Syndecan-4 Associates with α-Actinin*

Cell adhesion to the extracellular matrix influences many cellular functions. The integrin family of matrix receptors plays major roles in the formation of adhesions, but other proteins modulate integrin signaling. Syndecan-4, a transmembrane proteoglycan, cooperatively signals with integrins during the formation of focal adhesions. To date, a direct link between syndecan-4 and the cytoskeleton has remained elusive. We now demonstrate by Triton X-100 extraction immunoprecipitation and in vitro binding assays that the focal adhesion component α-actinin interacts with syndecan-4 in a β-integrin-independent manner.

In addition to binding soluble extracellular ligands such as growth factors (21,22), syndecans interact with the ECM, including fibronectin (20 -26). When cells are seeded onto coverslips coated with the 120-kDa proteolytic fragment of fibronectin (the cell-binding domain), cells attach and spread but do not form FAs and SFs (1,14). Focal adhesion formation requires additional signaling; ligation of syndecan-4 with the HepII domain of fibronectin (1,14,19) or recombinant peptides derived from HepII (30) or clustering syndecan-4 with antibodies (18) can circumvent the need for intact fibronectin to form FAs and SFs, implicating syndecan-4 in the process of FA and SF formation. The direct activation of protein kinase C (PKC) with phorbol esters in cells spread on the cell binding domain of fibronectin also bypasses the need for full-length fibronectin in the formation of FAs and SFs (31). The V region of syndecan-4 binds PKC␣ and potentiates its enzymatic activity (32,33). Furthermore, overexpression of syndecan-4 increases FA and SF formation (34,35), whereas expression of syndecan-4 truncated within its V region prevents their formation (34).
Although it is clear that syndecan-4 plays an important role in FA and SF formation, a direct role for the molecule in cytoskeletal organization has remained elusive. Here, we present novel evidence that syndecan-4 is linked to the microfilament cytoskeleton by association with the microfilament bundling protein ␣-actinin. Originally characterized in muscle cells, ␣-actinin cross-links actin stress fibers in both muscle and non-muscle cells (36 -38). ␣-Actinin is 100 kDa in size, and it contains a globular actin binding domain attached to a rod domain consisting of four spectrin-like repeats (36,37). ␣-Actinin associates with ␤ integrins (9), vinculin (39), zyxin (40), the cysteine-rich protein (41), and palladin (42) within FAs, and potential signaling roles for the molecule have emerged with the reports of interactions with phosphatidylinositol 4,5bisphosphate (PI(4,5)P 2 ) (43), MEKK1 (44), actinin-associated LIM proteins (45), and focal adhesion kinase (46).

EXPERIMENTAL PROCEDURES
Materials and Antibodies-All general chemicals were purchased from Sigma. Monoclonal antibodies used include those against ␣-acti-* This study was supported by National Institutes of Health Grant GM50194 (to A. W.) and by Sankyo, Co., Ltd. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Cells and Culture Conditions-Rat embryo fibroblast (REF) cells and human foreskin fibroblast cells were cultured in ␣-minimal essential medium (Mediatech; Fisher) containing 5% fetal bovine serum (Atlanta Biologicals; Norcross, GA, or Summit Biotechnology, Ft. Collins, CO) in 10% CO 2 . Cells were used before passage 20 and were shown to be free of mycoplasma by labeling with Hoechst 33258. All cell culture equipment was from Fisher. For immunostaining, cells were seeded onto 12-mm-diameter coverslips; for extractions and immunoprecipitations, cells were cultured in 25-, 75-, or 150-cm 2 culture flasks. For some experiments, cells were allowed to adhere to substrates coated with fibronectin (Collaborative Biomedical Products, Bedford, MA) at either 100 g/ml for 1 h at room temperature or 50 g/ml for 2-3 h in serum-free medium. No differences were seen in the results between the two coating concentrations.
Immunostaining-For staining with 150.9 antibody against syndecan-4, cells were fixed in 100% methanol at Ϫ20°C for 20 min and blocked with 2% whole goat serum (ICN/Cappel; Aurora, OH) for 45 min. Cells stained with all other antibodies were fixed for 10 min at 37°C with 3.5% paraformaldehyde. Nonextracted cells were permeabilized with 0.1% Tween 20 (Bio-Rad) after fixation. Primary antibodies were incubated with fixed cells (1:50 in PBS) for 45 min at 37°C or overnight at 4°C. After washing with PBS, fluorescein isothiocyanate-or Texas Red-conjugated goat-anti-mouse or anti-rabbit antibodies (1: 50, Organon Tekenika Corp., Durham, NC) were added for 45 min at 37°C. Preparations were examined on a Nikon Optiphot microscope equipped for epifluorescence, and images were recorded on Ilford HP-5 film. Co-localization studies were performed using a combination of 150.9 and rabbit anti-␣-actinin. Co-localization was analyzed on a Leitz Orthoplan microscope equipped for epifluorescence, and digital images were captured with a Vario-Orthomat II (High Resolution Imaging Facility, University of Alabama at Birmingham).
Immunoprecipitation-Co-immunoprecipitation studies were performed on confluent monolayers of REFs (ϳ10 7 cells/immunoprecipitation). Cells were scraped into lysis buffer containing 1% TX100 and the protease inhibitors leupeptin, phenylmethylsulfonyl fluoride, and benzamidine in PBS, incubated on ice for 30 min, then cleared by centrifugation (1000 ϫ g). Before immunoprecipitation, lysates were incubated with rabbit anti-mouse antibodies (1:5000; Dako, Glostrup, Denmark) followed by protein-A-Sepharose beads (Amersham Biosciences) preblocked in 10% fetal bovine serum to pre-clear the lysate. Immunoprecipitations were performed by sequentially incubating lysates with primary antibody followed by rabbit-anti-mouse antibodies and fresh preblocked protein-A-Sepharose beads. Beads were washed 4 times in lysis buffer, once with 1 M NaCl, and 3 times with PBS. Some cells were lysed in radioimmunoprecipitation assay (RIPA) buffer containing 1% TX100, 1% deoxycholate, and 0.2% SDS. Co-immunoprecipitated proteins were eluted by boiling in SDS sample buffer for 5 min and subjected to SDS-PAGE and immunoblotting.
Binding Studies-Binding studies with synthetic peptides (Synpep, Dublin, CA) coupled to Sulfolink Coupling Gel (Pierce) were performed in 1 ml of REF cells lysed in 1% TX100 immunoprecipitation lysis buffer at 4°C for 30 min. Binding assays using 0.5 g of chicken gizzard ␣-actinin (Sigma) were performed at room temperature in 0.01% TX100 in PBS. Bound material was eluted with SDS sample buffer and analyzed by immunoblotting. In competition experiments, V region peptides (50) were added to a final concentration of 50 g/ml to cells immediately after lysis and were present throughout.

Syndecan-2 and Syndecan-4 Have Distinct Cell Surface Distributions and Susceptibilities to TX100 Extraction-Synde-
can-2 and syndecan-4 are involved in cell-ECM communication; syndecan-2 modulates the assembly of the ECM (47), and syndecan-4 regulates FA and SF formation (19,34,35). When REF cells are examined by indirect immunofluorescence microscopy, syndecan-2 labeling revealed a punctate pattern on the cell body (Fig. 1A). In contrast, syndecan-4 staining is restricted to FAs and the membrane overlying SFs (Fig. 1B). The amount of labeling over stress fibers is variable, dependent on cell type, and also dependent on whether the syndecan-4 antibody is directed against the cytoplasmic domain (29) or the extracellular domain (19,34,56). Syndecan-2 and syndecan-4 also show differential susceptibility to TX100 extraction. After TX100 extraction of live cells, syndecan-2 labeling is lost (Fig. 1C), but most syndecan-4 remains (Fig. 1D). These results confirm that syndecan-4 resides in the TX100-resistant cytoskeleton and matrix residue, whereas syndecan-2 does not.
Syndecan-4 Association with ␣-Actinin Is Integrin ␤ 1 -independent-Previous studies indicated that integrins ␤ 1 and ␤ 3 associate with ␣-actinin (9), and syndecan-4 co-localizes with both integrins in nonextracted cells (29), leading to the possibility that the association between syndecan-4 and ␣-actinin requires the presence of integrins. Cells grown on fibronectin substrates in the absence of serum contain ␤ 1 integrins in their FAs (53). Human foreskin fibroblast cells were grown on fibronectin-coated coverslips and extracted to determine whether ␤ 1 integrin was resistant to TX100 treatment. Labeling for ␤ 1 integrin is seen in FAs in unextracted cells (Fig. 5A), but this is lost after TX100 treatment (Fig. 5B). This was confirmed by immunoblotting with anti-␤ 1 integrin antibodies (Fig. 5C); Western blotting of REF cell lysate reveals two bands, probably mature glycanated and immature nonglycanated forms of ␤ 1 integrin (54), both of which are lost after TX100 treatment. Similar results were seen for ␤ 3 integrin, which is the integrin found most prominently in focal adhesions when cells were grown in serum (53) (data not shown). A ␤ integrin-independent association between syndecan-4 and ␣-actinin was confirmed by preserved co-immunoprecipitation of ␣-actinin with syndecan-4 in TX100-resistant fractions that lack ␤ integrins (Fig. 5D). ␣-Actinin co-immunoprecipitates with ␤ 1 integrin in REF cells lysed in CHAPS buffer (51). However, we are unable to detect ␤ 1 integrin in ␣-actinin immunoprecipitates under more stringent conditions using TX100 buffer (data not shown), confirming that an association between syndecan-4 and ␣-actinin does not require the presence of integrin ␤ subunits.
Syndecan-4 Synthetic Peptides Are Able to Capture ␣-Actinin-To investigate the interaction between syndecan-4 and ␣-actinin, we utilized in vitro binding assays similar to those used to characterize associations between ␣-actinin and integrins (9), zyxin (40), vinculin (39), and the cysteine-rich protein (41). Synthetic peptides mimicking the cytoplasmic tails of syndecan-4 and syndecan-2 (Fig. 6A) were coupled to beads. Immunoblotting of material bound to beads coated with syndecan-2 or syndecan-4 cytoplasmic domain peptides reveals that ␣-actinin is captured specifically by syndecan-4 from cell lysates (Fig. 6B). ␣-Actinin from a chicken gizzard preparation is also captured specifically by syndecan-4 cytoplasmic domain (Fig. 6C). Because both syndecan-4 (33) and ␣-actinin (43) associate with PI(4,5)P 2 , we attempted to determine whether this chicken gizzard preparation of ␣-actinin was already FIG. 4. Syndecan-4 and ␣-actinin co-immunoprecipitate. The detergent present in the lysis buffer as well as whether each lane is a sample of beads used to pre-clear (PC) or immunoprecipitate (IP) is listed above each blot (WB). A, paxillin co-immunoprecipitates with syndecan-4 in TX100 but not RIPA buffer. B, ␣-actinin (␣-A) co-immunoprecipitates with syndecan-4 (S4) in both TX100 and RIPA buffer, but vinculin is not detected. C, syndecan-4 co-immunoprecipitates with ␣-actinin in TX100 and in RIPA buffer. Samples in A and B were boiled in reducing SDS sample buffer, whereas those in C were boiled in nonreducing buffer. PI(4,5)P 2 -bound, possibly contributing to the interaction. Anti-PI(4,5)P 2 antibodies were used to detect the presence of PI(4,5)P 2 in Western blots of REF lysate and chicken gizzard ␣-actinin (Fig. 6D). Immunodetection reveals that the ␣-actinin preparation is not bound to PI(4,5)P 2 ; however, immunoblotting a REF cell lysate or ␣-actinin that had been preincubated with PI(4,5)P 2 reveals a 100-kDa band that corresponds to ␣-actinin.
Syndecan-4 Association with ␣-Actinin Can Occur through the V Region-To map the site of interaction between syndecan-4 and ␣-actinin, binding assays were performed with peptides encompassing the cytoplasmic domains of syndecan-2 (Cys2L) or -4 (Cys4L), the V region of syndecan-4 (Cys4V), or a truncated syndecan-4 that lacks the PDZ domain binding FYA (21-27) tripeptide (Cys4E). Cell lysates were incubated with syndecan cytoplasmic domain-coated beads, and bound material was immunoblotted with ␣-actinin antibodies (Fig. 7B). Syndecan-2 cytoplasmic domain, which contains homologous C1 and C2 domains, fails to capture ␣-actinin, although all syndecan-4 synthetic peptides bind ␣-actinin. This suggests the syndecan-4 V region plays a significant role in the interaction of ␣-actinin, but the FYA region is not required. To verify the importance of the V region with respect to ␣-actinin binding, competition assays with mutated V region peptides were used (Fig. 7C). REF cells were lysed in the presence of competing peptide, Cys4L beads were added to the lysates, and bound material was examined for the presence of ␣-actinin. Native V region peptides (Cys4V) and V region peptides harboring a proline to alanine mutation (4VPA) or a tyrosine to phenylalanine mutation (4VYF) competed for ␣-actinin binding. How-ever, a V region peptide with a scrambled sequence (4Vscr) and a peptide containing two lysine to arginine mutations (4VKR) failed to compete.
PKC␣ and ␣-Actinin Translocation after PMA Treatment-PKC␣ binds to the V region of syndecan-4 (33) and co-precipitates with syndecan-4 from cells pretreated with PMA to activate PKC (32). Our data here demonstrate that ␣-actinin also associates with the V region of syndecan-4, suggesting that competition for binding to syndecan-4 may occur. Localization of PKC␣ and ␣-actinin was monitored after PKC activation by PMA. Triton X-100-insoluble fractions Ϯ PMA were analyzed by Western blotting with antibodies against PKC␣, ␣-actinin, syndecan-4, and actin. When compared with Me 2 SO-treated control cells, pretreating cells with PMA decreases the amount of ␣-actinin in TX100-resistant fractions (Fig. 8A) and causes the translocation of PKC␣ to TX100-resistant preparations (55) (Fig. 8B). However, the amount of syndecan-4 present does not appear to be affected by PMA treatment (Fig. 8C). These differences are not due to unequal protein loading, as revealed by immunoblotting by actin antibodies (Fig. 8D). DISCUSSION The role of syndecans in the regulation of cell morphology and cytoskeletal organization has been addressed in previous studies (18,34,35,56). Although these studies implicated a link to the microfilament cytoskeleton, the nature of an association to the cytoskeleton had not been determined. Here, we report the association of the transmembrane heparan sulfate proteoglycan syndecan-4 with ␣-actinin. These components co- localize in cells, show similar resistance to TX100 extraction, co-immunoprecipitate, and interact in in vitro binding assays. It should be noted that both ␣-actinin and syndecan-4 are retained after TX100 extraction at 37°C, which appears to be more stringent than at 4°C where lipid rafts (57), paxillin, talin, and vinculin (data not shown) are retained in TX100resistant fractions. The specific association between syndecan-4 and ␣-actinin was confirmed by co-immunoprecipitation since other FA components co-precipitate with syndecan-4 in milder conditions (in TX100), but only ␣-actinin co-immunoprecipitates in RIPA buffer. This was not dependent on the presence of integrin ␤ 1 or ␤ 3 , because the interaction between syndecan-4 and ␣-actinin is preserved after the removal of ␤ integrins, and unlike ␣-actinin/␤ 1 integrin interaction (9), is retained after washes with 1 M NaCl.
The interaction with ␣-actinin was specific for syndecan-4. Syndecan-2 and syndecan-4 have distinct subcellular distributions and differential susceptibilities to TX100 extraction, indicating different intracellular binding partners. Syndecan-2, through its C1 region, interacts with ezrin (58), a member of the ERM family of proteins that mediate cytoskeleton-cell membrane associations (59), but this does not result in resistance to extraction. The cytoplasmic tails of syndecan-2 and -4 share a COOH-terminal FYA motif that interacts with several PDZ domain proteins (60 -62) whose overexpression can alter cellular morphology. However, our studies indicate this motif does not mediate SF and FA formation. When full-length syndecan-4 is overexpressed in Chinese hamster ovary cells (34) or REF cells, 2 spreading and FA and SF formation are promoted.
A similar effect is observed in cells overexpressing the FYA deletion. 2 Deletion of this sequence in 4E peptides does not prevent ␣-actinin association in in vitro binding assays. The V region of syndecan-4 appears to control FA and SF formation, because a syndecan-4 construct that is truncated within this region acts as a dominant negative for spreading and FA formation (34). The V region of syndecan-4 binds and superactivates PKC␣ (32,33), and PKC activity is required for FA and SF formation (31,63,64). However, our in vitro binding assays indicate that ␣-actinin also binds the V region, and this may also contribute to the lack of FAs and SFs.
Synthetic peptides encompassing the V region of syndecan-4 can compete for binding of ␣-actinin to beads coated with the entire cytoplasmic domain. Peptides where the sequence of 4V was scrambled (4Vscr) or where two lysines were replaced with arginines (4VKR) did not compete for binding, whereas 4VPA and 4VYF did. Previous studies demonstrated that 4VPA and 4VYF also lack the ability to activate PKC␣, possibly because of a reduced ability to form oligomers (50). This indicates there are similarities between syndecan-4 binding to ␣-actinin and PKC␣. However, neither 4VPA nor 4VYF activated PKC␣, although they do compete for ␣-actinin binding. These data imply that the sites of interaction within syndecan-4 cytoplasmic domain V region for ␣-actinin and PKC␣ have some similarities but also some differences. This suggests that there may be some competition for binding, and further studies are under way. Interestingly. PMA-mediated PKC activation causes the translocation of PKC␣ to detergent-resistant preparations with Previous studies have indicated a possible interaction between ␣-actinin and syndecan-4. Both FA components can also be present in the myosin sheath found in some cells (65), both components are up-regulated in proliferative renal disease (56), and both components move coordinately into FA when cells spread on the cell binding domain of fibronectin are treated with soluble HepII domain that binds syndecan-4 (19). Embryonic fibroblasts from syndecan-4-deficient mice lack the ability to form FAs in response to HepII treatment (66), and a preliminary analysis of the distribution of FA and SF components indicates their cytoskeletal organization is abnormal, particularly with respect to SF organization and ␣-actinin distribution. In addition, several signaling molecules associate with ␣-actinin, including PI(4,5)P 2 (43), MEKK1 (44), and focal adhesion kinase (46), and it will be interesting to see how the interaction with syndecan-4 affects these associations.