Expression and renaturation of the N-terminal extracellular domain of torpedo nicotinic acetylcholine receptor alpha-subunit.

The N-terminal extracellular region (amino acids 1-209) of the alpha-subunit of the nicotinic acetylcholine receptor (nAChR) from Torpedo marmorata electric tissue was expressed as inclusion bodies in Escherichia coli using the pET 3a vector. Employing a novel protocol of unfolding and refolding, in the absence of detergent, a water-soluble globular protein of 25 kDa was obtained displaying approximately 15% alpha-helical and 45% beta-structure. The fragment bound alpha-[3H]bungarotoxin in 1:1 stoichiometry with a KD value of 0.5 nM as determined from kinetic measurements (4 nM from equilibrium binding). The kinetics of association of toxin and fragment were of second order, with a similar rate constant (8.2 x 10(5) M-1 s-1) as observed previously for the membrane-bound heteropentameric nAChR. Binding of small ligands was demonstrated by competition with alpha-[3H]bungarotoxin yielding the following KI values: acetylcholine, 69 microM; nicotine, 0.42 microM; anatoxin-a, 3 miroM; tubocurarine, 400 microM; and methyllycaconitine, 0.12 microM. The results demonstrate that the N-terminal extracellular region of the nAChR alpha-subunit forms a self-assembling domain that functionally expresses major elements of the ligand binding sites of the receptor.

So far, most of our knowledge on the three-dimensional structure of neuroreceptors is based on a combination of electronmicroscopical, biochemical and immunological data obtained for the nicotinic acetylcholine receptor (nAChR) 1 from the electric ray Torpedo (1,2). These studies have elucidated the overall dimensions of the receptor protein (3), its position with respect to the surrounding lipid bilayer, the locations of functional domains and amino acid residues belonging to the integral ion channel (4,5), its gating structures (6,7), and the binding sites for several classes of ligands (1,2,(7)(8)(9)(10)(11)(12). However, a high resolution three-dimensional structure of the nAChR or any other neuroreceptor is still missing. If available, it could provide a molecular correlate for the recognition function of the receptor and thereby also for rational drug design.
Prompted by the fact that all attempts to crystallize the detergent-solubilized whole nAChR protein have been unsuccessful for the past more than 20 years, we have begun to overexpress selected domains of the receptor in bacterial expression systems. If successfully renatured, the domains that are not transmembranous should be water-soluble and thus should provide a better material for protein crystallization than the detergent-solubilized whole receptor protein. Moreover, if small enough in size, such fragments should be suited for multidimensional NMR analysis (13). As presented here for the N-terminal extracellular region of the nAChR ␣-subunit, we attempted and achieved expression as a water-soluble globular protein that displays ligand binding properties comparable to or better than those of the SDS gel-isolated ␣-subunit. To achieve appropriate renaturation, we developed a protocol for the complete unfolding (by means of chaotropic agents and disulfide-reducing agents) and refolding (by means of an oxido shuffling system and L-arginine as structure-stabilizing agent) of the expressed polypeptide (14). The experimental conditions were such that self-organization of the unfolded protein was favored at the expense of (unwanted) aggregation and denaturation. In this way, we are able to prepare large quantities of the functional ligand binding domain from inclusion bodies of transformed Escherichia coli bacteria. Our results confirm that the N-terminal extracellular domain indeed harbors major elements of the ligand recognition function of the nAChR, as has long been suggested on the basis of affinity labeling and immunological studies (for recent reviews, see Refs. 1, 2, 8, 9, 15, and 16).

EXPERIMENTAL PROCEDURES
Materials-nAChR-rich Torpedo membrane fragments (3 mg/ml; 5 nmol of ACh sites/mg) were obtained by homogenization of Torpedo marmorata electric organ, followed by several centrifugation steps and alkaline treatment as described (11). N-␣-[propionyl-3 H]bungarotoxin (54 -68 Ci/mmol) was obtained from Amersham Pharmacia Biotech. All other chemicals were of analytical grade and were obtained from major commercial sources.
Expression of ␣ nAChR  in E. coli-A cDNA fragment coding for amino acids 1-209 of the nAChR ␣-subunit from T. marmorata was obtained by polymerase chain reaction amplification from a full-length cDNA clone (provided by Dr. Kretschmer, Leverkusen, Germany) using primers complementary to the nucleotide sequences 1-18 and 610 -627. The primers were supplemented with restriction sites for NdeI and BamHI, and using the appropriate restriction endonucleases, the purified cDNA fragment was subsequently ligated into pET3a vector. Plasmids were isolated according to the method of Birnboim (17) and subsequently used to transfect competent E. coli bacteria (strain BL 21 (DE3) pLysS). Expression of the ␣ nAChR 1-209 fragment was achieved by means of the T7 expression system following IPTG induction of transformed bacteria. The ␣ nAChR 1-209 polypeptide was found almost exclusively in E. coli inclusion bodies. Bacteria were disrupted by ultrasonic treatment in extraction buffer containing Tris-HCl (10 mM), NaCl (10 mM), MgCl 2 (10 mM), pH 8.0, and inclusion bodies were harvested by centrifugation at 12,000 ϫ g for 50 min and subsequently washed with guanidinium chloride (3 M), NaCl (0.1 M), Tris (0.01 M), pH 8.4.
SDS Gel Electrophoresis and Isoelectric Focusing-SDS-PAGE (12.5% total acrylamide) was performed according to Laemmli (20). Isoelectric focusing was performed using nonlinear pH 3-10 Immobi-line® DryStrips in a MultiphorII electrophoresis system from Amersham Pharmacia Biotech according to a protocol of the supplier. Theoretical pI calculations were performed using Expasy facilities at the Swiss Institute of Bioinformatics (21).
Sequence Analysis-After SDS-PAGE, the ␣ nAChR 1-209 protein was electroblotted onto polyvinylidene fluoride membranes as described previously (22), and the membrane was directly applied to an Applied Biosystems pulsed liquid-phase (ABI 477A) sequencer. Aliquots of the phenylthiohydantoin-amino acids released from the sequencer at each cycle were analyzed on-line using the ABI 120A reverse-phase HPLC system (Applied Biosystems).
Mass Spectroscopy-Peptide samples were dissolved in 5 l of 50% acetonitrile/0.1% trifluoroacetic acid and sonicated for 5 min. Aliquots of 0.5 l were applied onto a target disc and allowed to air-dry. Subsequently, 0.3 l of matrix solution (1% w/v ␣-cyano-4-hydroxycynnamic acid in 50% acetonitrile/0.1% trifluoroacetic acid v/v) was added to the sample, which was again allowed to dry. Peptide spectra were obtained using a Bruker Biflex time-of-flight mass spectrometer.
Titration of Free Sufhydryl Groups-Free sufhydryl groups in a 300 M solution of ␣ nAChR 1-209 protein were determined in a spectrophotometric assay using Ellman's reagent (dithiobis(nitrobenzoic acid)). The presence of free sulfhydryl groups was detected by equimolar formation of a colored product, which was measured at 405 nm with a sensitivity of approximately 1 nM (23).
Determination of Extinction Coefficient-The extinction coefficient was obtained from measurement of the absorbance of a solution of ␣ nAChR 1-209 fragment (40 g/ml, i.e. 1.6 M) at a wavelength of 280 nm. The theoretical values were obtained using the program PEPTIDES-ORT (21).
CD Measurements-CD spectra were measured at 24°C using a Jasco J-710 Spectropolarimeter. The cell path length was 0.1 mm. The scan speed was 50 nm/min; the response time was 4 s, and the range was 190 -250 nm. Optical activity was expressed as mean residue ellipticity in degrees ϫ cm 2 ϫ dmol Ϫ1 , based on a mean residue weight of 118. The concentration of ␣ nAChR 1-209 fragment was 40 M, and secondary structure was calculated using the CONTIN and JFIT programs (24,25).
Ligand Binding Studies-Radioligand binding studies were performed using a DEAE filter disc assay (26). Briefly, appropriate concentrations of ␣ nAChR 1-209 fragment in a buffer containing Tris-HCl (10 mM), NaCl (50 mM), sucrose (10 mM), pH 7.4, were incubated with selected amounts of N-␣-[propionyl-3 H]bungarotoxin for 3 h at room temperature. Filters were subsequently washed twice for 1 min with 0.1% Triton X-100 in above incubation buffer. The concentration of free radioligand was determined from the filtrate, that of bound ligand from the filter, for which purpose these were dried and dissolved in a tissue solubilizer (Solvable, Packard Instrument Co.) followed by liquid scintillation counting. Competition studies with small nicotinic ligands were performed at fixed concentrations of ␣-[ 3 H]bungarotoxin (6 nM) and ␣ nAChR 1-209 fragment (1 nM). The amount of bound small ligand was then calculated from the reduction in ␣-[ 3 H]bungarotoxin binding, as compared with control incubations in the absence of small ligand.
For kinetic experiments, aliquots were removed at indicated times and analyzed for their content of receptor/ligand complex by liquid scintillation counting. After complete saturation of toxin binding was achieved, a large excess of nonlabeled ␣-bungarotoxin (to a final concentration of 10 M) was added. The second order rate constant of receptor fragment/toxin-complex formation (k on ) was derived according to Equation 2 from the slope of the plot, against the incubation time in s (26) (C b eq , concentration of bound ligand at equilibrium; C b (t) , concentration of bound ligand at time t; R 0 , total concentration of binding protein; L 0 , total concentration of radioligand).
The first order rate constant of dissociation was calculated from Equation 3 (calculations according to Ref. 26). Division of k off by k on gave the K D value.
The kinetic data obtained determined the appropriate conditions for competition studies with small nicotinic ligands. In these experiments, the concentrations of Torpedo ␣ nAChR 1-209 fragment and tritiated ␣-bungarotoxin were kept constant at 1 and 6 nM, respectively. K I values were calculated from the IC 50 concentrations of competing ligands according to Cheng and Prusoff (27). Data were fitted using appropriate equations of the program Origin 5.0.

RESULTS
Expression, Unfolding, and Refolding of the ␣ nAChR 1-209 Fragment-After induction by IPTG, the ␣ AChR 1-209 -pET 3atransformed competent E. coli bacteria expressed large quantities (100 -200 mg from 1 liter of bacterial culture) of Torpedo ␣ nAChR  , largely in aggregated, denatured state as inclusion bodies. Overexpression was demonstrated in SDS gels of total bacterial extracts and of washed and solubilized inclusion bodies (Fig. 1). In particular, as shown in Fig. 1, lane 4, the washed inclusion bodies contained a very high percentage of ␣ AChR 1-209 .
To obtain the ␣ AChR 1-209 fragment in water-soluble and purified form, we developed an unfolding/refolding procedure that made use of strategies recently developed for other pro- teins overexpressed in E. coli as inclusion bodies (14,28,29). After extensive washing, in order to remove bacterial impurities, the inclusion bodies were incubated under moderate alkaline conditions with a large excess of chaotropic and disulfide-reducing agents (6 M guanidinium chloride, 0.1 M dithioerythritol, pH 8.5). The completely unfolded polypeptide was then dialyzed against 6 M guanidinium chloride, 10 mM EDTA, 100 mM Tris, in order to remove the disulfide-reducing agent and to prevent metal ion-catalyzed reoxidation. The solution was then diluted 200-fold, so as to provide optimal conditions for intramolecular interactions (self-organization) as compared with intermolecular interactions (aggregation). The renaturation buffer also contained an oxido shuffling system (5 mM reduced glutathione and 0.5 mM oxidized glutathione) and L-arginine (1 M) as a protein-stabilizing agent. Incubation was performed at 10°C for 24 h. Finally, the renatured ␣ nAChR 1-209 fragment was dialyzed against storage buffer consisting of Tris-HCl (10 mM), NaCl (100 mM), sucrose (10 mM), pH 7.4, and concentrated by ultrafiltration to the wanted level (up to 10 mg/ml, i.e. 400 M). No free sulfhydryl groups were detectable in the concentrated renatured solution by Ellman's reaction.
Biochemical and Biophysical Characterization of the Renatured ␣ nAChR  Fragment-As a result of the refolding procedure described above, a single polypeptide of apparent molecular mass of approximately 26 kDa was obtained, as demonstrated by SDS-PAGE (Fig. 1, lane 6). The band was electroblotted onto a polyvinylidene fluoride membrane and subsequently submitted to automated Edman degradation. A single sequence was obtained (SEHETRLVANLLENY) with an initial yield of 8 pmol. It corresponded to the first 15 amino acids of the N-terminal region of the Torpedo nAChR ␣-subunit (57). The molecular mass of the fragment protein, as determined by matrix-assisted laser desorption/ionization mass spectroscopy, was 24,896.2 Da, which conforms quite well with the calculated mass of 24,796.3. The difference of 100 is not unusual, as it may be due to random acrylamide adducts on cysteines (ϩ 70) and/or ionization involving K ϩ ions (ϩ 39); the accuracy of the method in this range of masses is approximately Ϯ50 (58). The mass spectrogram (not shown) did not indicate the presence of any significant impurities.
Additional purification schemes involving size exclusion gel permeation materials (Sephadex G50, G75, and G100, and Superdex 75) and subsequent SDS-PAGE analysis of fractions indicated a monomeric state of the Torpedo-␣nAChR 1-209 fragment (data not shown). UV absorption measurements at 280 nm of a 10 M solution of Torpedo-␣nAChR 1-209 fragment yielded an extinction coefficient of 61,000 Ϯ 500 cm 2 mM Ϫ1 , which conforms satisfactorily well to the value calculated from the known content of aromatic amino acid residues (59,840 cm 2 mM Ϫ1 ) (21).
After the renatured polypeptide was identified as the ␣ 1-209 fragment of Torpedo nAChR, it was further characterized by biochemical and biophysical means. Isoelectric focusing (not shown) yielded an isoelectric point of 5.5, which is close to the value of 5.4 calculated on the basis of the amino acid composition (21).
Taken together, the biophysical data suggest that the renaturation procedure resulted in a refolded ␣ nAChR 1-209 fragment that appeared to be structurally homogenous and to display a secondary structure composition that is not in contradiction to previous predictions.
Interaction of ␣ nAChR  Fragment with ␣-Bungarotoxin-␣-Bungarotoxin, the major neurotoxin in the venom of the elapidae Bungarus multicinctus is an established highaffinity competitive inhibitor of the binding of acetylcholine to Torpedo nAChR. We employed the toxin to test whether renaturation of the ␣ nAChR 1-209 fragment restored the ligand binding function of the fragment. Binding studies were performed with a filter disc assay (26) employing N-␣-[propionyl-3 H]bungarotoxin as radioligand. The assay makes use of the acidic nature of the fragment and the fragment-toxin complex in that these are collected by binding to a DEAE-cellulose filter matrix, whereas the free toxin is washed through. The assay is applicable only to complexes that do not dissociate to any significant extend during the time of filtration and washing (1-3 min). Fig. 3 shows a typical binding curve, the related Scatchard plot, and a binding capacity determination for the ␣ nAChR 1-209 fragment. As is demonstrated by the linearity of the Scatchard plot, the fragment preparation exhibited a single class of binding sites for the toxin, with an equilibrium dissociation constant K D ϭ 4 Ϯ 0.8 nM. The B max value derived from the Scatchard plot (2.7 nM) is very close to the concentration of fragment employed in the assays as determined by the bicinchoninic acid method (3 nM), demonstrating a stoichiometry of 1:1 for the toxin-fragment complex. Clearly, the ␣ nAChR 1-209 fragment binds the neurotoxin with high affinity, suggesting that the water-soluble renatured ligand binding domain assumes a conformation that in regard to toxin binding is close to the conformation in the native ␣-subunit. Furthermore, the fragment preparation appeared to consist of a single homogenous population of toxin binding sites.
Using the same DEAE filter disk assay as for the equilibrium binding experiments described above, the kinetics of association and dissociation of toxin and fragment were assayed. As shown in Fig. 4, the association kinetics were of second order, yielding an association rate constant k on ϭ 8.2 ϫ 10 5 M Ϫ1 s Ϫ1 .   in Tris-HCl (10 mM), NaCl (50 mM), pH 7.4, was prepared, and the CD spectrum was obtained as described under "Experimental Procedures." Using the CONTIN and JFIT programs (24,25), we calculated 15% ␣-helical and 45% ␤-sheet structure. These values imply a relatively low content of ␣-helix as compared with those obtained by a variety of methods (see Table I) for the membrane-bound and solubilized whole Torpedo nAChR, suggesting 20 -35% ␣-helical and 30 -55% ␤-pleated sheet. This rate constant is very similar to that obtained for the association of neurotoxin and the native membrane-bound receptor protein (26). Dissociation of toxin-fragment complexes was studied following addition of a 300-fold excess of unlabeled toxin to the toxin-fragment reaction mixture. The resulting dissociation kinetics were of first order, yielding a rate constant k off ϭ 0.4 ϫ 10 Ϫ3 min. The quotient of k off and k on , 0.5 nM, is lower than the K D value of 4 nM obtained from equilibrium binding experiments. The difference can be explained taking into account the second order kinetics of association: in the range of lower concentrations, on is slower than in the case of near-saturating concentrations (38,59), which results in apparently lower concentrations of bound radioligand.
As summarized in Table II the K D value of 0.5 nM may be compared with those obtained previously with the same toxin and the whole solubilized receptor protein (K D ϭ 0.01 nM) (39), the detergent-solubilized isolated ␣-subunit (K D ϭ 100 nM) (40,41), or synthetic peptides matching in sequence the region around cysteines 192 and 193 (K D ϭ 10 M) (42). From the dissociation rate constant, a half-life of the toxin-fragment complexes off ϭ 30 min was obtained.
Equilibrium competition studies were performed using N-␣-[propionyl-3 H]bungarotoxin (6 nM) as radioligand and methyllycaconitine, anatoxin-a, nicotine, D-tubocurarine, and acetylcholine as competing nicotinic ligands. The concentration of Torpedo ␣ nAChR 1-209 fragment was 1 nM. Fig. 5 shows results: K I values were calculated from the IC 50 concentrations of competing ligands (Fig. 4) according to Cheng and Prusoff (27) as 1.2 ϫ 10 Ϫ7 , 3.0 ϫ 10 Ϫ6 , 4.2 ϫ 10 Ϫ7 , 4.0 ϫ 10 Ϫ4 , and 6.9 ϫ 10 Ϫ5 M, for methyllycaconitine, anatoxin-a, nicotine, tubocurarine and acetylcholine, respectively (see Table II). As controls, atropin (a ligand of muscarinic acetylcholine receptors) and procain (a local anesthetic with binding sites within the open channel pore of the receptor) were included. Neither substance was able to displace tritiated ␣-bungarotoxin binding to the ␣ nAChR 1-209 fragment. DISCUSSION The Torpedo nAChR has been established as the prototypic model for fast ligand-gated ion channels in the central and peripheral nervous system due to its abundance in electric organs of electric fish (Torpedo and Electrophorus), which made it accessible to detailed biochemical and biophysical studies. To date, it has been impossible to obtain high resolution structural information at the atomic level due to resistance of the receptor protein to crystallization attempts (for recent reviews, see Refs. 1, 2, 8, 9, 15, and 16).
Therefore, attention has focused on the expression of relevant functional domains of the nAChR (such as the large extracellular loops of nicotinic ␣-subunits, which contain most of the determinants for binding of agonists and competitive antagonists) in heterologous systems, always with the hope of finding a general method to obtain substantial amounts of material, suitable for structural analysis at the atomic level by NMR or x-ray/crystallography.
Large scale expression of recombinant proteins in E. coli often results in confinement of the desired protein in so-called inclusion bodies, which concentrate the heterologous protein in a denatured and aggregated conformation. Recently, methods have been developed to restore the functional native conformation of recombinant proteins by ensuring conditions that allow correct refolding and disulfide bond formation starting from  B, the Scatchard plot of the binding data shown in A is linear, suggesting a single binding site for ␣-bungarotoxin and confirming the concentration of the ␣nAChR 1-209 fragment (3 nM) with a similar value for maximal binding sites (B max ϭ 2.7 nM). C shows the linear relationship of protein concentration of ␣ nAChR 1-209 fragment solutions (see under "Experimental Procedures") and radioligand binding at saturating concentrations of radioligand (500 nM). The slope of the plot is 0.7, again indicating a highly homogeneous solution of the protein in a native conformation and a 1:1 stoichiometry of ␣-bungarotoxin-binding. solubilized denatured material (14,28,29,43).
Using an oxido shuffling system with reduced and oxidized glutathione and L-arginine as a stabilizing agent, we have been able to obtain large amounts of an ␣-nAChR fragment that is soluble in aqueous buffers without the use of detergents and contains the first 209 N-terminal amino acids, including relevant binding sites for agonists and competitive antagonists, as demonstrated by radioligand binding experiments. The affinity for ␣-bungarotoxin (K D value, 0.5 nM) is lower than for whole membrane-bound receptor (0.01 nM) (26,39), but higher than for detergent-solubilized isolated ␣-subunit (100 nM) (40,41). These differences in affinity probably arise from the fact that in native nAChR, ligand binding sites are composed of discontinuous subsites, which contribute to the actual ligand binding domain. Most of these determinants reside within ␣-subunits, although there is also some contribution from neighboring (mainly ␤) subunits. Moreover, evidence from photoaffinity labeling experiments seemed to indicate location of toxin binding sites at the interface of subunits (for recent reviews, see Refs. 1, 2, and 16). The missing interaction of ligands with these additional subsites could be the explanation for the 40fold lower affinity of ␣-bungarotoxin to ␣ nAChR 1-209 fragment compared with the native pentameric receptor. On the other hand, the affinity of ␣-bungarotoxin to isolated and solubilized nAChR ␣-subunit is even 200-fold weaker when compared with ␣ nAChR 1-209 fragment, probably because the use of detergents is disturbing crucial secondary structure elements (see also  Table II).
In native nAChR, there are different agonist-binding states, namely low and high affinity states. Events leading to opening of the receptor intrinsic ion channel involve sequential binding of two molecules of agonist to the two ␣-subunits at relatively low affinity (EC 50 of 0.1-10 M, depending upon agonist used), followed by conformational changes that are associated with an approximately 100 -1000-fold increase in affinity for nicotine, for example, and a desensitized state of the receptor, which can no longer be activated by agonist (for recent reviews, see Refs. 1, 2, 15, 16, and 55).   table refer to Torpedo receptor, with the exception of methyllycaconitine and anatoxin-a. The data presented have been obtained by radioligand binding studies, with the exception of methyllycaconitine, which is an estimate from blockade of nicotinic currents by this compound in electrophysiological measurements using a frog muscle preparation. The value for anatoxin-a is also derived from electrophysiological measurements, using a mouse muscle nAChR expressed in Xenopus oocytes (for recent reviews, see Refs. 1,2,8,9,15 Table II for summary and comparison of K I and K D values). Secondary structure determination of the renatured ␣ nAChR 1-209 fragment by CD spectroscopy indicated 15% ␣-helical and 45% ␤-strand structure. Whereas the value for ␤-strand structure is consistent with data obtained for the membrane-bound and solubilized whole Torpedo nAChR by CD measurements (30 -32), results from Fourier transformed infrared and Raman spectroscopy (33,34) and theoretical modeling (35,36) (Table I) show that 15% ␣-helical structure is lower than the values of approximately 30 (37) and 20% (35) that have been obtained in two studies published so far explicitly concerning the extracellular N-terminal domain of nicotinic ␣-subunits. The first study used cryoelectron microscopy (36), and the differences observed could be explained in terms of subtle conformational changes due to interactions at interfaces to other subunits; the theoretical model makes a prediction (20% ␣-helix) that is nearer to the value we obtained (15%) than to the 30% ␣-helix derived from cryoelectron microscopy. Taken together, the biophysical data suggest that the renaturation procedure resulted in a ␣ nAChR 1-209 fragment that appeared to be structurally homogenous and to display a secondary structure composition that is consistent with previous predictions.
Expression of recombinant proteins in E. coli results in the absence of posttranslational modifications, like N-glycosylation, which are not part of the synthetic repertoire of bacteria. Native ␣-subunits of Torpedo nAChR are glycosylated on Asn 141, glycosylation does not affect ligand binding properties but rather seems to play an important role in transport of the different types of subunits to the plasma membrane and in the assembly of the receptor pentamer (see, e.g. Ref. 56 and, for reviews, Refs. 1, 2, 15, and 16).
Thus, the expression of the extracellular ligand binding domain of the ␣-subunit of Torpedo nAChR in E. coli and subsequent renaturation to a native conformation resulted in a water-soluble protein with ligand-binding properties very similar to the native receptor. The biochemical parameters confirm the homogeneity and functionality of the protein, which presents an example for the production and characterization of ligand binding domains of other nAChR, namely neuronal subtypes, which are not available from natural sources in high amounts, or from other members of the multigene superfamily of fivesubunit ionotropic receptors, like ␥-aminobutyric acid A , gly-cine, 5-hydroxytryptamine 3 or glutamate receptors.
To date, only the expression of short nicotinic receptor-fragment peptides, comprising mainly the assumed acetylcholinebinding site around the two adjacent cysteines 192 and 193, has been achieved in E. coli (41), and a NMR solution structure of such peptides in complex with ␣-bungarotoxin has been published (44). Recently, there were entries in the Protein Data Bank at Brookhaven National Laboratory of the NMR solution structures of the M2 membrane-spanning domains (which are assumed to contribute to the ion channel pore) of rat nAChR and human glutamate receptor of N-methyl-D-aspartate-subtype NR1, which have been expressed in E. coli and reconstituted in artificial micelles (Protein Data Bank accession numbers 1A11 and 2NR1). In summary, expression in E. coli and reconstitution of functional domains of neurotransmitter receptors appears to be an attractive possibility to obtain detailed structural information in the absence of successful crystallization of the whole receptor protein. Alternative strategies include recent successful attempts to express the N-terminal domain of mouse muscle ␣-nAChR in the membranes of Xenopus oocytes or Chinese hamster ovary cells (53) and rat ␣7 nAChR in Xenopus oocytes (54). Solubility and yield remained problematic, although these systems have advantages with respect to posttranslational modifications.
Altogether, this work demonstrates the feasibility of large scale production of functional ligand binding domains of neuroreceptors in a prokaryotic expression system for structural investigations applying NMR or x-ray crystallography, at the same time offering possibilities with regard to pharmacological screening systems.