Regulation of Inositol Phospholipid Binding and Signaling through Syndecan-4*

Syndecan-4 is a transmembrane heparan sulfate proteoglycan that can regulate cell-matrix interactions and is enriched in focal adhesions. Its cytoplasmic domain contains a central region unlike that of any other vertebrate or invertebrate syndecan core protein with a cationic motif that binds inositol phospholipids. In turn, lipid binding stabilizes the syndecan in oligomeric form, with subsequent binding and activation of protein kinase C. The specificity of phospholipid binding and its potential regulation are investigated here. Highest affinity of the syndecan-4 cytoplasmic domain was seen with phosphatidylinositol 4,5-bisphosphate (PtdIns(4,5P)2) and phosphatidylinositol 4-phosphate, and both promoted syndecan-4 oligomerization. Affinity was much reduced for 3-phosphorylated inositides while no binding of diacylglycerol was detected. Syndecan-2 cytoplasmic domain had negligible affinity for any lipid examined. Inositol hexakisphosphate, but not inositol tetrakisphosphate, also had high affinity for the syndecan-4 cytoplasmic domain and could compete effectively with PtdIns(4,5)P2. Since inositol hexaphosphate binding to syndecan-4 does not promote oligomer formation, it is a potential down-regulator of syndecan-4 signaling. Similarly, phosphorylation of serine 183 in syndecan-4 cytoplasmic domain reduced PtdIns(4,5)P2 binding affinity by over 100-fold, although interaction could still be detected by nuclear magnetic resonance spectroscopy. Only protein kinase Cα was up-regulated in activity by the combination of syndecan-4 and PtdIns(4,5)P2, with all other isoforms tested showing minimal response. This is consistent with the codistribution of syndecan-4 with the α isoform of protein kinase C in focal adhesions.

One cell surface heparan sulfate proteoglycan, syndecan-4, also binds PtdIns(4,5)P 2 through a motif in the central portion of its cytoplasmic domain, known as the V (variable) region. The V region of syndecan-4 is unlike that of any other family member and has the sequence LGKKPIYKKA (18,19). The two pairs of lysine residues appear to be critical for this activity, and the motif KKXXXKK is known to bind inositol polyphosphate based on previous studies, for example of synaptotagmin II (20). The functional consequences of inositol phospholipid binding at this site appears to be the stabilization of a dimer of syndecan-4 cytoplasmic domain in an unusual twisted clamp motif, determined by nuclear magnetic resonance (NMR) spectroscopy (21,22). Dimers, or more probably, higher order oligomers (23), then can bind protein kinase C␣ (PKC␣), and cause it to be strongly activated (18,23,24). This may explain why syndecan-4 and PKC␣ are both focal adhesion components in many cell types (25,26). Our current hypothesis is that syndecan-4, when clustered, signals through the kinase at nascent focal adhesions and contributes to focal adhesion assembly, with possible involvement of the G protein RhoA (19,(23)(24)(25)27).
One unresolved issue is the regulation of the syndecan-4, PtdIns(4,5)P 2 , and PKC␣ signaling complex. There are several possibilities. First, all syndecans can interact with PDZ domain proteins through a C-terminal FYA sequence (19,28). Such interactions could lead to complex disassembly or internalization or to stabilization. This has not been examined. Second, Horowitz et al. (24,29,30) showed that phosphorylation of the single serine residue at the boundary of the membrane proximal C1 region and V region of syndecan-4 cytoplasmic domain (Ser 183 ) can lead to decreased signaling through PKC␣. In part this may be due to decreased affinity of PtdIns(4,5)P 2 for the phosphorylated syndecan-4 cytoplasmic domain (24). A third, alternate, hypothesis is that PtdIns(4,5)P 2 may be displaced by another compound, yielding a form of syndecan-4 cytoplasmic domain unable to bind or activate PKC␣. One possibility is InsP 6 . The KKXXXKK motif within syndecan-4 V region provides a potential site for InsP 6 binding (31), although we showed previously that this inositol phosphate neither promotes syndecan-4 cytoplasmic domain oligomerization nor activates PKC␣ (18).
Here we have examined these hypotheses by analyzing the specificity of phospholipid and inositol phosphate binding to syndecan-4 cytoplasmic domain peptides, whether InsP 6 can compete with PtdIns(4,5)P 2 for binding to the cytoplasmic domain of syndecan-4 core protein and the role of Ser 183 phosphorylation. Both InsP 6 and phosphorylation may down-regulate syndecan-4-mediated signaling, since both strongly affect PtdIns(4,5)P 2 interactions. In addition, InsP 6 diminished microfilament bundle formation in fibroblasts, under experimental conditions where syndecan-4 and PKC␣ are involved.

EXPERIMENTAL PROCEDURES
Materials-Syndecan-2 and Ϫ4 peptides corresponding to the entire cytoplasmic domains of human syndecan-4 (4L) and syndecan-2 (2L) were synthesized by SynPep (Dublin, CA), and their sequences confirmed by mass spectroscopy. These sequences were (C)RMKKKDEG-SYDLGKKPIYKKAPTNEFYA and RMRKKDEGSYDLGERKPSSAA-YQKAPTKEFYA, respectively. Two modified 4L peptides were also synthesized, one lacking the three C-terminal amino acids (FYA; denoted 4⌬E) and a second incorporating a phosphoserine at position 183 in place of serine (p-4L). Also used were peptides corresponding to the central, variable (V) region of syndecan-4 and Ϫ2 with the sequences (C)LGKKPIYKK and (C)LGERKPSSAAYQ, respectively. At least two different batches of each peptide were used. All inositol phospholipids and diacylglycerols were purchased from Biomol (Plymouth Meeting, PA). PtdIns(4,5)P 2 was also purchased from Avanti Polar Lipids (Alabaster, AL), as were phosphatidylethanolamine, phosphatidylserine, and phosphatidylcholine. InsP 6 , InsP 4 , InsP 3 , and inositol hexasulfate (InsS 6 ), HEPES, CHAPS, and one batch of Ptd-Ins(4,5)P 2 were from Sigma-Aldrich.
Lipid Binding Assay-Lipids were dissolved in a chloroform/methanol solution at 2 mg/ml as described previously (29). Solubilized lipids were dried under N 2 and sonicated for 2 min in ice-cold H 2 O at a final concentration of 1 mg/ml. Peptides were incubated on ice for 30 min with the indicated concentrations of lipid in 10 mM Tris-HCl (pH 7.5) in a reaction volume of 100 l. The samples were spun in filter units (Ultrafree-MC, 30,000 NMWL, Millipore, Bedford, MA) at 2000 ϫ g for 70 s following the method described by Haarer et al. (32). The filtrates (40 l of each in 4% SDS, 10% glycerol, 1.5% dithiothreitol, 0.004% bromphenol blue, 50 mM Tris-HCl, pH 6.8) were resolved by SDS-PAGE on 16.5% Tris-Tricine gels, stained with Coomassie Brilliant Blue G-250 (BioRad). The stained peptides were scanned on a BioRad GS670 imaging densitometer and quantitated. All experiments were performed at least in triplicate. Bound and free phospholipid were then calculated and averaged from replicates. The data were used to generate Scatchard plots, from which affinities were estimated. In addition, two forms of displacement experiments were performed. Peptides were first incubated with differing concentrations of inositol phosphate or sulfate for 30 min, then with 50 M inositol phospholipid for an additional 30 min, or PtdIns(4,5)P 2 was incubated first with the peptide, then the inositol phosphate or sulfate was added.
Photoaffinity Labeling-The preparation and use of [ 3 H]benzoyldihydrocinnamide (BZDC)-InsP 4 and -InsP 6 probes has been described previously (33). Experiments were performed in 96-well plates. To 21 l of buffer, consisting of 25 mM Tris-HCl, pH 7.4, 1 mM EDTA, and 1 mM dibasic potassium phosphate, was added 1 g of peptide in 5 l of buffer and 3 l of water, with or without unlabeled InsP 4 or InsP 6 , as appropriate, to a final concentration of 20 M. After a 10-min incubation on ice, 1 l of photoactivable probe (0.5 Ci, 20 -70 nM) was added. The plate was incubated for 60 min on ice before exposure to UV irradiation. The reaction was stopped by the addition of 4ϫ SDS-PAGE sample buffer. Samples were resolved by 20% SDS-PAGE at ϳ30 mA, with prestained low molecular mass standards (Invitrogen) in one or two lanes. The gels were impregnated with Entensify (DuPont), dried and fluorographs exposed for 1 week. In other experiments, increasing concentrations of unlabeled InsP 6 were included in the assays, in order to ascertain the K i as an approximate measure of affinity of the inositol phosphate for the V region of syndecan-4.
Size Exclusion Chromatography-Gel filtration procedures were as previously (22). Synthetic peptides or a mixture of synthetic peptide and phosphoinositide were loaded onto a Sephadex G-50 gel filtration column (0.7 ϫ 50 cm) equilibrated with 50 mM HEPES (pH 7.3), 0.1% CHAPS, and 150 mM NaCl. Peptides were eluted with the same buffer at a flow rate of 3 ml/h at room temperature, and 1-ml fractions were assayed on a UV monitor at 280 nm. The column was calibrated with molecular standards containing thyroglobulin (670 kDa), bovine ␥-globulin (158 kDa), chicken ovalbumin (44 kDa), equine myoglobin (17 kDa), and vitamin B 12 (1.3 kDa).
Cell Adhesion Studies-Primary rat embryo fibroblasts were seeded onto fibronectin-coated glass coverslips in 24-well plates as previously (34) for 30 min. At this time InsP 4 , InsP 6 , or InsS 6 were added to the serum-free medium at various concentrations for up to an additional 2.5 h. In some cases, after 1.5 h, adherent cells were returned to normal growth medium, as a control to ascertain that adhesion inhibition was reversible. For interference reflection microscopy analysis of focal adhesion formation (35), cultures were fixed in 3% glutaraldehyde in phosphate-buffered saline for 15 min and washed and mounted in serum-free medium. Other cultures were fixed in 4% paraformaldehyde in phosphate-buffered saline, containing 0.1% Triton X-100. Texas Redconjugated phalloidin (Molecular Probes, Eugene, OR) staining of microfilaments was performed as before (35). The percentage of cells with focal adhesions or microfilament bundles was counted in three separate
replicates, with at least 100 cells per coverslip counted in each case. In further studies, cells were seeded on the integrin-binding 110-kDa fragment of plasma fibronectin (  bated with, or without, the addition of PtdIns(4,5)P 2 and syndecan-4L cytoplasmic domain peptide, using identical methods as previously (18,23). Phosphorylated histone IIIS was resolved by 20% SDS-PAGE and autoradiography. Quantitation was by laser scanning densitometry as before (18).
Nuclear Magnetic Resonance Spectroscopy-Phosphoinositide titration experiments were performed on a Bruker DRX500 spectrometer in quadrature detection mode, equipped with a triple resonance probe head with triple-axis gradient coils. A series of one-dimensional NMR spectra were recorded for 4L, p-4L, and 2L peptides with different concentrations of PtdIns(4,5)P 2 . All data were collected at 25°C, and the strong solvent resonance was suppressed by water-gated pulse sequence combined with pulsed-field gradient (PFG) pulses. All NMR data were processed using Bruker XWIN-NMR (Bruker Instruments) software on an SGI Indigo 2 work station. The proton chemical shifts were referenced to internal sodium 4,4-dimethyl-4-silapentane 1-sulfonate (DSS). Fig. 1 shows the sequences of syndecans-2 and Ϫ4 with their constant (C1 and C2) and variable (V) regions denoted. Lipid micelles were allowed to bind to peptides corresponding either to the full-length cytoplasmic domain of syndecan-4 or syndecan-2, and the unbound peptide recovered and quantified after centrifugation through a molecular mass cut-off filter. Affinities of the lipids for the peptides are shown in Table I. There was a significant preference of syndecan-4 cytoplasmic domain for PtdIns(4,5)P 2 and PtdIns(4)P over inositol phospholipids con-taining phosphate at the 3-position on the inositol ring. Phosphatidylserine had weak affinity for syndecan-4, while no binding to phosphatidylethanolamine or two forms of diacylglycerol was noted. Since diacylglycerol is a cleavage product of PtdIns(4,5)P 2 generated by phospholipase C, the data suggest that interactions of the inositol phosphate moiety of PtdIns(4,5)P 2 with syndecan-4 are key. Very little binding of any lipid to syndecan-2 cytoplasmic domain was seen, confirming that only the central V region of syndecan-4 contains the lipid binding site(s), since the C1 and C2 regions of these two core proteins are highly homologous. This was confirmed with a synthetic peptide corresponding to the V region of syndecan-4 cytoplasmic domains, which had similar affinity to 4L for PtdIns(4,5)P 2 (not shown).

Specificity of Phospholipid Binding to Syndecan-4 -
The almost identical affinities of PtdIns(4,5)P 2 and PtdIns(4)P for the cytoplasmic domain of syndecan-4 (4L) raised the question of whether both could stabilize the peptide in dimeric conformation. Fig. 2 shows that both phospholipids markedly promoted 4L dimer formation, which is only limited in their absence. Syndecan-2 cytoplasmic domain (2L) showed no tendency to form dimers (Fig. 2), even in the presence of PtdIns (4)P or PtdIns(4,5)P 2 (not shown).

Modification of Syndecan-4 Cytoplasmic Domain and Its
Influence on PtdIns(4,5)P 2 Binding-Phosphorylation of the single Ser 183 residue in syndecan-4 cytoplasmic domain has been reported to decrease phospholipid binding and subsequent PKC␣ activity (24). A synthetic peptide corresponding to the entire cytoplasmic domain of syndecan-4, but incorporating phosphorylated Ser 183 had markedly lower affinity for PtdIns(4,5)P 2 than the unphosphorylated peptide (570 M compared with 5 M; Table I). Moreover, this decrease was not accompanied by any change in preference for phospholipid interactions. There was still no detectable binding to D3 inositol phospholipids, and the affinity of the 4L peptides for PtdIns (4)P was also reduced to undetectable levels (Table I).
One-dimensional NMR spectroscopy was used as a further sensitive indicator of syndecan cytoplasmic domain interactions with PtdIns(4,5)P 2 (Fig. 3). The spectra demonstrated that each of the three peptides, 4L, p-4L and 2L, have different characteristics with respect to PtdIns(4,5)P 2 binding. Most resonances for 4L and p-4L were changed and broadened upon inositide titration, indicative of oligomerization in the presence of the phospholipid. Therefore, even though the membrane filter assay showed a much decreased affinity of p-4L for PtdIns(4,5)P 2 , an interaction was still clearly detectable by NMR spectroscopy. However, the spectra of 4L and p-4L were distinct, which indicate differences in oligomer organization. Consistent with results of gel chromatography and lipid binding assays, the spectra of 2L peptide were unchanged with increasing inositide, indicative of no detectable interaction (Fig. 3C).
Published data suggest that the C2 region of syndecan-4 cytoplasmic domain is flexible and may not participate in interactions with inositides (22). This region, and particularly its terminal FYA motif, does interact with PDZ domain proteins (28). These may stabilize the C2 structure in situ. Deletion of the FYA sequence did not alter in vitro binding affinity for PtdIns(4,5)P 2 ( Table I).
Interactions of Inositol Phosphates with Syndecan Cytoplasmic Domains-Peptides corresponding to the V region and entire cytoplasmic domains of syndecan-2 and Ϫ4 were exposed to [ 3 H]BZDC-InsP 4 or [ 3 H]BZDC-InsP 6 . Fig. 4 shows the structure of these compounds. Covalent cross-linking was achieved by UV irradiation, with subsequent SDS-PAGE, fluorography, and quantitation. Control experiments contained an ϳ400-fold excess of unlabeled InsP 4 or InsP 6  reduced by 58 -61% by excess unlabeled compound. Consistent with the fact that the syndecan-4 V region bound the InsP 6 probe, no specific binding of syndecan-2 peptides (with their dissimilar V region; Fig. 1) to either probe was detected (not shown).
To gauge the affinity of InsP 6 for syndecan-4 cytoplasmic domain, 50 nM [ 3 H]BZDC-InsP 6 was incubated with increasing concentrations of unlabeled InsP 6 in the presence of 5 g of 4V peptide. The data show (Fig. 5B) the IC 50 was ϳ4.5 M. However, it should be emphasized that the radiolabeled compound is more hydrophobic than the competitor, and so the true affinity of InsP 6 itself may be greater.
InsP 6 Is an Effective Competitor of PtdIns(4,5)P 2 Binding to Syndecan-4 Cytoplasmic Domain-As a second measure of InsP 6 affinity for syndecan-4 cytoplasmic domain, two assays were performed. In the first, the ability of PtdIns(4,5)P 2 to displace bound InsP 6 from the 4L peptide, was measured. Inositol polyphosphate was partially displaced (Fig. 6A) by Ptdins(4,5)P 2 (just 5% of added PtdIns(4,5)P 2 bound to peptide that was preincubated with 80 M InsP 6 , whereas over 50% bound in the absence of InsP 6 ). In the second assay, the ability of InsP 6 to displace already bound PtdIns(4,5)P 2 was calculated. The higher affinity of InsP 6 for syndecan-4 cytoplasmic domain was again suggested, since the inositol phospholipid was readily displaced in a dose-dependent manner ( Fig. 6B; PtdIns(4,5)P 2 binding was reduced to 30% by 80 M InsP 6 ). In a control experiment for that shown in Fig. 6A, InsP 3 was used in place of InsP 6 (Fig. 6C). Here, even preincubation of peptide with high quantities of InsP 3 did not reduce the amount of PtdIns(4,5)P 2 that subsequently bound. In another control experiment, inositol hexasulfate was used in place of InsP 6 (Fig.  6D). Preincubation with sulfated inositol was less efficient than with InsP 6 in reducing PtdIns(4,5)P 2 binding to the syndecan-4 cytoplasmic peptide, but more efficient than InsP 3 . This suggests that ionic interactions are largely responsible for interactions between InsP 6 and the syndecan-4 cytoplasmic domain and that this interaction is of higher affinity than that between PtdIns(4,5)P 2 and the 4L peptide.
Inositol Hexaphosphate Inhibits Cell Adhesion-Since InsP 6 at levels consistent with those found in vivo (ranging as high as 15-60 M;37) could compete with PtdIns(4,5)P 2 for binding to the syndecan-4 cytoplasmic domain, cell adhesion studies were performed. Syndecan-4 and PtdIns(4,5)P 2 , together with PKC, have roles in focal adhesion formation in fibroblasts seeded on fibronectin substrates (13,19,25,27). Since InsP 6 , when bound to the core protein of syndecan-4, does not facilitate PKC␣ signaling, it was ascertained whether exogenous inositol phosphates influenced fibroblast adhesion. Table II and Fig. 7 show that while cell spreading was not affected, focal adhesion and microfilament bundle assembly were strongly inhibited by InsP 6 . This effect was not seen with InsP 4 , while partial inhibition was seen with InsS 6 . The effects of both InsP 6 and InsS 6 were reversible, as focal adhesion formation was completed within 1 h of returning cells to normal culture medium. A further test of InsP 6 was in a focal adhesion formation assay. Cells spread on the integrin-binding 110-kDa fragment of plasma fibronectin spread but do not form focal adhesions. This last stage can be stimulated by the 31-kDa HepII domain of fibronectin, or by antibodies against the core protein that cluster syndecan-4 (reviewed in Ref. 19). However, treatment of fibroblasts with 0.1 mM InsP 6, after 1.5 h of spreading on the 110-kDa integrin-binding domain of fibronectin, prevented a response to either the recombinant fibronectin HepII domain or clustering antibody against the N terminus of syndecan-4 (Fig.  8). Cells remained spread but did not form prominent microfilament bundles that terminate in focal adhesions. Equivalent treatment with InsS 6 had much reduced inhibitory effects (not shown).
FIG. 6. InsP 6 is a competitor for PtdIns(4,5)P 2 binding to syndecan-4 cytoplasmic domain. A, C, and D, syndecan 4L peptide was exposed to varying concentrations of InsP 6 (A), InsP 3 (C), or InsS 6 (D), then to 50 M PtdIns(4,5)P 2 . In B the peptide was exposed first to PtdIns(4,5)P 2 , then to InsP 6 . The amount of PtdIns(4,5)P 2 bound was calculated from a membrane filter separation of peptide that was either bound or unbound to lipid.   Specificity of PtdIns(4,5)P 2 /Syndecan-4-mediated PKC Activation-Many isoforms of PKC have a lipid binding capability, but only PKC␣ has been shown to co-distribute with syndecan-4. A number of recombinant PKC isoforms were tested to determine whether their activity could also be up-regulated by a combination of PtdIns(4,5)P 2 and syndecan-4 cytoplasmic domain. As shown in Fig. 9, only PKC␣ was markedly activated, but several other conventional (␤I, ␤II, and ␥) and novel isoforms (␦ and ⑀) could be modestly activated. As expected from a lack of inositide or diacylglycerol binding site, the atypical PKC isoform was not affected by the presence of PtdIns(4,5)P2 and syndecan-4 cytoplasmic domain. DISCUSSION Syndecan-4 is alone among the family of cell surface heparan sulfate proteoglycans in having a cationic motif in its cytoplasmic V region that can bind inositol phospholipids and PKC␣ (18,19,(21)(22)(23)(24). We have shown previously by NMR spectroscopy that a dimer of V region is stabilized by one molecule of PtdIns(4,5)P 2 , and the cytoplasmic domain forms a unique "twisted clamp" structure that may be relevant to its function (21). Syndecan-4 has a role in focal adhesion formation, and evidence also suggests that PKC activity is required both for focal adhesion formation and syndecan-4 insertion into forming adhesions (25,26,38). Here we show specificity in terms of inositol phospholipid binding, with a distinct preference for PtdIns(4,5)P 2 over alternately phosphorylated inositol phospholipids. In particular, D3 forms had low, or no measurable affinity for syndecan-4 cytoplasmic domain peptides. This is unlike the situation with some proteins that bind inositol phos-pholipids through pleckstrin homology domains, where either D3 or D4 lipids may bind (11,16). Our preliminary two-dimensional NMR data provide a possible explanation; the 3-hydroxyl group of PtdIns(4,5)P 2 may interact with a residue of syndecan-4 core protein. 2 In turn, this would be disrupted by 3-phosphorylation. This raises the possibility that PI 3-kinase activity could down-regulate syndecan-4 signaling if the PtdIns(4,5)P 2 bound to its cytoplasmic domain is susceptible to 3-phosphorylation. In this context, PI 3-kinase activity has been implicated in focal adhesion disassembly (39), and actin polymerization related to membrane protrusion can be controlled by PtdIns(3,4,5)P 3 (11).
Previously we showed that InsP 6 could not stimulate PKC␣ activity in the presence of syndecan-4 cytoplasmic domain (18) since it did not promote 4L oligomerization (23) that is required for PKC activation. However, as we show here, the 4V peptide does bind to InsP 6 with a K i of ϳ4.5 M. This is consistent with the presence of a KKXXXKK motif, shown in synaptotagmin C2B domain to be a binding site for inositol polyphosphate (20,31). Since the amount of ligand in these assays was very low, we can assume that the K i may approximate to the affinity of InsP 6 for syndecan-4. This is approximately the same as the measured affinity of PtdIns(4,5)P2 for this core protein. Two different displacement assays indicated, however, that the affinity of InsP 6 for syndecan-4 cytoplasmic domain is higher than that of the inositol phospholipid. Importantly, the assays for inositol phospholipid and inositol polyphosphate binding were different and cannot be strictly compared. Moreover, measurement of InsP 6 affinity utilized unlabeled compound to compete with a BZDC-InsP 6 derivative. The latter is more hydrophobic, and this may have led to a requirement for higher concentrations of InsP 6 to compete. In turn, the true affinity of InsP 6 may be higher than that calculated here, consistent with the fact that InsP 6 could compete with PtdIns(4,5)P 2 for the syndecan-4 cytoplasmic domain. The less charged variant, inositol hexasulfate, was accordingly less efficient as a competitor of PtdIns(4,5)P 2 .
Proposed roles for InsP 6 in mammalian cells include regulation of vesicular traffic, calcium influx, and tumor cell proliferation (7)(8)(9). It has been suggested that inositol polyphosphates may compete with phosphoinositides in the regulation of phospholipase C␦ (40), AP-2, Bruton's tyrosine kinase, synaptotagmin II (20), and AMP deaminase (reviewed in Ref. 41). Levels of InsP 6 in some cells may reach 15-60 M (37), and it is the most abundant inositol polyphosphate in most vertebrate cells. At these levels it could compete with PtdInds(4,5)P 2 for syndecan-4 cytoplasmic domain. Experiments here showed that exogenous InsP 6 does not block fibroblast attachment or spreading on fibronectin substrata, but focal adhesion and microfilament bundle formation were severely limited. These  A and B) do not form microfilament bundles or focal adhesions, but these are stimulated by fibronectin HepII domain (C) or clustering antibodies binding to the N terminus of syndecan-4 core protein (E). In both cases, 0.1 mM InsP 6 blocks these events (D and F, respectively). Original magnification, ϫ60.

FIG. 9. Protein kinase C␣ is activated by a combination of syndecan-4 cytoplasmic domain and PtdIns(4,5)P 2 .
Recombinant isoforms of PKC were assayed for phosphorylation of histone IIIS substrate in the presence or absence of syndecan 4L peptide and PtdIns(4,5)P 2 . All isoforms except (which had almost no activity) showed low activity while only ␣ was strongly up-regulated. data are entirely consistent with the early stages of adhesion being integrin-dependent, but the late stages requiring syndecan-4 (19,27). Moreover, InsP 6 could act as an inhibitor in an assay where antibody clustering of syndecan-4 was used to stimulate microfilament bundle formation in cells prespread on the 110-kDa integrin-binding domain of fibronectin. Similarly, InsP 6 also blocked microfilament bundle formation promoted by recombinant HepII domain of fibronectin. InsP 4 had no impact, consistent with in vitro data showing limited binding of this inositol polyphosphate to the syndecan-4 core protein. It cannot be ruled out that InsP 6 competes not only for PtdIns(4,5)P 2 binding to syndecan-4, but other inositide-mediated processes. However, the ability of InsP 6 to affect stages of cell adhesion suspected to have a syndecan-4 dependence is a striking result, and these effects on the cytoskeleton have not been noted previously.
One effect of InsP 6 binding to syndecan-4 may be to limit oligomerization. Syndecan-4 cytoplasmic domain peptides do not oligomerize in response to InsP 6 , and correspondingly, do not activate PKC␣ (23). Evidence points to a critical role of PtdIns(4,5)P 2 in promoting, or stabilizing, oligomers of syndecan-4 that can then interact with PKC␣ and strongly activate it ( Fig. 8 and Refs. 19 and 25). Protein binding specificity for InsP 6 over InsP 4 has been reported previously, for example with myelin proteolipid protein (42), and, in some cases, InsP 4binding correlates with an ability to interact with PtdIns(3,4,5)P 3 , rather than phosphatidylinositol bisphosphates (43). Similarly, rabphilin 3A, a regulator of vesicle traffic and linkage to the actin cytoskeleton, binds PtdIns(4,5P) 2 (44), but does not have high affinity for InsP 4 (31). Consistent with this, we have found little binding of syndecan-4 cytoplasmic domain to PtdIns(3,4,5)P 3 , or InsP 4 .
Syndecan-4 cytoplasmic domain of 3T3 cells can be serinephosphorylated in response to serum-free conditions, and reversed by a heparin-binding growth factor (29). The kinase involved was thought not to be PKC␣, but perhaps a novel isoform of PKC, such as PKC␦ or PKC⑀ (30). As previously suggested by Horowitz et al. (24), phosphorylation of Ser 183 markedly decreased the binding of PtdIns(4,5)P 2 in our assays and abrogated binding of PtdIns (4)P. Nor was there increased binding of other phosphoinositides. NMR spectroscopy showed that despite decreased affinity, p-4L was subject to oligomerization in the presence of PtdIns(4,5)P 2 . Since the spectra generated by 4L and p-4L were distinct in these experiments, it suggests that the structures of the 4L and p-4L dimers differ from each other.
The ability of the syndecan-4 cytoplasmic domain to bind and activate PKC isoforms seems to be restricted to PKC␣ alone. The up-regulation of novel PKC isoforms is modest by comparison, and this effect was almost completely due to the inositol phospholipid, rather than an impact of the syndecan-4 peptide (not shown). Consistent with this is the observation that PKC, an atypical isoform lacking a diacylglycerol and phorbol ester binding site, was unaffected by the addition of PtdIns(4,5)P 2 and syndecan 4L peptide. If PKC␦ or PKC⑀ is responsible for phosphorylation of syndecan-4 cytoplasmic domain, it is presumably regulated in a manner distinct from that shown here for PKC␣.
Syndecan-4 regulates focal adhesion formation (19,(25)(26)(27). Over-expression of full-length syndecan-4 in CHO cells promotes increased focal adhesion formation, with concomitantly enhanced microfilament bundle assembly (36). In contrast, transfection of constructs encoding syndecan-4 truncated to delete the inositol phospholipid and PKC-binding sites act as dominant negatives. These decrease both spreading and microfilament bundle assembly, together with a reduction in the number and size of focal adhesions (36). It has been shown that PtdIns(4,5)P 2 is essential for focal adhesion assembly (13), and our data suggest that one function may revolve around the capacity of syndecan-4 to signal through PKC␣. Syndecan-4 may, therefore, be a nidus for cytoskeleton-membrane association and stabilization.