Syndecan-2 Regulates Transforming Growth Factor- (cid:1) Signaling*

Transforming growth factor- (cid:1) (TGF- (cid:1) ) has multiple functions including increasing extracellular matrix deposition in fibrosis. It functions through a complex family of cell surface receptors that mediate downstream signaling. We report here that a transmembrane heparan sulfate proteoglycan, syndecan-2 (S2), can regulate TGF- (cid:1) signaling. S2 protein increased in the renal interstitium in diabetes and regulated TGF- (cid:1) -mediated increased matrix deposition in vitro . Transfection of renal papillary fibroblasts with S2 or a S2 construct that has a truncated cytoplasmic domain (S2 (cid:2) S) promoted TGF- (cid:1) binding and S2 core protein ectodomain directly bound TGF- (cid:1) . Transfection with S2 increased the amounts of type I and type II TGF- (cid:1) receptors (T (cid:1) RI and T (cid:1) RII), whereas S2 (cid:2) S was much less effective. In con-trast, S2 (cid:2) S dramatically increased the level of type III TGF- (cid:1) receptor (T (cid:1) RIII), betaglycan, whereas S2 re-sulted in a decrease. Syndecan-2 specifically co-immu-noprecipitated with betaglycan but not with T (cid:1) RI or T (cid:1) RII. This is a novel mechanism of control of TGF- (cid:1) action that

Transforming growth factor-␤ (TGF-␤) has multiple functions including increasing extracellular matrix deposition in fibrosis. It functions through a complex family of cell surface receptors that mediate downstream signaling. We report here that a transmembrane heparan sulfate proteoglycan, syndecan-2 (S2), can regulate TGF-␤ signaling. S2 protein increased in the renal interstitium in diabetes and regulated TGF-␤-mediated increased matrix deposition in vitro. Transfection of renal papillary fibroblasts with S2 or a S2 construct that has a truncated cytoplasmic domain (S2⌬S) promoted TGF-␤ binding and S2 core protein ectodomain directly bound TGF-␤. Transfection with S2 increased the amounts of type I and type II TGF-␤ receptors (T␤RI and T␤RII), whereas S2⌬S was much less effective. In contrast, S2⌬S dramatically increased the level of type III TGF-␤ receptor (T␤RIII), betaglycan, whereas S2 resulted in a decrease. Syndecan-2 specifically co-immunoprecipitated with betaglycan but not with T␤RI or T␤RII. This is a novel mechanism of control of TGF-␤ action that may be important in fibrosis.
Syndecan-2 (S2) is one of four mammalian members of a transmembrane proteoglycan family (14). Syndecans act as coreceptors for growth factor binding, cell-matrix interactions, and cell-cell interactions (15)(16)(17)(18). They have divergent ectodomains and highly homologous transmembrane domains, and their cytoplasmic domains have two regions of homology (C1 and C2) flanking V regions unique to each syndecan. S2 interacts with matrix proteins such as laminin (19) and fibronectin (20), and its cytoplasmic V region ( Fig. 1) controls matrix assembly at the cell surface (21). This may be due to oligomerization-dependent V region phosphorylation 2 (22). Cells expressing an S2 construct that is truncated in the V region (S2⌬S) lack matrix deposition (21,23). The V region of S2 also controls left-right asymmetry during Xenopus development through the actions of TGF-␤ family members (23), with parallel phosphorylation requirements (24). Interestingly, S2 has similar motifs to betaglycan ( Fig. 1) including the GAG attachment sites, a possible serine phosphorylation site (Ser-ser-ala-ala; SSAA) and a C-terminal PDZdomain binding motif that binds GIPC. Since both fibrosis and determination of left-right asymmetry involve TGF-␤ family members (5, 23), we investigated whether S2 and TGF-␤ act in concert in fibrosis.
Analysis of Fibronectin Matrix Deposition-Cells were treated with TGF-␤1 as above for 48 h, fixed, and stained with anti-fibronectin, followed by FITC-conjugated goat anti-rabbit IgG (1:50). Levels of matrix deposition were semiquantified using camera exposure time.
Cell Surface Biotinylation and Immunoprecipitation-Confluent cultures (T-75 flask) were washed with cold PBS three times, incubated with NHS-Sulfo-Biotin (2.5 mg) in PBS (5 ml, 30 min, 4°C), washed with cold PBS three times, and scraped into lysis buffer (5 ml). The cell number of parallel cultures was used to confirm equal loading. After centrifugation (10,000 rpm, 10 min, 4°C), the supernatant was incubated (1 h, 4°C) with protein-G-Sepharose beads preblocked with 5% fetal bovine serum. Beads were removed by centrifugation (400 ϫ g, 10 min, 4°C) and the supernatant incubated with fresh preblocked beads and 5 g/ml antibody (2 h, 4°C). Beads were washed three times with lysis buffer, three times with PBS, and bound proteins were eluted by boiling with SDS sample buffer (5 min), subjected to SDS-PAGE, and blotted with streptavidin-HRP and ECL. Membranes were stripped and blotted with antibodies against S2.

RESULTS
When renal sections from age-matched controls ( Fig. 2A, left  panel) were compared with those from type II diabetic patients ( Fig. 2A, second panel), S2 was increased in the tubulointerstitium. In contrast, despite some increase in the glomerulus, syndecan-4 (right two panels) was not dramatically increased in the tubulointerstitium. Similar results were obtained with streptozotocin-induced diabetic rats (data not shown). TGF-␤ increases S2 expression in several cell types (28,29). We, therefore, tested whether the increase in S2 was due to TGF-␤, a known mediator of fibrosis (5). However, S2 mRNA or protein levels were not increased ( Fig. 2B and data not shown) in RPF cells treated with TGF-␤ at concentrations sufficient to increase fibronectin matrix in these (Fig. 2C) and other cells (30 -32).

Syndecan-2 Is a Co-receptor for TGF-␤ 15716
We then tested whether S2 can mediate TGF-␤-induced increased matrix deposition. Immunofluorescent labeling indicated that overexpression of S2 in RPF increased fibronectin matrix deposition with more fibrils/cell and larger fibrils (arrows) than in cells transfected with vector only (Fig. 2C). This difference was not statistically significant in a semiquantitative assay using exposure time. Construct expression levels vary among cells, although immunoblotting (data not shown) indicated transfection with S2 or S2⌬S increased S2 levels 2.5-fold overall. In contrast, expression of S2⌬S decreased basal fibronectin deposition (Fig. 2C), compared with cells transfected with vector only (102 Ϯ 11 versus 76 Ϯ 6.0, p Ͻ 0.001), and several cells (arrowheads) lacked fibrils. This suggests that S2 cytoplasmic domain regulates matrix deposition in RPF cells, as it does in CHO-K1 cells (21). Differences in matrix deposition were exaggerated after TGF-␤ treatment. The percentage decrease of exposure time after TGF-␤ treatment parallels increased matrix deposition. Treatment of vector-only transfected cells with TGF-␤ (Fig. 2C) increased matrix deposition (18% Ϯ 9.7% decrease in exposure time). Increased matrix deposition in S2 cells was exaggerated (21% Ϯ 7.0%), but S2⌬S cells showed (Fig. 2C) less increase (14% Ϯ 9.1%).
These results suggest that S2 regulates TGF-␤-mediated matrix increase. Transfection of cells with either S2 or S2⌬S increased surface binding of TGF-␤ over that to cells transfected with empty vector (Fig. 3, A and B). Cleavage of HS chains reduced TGF-␤ binding (Fig. 3B), suggesting that some binding is through HS chains. Surprisingly, recombinant S2 ectodomain bound biotinylated TGF-␤1, and TGF-␤1 could capture S2 from RPF cell lysates (Fig. 3, C and D). Thus, the core protein of both S2 and S2⌬S binds TGF-␤, and binding is independent of the cytoplasmic domain.
Immunoblotting of total cell lysates demonstrated that S2⌬S slightly, and S2 dramatically, increased protein levels of T␤RI and T␤RII (Fig. 4, A, B, D, and E). The expression of T␤RIII, betaglycan, was increased in S2⌬S cells but decreased in S2 cells (Fig. 4, C and F). When cell surface receptor levels were monitored by biotinylation followed by immunoprecipitation (Fig. 5, A-C), these were consistent with the total levels. In Fig.  5C, the lowest species is coprecipitated S2, as confirmed in Fig. 5D.
The regulation of the expression of TGF-␤ receptors by S2 may be through formation of a complex between S2 and the receptor(s). Syndecan-2 was not detected in T␤RI or T␤RII immunoprecipitates (data not shown) but was detected in T␤RIII immunoprecipitates (Fig. 5D). Although S2⌬S cells expressed the highest levels of betaglycan (Fig. 4C), less S2 coprecipitated with betaglycan from S2⌬S cells than from vectoronly transfected cells, and that detected may be endogenous full-length S2. Thus, the cytoplasmic domain of S2 may be Syndecan-2 Is a Co-receptor for TGF-␤ 15717 needed for association with betaglycan. Low amounts of coprecipitated S2 were detected from S2 cells, probably due to the decreased level of betaglycan in S2 cells (Fig. 4C). To confirm differential association of betaglycan with S2⌬S and S2, further coprecipitation experiments were performed with increased loading on SDS-PAGE and longer time exposure on the film (Fig. 5E).

DISCUSSION
Syndecans can act as co-receptors for growth factors (16 -18) by delivery of ligands to their signaling receptors and the formation of a ternary complex (18,33). Most growth factor binding has been thought to be through the glycosaminoglycan chains (16 -18). Here, for the first time, we report that S2 can bind and regulate the signaling of TGF-␤, and the binding is through the core protein of the ectodomain, with the HS GAGs having only a minor role. The binding is independent of the cytoplasmic domain. However, the cytoplasmic domain of S2 is crucial in the regulation of TGF-␤ receptor levels and in association of S2 with betaglycan.
Total and surface levels of T␤RI and T␤RII were dramatically increased by S2 but not by S2⌬S. The cell surface level of receptors is balanced by rates of synthesis, cleavage, and internalization-dependent down-regulation and recycling. There are conflicting reports whether TGF-␤ up-or down-regulates its receptor expression (27,(34)(35)(36). Additionally, other S2 ligands, such as FGF2 and GM-CSF (granulocyte-macrophage colony-stimulating factor), can up-regulate the expression of T␤RI and T␤RII (37,38). S2 and S2⌬S may differentially regulate cell surface degradation of TGF-␤ receptors; T␤RIII can be cleaved at the cell surface (39), although cleavage of T␤RI and T␤RII was not detected (35). Finally, S2 and S2⌬S may differentially regulate endocytosis-dependent degradation and recycling of TGF-␤ receptors. The distinct effects of S2 and S2⌬S on betaglycan levels are intriguing. Effects on betaglycan (increased by S2⌬S but not by S2) may be explained by similarities between S2 and betaglycan cytoplasmic domains. Both contain a PDZ-domain binding motif (Fig. 1) that binds GIPC (9,40). GIPC binding up-regulates betaglycan levels on the cell surface by preventing proteosome degradation (9). If S2 competes for GIPC binding, less GIPC will be available for betaglycan in S2 cells but more in S2⌬S cells. All three receptors may be required for efficient receptor down-regulation (36). Decreased cell surface betaglycan in S2 cells may then reduce down-regulation of T␤RI and T␤RII. Whatever the mechanism, the finding that S2 can regulate the cell surface levels of TGF-␤ receptors adds a new level of complexity to TGF-␤ action.
Despite increased TGF-␤ binding and betaglycan level in S2⌬S cells, TGF-␤ did not increase matrix deposition in S2⌬S cells, suggestive of defective downstream signaling. Both betaglycan and S2 cytoplasmic domain contain a SSAA sequence (Fig. 1). Phosphorylation of betaglycan cytoplasmic domain is needed for downstream signaling (8), and, although the site has not been determined, this can be by T␤RII. The SSAA sequence in S2 can be phosphorylated if not oligomerized (22), and phosphorylation is needed for both matrix assembly 2 and the regulation of left-right asymmetry in Xenopus embryos (23,24). Whether S2 and betaglycan compete for the same kinase(s) is not known, but overexpression of S2 may promote oligomerization, as it does with syndecan-4 (41), and thus reduce its phosphorylation potential. Finally, we found that betaglycan coprecipitates with S2, and the cytoplasmic domain of S2 is needed for this association. Whether this is a direct interaction is now being determined. There may be two distinct signaling mech-anisms, involving either S2 or betaglycan, with coprecipitation due to common binding to TGF-␤. While much remains to be determined, it is clear that, as is the case with betaglycan (8), cytoplasmic domain truncation of S2 has no effect on TGF-␤ binding but reduces downstream effects. In summary, the data presented here demonstrate a mechanism of TGF-␤ regulation by syndecan-2. Since TGF-␤ is a major fibrogenic agent, this highlights a possible role of syndecan-2 in fibrosis.