Synaptic multiprotein complexes associated with 5-HT2C receptors: a proteomic approach
2002; Springer Nature; Volume: 21; Issue: 10 Linguagem: Inglês
10.1093/emboj/21.10.2332
ISSN1460-2075
Autores Tópico(s)Neurobiology and Insect Physiology Research
ResumoArticle15 May 2002free access Synaptic multiprotein complexes associated with 5-HT2C receptors: a proteomic approach Carine Bécamel Carine Bécamel CNRS UPR9023, CCIPE 141 rue de la Cardonille, F-34094 Montpellier, Cedex 05, France Search for more papers by this author Gérard Alonso Gérard Alonso CNRS UMR 5101, CCIPE 141 rue de la Cardonille, F-34094 Montpellier, Cedex 05, France Search for more papers by this author Nathalie Galéotti Nathalie Galéotti CNRS UPR9023, CCIPE 141 rue de la Cardonille, F-34094 Montpellier, Cedex 05, France Search for more papers by this author Emmanuelle Demey Emmanuelle Demey CNRS UPR9023, CCIPE 141 rue de la Cardonille, F-34094 Montpellier, Cedex 05, France Search for more papers by this author Patrick Jouin Patrick Jouin CNRS UPR9023, CCIPE 141 rue de la Cardonille, F-34094 Montpellier, Cedex 05, France Search for more papers by this author Christoph Ullmer Christoph Ullmer Biofrontera Pharmaceuticals GmbH, Hemmelratherweg 201, D-51377 Leverkusen, Germany Search for more papers by this author Aline Dumuis Aline Dumuis CNRS UPR9023, CCIPE 141 rue de la Cardonille, F-34094 Montpellier, Cedex 05, France Search for more papers by this author Joël Bockaert Joël Bockaert CNRS UPR9023, CCIPE 141 rue de la Cardonille, F-34094 Montpellier, Cedex 05, France Search for more papers by this author Philippe Marin Corresponding Author Philippe Marin CNRS UPR9023, CCIPE 141 rue de la Cardonille, F-34094 Montpellier, Cedex 05, France Search for more papers by this author Carine Bécamel Carine Bécamel CNRS UPR9023, CCIPE 141 rue de la Cardonille, F-34094 Montpellier, Cedex 05, France Search for more papers by this author Gérard Alonso Gérard Alonso CNRS UMR 5101, CCIPE 141 rue de la Cardonille, F-34094 Montpellier, Cedex 05, France Search for more papers by this author Nathalie Galéotti Nathalie Galéotti CNRS UPR9023, CCIPE 141 rue de la Cardonille, F-34094 Montpellier, Cedex 05, France Search for more papers by this author Emmanuelle Demey Emmanuelle Demey CNRS UPR9023, CCIPE 141 rue de la Cardonille, F-34094 Montpellier, Cedex 05, France Search for more papers by this author Patrick Jouin Patrick Jouin CNRS UPR9023, CCIPE 141 rue de la Cardonille, F-34094 Montpellier, Cedex 05, France Search for more papers by this author Christoph Ullmer Christoph Ullmer Biofrontera Pharmaceuticals GmbH, Hemmelratherweg 201, D-51377 Leverkusen, Germany Search for more papers by this author Aline Dumuis Aline Dumuis CNRS UPR9023, CCIPE 141 rue de la Cardonille, F-34094 Montpellier, Cedex 05, France Search for more papers by this author Joël Bockaert Joël Bockaert CNRS UPR9023, CCIPE 141 rue de la Cardonille, F-34094 Montpellier, Cedex 05, France Search for more papers by this author Philippe Marin Corresponding Author Philippe Marin CNRS UPR9023, CCIPE 141 rue de la Cardonille, F-34094 Montpellier, Cedex 05, France Search for more papers by this author Author Information Carine Bécamel1, Gérard Alonso2, Nathalie Galéotti1, Emmanuelle Demey1, Patrick Jouin1, Christoph Ullmer3, Aline Dumuis1, Joël Bockaert1 and Philippe Marin 1 1CNRS UPR9023, CCIPE 141 rue de la Cardonille, F-34094 Montpellier, Cedex 05, France 2CNRS UMR 5101, CCIPE 141 rue de la Cardonille, F-34094 Montpellier, Cedex 05, France 3Biofrontera Pharmaceuticals GmbH, Hemmelratherweg 201, D-51377 Leverkusen, Germany *Corresponding author. E-mail: [email protected] The EMBO Journal (2002)21:2332-2342https://doi.org/10.1093/emboj/21.10.2332 PDFDownload PDF of article text and main figures. ToolsAdd to favoritesDownload CitationsTrack CitationsPermissions ShareFacebookTwitterLinked InMendeleyWechatReddit Figures & Info Membrane-bound receptors such as tyrosine kinases and ionotropic receptors are associated with large protein networks structured by protein–protein interactions involving multidomain proteins. Although these networks have emerged as a general mechanism of cellular signalling, much less is known about the protein complexes associated with G-protein-coupled receptors (GPCRs). Using a proteomic approach based on peptide affinity chromatography followed by mass spectrometry and immunoblotting, we have identified 15 proteins that interact with the C- terminal tail of the 5-hydroxytryptamine 2C (5-HT2C) receptor, a GPCR. These proteins include several synaptic multidomain proteins containing one or several PDZ domains (PSD95 and the proteins of the tripartite complex Veli3–CASK–Mint1), proteins of the actin/spectrin cytoskeleton and signalling proteins. Coimmunoprecipitation experiments showed that 5-HT2C receptors interact with PSD95 and the Veli3–CASK–Mint1 complex in vivo. Electron microscopy also indicated a synaptic enrichment of Veli3 and 5-HT2C receptors and their colocalization in microvilli of choroidal cells. These results indicate that the 5-HT2C receptor is associated with protein networks that are important for its synaptic localization and its coupling to the signalling machinery. Introduction In the classical model of G-protein-coupled receptor (GPCR)-mediated signalling, the receptor is stabilized under an active conformation able to activate heterotrimeric G-proteins following ligand binding. This activation consists of a catalytic GDP/GTP exchange on the α subunit, triggering its dissociation from βγ subunits. Both α-GTP and βγ stimulate and/or inhibit a large variety of intracellular effectors (Hepler and Gilman, 1992). Over the last decade, several studies have shown that other intracellular proteins are physically associated with GPCRs (Hall et al., 1999). These proteins interact directly, or indirectly, via scaffold proteins, with intracellular domains of GPCRs including their C-terminal tail. These interactions are important for some GPCR functions such as clustering, compartmentalization, optimization of transduction and signalling in the absence of G-protein coupling (Scott and Zuker, 1998; Dev et al., 2001; Milligan and White, 2001). A new concept of 'signaling at zero G' mediated by functional multiprotein complexes associated with GPCRs has emerged from these findings (Brzostowski and Kimmel, 2001). There is accumulating evidence that ionotropic receptors and ionic channels are also clustered in supramolecular complexes, maintained via networks of protein–protein interactions that delimitate specialized sub-membrane microdomains. Some of these complexes have been extensively characterized by proteomic analyses. Protein components of the N-methyl-D-aspartic acid (NMDA) receptor-associated complex, including scaffold proteins, channel subunits, signalling, adhesion and cytoskeletal proteins, have been identified by a combination of mass spectrometry and large-scale immunoblotting (Husi et al., 2000; Husi and Grant, 2001). The large number of protein interactions revealed by this study indicated a more complicated degree of organization of this complex than that predicted by other approaches. Similarly, a complex associated with the P2X7 ATP receptor has been isolated recently by immunoprecipitation, allowing the identification by MALDI-TOF mass spectrometry of a set of proteins interacting with the receptor (Kim et al., 2001). While several proteins are associated with both NMDA receptor and P2X7 receptor complexes, many of them are specific to one of these complexes. This suggests a specificity in the protein composition and organization of protein networks associated with channel-linked receptors. In contrast, identification of proteins associated with a particular GPCR has, so far, only been achieved in a piecemeal fashion, often based on yeast two-hybrid screens (Hall et al., 1998; Ullmer et al., 1998; Zitzer et al., 1999a, b). 5-hydroxytryptamine type 2C (5-HT2C) receptors are broadly expressed in the central nervous system (CNS) (Abramowski et al., 1995; Clemett et al., 2000) and modulate a large variety of behavioural and physiological processes, such as nociception, motor behaviour, thermoregulation and modulation of appetite (Lucki et al., 1989; Fone et al., 1998). Activation of 5-HT2C receptors exerts a phasic and tonic inhibition of the mesocorticolimbic dopamine function. This suggests that 5-HT2C receptor antagonists may be useful for the treatment of negative schizophrenia symptoms (Di Matteo et al., 2001). Although initial studies of 5-HT2C receptor signalling showed that they activate phospholipase Cβ (Chen et al., 1994), the molecular events initiated by these receptors remain largely unknown. 5-HT2C receptors contain a C-terminal sequence (SSV) corresponding to one of the class 1 PDZ (PSD95/disc large/ZO-1) recognition motifs (X-S/T-X-I/L/V). MUPP1, a multivalent PDZ domain adaptor protein, has recently been identified as a binding partner of 5-HT2C receptors by a two-hybrid screen (Ullmer et al., 1998; Bécamel et al., 2001). This suggests that 5-HT2C receptors participate in a protein network organized around a PDZ domain-based scaffold. The present study was performed to identify additional proteins that interact with the C-terminal tail of 5-HT2C receptors. For that purpose, we used a proteomic approach based on peptide affinity chromatography followed by mass spectrometry and/or immunoblotting. We report that 5-HT2C receptors are associated with at least 15 proteins, including synapse-enriched multidomain proteins containing one or several PDZ domains, such as the Veli3–CASK–Mint1 complex and PSD95. We also provide evidence that these interactions take place in vivo. Results Isolation of proteins interacting with 5-HT2C receptors via a PDZ domain-based scaffold Proteins physically associated with the 5-HT2C receptor were purified by affinity chromatography using the entire C-terminal tail (90 amino acids) of the receptor fused to glutathione S-transferase (GST–5HT90SSV) as a bait. As 5-HT2C receptors display a wide but discrete distribution in the CNS (Abramowski et al., 1995), whole-brain extracts were prepared and loaded on to glutathione–Sepharose 4B beads coated with the GST–5HT90SSV fusion protein. The bound proteins were eluted and separated by two-dimensional (2D) electrophoresis. A typical 2D gel is illustrated in Figure 1A. Numerous protein spots, including the GST fusion proteins and some bacterial proteins, were apparent on silver-stained 2D gels, due to the high sensitivity of this staining method. To detect proteins that interact specifically with the PDZ ligand of 5-HT2C receptors, we performed a control experiment using GST fused to the C-terminal tail of the receptor, which was mutated in this motif (GST–5HT90SSA). The mutated residue is critical for the interaction with target PDZ domains (Bécamel et al., 2001). The analysis of protein patterns in the 2D gels obtained with the GST–5HT90SSV and GST–5HT90SSA baits indicated a marked difference in normalized volume of five spots or groups of spots (spots 1–4 and 12, indicated by arrows, Figure 1A). The differential binding of these proteins to the GST–5HT90SSV and GST–5HT90SSA baits is illustrated in detailed gels (Figure 1B). These proteins were unambiguously identified by MALDI-TOF mass spectrometry (see Table I). Figure 2 shows the peptide mass maps of three of them. Eight of the measured peptide masses obtained for spot 1 matched the theoretical tryptic peptide mass calculated for Veli3 (one of the vertebrate homologues of LIN-7), a 22 kDa protein containing one PDZ domain and enriched in the brain (Table II; Jo et al., 1999). The peptide masses exclude a match with the two other members of the Veli protein family, Veli1 and Veli2. As Veli2 and Veli3 display similar molecular weights and isoelectric points, we also performed 2D immunoblotting experiments using Veli2-and Veli3-specific antibodies. These experiments confirmed that the C-terminal tail of 5-HT2C receptors recruited Veli3 but not Veli2 (Figure 3A). Taken together, these results illustrate the power of MALDI-TOF technology to identify a protein isoform unambiguously. Spots 2 and 3 were identified as a single protein, the post-synaptic density-enriched protein PSD95, and spot 4 as a Dlgh3 protein, a brain-enriched scaffold protein of the p55 family (Figure 2, Tables I and II; Lin et al., 1998). Both proteins belong to a superfamily of modular proteins dubbed membrane-associated guanylate kinases (MAGUKs). These proteins contain an SH3 domain and a C-terminal guanylate kinase domain, in addition to one or several PDZ domains. The specific association of PSD95 with the wild-type bait, but not with the mutated one, was confirmed by immunoblotting (Figure 3A). Two major forms of the protein with different molecular weights and isoelectric points were detected by 2D immunoblotting. These immunoreactive signals matched spots 2 and 3, identified as PSD95 from their peptide mass fingerprint, on silver-stained gels (Figures 1A and 3A). The identification of dynamin 1 (spot 12) as a protein interacting with the 5-HT2C receptor through a PDZ domain-mediated scaffold was more unexpected, as this protein is devoid of PDZ domain. Nevertheless, the PDZ domain-dependent interaction of dynamin 1 with the 5-HT2C receptor was confirmed by immunoblotting (Figure 3A). Figure 1.Two-dimensional analysis of the 5-HT2C receptor protein complex. (A) Proteins that bind to the C-terminal tail of the 5-HT2C receptor were purified by affinity chromatography using the 90-amino-acid C-terminal sequence of the receptor fused to GST (GST–5HT90SSV), separated by 2D electrophoresis and silver stained. A typical 2D gel is illustrated. Proteins that interact specifically with the PDZ ligand of the receptor (arrows) were detected by comparing protein patterns obtained with GST–5HT90SSV and a mutant bait in which the last residue was replaced by alanine (GST–5HT90SSA). Arrowheads indicate proteins that interact equally with both wild-type and mutated baits but that were less represented in gels from experiments using GST alone. (B and C) Areas of interest of gels obtained in experiments performed with GST–5HT90SSV, GST–5HT90SSA and GST alone. The quantification of proteins (spot volume relative to the volume of all spots) was performed with Image Master. Data (means ± SEM of values from four gels) were normalized for each spot to the value measured in experiments using the wild-type bait (ND, not detectable). *P < 0.05 versus GST–5HT90SSV (ANOVA followed by Student–Newman–Keul's test). Download figure Download PowerPoint Figure 2.MALDI-TOF peptide mass maps obtained from spots 1, 2 and 4. Ion signals with measured masses (Table II) that matched calculated masses of protonated tryptic peptides of mouse Veli3, PSD95 and Dlgh3 are indicated by arrows. T indicates the ion signals corresponding to the autolysis products of trypsin that were used for internal calibration of spectra (mol. wts 842.51, 1045.56 and 2211.10, respectively). Download figure Download PowerPoint Figure 3.Detection of proteins interacting with the 5-HT2C receptor by western blotting. (A) CHAPS-solubilized proteins from whole brain, retained by the GST–5HT90SSV and GST–5HT90SSA baits, were resolved on 2D gels and transferred electrophoretically on to nitrocellulose membranes. (B) Proteins were solubilized with 1% SDS instead of CHAPS and incubated with the GST–5HT90SSV and GST–5HT90SSA baits. (C) CHAPS-solubilized proteins from whole brain were passed over affinity columns containing the C-terminal tails of 5-HT2C and 5-HT2A receptors fused to GST. (D) CHAPS-solubilized proteins from choroid plexus were incubated with the GST–5HT90SSV and GST–5HT90SSA baits. Immunoblotting was performed with antibodies raised against the indicated proteins. For each protein, the immunoreactive signals were found at the expected isoelectric points (A) and molecular weights. The data illustrated are representative of three experiments. Download figure Download PowerPoint Table 1. Proteomic analysis of proteins interacting with 5-HT2C receptors Spota Protein identified Accession numberb Protein paramaters Observed MALDI-TOF MS PDZ domain Mol. wt (kDa) pI Mol. wt (kDa) pI Matching peptides Protein coverage (%) 1 Veli3 O88952 21.8 8.5 24 8.5 8 50.0 + 2 Post-synaptic density protein 95 (PSD95) P31016 80.5 5.6 75 5.8 27 45.4 + 3 Post-synaptic density protein 95 (PSD95) P31016 80.5 5.6 86 5.4 24 38.1 + 4 Dlgh3 protein (MPP3) O88910 64.5 5.8 66 5.8 20 43.0 + 5 Calmodulin P02593 16.7 4.1 17 3.5 7 59.5 − 6 F-actin capping protein β subunit (CAPZ β) P47757 31.3 5.5 30 5.4 13 47.0 − 7 F-actin capping protein α-2 subunit (CAPZ α-2) P47754 33.0 5.6 35 5.4 7 41.3 − 8 PKCθ-interacting protein PICOT Q9JLZ2 37.8 5.4 40 5.3 9 38.6 − 9 Actin, cytoplasmic 1 (β-actin) P02570 41.6 5.3 45 5.2 15 56.0 − 10 2810409H07Rik protein Q9CWE2 44.8 7.6 50 7.7 7 21.5 − 11 2810409H07Rik protein Q9CWE2 44.8 7.6 50 8.1 8 27.0 − 12 Dynamin 1 P39053 97.4 8.2 97 7.0 31 36.6 − 13 Spectrin α II chain (α-fodrin) O88663 285 5.2 250 5.2 36 22.2 − a The numbers correspond to those illustrated in Figure 1. b SWISS-PROT and TrEMBL accession numbers are listed. Table 2. Veli3, PSD95 and Dlgh3 peptides identified from the MALDI-TOF peptide mass maps shown in Figure 2 Measured mass Matching mass Δmass (p.p.m.) Missed cleavage Modification Position Peptide Veli3 911.54 911.53 −10.02 0 122–130 IIPGGIADR 1206.56 1206.61 42.53 0 112–121 EQNSPIYISR 1322.60 1322.65 39.40 0 Cys-CAM 41–51 VLQSEFCNAVR 1352.59 1352.65 45.48 0 99–111 TEEGLGFNIMGGK 1608.75 1608.79 23.43 0 77–92 ATVAAFAASEGHSHPR 2034.01 2033.99 −10.53 0 137–155 GDQLLSVNGVSVEGEHHEK 2190.10 2190.09 −4.66 1 136–155 RGDQLLSVNGVSVEGEHHEK 2238.04 2238.06 7.18 0 52–70 EVYEHVYETVDISSSPEVR PSD95 1010.49 1010.49 −17.55 0 579–586 DYHFVSSR 1037.52 1037.50 −19.51 1 571–578 REYEIDGR 1114.53 1114.58 40.56 0 234–242 NTYDVVYLK 1123.47 1123.48 6.74 0 506–516 DWGSSSGSQGR 1125.58 1125.60 18.38 0 99–110 IIPGGAAAQDGR 1156.57 1156.58 16.68 0 300–309 DLLGEEDIPR 1252.63 1252.66 27.30 0 369–380 NASHEQAAIALK 1314.72 1314.74 13.75 1 654–664 SLENVLEINKR 1322.59 1322.60 15.26 1 504–516 AKDWGSSSGSQGR 1354.63 1354.69 44.30 0 Cys-CAM 625–636 HCILDVSANAVR 1368.68 1368.67 −5.86 1 2×CysCAM 1–11 MDCLCIVTTKK 1386.69 1386.69 6.11 1 Cys-CAM 559–570 FGSCVPHTTRPK 1513.83 1513.83 1.70 1 355–368 KGDQILSVNGVDLR 1538.79 1538.78 −7.16 1 300–312 DLLGEEDIPREPR 1618.86 1618.84 −10.58 0 113–126 VNDSILFVNEVDVR 1650.84 1650.81 −16.49 0 409–424 EQLMNSSLGSGTASLR 1666.86 1666.81 −29.96 0 MSO 409–424 EQLMNSSLGSGTASLR 1714.95 1714.92 −17.1 0 707–721 VIEDLSGPYIWVPAR 1746.92 1746.82 −62.15 0 476–491 VHSDSETDDIGFIPSK 1853.13 1853.10 −17.72 1 638–653 LQAAHLHPIAIFIRPR 1902.96 1902.92 −20.09 1 475–491 RVHSDSETDDIGFIPSK 2009.30 2009.20 −50.55 2 637–653 RLQAAHLHPIAIFIRPR 2168.10 2168.10 −1.62 1 381–399 NAGQTVTIIAQYKPEEYSR 2256.15 2256.10 −20.31 0 598–617 FIEAGQYNSHLYGTSVQSVR 2294.12 2294.21 36.18 0 212–233 ILAVNSVGLEDVMHEDAVAALK 2501.25 2501.28 9.43 0 169–193 GLGFSIAGGVGNQHIPGDNSIYVTK 2787.18 2787.36 64.09 0 71–98 GNSGLGFSIAGGTDNPHIGDDPSIFITK Dlgh3 1018.50 1018.51 8.91 0 233–240 ALFHYDPR 1098.60 1098.63 25.85 1 143–152 NKEPLGATIR 1139.55 1139.58 24.65 0 154–164 DEHSGAVVVAR 1181.58 1181.63 39.7 0 174–184 SGLVHVGDELR 1252.54 1252.45 −76.16 0 323–332 ETCDCDEYFK 1267.62 1267.63 5.5 1 377–386 YQHQPGERPR 1290.65 1290.67 10.13 0 301–312 TTGTLPSPQNFK 1295.67 1295.68 6.93 1 153–154 RDEHSGAVVVAR 1323.65 1323.62 −27.08 0 440–450 QAFEADVHHNR 1399.75 1399.72 −17.6 0 Cys-CAM 478–489 VCLVDVEPEALR 1419.84 1419.83 −8.33 0 90–101 ELLQLLSTPHLR 1446.79 1446.77 −15.78 1 300–312 RTTGTLPSPQNFK 1486.79 1486.75 −26.19 0 Cys-CAM 244–256 AIPCQEAGLPFQR 1699.07 1699.03 −27.16 0 387–402 LVVLIGSLGAHLHELK 1962.08 1962.02 −31.82 0 359–376 VPTGAESQVLLTYEEVAR 2029.07 2029.00 −32.2 0 258–274 QVLEVVSQDDPTWWQAK 2084.16 2084.13 −14.31 1 195–213 RPDEISQILAQSQGSITLK 2162.24 2162.20 −17.26 2 193–510 TPEFKPYVIFVKPAIQER 2185.10 2185.10 −0.19 1 257–274 RQVLEVVSQDDPTWWQAK 2218.14 2218.17 14.15 1 405–424 VVAEDPQQFAVAVPHTTRPR MSO, oxidized methionine; Cys-CAM, carbamidomethyl cysteine. An immunoblotting screen was then performed to identify additional partners of 5-HT2C receptors. Veli proteins form a stable tripartite complex with two other neuron-enriched modular proteins: CASK, a MAGUK that contains an N-terminal calmodulin kinase II domain in addition to one PDZ domain; and Mint1, which contains two PDZ domains (Borg et al., 1998; Butz et al., 1998). Veli proteins, CASK and Mint1 are the mammalian orthologues of the Caenorhabditis elegans proteins LIN-7, LIN-2 and LIN-10. In C.elegans, this complex is required for the normal basolateral localization of the tyrosine kinase receptor LET23 in vulval cells. The whole complex is necessary for the proliferation and differentiation of vulval cells (Simske et al., 1996; Kaech et al., 1998). CASK binds directly to both Veli proteins and Mint1 through PDZ-independent interactions, leaving their PDZ domains free to interact with other proteins, such as cell adhesion molecules, receptors and signalling proteins (Butz et al., 1998). CASK and Mint1 were not detectable in silver-stained 2D gels obtained from the pull-down experiment using GST–5HT90SSV and GST–5HT90SSA fusion proteins. However, 2D immunoblotting indicated that both CASK and Mint1 were recruited by the wild-type bait but not by the mutated one (Figure 3A). This suggests that 5-HT2C receptors interact with the entire tripartite complex through a PDZ-based mechanism. Pull-down experiments performed with the C-terminal tail of the 5-HT2A receptor, a group 2 5-HT receptor expressed in the CNS and containing a PDZ ligand domain, indicated that the Veli3–CASK–Mint1 complex does not associate with this 5-HT2 receptor subclass (Figure 3C). This result further supports the idea that the proteins recruited by the C-terminal tail of 5-HT2C receptors are specific binding partners of these receptors rather than proteins interacting with any PDZ domain recognition motif. Mint1 can bind to Munc18, a presynaptic protein that is essential for exocytosis of synaptic vesicles (Okamoto and Südhof, 1997; Borg et al., 1999; Biederer and Südhof, 2000; Verhage et al., 2000). As shown by immunoblotting, Munc18 was also recruited by 5-HT2C receptors via a PDZ domain-dependent scaffold (Figure 3A). PSD95 can couple functionally specific signalling proteins to post-synaptic receptors. For example, PSD95 binds to neuronal NO-synthase (nNOS) via a PDZ–β-finger interaction and the C-terminal PDZ domain-binding motif of NMDA receptor subunits (NR2A and NR2B), allowing a physical coupling of nNOS to the NMDA receptor (Brenman et al., 1996; Christopherson et al., 1999; Tochio et al., 2000). Moreover, activation of 5-HT2 receptors, including the 5-HT2C subtype, induces cGMP production through activation of nNOS in various cell populations (Kaufman et al., 1995). We found that nNOS was specifically retained by the GST–5HT90SSV bait, suggesting that 5-HT2C receptors are physically associated with nNOS through a PSD95-dependent scaffold (Figure 3A). PSD95 is a major scaffolding protein involved in the assembly of NMDA receptor-associated protein complex (Husi et al., 2000). Thus, the 5-HT2C receptor, by interacting with PSD95, may also belong to this complex. To determine whether the 5-HT2C receptor associates with NMDA receptors, brain proteins were solubilized with 1% SDS instead of CHAPS (Ehlers et al., 1998) and incubated with the GST–5HT90SSV and GST–5HT90SSA baits. In this experimental condition, Veli3 was recruited by the PDZ ligand of 5-HT2C receptors (Figure 3B). However, the C-terminal tail of 5-HT2C receptors did not retain any detectable amount of the NMDA receptor subunit NR1, which is essential for the formation of functional NMDA receptors (Figure 3B). Finally, in agreement with our previous findings (Bécamel et al., 2001), the 5-HT2C receptor associated with MUPP1 via its C-terminal PDZ domain recognition motif (Figure 3A). Isolation of proteins interacting with the 5-HT2C receptor independently of PDZ domain-based scaffolds To identify proteins that bind to residues not located in the PDZ ligand of the 5-HT2C receptor, the 2D gel protein patterns resulting from pull-down experiments using GST–5HT90SSV and GST–5HT90SSA fusion proteins were compared with that obtained with GST alone (control). Using this approach, we detected eight proteins (indicated by arrowheads, Figure 1A) that were recruited to a similar extent by GST–5HT90SSV and GST–5HT90SSA baits but were less represented in gels from experiments performed with GST alone (see, for example, the quantification of spots 5 and 8, Figure 1C). These comprised calmodulin, PICOT, a protein that was recently identified as a binding partner of protein kinase Cθ (PKCθ) (Witte et al., 2000) and cytoskeletal proteins [β-actin, spectrin α II chain (α-fodrin) and both α and β chains of CAPZ (Table I)]. CAPZ is a capping protein that binds as a dimer to the barbed end of actin filaments and inhibits the growth of actin microfilaments (Xu et al., 1999). These results suggest that 5-HT2C receptors are physically associated with actin/spectrin microfilaments. Association of actin filaments with tetrameric spectrin is promoted by the 4.1 proteins, a family of peripheral membrane proteins identified initially in erythrocytes (Conboy et al., 1986). 4.1 proteins are critical for the attachment of the actin cytoskeleton to the plasma membrane through interaction with integral membrane proteins such as glycophorin C (Takakuwa, 2000). 4.1 proteins can also interact with other proteins through specific domains, including calmodulin and MAGUKs of the p55 subfamily such as CASK (Hoover and Bryant, 2000). We thus examined by western blotting whether 5-HT2C receptors interact with the neuronal form of 4.1 protein (4.1N). Identical amounts of 4.1N were recruited by the wild type and the mutated C-terminal tail of 5-HT2C receptors (Figure 3A). This result is consistent with the aforementioned findings that the 5-HT2C receptor binds to cytoskeletal proteins through a PDZ domain-independent scaffold and suggests that 4.1N protein may be a critical component that anchors 5-HT2C receptors to the actin cytoskeleton. The two last spots recruited by both fusion proteins but not by GST were identified as a single protein with an unknown function (2810409H07Rik protein; Kawai et al., 2001). Coimmunoprecipitation of PSD95 and the Veli3–CASK–Mint1 complex with 5-HT2C receptors Taken together, our results suggest that 5-HT2C receptors are part of multiprotein complexes including synapse-enriched scaffold proteins, containing one or several PDZ domains, such as the Veli3–CASK–Mint1 complex and PSD95. In order to examine the association of these proteins with 5-HT2C receptors in brain extracts, we performed coimmunoprecipitation studies. CHAPS- soluble lysates of mice brains were immunoprecipitated with the anti-5-HT2C receptor 522 antibody. We found that PSD95, Veli3, CASK and, to a lesser extent, Mint1 coimmunoprecipitated with the 5-HT2C receptor (Figure 4). In contrast, Veli2 was not coimmunoprecipitated by the 5-HT2C receptor antibody, indicating that 5-HT2C receptors interact preferentially with Veli3 in vivo. This result is consistent with the specific recruitment of this Veli isoform in pull-down experiments and, thus, supports the specificity of this in vitro binding assay. In a similar manner, we immunoprecipitated Veli3 and PSD95 from brain extract with Veli3 and PSD95 antibodies. We found that 5-HT2C receptors coimmunoprecipitated with both proteins (Figure 4, bottom). CASK and Mint1 also coimmunoprecipitated with Veli3 (Figure 4). This is consistent with previous findings that demonstrate that CASK, Mint1 and Veli proteins form a stable tripartite complex (Borg et al., 1998; Butz et al., 1998). Taken together, these results indicate that 5-HT2C receptors are associated with both the Veli3–CASK–Mint1 complex and PSD95 in vivo. In agreement with a previous report indicating that Veli proteins are clustered with PSD95 and NMDA-type glutamate receptors (Jo et al., 1999), we also found that PSD95 coimmunoprecipitated with Veli3 (Figure 4). Figure 4.Association of PSD95 and the Veli3–CASK–Mint1 complex with 5-HT2C receptors of mice brain. Solubilized membranes of mice brain were immunoprecipitated with either the anti-5-HT2C, the anti-Veli3 or the anti-PSD95 antibody. Immunoprecipitated proteins were analysed by western blotting using antibodies to Veli3, CASK, Mint1, PSD95, Veli2 and the 5-HT2C receptor. Input (CHAPS-soluble extract) represents 10% of the total protein used for the immunoprecipitation. Download figure Download PowerPoint PDZ domain-based interactions of the 5-HT2C receptor with Veli3 and PSD95 within intact cells Next, we examined whether the entire 5-HT2C receptor interacts with PSD95 and Veli3 within intact cells. For that purpose, we performed immunostaining of COS-7 cells transiently expressing the c-Myc epitope-tagged version of the human 5-HT2C receptor or its mutated form 5-HT2C-SSA, in the presence and absence of either Veli3 protein or GFP-tagged PSD95. Staining COS-7 cells transfected with either the wild type or mutated 5-HT2C receptor with an anti-c-Myc antibody revealed a random distribution of 5-HT2C receptors and 5-HT2C-SSA receptors on membrane-type structures, including intracellular membranes, consistent with our previous findings (Figure 5A; Bécamel et al., 2001). Staining of COS-7 cells expressing Veli3 with a Veli3 antibody indicated a homogeneous distribution of the protein. Similarly, GFP-tagged PSD95 was homogeneous
Referência(s)