Role of Rapsyn Tetratricopeptide Repeat and Coiled-coil Domains in Self-association and Nicotinic Acetylcholine Receptor Clustering*

Rapsyn, a 43-kDa peripheral membrane protein of skeletal muscle, is essential for clustering nicotinic acetylcholine receptors (nAChR) in the postsynaptic membrane. Previous studies with rapsyn NH2-terminal fragments fused to green fluorescent protein, expressed in 293T cells along with nAChRs, establish the following: Rapsyn-(1–90), containing the myristoylated amino terminus and two tetratricopeptide repeats (TPRs), was sufficient for self-association at the plasma membrane; rapsyn-(1–287), containing seven TPRs, did not cluster nAChRs; whereas rapsyn-(1–360), containing a coiled-coil domain (rapsyn-(298–331)), clustered nAChRs. To further analyze the role of rapsyn structural domains in self-association and nAChR clustering, we have characterized the clustering properties of additional rapsyn mutants containing deletions and substitutions within the TPR and coiled-coil domains. A mutant lacking the coiled-coil domain alone (rapsyn-(▴288–348)), failed to cluster nAChRs. Within the coiled-coil domain neutralization of the charged side chains was tolerated, while alanine substitutions of large hydrophobic residues resulted in the loss of nAChR clustering. Rapsyn self-association requires at least two TPRs, as a single TPR (TPR1 or TPR2 alone) was not sufficient. While TPRs 1 and 2 are sufficient for self-association, they are not necessary, as TPRs 3–7 also formed clusters similar to wild-type rapsyn. Fragments containing TPRs co-localized with full-length rapsyn, while the expressed coiled-coil or RING-H2 domain did not. These results are discussed in terms of a homology model of rapsyn, based on the three-dimensional structure of the TPR domain of protein phosphatase 5.

A characteristic and important feature of the vertebrate neuromuscular junction is the clustering of nicotinic acetylcholine receptors (nAChR) 1 at high surface density (ϳ10,000/ 2 ) in the post-synaptic membrane underlying the nerve terminal, with the nAChR surface density decreasing by a factor of 100 -1000 within a few microns. Rapsyn, a peripheral membrane protein of muscle, plays a critical role in the clustering of nAChRs and in the organization of the postsynaptic cytoskeletal complex (for a recent review, see Ref. 1). Mutant mice lacking rapsyn show severe neuromuscular abnormality marked by the absence of nAChR clusters in the postsynaptic membrane as well as the absence of cytoskeletal components such as utrophin and dystroglycan (2).
We recently provided a first characterization of the rapsyn structural domains involved in membrane targeting, self-association, and nAChR clustering (6). Our results indicated that rapsyn NH 2 -terminal fatty acylation is both required and sufficient for membrane targeting and that rapsyn-(1-90), containing two TPR domains, is sufficient for self-association. In addition, while rapsyn-(1-360) clustered nAChRs, rapsyn-(1-287) did not, indicating that rapsyn-(287-360) is important for nAChR clustering. Within rapsyn-(287-360) we identified a putative coiled-coil domain, rapsyn-(298 -331), and showed that disruption of the coiled-coil propensity by alanine insertions prevented nAChR clustering. These results indicated that the COOH-terminal rapsyn RING-H2 domain and the adjacent phosphorylation sites were not essential for either rapsyn or nAChR clustering.
We report here further mutational analyses of rapsyn to elucidate the role of the coiled-coil domain in nAChR clustering and the role of the TPRs in rapsyn self-association. These rapsyn mutants, fused at their COOH termini to GFP, were expressed transiently in 293T cells along with nAChR subunits and were visualized by fluorescence microscopy 24 -36 h after transfection. The results indicate that nAChR clustering depends critically on the structure of the hydrophobic surface of the coiled-coil domain but not on the hydrophilic surface. While TPRs mediate rapsyn self-association, nAChR clustering is retained in mutants lacking as many as four TPRs.

Materials and Methods
All restriction enzymes were purchased from New England Biolabs except Bpu1102I (Life Technologies, Inc.). T4 DNA ligase and oligonucleotides were obtained from Life Technologies, Inc. Polymerase chain reactions (PCR) were carried out in 100 l using 20 ng of the template pGL-rapsyn (mouse rapsyn subcloned into pGreenLantern vector; Life Technologies, Inc.), 50 -100 pmol of each primer, 250 M each dNTP, and 5 units of Pfu DNA polymerase (Stratagene) for 30 cycles at 94°C for 2 min, 50°C for 1 min, and 72°C for 1 min. All constructs were tested both by restriction enzyme analysis and by sequencing across the full-length of the inserted fragments.

Plasmid Construction and Mutagenesis
Rapsyn- -GFP-This construct encodes rapsyn amino acid sequence 1-41 containing the consensus sequence for N-myristoylation and TPR1. Primers GACACTATAGAAGGTACG and GCAAGTACTGC-CCACGAGGTCAGAGCCCTT were used along with the template pGLrapsyn-(1-412)-GFP (6) in a PCR to amplify rapsyn amino acids 1-41. The PCR product with SpeI and ScaI ends was subcloned into pBK-CMV vector (Stratagene). The resulting plasmid was digested with SpeI and NotI, and the fragment containing the rapsyn sequence was subcloned into pGL vector lacking the GFP insert (made by digesting pGreenLantern with NotI, removing the GFP insert, and religating the vector fragment). GFP cDNA was then inserted at the NotI site to result in pGL-rapsyn-(1-41)-GFP.
Rapsyn-(1-340)-GFP-This construct, which encodes rapsyn amino acids 1-340, terminates immediately after the coiled-coil domain and lacks the RING-H2 domain as well as the 20 amino acids between the coiled-coil and the RING-H2 domains. This construct differs from rapsyn-(1-360)-GFP (6), which includes these 20 amino acids. Primers GATGGGCCCGCGGCCGCTCTAGAAGTCCCTTTGCTGCGGTAGAT-GCTCTCACT and TACGACTCCGCTATGAGC were used along with pGL-rapsyn as a template in a PCR to amplify rapsyn cDNA sequence from 805-1020. The underlined letters denote the introduced XbaI site. The resulting PCR product was digested with PmlI and XbaI, and it was substituted in place of the PmlI-XbaI fragment of pGL-rapsyn-GFP, resulting in pGL-rapsyn-(1-340)-GFP.
Rapsyn-(287-340)-GFP-This construct encodes the coiled-coil domain of rapsyn in fusion with GFP. Primers AGATAAAATAGAAT-AACTAGTGCCACCATGCACGTGCTGCTGGGTGTGGCCAAGTGCT-GGATG and GCCATCCAGTTCCACGAG were used along with pGLrapsyn-(1-340)-GFP as a template in a PCR to amplify rapsyn cDNA sequence from 859 to 1020. The resulting PCR product was digested with SpeI and XbaI and then inserted into pGL vector.
Rapsyn-(298 -331)-LIV 3 A-GFP-This construct encodes a fulllength rapsyn mutant with Ala replacements of the hydrophobic residues (Leu, Ile, and Val) within rapsyn-(298 -331), the coiled-coil domain. Mutations were introduced by sequential PCR steps. The first PCR was done using the primers AGCCCTGTTCTTTCCCTG and GTC-CTGGGCTTTCTCAGCGGCATCCGCAGCCTTGTCTTGCGCCTTCCGGG-CCATCCA with pGL-rapsyn-GFP as a template. The highlighted codons represent Ala replacements of Ile 309 (12), Leu 306 (9), and Val 301 (4) where the numbers in parentheses refer to the position within the coiled-coil domain with number 1 being rapsyn-(298) (Ala). The second PCR was done using the primers GAGAAAGCCCAGGACGCAGCTGAAGA-GGCTGGCAATAAGGCGAGCCAGGCCAAGGCGCATTGCGCGAG-TGAGAGCATCTA and ACCACGCCAGTGAACAGT with pGL-rapsyn-GFP as a template. In the above primers, codons written in small letters represent the overlapping sequence between the two PCR products. The highlighted codons represent Ala replacements of Leu 315 (18), Val 319 (22), Leu 323 (26), Leu 326 (29), Leu 328 (31), and Leu 331 (34), with the number in the parentheses corresponding to the numbers used in Fig.  3a. The two fragments encompassing the mutation were annealed with each other, extended by mutually primed synthesis, and amplified by a second PCR. The resulting fragment was then digested with PmlI and FIG. 1. Schematic diagram of the structural domains of rapsyn. Indicated are the myristoylated NH 2 terminus (N-myr) and the borders of the seven putative TPRs, the coiled-coil domain, the RING-H2 domain, and the consensus sequence for phosphorylation.
XbaI and then substituted in place of the corresponding wild-type rapsyn sequence in pGL-rapsyn-GFP.

Expression of Rapsyn, Rapsyn Mutants, and nAChR in 293T Cells
Transfection of 293T cells by the calcium phosphate method, cell staining, and immunofluorescence experiments were done as described (6), except that the staining was done on cells fixed for 20 min with 2% paraformaldehyde followed by an incubation with a buffer containing PBS, 10% calf serum, 4% bovine serum albumin, and 100 mM L-lysine for 1 h. Under these conditions, ␣-bungarotoxin binding to surface nAChRs was similar to that of unfixed cells, and the distribution of nAChRs on the cell surface was generally uniform with submicron granularity. Most of the rapsyn clusters ranged from 1 to 5 in size with occasional larger plaques of 5-10 . Cells were visualized using a Nikon Eclipse E800 epifluorescence microscope with a Nikon 100X Plan Fluor objective (NA1.3). Green and red fluorescence were visualized through Nikon 91617 (excitation, 480 Ϯ 20 nm; emission, 515-560 nm) and 96171 filters (excitation, 540 Ϯ 10 nm; emission, 590 -650 nm), respectively. In some experiments photographs were taken with Kodak 160T film and digitized. Most of the images were acquired with a Micromax CH250 CCD camera (Princeton Instruments) using Meta-Morph software. Figures were prepared from the digitized images using Adobe Photoshop.
Anti-rapsyn monoclonal MAb19F4A, used in the identification of rapsyn in Fig. 6, was prepared by using Torpedo rapsyn as immunogen (16). Based upon epitope mapping studies it has been shown to recognize rapsyn amino acid sequence 396 -411. 2 In our immunofluorescence studies, it did not recognize rapsyn-(1-360) or other smaller fragments (data not shown).

Quantification of Cells with Rapsyn or nAChR Clusters
In general, within each cell expressing both rapsyn and nAChR, not all rapsyn clusters are associated with nAChRs. This is expected since only the surface nAChRs are labeled with ␣-bungarotoxin, while rapsyn-GFP is distributed at the plasma membrane and also intracellularly. For experiments involving expression of rapsyn-GFP (or mutants) and nAChRs in 293T cells, quantification of nAChR clusters relative to rapsyn clusters was done as follows. In each experiment, 100 cells positive for both rapsyn and nAChR expression were identified. For these we quantified the number of cells with nAChRs that colocalized with rapsyn clusters and the number of cells with nAChRs distributed diffusely. In experiments involving rapsyn mutants with TPR deletions, in some cells nAChRs were seen both clustered with rapsyn and also distributed diffusely on the surface. For quantification, these cells were classified as containing clustered nAChRs. Each experiment was repeated at least four times, cells were scored as above, and the results are presented as the % of cells with nAChR clusters (mean and S.D.).

Molecular Modeling of Rapsyn
All molecular modeling studies were done with Insight II, Version 98 (MSI, San Diego, CA) on a Silicon Graphics IRIS work station. The mouse rapsyn TPR domain was homology modeled based upon the three-dimensional crystal structure of the TPR domain of protein phosphatase 5 (17). The coordinates of this domain, containing 3 TPRs, were obtained from the Brookhaven Protein Data Base (PDB number 1A17). The sequences of the rapsyn TPRs 1-7 (14) were used sequentially to create the model. TPRs 1-3 and 4 -6 were each modeled individually, and then they were combined in register. TPR 7, initially modeled with TPR 8, was then added. For rapsyn, the intervening sequences between TPRs 2 and 7 each contain 6 -9 amino acids, while in protein phosphatase 5 there are no residues between TPRs. For rapsyn random loops were generated for these sequences, and these loops were then fused with the TPR structures. Rapsyn amino acids 286 -331 were modeled as an energy minimized, single long helix, based on our previous prediction that rapsyn 298 -331 has a high propensity to be organized as an ␣-helical coiled-coil (6). The orientation of the coiled-coil domain with respect to the TPRs was done arbitrarily and represents only one of the possibilities. The entire rapsyn model was then energy minimized using the Discover module, first with steepest descent algorithm and then by conjugate gradient algorithm until the energy of the entire molecule was low.

Rapsyn Coiled-coil Domain Is Essential for
Clustering nAChRs-Previously we had shown (6) that rapsyn-(1-287)-GFP could self-associate but not cluster nAChRs, while rapsyn-(1-360)-GFP could cluster nAChRs. Furthermore, rapsyn amino acids 298 -331 were predicted to have a high propensity to form an ␣-helical coiled-coil, and alanine insertions within this region that disrupted coiled-coil propensity also disrupted rapsyn's ability to cluster nAChRs without affecting rapsyn self-association. To further assess the importance of the rapsyn coiled-coil domain in nAChR clustering, we made two additional constructs: (i) rapsyn-(1-340)-GFP, with the rapsyn sequence terminated immediately after the coiled-coil domain; and (ii) rapsyn-(OE288 -348)-GFP, full-length rapsyn with the coiled-coil domain deleted (Fig. 2a). When expressed in 293T cells, rapsyn-(1-340)-GFP formed distinct membrane-associated clusters in all transfected cells and clustered nAChRs at the cell surface (Fig. 2, b and c) in 95 Ϯ 5% of cells expressing 2 J. Cohen, unpublished data. both proteins. The appearance of rapsyn/nAChR clusters was indistinguishable from that of wild-type rapsyn. In contrast, the construct lacking the coiled-coil domain, rapsyn-(OE288 -348)-GFP, formed membrane-associated clusters similar to wild type but did not cluster nAChRs (Fig. 2, d and e). When nAChR distribution was quantified in six separate transfection experiments, nAChRs were clustered with rapsyn in only 8 Ϯ 2% of cells. These results establish that the coiled-coil domain is essential for nAChR clustering, while the adjacent amino acids 340 -360 of rapsyn are not necessary.
The Structure of Hydrophobic Face of the Rapsyn Coiled-coil Domain Is Important for nAChR Clustering-Rapsyn-(298 -331) has a high propensity to organize as a coiled-coil domain, since it can form a 10-turn amphipathic ␣-helix with a hydrophobic moment of 0.6 (6) (Fig. 3a). When modeled as a righthanded ␣-helix, the amphipathic character of the helix is clearly visualized with a hydrophobic face (Fig. 3b, left) and a hydrophilic face (Fig. 3b, right). Striking features of the hydrophilic surface are the presence of a strip of six-acidic side chains forming a continuous surface over four helix turns and the presence of 5 lysine residues spaced to form a lysine ladder. These lysines are conserved in rapsyn from different species. The hydrophobic surface, which extends over 10 turns, consists of 5 alanine residues in the first four helical turns and nine bulky side chains (Leu, Ile, and Val).
To analyze the importance of the particular amino acids that contribute to the hydrophobic and hydrophilic helix surfaces, three constructs with point mutations were tested (Fig. 4a) (Fig. 4, b-d) in 72 Ϯ 7% cells expressing both proteins and were diffusely distributed in the other cells. Similarly, rapsyn-(298 -322)-KR 3 Q-GFP also formed clusters in all transfected cells, and nAChRs were associated with these clusters (Fig. 4, e-g) in 81 Ϯ 11% cells expressing both proteins. In contrast, while rapsyn-(298 -331)-LIV 3 A-GFP did form clusters at the cell surface in all the cells, only in 17 Ϯ 8% of these cells were nAChRs associated with any of these clusters. In a representative cell (Fig. 4, h-j) nAChRs are not associated with rapsyn clusters although there is a microgranular distribution of nAChRs in some parts of the cell surface (see also Ref. 6). These results establish that despite the conservation of the charged residues in rapsyn-(298 -331), their presence is not necessary for nAChR clustering. In contrast, alteration of the structure of the hydrophobic surface results in the lack of nAChR clustering.
The Interaction of Rapsyn Domains with Wild-type Rapsyn-We also characterized the capacity of constructs encoding rapsyn domains to associate with wild-type, full-length rapsyn clustered at the cell surface. The distribution of rapsyn domains fused to GFP was visualized by fluorescein isothiocyanate optics, while wild-type rapsyn was visualized by the binding of rhodamine-conjugated goat anti-mouse antibody to mAb 19F4A, a mouse monoclonal that recognizes rapsyn of many species (16). Based upon epitope mapping studies with Torpedo rapsyn, mAb194A recognizes a COOH-terminal epitope (rapsyn-(396 -411)), 3 and based upon immunofluorescence it did not recognize rapsyn-(1-360) or other smaller fragments expressed in 293T cells (data not shown). When coexpressed with full-length rapsyn, rapsyn-(1-15)-GFP was distributed diffusely at the cell surface, but it was not associated with the rapsyn clusters (Fig. 6, a and b). Rapsyn-(1-90)-GFP, containing TPRs 1 and 2, formed clusters, as previously shown ( Fig. 5b and Ref. 6), and these clusters were co-localized with the clusters of full-length rapsyn (Fig. 6, c and d). Rapsyn-(G2A-360)-GFP, which lacks the myristoylation consensus sequence and the RING-H2 domain but contains TPRs 1-7 and the coiled-coil domain, formed clusters co-localized with the rapsyn clusters (Fig. 6, e and f). Rapsyn-(287-340)-GFP, comprising the coiledcoil domain, was distributed diffusely in the cell and did not associate with rapsyn clusters (Fig. 6, g and h). Rapsyn-(351-411)-GFP, containing the RING-H2 domain, was distributed diffusely throughout the cell and was enriched in the nucleus, but it did not associate with rapsyn clusters (Fig. 6, i and j) similar to the epitope-tagged rapsyn RING-H2 domain (19). These results further demonstrate that rapsyn self-association is specifically mediated by the TPR domain. The GFP-tagged fragments of rapsyn containing only the coiled-coil domain, the RING-H2 domain, or the myristoylated amino terminus were not recruited to clusters of wild type rapsyn at the cell surface, while rapsyn-(1-90) or the nonmyristoylated rapsyn-(G2A-360) both associated with full-length rapsyn.
Rapsyn TPRs 1-7 and nAChR Clustering-To further analyze the role of the rapsyn TPR domain in nAChR clustering, a series of constructs with internal deletions corresponding to 3  specific TPRs were created and expressed in 293T cells along with nAChRs (Fig. 7a). nAChRs were clustered by rapsyn-(OE16 -90)-GFP, the construct lacking the first two TPRs, in 62 Ϯ 6% of cells expressing both proteins, with nAChRs distributed uniformly in other cells. Fig. 7, b and c, show representative images of a cell containing clustered nAChRs, and Fig. 7, d and e, show a cell where nAChRs, distributed on the surface with some granularity, are not associated with the rapsyn clusters. Similarly, rapsyn-(OE91-254)-GFP, a construct lacking TPRs 3-6, formed distinct clusters in all the cells, with nAChRs co-clustered (Fig. 7, f and g) in 44 Ϯ 18% of cells. Other rapsyn mutants lacking either TPR6 (rapsyn-(OE194 -254)-GFP; Fig. 7, h and i) or TPR7 (rapsyn-(OE255-287)-GFP; Fig. 7, j and k) were clustered in all the cells but clustered nAChRs in 52 Ϯ 11 and 41 Ϯ 15%, respectively. These results indicate that a mutant rapsyn containing as few as three TPRs (TPRs 1, 2, and 7) is able to cluster nAChRs but that neither TPR 1, 2, or 7 is required. However, for these TPR deletion mutants there were two populations of cells, those with nAChR clusters and those without. Further studies are required to identify the differences between these two groups.
Nuclear Localization of Non-myristoylated Rapsyn-Rapsyn-(G2A)-GFP, with the amino-terminal glycine mutated to alanine to prevent N-myristoylation, is preferentially targeted to the nucleus (6,20). To determine whether it is the COOHterminal region of rapsyn containing the RING-H2 domain that contributes to the nuclear retention of rapsyn-(G2A)-GFP, we examined the distribution in 293T cells of rapsyn-(G2A-360)-GFP that contained the G2A mutation and lacked the RING-H2 domain (amino acids 361-412; Fig. 8a). When coexpressed with nAChRs in 293T cells, as expected, rapsyn-(G2A)-GFP was targeted to the nucleus, while nAChRs were distributed diffusely on the plasma membrane (Fig. 8, b and c). In contrast, rapsyn-(G2A-360)-GFP was distributed in the cytoplasm and excluded from the nucleus, with occasional enrichment at the surface. Most of the nAChRs appeared uniformly distributed on the cell surface (Fig. 8, d and e). DISCUSSION The results presented here extend our understanding of the role of the rapsyn coiled-coil and TPR domains in nAChR clus-  and h), or rapsyn-(351-411)-GFP (i and j) to interact with full-length, wild-type rapsyn was tested by transfecting 293T cells with the corresponding rapsyn constructs along with wild-type rapsyn. 24 h after transfection, the distribution of the rapsyn domains fused to GFP (left) was visualized by fluorescein isothiocyanate optics, while wild-type rapsyn (right) was visualized in rhodamine optics by the binding of anti-rapsyn mouse monoclonal 19F14A and a rhodamine-conjugated goat anti-mouse secondary antibody. mAb 19F4A, which recognizes the COOH terminus of rapsyn, does not recognize rapsyn-(1-360) or other smaller fragments. Rapsyn-(1-15)-GFP was distributed diffusely at the plasma membrane (a) and did not co-localize with rapsyn clusters (b). Rapsyn-(1-90)-GFP, containing TPRs 1 and 2, formed clusters (c) that were co-localized with wild-type rapsyn clusters (d). Rapsyn-(G2A-360)-GFP formed clusters (e) co-localized with rapsyn clusters (f). Rapsyn-(287-340)-GFP, the coiled-coil domain, was distributed diffusely (g) and did not co-localize with rapsyn clusters (h). Rapsyn-(351-411)-GFP, containing the RING-H2 domain, was distributed throughout the cell and enriched in the nucleus (i), but it did not co-localize with rapsyn clusters (j). Bar ϭ 10 m.

FIG. 7.
Rapsyn mutants containing TPR deletions can cluster nAChRs. a, schematic representation of the rapsyn deletion mutants. b-k, 293T cells were transfected with cDNAs encoding rapsyn mutants tagged with GFP, along with nAChR subunits. Distributions of rapsyn (left) and surface nAChRs (right) were visualized in cells expressing both proteins. b-e, rapsyn-(OE16 -90)-GFP, lacking TPRs 1 and 2, formed clusters in all the cells (b and d, two representative cells). In 62 Ϯ 6% cells nAChRs were co-localized with these clusters as represented by the cell in b and c. In about 35% of cells, nAChRs were distributed diffusely on the surface as represented by the cell in d and e, with some granularity (arrow) not co-localized with rapsyn clusters (arrowhead). f-k, rapsyn-(OE91-254)-GFP (f and g), lacking TPRs 3-6, rapsyn-(OE194 -254)-GFP (h and i), lacking TPR 6, and rapsyn-(OE255-287)-GFP (j and k), lacking TPR 7, formed clusters in all the cells with surface nAChRs (g, i, and k) associated with these clusters in 44 Ϯ 18, 52 Ϯ 11, 41 Ϯ 15% of the cells. The filled arrowheads indicate nAChRs co-clustered with the mutant rapsyn. The arrows in j and k indicate nAChRs distributed uniformly at the surface without rapsyn associated clusters (j). Bar ϭ 10 m.
tering and self-association, respectively. Internal deletion of the rapsyn sequence corresponding to the coiled-coil domain (rapsyn-(OE288 -348)) completely abolished nAChR clustering without affecting rapsyn self-association. This result further emphasizes that the rapsyn coiled-coil domain is the single most important domain required for nAChR clustering. While the COOH terminus of rapsyn containing the RING-H2 domain may be a determinant of the size of rapsyn and nAChR clusters (21), no nAChR clustering was seen for a rapsyn mutant containing seven TPRs and the RING-H2 domain but lacking the coiled-coil domain.
The Rapsyn Coiled-coil Domain and nAChR Clustering-Mutational analyses of the hydrophobic and hydrophilic surfaces of the rapsyn coiled-coil domain establish that it is the specific structure of the hydrophobic surface that is crucial for rapsyn to cluster nAChRs (Fig. 4). A classical coiled-coil domain is characterized by a heptad repeat sequence (a b c d e f g) n in which amino acids at positions a and d are typically hydrophobic, residues at e and g are often charged, and those at the solvent-exposed b, c, and f positions are predominantly polar. In the rapsyn coiled-coil domain there is a continuous hydrophobic surface (Fig. 3b) defined by side chains at position d (Ala 1 , Ala 8 , Ala 15 , Val 22 , and Leu 29 ) and position a (Ile 12 , Ala 19 , and Leu 26 ), along with Val 4 and Leu 18 (position g) and Leu 9 (position e). The structure of this surface is essential for nAChR clustering, since mutations of the nine large hydrophobic residues in this region, some of which are conserved across species, to alanine resulted in the loss of nAChR clustering. While this result demonstrates the importance of the hydrophobic surface, additional point mutations are necessary to define the contribution made by individual amino acids in this region.
It was surprising that either the charge-neutralizing substitutions of the highly conserved positively charged residues (lysine ladder) or the negatively charged residues had no significant effect on nAChR clustering. While our results suggest clearly that charges are not required at these positions, it is possible that the size, hydrophilicity, or hydrogen bonds forming capabilities of the side chains at these positions are important. However, the high degree of charge conservation at these positions in rapsyn from species ranging from human to Caenorhabditis elegans suggests that these side chains may be involved in ionic interactions contributing to the stability of a helix bundle as seen in other homo-oligomeric or hetero-oligomeric coiled-coil structures (22,23). Significantly, rapsyn-(287-340)-GFP, which contains only the coiled-coil domain, did not interact with full-length rapsyn. This result indicates that the coiled-coil domain is unlikely to self-associate, and it is consistent with the hypothesis that the rapsyn coiled-coil domain associates with a coiled-coil domain in the nAChR.
There is growing evidence for direct interaction between rapsyn and nAChRs. Early chemical cross-linking studies and freeze fracture immunoelectron microscopy both supported the view that rapsyn is in close proximity to the cytoplasmic domains of nAChR (24,25) (for a review, see Ref. 26). The 4.6-Å structure recently determined by electron microscopy of tubular crystals of Torpedo postsynaptic membranes included on the cytoplasmic surface electron dense regions immediately underneath the nAChR which were attributed to rapsyn (27). nAChR subunits expressed individually in nonmuscle cells can be clustered by rapsyn (28), and recent studies indicate that this clustering is mediated by the large cytoplasmic loop between the M3 and M4 transmembrane regions of the nAChR ␣ subunit (29). In addition, overlay experiments with 125 I-rapsyn suggest that rapsyn binds to itself with high affinity and to full-length nAChR ␣, ␥, or ⑀ subunit, but not to truncated ␤ or ␦ subunit lacking the cytoplasmic loop (30).
TPR domains have been widely recognized as protein interaction domains (reviewed in Ref. 31 and 32). Identified TPR binding motifs include short peptide sequences or helix bundles. Within protein chaperone complexes, the EEVD sequences at the COOH termini of Hsp70 and Hsp90 bind to distinct TPRs in the NH 2 -and COOH-terminal domains of Hop (33). In a yeast transcription repressor complex, TPRs 1-3 within SSN6 can bind to a four helix bundle at the NH 2 terminus of Tup1 (34) or to a tripeptide sequence within the homeo domain of a cell type regulator, ␣2 (35). However, a TPR domain can also be involved in more distributed binding interactions. Amino acid side chains distributed in 7 TPRs in the ␣-subunit of protein prenyltransferase make extensive contacts with the ␤-subunit side chains (36), and based upon yeast two-hybrid and biochemical assays, the TPR domain of the serine/threonine phosphatase protein 5 is involved in binding to the TPR domains of the CDC16 or CDC27 subunits of the anaphase promoting complex (37), but not to PEX5, an unrelated TPR protein. Since there are no previous reports of TPR domains mediating protein self-association, it will be of particular interest to define the structure of the rapsyn-(1-90), the minimal self-association domain, and also to determine whether the rapsyn TPR domain is involved in rapsyn binding to other proteins.
Rapsyn TPRs and nAChR Clustering-Analysis of rapsyn mutants containing TPR deletions revealed that nAChR clustering was still seen for mutant rapsyns lacking TPRs 3-6 and, furthermore, that deletions of TPRs 1 and 2 or TPR 7 (proximal to the coiled-coil domain) were tolerated. Changes in the relative orientation of rapsyn's coiled-coil domain or in the distance between the membrane anchoring site (N-Myr) and the coiledcoil domain of rapsyn due to internal TPR deletions may affect the ability of rapsyn to interact with nAChRs efficiently. Alternatively, it may be that for these deletion mutants the ability to cluster nAChRs might be particularly sensitive to the levels of protein expression or turnover.
Properties of the RING-H2 Domain-Although nonmyristoylated rapsyn (20) or rapsyn-GFP (6) is targeted to the nucleus, the primary structure of rapsyn does not contain any known consensus sequence for nuclear localization. However, the cysteine-rich domain of rapsyn is a RING-H2 domain, a structural motif found in many transcription factors localized in the nucleus (15). Hence, we reasoned that in the absence of a dominant membrane-targeting signal such as N-Myr, the RING-H2 domain of rapsyn might contribute to its nuclear retention. Consistent with this, rapsyn-(G2A-360), which lacks the RING-H2 domain, was found in the cytosol and was excluded from the nucleus. Thus, with the myristoylated NH 2 -terminal domain of rapsyn acting as dominant membrane targeting signal, the RING-H2 domain can potentially interact with other proteins at the cell surface. In the absence of the membrane-targeting signal, rapsyn's RING-H2 domain may interact with proteins that are normally localized to the nucleus and by virtue of this interaction be targeted to the nucleus.
An Homology Model of the Rapsyn TPR Domain-To provide a structural rationale for rapsyn activity, we constructed a model of rapsyn (Fig. 9A) based upon the three-dimensional structure of the 3 TPRs in the serine/threonine protein phosphatase 5 (17). Rapsyn amino acid sequence 6 -279 was modeled as having 7 TPRs (14). Rapsyn amino acids 298 -331 was modeled as a single ␣-helix, based on our previous results (6) indicating that rapsyn-(298 -331) formed a helix. TPRs 1-7 (rapsyn-(1-287)) form a super-helix extending ϳ70 Å in length. The concave (inner) surface of the superhelix has a diameter of ϳ25 Å, while that of the external surface is ϳ40 Å. In this context, it is interesting to note that in the Torpedo postsynaptic membrane, the diameter of the cytoplasmic projection is ϳ65 Å (38). In our model, helix B of TPR 7 ends at rapsyn-(279), and rapsyn-(286 -331) is modeled as a single helix beginning with three turns (rapsyn-(286 -297)) packed along helix B of TPR7 that precede the predicted coiled-coil domain (rapsyn-(298 -331)). With this packing the helix extends ϳ70 Å beyond the TPR domain with the hydrophobic surface of the coiled-coil domain oriented toward the concave surface of the TPR superhelix. We should emphasize that we have no experimental data that constrain the orientation of the coiled-coil domain relative to the TPR domain. An interesting alternative packing would be for the coiled-coil domain of rapsyn (potentially in association with a coiled-coil domain from nAChR subunit(s)) to be actually encompassed within the TPR superhelix. Such a structure has been proposed for a yeast transcription repressor complex, where a four-helix bundle at the NH 2 terminus of Tup1 is bound within the superhelix groove of TPRs 1-3 of SSN6 (34). Fig. 9a is a surface representation of rapsyn-(1-340) with acidic side chains in blue and basic in red. The two images rotated by 180°are presented to emphasize the surface structure, and the arrows point to the interior (concave) surface of the superhelix at the level of TPR2 (left) and TPR7 (right). Except for Glu 83 and Arg 90 (in TPR3) and Asp 254 (in TPR7), whose side chains point directly into the groove of the superhelix, all the other charged residues (18 Glu, 12 Asp, 18 Lys, and 14 Arg) within rapsyn-(1-287) are located either on the FIG. 9. Homology model of mouse rapsyn. Rapsyn-(1-279), consisting of seven TPRs, was homology modeled based on the known three-dimensional structure of the TPR domain of protein phosphatase 5. Rapsyn-(298 -331) was modeled as an ␣ helix based on its high coiled-coil propensity and our experimental results. a, surface representation of rapsyn-(1-340) with acidic side chains in blue and basic in red. The two images are rotated 180°to emphasize the surface structure, and the arrows point to the interior (concave) surface of the superhelix at the level of TPR2 (left) and TPR7 (right). The numbers refer to the corresponding TPR and the coiled-coil domain is denoted by cc. B, an enlarged view of the rapsyn TPRs 1-3. The CPK representation of the lysines in TPR1 is shown in red. Lysine at positions 6, 11, 23, 30, and 34 are oriented on the same face of the TPR1 helices and can potentially interact with the acidic phospholipid head groups. outer surface or in the loops connecting the TPR domains, often as ion pairs. Among these charges, it is striking that there are five noncontiguous lysines at positions 6, 11, 23, 30, and 34 in TPR1 on the external surface (Fig. 9b). Contiguous stretches of positively charged residues have been shown to be required for the membrane association of other myristoylated (src and MARCKS) or palmitoylated (GAP43) proteins (reviewed in Ref. 39). Because rapsyn primary structure does not have such a contiguous stretch of positively charged residues, it is possible that the three-dimensional alignment of these lysine residues close to the myristoylated amino terminus plays the equivalent role in rapsyn's plasma membrane association.
Based on the primary structure it has been suggested (40) that rapsyn-(82-110), which contains a heptad repeat of Leu/ Ile residues, would be organized as an ␣ helix and function as a leucine zipper coiled-coil motif potentially important for nAChR clustering (20). However, in our model, this sequence is within TPR3, with the leucines distributed on the two helices and contributing to the helix packing.
Our data clearly demonstrate that rapsyn self-association is mediated by the TPR domain. It remains to be determined whether this self-association is mediated by interactions involving the loops between the TPRs or either the external or the internal surfaces of the superhelix. While there has been no evidence of self-association in the known TPR structures, early studies of the TPR domain of the yeast nuc 2 ϩ protein, a truncated polypeptide with nine tandem TPRs, provided evidence that aggregates of as many as several hundred molecules could be formed (41). The crystal structure of the armadillo (arm) repeat domain of the nuclear import factor karyopherin ␣, consisting of 10 repeats of a three-helix bundle, reveals a homodimer with extensive contacts within the groove (42).
In the present study, we have provided evidence that the rapsyn coiled-coil domain is essential for nAChR clustering and at least two TPR repeats are necessary and sufficient for rapsyn self-association. nAChR clustering is retained for rapsyn constructs that contained the coiled-coil domain and as few as three TPRs. In the future, it will be of interest to identify the nAChR domain(s) involved in interaction with rapsyn and to determine the structure of the rapsyn coiled-coil domain in association with the nAChR domains.