The ERBB2/HER2 Receptor Differentially Interacts with ERBIN and PICK1 PSD-95/DLG/ZO-1 Domain Proteins

Identification of protein complexes associated with the ERBB2/HER2 receptor may help unravel the mechanisms of its activation and regulation in normal and pathological situations. Interactions between ERBB2/HER2 and Src homology 2 or phosphotyrosine binding domain signaling proteins have been extensively studied. We have identified ERBIN and PICK1 as new binding partners for ERBB2/HER2 that associate with its carboxyl-terminal sequence through a PDZ (PSD-95/DLG/ZO-1) domain. This peptide sequence acts as a dominant retention or targeting basolateral signal for receptors in epithelial cells. ERBIN belongs to the newly described LAP (LRR and PDZ) protein family, whose function is crucial in non vertebrates for epithelial homeostasis. Whereas ERBIN appears to locate ERBB2/HER2 to the basolateral epithelium, PICK1 is thought to be involved in the clustering of receptors. We show here that ERBIN and PICK1 bind to ERBB2/HER2 with different mechanisms, and we propose that these interactions are regulated in cells. Since ERBIN and PICK1 tend to oligomerize, further complexity of protein networks may participate in ERBB2/HER2 functions and specificity. Identification of protein complexes associated with the ERBB2/HER2 receptor may help unravel the mechanisms of its activation and regulation in normal and pathological situations. Interactions between ERBB2/HER2 and Src homology 2 or phosphotyrosine binding domain signaling proteins have been extensively studied. We have identified ERBIN and PICK1 as new binding partners for ERBB2/HER2 that associate with its carboxyl-terminal sequence through a PDZ (PSD-95/DLG/ZO-1) domain. This peptide sequence acts as a dominant retention or targeting basolateral signal for receptors in epithelial cells. ERBIN belongs to the newly described LAP (LRR and PDZ) protein family, whose function is crucial in non vertebrates for epithelial homeostasis. Whereas ERBIN appears to locate ERBB2/HER2 to the basolateral epithelium, PICK1 is thought to be involved in the clustering of receptors. We show here that ERBIN and PICK1 bind to ERBB2/HER2 with different mechanisms, and we propose that these interactions are regulated in cells. Since ERBIN and PICK1 tend to oligomerize, further complexity of protein networks may participate in ERBB2/HER2 functions and specificity. receptor with tyrosine kinase activity PSD-95/DLG/ZO-1 Src homology 2 ERBB2-interacting protein LRR and PDZ human neurotrophin receptor P75 Madin-Darby canine kidney glutathione S-transferase nitric-oxide synthase epidermal growth factor epidermal growth factor receptor 3-aminotriazole Tris-buffered saline binding domain activation domain ephrin receptor The ERBB2/HER-2 gene is overexpressed in about 30% of human breast cancers and is also frequently altered in carcinoma of lungs and kidneys. The ERBB2 protein belongs to the large family of receptors with tyrosine kinase activity (RTK)1 and more precisely to the ERBB subfamily of four receptors which includes the EGF receptor. Upon dimerization and activation, it phosphorylates many substrates including itself. This step induces a network of protein interactions. For example, phosphotyrosine binding and SH2 domains found in cytosolic proteins such as GRB2 and SHC adaptors bind to phosphorylated ERBB2 and lead to the activation of the RAS pathway (1Ullrich A. Schlessinger J. Cell. 1990; 61: 203-212Abstract Full Text PDF PubMed Scopus (4594) Google Scholar, 2Pawson T. Scott J.D. Science. 1997; 278: 2075-2080Crossref PubMed Scopus (1891) Google Scholar). RTKs also interact with cytoplasmic proteins in a non-phosphorylated dependent manner by means of PDZ protein modules found in adaptor proteins (3Simske J.S. Kaech S.M. Harp S.A. Kim S.K. Cell. 1996; 85: 195-204Abstract Full Text Full Text PDF PubMed Scopus (234) Google Scholar, 4Hock B. Bohme B. Karn T. Yamamoto T. Kaibuchi K. Holtrich U. Holland S. Pawson T. Rubsamen-Waigmann H. Strebhardt K. Proc. Natl. Acad. Sci. U. S. A. 1998; 95: 9779-9784Crossref PubMed Scopus (174) Google Scholar, 5Torres R. Firestein B.L. Dong H. Staudinger J. Olson E.N. Huganir R.L. Bredt D.S. Gale N.W. Yancopoulos G.D. Neuron. 1998; 21: 1453-1463Abstract Full Text Full Text PDF PubMed Scopus (397) Google Scholar, 6Borg J.-P. Marchetto S. Le Bivic A. Ollendorff V. Jaulin-Bastard F. Saito H. Fournier E. Adélaı̈de J. Margolis B. Birnbaum D. Nat. Cell Biol. 2000; 7: 407-414Crossref Scopus (255) Google Scholar, 7Garcia R.A. Vasudevan K. Buonanno A. Proc. Natl. Acad. Sci. U. S. A. 2000; 97: 3596-3601Crossref PubMed Scopus (243) Google Scholar, 8Huang Y.Z. Won S. Ali D.W. Wang Q. Tanowitz M. Du Q.S. Pelkey K.A. Yang D.J. Xiong W.C. Salter M.W. Mei L. Neuron. 2000; 26: 443-455Abstract Full Text Full Text PDF PubMed Scopus (318) Google Scholar). PDZ domains are 90–100-amino acid domains that interact with short peptides usually found at the carboxyl terminus of proteins. Proteins bearing a S/TXV motif, such as the glutamate receptors (NMDAR) and K+ channels, interact with class I PDZ domains found in LIN-7 and PSD-95 (9Songyang 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 (1212) Google Scholar, 10Fanning A.S. Anderson J.M Curr. Top. Microbiol. Immunol. 1998; 228: 209-233PubMed Google Scholar). The second class of PDZ domains including the LIN-2/CASK PDZ domain prefers a carboxyl-terminal ΨXΨ sequence, where Ψ is a hydrophobic residue. A third class of PDZ domain including the NOS PDZ domain binds to DXV peptides and PDZ domains (11Brenman 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 (1437) Google Scholar, 12Stricker N.L. Christopherson K.S. Yi B.A. Schatz P.J. Raab R.W. Dawes G. Bassett Jr, D.E. Bredt D.S. Li M. Nat. Biotechnol. 1997; 15: 336-342Crossref PubMed Scopus (218) Google Scholar). Within the RTK superfamily, several members have a canonical PDZ domain binding site in carboxyl-terminal position, i.e. two ERBB receptors (ERBB2/HER2 and ERBB4/HER4), MUSK (a muscle-specific receptor), and several EPH receptor family members involved in developmental processes such as axon guidance. Kim and colleagues have demonstrated the importance of PDZ domain proteins for RTKs function in nonvertebrates. LET-23, the homolog of ERBB receptors in C. elegans, is localized to the basolateral epithelial membrane upon interaction with LIN-7, a PDZ protein that associates with LIN-2 and LIN-10, two other PDZ domain proteins (3Simske J.S. Kaech S.M. Harp S.A. Kim S.K. Cell. 1996; 85: 195-204Abstract Full Text Full Text PDF PubMed Scopus (234) Google Scholar, 13Kaech S.M. Whitfield C.W. Kim S.K. Cell. 1998; 94: 761-771Abstract Full Text Full Text PDF PubMed Scopus (313) Google Scholar). The LIN-7/LIN-2/LIN-10 complex is crucial for proper LET-23 function in insuring a basolateral localization of LET-23 molecules in epithelial cells. Subsequent work has inferred that a subset of mammalian EPH receptors are recruited to specific subcellular compartments in neurons presumably by contacting AF6 and PICK1, two PDZ proteins (4Hock B. Bohme B. Karn T. Yamamoto T. Kaibuchi K. Holtrich U. Holland S. Pawson T. Rubsamen-Waigmann H. Strebhardt K. Proc. Natl. Acad. Sci. U. S. A. 1998; 95: 9779-9784Crossref PubMed Scopus (174) Google Scholar, 5Torres R. Firestein B.L. Dong H. Staudinger J. Olson E.N. Huganir R.L. Bredt D.S. Gale N.W. Yancopoulos G.D. Neuron. 1998; 21: 1453-1463Abstract Full Text Full Text PDF PubMed Scopus (397) Google Scholar). These adaptors are also targets for the tyrosine kinase activity of EPHR and may participate in their signaling. Among ERBB family members, ERBB4/HER4 is a binding partner for PSD-95, a membrane-associated guanylate kinase also associated with K+ channels and NMDAR in neurons (7Garcia R.A. Vasudevan K. Buonanno A. Proc. Natl. Acad. Sci. U. S. A. 2000; 97: 3596-3601Crossref PubMed Scopus (243) Google Scholar, 8Huang Y.Z. Won S. Ali D.W. Wang Q. Tanowitz M. Du Q.S. Pelkey K.A. Yang D.J. Xiong W.C. Salter M.W. Mei L. Neuron. 2000; 26: 443-455Abstract Full Text Full Text PDF PubMed Scopus (318) Google Scholar). It is thought that these PDZ interactions both restrict proteins to specific subcellular compartments and actively participate to signaling events. For example, PSD-95 −/− mice have an impaired long term potentiation response attributed to a defect in NMDAR signaling (14Migaud M. Charlesworth P. Dempster M. Webster L.C. Watabe A.M. Makhinson M. He Y. Ramsay M.F. Morris R.G. Morrison J.H. O'Dell T.J. Grant S.G. Nature. 1998; 396: 433-439Crossref PubMed Scopus (955) Google Scholar). Ligands of RTKs including transforming growth factor α, a ligand for EGFR, and ephrins, ligands for EPHR, are also regulated by interacting with GRIP1 and other PDZ proteins including syntenin (15Bruckner K. Pablo Labrador J. Scheiffele P. Herb A. Seeburg P.H. Klein R. Neuron. 1999; 22: 511-524Abstract Full Text Full Text PDF PubMed Scopus (305) Google Scholar, 16Fernandez-Larrea J. Merlos-Suarez A. Urena J.M. Baselga J. Arribas J. Mol. Cell. 1999; 3: 423-433Abstract Full Text Full Text PDF PubMed Scopus (132) Google Scholar, 17Lin D. Gish G.D. Songyang Z. Pawson T. J. Biol. Chem. 1999; 274: 3726-3733Abstract Full Text Full Text PDF PubMed Scopus (211) Google Scholar). ERBB2/HER2 contains a bona fide PDZ domain binding site in its carboxyl terminus and we previously characterized ERBIN (ERBB2-interacting protein), a novel PDZ domain protein, as a partner for the receptor in epithelia (6Borg J.-P. Marchetto S. Le Bivic A. Ollendorff V. Jaulin-Bastard F. Saito H. Fournier E. Adélaı̈de J. Margolis B. Birnbaum D. Nat. Cell Biol. 2000; 7: 407-414Crossref Scopus (255) Google Scholar). To identify additional ERBB2/HER2 binding partners, we screened a mouse muscle library by the two-hybrid procedure and identified a second partner for ERBB2/HER2 named PICK1. PICK1 is a single PDZ domain protein originally described as an interactor for PKCα (18Staudinger J. Zhou J. Burgess R. Elledge S.J. Olson E.N. J. Cell Biol. 1995; 128: 263-271Crossref PubMed Scopus (261) Google Scholar). PICK1 is thought to act as a scaffold protein able to cluster receptors and components of signaling or targeting machineries at specific sites in cells (19Xia J. Zhang X. Staudinger J. Huganir R.L. Neuron. 1999; 22: 179-187Abstract Full Text Full Text PDF PubMed Scopus (491) Google Scholar). We show by using the two-hybrid system, GST pull-down, and coimmunoprecipitation analysis in living cells that PICK1 specifically interacts with ERBB2/HER2. PICK1 binds to the VXV motif found in ERBB2/HER2 but, in contrast to ERBIN, the PDZ domain alone is not sufficient for the interaction. PICK1 also interacts with the VXV motif found in MUSK and ERBB4/HER4 receptors, two RTKs important for the function of neuromuscular junctions, whereas ERBIN prefers a S/T/VXV motif present in ERBB2/HER2 and proteins unrelated to RTK. Furthermore, we show that ERBIN has a propensity to oligomerize through a PDZ-PDZ domain interaction. Activation of ERBB2 interferes with ERBIN, but not with PICK1, ability to interact with the receptor. We finally show that the carboxyl-terminal sequence of ERBB2/HER2 is a dominant basolateral localization sequence per se since it is able to redirect an apical receptor to the basolateral surface of MDCK cells. This study suggests that diverse PDZ domain proteins are involved in the regulation of ERBB2/HER2 functions. To prepare the baits used in this paper, the 15 last amino acids of ERBB2 wild type or mutant or the 15 last amino acids of various proteins were fused to the LexA-BD subunit using pBTM116 vector, which carries Trp1. For library screening, an oligo(dT) primed mouse cDNA muscle library cloned in pACT2 vector, which carries LEU2 as a selection marker, was screened by using the LexA-ERBB2 as a bait and the yeast strain L40 following the lithium-acetate protocol. Approximately 106 TRP + LEU + transformants were selected on plates with supplemented minimum medium that lacked tryptophan, leucine, and histidine in the primary screening and contained 10 mm 3-aminotriazole (3-AT) and then tested for the β-galactosidase activity by the filter method in the secondary screening. After rescue, the DNA of selected clones was retransformed in L40 yeast containing LexA-ERBB2 or LexA fused to control peptides. Specific clones were positive for growth in histidine-deficient medium and β-galactosidase activity and encoded for PICK1. Cells were washed twice with cold phosphate-buffered saline and lysed in lysis buffer (50 mmHEPES, pH 7.5, 10% glycerol, 150 mm NaCl, 1% Triton X-100, 1.5 mm MgCl2, 1 mm EGTA) supplemented with 1 mm phenylmethylsulfonyl fluoride, 10 μg·ml−1 aprotinin and 10 μg·ml−1 leupeptin. Sodium orthovanadate at 200 μm final was added to lysis buffer when cells were stimulated by EGF. After centrifugation at 16,000 × gfor 20 min, supernatants were saved for further procedures. For immunoprecipitation, lysates were incubated with antibodies overnight at 4 °C. Protein A-agarose was added and immune complexes bound to beads were recovered after 1 h, washed three times with HNTG buffer (50 mm HEPES, pH 7.5, 10% glycerol, 150 mm NaCl, 0.1% Triton X-100), boiled in 1× sample buffer, and separated by SDS-PAGE. Transfer and immunoblotting on nitrocellulose using horseradish peroxidase anti-rabbit or horseradish peroxidase anti-mouse antibody/chemiluminescence method were performed as described (20Borg J.-P. Ooi J. Levy E. Margolis B. Mol. Cell. Biol. 1996; 16: 6229-6241Crossref PubMed Scopus (430) Google Scholar). For overlay assays, the membrane was incubated 2 h at room temperature with soluble GST fusion proteins labeled with protein kinase A and [γ-32P]ATP diluted in TBS, 5% dried milk, 1 mm dithiothreitol (106cpm/ml). After rinsing with TBS, 0.1% Triton X-100 and TBS buffers, bound GST was revealed by autoradiography. GST production and GST binding assays were performed as described previously (20Borg J.-P. Ooi J. Levy E. Margolis B. Mol. Cell. Biol. 1996; 16: 6229-6241Crossref PubMed Scopus (430) Google Scholar). COS-1 and MDCK II cells were grown in Dulbecco's modified Eagle's medium containing 100 units·ml−1 penicillin and 100 μg·ml−1 streptomycin sulfate, supplemented with 10% fetal calf serum. All cell transfections were made using Fugene 6 reagent according to manufacturer's recommendations (Roche Molecular Biochemicals). Human ERBIN cDNA was used to create different constructs allowing expression of the protein in procaryote or eucaryote cells (GST, Myc, LexA-BD-tagged proteins). The RK5-myc vector was used to express proteins fused to the Myc epitope (20Borg J.-P. Ooi J. Levy E. Margolis B. Mol. Cell. Biol. 1996; 16: 6229-6241Crossref PubMed Scopus (430) Google Scholar). The pGEX-Tag vector was used to produce all GST fusion proteins. Site-directed mutagenesis were performed using the Quick-Change kit (Stratagene). All constructs were sequenced by Genome Express, SA (Grenoble, France). A rat cDNA library was used as a template to amplify a fragment encoding the DENSIN-180 PDZ domain (residues 1342–1492) and AF6 PDZ domain (residues 1007–1124). Chimeras between the human P75NTR and human ERBB2/HER2 or GBT-1 were done by PCR using reverse primers coding for the last 15 amino acids of either ERBB2/HER2 or GBT-1 in order to replace the last 15 residues of P75NTR. All constructs were sequenced and subcloned into pIRES (CLONTECH). The resulting plasmids were transfected into MDCK II cells, and stable populations were obtained after G418 selection. Anti-Myc 9E10 (Oncogene Research Products, Cambridge, MA) monoclonal antibody was used for immunoprecipitation and immunoblotting. 4G10 (anti-PY) monoclonal antibody was from Upstate Biotechnology Inc. Goat anti-rabbit and anti-mouse IgG coupled to horseradish peroxidase were purchased from Jackson Laboratory and Dako, respectively. A rabbit anti-ERBIN polyclonal antibody was produced by injecting a soluble GST-ERBIN-(914–1371) fusion protein. Anti-PICK1 goat antibody was from Santa Cruz Inc. Transfected MDCK cells were grown on coverslips for 3 days after confluence and then treated as described (21Le Bivic A. Real F.X. Rodriguez-Boulan E. Proc. Natl. Acad. Sci. U. S. A. 1989; 86: 9313-9317Crossref PubMed Scopus (148) Google Scholar) using ME 20-4, a monoclonal antibody against the extracellular domain of human P75NTR and a rabbit polyclonal antibody against Gp114, an apical glycoprotein of MDCK II cells (22Le Bivic A. Sambuy Y. Mostov K. Rodriguez-Boulan E. J. Cell Biol. 1990; 110: 1533-1539Crossref PubMed Scopus (102) Google Scholar). Transfected MDCK II cells were grown on Transwell™ filters for 5 days after seeding and processed for selective cell surface biotinylation as described before (23Monlauzeur L. Rajasekaran A. Chao M. Rodriguez-Boulan E. Le Bivic A. J. Biol. Chem. 1995; 270: 12219-12225Abstract Full Text Full Text PDF PubMed Scopus (38) Google Scholar). P75NTR or chimeras were immunoprecipitated using the ME 20-4 antibody followed by streptavidin precipitation. Quantitative analysis was done using the Bio-Image IQ software. We have recently isolated ERBIN, a partner for ERBB2/HER2 in epithelia. ERBIN is a novel PDZ domain protein belonging to the LAP (LRR and PDZ) protein family that includes DENSIN-180, SCRIBBLE, and LET-413 (6Borg J.-P. Marchetto S. Le Bivic A. Ollendorff V. Jaulin-Bastard F. Saito H. Fournier E. Adélaı̈de J. Margolis B. Birnbaum D. Nat. Cell Biol. 2000; 7: 407-414Crossref Scopus (255) Google Scholar, 24Apperson M.L. Moon I.S. Kennedy M.B. J. Neurosci. 1996; 16: 6839-6852Crossref PubMed Google Scholar, 25Bilder D. Perrimon N. Nature. 2000; 403: 676-680Crossref PubMed Scopus (551) Google Scholar, 26Bilder D.B. Birnbaum D. Borg J.-P. Bryant P. Huigbretse J. Jansen E. Kennedy M.B. Labouesse M. Legouis R. Mechler B. Perrimon N. Petit N. Sinha P. Nat. Cell Biol. 2000; 7: E114Crossref Scopus (50) Google Scholar, 27Legouis R. Gansmuller A. Sookhareea S. Bosher J.M. Baillie D.L. Labouesse M. Nat. Cell Biol. 2000; 2: 415-422Crossref PubMed Scopus (149) Google Scholar). The ERBIN PDZ domain interacts with the 15 carboxyl-terminal residues in ERBB2, which contain a VXV motif. We used the two-hybrid system in yeast to further delineate critical residues important for this interaction. Position for each residue is given starting position (0) for the carboxyl-terminal valine, followed by (−1) for the proline, (−2) for the valine, and so on (Fig.1 A). Each amino acid in the ERBB2 peptide was changed to alanine except for alanine in −12 position changed to glycine. Peptides fused to the LexA binding domain (LexA-BD) were challenged against the GAL4 activation domain (GAL4-AD) fused to the ERBIN PDZ domain (Fig. 1 A). Interaction was evidenced by the capacity of cotransformed L40 yeast to grow on plates depleted of histidine (−HIS) supplemented with 10 mm 3-AT and to present a positive β-galactosidase activity. As previously shown, mutation of the carboxyl-terminal valine (mutant VA) and deletion of the 6 last residues (data not shown) in ERBB2 abrogated the interaction with ERBIN (6Borg J.-P. Marchetto S. Le Bivic A. Ollendorff V. Jaulin-Bastard F. Saito H. Fournier E. Adélaı̈de J. Margolis B. Birnbaum D. Nat. Cell Biol. 2000; 7: 407-414Crossref Scopus (255) Google Scholar). Interaction was also altered when Val in −2 position, Asp in −3 position, Leu in −4 position, and Tyr in −7 position were mutated (Fig. 1 A). We expressed ERBB2 (wild type and mutants) in COS cells and precipitated receptors with a GST-ERBIN PDZ domain in a pull-down assay (Fig.1 B). Replacement of the COOH-terminal valine, aspartic acid in −3 position, leucine in −4 position, and tyrosine in −7 position to alanine as well as truncation of ERBB2 (Δ6 mutant) reduced the binding to the ERBIN PDZ domain. No change was found when asparagine in −10 position was mutated to alanine. Mutation of valine (−2) did not modify the interaction in the pull-down assay contrasting with the two-hybrid data. This may imply that this residue is not as crucial for ERBIN-ERBB2 interaction in vivo. To obtain further information on the ERBIN PDZ domain binding specificity, we used a Far Western strategy to test the interaction of this domain with peptides found in the carboxyl terminus of receptors and ion channels (Fig.1 C). These sequences, except for EGFR and ERBB3, are known ligands for PDZ domains. Nine-amino acid peptides were fused to the GST protein, and equivalent amount of proteins were resolved by Western blot. The membrane was incubated with a soluble and radiolabeled32P-GST-ERBIN PDZ. After washing, the membrane was exposed for autoradiography to detect bound radiolabeled proteins. No signal was found when we used a radiolabeled GST alone or GST-AF6 PDZ domain as a probe (data not shown). As expected, the ERBIN PDZ domain interacted with ERBB2 but not with other human EGFR family members while [32P]GST-LIN-7 efficiently bound to the worm LET-23 (data not shown). Among the other fusion proteins tested, the carboxyl termini of Kv1.4, NR2C and NR2B interacted strongly with ERBIN (Fig.1 C). Interestingly, these peptides have sequence similarities to the ERBB2 peptide at the crucial positions found for ERBIN interaction, namely a carboxyl-terminal valine, an acidic residue in −3 position (Asp or Glu), and hydrophobic residues in −4 and −7 positions (tyrosine, isoleucine, or leucine). Position −2 can support a serine, threonine, or valine. Taken together, these data demonstrate that the ERBIN PDZ domain binds to a ΨXXΨ[ED][STV]XV peptide, where Ψ represents hydrophobic residue and X any amino acid. We searched in data bases for other mammalian proteins containing such a motif and found the proteins listed in Fig. 1 D, including a transmembrane phosphatase (RPTPζ), PMCA4, a calcium pump, and MKP5, a dual specificity phosphatase specific for p38 and stress-activated protein kinase/c-Jun NH2-terminal kinase. These are potential binding partners for ERBIN. Interestingly, brain-specific angiogenesis inhibitor 1 (BAI1), a potential ERBIN interactor with seven-span transmembrane domains, has indeed been shown to bind PDZ domain proteins (28Shiratsuchi T. Futamura M. Oda K. Nishimori H. Nakamura Y. Tokino T. Biochem. Biophys. Res. Commun. 1998; 247: 597-604Crossref PubMed Scopus (83) Google Scholar). X-ray crystallographic studies of PSD-95, DLG, and CASK/LIN-2 show that PDZ domains comprise two α helixes and six β sheets (29Doyle D.A. Lee A. Lewis J. Kim E. Sheng M. MacKinnon R. Cell. 1996; 85: 1067-1076Abstract Full Text Full Text PDF PubMed Scopus (964) Google Scholar, 30Morais Cabral J.H. Petosa C. Sutcliffe M.J. Raza S. Byron O. Poy F. Marfatia S.M. Chishti A.H. Liddington R.C. Nature. 1996; 382: 649-652Crossref PubMed Scopus (290) Google Scholar, 31Daniels D.L. Cohen A.R. Anderson J.M. Brunger A.T. Nat. Struct. Biol. 1998; 5: 317-325Crossref PubMed Scopus (161) Google Scholar). The COOH-terminal peptide of receptors binds to the groove between the second α helix and the second β sheet with a GLGF motif providing a carboxylate-binding loop. A residue found in the second α helix (His for class I; Val, Leu, or Gln for class II; and Tyr for NOS class III PDZ domain) selects a certain type of ligand depending on the residue found in −2 position in the peptide. Hence, the PSD-95 class I PDZ domain prefers a Ser or Thr in −2 position in the peptide ligand and the CASK/LIN-2 class II PDZ domain selects hydrophobic residues in −2 position while class III PDZ domain in NOS binds to an aspartic residue (Asp) in −2 position. In a previous report, we identified a mutation within the ERBIN PDZ domain (mutant YD) changing the His residue in αB helix to Tyr as found in NOS, which abrogates the interaction with ERBB2 (6Borg J.-P. Marchetto S. Le Bivic A. Ollendorff V. Jaulin-Bastard F. Saito H. Fournier E. Adélaı̈de J. Margolis B. Birnbaum D. Nat. Cell Biol. 2000; 7: 407-414Crossref Scopus (255) Google Scholar). The NOS PDZ domain interacts with DXV motifs. We mutated the Val-2 residue to Asp-2 in ERBB2 but no interaction was found with the ERBIN YD mutant by using the two-hybrid system (Fig.2 A). Thus, mutation YD abrogates interaction with ERBB2 but does not shift the ERBIN PDZ domain specificity from class I to class III. This result can be explained by the fact that the functional NOS PDZ domain requires additional amino acids beyond the conserved PDZ domain consensus domain (12Stricker N.L. Christopherson K.S. Yi B.A. Schatz P.J. Raab R.W. Dawes G. Bassett Jr, D.E. Bredt D.S. Li M. Nat. Biotechnol. 1997; 15: 336-342Crossref PubMed Scopus (218) Google Scholar). This additional region is not present in the ERBIN PDZ domain. We also mutated the histidine to leucine as found in class II PDZ domain (mutant HL). Surprisingly, interaction with ERBB2 was improved as L40 yeast co-expressing LexA-ERBB2/GAL4-ERBIN HL grew on −HIS plates containing 50 mm 3-AT in contrast to LexA-ERBB2/GAL4-ERBIN PDZ domain wild type. Growth was identical on 10 mm 3-AT plates. This result was confirmed in a GST pull down assay where ERBB2 was more efficiently precipitated by ERBIN HL mutant than by ERBIN wild type (Fig. 2 C). The VXV motif in ERBB2 binds to the class I ERBIN PDZ domain, although it conforms to the ΨXΨ motif binding site for class II PDZ domains. Changing His to Leu in ERBIN PDZ domain αB helix improves the binding to VXV-containing peptides. Accordingly, the ERBIN PDZ domain mutant HL now binds to the VXV motif found in EPHB2 (data not shown). We have thus identified important residues within ERBB2 and ERBIN peptide sequences responsible for the interaction between the PDZ domain and its ligand. Promiscuous interaction with various receptors is a frequent theme for PDZ domains. For example, the CASK/LIN-2 PDZ domain binds to neurexins and syndecans (32Hata Y. Butz S. Sudhof T.C. J. Neurosci. 1996; 16: 2488-2494Crossref PubMed Google Scholar, 33Cohen A.R. Wood D.F. Marfatia S.M. Walther Z. Chishti A.H. Anderson J.M. J. Cell Biol. 1998; 142: 129-138Crossref PubMed Scopus (319) Google Scholar, 34Hsueh Y.P. Yang F.C. Kharazia V. Naisbitt S. Cohen A.R. Weinberg R.J. Sheng M. J. Cell Biol. 1998; 142: 139-151Crossref PubMed Scopus (284) Google Scholar) and PSD-95 interacts with glutamate receptors, K+channels, and adenomatous polyposis coli (35Kim E. Niethammer M. Rothschild A. Jan Y.N. Sheng M. Nature. 1995; 378: 85-88Crossref PubMed Scopus (892) Google Scholar, 36Kornau H.-C. Schenker L.T. Kennedy M.B. Seeburg P.H. Science. 1995; 269: 1737-1740Crossref PubMed Scopus (1622) Google Scholar, 37Matsumine A. Ogai A. Senda T. Okumura N. Satoh K. Baeg G.H. Kawahara T. Kobayashi S. Okada M. Toyoshima K. Akiyama T. Science. 1996; 272: 1020-1023Crossref PubMed Scopus (407) Google Scholar). On the other hand, receptors can bind different PDZ proteins, e.g. a class of EPH receptors binds to syntenin and AF6 PDZ domains (4Hock B. Bohme B. Karn T. Yamamoto T. Kaibuchi K. Holtrich U. Holland S. Pawson T. Rubsamen-Waigmann H. Strebhardt K. Proc. Natl. Acad. Sci. U. S. A. 1998; 95: 9779-9784Crossref PubMed Scopus (174) Google Scholar, 5Torres R. Firestein B.L. Dong H. Staudinger J. Olson E.N. Huganir R.L. Bredt D.S. Gale N.W. Yancopoulos G.D. Neuron. 1998; 21: 1453-1463Abstract Full Text Full Text PDF PubMed Scopus (397) Google Scholar). We have shown that the ERBIN PDZ domain interacts in vitro with several peptide ligands (Fig. 1 C). Conversely, it is likely that ERBB2 binds to other PDZ domain-containing proteins. To search for other ERBB2 partners, we screened a murine muscle cDNA library with LexA-ERBB2 as a bait. Four different clones were pulled out, each encoding the full-length PICK1 protein (Fig.3 A). PICK1 is an ubiquitous 50-kDa protein containing an amino-terminal PDZ domain and a coiled-coil region in its carboxyl terminus. PICK1 was originally described as a partner for PKCα, AMPA, EPH receptors, and, more recently, water channel aquaporins through a PDZ domain interaction (5Torres R. Firestein B.L. Dong H. Staudinger J. Olson E.N. Huganir R.L. Bredt D.S. Gale N.W. Yancopoulos G.D. Neuron. 1998; 21: 1453-1463Abstract Full Text Full Text PDF PubMed Scopus (397) Google Scholar,18Staudinger J. Zhou J. Burgess R. Elledge S.J. Olson E.N. J. Cell Biol. 1995; 128: 263-271Crossref PubMed Scopus (261) Google Scholar, 19Xia J. Zhang X. Staudinger J. Huganir R.L. Neuron. 1999; 22: 179-187Abstract Full Text Full Text PDF PubMed Scopus (491) Google Scholar, 38Cowan C.A. Yokoyama N. Bianchi L.M. Henkemeyer M. Fritzsch B. Neuron. 2000; 26: 417-430Abstract Full Text Full Text PDF PubMed Scopus (195) Google Scholar). We found no binding between PICK1 (or ERBIN) and ERBB2 when the carboxyl-terminal valine in ERBB2 was replaced by an alanine (LexA-ERBB2.VA) (Fig. 3 A). To further delineate the region of PICK1 involved in the interaction, we introduced a 100-amino acid carboxyl-terminal deletion in the protein and failed to detect an interaction between deleted PICK1-(1–305) and ERBB2. Additionally, a previously described mutation within the PICK1 PDZ domain changing K27D28 to AA also disrupted the interaction PICK1-ERBB2 (Fig. 3 A) (39Staudinger J. Lu J. Olson E.N. J. Biol. Chem. 1997; 272: 32019-32024Abstract Full Text Full Text PDF PubMed Scopus (210) Google Scholar). As in the case of NOS PDZ domain, these results suggest that the PICK1 PDZ domain is necessary but not sufficient to bind to ERBB2 (12Stricker N.L. Christopherson K.S. Yi B.A. Schatz P.J. Raab R.W. Dawes G. Bassett Jr, D.E. Bredt D.S. Li M. Nat. Biotechnol. 1997; 15: 336-342Crossref PubMed Scopus (218) Google Scholar). To confirm our two-hybrid data, we produced GST fusion proteins encompassing PICK1 (GST-PICK1), deleted PICK1 (GST-PICK1 1–305), mutated PICK1 (GST-PICK1mut), and the PICK1 PDZ domain (GST-PICK1 PDZ). COS