PDZ Domain Protein GIPC Interacts with the Cytoplasmic Tail of Melanosomal Membrane Protein gp75 (Tyrosinase-related Protein-1)*

Tyrosinase and tyrosinase-related proteins (TRPs) are a family of melanosomal membrane proteins involved in mammalian pigmentation. Whereas the melanogenic functions of TRPs are localized in their amino-terminal domains that reside within the lumen of melanosomes, the sorting and targeting of these proteins to melanosomes is mediated by signals in their cytoplasmic domains. To identify proteins that interact with the cytoplasmic tail of gp75 (TRP-1), the most abundant melanosomal membrane protein, we performed yeast two-hybrid screening of a melanocyte cDNA library. Here, we show that the cytoplasmic domain of gp75 interacts with a PDZ domain-containing protein. The gp75-interacting protein is identical to GIPC, an RGS (regulator of G protein signaling)/GAIP-interacting protein, and to SEMCAP-1, a transmembrane semaphorin-binding protein. Carboxyl-terminal amino acid residues, Ser-Val-Val, of gp75 are necessary and sufficient for interaction of gp75 with the single PDZ domain in GIPC. Although endogenous and transfected GIPCs bind efficiently to transiently expressed gp75, only a small amount of GIPC is found associated with gp75 at steady state. Using a strategy to selectively synchronize the biosynthesis of endogenous gp75, we demonstrate that only newly synthesized gp75 associates with GIPC, primarily in the juxtanuclear Golgi region. Our data suggest that GIPC/SEMCAP-1 plays a role in biosynthetic sorting of proteins, specifically gp75, to melanosomes.

In mammalian melanocytes, melanin pigment is synthesized in specialized organelles known as melanosomes. Tyrosinase is the critical enzyme required for melanin synthesis, and tyrosinase-related protein-1 (TRP-1) 1 and TRP-2 (dopachrome tautomerase) influence the nature of the pigment produced (1,2). Tyrosinase and TRP-2 catalyze specific steps in melanin synthesis, and their enzymatic functions are conserved between mouse and human (3)(4)(5). In contrast, the function of TRP-1 (also known as gp75 or brown locus protein) in pigmentation is less clear. Although the molecular genetic basis for the brown coat-color phenotype in mouse and the enzyme activity (5,6dihydroxy indole 2-carboxylic acid oxidase) associated with mouse gp75 has been characterized, the exact role gp75 plays in the biology of human pigmentation is not clear (6 -12). In human melanocytes and pigmented melanoma cells, gp75 is one of the most abundant membrane proteins (13). The 537amino acid-long gp75 polypeptide consists of a large aminoterminal lumenal domain (which shares sequence identity and significant structural homology with tyrosinase), a membraneanchoring domain, and a short cytoplasmic tail. An endosomal/ melanosomal-targeting signal, Glu-Xaa 2-3 -Pro-Leu-Leu, first identified in the cytoplasmic tail of gp75, is also conserved among TRPs and several human and mouse melanosomal membrane proteins (14). Interaction of adaptor complex AP-3 with this dileucine motif seems to mediate intracellular sorting of tyrosinase and TRPs to melanosomes (15,16).
Despite many structural similarities and functional relatedness, TRPs show distinct biochemical and cell-biological characteristics (13,17,18). There is evidence that amino acid sequences in the cytoplasmic tails of tyrosinase and TRPs contribute to their biological properties. For example, the phosphorylation of serine residues (Ser 505 and Ser 509 ) in the cytoplasmic tail of tyrosinase by protein kinase C-␤ seems to augment the enzymatic activity of tyrosinase (19). Amino acid sequences in the cytoplasmic tail of gp75 have been shown to determine the intracellular stability and export from the endoplasmic reticulum, suggesting the presence of several signals for intracellular transport in the cytoplasmic tail (20). Thus, through their interactions with cytosolic proteins, cytoplasmically exposed amino acid sequences seem to influence the biological functions of mammalian pigmentation-related proteins.
In this study, by using the cytoplasmic tail of gp75 as bait for yeast two-hybrid screening we identified a PDZ domain-containing protein, GIPC/SEMCAP-1 (referred to as GIPC hereafter), as a gp75-interacting protein. GIPC, which was identified earlier based on its interaction with GAIP, a regulator of G protein signaling (21), and as neural semaphorin-interacting protein SEMCAP-1 (22), binds specifically to the carboxyl terminus of newly synthesized gp75 but not tyrosinase. Binding of GIPC with the melanosomal gp75 during biosynthesis suggests a possible role for this interaction in intracellular sorting and targeting of melanosomal membrane proteins.

EXPERIMENTAL PROCEDURES
Yeast Two-hybrid Analysis-For bait plasmid construction a DNA fragment encoding the cytoplasmic tail of gp75 was amplified by polymerase chain reaction using the plasmid pSVK3hgp75 containing the full-length gp75 cDNA (14) as template and primers with 5Ј-EcoRI and 3Ј-XhoI restriction sites and cloned into EcoRI-XhoI-digested pHybLex/ Zeo plasmid (Invitrogen, Carlsbad, CA) to produce an in-frame fusion cDNA with the LexA DNA-binding domain protein. The transformation of Saccharomyces cerevisiae strain EGY48 and selection of Zeocin-resistant colonies were performed according to manufacturer instructions. Zeocin-resistant colonies were transformed with the reporter plasmid pSH18 -34 (Invitrogen). The transformants were selected on Zeo ϩ /Ura Ϫ plates and then plated on Zeo ϩ /ura Ϫ /leu Ϫ plates containing 2% galactose, 1% raffinose, and 80 mg/ml X-gal to test background activation of the reporter genes by the bait fusion protein alone.
The human melanocyte cDNA library cloned in pJG vector (Invitrogen) as B42 activation domain fusion constructs was a gift from Dr. Tony Zervos (Cutaneous Biology Research Center, Cambridge, MA). A freshly isolated colony of EGY48 transformant selected for bait and ␤-galactosidase (␤-gal) plasmids was transformed with 150 g of melanocyte cDNA library plasmid. Aliquots (200 g) of the transformation mixture containing a total of 13.6 ϫ 10 6 transformants were plated on 40 Zeo ϩ /Ura Ϫ /Trp Ϫ plates containing 2% dextrose and incubated at 30°C for 2 days. An amplified library was titered and stored at Ϫ80°C. The transformants were screened for the interacting proteins by plating 1.15 ϫ 10 8 colony-forming units on 40 gal ϩ /raf ϩ /ura Ϫ /trp Ϫ /leu Ϫ /Zeo ϩ plates (150 mm) containing X-gal. Blue colonies that appeared on day 3 of incubation were marked and allowed to grow for 2 more days. All 27 positive clones obtained were analyzed further after restreaking on selection plates. pJG plasmids containing library inserts were isolated from two colonies of each positive clone and used to transform Escherichia coli DH5␣. Plasmids isolated from four ampicillin-resistant colonies were analyzed by restriction digestion and sequencing.
␤-Galactosidase Activity-Three colonies of yeast EGY48 transformed with the plasmids pJG-GIPC, pSH18 -34, and pHybLexA containing wild-type or mutant cDNA inserts of gp75 or tyrosinase were grown overnight in Zeo ϩ /ura Ϫ /trp Ϫ /dextrose medium. Cultures were diluted to an A 600 of 0.4 in the selection medium Zeo ϩ /leu Ϫ /ura Ϫ /trp Ϫ with galactose and raffinose and incubated for 6 -10 h at 30°C until the A 600 reached 0.5-0.8. The cells were collected, washed, and suspended in phosphate buffer in a volume equivalent to one-fifth of the culture volume, and the ␤-gal assays were performed according to manufacturer instructions (CLONTECH). Units of ␤-gal activity were calculated as units ϭ 1,000 ϫ A 420 /(color development time (min) ϫ culture volume used (ml) ϫ A 600 ).
Construction of Expression Plasmids-Full-length gp75 cDNA was cloned in pCMV5a vector (Sigma). A 0.7-kb cDNA encoding the 243amino acid-long partial GIPC fragment lacking the amino-terminal sequences (NH 2 ⌬GIPC), identified by yeast two-hybrid screening, and a 1.0-kb fragment encoding the entire open reading frame of 333-amino acid residues of GIPC were amplified from human melanoma cDNA using GIPC-specific primers (GenBank TM accession no. AF089816) that incorporate 5Ј-HindIII and 3Ј-XbaI (following the stop codon) or 3Ј-KpnI (without the stop codon) restriction sites were cloned into pFLAG-CMV vectors (Sigma) to produce amino-or carboxyl-terminal FLAG epitopetagged GIPC proteins. A gp75/tyrosinase chimera was constructed using the following strategy. Briefly, a mutant tyrosinase expression plasmid pSVK3hTyr containing a HindIII site at the junction of the transmembrane and the cytoplasmic domains was constructed, from which a HindIII-XbaI DNA fragment encoding the cytoplasmic domain and the 3Ј-untranslated sequences of tyrosinase was isolated and ligated to the large pSVK3hgp75 plasmid fragment containing the lumenal and transmembrane domains of gp75. This resulted in an expression plasmid encoding a chimeric gp75 with the cytoplasmic tail of tyrosinase. The chimeric cDNA was sequenced to confirm restoration of the tyrosinase open reading frame.
Preparation of Antiserum-Polymerase chain reaction-amplified fulllength GIPC was cloned into the EcoRI site of pGEX4T (Amersham Pharmacia Biotech) to generate a glutathione S-transferase (GST)-GIPC fusion protein. The fusion protein for the immunization of rabbits was prepared using GST bulk-purification modules according to manufacturer instructions (Amersham Pharmacia Biotech). The GST-GIPC fusion protein was purified by glutathione-Sepharose chromatography followed by elution from acrylamide gels. Immunization of rabbits with the GST-GIPC protein was performed by Cocalico Biologicals (Reamstown, PA). Briefly, rabbits were immunized with 100 g of purified protein mixed with complete Freund's adjuvant. After three booster injections of 50 g of antigen in incomplete Freund's adjuvant, sera were obtained and tested for anti-GIPC reactivity by Western blotting. Two weeks after the third boost, responding rabbits were boosted twice at 1-week intervals, and serum was collected 1 week after the last boost by sacrificing the animal.
For the isolation of anti-GIPC IgG, sera were first precleared with GST-Sepharose, and IgG from GST-precleared preimmune and immune sera was purified using protein A-Sepharose affinity chromatography (Amersham Pharmacia Biotech). GIPC-specific IgG was isolated by GST-GIPC-Sepharose affinity chromatography (23). GST-precleared sera were used for immunoblotting analyses, and GST-precleared IgG fraction or GST-GIPC affinity-purified IgG were used for immunofluorescence staining.
Cell Lysis and Fractionation-Cells were harvested, washed in phosphate-buffered saline (PBS), and lysed in 50 mM phosphate buffer, pH 7.0, containing 150 mM NaCl, 1% Triton X-100, and a mixture of protease inhibitors (Roche Diagnostics, Indianapolis, IN). Detergent lysates were cleared by centrifuging at 15,000 ϫ g for 20 min. For preparation of cytosolic and membrane-bound proteins, cells were resuspended in PBS containing protein inhibitor mixture and lysed by freeze/thaw cycles in liquid nitrogen. After centrifugation for 1 h at 100,000 ϫ g at 4°C, the cytosol fraction was collected and adjusted to 1% Triton X-100. The membrane pellet was solubilized in lysis buffer and cleared as described above. In experiments to determine the nature of membrane-associated GIPC, the membrane fraction was washed with buffer containing 0.5 M NaCl for 1 h and centrifuged at 100,000 ϫ g (25). The supernatant was collected, and the membrane pellet was lysed and cleared as described above. Nuclei were isolated, and nuclear extracts were prepared as described (26).
Sucrose Density Gradient Fractionation-SK-MEL-19 melanoma cells were harvested, resuspended in gradient buffer (20 mM Tris-HCl, pH 7.2, 5 mM KCl, 5 mM NaCl, 0.1 mM EDTA, and protease inhibitor mixture), homogenized in Dounce homogenizer, and centrifuged at 700 ϫ g for 10 min in gradient buffer containing 0.25 M sucrose. The postnuclear supernatant was fractionated by discontinuous sucrose density gradient as described (27). Fractions (1 ml) were collected from the top and stored at Ϫ20°C until used for immunoblot analysis. Fraction 12, which included the pellet obtained at the bottom of the tube, presumably consisting cellular debris, contained all proteins analyzed and is not shown.
Transfection and Co-immunoprecipitation-COS-7 and clone 22a melanoma cells were co-transfected with 8 g of pCMV5a-gp75 or pCMV5a-gp75/tyr (chimeric gp75 with tyrosinase cytoplasmic tail) or 18 g of pSVK3-gp75⌬14 (gp75 truncated its last 14 amino acids of the cytoplasmic tail) and 2 g of pCMV5a-full-length GIPC-FLAG or pCMV5a-NH 2 ⌬GIPC-FLAG plasmids using LipofectAMINE Plus reagent (Life Technologies, Inc.) according to manufacturer instructions. pCMV2-FLAG-control (an unrelated 82-kDa soluble protein, 2 g) was used for control transfections. After 36 h of transfection, the cells were washed with cold PBS and lysed in 1 ml of lysis buffer as described above. Clear centrifuged lysates were incubated with mAb TA99 (ϳ6 g/ml) on ice for 2 h followed by protein A-Sepharose for 1 h. Immunoprecipitates were washed three times with lysis buffer, twice with lysis buffer containing 0.5 M NaCl, and analyzed by SDS-PAGE and immunoblotting with anti-FLAG mAb M2 (Sigma).
Co-precipitation of Endogenous gp75 and GIPC-Melanoma cells were cultured in the presence or absence of 5 mM hexamethylene bisacetamide (HMBA) for 5 days and harvested immediately or after allowing to recover from HMBA treatment for 8 -48 h. gp75 in detergent lysates was precipitated with mAb TA99, immunoprecipitates were washed twice with lysis buffer, and GIPC in the immunoprecipitates was detected by immunoblot analysis with anti-GIPC serum.
Immunofluorescence Microscopy-SK-MEL-19 and clone 22 cells plated in 8-well chamber slides were left untreated or treated with 5 mM HMBA for 5 days and then allowed to recover from HMBA treatment for 15 or 36 h. The cells were fixed with 2% paraformaldehyde in PBS for 1 h at room temperature and permeabilized with 0.1% Triton X-100 in PBS for 2 min. After blocking with 3% bovine serum albumin in PBS, the cells were incubated overnight at 4°C with GST-precleared or affinity-purified anti-GIPC IgG and preimmune IgG (1:20). After washing and incubation with tetramethylrhodamine isothiocyanate (TRITC)-conjugated goat anti-rabbit IgG (1:100, Jackson ImmunoResearch Laboratories, Inc., West Grove, PA) for 1 h, the cells were finally stained with fluorescein isothiocyanate (FITC)-conjugated anti-gp75 mAb TA99 (1:40) for 1 h at room temperature. Golgi structures were visualized by staining with red (Alexa 594) and green (Alexa 488) fluorescent dye-conjugated Golgi-specific lectin GS-II (Molecular Probes, Eugene, OR), which recognizes nonreducing ␣or ␤-linked N-acetyl-D-glucosamine residues. After each antibody incubation, the cells were washed for 15 min with three changes of 1% Triton X-100/ PBS. Immunofluorescence was captured using a laser scanning confocal image system (LSM510, Carl Zeiss, Jena, Germany).

Identification of gp75 Cytoplasmic Domain-interacting Protein GIPC-
We employed the yeast two-hybrid method to screen a human melanocyte cDNA expression library with a LexA DNA-binding domain-gp75 cytoplasmic tail fusion protein as bait. We obtained 27 positive clones from 13 ϫ 10 6 primary transformants. DNA sequence analysis showed that these 16 positive clones contained two overlapping cDNAs of 1,437 and 1,348 base pairs differing at their 5Ј ends. Fig. 1A shows two representative yeast clones streaked on leucinedeficient (leu Ϫ ) plates. In the absence of galactose (gal), which is required for the expression of the prey protein, yeast clones failed to grow on leu Ϫ /dextrose-containing plates (Fig. 1A, leu Ϫ / dex ϩ ). Thus, the growth of these clones on leu Ϫ plates (Fig. 1A, leu Ϫ /gal ϩ ) and activation of the ␤-gal reporter gene (Fig. 1A, leu Ϫ /gal ϩ /X-gal) depend on the expression and interaction of the protein product encoded by melanocyte-derived cDNA with the LexA-gp75 tail fusion protein.
A BLAST search of GenBank TM data base using the 243amino acid sequence predicted from the 1.4-kb partial melanocyte cDNA showed sequence identity with three PDZ domaincontaining mammalian protein entries. These are 1) a protein fragment encoded by TIP-2 (GenBank TM accession no. AF02884), a cDNA isolated by yeast two-hybrid screen using human T cell leukemia virus transactivator Tax as a bait (28), 2) a human homologue of GIPC (GenBank TM accession no. AF089816), a protein identified by virtue of its binding intracellular G ␣i3 -interacting protein RGS-GAIP (21), and 3) SEM-CAP-1 (GenBank TM accession no. AF061263), a neural semaphorin-binding protein (22). The gp75-interacting protein fragment contained a PDZ domain. This domain, which was originally identified in the post-synaptic density protein PSD-95, disc large protein (Dlg) and zonula occludens-1 (ZO-1), consists of a stretch of 80 -90-amino acid residues. PDZ domain proteins are known to interact with the cytoplasmic tails of membrane proteins and are found in all eukaryotic organisms ranging from Caenorhabditis elegans to humans (reviewed in Refs. 29 -31).
Co-immunoprecipitation of GIPC with gp75-That the interaction of the protein fragment encoded by GIPC cDNA with the cytoplasmic tail of melanocyte-specific gp75 may also be mediated by the PDZ domain is suggested by the following observations. First, class I PDZ domain proteins, including GIPC, are known to specifically bind the carboxyl-terminal amino acid motif Xaa-Ser/Thr-Xaa-Val (where Xaa is any amino acid). The carboxyl-terminal sequence of gp75 is Gln-Ser-Val-Val (14). Second, the protein fragment encoded by all the 16 partial cDNAs contained no protein-protein interaction domain other than the PDZ domain. To test whether the full-length, intracellular transmembrane gp75 also interacts with GIPC, we cloned 0.7-and 1.0-kb GIPC cDNAs encoding partial and com- plete open reading frames in the pFLAG-CMV plasmid to generate FLAG-epitope fusion proteins. COS-7 cells were transiently transfected with the gp75 cDNA expression plasmid together with partial and full-length GIPC-FLAG fusion proteins. Detergent lysates of transfected cells were immunoprecipitated with gp75-specific mAb TA99. Western blot analysis of immunoprecipitated proteins with anti-FLAG mAb M2 (Fig.  1B) showed that both the partial ϳ28-kDa ⌬NH 2 GIPC and full-length 38-kDa GIPC co-precipitated with gp75 (lanes 1 and 3 in the upper panel). No M2 antibody-reactive proteins were precipitated from detergent extracts of cells cotransfected with either empty vector pFLAG (Fig. 1B, lane 2 in the upper panel) or the unrelated control protein control-FLAG fusion plasmid pControl-FLAG (Fig. 1B, lane 4 in the upper panel ). These data show that the PDZ domain of GIPC efficiently binds intracellular membrane-bound gp75.
Carboxyl-terminal Amino Acids of gp75 Are Necessary and Sufficient for Interaction with GIPC in Melanocytic Cells-In melanocytic cells, the cytoplasmic tail of gp75 is involved in the sorting and targeting of this protein to melanosomes. Therefore, we tested whether in melanocytic cells the cytoplasmic tail of gp75 is accessible for interaction with GIPC. We used a nonpigmented gp75-negative melanoma cell line clone 22a to coexpress FLAG-GIPC with either full-length gp75 or a deletion mutant lacking carboxyl-terminal 14 amino acid (⌬C14). In earlier studies we showed that 1) the deletion of carboxylterminal amino acids of gp75 does not affect its recognition by anti-gp75 mAb TA99 and 2) in transiently transfected melanoma cells, ⌬C14 mutant, which lacks the PDZ-binding motif but contains the endosomes/lysosomes sorting signal, displays an intracellular distribution similar but not identical to the wild-type protein (14). Detergent extracts of transfected cells were immunoprecipitated with anti-gp75 mAb TA99 and immunoblotted with anti-FLAG mAb M2 (Fig. 2, upper panel). In cells coexpressing wild-type gp75 and GIPC-FLAG fusion protein, the mAb M2-reactive 38-kDa band corresponding to the co-precipitated GIPC-FLAG fusion protein could be detected (lane1). This interaction was specific as shown by the failure of co-precipitation of control-FLAG with gp75 (lane 3). The coexpression of GIPC with gp75⌬C14 did not result in its co-precipitation with gp75 (lane 2). Immunoblotting of total cell lysates showed approximately equal expression of GIPC-FLAG fusion protein (Fig. 2, middle panel) and full-length and truncated gp75 (bottom panel) in the transfectants. These data demonstrate that the PDZ domain protein GIPC binds efficiently to the carboxyl-terminal amino acids of gp75 in melanocytic cells. This is consistent with the canonical mode of PDZ domain interaction with the carboxyl-terminal amino acids of target membrane proteins (32).
Specific Interaction of GIPC with the Unique Carboxyl Domain of gp75-gp75 is a member of the tyrosinase family of melanosomal membrane proteins, and its cytoplasmic tail, which contains the conserved dileucine sorting motif, shares limited sequence homology with other members of the tyrosinase family (14). Examination of amino acid sequences of the cytoplasmic domains of gp75, tyrosinase, TRP-2, and other known melanosomal proteins including gp100, Melan A/MART-1, and P (pink-eyed dilution) protein showed that the PDZ-binding motif, however, is found only in the gp75 tail (Fig. 3A). It therefore seems that the unique carboxyl-terminal amino acid sequence of gp75 is sufficient for its interaction with GIPC. We hypothesized that other tyrosinase family proteins, which lack the extended carboxyl-terminal PDZ-binding motif, do not interact with GIPC. We tested this by yeast two-hybrid analysis and co-precipitation experiments using wild-type and carboxyl-terminal sequence mutant proteins. In Fig. 3B, ␤-gal activity in two representative clones of double transformants of S. cerevisiae expressing GIPC and wild-type or mutant cytoplasmic tails of gp75 and tyrosinase is shown. As predicted, wtgp75 tail (Fig. 3B, wtgp75) showed a strong interaction with GIPC as seen by the activation of ␤-gal, and the gp75 tail lacking terminal residues Gln-Ser-Val-Val did not interact with GIPC (Fig. 3B, gp75⌬QSVV). The cytoplasmic tail of tyrosinase did not show interaction with GIPC (Fig. 3B,  Tyr). Substitution of the tyrosinase carboxyl-terminal amino acid residues His-Leu with valine, which results in a carboxyl-terminal sequence Ser-Val-Val, allowed efficient interaction of the tyrosinase tail with GIPC (Fig. 3B, Tyr-QSVV). Although interaction of GIPC with the mutant tyrosinase-QSVV tail was qualitatively comparable with its interaction with the gp75 tail (as shown by the in-plate ␤-gal assay in Fig. 3B), a quantitative difference between these interactions could be noted (Fig. 3C). However, ␤-gal activity of both the wild-type gp75 tail and tyr-QSVV transformants was significantly higher than the activity seen for the positive control (Fig. 3C, junϩfos). Consistent with the in-plate assay, ␤-gal activity was undetectable in gp75⌬QSVV and tyrosinase transformants. These data suggest that although the terminal three amino acids Ser-Val-Val of gp75 are necessary for GIPC binding, other residues in the gp75 tail may also influence this binding. A lack of interaction between the tyrosinase tail and GIPC in vivo was confirmed by coexpression of GIPC with a chimeric gp75 containing the lumenal and transmembrane domains of gp75 and the cytoplasmic tail of tyrosinase in 22a cells (Fig. 3, D and E). As shown in Fig. 3E, although the immunoprecipitation of gp75 from wtgp75 transfectants co-precipitated GIPC (lane 1, upper panel), no co-precipitated GIPC band could be seen from chimeric gp75/tyr transfectants (lane 2, upper panel). Utilization of anti-gp75 mAb TA99 for immunoprecipitation of both wild-type gp75 and gp75/tyr chimera transfectants allowed us to conclude that the lumenal domain of gp75 does not contribute to its interaction with GIPC.
Expression of GIPC in Melanocytic Cells-Northern analysis of primary melanocytes and a panel of pigmented and nonpigmented melanoma cells using the 1.4-kb cDNA obtained from the yeast two-hybrid screen revealed a single band of 1.8-kb RNA in all cell lines tested (data not shown). To characterize the human GIPC protein, we generated polyclonal antibodies against the GST-GIPC fusion protein. Fig. 4A shows Western blot analysis of melanoma cell extracts with preimmune and anti-GST-GIPC immune serum. Immune serum from GST-GIPC-immunized rabbits detected three major bands including the prominent band at 38 kDa, the expected molecular mass of GIPC. A one-step GST-Sepharose affinity chromatography produced a serum that specifically detected the 38-kDa GIPC band. We used GST-precleared serum to study the expression of GIPC in human melanocytic cells. The steady-state level of GIPC expression in neonatal foreskin melanocytes (NMC), primary (WM75, WM35, and WM98-1) and metastatic (WM451, WM1205, clone 22 (WM35, WM98-1, WM451, WM1205, and clone 22a) cell lines is shown in Fig. 4B. Nearly equal amounts of GIPC are expressed in normal and malignant melanocytes, and GIPC seems to be regulated independently of gp75, pigmentation, or melanoma tumor progression.
Intracellular Distribution of GIPC-To study the intracellular distribution of GIPC and investigate whether it colocalizes with gp75 in human melanoma cells, we performed double immunofluorescence staining. Paraformaldehyde-fixed and detergent-permeabilized SK-MEL-19 melanoma cells were incubated with either preimmune serum or anti-GIPC antibody followed by TRITC-conjugated ant-rabbit IgG and then FITCconjugated anti-gp75 mAb TA99. The anti-GIPC antibody showed patchy staining in the juxtanuclear region, punctate staining in the cytoplasm, and a variable nuclear staining ( No specific staining could be seen with the preimmune serum ( Fig. 5A-a). gp75 showed similar juxtanuclear and punctate staining, consistent with its localization in the Golgi, endosomes, and melanosomes (Fig. 5, A-b and  e). Staining for the two proteins showed only limited overlap localized primarily to a few vesicles near the intense patchy areas of GIPC staining (Fig. 5A-f). Accordingly, immunoprecipitation of endogenous gp75 with mAb TA99 showed co-precipitation of only a weak band corresponding to GIPC (data not shown). Because coexpression studies in COS and melanoma cells showed that GIPC interacts with the membrane-bound gp75, we investigated whether GIPC in melanoma cells is associated with membranes. The distribution of GIPC in postnuclear supernatant, crude membrane, and cytosolic fractions is shown in Fig. 5B. A significant amount (ϳ50%) of GIPC was found to be associated with the membrane fraction. After incubation of the crude membrane fraction with 0.5 M NaCl for 1 h, the bulk of GIPC remained in the insoluble membrane fraction. Almost all the cellular gp75, on the other hand, is present in the membranes and could not be dissociated by high salt. Thus, although a significant amount of GIPC is found associated with membranes in melanoma cells, only a small fraction of this protein appeared to interact with the cytoplasmic domain of gp75. The affinity-purified anti-GIPC antibody showed variable nuclear staining (Fig. 5, A-d and f). We therefore investigated whether GIPC could be detected biochemically in melanoma nuclei. Western blotting showed a weak band corresponding to the molecular mass of GIPC in melanoma nuclear extracts (Fig.  5B, right panel) confirming the immunofluorescence observation (see "Discussion").
To further explore whether endogenous GIPC and gp75 can be co-sedimented in same-cellular fractions, we performed sucrose density gradient fractionation. The postnuclear supernatant isolated from SK-MEL-19 cells was fractionated on discontinuous sucrose density gradient (0.25-1.5 M), and the distribution of various proteins was studied by Western blot analysis (Fig. 6). Based on the enrichment of specific proteins, fractions collected from the top to bottom were designated as soluble and low density (fractions 1-3), intermediate density (fractions 4 -8), and high density (fractions 9 -11). GIPC showed a tri-partite distribution. Although a significant proportion of the protein was distributed among low and high density fractions, a small amount of GIPC could also be detected in the intermediate fractions (low Ͼ Ͼ dense Ͼ intermediate). Low density fractions contained the bulk of cellular clathrin, and high density fractions were enriched for the lysosomal membrane protein LAMP-1, whereas intermediate fractions were enriched for adaptor protein AP-1. gp75 and tyrosinase were enriched in intermediate and high density fractions, which also contained varying amounts of melanin pigment presumably associated with melanosomes at various stages of maturation. A small amount of mature gp75 and tyrosinase could also be seen in clathrin-enriched low density

FIG. 4. Characterization of anti-GIPC antiserum and expression of GIPC in melanocytic cells.
A, equal volumes of human melanoma SK-MEL-19 cell lysate were electrophoresed in multiple lanes by SDS-PAGE, transferred to polyvinylidene difluoride membranes, and probed with preimmune and immune sera and GST-precleared immune serum (1:800) followed by horseradish peroxidaseconjugated donkey anti-rabbit antibody (1:10,000). A specific band of endogenous GIPC detected by immune serum and GST-precleared antiserum but not by preimmune serum is shown by the arrow on the right. B, GIPC expression in melanocyte and melanoma cells. Total cell lysates (50 g of protein) from neonatal foreskin melanocytes and primary and metastatic melanoma cells were electrophoresed by SDS-PAGE and immunoblotted with GST precleared anti-GIPC serum (1: 1,000). Western blotting for ␥-tubulin is shown as control for protein loading. Prestained molecular mass markers are shown on the right. GIPC Interacts with Newly Synthesized gp75-We consistently found that only a small amount of endogenous GIPC associates with gp75 at steady state. However, because we also found that endogenous GIPC in melanoma cells readily coprecipitated with transiently expressed gp75 (data not shown), we considered the possibility that interaction of GIPC with gp75 may be transient. The presence of a small amount of GIPC and gp75 in intermediate density clathrin-containing fractions (fractions 6 and 7), in which AP-1 is also enriched (Fig. 6A), suggested that such an interaction might be restricted to a transport compartment.
To examine this possibility, we used a strategy to synchronize biosynthesis of endogenous gp75 by selective repression and reactivation of the gp75 gene. We showed earlier that treatment of pigmented melanoma cells with a polar planar compound HMBA selectively and completely extinguishes gp75 (33). This inhibition of gp75 expression occurs at the level of transcription and is reversible (34). We tested the interaction of synchronously and newly synthesized gp75 with GIPC in cells recovering from inhibition by HMBA (Fig. 7). The treatment of pigmented SK-MEL-19 melanoma cells with HMBA for 5 days extinguishes expression of gp75, and removal of this drug allows gp75 re-expression (Fig. 7, bottom). The expression of GIPC was not affected by treatment with HMBA or during recovery from HMBA (Fig. 7, middle). Immunoprecipitation with anti-gp75 mAb TA99 followed by Western blotting with the rabbit ant-GIPC antibody showed a weak co-precipitated band of ϳ38 kDa in control untreated cells (arrow in lane C, upper panel). As expected, after the extinction of gp75 expression by HMBA the co-precipitated 38-kDa band was undetectable ( lane H, upper panel). Immunoprecipitation of newly accumulating gp75 from lysates of cells recovering from HMBA treatment clearly showed a 38-kDa GIPC band. The intensity of this band increased with prolonged recovery up to 36 h (lanes 15- h and 36-h), but decreased after 36 h to a level nearly comparable with that of untreated cells (arrow in lane 48 h, upper panel). Similar data were obtained using clone 22, another pigmented melanoma cell line (data not shown). These results show that GIPC only binds newly accumulating gp75.
At steady state, the transient interaction of GIPC with newly synthesized gp75 molecules, which constitute a relatively small proportion of intracellular gp75 compared with a large pool of preexisting gp75 protein, accounts for co-precipitation of a small amount of GIPC. Synchronization of gp75 synthesis either by transient transfection or selective repression and reactivation increased the pool of newly synthesized protein, allowing for visualization of this interaction.
Double immunofluorescence microscopy of HMBA-treated and untreated SK-MEL-19 cells and cells recovering from HMBA confirmed these biochemical observations. Fig. 8A, a-e, shows the staining of untreated cells. a shows background staining with the preimmune rabbit IgG. The staining of en- Interestingly, intracellular vesicles in which the two proteins are colocalized appeared to show a polarized intracellular distribution. Fig. 8, B and C, shows staining of GIPC and newly synthesized gp75 in clone 22 melanoma cells (recovering from HMBA for 36 h). Most vesicles showing staining for both gp75 and GIPC appeared to be localized predominantly in an area of the cytoplasm (Fig. 8C, right panel) distinct from gp75-enriched structures (Fig. 8C, left panel).
Double immunofluorescence staining with the anti-GIPC antibody and a Golgi-specific lectin GS-II showed colocalization of GIPC with the Golgi marker in both control SK-MEL-19 cells and cells recovering from HMBA (Fig. 9, bottom panels). Similarly, gp75 showed colocalization with the Golgi marker ( Fig.  9, left middle panel). In cells recovering from HMBA, the bulk of newly synthesized gp75 can be seen localized to the Golgi (Fig. 9, right middle panel). Association of GIPC with this newly synthesized gp75 predominantly in the Golgi is shown in the top right panel in Fig. 9. Minimal colocalization of steadystate gp75 with GIPC was noted in untreated cells (Fig. 9, left top panel) similar to data shown in Figs. 5 and 8. Together with biochemical data, these studies demonstrate that the PDZ domain protein GIPC/SEMCAP-1 interacts with newly synthesized melanocyte-specific membrane protein gp75. DISCUSSION We used the yeast two-hybrid system to identify proteins that interact with the cytoplasmic domain of melanosomal membrane protein gp75 and found GIPC/SEMCAP-1 as a gp75-interacting protein. In melanocytic cells, GIPC co-precipitated and colocalized with newly accumulated gp75 in the Golgi region, suggesting a role for GIPC in intracellular transport of gp75 during biosynthesis.
Specific Interaction of gp75 with PDZ Domain of GIPC-GIPC is a member of a large family of PDZ domain proteins. PDZ domains that recognize the sequence motif Ser/Thr-Xaa-Val are classified as class I PDZ domains (35). The requirement of gp75 carboxyl-terminal Ser-Val-Val residues for its interaction with the single PDZ domain present in GIPC is consistent with a canonical mode of class I PDZ domain interaction (32,35). It is interesting to note that although GIPC/SEMCAP-1, which was isolated independently as a GAIP-binding protein from a rat pituitary library and as neurally enriched semaphorin-binding protein, it is expressed widely in mammalian tissues. Thus, the specificity of GIPC interactions may be determined by tissue distribution of the interacting target proteins. For example, because GIPC/SEMCAP-1 and M-SemF, but not GAIP, are expressed in the brain, it has been proposed that M-SemF (but not GAIP) is the biologically relevant target for GIPC in this tissue (22). The expression of GIPC targets GAIP and M-SemF in melanocytes has not been investigated. Although the carboxyl-terminal sequence motif of gp75 is a potential target for other class I PDZ domain proteins, coimmunoprecipitation of endogenous GIPC with gp75 and immunofluorescence data presented here support its specific interaction with GIPC in melanocytic cells. Additionally, among 27 positive clones isolated by yeast two-hybrid screening, we found only GIPC (16 of 27 positives) but no other PDZ domain proteins.
Expression of GIPC in Melanocytic Cells-Although gp75 expression in melanoma cells is widely variable and readily modulated (33,36), nearly identical amounts of GIPC are present in all mouse and human melanocytic cells and nonmelanocytic cells (e.g. COS cells, mouse fibroblasts, and rat PC12 cells; data not shown). This is reminiscent of a tightly regulated expression pattern of proteins with essential or housekeeping functions (e.g. GAPDH and tubulin) and suggests that PDZ interactions mediated by GIPC play important roles in the biology of melanocytic cells. Because we isolated multiple clones of GIPC by yeast two-hybrid analysis and found that GIPC efficiently associates with gp75 when coexpressed by transient transfection in both melanoma and nonmelanoma cells, we were intrigued by the fact that only a small amount of GIPC associated with endogenous gp75. However, nearly 50% of cellular GIPC is associated with membranes in melanoma cells, and a significant amount of GIPC co-sedimented with gp75 and melanin-containing dense vesicles. Thus, at steady state the bulk of membrane-bound GIPC is not associated with gp75, suggesting that in melanocytes GIPC interacts with multiple target proteins.
Interestingly, a small amount of GIPC was also found to co-fractionate with AP-1 and clathrin-containing vesicles. In this context, it is of interest to note that the GIPC-interacting protein RGS-GAIP seems to be localized to clathrin-coated vesicles, and in HeLa cells immunogold electron microscopic studies showed that GIPC is associated with small vesicles located near the cell membrane (21). Interaction of GIPC/SEM-CAP-1 with SemF, on the other hand, seems to create complex protein aggregates at specialized domains on the plasma membrane (22).
Nuclear Localization of GIPC-Using the antisera raised against melanoma GIPC, we found variable staining of nuclei and a small amount of protein in the nuclear extracts. It seems that the intracellular distribution of GIPC is also variable, suggesting a possible redistribution of the protein in response FIG. 9. Intracellular localization of GIPC, gp75, and the Golgi marker lectin GS-II. SK-MEL-19 melanoma cells grown in 8-well chamber slides were left untreated (left panels) or treated with HMBA for 5 days and allowed to recover from treatment with HMBA for 15 h (right panels). After fixation and permeabilization, the cells were stained with anti-GIPC IgG followed by TRITC-conjugated anti-rabbit antibody and FITC-conjugated anti-gp75 mAb TA99 (top panels) or FITC-conjugated anti-gp75 mAb TA99 followed by Alexa Fluor 594conjugated lectin GS-II (1:100) (middle panels) or anti-GIPC antibody followed by TRITC-conjugated anti-rabbit antibody and Alexa Fluor 488-conjugated lectin GS-II (1:50). The distribution of GIPC, gp75, and the Golgi marker were examined by confocal microscopy. Scale bar, 20 m.
to increased synthesis of interacting protein gp75 (compare GIPC distribution in steady-state cells in Fig. 5D and the top left panel in Fig. 9 versus cells recovering from HMBA in Figs. 8j and the top right panel in Fig. 9). Because, GIPC lacks recognizable nuclear localization signals it is possible that GIPC is translocated to the nucleus in association with other cytosolic protein(s). This possibility is highlighted by the following observations. 1) GIPC, indeed, was first isolated as a partial protein fragment named TIP based on its interaction with the carboxyl terminus of human T cell leukemia virus-1encoded protein Tax, a transcriptional activator (28). However, endogenous cellular transcription factors to which GIPC can bind have not yet been identified. 2) The PDZ domain protein CASK has been shown to be translocated to the nucleus in association with the Tbr-1 transcription factor after the binding of CASK to the cytoplasmic tail of the transmembrane adhesion molecule syndecan (37).
Transient Interaction of Newly Synthesized gp75 with GIPC-We found that endogenous and transfected GIPC efficiently associate with coexpressed gp75 in both melanoma or nonmelanoma cells. Moreover, GIPC co-sedimented with gp75containing vesicles. However, immunoprecipitation of gp75 from total cellular lysates of melanoma cells or from gp75enriched fractions showed only a weak co-precipitated band of GIPC. We considered several possible explanations for the discrepancy in interaction of GIPC with transfected and endogenous gp75. It is possible that GIPC binds only a small distinct subpopulation of mature gp75 molecules, or GIPC may bind all gp75 molecules only transiently either during the biosynthesis and/or degradation of gp75. Alternatively, the inefficient coprecipitation of GIPC could be caused by the rapid degradation of GIPC upon its association with membranes and/or gp75. To examine whether membrane-bound GIPC is degraded rapidly, we estimated the half-life of total, cytosolic, and membraneassociated GIPC in gp75 ϩ and gp75 Ϫ melanoma cells by pulsechase metabolic labeling and immunoprecipitation. Newly synthesized cytosolic and membrane-bound GIPC was degraded with a similar half-life (ϳ18 h) in both gp75 ϩ and gp75 Ϫ cell lines (data not shown), suggesting that our inability to efficiently coprecipitate GIPC with gp75 was not caused by a preferential degradation of membrane or gp75-associated GIPC.
Our strategy to selectively synchronize synthesis of gp75 by inhibiting gp75 transcription and then releasing the inhibition allowed us to demonstrate that endogenous GIPC in melanocytic cells interacts only with newly synthesized gp75 molecules. We examined the possibility that this could be demonstrated by pulse-chase metabolic labeling followed by immunoprecipitation with anti-gp75 mAb TA99. However, only a weak co-precipitating radioactive band corresponding to GIPC could be detected at the 8-h chase point (data not shown). This is consistent with the data from the co-immunoprecipitation of steady-state proteins. This highlights the importance of the unique strategy we employed to demonstrate this transient interaction by synchronization of endogenous gp75 biosynthesis.
Role of GIPC in Intracellular Sorting of gp75-We showed that the interaction of GIPC with the gp75 cytoplasmic tail is dynamic and limited to newly synthesized gp75 protein localized in and near the Golgi. A transient interaction during biosynthesis of gp75 suggests that the carboxyl PDZ-binding motif plays a role in intracellular sorting. A sequence motif in the cytoplasmic tail of gp75 (and other melanosomal membrane proteins) that is known to interact with cytosolic proteins is the dileucine sorting sequence (14). The binding of adaptor complexes, specifically AP-3, to this sequence is known to mediate the sorting of melanosomal proteins along the lysosome/melanosome pathway (15). Mutations in AP-3 produce the mocha phenotype in mouse and Hermansky-Pudlak-like syndrome in man (reviewed in Ref. 16). The role GIPC plays in intracellular protein sorting remains to be investigated. In transiently transfected mouse fibroblasts and COS cells, the intracellular distribution of gp75⌬C14 protein lacking the PDZ-binding motif appears grossly similar to but distinct from that of wild-type protein (14). Preliminary pulse-chase experiments indicated that carboxyl-terminal residues determine the stability of newly synthesized gp75 expressed in nonpigmented melanoma cells (data not shown). It is tempting to speculate that GIPC participates in the sorting of gp75 to a vesicular compartment, presumably premelanosomes. The observation that GIPC binds GAIP, a G ␣i3 -interacting protein localized to vesicles, implicates GIPC in G protein-mediated control of vesicular trafficking (21). Similarly, the interaction of GIPC/SEMCAP-1 with the plasma membrane protein SemF appears to link SemF to G protein-signaling pathways (22). Recently, it has been reported that mutations in the novel intracellular G protein-coupled receptor OA1 are associated with an X-linked ocular albinism type 1 (OA1) (38). A defect in melanosome biogenesis in skin melanocytes and retinal pigment epithelial cells seems responsible for the hypopigmentary phenotype in OA1 (39). It is not clear whether GIPC or its interaction with gp75 plays a role in OA1-mediated G protein signaling and pigment regulation.
3) The 36-amino acid residue-long tail of gp75 is the longest among TRPs and contains some unique features as well as some shared with other TRPs (14). The relative abundance of GIPC in melanocytic cells and its interaction with the unique carboxyl terminus of gp75 suggests a possible role for this interaction in the biological functions of the enigmatic brown locus protein gp75. In this context it is worth noting that we have mapped the human GIPC gene to chromosome 19p13.1, where BRHC (OMIM locus 113750), the human brown hair color 1 locus, is also mapped (40,41). 2 Alternative Functions of GIPC-The best known function of PDZ domains is to organize the assembly of protein complexes, especially in submembranous areas, by binding to the cytoplasmic tails of membrane proteins (42). Generally, multiple PDZ domains present in these proteins allow binding and assembly of protein complexes. Other functional domains such as the Src homology (SH)-3 domain, tyrosine phosphatase, guanylate kinase, and calcium/calmodulin protein kinase domains on PDZ proteins or the interacting target proteins are thought to mediate the signaling functions of these complexes. For example, clustering of NMDA (N-methyl-D-aspartate) receptors in neuronal cells by PSD-95, a protein with multiple PDZ domains, has been shown to couple the receptor activation to specific excitotoxic signaling (30). GIPC, however, has only one PDZ domain and no other recognizable sequence motif involved in signaling (21,22). This raises the possibility that gp75-bound GIPC interacts with other proteins through novel non-PDZ interactions. Alternatively, the functions of GIPC may involve the acyl carrier protein (ACP) domain present at its C terminus. De Vries et al. (21) suggested that GIPC could act as a carrier for palmitoyl moieties for palmitoylation of target proteins including GAIP and the ␣ subunits of G protein. Thus, palmitoylation of the cytoplasmic tails of TRPs and/or other melanosomal membrane proteins by GIPC also merits investigation.