A role for the PDZ-binding domain of the coxsackie B virus and adenovirus receptor (CAR) in cell adhesion and growth

The coxsackie and adenovirus receptor (CAR) is an integral membrane protein that plays a role in both viral infection and cell adhesion. As a viral receptor, the intracellular C-terminal tail (C-terminus) of CAR plays no apparent role because deletion of the intracellular domain supports a similar adenovirus (Ad) infection to wild type (Wang and Bergelson, 1999; Walters et al., 2001). The C-terminus of CAR, however, is important for the endogenous function of CAR. Deletion of, and specific mutations within, the C-terminus affect cellular localization (Walters et al., 2001; Cohen et al., 2001a) and may affect many different protein-protein interactions.

The last four amino acids (aa) of CAR are GSIV. The C-terminal motif XS/TXV (where X is any aa) represents a class 1 PDZ (PSD-95/Disc-large/ZO-1) binding motif for interacting with proteins containing PDZ domains (Sheng and Sala, 2001; Hung and Sheng, 2002). PDZ domains are approximately 90 aa long modular-protein-interacting domains present once or multiple times within certain proteins and are involved in the assembly of large protein complexes, cell signaling, or intracellular trafficking. PDZ-domain-containing proteins generally contain additional protein interaction domains. For example, members of the membrane-associated guanylate kinase (MAGUK) family generally contain an SH3 domain, a guanylate kinase-like domain and one or several PDZ domains. PDZ domains bind to their specific target proteins through either the terminal four aa or an internal peptide that bends in a β-hairpin structure mimicking a four aa peptide.

Previously, CAR has been shown to interact with the prototypical PDZ-domain-containing tight junction protein ZO-1, suggesting a role for CAR in the formation of epithelial tight junctions and paracellular permeability (Cohen et al., 2001b). It has also been shown to interact with ligand-of-numb protein-X by the yeast two-hybrid screen. This protein interacts with the cell-fate determinate numb and potentially suggests a role for CAR in development (Sollerbrant et al., 2003). However, the presence of CAR in the junction adhesion complex in human airway epithelia suggests potential interactions with other PDZ-domain-containing proteins present in this region, such as MAGI-1 (Ide et al., 1999), PICK1 (Jaulin-Bastard et al., 2001), SAP97 (Muller et al., 1995; Ide et al., 1999), GRIP1 (Bladt et al., 2002), LIN-7 (Perego et al., 1999), or PIST (Neudauer et al., 2001).

The role of CAR as an adhesion protein interacting homotypically via its most distal IgG-like domain between adjacent cells has recently been established. CHO cells expressing CAR cluster more rapidly than parental cells that do not express CAR (Honda et al., 2000; Cohen et al., 2001b). Increased CAR expression results in a significant accumulation of CAR protein at the junctions between adjacent overexpressing cells (Cohen et al., 2001b; Walters et al., 2002). Additionally, expression of CAR in well-differentiated human airway epithelia (HAE) increases the transepithelial resistance or `tightness' of these epithelia (Walters et al., 2002). By contrast, antibody to the extracellular domain of CAR, Ad, or Ad fiber alone disrupts the epithelial junctions.

CAR expression also has an affect on cellular growth properties. Ablation of CAR expression is found in several types of cancer, such as bladder, prostate, melanoma and colon (Li et al., 1999; Okegawa et al., 2000; Okegawa et al., 2001). Accordingly, reintroducing or increasing CAR expression in some transformed cell models is sufficient to decrease the growth of these cells. The current mechanism proposed for this is not clearly defined but it is thought to be a result of contact inhibition signaled through CAR:CAR interactions between adjacent cells (Okegawa et al., 2001).

We wanted to investigate the functional importance of the PDZ-binding motif of CAR in these different roles for CAR. We hypothesized that deletion of the GSIV motif would have no consequence for Ad infection. Considering the potential interactions mediated by this peptide motif, however, we hypothesized that deletion of this sequence may play a role both in cell adhesion and in growth regulation. Furthermore, elucidation of additional partners that interact with CAR might lead to a better understanding of the endogenous roles for CAR and provide novel mechanisms relevant to cancer biology.

FLAG M2 and LIN-7 antibodies (Ab) were purchased from Sigma (F3165, L1538, St Louis, MO), rabbit anti-FLAG (Immunology Consultants Laboratory, Newberg, OR), GRIP1 (06-986, Upstate Biotechnology, Lake Placid, NY), HA (Boehringer Mannheim), Alexa-488 and -568 conjugated goat anti-mouse or anti-rabbit Abs, and rabbit anti-GFP were from Molecular Probes (Eugene, OR). Myc 9E10 and RmcB Ab (CRL-2379, ATCC, Manassas, VA) were produced by the University of Iowa Hybridoma Core. COS-7 and L cells were from ATCC (Manassas, VA) and maintained under standard culture conditions (D-MEM with 10% FCS, penicillin and streptomycin). Ad serotype 5 containing the β-galactosidase (Ad-βGal), eGFP, RFP (peGFP-N1, pDSRed1, Clontech, Palo Alto, CA), or CAR gene have previously been described (Walters et al., 1999; Ashbourne Excoffon et al., 2003). The University of Iowa Gene Transfer Vector Core produced all viruses. Several cDNAs were kind gifts from the following investigators: pcDNA3.1(–)-FLAG-hCAR was from Ronald Crystal; peGFP-MAGI-1b was from Irina Dobrosotskaya; SAP97-GFP was from Johannes Hell; PSD-95-GFP was from David Bredt; pRK5-myc-mLIN-7b was from Ben Margolis; pGRIP1-myc was from Moran Sheng; pKH3mPIST was from Ian Macara. The cDNA for PICK1-GFP has previously been described (Hruska-Hageman et al., 2002) and MAGI-1b-CMV-myc was subcloned and contained aa 642-1287.

Primary HAE were isolated from trachea and bronchi of donor lungs and seeded onto collagen-coated semi-permeable membranes (Millipore, Bedford, MA) and grown at the air-liquid interface as previously described (Zabner et al., 1996; Karp et al., 2002). Approximately 2 weeks after seeding, cultures were well-differentiated and attained a measurable transepithelial resistance.

Cultures of well-differentiated HAE from eight separate donors were mounted in Ussing chambers and transepithelial resistance was measured as previously described (Smith et al., 1994; Zabner et al., 1996). Total RNA was extracted from HAE cultures from the same eight separate donors using TRIzol reagent (Invitrogen, Carlsbad, CA) according to manufacturer's recommendations. Following a 30 minute treatment with 1U/10 μg RNA with DNase I (Invitrogen, Carlsbad, CA), an additional RNA cleanup step was performed using the Qiagen RNeasy total RNA isolation kit (Chatsworth, CA). The RNA was processed for expression array according to manufacturer's recommended protocols using Affymetrix U133A genechip (Reese et al., 2001).

The PDZ-binding motif of CAR was deleted via PCR by using a primer to the 5′ end of CAR (CAR5′: 5′ TGGAATTCCCAGGAGCGAGAG 3′) and a primer that excluded the last four aa of the C-terminus (CAR^ΔGSIV: 5′ CGGGATCCCTAATCCTTGCTCTGTGCTGG 3′) (95°C, 5 minutes; 30 rounds of 94°C, 30 seconds, 58°C, 45 seconds, 72°C, 45 seconds; 72°C, 5 minutes). The fragment was digested with EcoRI and BamHI and cloned into pcDNA3.1(–) or an Ad5-CMV shuttle plasmid. DNA sequencing and western blot confirmed the expected protein. Ad-CAR^ΔGSIV was generated as previously described (Ashbourne Excoffon et al., 2003).

Primary HAE were treated apically with 5 mM EGTA in water for 5 minutes when the transepithelial resistance was reduced to background levels. EGTA was then removed and virus (MOI=200) in 40 μl of EMEM was added apically and left on the cells for 1 hour. Virus was removed and the basolateral media was changed. Transepithelial resistance was evaluated 3 days later using an Ohm meter (World Precision Instruments, Sarasota, FL). Greater than 60% transfection rate is routinely achieved by this method (Walters et al., 2001).

Mouse fibroblast L-cells that do not make epithelial junctions were selected to evaluate the unique contribution of CAR. Cells were plated in six-well dishes (2×10⁵ cells/well) and transfected the next day with CAR, CAR^ΔGSIV, or eGFP-N1 using Lipofectamine Plus (Invitrogen, Carlsbad, CA) by standard procedures. After 48 hours cells were selectively grown in the presence of 1 mg/ml Geneticin (Invitrogen, Carlsbad, CA). After approximately 3 weeks in culture, cells were tested for the presence of CAR or CAR^ΔGSIV by RT-PCR and western blot. Cells stably expressing eGFP were confirmed by fluorescence microscopy. Ad infection was investigated by seeding cells into 24 well dishes (n=6) and infecting with Ad-β-gal (MOI=100) for 1 hour at 37°C. After 48 hours cells were lysed and β-galactosidase expression per mg protein was determined as previously described (Ashbourne Excoffon et al., 2003). Growth curves were generated by seeding 5×10³ cells per well on a 24 well dish and protein levels were analyzed by Bio-Rad Protein Assay (Bio-Rad Laboratories, Hercules, CA, n=5) or counted by hemacytometer (Hausser Scientific Co., Horsham, PA, n=3) or Coulter Counter (Z1, Coulter, Miami, FL, n=3).

COS-7 cells were electroporated by standard methodologies. Briefly, 10 million cells were mixed with 20 μg of plasmid DNA for single transfection or 15 μg of each DNA for double transfections, 400 μl of cytomix (120 mM KCl, 0.15 mM CaCl₂, 10 mM K₂HPO₄, 10 mM KH₂PO₄, 25 mM HEPES, 2 mM EGTA, 5 mM MgCl₂, 2 mM ATP and glutathione) and put in an electroporation cuvette (Bio-Rad Laboratories, Hercules, CA) for 30 minutes on ice. After electroporation, cells were seeded onto 10 cm dishes for immunoprecipitation (IP) and collagen coated glass chamber slides for immunofluorescence studies 2 days later.

COS-7 cells grown on collagen coated chamber slides were washed once with PBS, fixed with 4% paraformaldehyde, permabilized with 0.1% Triton X-100, and blocked with 2% BSA in SuperBlock (Pierce, Rockford, IL). Cells were incubated with primary Ab, washed extensively and incubated with fluorescently labeled secondary Ab. After washing, slides were coverslipped with Vectashield mounting media (Vector Laboratories, Burlingame, CA). Images were acquired with a BioRad MRC-1024 Laser Scanning Confocal Microscope (Hercules, CA) mounted on a Nikon E600 microscope (Melville, NY) using a 60× oil immersion lens.

Cells from 2-100 mm plates were placed on ice, washed once with ice cold PBS, and lysed with lysis buffer [50 mM Tris-HCl, pH 7.5, 137 mM NaCl, 1% Triton X-100, 5 mM EDTA, 1 mM EGTA, protease inhibitors (10 μg/ml) leupeptin, aprotinin, pepstatin, and 1 mM phenylmethylsulfonyl fluoride] by rocking at 4°C. Cells were scraped, sonicated 5× and spun in a microcentrifuge at full speed for 10 minutes. Supernatant was incubated with the indicated Ab with rotation at 4°C overnight. Protein A or G conjugated sephararose (Amersham Biosciences, Uppsala Sweden) was added for 1-2 hours followed by a wash with lysis buffer, 10% lysis buffer in TBS (50 mM Tris-HCl, pH 7.5, 137 mM NaCl), and TBS. Beads were suspended in loading buffer (4% sodium dodecyl sulfate, 100 mM dithiothreitol, 20% Glycerol, 65 mM Tris, pH 6.8, 0.005% bromophenol blue) and proteins were separated by SDS-poly acrylamide gel electrophoresis. Gels were transferred to a polyvinylidene difluoride membrane (Millipore, Bedford, MA), blocked with 5% BSA, washed, probed with primary Ab as indicated, followed by washing and incubation with protein A or G conjugated HRP (Pierce, Rockford, IL). Bands were detected with ECL reagents (Pierce, Rockford, IL).

To evaluate the effect of basal transcript levels of CAR on transepithelial resistance, we correlated transepithelial resistance with the CAR mRNA levels on well-differentiated primary human airway epithelia (HAE) from eight separate donors. In cultures from each donor, Ussing chamber studies were performed to determine various parameters including transepithelial resistance. The transepithelial resistance ranged from 378-880 Ohm. Concurrently, RNA was extracted from cultures from each donor and analyzed by genechip. Increasing RNA transcript levels for CAR correlated significantly with increasing resistance in these samples (R²=0.80, Fig. 1A). In comparison, there was little or no relationship with other adhesion proteins [e.g. occludin (R²=0.28), JAM (R²=0.03) and E-cadherin (R²=0.05)] or PDZ-domain-containing proteins [e.g. ZO-1 (R²=0.31), LIN7 (R²=0.21)]. An inverse relationship was observed with some PDZ-domain-containing proteins [e.g. MAGI-1 (R²=0.49), GRIP1 (R²=0.38)]. No relationship was found between the transcript levels of housekeeping genes, such as β-actin (R²=0.00, Fig. 1B) or GAPDH (R²=0.002) with transepithelial resistance. Because mRNA does not necessarily correlate with protein levels, this indicates that transcriptional regulation of CAR or similarly regulated proteins correlate with the electrical tightness of airway epithelia.

We have previously shown that overexpression of CAR significantly increases the transepithelial resistance of HAE (Walters et al., 2002). It is currently thought that this increase is mainly due to increased homotypic extracellular interactions between CAR molecules present on adjacent cells. However, a portion of this characteristic may be due to intracellular interactions between CAR and other interacting proteins. To determine the effect of the PDZ-binding motif of CAR on transepithelial resistance in HAE, an adenoviral vector was generated containing the cDNA for CAR with a deletion of the last four amino acids (GSIV) that comprise the PDZ-binding motif (CAR^ΔGSIV). HAE were infected with Ad-CAR, -CAR^ΔGSIV, or -eGFP, and transepithelial resistance was measured (Fig. 2). As expected, CAR significantly increased airway resistance over baseline (data not shown) or control values (Ad-GFP, P=0.0005). Although expression of CAR^ΔGSIV significantly increased resistance (P=0.007 versus Ad-GFP), it failed to increase transepithelial resistance to the same degree as wild-type CAR (P=0.002). Similar results were obtained from HAE cultures derived from three separate donors. These data highlight an important role for the GSIV motif, intracellular interactions, and potentially membrane microdomain structure and stability in CAR-mediated adhesion.

To determine whether the PDZ-binding motif of CAR has any effect on adenovirus infection or cell growth, CAR-negative mouse L-cells were stably transfected with CAR, CAR^ΔGSIV, or eGFP and infected with Ad-β-gal (MOI=100). In agreement with previous studies that have shown that alterations to the cytoplasmic tail of CAR have no apparent effect on adenovirus infection (Wang and Bergelson, 1999; Walters et al., 2001), there was no difference between cells expressing CAR or CAR^ΔGSIV. However, cells expressing eGFP showed significantly less infection (Fig. 3A, P<0.001). These data indicate that a GSIV-PDZ domain interaction is not required for Ad infection.

Other recent studies have shown that overexpressing CAR in CAR-deficient cell lines or primary cancer cells is sufficient to reduce the cell growth rate of transformed cells (Okegawa et al., 2000; Okegawa et al., 2001). The growth of CAR, CAR^ΔGSIV, or eGFP stably transfected L-cells was monitored by the amount of protein or cell number increase over time (Fig. 3B,C). L-cells expressing CAR grew significantly slower than control eGFP cells (P<0.001). Interestingly, cells expressing CAR^ΔGSIV grew similarly to control eGFP cells. This difference in growth modulation is probably not due to either a difference in cell surface expression or localization, because Ad infection (Fig. 3A) and immunofluorescence (see Figs 4, 5) appears to be identical between CAR and CAR^ΔGSIV. Thus, this difference may be attributed to differences in intracellular interactions or microdomain complex formation.

The CAR PDZ-binding motif is a class I motif that could interact with many different PDZ-domain-containing proteins. Although others have found interactions between CAR and PDZ-domain-containing proteins (Cohen et al., 2001b; Sollerbrant et al., 2003), these proteins do not reside in the adherens junctions where most CAR expression is found in well-differentiated HAE (Walters et al., 1999). We took a candidate gene approach to determine what proteins CAR may be interacting with in HAE. Each candidate has been reported to interact with a class I PDZ-binding motif and all, with the exception of PSD-95, have been shown to be present at the junction adhesion complex in epithelia. Cells were transfected with CAR, CAR^ΔGSIV, MAGI-1b-GFP, PICK1-GFP, myc-LIN-7b, myc-GRIP1, HA-PIST, SAP97-GFP or PSD-95-GFP, or a combination of each of the PDZ-domain-containing proteins with CAR or CAR^ΔGSIV. After 2 days cells were analyzed for colocalization using immunofluorescence with confocal microscopy or co-IP/western blot. Results fell into three categories: (a) CAR and the PDZ-domain-containing protein resided in different compartments in the cell and did not interact (Fig. 4); (b) CAR and the PDZ-domain-containing protein resided in a similar compartment of the cell and did not interact (Fig. 5); or (c) A GSIV dependent interaction existed between CAR and the PDZ-domain-containing protein (Figs 6, 7).

We have previously shown by immunofluorescent labeling and confocal microscopy that CAR containing a FLAG epitope localizes primarily to the junctions between COS-7 cells (Ashbourne Excoffon et al., 2003). CAR^ΔGSIV localization is similar to CAR (Fig. 5A). It is thought that most interactions between PDZ-binding motifs and PDZ domains require additional signals, interactions, or localization to strengthen this defined interaction. Thus, not all proteins containing PDZ domains reported to bind a particular class of PDZ-binding motif sequences will bind all proteins containing that consensus motif. Currently, there are no guidelines to predict the interaction beyond the C-terminal peptide sequence. COS-7 cells were transfected as above with myc-LIN-7b, myc-GRIP1, or HA-PIST alone or in combination with CAR or CAR^ΔGSIV (data not shown) and analyzed with immunofluorescence (Fig. 4A,B) and co-IP/western blot (Fig. 4C). The overall localization of these proteins or CAR did not change significantly upon co-transfection (Fig. 4B). The small amount of colocalization at the membrane was no different with the expression of CAR or CAR^ΔGSIV and is probably due to being present in similar compartments without a direct interaction. Consistent with this, myc-LIN-7b, myc-GRIP1, or HA-PIST were successfully isolated by IP using a LIN7, GRIP1, or HA specific Ab respectively (Fig. 4C). However, none of these proteins were isolated after IP using the CAR specific RmcB Ab or a control Ab. This lack of interaction may be due to the specificity of the GSIV-domain interaction or could be a consequence of different compartmentalization in cells.

CAR is one of many proteins that localize at the junctions and membranes of cells. SAP97-GFP also showed a high degree of localization at the junctions of COS-7 cells (Fig. 5A). Both CAR and CAR^ΔGSIV colocalized with SAP97 at the cell-cell junctions (Fig. 5B). This suggests a GSIV-independent interaction or similar localization without a direct interaction. Consistent with these proteins residing in a similar domain but not interacting directly, CAR and SAP97-GFP did not co-IP (Fig. 5C). This indicates that similar localization is not sufficient to create an interaction between the PDZ-binding motif of CAR and a PDZ domain known to interact with class 1 PDZ-binding motifs.

In contrast to the junction localization of CAR and CAR^ΔGSIV, although a small amount of MAGI-1b-GFP appeared at the junctions between cells, the majority of MAGI-1b-GFP, PICK1-GFP and PSD-95-GFP expression was diffuse throughout the cell (Fig. 6A). When each of these candidate proteins was co-expressed with CAR, significant colocalization was observed (Fig. 6B,C,D). In CAR+MAGI-1b-GFP or CAR+PSD-95-GFP cells, CAR dramatically affected the localization of MAGI-1b-GFP or PSD-95-GFP by shifting a significant amount of the protein into the junctional region. Co-expression of CAR+PICK1-GFP produced colocalization at the membrane as well as the perinuclear region. COS-7 cells co-expressing CAR^ΔGSIV and MAGI-1b-GFP, PICK1-GFP, or PSD-95-GFP did not show an alteration of the original staining patterns (Fig. 6B-D).

Cells co-expressing CAR and MAGI-1b-GFP, PICK1-GFP, or PSD-95-GFP were also subjected to Triton X-100 lysis followed by IP and western blot (Fig. 7A). IP using a GFP Ab for the protein of interest resulted in a strong band corresponding to the GFP tagged protein. In agreement with the co-immunofluorescence, IP with the CAR specific mAb RmcB resulted in a co-IP between CAR and MAGI-1b-GFP, PICK1-GFP, or PSD-95-GFP. IP using a control Ab (rabbit IgG or myc) did not produce any corresponding band. RmcB is able to IP both the endogenous CAR expressed in COS-7 cells as well as the FLAG-tagged CAR. Thus, IP with RmcB for endogenous CAR from cells singly transfected with MAGI-1b-GFP or PSD-95-GFP resulted in a co-IP (data not shown). In addition, cells co-transfected with CAR^ΔGSIV and MAGI-1b-GFP or PSD-95-GFP also resulted in a co-IP apparently due to the endogenous CAR (data not shown). IP for the FLAG epitope on the N-terminus of CAR with FLAG Ab in lysates from cells co-transfected with MAGI-1b-GFP and CAR resulted in co-IP whereas CAR^ΔGSIV did not, confirming the specificity of the interaction (Fig. 7B). Interestingly, RmcB-mediated co-IP of PICK1-GFP was apparent only when co-transfected with CAR. To ensure that these interactions were not an artifact of GFP labeling, CAR was co-transfected with myc-MAGI-1b, myc-PSD-95, or GFP. Co-immunofluorescence and co-IP for MAGI-1b and PSD-95 were similar to the results shown while no co-IP or strong colocalization by immunofluorescence was observed for GFP (data not shown). The myc-MAGI-1b construct contained only the last four PDZ domains indicating that at least one of these domains interacts with CAR.

These data show that the interaction of the CAR PDZ-binding motif with PDZ-domain-containing proteins is specific, does not rely on localization but may rely on other motifs within the C-terminus of CAR or other specific interacting proteins.

Our data confirm previous reports that CAR is an important adhesion protein. Moreover, our results highlight the important role that the CAR PDZ-binding motif, GSIV, plays in the modulation of the endogenous role of CAR. As predicted by prior experiments, the GSIV motif plays no role in adenovirus infection. It appears that the C-terminus of CAR is inconsequential for its function as a viral receptor. By contrast, the GSIV motif can modulate the adhesion characteristics of primary airway epithelia and the growth rate of fibroblasts in culture. We did not measure the stability of CAR^ΔGSIV, however, we found similar levels of CAR and CAR^ΔGSIV by western blotting, immunocytochemistry and Ad infection. These suggest similar cell surface receptor levels and cellular localization. An interaction between CAR and PDZ-domain-containing proteins could explain its role in both adhesion and cell growth regulation. In many instances, PDZ-domain-containing proteins provide a network or skeleton for modulating several interactions. CAR may anchor these proteins to the plasma membrane allowing the recruitment of more actin or other stabilizing elements while at the same time provide a connection to proteins important for cell growth regulation, including factors such as β-catenin.

To investigate the molecular mechanism explaining the function of this GSIV motif we co-transfected several candidate PDZ-domain-containing proteins, known to interact with class 1 PDZ-binding motifs, with CAR or CAR deleted for the PDZ-binding motif. Interactions with MAGI-1b, PICK1 and PSD-95 were observed. Each of these proteins contains one or more PDZ domains as well as other protein-interacting motifs. Although we cannot currently exclude these alternative interacting sites, these interactions require the GSIV motif for an alteration in localization by immunofluorescence studies and for co-IP, strongly indicating a PDZ-binding motif-dependent interaction. The specificity of this interaction with some but not all candidate proteins also suggests that the GSIV motif is required but may not be sufficient. These interactions may involve other associations between the C-terminus of CAR and the protein, or other intermediary interactions such as specific lipid modifications [e.g. CAR and PSD-95 are both palmitylated proteins (Topinka and Bredt, 1998; van't Hof and Crystal, 2002)].

We provide evidence for interactions between CAR and three different proteins: MAGI-1, PICK1 and PSD-95. MAGI-1 is a member of the MAGUK family and is composed of six PDZ domains, two WW domains, and one guanylate kinase domain (Ide et al., 1999). Three isoforms have been described (MAGI-1a, 1b, 1c). Although all forms are present in epithelial cells, MAGI-1b has been localized to epithelial cell junctions (Dobrosotskaya and James, 2000). Interestingly, the sixth PDZ domain of MAGI-1 interacts with β-catenin, a protein that we have previously shown to co-IP with CAR (Walters et al., 2002). Additionally, MAGI-1b is particularly interesting because it interacts with viral proteins from adenovirus and it is associated with cellular transformation (Glaunsinger et al., 2000; Nguyen et al., 2003). Protein interacting with protein C kinase (PICK1) contains a single PDZ domain but can form dimers at a site distinct from the PDZ domain (Hirbec et al., 2002). Considering the interaction with PKC, it has been proposed that PICK1 is not only a substrate for PKC phosphorylation but also serves as an adaptor between transmembrane receptors and cytoplasmic PKC. PICK1 is found in epithelial cells and has been found to interact with several proteins such as the ERBB2/HER2 receptor (Jaulin-Bastard et al., 2001). Postsynaptic density 95/synapse-associated protein 90 (PSD-95/SAP-90 or PSD-95) is a neuronal-specific protein that plays an important role in the formation of the synaptic multiprotein complex (Ide et al., 1999; Mok et al., 2002). PSD-95 is a member of a family of proteins that includes SAP97/human Discs-large tumor suppressor gene (SAP97), which is found in epithelial cells. This more ubiquitious isoform interacts with ZO-1 at adherens junctions. CAR interacts with PSD-95 and causes redistribution of PSD-95 to the cell junctions in COS-7 cells. However, CAR colocalizes with SAP97 but does not interact with it. Both PSD-95 and CAR are palmitylated proteins (Topinka and Bredt, 1998; van't Hof and Crystal, 2002) and this lipid modification may assist the CAR-PSD-95 interaction. CAR is expressed in neurons and may interact with PSD-95 in this tissue (Hotta et al., 2003) where the activity of CAR may be relevant to the formation of neuronal networks. The significance of these interactions in vivo is currently under investigation.

In summary, the GSIV motif of CAR appears to play a significant role in the endogenous function of CAR as a cell adhesion protein and growth modulatory protein. It is likely that CAR interacts with several different proteins via this motif and the determination of the key interacting proteins will probably yield novel insights into the role of CAR in cell adhesion, signaling and cancer biology.

We thank Jamie Kesselring and Theresa Mayhew for assistance with manuscript preparation, Michael Welsh and Johannes Hell for discussions, Mike and Sandy Shasby for the L-cells, the Central Microscopy Research Facility, the Gene Transfer Vector Core, the Gene Transfer Morphology Core [supported by the NIDDK(DK54759) and the Cystic Fibrosis Foundation, ENGELH98S0], the Hybridoma/Tissue Culture Facility, the In Vitro Cell Models Core [supported by the National Heart, Lung and Blood Institute, the Cystic Fibrosis Foundation, and the National Institutes of Diabetes and Digestive and Kidney Diseases (DK54759)], the Iowa Statewide Organ Procurement. This work was supported by a PPG grant from the NIH (HL51670-10). KAE is supported by a fellowship from the Parker B. Francis Foundation (PBF).