is thus due to permanent denaturation of a fraction of the AChR channels during the reconstitution process. Isopycnic sucrose gradient centrifugation of vesicles formed f i s t under suboptimal reconstitution conditions and subsequently under optimal conditions reveals an apparent packing density of AChRs intermediate between the isopycnic density expected under optimal conditions and that measured after the f i s t reconstitution under suboptimal conditions (Fig. 11). We cannot, however, distinguish whether this peak represents a single population of vesicles or whether it results from the summation of equal amounts of two components, banding at 20% (w/w) and 16% (w/w) sucrose, respectively.

Soybean lipid con-
Denaturation of AChR channels might be acccmpanied by aggregation of the inactive AChRs, which could account for the relatively higher packing density of the AChRs during the second reconstitution. We tested whether aggregation of the reconstituted AChRs was detectable in cholate-lipid prior to a second reconstitution. First we dissolved the vesicles in 2% cholate and 5 mg/ml of soybean lipid, and we then performed sucrose gradient centrifugation essentially as previously described @), except that 5 mg/ml of lipid was present in the sucrose gradient. Both AChRs reconstituted under optimal as well as suboptimal conditions were predominantly present as dimers (Fig. 12, Ref. 8). Aggregates near the bottom of the gradients below the dimer peaks were not discernible. Therefore, we did not detect aggregates which persisted in cholatelipid solution.
We also attempted to detect aggregated AChRs in the

CONCENTRATION OF 2-MERCAPTOETHANOL (mM)
reconstituted membranes by cross-linking with glutaraldehyde. Reconstituted vesicles were incubated for 5 min at 4 "C with 0.4% glutaraldehyde. The reaction was quenched with an equal volume of 1 M glycine in 10 mM Na phosphate buffer, pH 7.4, 100 mM NaC1,2% Triton X-100,0.5 M 2-mercaptoethanol, and the reaction products were analyzed by sucrose gradient centrifugation as previously described (8) incubation at 4 "C and dissolved in 2 8 Triton X-100 in the presence of the indicated concentrations of 2-mercaptoethanol. Sucrose gradient centrifugation was then performed as previously described (8). Relative amounts of dimers and monomers were measured by integration of the peaks, and the percentage of conversion of AChR dimers into monomers was calculated. AChRs subjected to suboptimal conditions, as well as control AChRs, displayed the same relative amounts of dimers (80 k 4%) and monomers (20 f 4%) in the absence of reducing agents. Reduction with 250 mM 2-mercaptoethanol, supplemented with 50 mM dithiothreitol, resulted in 72 & 4% conversion of dimers into monomers for AChRs which had been exposed to low lipid conditions during reconstitution, compared to 94 & 5% dimer to monomer conversion for control AChRs.   The calculated values were obtained using 250,000 as the molecular weight of the AChR monomer (1) and assuming an average M, = 750 for soybean lipid.

AChR-rich
* Calculations were based on the assumption that the intramembranous portion of the AChR molecule can be considered as a cylinder of 30 A diameter (29) and using 60 A* as estimate for the area occupied by a lipid molecule (36). N.D.. not determined.
linked AChRs were detected as monomers. About 45% of the AChRs which had been exposed to suboptimal reconstitution conditions were detected as a continuous smear in the gradient below the AChR dimer peak after glutaraldehyde treatment.
In contrast, only about 18% of the control AChRs (reconstituted under optimal conditions) displayed this sedimentation behavior after cross-linking under identical conditions (data not shown). This difference in cross-linking efficiency could, however, be fully accounted for by the greater packing density of AChRs reconstituted under suboptimal conditions (13% apparent weight fraction as compared to 7%, Fig. 7). Therefore, although aggregates of denatured receptors could not be readily demonstrated by centrifugation in cholate-lipid, their existence in the reconstituted membranes can still not be excluded.
We observed that the disulfide bond between the S subunits of the AChR dimer became less susceptible to reduction once AChRs had been subjected to suboptimal reconstitution conditions. Reducing agents, such as 2-mercaptoethanol and dithiothreitol, cause an efficient conversion of AChR dimers into monomers by reduction of the disulfide bond between the S subunits (4, 5 , 8). Fig. 13 demonstrates that this dimer to monomer conversion occurs less readily with AChRs after exposure to suboptimal lipid concentrations during reconstitution than with AChRs maintained under optimal conditions. The lower efficiency of AChR dimer reduction reflects a permanent conformational alteration of the protein, which persists after solubilization of the membranes in Triton and provides a protected configuration for the 6-8 disulfide bond. This is surprising, since this disulfide is not involved directly in channel function (8). However, since it must be at the periphery of adjacent AChR monomers, an alteration in its reactivity with reducing agents indicates that the association of adjacent AChR monomers is altered. This evidence, like the only partially reversible packing density of AChRs after exposure to suboptimal lipid conditions, suggests that this denaturation of the cation channel is accompanied by changes in AChR conformation that may affect its interaction with other AChRs. However, denaturation of the channel and denaturation of the area around the disulfide bond-linking monomers are probably not functionally related (8).

DISCUSSION
A schematic representation of the solubilization-reassembly cycle is shown in Fig. 14. The approximate molar compositions of the different AChR-lipid assemblies a t each stage of this process, calculated on the basis of our data, are presented in Table IV. By electron microscopy of the native membrane, AChRs appear as densely packed 85-a diameter doughnuts (26,27) which protrude as mushroom-like projections from the extracellular surface of the membrane (28,29). We suggest that AChRs solubilized in cholate-lipid mixtures are surrounded by an annulus of lipid in the form of a bilayer, and that an annulus of cholate solubilizes this infinitesimal patch of AChR-containing membrane (Fig. 14). Mixed micelles of cholate and lipid are thought to consist of various sizes of small lipid bilayer discs surrounded by an annulus of cholate (Ref. 24,Fig. 14). We propose that during dialysis some of the cholate molecules dissociate from these bilayer discs, leaving a hydrophobic region which may then fuse with another cholate-depleted region of lipid bilayer. About 68% of the AChRs can be extracted in mixed micelles of cholate and soybean lipids. It is not clear whether the remaining unsolubilked AChRs are present in the native membranes in a different structural arrangement. A 43,000-molecular weight protein has been suggested to associate with the cytoplasmic surface of AChRs (30). We used alkaline extraction methods to remove this protein (11,31,32), but this did not increase the extent of the extraction (data not shown).
The channel denaturation observed at elevated concentrations of cholate ( Fig. 1B) is probably due to the concentration of lipid-free cholate micelles which displace the lipid protecting the AChR channel. Although even at 2% cholate, 2 mg/ml of lipid, cholate is present in 20-fold molar excess over the lipid (Table IV), the absolute concentration of disruptive micelle configurations probably becomes significant above 3% cholate. Heidmann et al. (12) have shown that in sodium cholate in the absence of protective lipids, the AChR is stabilized in a low affinity state for agonists, which can no longer be triggered into a slow interconversion to a high affinity state by agonists or local anesthetics. The characteristic allosteric properties of the membrane-bound AChR could be preserved by the presence of protective lipids in the micellar solution. These observations, like ours, suggest that the AChR channel must be shielded from the detergent molecules and the aqueous medium by a protective annulus of lipids as indicated in Fig. 14. Taking into account the 20-fold molar excess of cholate over lipid in the micellar solution (Table IV), it appears evident that under optimal conditions (2% cholate, 2 mg/ml of soybean lipid) the binding affinity of the lipids for the AChR must be significantly higher than the binding affinity of cholate for the AChN. Irreversible denaturation of the ion channel at elevated cholate concentrations might be due to: 1) a direct interaction between cholate and specific regions of the protein; 2) denaturation of hydrophobic regions of the channel by direct contact with the aqueous environment; or 3) displacement of lipids essential to maintain a native conformation of the channel. Local anesthetic-like effects of the detergent on the channel (33,34) could not account for this irreversible inactivation, since such effects would also take place under optimally protective conditions, because the cholate is present in 20-fold molar excess. This large excess of detergent would probably also prevent access of water to a delipidated hydrophobic protein moiety. O w data therefore favor the third possibility.
Our observation that octylglucoside denatures the channel in the presence (or initially in the absence) of supplementary lipids is a t variance with a recent report by . They reported reconstitution of AChRs after solubilization and purification of the AChRs in octylglucoside in the absence of supplementary lipids. The incorporation of AChRs into reconstituted lipid vesicles is well documented in this report. However, since these investigators used a high AChR concentration and a low lipid/protein ratio (-2) during their Acetylcholine Receptor Reconstitution reconstitution, the carbamylcholine-induced "Na+ flux which they observed may result from only a very small fraction of intact AChR channels.
Adequate data are not available yet to allow a clear description of the molecular interactions which occur during the reassembly process. We hope, however, that our data will provide a useful conceptual framework for further studies on the mechanism of AChR reconstitution. Any hypothesis concerning the molecular mechanism of vesicle formation during the reconstitution process must accommodate the following experimental observations: 1) incorporation of a large fraction of the lipid into small unilamellar liposomes which do not contain AChRs; 2) incorporation of AChRs into large unilamellar vesicles at a constant AChR/lipid ratio independent of the AChR/lipid ratio in the micellar solution, when lipid is present in excess; and 3) incorporation of AChRs in the reconstituted membranes with their extracellular portions preferentially oriented outward and with all the AChRs in a single vesicle oriented in the same direction. It is likely that AChR-AChR interactions play an important role in the reassembly process. One hypothetical mechanism which may account for our observations employs the idea of "pseudo-crystalline spherical arrays" of AChRs being formed during the nucleation stage of the reconstitution process. This hypothesis postulates that as a result of interactions between the extracellular surfaces of the AChRs, they pack into spherical arrays, in most cases (-70%) with their [lZ5I]a-BGT binding sites oriented toward the exterior of the nascent sphere. We suppose that subsequent to the formation of such pseudo-crystalline spherical arrays of AChRs, lipid is incorporated until a surface tension is reached which allows the complex to be sealed as a stable vesicle. We suggest that under conditions of limiting lipid, vesicles would close at unusually high surface tensions which would denature a fraction of the channels. Although this theory can account for a number of experimental observations, we cannot exclude other possible mechanisms. Moreover, independent mechanisms might account for the constant AChR/lipid ratio and the orientation of the AChRs in the vesicles. For example, electrostatic repulsion between AChRs might limit their packing density, while the differential surface tension of the inner and outer layers of the lipid bilayer might cause AChRs to incorporate in a particular orientation.
AChRs are &fold less densely packed in the reconstituted vesicles than in the native membranes. This prevents equilibration of the reconstituted vesicles during the permeability response, so that desensitization of the AChR is the only factor limiting the carbamylcholine-regulated "Na+ influx (Fig. 10, Refs. 8 and 11). From the values in Table IV one can calculate that vesicles of about 4000-A diameter (11) will contain approximately 200 AChRs/vesicle. An important caveat for the interpretation of our data is that the AChR-enriched membranes used in our experiments contain some additional membrane components (11). Although the purity of our preparations has not been taken into account in calculating the compositions of the membranes, this inaccuracy will still permit a relative comparison between the packing density of AChRs in the native membrane with that in the reconstituted membranes. Moreover, the calculated value of 7% for the apparent weight percentage of AChRs in the reconstituted vesicles is in fair agreement with the 9% value as measured by inclusion of a radioactive lipid tracer during the reconstitution.
The conditions that lead to irreversible inactivation of AChR channels during reconstitution at suboptimal lipid/ protein ratios remain still unresolved. Our observations suggest, however, that this channel denaturation does not result from limited availability of an essential lipid component. Channel inactivation during reconstitution would probably be a reversible process if it were due to limited amounts of an essential chemical factor, since AChR channels survive in micellar solution in the presence of even lower lipid concentrations. Formation of aggregates is, although not excluded, not readily demonstrable. We suggest that denaturation of the channel per se or impairment of its gating mechanism may result from unfavorable surface tension and/or electrostatic interactions during reassembly under limiting lipid conditions. This loss of channel function is probably concomitant with a substantial permanent alteration in AChR conformation reflected in the lowered susceptibility of the 8-8 disulfide bond to reducing agents added during solubilization of the reconstituted membranes in detergent.
The AChR represents an archetype for other neurotransmitter and hormone receptors and is in fact the fist neurotransmitter receptor to be incorporated into model membranes. Stabilization of the tetrodotoxin-binding component of the sodium channel from Electrophorus electricus by mixed lipid-detergent micelles has been reported by Agnew et al. (35). We believe that establishing optimal conditions for the stability of the AChR channel in mixed micellar solution, and a careful investigation of the molecular interactions occurring during the reassembly process may be of general use for future studies on other transmembrane proteins.