Requirement of N-terminal Cysteines of PSD-95 for PSD-95 Multimerization and Ternary Complex Formation, but Not for Binding to Potassium Channel Kv1.4
1999; Elsevier BV; Volume: 274; Issue: 1 Linguagem: Inglês
10.1074/jbc.274.1.532
ISSN1083-351X
Autores Tópico(s)Neuroscience and Neuropharmacology Research
ResumoThe PSD-95 family of PSD-95/Discs large/ZO-1 (PDZ) domain-containing proteins plays a role in the clustering and localization of specific ion channels and receptors at synapses. Previous studies have shown that PSD-95 forms multimers through an N-terminal region (termed the N-segment) and that the multimerization of PSD-95 is critical for its ability to cluster Shaker-type potassium channel Kv1.4 in heterologous cells. We show here that the PSD-95 N-segment functions as a multimerization domain only when located at the N-terminal end of a heterologous protein. A pair of N-terminal cysteines, Cys3 and Cys5, is essential for the ability of PSD-95 to self-associate and to form cell surface clusters with Kv1.4. However, PSD-95 mutants lacking these cysteine residues retain their ability to associate with membranes and to bind to Kv1.4. Unlike wild type PSD-95, the cysteine mutant of PSD-95 cannot form a ternary complex with Kv1.4 and the cell adhesion molecule Fasciclin II. These results suggest that the N-terminal cysteines are essential for PSD-95 multimerization and that multimerization is required for simultaneous binding of multiple membrane protein ligands by PSD-95. The PSD-95 family of PSD-95/Discs large/ZO-1 (PDZ) domain-containing proteins plays a role in the clustering and localization of specific ion channels and receptors at synapses. Previous studies have shown that PSD-95 forms multimers through an N-terminal region (termed the N-segment) and that the multimerization of PSD-95 is critical for its ability to cluster Shaker-type potassium channel Kv1.4 in heterologous cells. We show here that the PSD-95 N-segment functions as a multimerization domain only when located at the N-terminal end of a heterologous protein. A pair of N-terminal cysteines, Cys3 and Cys5, is essential for the ability of PSD-95 to self-associate and to form cell surface clusters with Kv1.4. However, PSD-95 mutants lacking these cysteine residues retain their ability to associate with membranes and to bind to Kv1.4. Unlike wild type PSD-95, the cysteine mutant of PSD-95 cannot form a ternary complex with Kv1.4 and the cell adhesion molecule Fasciclin II. These results suggest that the N-terminal cysteines are essential for PSD-95 multimerization and that multimerization is required for simultaneous binding of multiple membrane protein ligands by PSD-95. PSD-95/Discs large/ZO-1 membrane-associated guanylate kinase hemagglutinin wild type green fluorescence protein N-ethylmaleimide Drosophila Discs large protein Fasciclin II. PDZ1 domains are modular protein-protein interaction domains that are specialized for binding to specific C-terminal peptide sequences (1Sheng M. Kim E. Curr. Opin. Neurobiol. 1996; 6: 602-608Crossref PubMed Scopus (100) Google Scholar, 2Doyle D.A. Lee A. Lewis J. Kim E. Sheng M. MacKinnon R. Cell. 1996; 85: 1067-1076Abstract Full Text Full Text PDF PubMed Scopus (972) Google Scholar, 3Ponting C.P. Phillips C. Davies K.E. Blake D.J. Bioessays. 1997; 19: 469-479Crossref PubMed Scopus (355) Google Scholar, 4Songyang Z. Fanning A.S. Fu C. Xu J. Marfatia S.M. Chishti A.H. Crompton A. Chan A.C. Anderson J.M. Cantley L.C. Science. 1997; 275: 73-77Crossref PubMed Scopus (1222) Google Scholar) and to other PDZ domains (5Brenman J.E. Chao D.S. Gee S.H. McGee A.W. Craven S.E. Santillano D.R. Wu Z. Huang F. Xia H. Peters M.F. Froehner S.C. Bredt D.S. Cell. 1996; 84: 757-767Abstract Full Text Full Text PDF PubMed Scopus (1443) Google Scholar). Many proteins contain multiple PDZ domains, thereby allowing them to function as multivalent scaffolds for organizing large protein complexes (6Pawson T. Scott J.D. Science. 1997; 278: 2075-2080Crossref PubMed Scopus (1900) Google Scholar, 7Tsunoda S. Sierralta J. Sun Y. Bodner R. Suzuki E. Becker A. Socolich M. Zuker C.S. Nature. 1997; 388: 243-249Crossref PubMed Scopus (552) Google Scholar, 8Niethammer M. Valtschanoff J.G. Kapoor T.M. Allison D.W. Weinberg R.J. Craig A.M. Sheng M. Neuron. 1998; 20: 693-707Abstract Full Text Full Text PDF PubMed Scopus (252) Google Scholar, 9Kim J.H. Liao D. Lau L.F. Huganir R.L. Neuron. 1998; 20: 683-691Abstract Full Text Full Text PDF PubMed Scopus (507) Google Scholar). One major class of PDZ-containing proteins, known as MAGUKs (membrane-associatedguanylate kinases) (10Anderson J.M. Curr. Biol. 1996; 6: 382-384Abstract Full Text Full Text PDF PubMed Scopus (219) Google Scholar), is characterized by an SH3 domain and a guanylate kinase-like domain in their C-terminal region. MAGUK proteins are typically localized in specific membrane domains such as cell junctions in epithelial cells (10Anderson J.M. Curr. Biol. 1996; 6: 382-384Abstract Full Text Full Text PDF PubMed Scopus (219) Google Scholar) and synaptic junctions in neurons (11Sheng M. Neuron. 1996; 17: 575-578Abstract Full Text Full Text PDF PubMed Scopus (296) Google Scholar, 12Kornau H. Seeburg P. Kennedy M. Curr. Opin. Neurobiol. 1997; 7: 368-373Crossref PubMed Scopus (313) Google Scholar, 13O'Brien R.J. Mammen A.L. Blackshaw S. Ehlers M.D. Rothstein J.D. Huganir R.L. J. Neurosci. 1997; 17: 7339-7350Crossref PubMed Google Scholar). MAGUK proteins are thought to play a central role in the organization of protein complexes at these specialized membrane domains. Much of what is known about PDZ protein function has come from studies of the PSD-95 family of MAGUK proteins that are predominantly localized in the postsynaptic density of excitatory synapses in the brain. The first two PDZ domains, PDZ1 and PDZ2, of PSD-95 bind specifically to a peptide sequence at the very C terminus of Shaker-type potassium channels and of N-methyl-d-aspartate receptor NR2 subunits (14Kornau H.-C. Schenker L.T. Kennedy M.B. Seeburg P.H. Science. 1995; 269: 1737-1740Crossref PubMed Scopus (1628) Google Scholar, 15Kim E. Niethammer M. Rothschild A. Jan Y.N. Sheng M. Nature. 1995; 378: 85-88Crossref PubMed Scopus (897) Google Scholar, 16Niethammer M. Kim E. Sheng M. J. Neurosci. 1996; 16: 2157-2163Crossref PubMed Google Scholar). This PDZ-mediated interaction results in the clustering of the ion channel proteins in heterologous cells (15Kim E. Niethammer M. Rothschild A. Jan Y.N. Sheng M. Nature. 1995; 378: 85-88Crossref PubMed Scopus (897) Google Scholar, 17Kim E. Cho K.-O. Rothschild A. Sheng M. Neuron. 1996; 17: 103-113Abstract Full Text Full Text PDF PubMed Scopus (476) Google Scholar,18Hsueh Y.-P. Kim E. Sheng M. Neuron. 1997; 18: 803-814Abstract Full Text Full Text PDF PubMed Scopus (185) Google Scholar) and is important for the synaptic localization of the membrane protein-binding partners in vivo (19Tejedor F.J. Bokhari A. Rogero O. Gorczyca M. Zhang J. Kim E. Sheng M. Budnik V. J. Neurosci. 1997; 17: 152-159Crossref PubMed Google Scholar, 20Thomas U. Kim E. Kuhlendahl S. Koh Y.H. Gundelfinger E.D. Sheng M. Garner C.C. Budnik V. Neuron. 1997; 19: 787-799Abstract Full Text Full Text PDF PubMed Scopus (179) Google Scholar, 21Zito K. Fetter R.D. Goodman C.S. Isacoff E.Y. Neuron. 1997; 19: 1007-1016Abstract Full Text Full Text PDF PubMed Scopus (182) Google Scholar). Recent studies have investigated the molecular mechanisms of ion channel clustering by PSD-95. Channel-clustering activity is dependent on the multimerization of PSD-95, which is mediated by a stretch of ∼64 amino acids in the N-terminal region of the protein (termed the N-segment) (18Hsueh Y.-P. Kim E. Sheng M. Neuron. 1997; 18: 803-814Abstract Full Text Full Text PDF PubMed Scopus (185) Google Scholar). Two members of the PSD-95 family, PSD-95 and chapsyn-110, can heteromultimerize with each other because they both contain the N-segment at their N-terminal ends. Moreover, a pair of cysteine residues, Cys3 and Cys5, is conserved in the N-terminal regions of PSD-95 and chapsyn-110. Mutation of either or both of these cysteines abolishes multimerization and the channel-clustering activity of PSD-95 (18Hsueh Y.-P. Kim E. Sheng M. Neuron. 1997; 18: 803-814Abstract Full Text Full Text PDF PubMed Scopus (185) Google Scholar). Based on biochemical experiments in vitro and in vivo, it was hypothesized that Cys3 and Cys5 formintermolecular disulfide bonds that stabilize the N-segment-mediated multimerization of PSD-95 (18Hsueh Y.-P. Kim E. Sheng M. Neuron. 1997; 18: 803-814Abstract Full Text Full Text PDF PubMed Scopus (185) Google Scholar). More recently, these N-terminal cysteines have also been found to be sites for fatty acid modification (palmitoylation) of PSD-95 (22Topinka J.R. Bredt D.S. Neuron. 1998; 20: 125-134Abstract Full Text Full Text PDF PubMed Scopus (245) Google Scholar). These investigators reported that mutation of Cys3 and Cys5 inhibited the association of PSD-95 with membranes and prevented PSD-95 from binding to the potassium channel Kv1.4 in heterologous cells. This would provide an alternative explanation of why Cys3 and Cys5 mutations abolish Kv1.4 clustering by PSD-95. The conclusion reached by Topinka and Bredt (22Topinka J.R. Bredt D.S. Neuron. 1998; 20: 125-134Abstract Full Text Full Text PDF PubMed Scopus (245) Google Scholar) is somewhat surprising because there are other members of the PSD-95 family that lack these N-terminal cysteines (e.g. SAP97 and the Drosophila homolog Discs large). SAP97 and Discs large (Dlg) are nevertheless membrane-associated proteins (17Kim E. Cho K.-O. Rothschild A. Sheng M. Neuron. 1996; 17: 103-113Abstract Full Text Full Text PDF PubMed Scopus (476) Google Scholar, 23Müller B.M. Kistner U. Veh R.W. Cases-Langhoff C. Becker B. Gundelfinger E.D. Garner C.C. J. Neurosci. 1995; 15: 2354-2366Crossref PubMed Google Scholar), and Dlg is known to play a critical role in the binding and clustering of Shaker ion channels and FasII at the synapse in vivo(19Tejedor F.J. Bokhari A. Rogero O. Gorczyca M. Zhang J. Kim E. Sheng M. Budnik V. J. Neurosci. 1997; 17: 152-159Crossref PubMed Google Scholar, 20Thomas U. Kim E. Kuhlendahl S. Koh Y.H. Gundelfinger E.D. Sheng M. Garner C.C. Budnik V. Neuron. 1997; 19: 787-799Abstract Full Text Full Text PDF PubMed Scopus (179) Google Scholar, 21Zito K. Fetter R.D. Goodman C.S. Isacoff E.Y. Neuron. 1997; 19: 1007-1016Abstract Full Text Full Text PDF PubMed Scopus (182) Google Scholar). In re-examining the role of the N-segment and of Cys3 and Cys5 in more detail, we report here that mutation of these N-terminal cysteines has no detectable effect on the PSD-95 binding of Kv1.4, although it abolishes the ability of PSD-95 to self-associate and to cluster Kv1.4 at the cell surface. Thus Kv1.4 binding and homomultimerization are separable functions of PSD-95, and the former activity is not dependent on the N-terminal cysteines. Moreover, we demonstrate that cysteine mutants of PSD-95 lose their ability to form a ternary complex with Kv1.4 and the cell adhesion molecule FasII, implying that multimerization is required for PSD-95 to bind simultaneously to two different PDZ ligands. All PSD-95 constructs used here contain a Myc epitope tag inserted between residue 9 and 10 of PSD-95. Expression constructs GW1-PSD-95, GW1-PSD-95 (C3S/C5S), GW1-N-GFP, GW1-N-PDZ12, and GW1-N-PDZ12(C3S/C5S) have been described previously (18Hsueh Y.-P. Kim E. Sheng M. Neuron. 1997; 18: 803-814Abstract Full Text Full Text PDF PubMed Scopus (185) Google Scholar). To construct HA-tagged PSD-95, the Myc cassette in Myc-tagged PSD-95 was replaced by an HA epitope. For the GFP-N construct, the Myc-tagged N-segment (corresponding to residues 1–64 of PSD-95) was polymerase chain reaction amplified from Myc-tagged PSD-95 and subcloned into plasmid EGFP-C1 (CLONTECH). To construct Myc-tagged SAP97, an AscI restriction site was introduced between residues 435 and 436 of SAP97 in GW1-CMV by inverse polymerase chain reaction, and a duplex oligonucleotide encoding the Myc epitope tag was inserted into the AscI site. GW1 expression constructs of WT Kv1.4, Kv1.4 C-terminal mutant (Kv1.4VA), WT FasII, and FasII C-terminal mutant (FasIIVA) have been described previously (15Kim E. Niethammer M. Rothschild A. Jan Y.N. Sheng M. Nature. 1995; 378: 85-88Crossref PubMed Scopus (897) Google Scholar,20Thomas U. Kim E. Kuhlendahl S. Koh Y.H. Gundelfinger E.D. Sheng M. Garner C.C. Budnik V. Neuron. 1997; 19: 787-799Abstract Full Text Full Text PDF PubMed Scopus (179) Google Scholar). Rabbit anti-PSD-95 ("CSK"), guinea pig anti-PSD-95 ("HM319") antibodies (15Kim E. Niethammer M. Rothschild A. Jan Y.N. Sheng M. Nature. 1995; 378: 85-88Crossref PubMed Scopus (897) Google Scholar, 17Kim E. Cho K.-O. Rothschild A. Sheng M. Neuron. 1996; 17: 103-113Abstract Full Text Full Text PDF PubMed Scopus (476) Google Scholar), and Kv1.4 antibodies (24Sheng M. Tsaur M.-L. Jan Y.N. Jan L.Y. Neuron. 1992; 9: 271-284Abstract Full Text PDF PubMed Scopus (403) Google Scholar) have been described previously. The sites in the PSD-95 protein recognized by CSK and HM319 antibodies are indicated in Fig. 1. Myc monoclonal antibody 9E10 and HA monoclonal antibody 12CA5 were purchased from Santa Cruz Biotechnology and Boehringer Mannheim, respectively. FasII monoclonal antibodies 1D4 were kindly provided by Dr. Corey S. Goodman. Transfection was performed using the LipofectAMINE reagent according to the manufacturer's directions (Life Technologies, Inc.). For immunoprecipitation, COS-7 cells in 35-mm plates at 50–70% confluency were incubated with a 1-ml Opti-MEM (Life Technologies, Inc.) mixture containing 1.6 μg of DNA and 6 μl of LipofectAMINE (Life Technologies, Inc.) for 5 h followed by incubation in Dulbecco's modified Eagle's medium. 48 h later, cells were harvested and lysed in 0.45 ml of radioimmune precipitation buffer. One-third of the cell lysate was incubated with CSK antibodies or Kv1.4 antibodies at a final concentration of 2 μg/ml at 4 °C for 2 h. 20 μl of 1:1 slurry of Protein A-Sepharose (Amersham Pharmacia Biotech) was added, and the mixture was rotated at 4 °C for 1–2 h. The immunoprecipitates were washed with a series of buffers: first, radioimmune precipitation buffer; second, 10 mm Tris (pH 7.5), 0.5% Nonidet P-40, 0.5 m LiCl; third, 10 mm Tris (pH 7.5), 0.5 m LiCl; and finally, 10 mm Tris (pH 7.5). The immunoprecipitates were eluted in SDS-loading buffer, separated by SDS-polyacrylamide gel electrophoresis, and analyzed by immunoblotting using 0.2 μg/ml Myc 9E10 or Kv1.4 antibodies. For immunocytochemistry, transfected COS-7 cells were fixed by 2% formaldehyde in phosphate-buffered saline for 15 min, permeabilized with 0.1% Triton X-100 in Tris-buffered saline for 1 min, and blocked with 3% horse serum and 0.1% bovine serum albumin in Tris-buffered saline for 30 min at room temperature. The cells were incubated with primary antibodies at 1 μg/ml at room temperature for 1 h, followed by Cy3- and fluorescein isothiocyanate-conjugated secondary antibodies (Jackson ImmunoResearch) for 1 h. Results were viewed with a Zeiss Axioskop microscope, and images were prepared for publication using Adobe Photoshop. Transfected COS-7 cells were suspended in 800 μl of homogenization buffer (50 mm Tris-HCl, pH 8, 150 mm NaCl, 0.1 mm EDTA, 0.1 mm EGTA, 1 mmdithiothreitol, and protease inhibitors) and broken by Dounce homogenization. Total cell extracts were centrifuged at 1,000 ×g for 10 min twice. The supernatants were centrifuged at 100,000 × g for 1 h at 4 °C to separate crude membrane pellet and S100 soluble fraction. Crude membrane pellets were further washed with either homogenization buffer, 1 m NaCl, or 100 mm sodium carbonate-buffered homogenization buffer and centrifuged at 100,000 × g for 1 h. Equal fractions of crude membrane pellet, S100 cytosolic fraction, final washed pellet, and soluble fractions were analyzed by immunoblotting using the HA antibody for WT PSD-95 and the Myc antibody for N-terminal cysteine mutant, PSD-95-CS. In an earlier study, we showed that deletion mutants of PSD-95 lacking a short N-terminal region of the protein (residues 1–64, termed the N-segment) (Fig.1) lost the ability to form homomultimers (18Hsueh Y.-P. Kim E. Sheng M. Neuron. 1997; 18: 803-814Abstract Full Text Full Text PDF PubMed Scopus (185) Google Scholar). Thus the N-segment is required for self-association of PSD-95. Moreover, the N-segment is sufficient to confer multimerization on a heterologous protein. A chimeric protein containing the N-segment from PSD-95 linked to the green fluorescent protein (Fig. 1,N-GFP) could be coimmunoprecipitated with wild type full-length PSD-95 using CSK antibodies directed against the SH3 domain of PSD-95 (Fig. 2 A) (18Hsueh Y.-P. Kim E. Sheng M. Neuron. 1997; 18: 803-814Abstract Full Text Full Text PDF PubMed Scopus (185) Google Scholar). In the absence of PSD-95, N-GFP was not precipitated by CSK antibodies; furthermore, GFP itself (lacking the N-segment) cannot be coimmunoprecipitated with PSD-95 (data not shown) (18Hsueh Y.-P. Kim E. Sheng M. Neuron. 1997; 18: 803-814Abstract Full Text Full Text PDF PubMed Scopus (185) Google Scholar). These results indicate that the N-segment alone can confer on a heterologous protein (GFP) the ability to associate with PSD-95. Interestingly, a chimeric protein in which the PSD-95 N-segment is fused to the C-terminal end of GFP (diagrammed in Fig. 1,GFP-N) was unable to associate with full-length PSD-95 by the co-immunoprecipitation assay, despite being efficiently expressed (Fig. 2 A). This suggests that the N-segment has to be placed at an N-terminal location to self-associate. The GFP moiety, when placed at the N-terminal end but not when placed at the C-terminal end of the N-segment, may sterically interfere with self-association mediated by the N-segment. The N-terminal cysteine residues Cys3and Cys5 are conserved between PSD-95 and chapsyn-110; these members of the PSD-95 family can form homomultimers with themselves and heteromultimers with each other (17Kim E. Cho K.-O. Rothschild A. Sheng M. Neuron. 1996; 17: 103-113Abstract Full Text Full Text PDF PubMed Scopus (476) Google Scholar, 18Hsueh Y.-P. Kim E. Sheng M. Neuron. 1997; 18: 803-814Abstract Full Text Full Text PDF PubMed Scopus (185) Google Scholar). To confirm that Cys3 and Cys5 are important for multimerization of PSD-95, both these cysteines were substituted with serines in the context of two different constructs of PSD-95 (generating N-PDZ12-CS and PSD-95-CS) (see Fig. 1). These double cysteine mutants were assayed for multimerization in coimmunoprecipitation experiments. Unlike wild type N-PDZ12, the N-PDZ12-CS mutant could not be precipitated by CSK antibodies in the presence of full-length PSD-95 (Fig. 2 B), suggesting that the Cys3 and Cys5 residues are essential for PSD-95 self-association. In previous studies, we presented evidence that PSD-95 can form disulfide-linked multimeric complexes through residues Cys3 and Cys5 (18Hsueh Y.-P. Kim E. Sheng M. Neuron. 1997; 18: 803-814Abstract Full Text Full Text PDF PubMed Scopus (185) Google Scholar). To make sure that the association of full-length PSD-95 and N-PDZ12 is not because of artifactual disulfide bonding that occurred during cell lysis,N-ethylmaleimide (NEM) was included in the lysis buffer at 10 mm to inactivate free sulfhydryl groups. Even in the presence of NEM, N-PDZ12 was still as efficiently associated with full-length PSD-95 as in standard lysis conditions (Fig.2 C). The same result was obtained with 20 mm NEM added to the cell culture for 20 min prior to cell harvesting (data not shown). Taken together, these results indicate that PSD-95 multimerization requires Cys3 and Cys5; however, PSD-95 multimerization is not because of artifactual oxidation of these residues during or after cell extract preparation. The double cysteine mutants N-PDZ12-CS and PSD-95-CS lack multimerization activity (Figs. 1 and 2). In addition, these mutants are incapable of forming cell surface-associated clusters with Shaker-type potassium channel Kv1.4 in heterologous cells (Fig.3 A) (data not shown for N-PDZ12-CS). Wild type PSD-95 formed plaque-like coclusters with Kv1.4 in doubly transfected COS-7 cells (Fig. 3 A, a), as has been shown previously (15Kim E. Niethammer M. Rothschild A. Jan Y.N. Sheng M. Nature. 1995; 378: 85-88Crossref PubMed Scopus (897) Google Scholar, 17Kim E. Cho K.-O. Rothschild A. Sheng M. Neuron. 1996; 17: 103-113Abstract Full Text Full Text PDF PubMed Scopus (476) Google Scholar, 18Hsueh Y.-P. Kim E. Sheng M. Neuron. 1997; 18: 803-814Abstract Full Text Full Text PDF PubMed Scopus (185) Google Scholar). These PSD-95-Kv1.4 coclusters are on or closely associated with the cell surface (25Kim E. Sheng M. Neuropharmacology. 1996; 35: 993-1000Crossref PubMed Scopus (135) Google Scholar). However, in COS cells coexpressing Kv1.4 and double cysteine PSD-95 mutants, the typical plaque-like coclusters did not form (Fig. 3 A, b) (data for N-PDZ12-CS not shown). Instead, both Kv1.4 and PSD-95 cysteine mutants were often concentrated in brightly staining intracellular structures that colocalized with each other. These intracellular structures were typically spherical and had brightly immunoreactive perimeters, suggesting that they were membrane-bound. The membranous nature of these stained structures is consistent with the fact that the Kv1.4 ion channel is an integral membrane protein. The appearance of the intracellular staining seen with Kv1.4 and PSD-95-CS is strikingly reminiscent of that seen upon coexpression of Kv1.4 and SAP97 in COS-7 cells (25Kim E. Sheng M. Neuropharmacology. 1996; 35: 993-1000Crossref PubMed Scopus (135) Google Scholar). This may reflect a similarity between SAP97 and PSD-95-CS in that SAP97 naturally lacks any N-terminal cysteines equivalent to Cys3 and Cys5. A recent report suggested that mutations of Cys3 and Cys5 prevent PSD-95 from binding to Kv1.4 in COS cells, as measured by coimmunoprecipitation assays (22Topinka J.R. Bredt D.S. Neuron. 1998; 20: 125-134Abstract Full Text Full Text PDF PubMed Scopus (245) Google Scholar). However, we find that double cysteine mutants of PSD-95 coimmunoprecipitate as efficiently with Kv1.4 as does wild type PSD-95 (Fig. 3 B). Thus, the inability of PSD-95 cysteine mutants to form surface clusters with Kv1.4 is not because of their inability to bind to the potassium channel. The ability of double cysteine mutants of PSD-95 to bind Kv1.4 is further supported by the colocalization of these proteins in the brightly staining intracellular structures. Another member of the PSD-95 family, SAP97, has an additional 94 amino acids preceding its N-segment-like region, and it does not contain a pair of cysteine residues at its N terminus. We tested for an interaction between SAP97 and Kv1.4 by the coimmunoprecipitation assay. When coexpressed in COS-7 cells, SAP97 was readily co-precipitated by anti-Kv1.4 antibodies in the presence of Kv1.4 (Fig. 3 C), suggesting that SAP97 and Kv1.4 can form a stable complex. This result reinforces the conclusion that N-terminal cysteines are not required for biochemical association of PSD-95 family proteins with Kv1.4 and perhaps other membrane protein-binding partners. Taken together, the above data indicate that residues Cys3and Cys5 of PSD-95 are required for PSD-95 multimerization and for formation of surface-associated clusters with Kv1.4. However, unlike Topinka and Bredt (22Topinka J.R. Bredt D.S. Neuron. 1998; 20: 125-134Abstract Full Text Full Text PDF PubMed Scopus (245) Google Scholar), we find that these N-terminal cysteines are not important for Kv1.4 channel binding by PSD-95 or its relatives. Because Cys3 and Cys5 are the defined sites of palmitoylation in PSD-95 (22Topinka J.R. Bredt D.S. Neuron. 1998; 20: 125-134Abstract Full Text Full Text PDF PubMed Scopus (245) Google Scholar), our data imply that palmitoylation is not essential for the interaction of PSD-95 with membrane ion channels. To examine whether Cys3and Cys5 play a role in the membrane association of PSD-95, we compared the biochemical fractionation of HA-tagged wild type PSD-95 and Myc-tagged PSD-95-CS mutant coexpressed in COS-7 cells. The relative distribution of wild type and cysteine mutant PSD-95 could then be followed in the same extracts by immunoblotting with HA and Myc antibodies, respectively. In COS cells, the majority of both the wild type PSD-95 and the PSD-95-CS mutant were found in the cytosolic (S100) fraction, although substantial amounts were associated with the crude membrane fraction (Fig. 4). The association of PSD-95 with membranes was investigated further by extracting with high salt or high pH buffers. Sodium carbonate buffer (PH11) was slightly more efficient than 1 m NaCl in extracting PSD-95 from the crude membrane fraction; however, there was no major difference between wild type PSD-95 and PSD-95-CS with regard to sensitivity to these conditions (Fig. 4). Although PSD-95-CS was perhaps more efficiently extracted than wild type PSD-95, the difference was slight in contrast to the qualitative difference observed by Topinka and Bredt (22Topinka J.R. Bredt D.S. Neuron. 1998; 20: 125-134Abstract Full Text Full Text PDF PubMed Scopus (245) Google Scholar). The amount of PSD-95-CS that remained associated with membranes after extraction with high salt or PH11 buffer was similar to that of wild type PSD-95 (Fig. 4). PSD-95 and its Drosophila homolog Dlg have recently been shown to bind via their PDZ domains to the cell adhesion molecule FasII as well as to the Shaker potassium channel (20Thomas U. Kim E. Kuhlendahl S. Koh Y.H. Gundelfinger E.D. Sheng M. Garner C.C. Budnik V. Neuron. 1997; 19: 787-799Abstract Full Text Full Text PDF PubMed Scopus (179) Google Scholar, 21Zito K. Fetter R.D. Goodman C.S. Isacoff E.Y. Neuron. 1997; 19: 1007-1016Abstract Full Text Full Text PDF PubMed Scopus (182) Google Scholar). Indeed, Dlg can form a ternary complex with Shaker and FasII proteins in heterologous cells, although Shaker and FasII have the same binding specificity for PDZ2 of Dlg/PSD-95 (20Thomas U. Kim E. Kuhlendahl S. Koh Y.H. Gundelfinger E.D. Sheng M. Garner C.C. Budnik V. Neuron. 1997; 19: 787-799Abstract Full Text Full Text PDF PubMed Scopus (179) Google Scholar). We asked whether PSD-95 can form a ternary complex with Kv1.4 and FasII and whether this ability depended on multimerization of PSD-95. That is, can a cysteine mutant of PSD-95 that cannot multimerize bind simultaneously to Kv1.4 and FasII? COS-7 cells were triply transfected with Kv1.4, FasII, and either WT PSD-95 or PSD-95-CS mutant, and the cell extracts were immunoprecipitated with Kv1.4 antibodies. The presence of FasII in the Kv1.4 immunoprecipitate was taken to imply existence of the Kv1.4-PSD-95-FasII ternary complex (Fig.5). FasII was precipitated by Kv1.4 antibodies in the presence of wild type PSD-95, but not PSD-95-CS mutant (Fig. 5, lanes 7 and 10), indicating that residues Cys3 and Cys5 of PSD-95 are required for ternary complex formation. The C-terminal FasII(VA) mutant, which cannot interact with the PDZ2 domain of PSD-95, could not be coprecipitated with Kv1.4 and PSD-95 (Fig. 5, lane 11). Moreover, the C-terminal Kv1.4(VA) mutant, which cannot interact with PSD-95, could not coprecipitate either PSD-95 or FasII (Fig. 5,lane 12). These controls confirm that the ternary complex is mediated by C-terminal tail interactions with PDZ domains and that the immunoprecipitating antibodies (anti-Kv1.4) do not cross-react with PSD-95 or FasII. The requirement for N-terminal residues Cys3 and Cys5 in formation of the ternary complex suggests that a single monomer of PSD-95 cannot bind simultaneously to FasII and Kv1.4. Two properties appear to be essential for ion channel clustering by PSD-95 in heterologous cells: 1) the ability of PSD-95 to bind directly to the ion channel by a PDZ-mediated interaction (17Kim E. Cho K.-O. Rothschild A. Sheng M. Neuron. 1996; 17: 103-113Abstract Full Text Full Text PDF PubMed Scopus (476) Google Scholar, 18Hsueh Y.-P. Kim E. Sheng M. Neuron. 1997; 18: 803-814Abstract Full Text Full Text PDF PubMed Scopus (185) Google Scholar, 25Kim E. Sheng M. Neuropharmacology. 1996; 35: 993-1000Crossref PubMed Scopus (135) Google Scholar); and 2) the ability of PSD-95 to self-associate into multimers (18Hsueh Y.-P. Kim E. Sheng M. Neuron. 1997; 18: 803-814Abstract Full Text Full Text PDF PubMed Scopus (185) Google Scholar). Deletion mutants of PSD-95 that lack the Kv1.4-binding domains PDZ1 and PDZ2 lose Kv1.4-clustering activity (18Hsueh Y.-P. Kim E. Sheng M. Neuron. 1997; 18: 803-814Abstract Full Text Full Text PDF PubMed Scopus (185) Google Scholar). Deletion mutants of PSD-95 that lack the N-segment responsible for multimerization are also unable to cluster Kv1.4 (18Hsueh Y.-P. Kim E. Sheng M. Neuron. 1997; 18: 803-814Abstract Full Text Full Text PDF PubMed Scopus (185) Google Scholar). Multimerization of PSD-95 and surface clustering of Kv1.4 are dependent on a pair of cysteines in the N-segment of PSD-95 (Cys3 and Cys5). These cysteines are potential sites for disulfide bonding and/or palmitoylation. Palmitoylation of these cysteine residues has been proposed to mediate membrane association of PSD-95 and to allow its binding to Kv1.4, a membrane ion channel (22Topinka J.R. Bredt D.S. Neuron. 1998; 20: 125-134Abstract Full Text Full Text PDF PubMed Scopus (245) Google Scholar). However, we find that these palmitoylation sites are not required for Kv1.4 binding. Nor did the presence or absence of these cysteines have a major effect on the biochemical association of PSD-95 with membranes, at least when tested in heterologous cells. Perhaps most significantly, natural homologs of PSD-95 that lack these cysteines (such as SAP97 and Dlg) interact efficiently with Shaker-type potassium channels in heterologous cells and in native tissues (19Tejedor F.J. Bokhari A. Rogero O. Gorczyca M. Zhang J. Kim E. Sheng M. Budnik V. J. Neurosci. 1997; 17: 152-159Crossref PubMed Google Scholar, 25Kim E. Sheng M. Neuropharmacology. 1996; 35: 993-1000Crossref PubMed Scopus (135) Google Scholar). Taken together, it seems unlikely that palmitoylation of Cys3 and Cys5 is essential for membrane association of PSD-95 or for PSD-95 interaction with Kv1.4 or other binding partners in the membrane. Although palmitoylation is not critical for membrane interactions of PSD-95 in general, it may facilitate the targeting of PSD-95 specifically to the plasma membrane. We noted that cysteine mutants of PSD-95 tended to be trapped in intracellular membranous organelles (Fig. 3 A). Our results are still consistent with the possibility that disulfide bond formation involving Cys3 and Cys5 is important in the ability of PSD-95 to organize large two-dimensional clusters of ion channels at the cell surface. Given the potential significance of these two cysteines for PSD-95 and chapsyn-110 function, it would be valuable to examine in greater detail the relative roles played by palmitoylation and disulfide bonding of these two residues. Whatever the chemical modifications of Cys3 and Cys5, it is clear that these cysteine residues are required for the N-segment-mediated multimerization of PSD-95. Mutations of these residues abolish PSD-95 self-association (Fig. 2 B). The coimmunoprecipitation experiments in the presence of NEM are consistent with PSD-95 being multimerized prior to cell lysis via disulfide bonds involving Cys3 and Cys5. Alternatively, N-segment multimerization may be mediated by noncovalent bonds, and the Cys3 and Cys5 residues could be important in some other way for the structure and function of the N-segment. It is also possible that palmitoylation is required for the multimerization mediated by the N-segment, although the mechanism of such involvement is unclear. The other major conclusion of this study is that the N-terminal cysteines are required for PSD-95 to coordinate a ternary complex of two distinct membrane proteins, FasII and Kv1.4. Mutation of Cys3 and Cys5 does not affect Kv1.4 binding by PSD-95, but it prevents association of Kv1.4 and FasII together in a PSD-95-based complex (Fig. 5). The simplest explanation of these results is that a single monomer of PSD-95 cannot bind simultaneously to both FasII and Kv1.4. Rather, the ternary complex is constructed on a multimer of PSD-95, each monomer of PSD-95 binding individually to either Kv1.4 or FasII. Such a mechanism neatly explains how PSD-95 can form a ternary complex with two distinct membrane proteins, both of which have the same binding preference for PDZ2 of PSD-95. In vivo, there are multiple proteins that have overlapping specificities for the various PDZ domains of PSD-95. A multimeric scaffold of PSD-95 based on N-segment-mediated association would provide a simple mechanism for several different proteins with the same PDZ specificity to be brought together in a physical complex. We thank Dr. Corey S. Goodman for the FasII antibody and Fu-Chia Yang for technical assistance.
Referência(s)