A Peptide Derived from Tenascin-C Induces β1 Integrin Activation through Syndecan-4*

Tenascin-C (TN-C) is unique for its cell adhesion modulatory function. We have shown that TNIIIA2, a synthetic 22-mer peptide derived from TN-C, stimulated β1 integrin-mediated cell adhesion of nonadherent and adherent cell types, by inducing activation of β1 integrin. The active site of TNIIIA2 appeared cryptic in the TN-C molecule but was exposed by MMP-2 processing of TN-C. The following results suggest that cell surface heparan sulfate (HS) proteoglycan (HSPG), including syndecan-4, participated in TNIIIA2-induced β1 integrin activation: 1) TNIIIA2 bound to cell surface HSPG via its HS chains, as examined by photoaffinity labeling; 2) heparitinase I treatment of cells abrogated β1 integrin activation induced by TNIIIA2; 3) syndecan-4 was isolated by affinity chromatography using TNIIIA2-immobilized beads; 4) small interfering RNA-based down-regulation of syndecan-4 expression reduced TNIIIA2-induced β1 integrin activation, and consequent cell adhesion to fibronectin; 5) overexpression of syndecan-4 core protein enhanced TNIIIA2-induced activation of β1 integrin. However, treatments that targeted the cytoplasmic region of syndecan-4, including ectopic expression of its mutant truncated with the cytoplasmic domains and treatment with protein kinase Cα inhibitor Gö6976, did not influence the TNIIIA2 activity. These results suggest that a TNIIIA2-related matricryptic site of the TN-C molecule, exposed by MMP-2 processing, may have bound to syndecan-4 via its HS chains and then induced conformational change in β1 integrin necessary for its functional activation. A lateral interaction of β1 integrin with the extracellular region of the syndecan-4 molecule may be involved in this conformation change.

Tenascin-C (TN-C) is unique for its cell adhesion modulatory function. We have shown that TNIIIA2, a synthetic 22-mer peptide derived from TN-C, stimulated ␤1 integrinmediated cell adhesion of nonadherent and adherent cell types, by inducing activation of ␤1 integrin. The active site of TNIIIA2 appeared cryptic in the TN-C molecule but was exposed by MMP-2 processing of TN-C. The following results suggest that cell surface heparan sulfate (HS) proteoglycan (HSPG), including syndecan-4, participated in TNIIIA2-induced ␤1 integrin activation: 1) TNIIIA2 bound to cell surface HSPG via its HS chains, as examined by photoaffinity labeling; 2) heparitinase I treatment of cells abrogated ␤1 integrin activation induced by TNIIIA2; 3) syndecan-4 was isolated by affinity chromatography using TNIIIA2-immobilized beads; 4) small interfering RNA-based down-regulation of syndecan-4 expression reduced TNIIIA2-induced ␤1 integrin activation, and consequent cell adhesion to fibronectin; 5) overexpression of syndecan-4 core protein enhanced TNIIIA2-induced activation of ␤1 integrin. However, treatments that targeted the cytoplasmic region of syndecan-4, including ectopic expression of its mutant truncated with the cytoplasmic domains and treatment with protein kinase C␣ inhibitor Gö6976, did not influence the TNIIIA2 activity. These results suggest that a TNIIIA2-related matricryptic site of the TN-C molecule, exposed by MMP-2 processing, may have bound to syndecan-4 via its HS chains and then induced conformational change in ␤1 integrin necessary for its functional activation. A lateral interaction of ␤1 integrin with the extracellular region of the syndecan-4 molecule may be involved in this conformation change.
Tenascin (TN)-C 2 is one of the most intriguing extracellular matrix (ECM) proteins (1)(2)(3). TN-C is expressed predominantly during embryogenesis, wound healing, and neoplastic processes, in which alternative mRNA splicing within the fibronectin (FN) type III-like (FN-III) repeats can generate different TN-C isoforms (4). Multifunctional properties have been identified for TN-C, including effects on cell adhesion, migration, proliferation, survival, and differentiation. The effects of TN-C on cell adhesion are particularly complex; the TN-C substrate supports attachment of some cell types, but is nonadhesive or even repulsive for other cell types (5)(6)(7). Based on these antipodal effects on cell adhesion, TN-C is multifunctional and is therefore classified as an adhesion modulatory ECM protein, a so-called "matricellular" protein (8). Various TN-C molecule domains, especially FN-III repeats, including the alternative splicing domains, have been implicated in its function as a matricellular protein. However, their contributions to the adhesion modulatory effects of TN-C are not completely understood.
Interactions of cells with the ECM are largely mediated by members of the integrin superfamily of adhesive receptors. The most unique feature of integrins is their ability to alter ligand binding and signaling activities. Because integrin-mediated cell-ECM interactions play key roles in maintaining normal cellular functions, affinity changes in integrins are critical for the anchorage-dependent cellular processes, such as growth, survival, migration, and differentiation. Integrin activation is considered to be regulated by "inside-out" signals from the cell interior, that is, those triggered by extracellular stimuli. However, the molecular pathways leading to inside-out activation of integrins are still poorly understood.
There have been many studies indicating that ECM proteins such as FNs, laminins, and collagens have biologically active cryptic sites. These matricryptic sites are exposed through bio-logical processes such as proteolytic cleavage and conformational change in response to multimerization of ECM proteins, binding to other molecules, or cell-mediated mechanical forces (9,10). Davis et al. (11) coined the term "matricryptins" to describe biologically active proteolytic fragments of ECM proteins. Matricryptic sites and matricryptins of ECM proteins have been implicated in a variety of events governed by cell-ECM interactions.
We have previously found that FN has both cell adhesion sites and a matricryptic site opposing cell adhesion. The 22-mer matricryptin, FNIII14, derived from the 14th FN-III repeat, strongly suppresses ␤1 integrin-mediated cell adhesion to FN, whose activity depends on its C-terminal amino acid sequence, YTIYVIAL (12). This cryptic antiadhesive site is exposed by either FN degradation with MMP-2, or FN interaction with heparin (13). As a negative modulator of cell-ECM interaction, FNIII14 influences physiological cellular processes such as survival (14) and differentiation (15), as well as pathological events such as tumor metastasis (16).
There are several sequences similar to the YTIYVIAL sequence of FN in other ECM proteins including TN-C. Two analogous sequences, YTITIRGV and YTIYLNGD, are present in the FN-III repeat A2 of the alternative splicing region and the C terminus fibrinogen-globe, respectively, of the human TN-C molecule (supplemental Fig. S1). These analogous sequences may be involved in TN-C cell adhesion modulatory activity. We have investigated the effects of synthetic TN-C peptides containing these analogous sequences on cell adhesion to FN. Surprisingly, TNIIIA2, a 22-mer TN-C peptide containing YTITIRGV, stimulated cell adhesion to FN by inducing conformational and functional activation of ␤1 integrin. The active site of TNIIIA2 appears cryptic but was exposed by MMP-2 processing. Some of our results suggest that cell surface heparan sulfate (HS) proteoglycans (HSPG), including syndecan-4, participated in ␤1 integrin activation in response to TNIIIA2, without requiring its cytoplasmic domains. Thus, a TNIIIA2-related matricryptic site of TN-C molecule, exposed by MMP-2 processing, may induce a lateral interaction of ␤1 integrin with the syndecan-4 ectodomain, resulting in conformational change in ␤1 integrin necessary for its functional activation. The cell adhesion-modulatory activity of TN-C, at least in part, may be due to the TNIIIA2-related matricryptic site/matricryptin.
Peptides-The GRGDSP was purchased from IWAKI. Peptides derived from human FN (FNIII14: TEATITGL EPGTEYTIYVIAL), TN-C (TNIIIA2), their mutants (supplemental Fig. S1 and supplemental Table S1), and CS-1 (LHPGEILDVPST) were synthesized using the solid phase strategy combined with the Boc and Fmoc chemistry, in which a Cys was added to the C terminus of each peptide to increase their activity by dimerization and to facilitate coupling to SGbeads. For a photoaffinity cross-linking of a putative cell surface receptor of TNIIIA2, a photosensitive derivative of TNIIIA2, RSTDLPGLKAAT-p-benzoyl-phenylalanine (Bpa)-YTITIR-GVK (biotin) C (biotinylated TNIIIA2-Bpa), which has been shown to retain the TNIIIA2 activity (17), was synthesized using standard solid state methodology (18). These synthetic peptides were purified by reversed-phase HPLC and characterized by mass spectrometry.
Confocal Microscopy-Cells were seeded on FN-coated cover glasses, incubated with or without TNIIIA2 for 1 h, fixed with 4% paraformaldehyde and permeabilized with 0.03% Triton X-100. For immunostaining of ␤1 integrins, the cells were not permeabilized. The cells were stained with anti-vinculin Ab (Simga) and rhodamine-phalloidin (Invitrogen), or an anti-integrin ␤1 mAb (AG89 and HUTS-4), and mounted on slides. Confocal microscope images were obtained with an inverted microscope (Fluoview FV1000, Olympus) fitted with a ϫ60 UPlanSApo oil-immersion objective (NA 1.35) and Fluoview software.
Cell Adhesion Assay-Cell adhesion assay was performed as described previously (19). Ramos cell adhesion to HUVEC was performed as reported previously (16).

Recombinant TN-C Protein and Its Processing with
Proteinases-A recombinant protein containing the FN-III repeats A1-4 of human TN-C (rA1-4) was prepared as described previously (20). Briefly, cDNA encoding the region was generated by PCR using human fetal brain Marathon-Ready cDNA (BD Biosciences Clontech) as the template and primers consisting of 5Ј-AGTGGATCCACTGAACAAGC-CCCTGAGC-3Ј and 5Ј-CCCAAGCTTGGGCAGTTCGT-TCAGCACCAGAGA-3Ј. rA1-4 was processed in a dialysis bag (Sanko) at room temperature as follows. rA1-4 mixed with trypsin (1:200, w/w) was dialyzed first against PBS(Ϫ) for 6 -24 h, next against PBS(Ϫ) containing 1 mM diisopropyl fluorophosphate and then against the serum-free medium. rA1-4 protein mixed with MMP-2 (1:100, w/w) and 2 mM p-aminophenylmercuric acetate was dialyzed first against the buffer containing 2 mM CaCl 2 for 6 -24 h, next against the buffer containing MMP-2 inhibitor GM6001 (10 M) and then against the serum-free medium. As a control, rA1-4 protein was treated under the same conditions without MMP-2.
Photoaffinity Labeling-WI38VA13 cells (3 ϫ 10 5 ) spread on a 6-well plate were incubated with biotinylated TNIIIA2-Bpa (5 g/ml) at 37°C for 1 h in the presence or absence of heparin (20 g/ml) and then irradiated with UV (two 15-watt lamps, 365 nm) for 30 min on ice at a distance of 10 cm. Cells were dissolved with 1 ml of lysis buffer (10 mM Tris-HCl buffer (pH 7.4) containing 0.15 M NaCl, 1% Nonidet P-40 and 2 mM phenylmethylsulfonyl fluoride). Aliquots of the lysates were subjected to immunoblot analysis using avidin-conjugated peroxidase. Another aliquot of the lysate was mixed with avidin-conjugated beads to precipitate the biotinylated TNIIIA2-Bpa-linked molecules. After washing the beads extensively with the lysis buffer, the bound materials were eluted with Laemmli's buffer and then subjected to immunoblot analysis using anti-HS and anti-␤1 (DE9) mAbs.

Stimulation of Cell Adhesion to FN by a TN-C-derived
Peptide-We first examined whether synthetic TN-C peptides containing analogous sequences, YTITIRGV and YTIYLNGD in their C terminus, termed TNIIIA2 and TNfbg, respectively (supplemental Fig. S1), exerted any effect on cell adhesion to FN. WI38VA13 cells were seeded onto FN-coated plates, with or without one of the synthetic peptides. In the absence of these peptides, the cells began attaching to the FN substrate about 30 min after seeding, while many cells had not spread, even after 60 min (Fig. 1A). Focal contact did not appear in these attached cells (Fig. 1B). In the presence of TNIIIA2, most cells had attached completely within 60 min, and many had already spread on the FN (Fig. 1A). There was a concomitant appearance of focal contact and actin stress fiber reorganization (Fig.  1B). A control peptide, TNIIIA2scr, in which the YTITIRGV sequence is shuffled (RITYITGV), and TNfbg were inactive (Figs. 1A and 2A). TNIIIA2-stimulated cell adhesion in a dosedependent manner, and this was inhibited by function-blocking mAbs directed to integrin subunits ␣4, ␣5, and ␤1, but not ␤2 (Fig. 2, A and B).
We then examined the effect of TNIIIA2 on adhesion of nonadherent hematopoietic progenitor cell lines, K562 (which exclusively expresses ␣5␤1 as a ␤1 class integrin) and Ramos (which only expresses ␣4␤1 of the ␤1 integrins), whose integrins are known to be in their inactive states. K562 cells attached specifically to FN only in the presence of Mn 2ϩ , an integrin activator (Fig. 2C). TNIIIA2 was capable of inducing K562 cell attachment to FN without Mn 2ϩ , and this was blocked by mAbs anti-␣5 and ␤1, but not ␣4 (Fig. 2D). TNIIIA2  NOVEMBER 30, 2007 • VOLUME 282 • NUMBER 48 JOURNAL OF BIOLOGICAL CHEMISTRY 34931 also induced Ramos cell attachment to FN without Mn 2ϩ , and this was blocked by CS-1 and an anti-␣4 mAb, but not by anti-␣5 mAb (Fig. 2E). Ramos cells attached to HUVEC via ␣4␤1-VCAM-1 interaction, for which Ramos cells and HUVEC had to be pretreated with Mn 2ϩ and TNF-␣, respectively (Fig.  2F). TNIIIA2 induced Ramos cell attachment onto HUVEC without Mn 2ϩ , in an ␣4␤1-VCAM-1 interaction-specific manner (Fig. 2G). Thus, the effects of TNIIIA2 on nonadherent cell adhesion lead us to speculate that this peptide may induce activation of ␤1 integrin.

␤1 Integrin Activation by Tenascin-C Peptide
TNIIIA2 Induces Activation of ␤1 Integrin-The status of ␤1 integrin activation can be evaluated by its accessibility to anti-␤1 integrin mAbs, such as AG89 (22) and HUTS-4 (23), both of which recognize the active conformation-specific epitope of ␤1 subunit, regardless of the ␣ subunit. We performed flow cytometric analysis with those mAbs to investigate whether TNIIIA2 was able to influence the conformational status of ␤1 integrin (Fig. 3, A and B).
Unstimulated K562 cells showed only low accessibility to AG89 (Fig. 3A, panel a), while Mn 2ϩ stimulation slightly increased expression of the AG89 epitope (Fig. 3A, panel b). Addition of GRGDSP alone, a ligand of integrin ␣5␤1, also caused a slight increase in expression of the AG89 epitope (Fig.  3A, panel c), and GRGDSP together with Mn 2ϩ further increased the AG89 epitope expression (Fig. 3A, panel d).
TNIIIA2 induced a remarkable increase in expression of the AG89 epitope in a dose-dependent manner even without Mn 2ϩ (Fig. 3, A, panels e and f). Addition of GRGDSP together with TNIIIA2 caused a further but slight increase in expression of the AG89 epitope (Fig. 3A, panel g). TNIIIA2 treatment (at least within 3 h) did not change expression of the ␤1 integrin subunit under the experimental conditions used here, as confirmed by both flow cytometric and immunoblot analyses (data not shown). Expression of the active ␤1-specific epitope, in response to TNIIIA2, was similarly confirmed by using either another mAb (HUTS-4) (supplemental Fig. S2A) or another nonadherent cell line, Ramos (supplemental Fig. S2B). Conformational change in the ␤1 integrin was also investigated using an adherent cell type WI38VA13. The results, characterized by a dramatic conformational change in response to TNIIIA2, were almost the same as those observed with nonadherent cell types, confirmed by using AG89 (Fig. 3B) and HUTS-4 (data not shown). Thus, in both nonadherent and adherent cell types, expression of the AG89 and HUTS-4 epitopes in response to TNIIIA2 correlated well with the effects of this peptide on cell adhesion to FN or HUVEC, suggesting that TNIIIA2 stimulated ␤1 integrin-mediated cell adhesion by inducing ␤1 integrin activation.
We next performed immunofluorescence microscopic analysis of ␤1 integrin, by using the HUTS-4 and AG89 mAbs. Without stimulation, only a low level of ␤1 integrins expressed the AG89 epitope on K562 cells (Fig. 3C). When the cells were stimulated with TNIIIA2, the dots of punctate staining appeared intensely on the basal surfaces, with a concomitant increase in the number of K562 cells attached to the FN (Fig.  3C). On the other hand, expression of the HUTS-4 epitope was detected peripherally on WI38VA13 cells spreading on the FN (Fig. 3D, left), but became much more remarkable after incubation with TNIIIA2 (Fig. 3D, right). The results show that TNIIIA2 induces a net increase in the quantity of activated ␤1 integrin on cell surfaces.
Active Site of TNIIIA2 Is Cryptic in the TN-C Molecule-To characterize the active site of TNIIIA2 in the TN-C molecule, we first performed alanine-scanning mutagenesis to identify the amino acid residues in the sequence YTITIRGV, which was essential for the TMIIIA2 activity. Supplemental Table S1 presents the mutant peptides tested and summarizes the results of their abilities to induce ␤1 integrin activation, as evaluated by , and Ramos (E-G) cells suspended in serum-free medium (1 ϫ 10 4 /ml) in the presence of TNIIIA2 or TNIIIA2scr (50 g/ml) were seeded with or without Mn 2ϩ (1 mM), anti-integrin mAb (20 g/ml), or CS-1 peptide (200 g/ml) into a 96-well plates coated with the FN (0.5 g/ml in A and B; 5 g/ml in C-E). F and G, Ramos cells were seeded onto a HUVEC monolayer which was prepared on a 48-well plate and treated as indicated. HUVEC were pretreated with or without TNF-␣ (10 ng/ml) for 24 h (16). In all experiments, cells were allowed to adhere for 1 h, and the number of cells either attached (white bars) or spread (gray bars) was counted as described previously (12). Each point represents the mean Ϯ S.E. of triplicate determinations. One of three individual experiments is shown.

␤1 Integrin Activation by Tenascin-C Peptide
flow cytometric analysis using AG89 (data not shown). The TNIIIA2 activity was highly sensitive to single Ala replacement of the two Ile and Val residues at positions 16, 18, and 21, respectively, and the resultant peptides were inactive for ␤1 integrin activation. The replacement of Arg 19 with Ala caused partial reduction in TNIIIA2 activity, while alternative replace-ment of Arg 19 with Glu, resulted in complete loss of the activity, suggesting the necessity of a positive net charge, as well as specific amino acids (two Ile and Val) for ␤1 integrin activation in this peptide.
To confirm whether the active site of TNIIIA2 is exposed on the TN-C molecule, we examined the effects of a recombinant protein composed of the FN-III repeats A1-4 (rA1-4) (supplemental Fig. S1) on K562 cell attachment to the FN substrate. K562 cell attachment to FN was not affected by rA1-4, while processing of rA1-4 with MMP-2, but not trypsin, caused a significant increase in K562 cell attachment to FN, even without Mn 2ϩ (Fig. 4A). The MMP-2 digest of rA1-4 also weakly induced ␤1 integrin activation (Fig. 4B). Thus, the active site appears to be buried within the TN-C molecule, but exposed by processing with MMP-2.
Involvement of Syndecan-4 in Expression of TNIIIA2 Activity-Assuming the presence of a membrane receptor mediating TNIIIA2 activity, we attempted to detect a cell surface molecule, with specific binding affinity toward TNIIIA2, by affinity labeling. We did not detect such a molecule by using conventional cross-linking chemicals containing succinimide, carbodiimide, or azide groups (data not shown). These can covalently connect peptide with protein based on its reactivity against a   NOVEMBER 30, 2007 • VOLUME 282 • NUMBER 48 primary amine. However, we successfully detected the molecule by photoaffinity labeling, using a photosensitive TNIIIA2 derivative containing the benzoyl group, biotinylated TNIIIA2-Bpa, which strongly reacts with macromolecules without primary amines, such as sugars (24).

␤1 Integrin Activation by Tenascin-C Peptide
By photoaffinity labeling of WI38VA13 cells using this photoreactive probe, a high molecular mass band (Ͼ200 kDa) and two bands of 110 and 70 kDa, were detected with avidin-conjugated peroxidase (Fig. 5A). Among these, only the high molecular mass band resulted from specific binding with biotinylated TNIIIA2-Bpa (Fig. 5A, lane 2), because the 110-and 70-kDa bands were visualized nonspecifically, even in the absence of biotinylated TNIIIA2-Bpa (Fig. 5A, lane 1). The high molecular mass band tagged with biotinylated TNIIIA2-Bpa was precipitated with avidin-immobilized beads and analyzed by immunoblot analysis. A high molecular mass band was also visualized by anti-HS mAb (Fig. 5B, lanes 1 and 2). When the affinity labeling was done in the presence of heparin, the band disappeared (Fig.  5, lanes 3 of A and B). TNIIIA2 was shown to bind to heparin with relatively high affinity (K d ϭ 1.1 ϫ 10 Ϫ7 M), compared with chondroitin sulfate (K d ϭ 4.9 ϫ 10 Ϫ7 M) and hyaluronic acid (K d Ͼ 10 Ϫ5 ) (supplemental Fig. S3). Furthermore, ␤1 integrin activation in response to TNIIIA2 was abolished in the presence of heparin (supplemental Fig. S3). These results suggest that TNIIIA2 expresses its activity through specific binding to cell surface HSPG via its HS chain. Interestingly, ␤1 integrin was also detected within precipitates with avidin beads, but not within control precipitates, as a band migrating at a position different from that of high molecular mass HSPG (Fig. 5B, lanes  4 and 5). This suggests an induced association of ␤1 integrin with cell surface HSPG in response to TNIIIA2.
Affinity chromatography, using the TNIIIA2-immobilized SG beads, was performed to isolate HSPG which had specific binding affinity to TNIIIA2. Immunoblot analysis of the purified material using the anti-HS mAb, showed a diffuse band at around 200 kDa (Fig. 5C, lane 2). This purified HSPG(s) was treated with heparitinase I and chondroitinase ABC to identify it based on its core protein. Immunoblot analysis using an anti-⌬HS mAb (3G10) showed a major intense band of 30 kDa and several weak bands, including a high molecular mass band migrating near the original HSPG band (Fig. 5C, lane 3). The major band of 30 kDa was recognized by either of two different mAbs against syndecan-4 core protein (lanes 4 and 5), but not syndecan-1 (lane 6). This suggests that, among the cell surface HSPGs, syndecan-4 may have the greatest involvement in TNIIIA2-induced ␤1 integrin activation, although involvement of other types of HSPG cannot be excluded completely.
To verify the involvement of syndecan-4/HSPG in the expression of TNIIIA2 activity, we examined the effect of HS chain degradation on TNIIIA2-induced ␤1 integrin activation. Treatment of WI38VA13 cells with heparitinase I abrogated ␤1 integrin activation in response to TNIIIA2 (Fig. 6A). We then examined the effect of syndecan-4 knockdown by RNA interference (RNAi). Flow cytometric and immunoblot analyses showed that syndecan-4 expression on WI38VA13 cells was reduced by about 70% upon introduction of siRNA targeted against syndecan-4 core protein (Fig. 6B). In cells transfected with control siRNA, TNIIIA2 clearly induced ␤1 integrin acti-vation (Fig. 6C, panel b) and cell adhesion (Fig. 6D, panel b), while syndecan-4 knockdown resulted in a remarkable reduction of that activation (Fig. 6C, panel d), with a concomitant decrease in cell adhesion/spreading to FN (Fig. 6D, panel d).
These results suggest that the syndecan-4 core protein, and its extracellular HS chains are required for ␤1 integrin activation by TNIIIA2.  1 and 2) or presence (lane 3) of heparin (20 g/ml), as described under "Experimental Procedures." Cell lysates were subjected to immunoblot analysis using avidin-peroxidase to detect a biotinylated-TNIIIA2-Bpa-linked molecule(s). B, WI38VA13 cells tagged with biotinylated TNIIIA2-Bpa (Bi-TNIIIA2) under the indicated conditions were dissolved, precipitated with avidin-immobilized beads, and then subjected to immunoblot analysis using anti-HS antibody (lanes 1-3) or anti-␤1 integrin mAb (DE9) (lanes 4 and 5). C, sample purified by affinity chromatography using the TNIIIA2-immobilized SG-beads was subjected to immunoblot analysis using normal mouse IgM

␤1 Integrin Activation by Tenascin-C Peptide
The cytoplasmic region of syndecan-4 and its effector enzyme PKC␣ play an indispensable role in the syndecan-4based signaling for focal contact and actin stress-fiber formation (25). We overexpressed syndecan-4 core protein or its mutant (S4⌬R), with complete deletion of the cytoplasmic domain (Fig. 7A). Overexpression of wild type syndecan-4 caused a significant increase in ␤1 integrin activation by TNIIIA2 (Fig. 7B, panel b). Unexpectedly, overexpression of S4⌬R did not reduce, but rather increased, ␤1 integrin activation in response to TNIIIA2 (Fig. 7B, panel c). Additionally, treating the cells with a specific PKC␣ inhibitor Gö6976 did not influence TNIIIA2-induced ␤1 integrin activation (Fig. 7C). These results suggest that activation f ␤1 integrin by TNIIIA2 may be dependent on cell surface HSPG including syndecan-4, but independent of its cytoplasmic region.

DISCUSSION
TN-C plays multiple roles in tissue regulation as a matricellular ECM protein, especially in certain pathophysiological situations such as embryonic development, inflammation, and tumorigenesis. A number of studies using isolated recombinant domains of the TN-C molecule show that regions with adhesive and antiadhesive activities exist on TN-C. A previous study  using a recombinant large TN-C variant has demonstrated an antiadhesive mechanism as a specific interference of TN-C, with cell binding to the HepII/syndecan-4 site in FN through direct binding of TN-C to the 13th FN-III (26). The research group developed their study and reported that TN-C blocks cell cycle progression of anchorage-dependent fibroblasts on FN, through inhibition of syndecan-4 (27). Midwood et al. (28) demonstrated that TN-C modulates cell behavior by interfering with the binding of FN to the HS chains of syndecan-4, and inhibiting down stream activation of RhoA and FAK. Consequently, TN-C and syndecan-4 co regulate FN signaling and matrix contraction in tissue repair. Thus, TN-C acts as a negative modulator for cell adhesion and function by interfering with cooperative regulation of cell adhesion by integrin ␣5␤1 and syndecan-4.
In contrast, the results of the present study suggested that the TN-C molecule harbors a matricryptic site that positively modulates cell adhesion to FN. A synthetic TN-C peptide, TNIIIA2, had the ability to induce activation of ␤1 integrin, as determined by active ␤1 integrin-specific mAbs AG89, HUTS-4, and 9EG7. We have observed that TNIIIA2 protects normal fibroblasts from anoikis-like apoptosis by stimulating the PI3-K/ Akt/Bcl-2 pathway. 3 This pathway is known to be immediately downstream of integrin ␣5␤1 in anchorage-dependent cell survival signaling (29). The result further provides evidence of its relevance to integrin function. Therefore, we conclude that conformational change in ␤1 integrin in response to TNIIIA2 caused its functional activation.
It has generally been considered that integrin activation is regulated by "inside-out" signals from the cell interior. This signaling switch is triggered by an extracellular soluble agonist such as ADP, thrombin, phorbor ester, or certain chemokines. However, several artificial factors not found in vivo are able to induce ␤1 integrin activation. These include divalent cations such as Mn 2ϩ and Mg 2ϩ and anti-␤1 integrin mAbs, including TS2/16 and 12G10, which directly bind to the extracellular domain of ␤1 integrin. It is possible that some physiological process targeting the extracellular region of ␤1 integrin may also be involved in ␤1 integrin activation. Cell surface molecules known to modulate integrin activity include tetraspan/ TM4SF, CD87/uPAR, and CD47/IAP, which associate with integrins via their extracellular domains. For example, integrin ␣3␤1 constitutively associates with CD151 via the ␣3 chain at amino acid residues 570 -705, corresponding to the "thigh" and "calf" domain of the ␣v chain (30). Nishiuchi et al. (31) demonstrated that CD151 serves as a constitutive potentiator for ␣3␤1 activity, through a direct association with this integrin. They speculate that the laterally associated CD151 acts as a "prop" to stabilize the extended high affinity conformation of ␣3␤1 and render the bending unfavorable.
The syndecan family may provide another example of the regulatory roles of ectodomains in cell adhesion signaling (32)(33)(34), although a number of studies have shown the importance of the transmembrane and cytoplasmic domains (35). A site within the ectodomain of syndecan-4 interacts with an uniden-tified surface molecule to promote adhesion without requiring the cytoplasmic domain (36). The integrin ␣v␤3 was shown to be dependent on syndecan-1 for activation and to mediate signals required for human mammary carcinoma cell spread on vitronectin (37). Recently, Whiteford and Couchman (38) have shown that a conserved NXIP motif is required for cell adhesion properties of the syndecan-4 ectodomain. Interestingly, they have also shown that cell adhesion to the syndecan-4 ectodomain involves ␤1 integrin.
The present study suggests that syndecan-4 may be involved in ␤1 integrin activation induced by TNIIIA2. This TNIIIA2induced ␤1 integrin activation required the extracellular region, not the cytoplasmic domains, of syndecan-4 molecule. Although the precise mechanisms underlying TNIIIA2-induced ␤1 integrin activation through syndecan-4 remains to be established, a direct interaction between the extracellular regions of syndecan-4 and ␤1 integrin may induce conformational change in ␤1 integrin. Indeed, the photoaffinity labeling experiment suggested a physical association of syndecan-4 with ␤1 integrin on cell surface (see Fig. 5B). It may be that a physical association with the syndecan-4 ectodomain may force ␤1 integrin to alter its conformation into a functionally active state. Alternatively, syndecan-4 may stabilize the active conformation of ␤1 integrin, as CD151 acts on integrin ␣3␤1 (31). Thus, TNIIIA2 induces a net increase in the number of activated ␤1 integrins on cell surfaces, resulting in enhanced adhesion of adherent and nonadherent cell types. However, adherent cell types, such as the WI38VA13 and NIH3T3 cells used in this study, can attach and spread spontaneously on FN without requiring TNIIIA2 stimulation, although a prolonged time is needed to allow complete spreading. TNIIIA2 may contribute to adhesion of nonadherent cell types, such as hematopoietic cells whose integrins are in their inactive states, rather than to that of adherent cell types. As well as transient expression of TN-C in tissues bearing pathological conditions, such as inflammation and tumorigenesis, significant constitutive expression of TN-C has been observed in adult lymphoid tissues, such as bone marrow, thymus, spleen, and lymph nodes (39,40), while little is known about a physiological role of TN-C expressed in the lymphoid tissues. A TNIIIA2-related matricyptic site/matricryptin may play a critical role in lymphocyte homing and in leukocyte infiltration, as both require integrin activation. In any case, the present study revealed a new route for ␤1 integrin activation via a matricryptic site on the TN-C molecule. Further investigation is needed to define how syndecan-4 ligation with TNIIIA2 can induce conformational change in ␤1 integrin necessary for its functional activation.
Our proposal that ␤1 integrin is activated through its lateral association with syndecan-4, does not exclude the importance of cell adhesion signaling via the cytoplasmic domains of syndecan-4 in focal adhesion formation and in cytoskeletal organization. Taking into account that the active site of TNIIIA2 appears to be cryptic in the TN-C molecule, it should be assumed that TNIIIA2 works mainly in certain pathophysiological states, in which exposure of matricryptic sites happens frequently, as discussed below.
The ECM generates a variety of signals for cell regulation. Recent studies have shown that some of those signals are derived from biologically active cryptic sites of ECM molecules (9 -11). In our study, the TNIIIA2-related matricryptic site was exposed by MMP-2. Siri et al. (41) have shown that large variants of TN-C are much more susceptible to MMP-2, and that MMP-2 completely digests a single FN-III repeat inside the splicing area. It should be noted that TN-C, especially its variants containing the alternative splicing domains, is highly expressed in developing tissues and in pathological tissues (20,42,43,44), where exposure of the matricryptic sites and release of matricryptins occurs frequently. Moreover, expression of syndecan-4 is also up-regulated under similar pathophysiological situations (45). Considering the similarities in specific expression pattern between TN-C and syndecan-4, it is tempting to speculate that the TNIIIA2-related matricryptic site/matricryptin may act, in conjunction with syndecan-4, as a decisive factor for the progression and/or termination of tissue injury. It was reported that TN concentrations in tissues may reach 0.2-2.0 mg/ml (4), which corresponds to the low mM range (46). Thus, significant interactions with syndecan-4 may be expected if the matricryptic site is exposed, based on the binding affinity of TNIIIA2 for HS chain such as heparin (K d ϭ 1.1 ϫ 10 Ϫ7 M), and even considering the controlled expression of the alternative splicing region. It is important to examine whether the TNIIIA2-related matricryptic site is exposed at its functional level in tissues.
We previously found that FN had a matricryptic site suppressing cell adhesion. In sharp contrast to TNIIIA2, peptide FNIII14 containing this matricryptic site is capable of inactivating ␤1 integrins (12,15). It is interesting that the adhesive ECM protein FN has a matricryptic site that negatively modulates ␤1 integrin-mediated cell adhesion and conversely, the representative antiadhesive ECM protein TN-C has a site that positively modulates it. This implies that FN and TN-C harbor active sites that function in opposition to their parental ECM proteins. These matricryptic sites of FN and TN-C may act as a negative feedback loop for preventing excessive cellular responses to these ECM proteins during tissue remodeling.