Assembly of ionotropic neurotransmitter receptors typified by acetylcholine receptors (AChRs) is thought to be directed by an N-terminal extracellular domain of a subunit. Consistent with this hypothesis, chimeras with the δ subunit N-terminal domain fused to the rest of the γ subunit can substitute for δ, but not γ, subunits during AChR assembly. However, chimeras with the γ subunit N-terminal domain fused to the rest of the δ subunit cannot substitute for γ or δ subunits during assembly. Furthermore, expression of this chimera with the four wild-type subunits prevents the formation of α-bungarotoxin (Bgt) binding sites. Instead of AChR pentamers, complexes are assembled containing only the chimera and either α or β subunits. Based on the results of additional γ-δ chimeras, there are at least two different regions within the C-terminal half of the chimera required for the dominant-negative effect. Our results indicate that the N-terminal domain of the γ subunit mediates the initial subunit associations, whereas signals in the C-terminal half of the subunit are required for subsequent subunit interactions.
As large oligomeric and polytopic proteins, the folding and assembly of ion channels are complex and must occur with high fidelity because misassembly of even a few channels, each controlling the flow of ∼107 ions/sec, can be disastrous for a cell. Moreover, folding and assembly are the rate-limiting steps during channel biogenesis (Green and Millar, 1995) and are likely to play an important role in regulating the number and targeting of functional channels. The muscle-type nicotinic acetylcholine receptor (AChR) was the first and remains among the few ion channels for which the subunit composition and stoichiometry are established definitively. As a member of the family of neurotransmitter-gated ion channels that includes neuronal AChRs, 5HT3, glycine, and GABAA and GABAC receptors, the muscle-type AChR has long served as a model for studies of ion channel assembly.
Muscle-type AChRs are composed of four homologous subunits (α, β, γ, and δ) that assemble into α2βγδ pentamers (Karlin and Akabas, 1995; Lindstrom, 1995; Changeux and Edelstein, 1998). According to the “heterodimer” model of AChR assembly (Blount et al., 1990; Gu et al., 1991; Saedi et al., 1991; Kreienkamp et al., 1995), α subunits fold into a conformation that can bind α-bungarotoxin (Bgt) before assembling with γ or δ subunits into heterodimers. ACh binding sites form on the heterodimers that associate together with β subunits into α2βγδ pentamers. Evidence for this model is based on complexes formed in cells expressing less than the full complement of subunits. When cells expressing all four AChR subunits are used (Green and Claudio, 1993;Green and Wanamaker, 1997), it has been found that AChR subunits rapidly assemble into αβγ trimers. Afterward, δ subunits are added to form αβγδ tetramers, and then a second α subunit is added to make α2βγδ pentamers. Another significant difference is that the first Bgt and ACh binding sites appear only after αβγ trimer assembly, and the second Bgt and ACh binding sites appear after α2βγδ pentamers form (Green and Wanamaker, 1998).
Regions within the N-terminal extracellular domain of AChR subunits are required for associations between subunits (Yu and Hall, 1991; Sumikawa and Gehle, 1992; Verrall and Hall, 1992; Kreienkamp et al., 1995). The N-terminal regions of glycine (Kuhse et al., 1993) and GABAC and GABAA (Hackam et al., 1998) receptors also are involved in receptor assembly. In support of these studies, we find that the chimeric subunit, which contains the N-terminal domain of the AChR δ subunit fused to the rest of the γ subunit, substitutes for δ, but not γ, subunits during AChR assembly. However, the reverse chimera with the N-terminal domain of the γ subunit fused to the rest of the δ subunit does not substitute for γ or δ subunits. Instead, when coexpressed with the four wild-type subunits, the chimera assembles with α or β subunits and blocks the formation of all Bgt binding sites. This dominant-negative effect is specific for the loss of γ subunit regions, not the presence of δ subunit regions, and requires at least two different γ subunit regions in the C-terminal half of the subunit. Our results indicate that regions within an N terminal domain of a subunit mediate the initial rapid subunit associations, whereas regions within the C-terminal half of the γ subunit are required for subsequent subunit interactions.
MATERIALS AND METHODS
cDNA constructs. The γ215δ and δ221γ chimeras were constructed via standard PCR methods (Yon and Fried, 1989). For γ215δ, an oligonucleotide was obtained consisting of the 15 base pairs (bp) of the γ cDNA preceding the region coding for the first transmembrane domain (M1), followed by the first 15 bp of M1 from the δ cDNA. Amplification of γ cDNA using this primer resulted in a product corresponding to the γ sequence up to the start of M1, followed by 15 bp of δ sequence. This product was used in a second PCR step to amplify the δ cDNA, with the short region of δ sequence serving to anneal the sequence to the δ template. The δ221γ chimera was made by the same method, except that both a forward and a reverse chimeric primer corresponding to the chimera junction were used. The γ215β chimera was constructed by using a sharedBclI restriction site located before M1 of the β and γ subunits to join the γ and β cDNAs at the BclI site.
All additional chimeras were constructed by an introduction of restriction sites into the subunit cDNA by PCR, using a mutagenic primer with a 1–3 bp mismatch (Ho et al., 1989). The γ215α chimera was produced by introducing the BclI site described above into the homologous location in the α subunit to join the γ and α cDNAs at the BclI site. To construct γ215δ467γ, we introduced a unique StyI restriction site within M4 of δ into the homologous location of the γ subunit to replace the C terminus of γ215δ, starting at amino acid 467, with that of γ. The γ322δ and γ215δ329γ chimeras were constructed by introducing the unique BglII site of the δ subunit into the corresponding location of the γ subunit. Full-length chimeric subunits were created by cutting with the appropriate restriction enzyme and ligating subunit fragments together at the common sites. Regions of mutant subunits constructed by PCR methods were sequenced to control for errors in the PCR products.
Transfections. The four Torpedo subunit cDNAs in the pRBG4 expression vector (Lee et al., 1991) were transiently transfected into 6 cm cultures of tsA201 cells (Margolskee et al., 1993), using a calcium phosphate method (Eertmoed et al., 1998). Unless otherwise indicated, 1 μg of α, 0.5 μg of β, 2.5 μg of γ, and 7.5 μg of δ subunit cDNAs were used. Cells were maintained at 37°C with 5% CO2 in DMEM supplemented with 10% fetal bovine serum. Because of the temperature dependence ofTorpedo AChR expression (Claudio et al., 1987), cells were shifted from 37 to 20°C 24 hr after the transfection. Maximal expression occurred 4 d after the temperature shift.
The γ215δ chimeric cDNA in pRBG4 was stably transfected into a cell line already expressing the four Torpedo subunits (Claudio et al., 1987). Cells were maintained in DMEM supplemented with 10% calf serum, HAT (15 μg/ml hypoxanthine, 1 μg/mlaminopterin, and 5 μg/ml thymidine), and G418 (1 mg/ml) to maintain thymidine kinase and neomycin selection of the stably transfected subunits. To enhance the expression of the four wild-type subunits under the control of SV40 promoters, we also supplemented the culture medium with 20 mm sodium butyrate for 1–5 d before assay. Using calcium phosphate transfection, we cotransfected each 10 cm culture with 15 μg of the γ215δ chimeric cDNA and 500 ng per plate of the neomycin resistance gene construct pSV2Neo. Clonal isolates of neomycin-resistant cells were obtained by adding 0.6 mg/ml G418 to the culture medium and were screened for the expression of AChR subunits by 125I-Bgt binding.
125 I-Bgt binding. To assay cell-surface125I-Bgt binding, we washed the cultures with PBS and incubated them at room temperature in PBS containing 4 nm 125I-Bgt (140–170 cpm/fmol) for 2.5 hr on a shaker table. Plates were washed again with PBS and solubilized; 125I-Bgt counts were determined by γ counting. To assay 125I-Bgt binding to solubilized AChRs (surface plus intracellular), we incubated cell lysates with 10 nm 125I-Bgt overnight at 4°C, followed by immunoprecipitation with monoclonal antibody (mAb) 35. Nonspecific binding was measured by binding to sham-transfected cells or by adding 10 mm carbamylcholine during binding.
Metabolic labeling and SDS-PAGE analysis. Cultures in 10 cm plates were labeled as described previously (Green and Claudio, 1993;Green and Wanamaker, 1997). Briefly, cultures were pulse-labeled in 2 ml of methionine–cysteine-free DMEM, supplemented with 333 μCi of a 35S-methionine35S-cysteine mixture (NEN EXPRE35S35S). The labeling was stopped with the addition of DMEM plus 5 mm methionine. To follow the subsequent changes in the labeled subunits, we “chased” the cells by incubation for the indicated times in NB medium at 20°C. The cells were solubilized in (in mm) 150 NaCl, 5 EDTA, 50 Tris, pH 7.4, 2 PMSF, and 2 N-ethylmaleimide plus 0.02% NaN3 and 1.83 mg/ml phosphatydlcholine and 1% Lubrol (LPC). Solubilized AChR subunits and chimeras were immunoprecipitated with rabbit polyclonal antisera, anti-α, anti-δ (Claudio and Raftery, 1977), or anti-β mAb 148, anti-γ mAb 168 (Gullick and Lindstrom, 1983), conformational-dependent mAb 14 (Tzartos and Lindstrom, 1980), or γ and δ subunit-specific mAb 88b (American Type Culture Collection, Rockville, MD).
Antibody–subunit complexes were precipitated with Protein G-Sepharose and electrophoresed on 7.5% SDS polyacrylamide gels, fixed, enhanced for 30 min, dried on a gel dryer, and exposed to film at −70°C with an intensifying screen. Autoradiographs were quantified by scanning densitometry, using a flatbed scanner, and analyzed with the Intelligent Quantifier software from BioImage (Ann Arbor, MI). A standard, provided by BioImage, with 21 bands ranging from 0.05 to 3.05 optical density units in 0.15 increments was used to calibrate the linearity of the densitometer before using it. To ensure that quantified bands were in the linear range and that the darker signals were not saturated, we took three to five exposures of each autoradiograph; each was scanned to insure that the scanned bands remained in the linear range of the film.
Sucrose gradients. Cell lysates were layered on a 5 ml 5–20% linear sucrose gradient prepared in LPC lysis buffer. Gradients were centrifuged in a Beckman SW 50.1 rotor at 40,000 rpm to ω2 t = 9.0 × 1011. Eighteen fractions of 300 μl each were collected from the top of the gradient. Then the fractions were counted in a gamma counter to determine the amount of 125I-Bgt bound to each fraction.
The δ subunit N-terminal extracellular domain is sufficient to direct δ subunit assembly
Of the four AChR subunits, γ and δ subunits are the most similar. Nonetheless, γ subunits rapidly associate with α and β subunits, while δ subunits remain unassembled (Green and Claudio, 1993). To determine regions on the subunits responsible for these differences, we constructed subunit chimeras in which homologous regions of the γ and δ subunits were swapped. In the chimera δ221γ, regions of the δ subunit were replaced by regions of the γ subunit starting at amino acid 221, which is at the junction between the extracellular N-terminal domain and the first transmembrane region, M1 (Fig. 1 A). As expected, the molecular weight of the chimera appears to be between that of the γ and δ subunits (Fig. 1 A). Cotransfection of the δ221γ chimera with wild-type α, β, and γ subunits yielded expression of Bgt binding receptors on the cell surface at levels similar to those seen with all four wild-type subunits (Fig. 1 B). Consistent with previous studies (Kurosaki et al., 1987; Sine and Claudio, 1991), surface Bgt binding receptors were observed with the transfection of the wild-type subunit cDNAs without either the γ or δ subunit cDNAs (Fig.1 B). However, the levels of surface Bgt receptors lacking γ or δ subunits were very low, ∼10% or less of control, reflecting very inefficient assembly without all four subunits. The addition of δ221γ subunits to wild-type α, β, and δ subunits resulted in little to no increase in the number of Bgt binding sites above what was observed with the expression of α, β, and δ subunits alone (Fig. 1 B).
Further experiments tested whether AChRs containing the δ221γ chimera were pentamers with two ACh binding sites. The size of surface receptors was examined by using sucrose gradients (Fig. 1 C). A single peak, at 9 S, was indistinguishable from the location where AChRs composed of the four wild-type subunits migrated. This finding indicates that δ221γ-containing AChRs are pentamers with a stoichiometry of α2βγ(δ221γ), with δ221γ replacing the δ subunit in the Torpedo, wild-type α2βγδ complexes. To test whether the δ221γ-containing AChRs have two distinguishable ACh binding sites like wild-type AChRs, we used mAb 247G, which recognizes and blocks Bgt binding to one ACh site, but not the other (Mihovilovic and Richman, 1987; Green and Wanamaker, 1998). When cells expressing δ221γ-containing AChRs were incubated with increasing amounts of mAb 247G, ∼50% of the Bgt binding sites were blocked, confirming the presence of two different ACh binding sites. Similar results were obtained previously by using a mouse δ221γ chimera (Sine, 1993) and, together with our results, indicate that the δ221γ chimera can substitute for the δ subunit, but not the γ subunit, during AChR assembly.
The C-terminal half of the γ subunit is required for Bgt binding site formation
The chimera γ215δ was constructed by replacing the δ subunit N-terminal domain with that of the γ subunit (Fig.2 A). Like the δ221γ chimera, γ215δ migrates as expected between the γ and δ subunits on SDS-PAGE gels (Fig. 2 A). Unlike the δ221γ chimera, the γ215δ chimera does not substitute for the γ subunit during AChR assembly because the addition of γ215δ to α, β, and δ subunits did not increase AChR expression (data not shown). However, the γ215δ chimera is not inert during assembly, as evidenced by a complete loss of surface AChRs when γ215δ chimera cDNAs are transfected with the four subunit cDNAs (Fig.2 B). In addition to blocking the expression of surface AChRs, γ215δ chimeras blocked the formation of all intracellular Bgt binding sites (Fig. 2 B). The loss of surface Bgt binding sites because of the γ215δ chimera thus is caused by a failure of the sites to form and not by a block of their transport to the cell surface. The dominant-negative effect of the γ215δ chimera differs from the inhibition of expression observed when a single AChR subunit is overexpressed relative to the other three subunits (Gu et al., 1991). Overexpression of the δ subunit, which most affected AChR expression, did not alter significantly the number of intracellular Bgt binding sites (Gu et al., 1991), unlike the effect of adding the γ215δ chimera (Fig. 2 B). Additionally, overexpression of the δ subunit caused, at most, a 40–50% decrease in expression (Gu et al., 1991). In our experiments, transfection of an additional 10 μg of the δ subunit cDNA caused a <20% decrease in AChR expression, whereas transfection of the same amount of the γ215δ chimera cDNA decreased expression by >90% (Fig. 2 B).
Two more chimeras were constructed to test whether the dominant-negative effect of the γ215δ chimera is caused by the loss of γ subunit regions or by the addition of δ subunit regions. The chimeras γ215α and γ215β contained the N-terminal domain of the γ subunit and the C-terminal half contributed by either the α or β subunit (Fig. 2 C). The translated product of both chimera cDNAs appeared to be full-length, because γ215α migrates on gels between the γ and α subunits and γ215β migrates between the γ and β subunits (Fig. 2 C). When cotransfected with all four wild-type subunits, both the γ215α and γ215β chimeras blocked the formation of surface AChRs (Fig. 2 D). The dominant-negative effect of the γ215β chimera was less pronounced than that of the γ215δ or γ215α chimeras. This difference was not caused by the assembly of any γ215β-containing AChRs, because the transfection of γ215β chimeras with α, β, and δ subunits did not increase the surface expression of AChRs (Fig.2 E). In summary, a dominant-negative effect occurs regardless of whether the replaced regions are from the α, β, or δ subunits. Regions from these subunits, therefore, do not contribute, and the dominant-negative effect results from a loss of signals within the C-terminal half of the γ subunit.
The effect of the γ215δ chimera on AChR assembly
Because Bgt binding sites form during AChR assembly (Merlie and Lindstrom, 1983; Green and Claudio, 1993; Green and Wanamaker, 1998), the block of Bgt binding site formation by the γ215δ chimera appears to be altering AChR assembly. To determine more precisely how the γ215δ chimera affects AChR assembly, we stably transfected the γ215δ chimera into a AChR-expressing cell line we have used previously to characterize AChR assembly (αβγδ cells; Claudio et al., 1987). Eighteen cell lines were isolated after stable transfection of the γ215δ cDNA into αβγδ cells. All isolated cell lines displayed a significant reduction in surface AChR expression as assayed by Bgt binding. A single cell line was chosen for characterization in which Bgt binding was not totally eliminated but was greatly reduced. As displayed in Figure3 A, cell-surface Bgt binding to these cells was reduced sevenfold as compared with the parent cell line. There was also an approximately twofold decrease in the levels of the wild-type subunits relative to the αβγδ cells as determined by metabolically labeling the subunits (Fig. 3 B). The decrease in wild-type subunit levels, apparently caused by the added γ215δ chimera synthesis, may be contributing to the loss of Bgt binding, but it is unlikely to be the major factor. The reduction in the subunit levels is much less than the decrease in Bgt binding. Moreover, we have shown in Figure 2 B that the γ215δ chimera causes a specific inhibition of AChR expression that is not duplicated by the overexpression of other AChR subunits.
Subunit assembly was assayed by metabolically labeling the subunits for 30 min and following changes in the labeled subunits for the indicated times (Fig. 3 C–E). When labeled subunits from the αβγδ cells were immunoprecipitated with α subunit-specific polyclonal antibodies, β and γ subunits coprecipitated with α subunits consistent with the assembly of αβγ trimers during the label period (Fig. 3 C, left). Initially, the α subunit signal was larger than that of the other subunits because both assembled and unassembled α subunits were precipitated by the anti-α antibodies. Excess unassembled α subunits degraded with a half-life of 8–10 hr at this temperature (Fig. 3 F) (Green and Claudio, 1993; Green and Wanamaker, 1997), whereas the assembled α subunits were stable. Labeled subunits also were immunoprecipitated with mAb 14 (Fig. 3 C, right), which recognizes a conformation-dependent epitope formed as δ subunits are added to αβγ trimers, and only precipitates complexed subunits well after the assembly of αβγ (Green and Claudio, 1993). Subunit assembly was changed significantly by the addition of γ215δ chimeras to the parent cell line (Fig. 3 D). When the labeled subunits were immunoprecipitated with the α subunit-specific antibodies, only γ215δ chimeras significantly coprecipitated (Fig.3 B,D). Also, β subunit-specific mAb coprecipitated predominantly γ215δ chimeras, and the γ/δ subunit-specific mAb coprecipitated α and β subunits with γ215δ chimeras (Fig.3 B,E). Thus, instead of AChR assembly, γ215δ chimeras associated with α or β subunits, and complexes with all three subunits were not observed.
The γ215δ-containing complexes were not stabilized like normal assembly intermediates and were degraded at the same rate as unassembled wild-type subunits (Fig. 3 F). Furthermore, subunit folding events that normally occur on partially assembled complexes during AChR assembly were not observed for the γ215δ-containing subunit complexes. None of the γ215δ-containing complexes was precipitated by mAb 14 (Fig.3 D, right) and thus lacked the mAb 14 epitope. In the γ215δ-expressing cells, a small number of subunit complexes were precipitated by mAb 14 (see last lane in Fig.3 D). Importantly, these complexes contained only the four wild-type subunits and appear to be the small number of AChRs that do assemble and are transported to the cell surface (Fig. 3 A). Because the addition of γ215δ chimeras also blocks the formation of Bgt binding sites (Figs. 2 B, 3 A), neither the mAb 14 epitope nor the Bgt binding site forms on γ215δ-containing complexes. All together, our data indicate that associations between either α or β subunits and γ215δ chimeras prevent the assembly of αβγ trimers, the first stable assembly intermediate. The mAb 14 epitope and Bgt binding sites fail to form on these α-γ215δ and β-γ215δ complexes and, as misfolded partially assembled complexes, they are degraded at a rate similar to unassembled subunits.
Multiple assembly signals within the C-terminal half of the γ subunit
Additional chimeras were constructed to identify discrete regions within the C-terminal domain important for assembly. It was reported previously that a chimera composed of the ε subunit N-terminal and β subunit C-terminal domains substituted during assembly for the ε subunit, which is the mammalian adult isoform of the γ subunit (Yu and Hall, 1991). The ε/β chimera differed from the γ215β chimera in that its short extracellular C terminus was from the ε subunit, not the β subunit. To test whether the C terminus contributes to the dominant-negative effect of γ215δ, we replaced the C terminus of γ215δ by that from the γ subunit to construct γ215δ467γ (Fig.4 A). Transfection of the γ215δ467γ chimera along with the four wild-type subunits completely blocked AChR expression (Fig. 4 B), and the addition of the γ subunit C terminus does not reverse the dominant-negative effect. It is unclear why our results differ from those with the ε/β chimera.
Two more chimeras were constructed and characterized. The γ subunit region that contains the first three transmembrane domains was replaced by the homologous region of the δ subunit to produce the γ215δ329γ chimera, and the γ subunit region just after the third transmembrane domain to the C terminus was replaced by the homologous δ subunit region to construct the γ322δ chimera (Fig.4). Surprisingly, expression of either the γ215δ329γ or γ322δ chimera along with the four wild-type subunits again completely blocked expression of cell-surface AChRs (Fig. 4 D). On the basis of these results, we conclude that multiple regions within the γ subunit C-terminal half are required to overcome the dominant-negative effect observed when this domain is replaced. Minimally, two separate regions are required: one that contains the first three transmembrane domains and a second that extends from the cytoplasmic loop just after the third transmembrane domain to the C terminus.
A large number of studies have identified regions and specific residues within the AChR subunit N-terminal domain that mediate subunit associations during assembly (Yu and Hall, 1991, 1994a; Sumikawa and Gehle, 1992; Verrall and Hall, 1992; Sumikawa and Nishizaki, 1994;Kreienkamp et al., 1995). These studies did not assay the process of receptor assembly directly. Instead, assembly was assayed indirectly, either by monitoring cell-surface expression or by studying associations between subunits and/or subunit fragments in isolation of the other subunits. Thus, it remains to be determined when N-terminal regions participate during assembly and whether other subunit regions also could be involved in assembly (see also Yu and Hall, 1994b).
We found that the membrane-spanning and the cytoplasmic domains of the γ subunit are essential for complete AChR assembly. When these regions of the γ subunit are replaced by the homologous regions of the α, β, or δ subunits, AChRs no longer fully assemble, as shown by a complete block of AChR surface expression and Bgt binding site formation. Our approach differed from previous work in that methods were used that directly assayed full AChR assembly and allowed us to determine which steps in assembly were altered. We found that, by altering the C-terminal half of the γ subunit, assembly was blocked at a step that prevented the appearance of αβγ trimers. Instead of assembling into trimers, γ215δ chimeras associated with α and β subunits, but not with γ and δ subunits. These results indicate that the initial subunit associations required for the assembly of αβγ trimers, i.e., the recognition of α and β by the γ subunit, are preserved when γ215δ chimeras replace γ subunits. Thus, the N-terminal domain of the γ subunit in the γ215δ chimera appears to mediate the initial subunit associations. Importantly, γ215δ chimeras only interact with α or β subunits, not both together. This finding further indicates that γ215δ chimeras block AChR assembly at a step in which heterodimers specifically recognize a third subunit for assembly into αβγ trimers.
The results from this paper, together with data from other studies, indicate that the first steps of AChR assembly occur as shown in Figure5. We have shown previously that α, β, and γ subunits rapidly associate into αβγ trimers during or just after subunit synthesis, and interactions between N-terminal domains may be in part cotranslational (Green and Claudio, 1993). The data from this study further indicate that αγ and βγ heterodimers precede the rapid assembly of αβγ trimers, which suggests that assembly is initiated when the N-terminal domains of two subunits interact. The space between ribosomes on the endoplasmic reticulum (ER) membrane is ∼500 Å, too large a distance to allow associations between two partially synthesized subunits (Hurtley and Helenius, 1989) but, as shown in Figure 5, would allow the N-terminal domain of a subunit undergoing synthesis to interact with a subunit no longer associated with the translocation machinery. A similar process also may initiate K+ channel assembly in which a region of the subunit N-terminal domain is required for rapid, perhaps cotranslational, interactions between subunits (Deal et al., 1994). According to the “heterodimer” model of AChR assembly (Blount et al., 1990; Gu et al., 1991; Saedi et al., 1991), α and the other subunits remain unassembled for a considerable time, during which the subunits fold and the Bgt binding site is created on the α subunits. This feature of the assembly model is difficult to reconcile with our results, in particular our finding that γ subunit chimeras prevent the formation of Bgt binding sites when added to wild-type subunits.
Interactions between subunits show a lack of subunit specificity, because homomeric associations occur when each of the four AChR subunits is expressed alone (Paulson et al., 1991) or when any heteromeric pair of subunits is expressed (Green and Claudio, 1993). These subunit associations also appear to form rapidly (Green and Claudio, 1993) and involve the N-terminal domain of the subunit (Verrall and Hall, 1992). The lack of subunit specificity, together with the rapidity by which associations occur, suggests that the function of the initial subunit associations is to protect critical subunit domains from exposure to the membrane or the aqueous environment and to prevent the misfolding of these domains. As shown in Figure 5, A and B, a lack of subunit specificity also can help to explain why AChRs assemble inefficiently such that only 20–30% of the synthesized subunits assemble into AChRs and the rest of the subunits are degraded (Merlie and Lindstrom, 1983; Ross et al., 1991). As modeled in Figure 5 A, productive subunit associations occur with the assembly of αβ, αγ, or βγ heterodimers. If each potential dimer assembles with equal probability, only 30% of the initial interactions would form productive subunit pairs. Dimers assembled from other subunit pairs, i.e., αα, ββ, γγ, δδ, αδ, βδ, or γδ, would not continue into αβγ trimers. Although these dimers can form, they are not detected when all four subunits are present (Green and Claudio, 1993;Green and Wanamaker, 1998), indicating that the associations are relatively weak and short-lived. The transient nature of these subunit interactions could allow some of the subunits to reassociate with other subunits, perhaps to form a productive pairing. However, many of the subunits that formed unproductive pairs would be expected to misfold and be rapidly degraded (Fig.5 B).
Because there appears to be little subunit specificity in the initial subunit associations among AChR subunits, we would expect the αβ, αγ, and βγ heterodimer associations to be no tighter than any other combination. What then stabilizes the productive αβ, αγ, and βγ heterodimers? In previous studies we showed that the addition of δ subunits to αβγ trimers is preceded by the folding of the trimer subunits, as evidenced by the appearance of the Bgt binding site and mAb 14 epitope. Similarly, the addition of the second α subunits to αβγδ tetramers is preceded by the folding of the tetramer subunits (Green and Claudio, 1993; Green and Wanamaker, 1997, 1998). We propose that a similar set of events occurs after a productive heterodimer forms and is required for αβγ trimer assembly (Fig. 5 A). The correct subunit combination in the dimer allows the subunits to fold into a conformation that strengthens the association and also allows the appropriate third subunit to bind tightly to the dimer (Fig. 5 A).
The α(γ215δ) and β(γ215δ) heterodimers form a tightly associated complex but fail to recognize the third subunit needed for assembly into αβγ trimers (Fig. 5 C). This finding indicates that the γ subunit C-terminal regions are involved in the subunit folding that creates a new subunit recognition site and are not necessarily part of the site where the third subunit associates with the heterodimer. Expression of the γ215δ chimera has a dominant-negative effect, therefore, because association with α or β subunits allows for strengthening of the association, but the dimer fails to fold into a conformation that will allow assembly of the next subunit. It is possible that the chimera can associate properly with α or β subunits but cannot undergo subsequent conformational changes because the C-terminal half of the chimera is misfolded. Although the difference between starting in a misfolded state versus a failure to fold correctly is subtle, several of our findings suggest that the γ215δ chimera is not misfolded. The chimera is not recognized by the cells as misfolded because it is degraded at the same rate as unassembled wild-type subunits (see Fig. 3 F). Furthermore, similar effects on surface expression were observed when different regions of the C-terminal half of the γ subunits were substituted by the corresponding δ subunit region (see Fig. 4). More generally, it is possible that all of the nonproductive dimers as well as the γ215δ-containing dimers are unable to fold correctly and contribute to the assembly of αβγ trimers because they lack the appropriate C-terminal regions. This conclusion about the role of the γ subunit C-terminal regions is supported by the dominant-negative effect of all of the additional γ-δ subunit chimeras described in Figure 4. We found that replacing any part of the C-terminal half of the δ subunit on the γ215δ chimera by the appropriate γ regions failed to prevent the dominant-negative effect. This failure to rescue cell-surface expression by replacing piece-by-piece all of the δ regions by γ regions argues that there is no general C-terminal motif that is part of the association site. Instead, it is consistent with the idea that the site where the third subunit associates is created by rearrangement of the heterodimer subunits, and the ability to undergo the rearrangement requires the C-terminal regions of the γ subunit.
The picture that emerges from this work is that associations between AChR subunits are at first weak and nondiscriminating. If the purpose of the initial subunit associations is, indeed, to protect critical subunit domains and prevent misfolding, then a relatively weak and promiscuous type of association fits with an apparent chaperone role of the subunit associations. Other proteins, such as immunoglobulin-binding protein (BiP) and calnexin, that are endogenous ER chaperones also transiently interact with newly synthesized AChR subunits (Blount and Merlie, 1991; Paulson et al., 1991; Forsayeth et al., 1992; Gelman et al., 1995; Keller et al., 1996). Only the correct associations between AChR subunits subsequently are strengthened and stabilized as the subunits continue to fold and mature during assembly of the AChR. Because these assembly events occur in the plane of ER membrane (Smith et al., 1987), cells must be solubilized to purify assembling subunits. Previously, we showed that the ability to purify intact assembly intermediates is dependent on the detergent used and the presence of phospholipids during the solubilization (Green and Claudio, 1993). These milder solubilization conditions allow us to observe intact αβγ trimers and αβγδ tetramers, but we have never observed intact heterodimer precursors. It is likely, therefore, that even the mildest of solubilization protocols disperses the heterodimers because of the weakness of the interaction and that subunit associations strengthen as assembly continues from dimer to trimer to tetramer to the final pentamer, which can be dispersed only by SDS.
This study was funded by grants from National Institutes of Health and the Brain Research Institute to W.N.G. We are most grateful to Dr. J. Lindstrom for mAbs 14, 148, and 168 and to Dr. T. Claudio for the cell line stably expressing the four Torpedo AChR subunits and the anti-α and δ antiserum. We also thank Dr. A. Fox and members of the Green laboratory for discussion and comments about this paper.
Correspondence should be addressed to Dr. William N. Green, Department of Pharmacological and Physiological Sciences, University of Chicago, 947 East 58th Street, Chicago, IL 60637.