Cell-free Expression and Functional Reconstitution of Homo-oligomeric α7 Nicotinic Acetylcholine Receptors into Planar Lipid Bilayers*

The α7 nicotinic acetylcholine receptor (nAChR) is a ligand-gated ion channel that modulates neurotransmitter release in the central nervous system. We show here that functional, homo-oligomeric α7 nAChRs can be synthesized in vitrowith a rabbit reticulocyte lysate translation system supplemented with endoplasmic reticulum microsomes, reconstituted into planar lipid bilayers, and evaluated using single-channel recording techniques. Because wild-type α7 nAChRs desensitize rapidly, we used a nondesensitizing form of the α7 receptor with mutations in the second transmembrane domain (S2′T and L9′T) to record channel activity in the continuous presence of agonist. Endoglycosidase H treatment of microsomes containing nascent α7 S2′T/L9′T nAChRs indicated that the receptors were glycosylated. A proteinase K protection assay revealed a 36-kDa fragment in the ER lumen, consistent with a large extracellular domain predicted by most topological models, indicating that the protein was folded integrally through the ER membrane. α7 S2′T/L9′T receptors reconstituted into planar lipid bilayers had a unitary conductance of ∼50 pS, were highly selective for monovalent cations over Cl−, were nonselective between K+ and Na+, and were blocked by α-bungarotoxin. This is the first demonstration that a functional ligand-gated ion channel can be synthesized using an in vitro expression system.

The ␣7 nicotinic acetylcholine receptor (nAChR) is a ligand-gated ion channel that modulates neurotransmitter release in the central nervous system. We show here that functional, homo-oligomeric ␣7 nAChRs can be synthesized in vitro with a rabbit reticulocyte lysate translation system supplemented with endoplasmic reticulum microsomes, reconstituted into planar lipid bilayers, and evaluated using single-channel recording techniques. Because wild-type ␣7 nAChRs desensitize rapidly, we used a nondesensitizing form of the ␣7 receptor with mutations in the second transmembrane domain (S2T and L9T) to record channel activity in the continuous presence of agonist. Endoglycosidase H treatment of microsomes containing nascent ␣7 S2T/L9T nAChRs indicated that the receptors were glycosylated. A proteinase K protection assay revealed a 36-kDa fragment in the ER lumen, consistent with a large extracellular domain predicted by most topological models, indicating that the protein was folded integrally through the ER membrane. ␣7 S2T/L9T receptors reconstituted into planar lipid bilayers had a unitary conductance of ϳ50 pS, were highly selective for monovalent cations over Cl ؊ , were nonselective between K ؉ and Na ؉ , and were blocked by ␣-bungarotoxin. This is the first demonstration that a functional ligand-gated ion channel can be synthesized using an in vitro expression system.
During biosynthesis of nAChRs, the polypeptide is translocated into the endoplasmic reticulum (ER) membrane. The ER contains enzymes necessary for signal sequence cleavage (13) and other post-translational modifications required for correct subunit folding, assembly, and ligand-binding site formation. These modifications include core glycosylation (13) and disulfide bond formation (14,15). In addition, ER and cytoplasmic chaperone proteins are thought to be involved in the maturation of muscle-type and ␣7 nAChRs (16 -19). Based on current topological models, nicotinic receptor subunits have four putative transmembrane domains, a large, glycosylated N-terminal extracellular domain that contains the agonist-binding site (14,20), and a short extracellular C terminus (1). The second transmembrane domain (M2) from each of the five subunits is postulated to line the ion-conducting pore (21).
The subunit composition of nAChRs containing the ␣7 gene product is not completely clear. The injection of ␣7 cRNA into Xenopus oocytes results in the formation of ACh-gated ion channels without requiring the co-expression of other neuronal nAChR subunit cRNAs (22), suggesting that ␣7 nAChRs are homo-oligomeric. However, Xenopus oocytes also express low levels of endogenous nAChR ␣ subunits, which can co-assemble with ␤, ␥, and ␦ muscle-type nAChR subunits to form functional nAChRs (23). These ␣-subunits could, potentially, coassemble with expressed ␣7 receptors in Xenopus oocytes as well. Based on co-immunoprecipitation experiments, native ␣7 receptors in rat brain appear to be homo-oligomeric (24), whereas native chick ␣7 subunits are thought to form both homo-oligomeric and hetero-oligomeric receptors, complexing with ␣8 (25-27) and other neuronal nAChR subunits (28). However, it is possible that the apparent homomeric ␣7 nAChRs in native tissues could represent heteromeric complexes containing yet unidentified nAChR subunits.
␣7 nAChRs have been difficult to express in several mammalian heterologous expression systems. The folding, assembly, and subcellular localization of heterologously expressed ␣7 nAChRs is deficient in some cell lines, due to misfolding and trapping of proteins in the ER (29). For example, human ␣7 nAChRs have been expressed in HEK-293 cells (30), but attempts to express chick or rat ␣7 nAChRs in HEK-293 cells have not yet been successful (29,31,32).
Another powerful approach to determine whether ␣7 nAChRs * This work was supported by National Institutes of Health Grant NS37317. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18  can form functional, homo-oligomeric receptors is to express them in a cell-free system. Muscle-type nAChRs translated in the presence of ER microsomes have been studied biochemically (13,33,34) but have not been examined for functional channel activity. Our goal was to express chick ␣7 nAChRs in vitro, where the co-expression of other nAChR subunits is extremely unlikely, and to study their biochemical and functional properties. Our experimental strategy was to express receptors using rabbit reticulocyte lysates in the presence of ER microsomes, reconstitute the channels into planar lipid bilayers, and record single-channel activity. This method has been used successfully to synthesize and reconstitute functional Shaker potassium channels (35), amiloride-sensitive sodium channels (36), and gap junction channels (37). In this paper, we show that ␣7 S2ЈT/L9ЈT nAChRs expressed in vitro were glycosylated, were processed integrally though the membrane, and formed functional channels when reconstituted into planar lipid bilayers. These data also show that ␣7 nAChR subunits can form functional, homo-oligomeric channels.

EXPERIMENTAL PROCEDURES
cDNAs and in Vitro Transcription of cRNA-Chick wild-type ␣7 nAChR cDNA was a gift from Mark Ballivet (University of Geneva, Geneva, Switzerland). Wild-type and S2ЈT/L9ЈT ␣7 nAChR cDNAs cloned into the pAMV vector (38) under control of the T7 promoter were kindly provided by Purnima Deshpande, Dr. Henry Lester, and Dr. Cesar Labarca (California Institute of Technology). The numbering of the residues in the M2 domain follows the convention of Miller (39), where the N-terminal (cytoplasmic) residue of the M2 domain is denoted as 1Ј. The Ser at the 2Ј position and Leu at the 9Ј position correspond to amino acids Ser 240 and Leu 247 , respectively, in the chick ␣7 cDNA sequence (22). The plasmid templates were digested with NotI. Capped cRNA transcripts were generated from the cDNA templates with T7 RNA polymerase using the mMessage mMachine kit (Ambion, Austin, TX) according to the manufacturer's instructions. The cRNA was resuspended in 100 mM KCl and stored at Ϫ70°C.
Two-electrode Voltage Clamp of Xenopus Oocytes-Two-electrode voltage clamp was performed with a GeneClamp 500 amplifier controlled by pCLAMP6 software (Axon Instruments, Foster City, CA). Microelectrodes were filled with 3 M KCl and had resistances of 0.5-2.1 M⍀. Oocytes were superfused with ES-EGTA (96 mM NaCl, 2 mM KCl, 1 mM MgCl 2 , 1 mM EGTA, 10 mM Na-HEPES, pH 7.5) in the presence and absence of ACh. Unless otherwise indicated, oocytes were voltage clamped at a holding potential of Ϫ60 mV. Currents from S2ЈT/L9ЈT receptors were filtered at 50 Hz (8-pole Bessel low pass) and digitized at 100 Hz with a Digidata 1200 A/D converter. Currents from wild-type receptors were filtered at 250 Hz and sampled at 500 Hz. Additional filtering was added to the traces for display purposes. To adjust for run-down during acquisition of dose-response data, the peak currents from repeated applications of a standard dose of ACh were used to normalize the test responses. Dose-response relationships were fitted to the Hill equation in Prism (GraphPad Software, San Diego, CA). Current-voltage relationships during the sustained maximal response to ACh were generated by applying a series of voltage steps (125 ms, 10 mV intervals) from the holding potential of Ϫ60 mV. Leak currents obtained in the absence of ACh were subtracted from ACh-evoked currents.
Preparation of Endoplasmic Reticulum-derived Quail Oviduct Microsomes-Endoplasmic reticulum microsomes were obtained from canine pancreas (Promega, Madison, WI) or were prepared from the oviducts of mature, laying Japanese quail (42). Briefly, the magnum portion of the oviduct was minced, suspended in 5 volumes of TKMD buffer (50 mM Tris-HCl, pH 7.7, 25 mM KCl, 2.5 mM MgCl 2 , 3 mM dithiothreitol) plus 0.88 M sucrose, homogenized, and centrifuged at 4000 ϫ g in an HB4 rotor for 10 min. The supernatant was diluted with TKMD buffer to a final sucrose concentration of 0.6 M and layered onto a sucrose step gradient of TMKD buffer containing 1.5 M and 2.0 M sucrose. After centrifugation for 16 -20 h at 100,000 ϫ g in an SW28 rotor, the microsomes were harvested from the 1.5 M and 2.0 M sucrose interface and resuspended in 20 mM Na-HEPES. The volume was adjusted with 20 mM Na-HEPES to obtain an A 260 reading of 0.6 in a sample containing 1% SDS. One-half volume of 750 mM sucrose, 20 mM Na-HEPES, pH 7.5 was added to the resuspended membranes, which were frozen in liquid nitrogen and stored at Ϫ70°C.
In Vitro Transcription, Translation, and Membrane Processing-Coupled in vitro transcription/translation was performed using the Promega TNT kit. For analysis of translation and processing efficiency, reaction mixtures (25 l) contained 12.5 l of rabbit reticulocyte lysate, 0.5 l of 10ϫ reaction buffer, 20 M amino acids minus methionine, 10 units of RNasin (Promega), 0.5 l of T7 RNA polymerase, and 10 Ci of [ 35 S]methionine (Amersham Pharmacia Biotech) with or without 1 g of ␣7 S2ЈT/L9ЈT nAChR cDNA. Reactions were also supplemented with 1 mM dithiothreitol and 1 mM oxidized glutathione. Some reactions contained microsomes from canine pancreas (1.8 l/25-l reaction) or quail oviduct (1.0 l/25-l reaction). Reactions were incubated for 90 min at 30°C and stopped by placing on ice. After reserving 3 l of the reaction mixture for gel analysis (labeled M in Figs. [2][3][4], the membranes were pelleted by centrifugation for 25 min at 14,000 rpm in a microcentrifuge, washed twice in solution DЈ (160 mM KCl, 20 mM MOPS, pH 7.4), and resuspended in water. Samples were heated for 30 s at ϳ95°C in SDS-PAGE sample buffer and electrophoresed on 10% or 12.5% SDS-polyacrylamide gels (43). Gels were treated with Fluoro-Hance (Research Products International, Mount Prospect, IL), dried, and exposed to Kodak X-Omat film at Ϫ70°C.
Endoglycosidase H Treatment-A coupled transcription/translation reaction (37.5 l) was performed as described above in the presence of ␣7 S2ЈT/L9ЈT nAChR cDNA and canine pancreatic microsomal vesicles. The membranes were pelleted (15 min at 14,000 rpm), resuspended in 50 mM sodium acetate, 1% (w/v) ␤-mercaptoethanol, pH 6.0, and divided into equal fractions. Endoglycosidase H (1 milliunit; Roche Molecular Biochemicals) or carrier buffer (25 mM EDTA, 0.05% sodium azide, 0.1% SDS, 50 mM NaH 2 PO 4 , pH 7.0) was added, and the samples were incubated for 2 h at 37°C. The reaction was stopped by chilling on ice. SDS-PAGE sample buffer was added, and the samples were heated at ϳ95°C for 30 s and analyzed by SDS-PAGE.
Proteinase K Analysis-Pelleted microsomal membranes from a 50-l coupled transcription/translation reaction were resuspended in 30 l of reaction buffer (160 mM NaCl, 5 mM CaCl 2 , 50 mM Tris-Cl, pH 7.5). Each aliquot was treated with 0.1 mg/ml proteinase K, 0.1 mg/ml proteinase K plus 1% Triton X-100, or water for 45 min on ice. The reactions were stopped by adding 2 mM phenylmethylsulfonyl fluoride and an equal volume of 2x SDS-PAGE sample buffer. The reactions were heated immediately at ϳ95°C for 30 s and analyzed by SDS-PAGE.
Sucrose Density Gradient Analysis-Coupled transcription/translations containing [ 35 S]Met were performed in the absence or presence of canine pancreatic microsomes as described above. Mixtures (25 l) from translations performed in the absence of microsomes were diluted to 40 l in a final concentration of 10 mM NaH 2 PO 4 , 50 mM NaCl, 1 mM EDTA, pH 7.0, and 0.5% Triton X-100. Translations performed in the presence of microsomes were scaled to a final volume of 100 l and centrifuged to pellet the microsomes. The microsomes were washed and resuspended in 40 l of solution DЈ and solubilized in 0.5% Triton X-100 for 60 min on ice. Samples were layered on top of 5-ml linear 5-20% (w/w) sucrose gradients containing 0.2% Triton X-100, 10 mM NaH 2 PO 4 , 50 mM NaCl, and 1 mM EDTA, pH 7.0 (44). Sucrose gradients run in parallel contained the standard proteins ovalbumin (3.6 S), bovine serum albumin (4.2 S), human gamma globulin (7 S), and catalase (11 S). The gradients were centrifuged for 12 h at 300,000 ϫ g at 4°C in a SW50.1 rotor. Fractions were collected from the top of the gradients, and the sucrose concentration of each fraction was determined by refractometry. Proteins from each fraction were separated by SDS-PAGE and visualized by fluorimetry ( 35 S-labeled protein) or Coomassie Blue staining (standard proteins). Gels or autoradiographs were digitized, and the relative amount of protein in each band was analyzed by densitometry using NIH Image software.
In Vitro Transcription/Translation for Reconstitution into Planar Lipid Bilayers-Coupled transcription/translations for the incorporation of microsomal vesicles into planar lipid bilayers were performed as described above except that [ 35 S]methionine was omitted, 20 M amino acids minus cysteine were added to provide the full complement of all unlabeled amino acids, and all volumes were scaled up to make a final volume of 50 l. After translation, the membranes were pelleted by centrifugation for 15 min at 9,000 rpm. The supernatant was removed immediately, and the pellet was washed twice with solution DЈ and resuspended thoroughly by gentle trituration in 5 l of 250 mM sucrose and 20 mM HEPES, pH 7.4. Bilayer experiments were generally performed on the same day as translations.
Single Channel Recording of Cell-free Expressed nAChRs Reconstituted into Planar Lipid Bilayers-Synthetic lipids (1-palmitoyl-2-oleoyl phosphatidylethanolamine and 1-palmitoyl-2-oleoyl phosphatidylserine; Avanti Polar Lipids, Alabaster, AL) were resuspended in n-decane at concentrations of 15 mg/ml and 5 mg/ml, respectively. Planar lipid bilayers were formed by applying the decane solution across a 200-m hole in a polyvinyldifluoride partition separating two aqueous chambers denoted cis and trans. Endoplasmic reticulum-derived microsomal vesicles from in vitro translations were incorporated into planar lipid bilayers by applying them directly onto the cis face of the bilayer with a fire-polished glass probe (45). To promote the fusion of vesicles with the bilayer, 0.5 mM CaCl 2 was present on the cis side. Acetylcholine was added to both chambers to activate all incorporated nAChRs regardless of membrane orientation. For each 50-l translation, three to five bilayer experiments could be performed, each of which lasted about 1 h. Bilayers were voltage-clamped with a Warner Instruments patch-clamp amplifier (Hamden, CT). Voltages were assigned as cis relative to trans, with trans corresponding to the luminal side of the ER (the extracellular face). Data were filtered at 200 Hz using a 4-pole Bessel low pass filter, digitized at 1 kHz, and analyzed off-line using in-house analysis programs written in AxoBasic (Axon Instruments, Foster City, CA).

Selection of a Nondesensitizing Mutant of the ␣7 nAChR-
The goal of these experiments was to characterize the biochemical and functional properties of ␣7 nAChRs synthesized in vitro and reconstituted into planar lipid bilayers. To study the activity of ligand-gated ion channels in a planar lipid bilayer, agonist is continually present to evoke channel opening events. Because wild-type ␣7 nAChRs desensitize very rapidly in the continued presence of ACh, we selected a nondesensitizing form of the ␣7 nAChR so that channel activity could be observed for extended periods of time. This receptor has a serine to threonine mutation at the 2Ј position and a leucine to threonine mutation at the 9Ј position (␣7 S2ЈT/L9ЈT nAChR). Fig.  1A shows that wild-type ␣7 nAChRs activated rapidly and desensitized completely within 1 s of ACh application. In contrast, ␣7 S2ЈT/L9ЈT nAChRs (Fig. 1B) did not desensitize during a 30-s application of ACh, behavior similar to that of ␣7 L9ЈT nAChRs described by Revah et al. (46). Like wild-type and ␣7 L9ЈT nAChRs (47), ␣7 S2ЈT/L9ЈT nAChRs were completely inhibited by ␣-BTX (Fig. 1B). Fig. 1C shows the ACh doseresponse characteristics of wild-type and ␣7 S2ЈT/L9ЈT nAChRs. As was seen with ␣7 L9ЈT nAChRs (46), the EC 50 of ␣7 S2ЈT/L9ЈT nAChRs (14.1 M) was much lower than that of wild-type ␣7 nAChRs (345 M). Fig. 1D shows that Ca 2ϩ permeates through ␣7 S2ЈT/L9ЈT nAChRs. In the absence of any permeant ions in the bath, no ACh-evoked currents were observed. In the presence of 10 mM Ca 2ϩ (and no other permeant ions), robust ACh-evoked currents were recorded. Thus, ␣7 S2ЈT/L9ЈT nAChRs displayed the nondesensitizing kinetics necessary to sustain channel activity in planar lipid bilayers in the continued presence of agonist. In addition, ␣7 S2ЈT/L9ЈT nAChRs displayed sensitivity to ␣-BTX, high potency of ACh, and Ca 2ϩ permeability, as expected.
In Vitro Translation and Processing of ␣7 S2ЈT/L9ЈT nAChRs-To determine whether the S2ЈT/L9ЈT ␣7 nAChRs synthesized in vitro were full-length, glycosylated, and correctly folded, we evaluated a number of its biochemical properties. Fig. 2 shows a fluorogram of [ 35 S]Met-labeled proteins generated from a coupled transcription/translation performed in the presence and absence of endoplasmic reticulum microsomes derived from quail oviduct. In the absence of microsomes, a ϳ41-kDa protein and a 28 kDa protein were observed (lane 3). The primary sequence of wild-type ␣7 nAChRs indicates a nonglycosylated molecular mass of 54 kDa (22, 25), a value substantially higher than the ϳ41 kDa observed. Other nAChR ␣-subunits also run anomalously fast on SDS-poly-acrylamide gels; nonglycosylated ␣1 subunits from Torpedo and mouse muscle nAChRs have calculated molecular masses of 50 kDa but migrate as 41-43-kDa proteins (33,34,48). Thus, the ϳ41-kDa protein expressed in vitro is likely to be the full-length ␣7 S2ЈT/L9ЈT nAChR. The 28-kDa protein could be either a premature translation stop or a proteolytic degradation product.
In translation mixtures containing endoplasmic reticulum microsomes, a ϳ51-kDa product was observed in addition to the ϳ41-kDa protein (Fig. 2, lane 4). When the translation products were centrifuged to separate proteins associated with the microsomal membranes from those in solution, the 51-kDa product was predominantly associated with the membrane pellets (P, lane 5), whereas the 41-kDa product remained primarily in the supernatant (S, lane 6). These results suggest that the 51-kDa product was the membrane-processed form of the receptor and the 41-kDa protein was the unprocessed form. Similar results were obtained when canine pancreatic microsomes were used instead of avian oviduct microsomes, except that higher yields of processed protein were obtained with the canine pancreatic microsomes.
To test the hypothesis that the slower migration of the 51-kDa membrane-associated protein was due to glycosylation, the microsome-associated proteins were treated with endoglycosidase H (Endo H), which cleaves high mannose oligosaccharides from glycoproteins (49).  4). These data indicate that the membrane-associated ␣7 S2ЈT/L9ЈT receptors were glycosylated with high mannose oligosaccharides.
To determine whether the membrane-associated proteins were co-translationally inserted into the membranes with the transmembrane configuration expected for nAChRs, or were simply associated with the membranes in a nonspecific manner, membrane-associated proteins were treated with proteinase K, a nonspecific serine protease, in the presence and absence of detergent (Fig. 4A). A coupled transcription/ translation of ␣7 S2ЈT/L9ЈT cDNA was performed in the presence of canine pancreatic microsomes (lanes 2-6). As before, the 51-kDa protein was found in the microsomal fraction (lane 4). Incubation of intact microsomal membranes with proteinase K produced a ϳ36-kDa fragment (lane 5), suggesting that a large part of the protein was protected from proteolysis by the membrane. In the presence of Triton X-100 to disrupt the membranes, proteinase K caused complete digestion of the membrane protein (lane 6), as expected. The size of the 36-kDa membrane-protected fragment was consistent with the expected size of the glycosylated extracellular N-terminal domain. Specifically, the mature N-terminal extracellular region plus the first transmembrane domain of the ␣7 nAChR has a calculated molecular mass of 27.5 kDa (22). The addition of carbohydrate to the three N-linked glycosylation sites in the N-terminal extracellular domain of ␣7 nAChRs (20) adds ϳ10 kDa, as evidenced by the apparent shift in gel migration (Fig.  3). Thus, the expected molecular mass for the glycosylated N terminus is ϳ37.5 kDa, a value similar to that of the ϳ36-kDa protein fragment observed. These results suggest that the nascent ␣7 nAChR proteins were co-translationally inserted across the membrane and folded into a transmembrane orientation with a large extracellular domain (Fig. 4B).
To determine whether the ␣7 subunits formed oligomeric assemblies, we performed sucrose gradient sedimentation analysis of in vitro translated ␣7 S2ЈT/L9ЈT nAChRs. On 5-20% linear sucrose gradients containing 0.2% Triton X-100, ␣7 S2ЈT/L9ЈT nAChRs translated in the absence of microsomes sedimented at 3.6 S (Fig. 5A). This value was similar to the sedimentation velocity of Torpedo ␣1 (50) and muscle ␣1 (51) nAChR subunits. The S2ЈT/L9ЈT nAChRs translated in the presence of microsomes formed multiple sizes of subunit com- In this membrane orientation, digestion with proteinase K in the absence of detergent is predicted to generate a protein fragment that contains the glycosylated N-terminal extracellular domain and first transmembrane domain, which has a predicted molecular mass of 37.5 kDa. plexes ranging from 3.6 S to 11 S (Fig. 5B). The multiple sizes of these complexes suggest that some but not all of the ␣7 complexes synthesized in vitro form pentameric complexes. Similar results were observed with oocyte-expressed ␣7 nAChRs, which form protein complexes composed of various numbers of subunits, but only those fractions corresponding to ϳ10 S complexes bind ␣-BTX (52). Native Torpedo californica nAChR and ␣7 nAChR pentamers sediment at ϳ9 and 10 S, respectively (44,52).
Functional Properties of ␣7 nAChRs Expressed in Vitro and Incorporated into Planar Lipid Bilayers-To determine whether the ␣7 nAChR protein translated in vitro formed functional ion channels, endoplasmic reticulum microsomes containing nascent ␣7 S2ЈT/L9ЈT nAChRs were reconstituted into planar lipid bilayers that were voltage clamped to measure single-channel current transitions from the incorporated ion channels. Fig. 6A shows a selection of current recordings at several membrane voltages with 300 mM KCl in the cis (intracellular) chamber and 50 mM KCl in the trans (extracellular) chamber. The channel had a high open probability, with bursts of opening and closing transitions.
In asymmetrical concentrations of KCl, the reversal potential from the current-voltage relationship provides information about the K ϩ -over-Cl Ϫ selectivity of the channels. A plot of current amplitude versus membrane potential (Fig. 6B) revealed a linear current-voltage relationship with a reversal potential of Ϫ41 mV, a value close to the equilibrium potential for K ϩ (E K ) of Ϫ43 mV, indicating the expected K ϩ -over-Cl Ϫ selectivity of the reconstituted channels. The unitary conductance of 48 pS was similar to the 45 pS value reported for oocyte-expressed wild-type ␣7 nAChRs (46) and to the conductance of oocyte-expressed ␣7 S2ЈT/L9ЈT receptors recorded from outside-out patches (ϳ41 pS; data not shown).
Another distinctive property of both muscle-type and neuronal nAChRs is their inability to select between small monovalent cations such as Na ϩ and K ϩ (21,53). After confirming that the incorporated channel was selective for K ϩ over Cl Ϫ in asymmetrical KCl concentrations, currents were recorded in the presence of both Na ϩ and K ϩ to determine whether the reconstituted ion channel was equally permeable to Na ϩ and K ϩ . Under these ionic conditions (see legend to Fig. 7), the reversal potential of a cation channel with equal permeabilities to Na ϩ and K ϩ would be ϳ0 mV. Examples of single-channel recordings (Fig. 7A) and the current-voltage plot (Fig. 7B) are shown. As expected for a nonselective cation channel, the reversal potential was Ϫ2.0 mV, producing a calculated K ϩ -to-Na ϩ permeability ratio (P K /P Na ) of 1.2. The unitary conductance of the channel was 51 pS.
A distinguishing characteristic of the ␣7 neuronal nAChR is its sensitivity to block by ␣-BTX (22). Macroscopic currents evoked by acetylcholine in Xenopus oocytes expressing ␣7 S2ЈT/ L9ЈT receptors were blocked after a 30-min incubation by 100 nM ␣-BTX (Fig. 1B). To test the functional block of ␣7 S2ЈT/L9ЈT receptors synthesized in vitro, ␣-BTX was added to an active FIG. 5. Sucrose gradient analysis of in vitro translated ␣7 S2T/ L9T nAChRs. ␣7 receptors translated in the absence (A) and presence (B) of canine pancreatic microsomes were solubilized in 0.5% Triton X-100 and sedimented on 5-20% linear sucrose gradients containing 0.2% Triton X-100. The migration (in Svedberg units) of standard proteins run on parallel gradients is indicated with arrows. Sucrose concentrations for each fraction were determined by refractometry to confirm that paired gradients were identical. Image densities were determined from densitometric analysis of autoradiographs ( 35 S-labeled ␣7 nAChR protein) or Coomassie Blue-stained gels (standard proteins). The 11 S peak in B indicates 35 S-labeled protein that had sedimented to the bottom of the gradient. channel incorporated into a planar lipid bilayer. Channel activity stopped 20 min after the addition of 375 nM ␣-BTX to both sides of the bilayer (not shown). The slow onset of channel block was similar to the long incubations usually required for the ␣-BTX block of muscle-type nAChRs (54) and oocyte-expressed wild-type ␣7 receptors (22).
No similar channel activity was observed in control bilayer experiments using microsomes from quail oviduct or canine pancreas. The criteria for identifying nAChR-like channel behavior were: 1) cation selectivity, 2) lack of selectivity between Na ϩ and K ϩ , 3) ϳ50 pS conductance, and 4) channel bursting kinetics. No channels with these characteristics were observed from microsome-processed ␣7 S2ЈT/L9ЈT receptors in the absence of agonist (11 experiments from 4 translations). In addition, no nAChR-like channel activity was observed following quail oviduct-processed translations of cDNA encoding the inward rectifier potassium channel (52 experiments from 37 translations). Finally, no nAChR-like channels were observed when quail oviduct microsomes that were not incubated with translation mixtures were incorporated into bilayers (319 experiments). These data suggest that the nAChR channel activity was a result of nascent protein synthesis and did not derive from channels present in the oviduct microsomes.
Occasionally, channels other than nAChRs were observed during the reconstitution experiments. We observed several Cl Ϫ -selective channels as described previously (35), a few cat-ion-selective channels with conductances of 20 -38 or Ͼ100 pS, and one channel type with no selectivity among Cl Ϫ , Na ϩ , and K ϩ . These channels were observed in reconstitution experiments whether or not the microsomes had been incubated with translation mixtures, indicating that they were endogenous to the microsomal membranes. DISCUSSION We have used a cell-free expression system consisting of an in vitro transcription and translation system derived from rabbit reticulocyte lysates and supplemented with ER microsomes to reconstitute functional ␣7 S2ЈT/L9ЈT nAChRs into planar lipid bilayers. We have shown that ␣7 S2ЈT/L9ЈT nAChRs synthesized in vitro were co-translationally processed into the ER-derived membranes and core glycosylated. When ER vesicles containing synthesized ␣7 S2ЈT/L9ЈT nAChRs were fused with planar lipid bilayers, functional ACh-gated ion channels were observed. The ion channels displayed the functional properties expected of the ␣7 nicotinic receptor: selectivity for K ϩ over Cl Ϫ , nonselectivity between Na ϩ and K ϩ , the expected single channel conductance (ϳ50 pS), and sensitivity to block by ␣-BTX. The nondesensitizing properties of the ␣7 S2ЈT/L9ЈT nAChR allowed recording of the single channel activity for extended periods of time (over 30 min).
The observation that functional ␣7 S2ЈT/L9ЈT nAChRs could be synthesized in vitro confirms that other nAChR subunits are not required for the formation of functional ␣7 receptors, assuming that no endogenous nAChR subunits are present in quail oviduct ER microsomes. These data are consistent with the observation that ␣7 wild-type (22) and ␣7 S2ЈT/L9ЈT nAChRs (Fig. 1) require no additional nAChR subunits to form ␣-BTX-sensitive, ACh-gated channels when expressed in Xenopus oocytes. In addition, these data imply that maturation of carbohydrate moieties by the Golgi apparatus is not required for functional ␣7 channel activity.
Functional ␣7 S2ЈT/L9ЈT nAChR channels were observed when translation products were processed by quail oviduct ER microsomes (120 experiments from 32 translations) but not when processed by canine pancreatic ER microsomes (98 experiments from 27 translations). This difference was not due to the amount of protein produced, because the level of membrane-processed ␣7 protein was higher in translations containing canine pancreatic microsomes than in translations containing avian oviduct microsomes. Vesicle fusion with the planar lipid bilayer was achieved with both processing systems, as evidenced by the appearance of endogenous channels in both types of microsomes; the incorporation rate of channels present in the microsomal membranes (chloride channels and cation channels) was 24% for quail oviduct microsomes and 44% for canine pancreatic microsomes. Protein factors such as foldases or chaperonins that are necessary for nicotinic receptor maturation (16 -19) may be more active in the oviduct microsomes than pancreatic microsomes. Perhaps additional factors or post-translational events that are required for functional expression of chick ␣7 nAChRs are more efficient in oviduct microsomes. Alternatively, commercial preparations of canine pancreatic microsomes may contain inhibitors that prevent the proper formation of functional ␣7 nAChRs channels.
The incorporation rate of the synthesized, reconstituted nAChR channels processed by quail oviduct microsomes was ϳ10%, similar to the incorporation rate of 12% for the Shaker potassium channel (35). The frequent appearance of endogenous channels suggests that vesicle incorporation into the bilayer was not the sole limiting factor for the low incorporation rate of functional ␣7 nAChRs. It seems likely that oligomerization may be a limiting step in the maturation of functional ␣7 channels in vitro, as it is in cultured muscle cells where only FIG. 7. ␣7 S2T/L9T nAChRs synthesized in vitro are nonselective between K ؉ and Na ؉ . A, channels were reconstituted into a planar lipid bilayer from an in vitro transcription/translation containing ␣7 S2ЈT/L9ЈT nAChR cDNA and quail oviduct microsomal vesicles. The solutions contained 300 mM KCl, 0.5 mM CaCl 2 and 10 mM HEPES, pH 7.0 (cis), and 50 mM KCl, 250 mM NaCl, and 10 mM HEPES, pH 7.0 (trans). 30 M ACh was included on both sides of the channel to evoke channel openings. The closed level is indicated by dotted lines. Cation selectivity was confirmed before adding NaCl by analyzing current transitions in asymmetrical concentrations of KCl. B, current-voltage plot in the presence of Na ϩ and K ϩ . The equilibrium potentials for K ϩ (E K ) and Na ϩ (E Na ) as calculated from the Nernst equation are indicated on the graph. The channel had a unitary conductance of 51 pS and a reversal potential of Ϫ2.0 mV. Using the Goldman-Hodgkin-Katz equation (21) and published activity coefficients (63,64), the calculated permeability ratio of K ϩ to Na ϩ (P K /P Na ) was 1.2, indicating the channel was almost equally permeable to Na ϩ and K ϩ . Current amplitudes were determined as the mean Ϯ S.D. of Gaussian fits to amplitude histograms. The line represents a linear regression of the data. 30% of ␣1 nAChR subunits synthesized in the ER bind ␣-BTX or are assembled into heteropentamers (55). The multiple sizes of ␣7 nAChR subunit complexes from in vitro translations and from Xenopus oocytes (52) suggest that many of the subunits synthesized both in ovo and in vitro may be trapped in nonfunctional oligomers containing too few or too many subunits. It is also possible that improperly folded subunits may oligomerize with correctly folded subunits and prevent them from forming functional channels.
The assembly of ␣7 nAChRs expressed in Xenopus oocytes requires cyclophilin, a prolyl isomerase and chaperone protein (19), the concentration of which is estimated to be 4 -8 M in reticulocyte lysates (56). The co-expression of cloned cyclophilin A did not improve the rate of synthesis of functional ␣7 nAChRs in vitro (not shown), suggesting that the amount of cyclophilin in lysates was sufficient and that mechanisms other than prolyl isomerization were limiting.
An appropriate redox potential is critical for the correct formation of the ␣-BTX-binding site (33) as well as the correct folding and assembly (57) of muscle nAChRs. Most of the translation reactions were supplemented with equal amounts of dithiothreitol and glutathione to provide a redox gradient across the membrane. Under these conditions, the lumen of the ER vesicle has an oxidizing environment to promote the formation of disulfide bonds (58). We optimized these conditions for protein yield, but it is possible that the conditions were not optimal for proper assembly, folding, and/or ␣-BTX-binding site formation (33).
The cell-free expression approach offers several advantages over the isolation and reconstitution of channels from native tissues or the expression of cloned ion channels in heterologous systems. Regulatory molecules such as kinases and phosphatases are often closely associated with native ion channels reconstituted from plasma membranes (59 -61). In contrast, ion channels newly synthesized in the ER are less likely to be assembled with regulatory protein complexes than ion channels purified from plasma membranes. Channels synthesized in vitro are dissociated from second messenger pathways in the host cell that can modulate the activity of the ion channel or activate other cellular components. In addition, the planar lipid bilayer technique offers precise control over the ionic conditions, isolation from other channels native to the host cell such as the Xenopus oocyte Ca 2ϩ -activated chloride channel (62), and the ability to co-reconstitute signaling pathways and modulatory proteins as desired. In addition, this in vitro translation and reconstitution approach can be combined with mutational analysis to evaluate structure-function relationships.