Pro-angiogenic Activities of CYR61 (CCN1) Mediated through Integrins αvβ3 and α6β1 in Human Umbilical Vein Endothelial Cells

CYR61 (CCN1) is an extracellular matrix-associated protein of the CCN family, which also includes CTGF (CCN2), NOV (CCN3), WISP-1 (CCN4), WISP-2 (CCN5), and WISP-3 (CCN6). Purified CYR61 induces neovascularization in corneal implants, and Cyr61-null mice suffer embryonic death due to vascular defects, thus establishing that CYR61 is an important regulator of angiogenesis. Aberrant expression ofCyr61 is associated with breast cancer, wound healing, and vascular diseases such as atherosclerosis and restenosis. In culture, CYR61 functions through integrin-mediated pathways to promote cell adhesion, migration, and proliferation. Here we show that CYR61 can also promote cell survival and tubule formation in human umbilical vein endothelial cells. Furthermore, we have dissected the integrin receptor requirements of CYR61 with respect to its pro-angiogenic activities. Thus, CYR61-induced cell adhesion and tubule formation occur through interaction with integrin α6β1 in early passage endothelial cells in which integrins have not been activated. By contrast, in endothelial cells in which integrins are activated by phorbol ester or vascular endothelial growth factor, CYR61-promoted cell adhesion, migration, survival, growth factor-induced mitogenesis, and endothelial tubule formation are all mediated through integrin αvβ3. These findings indicate that CYR61 is an activation-dependent ligand of integrin αvβ3 and an activation-independent ligand of integrin α6β1 and that these integrins differentially mediate the pro-angiogenic activities of CYR61. These findings help to define the mechanisms by which CYR61 acts as an angiogenic regulator, provide a molecular interpretation for the loss of vascular integrity and increased apoptosis of vascular cells inCyr61-null mice, and underscore the importance of CYR61 in the development and homeostasis of the vascular system.

Angiogenesis, the formation of new vessels by sprouting from pre-existing ones, is critical for embryonic development, pregnancy, and placentation as well as wound healing and tissue repair (1). Imbalance in angiogenesis can underlie or exacerbate a variety of diseases such as rheumatoid arthritis, diabetic retinopathy, and cancer (2). Angiogenesis begins with degrada-tion of the basement membrane that surrounds the parental vessel, followed by endothelial cell migration, proliferation, and assembly into tubular structures that convey blood supply to target tissues. These processes are regulated by the coordinated interaction of endothelial cells with both angiogenesisinducing factors and components of the extracellular matrix (3,4).
CYR61 is a secreted, extracellular matrix-associated, angiogenic regulator of the CCN protein family (5,6), which includes six members: CYR61, CTGF, 1 NOV, WISP-1, WISP-2, and WISP-3 (7)(8)(9). Members of the CCN family are organized into four distinct structural domains with sequence similarities to insulin-like growth factor-binding proteins, von Willebrand factor type C repeat, thrombospondin type I repeat, and carboxylterminal domains of extracellular matrix proteins such as von Willebrand factor and mucins. Encoded by a growth factorinducible immediate-early gene, CYR61 is a cysteine-rich matricellular protein that supports cell adhesion and induces adhesion signaling (10 -12). Furthermore, CYR61 stimulates endothelial cell migration and enhances growth factor-induced DNA synthesis in culture (10) and induces angiogenesis in vivo (13).
The essential nature of CYR61 as a regulator of vascular development has been established through gene targeting studies in mice (14). Cyr61-null mice suffer embryonic death due to vascular defects that include undervascularization of the placental labyrinth and loss of vascular integrity in the embryo. Large vessels of CYR61-deficient mice show a disorganized basal lamina, paucity of smooth muscle cells that normally comprise the vessel wall, and vascular cells undergoing apoptosis (14). Consistent with the angiogenic activity of CYR61, its expression is elevated in healing wounds (15,16). Furthermore, overexpression of Cyr61 promotes tumor growth and vascularization (13,17) and is associated with human breast cancer (17)(18)(19).
Mechanistically, CYR61 acts as a non-RGD-containing ligand of integrin receptors (8). Integrins are heterodimeric cellsurface receptors capable of transducing extracellular signals and functions to regulate cell adhesion, motility, proliferation, survival, and differentiation (20,21). Thus, extracellular ligand binding to integrins can transduce "outside-in signaling" similar to a growth factor receptor. Conversely, intracellular molecules that interact with the integrin cytoplasmic domains can also induce "inside-out signaling," thereby "activating" the integrin to adopt a conformational change that results in en-hanced affinity for ligands (22). Recent studies indicate that integrin ␣ v ␤ 3 , a prominent integrin in angiogenic endothelial cells, can be activated either through serial passage in culture or by stimulation with agonists such as growth factors and phorbol ester tumor promoters. Activation can also be accomplished by binding of monoclonal antibodies or manganese to the extracellular domains of integrins, thereby inducing a conformational change (22)(23)(24).
In this study, we have uncovered two additional pro-angiogenic activities for CYR61 that have not been previously described, namely the promotion of endothelial cell survival and tubule formation. To understand the mechanisms of CYR61 actions, we have identified the integrin receptors mediating each of five pro-angiogenic activities: endothelial cell adhesion, migration, proliferation, survival, and tubule formation. We show that integrin ␣ 6 ␤ 1 mediates CYR61-supported cell adhesion and tubule formation in unactivated endothelial cells. By contrast, when integrins are activated by stimulation of endothelial cells with agonists, CYR61 promotes cell adhesion, migration, proliferation, survival, and tubule formation through integrin ␣ v ␤ 3 .
Cell Culture and Adhesion Assay-Primary HUVECs were maintained at 37°C with 5% CO 2 in Medium 200 containing 2% serum and endothelial growth supplements according to the manufacturer's instructions (Cascade Biologics). Cells were used within passages 5-10 for all experiments. HUVEC adhesion assays were carried out essentially as described (10). Briefly, 96-well microtiter plates (BD Biosciences) were coated with test proteins diluted in phosphate-buffered saline at 50 l/well and then incubated at 4°C for 16 h, followed by blocking with 1% BSA at room temperature for 1 h. Subconfluent HUVECs were washed twice with phosphate-buffered saline containing 1 mM EDTA and 0.1% glucose and harvested by incubation in the same buffer for 15 min at room temperature. Cells were washed and resuspended in serum-free basal medium containing 0.2% BSA and 10 mM HEPES (pH 7.2) at 7 ϫ 10 5 cells/ml. Where indicated, cells were stimulated with PMA, VEGF, or Mn 2ϩ for 30 min to activate integrin receptors. EDTA, peptides, or function-blocking mAbs were mixed with cells for an additional 30 min prior to plating. The cell suspension was plated at 50 l/well and allowed to adhere to protein-coated wells at 37°C for 20 min, followed by washing twice with phosphate-buffered saline. Adherent cells were fixed with 10% Formalin and stained with methylene blue. Adhesion was quantified by dye extraction and measurement of absorbance at 620 nm as described (10). Vehicle buffers containing Me 2 SO (ranging from 0.001 to 0.005%) had no effect on cell adhesion.
Cell Migration Assay-Cell migration was monitored as described (25) using Transwell chambers (Corning Costar) with tissue culturetreated filter membranes separating the upper and lower chambers. The lower surfaces of polycarbonate filters were coated with BSA or CYR61 (15 g/ml) for 3 h at 37°C, washed, and placed on culture wells containing 300 l of serum-free basal medium plus 0.2% BSA in the lower chamber. Subconfluent HUVECs were harvested by mild trypsinization (0.02%) and then washed and resuspended in serum-free basal medium with 0.2% BSA at 7 ϫ 10 5 cells/ml. Cells were left untreated or stimulated with PMA for 30 min, followed by addition of test mAbs or peptides for another 30 min. Inhibitors (mAbs or peptides) were also added to the lower compartments. The cell suspension (250 l) was placed in the upper compartment of each chamber, and cells were allowed to migrate toward CYR61 (coated on the membrane facing the lower chamber) for 8 h in 5% CO 2 during incubation at 37°C. After incubation, unmigrated cells that remained attached to the upper surfaces of the filters were removed by cotton swabs, and cells that migrated to the lower surfaces of the filters were fixed with methanol and stained with hematoxylin. Cells were counted under a microscope (magnification ϫ200) in 10 random fields/filter. Results are expressed as numbers of migrated cells/field. In other sets of experiments, CYR61 was also tested as a soluble ligand added directly to the lower compartments. Both soluble and immobilized CYR61 yielded similar results as a chemotactic factor.
Thymidine Incorporation Assay-DNA synthesis was assessed as described (10) with minor modifications. Briefly, subconfluent HUVECs were replated on 24-well plates at 3 ϫ 10 4 cells/well in complete culture medium for 16 h and starved in serum-free basal medium with 0.5% BSA for 24 h. Cells were then either left untreated or incubated with mAbs in serum-free basal medium for 1 h as indicated prior to their removal and addition of test proteins. [ 3 H]Thymidine (5 Ci/ml), CYR61, or VEGF was simultaneously added to the wells in medium containing 0.5% charcoal-stripped serum (Sigma). After 48 h of incubation, cells were washed with phosphate-buffered saline and fixed with 10% trichloroacetic acid. DNA was extracted in 0.1 N NaOH, and thymidine incorporation was measured using a scintillation counter.
Measurement of Apoptosis-Subconfluent HUVECs were serumstarved for 24 h and harvested. Cells were suspended in serum-free medium with 0.2% BSA and stimulated with PMA or the integrin ␤ 3 -activating anti-LIBS-6 mAb for 1 h. Glass coverslips (18 ϫ 18 mm) were coated with poly-L-lysine, vitronectin, CYR61, or laminin for 16 h at 4°C and then blocked with 1% BSA. Cells were seeded on the glass coverslips at low cell density (10,000 cells/cm 2 ), allowed to attach to these different matrix coatings, and challenged with serum-free medium containing 0.2% BSA. After 20 h of incubation, cells were fixed in 4% paraformaldehyde for 30 min, and apoptosis was monitored by TUNEL assays using an in situ cell death detection kit (Roche Molecular Biochemicals). Prior to analysis of apoptotic nuclei, cells were lightly counterstained with hematoxylin, and total cell numbers were examined. The total numbers of cells attached to different matrix coatings were similar. The extent of apoptosis is represented as a percentage of apoptotic cells versus the total; at least 500 cells were counted from random fields in each slip.
Endothelial Tubule Formation-The formation of capillary-like tubules by HUVECs was evaluated using a three-dimensional collagen gel as described (26) with modifications. Briefly, cells were harvested and resuspended in serum-free basal medium with 0.2% BSA. A mixture of type I collagen gels, CYR61 protein stocks (0.5 mg/ml) or vehicle buffer (containing 0.75 M NaCl and 50 mM phosphate buffer), and neutralization buffer (260 mM NaHCO 3 , 200 mM HEPES, and 50 mM NaOH) was mixed at a ratio of 7:2:1 by volume and kept on ice to prevent gelation. Cold collagen mixtures (250 l) were put in 24-well plates, and gelation occurred at 37°C for 1 h. Unstimulated or PMA (5 nM)-stimulated HUVECs were inoculated on the wells at 125,000 cells/cm 2 and cultured at 37°C. After 16 h of incubation, the medium was removed, and a second collagen gel mixture of identical components was overlaid on the cells. After 10 min of gel polymerization at 37°C, 0.5 ml of supplemented serum-free medium was added to each well. Formation of apparent endothelial tubules was scored 16 -20 h thereafter.

RESULTS
Differential Utilization of Integrins ␣ v ␤ 3 and ␣ 6 ␤ 1 as Adhesion Receptors for CYR61 in HUVECs-CYR61 is a ligand of integrin ␣ v ␤ 3 and supports HUVEC adhesion through interaction with this integrin (27). To investigate the effects of integrin activation on CYR61 action, we examined the characteristics of adhesion of unstimulated early passage HUVECs to CYR61 compared with cells stimulated to activate integrins. As shown in Fig. 1A, unstimulated early passage HUVECs adhered to CYR61. However, only a minor fraction of this adhesion was inhibited by agents that disrupt integrin ␣ v ␤ 3 function such as GRGDSP peptide (24% inhibition) or LM609 (19% inhibition). Neither the control peptide GRGESP nor normal mouse IgG had any effect on cell adhesion to CYR61, whereas EDTA abolished cell adhesion completely (Fig. 1A). These results show that in naive early passage HUVECs, integrin ␣ v ␤ 3 is not the principal adhesion receptor for CYR61. Another cation-dependent receptor is required for adhesion to CYR61.
Treatment of endothelial cells with agonists such as the phorbol ester PMA, the divalent cation Mn 2ϩ , and the angiogenic growth factor VEGF can rapidly increase integrin ␣ v ␤ 3 activity via enhancement of ligand binding affinity without changing the receptor expression profile (28,29). When HU-VECs were stimulated with PMA to activate integrin ␣ v ␤ 3 , total cell adhesion to CYR61 was significantly increased (ϳ1.6fold) (Fig. 1B). This PMA-dependent enhancement of cell adhesion to CYR61 was inhibited by GRGDSP peptide or mAb LM609, indicating that this cell adhesion is mediated through integrin ␣ v ␤ 3 . Similarly, following stimulation with the angiogenic growth factor VEGF (10 ng/ml), HUVEC adhesion to CYR61 was increased by ϳ2-fold (Fig. 1C). This increment was also blocked by GRGDSP peptide or mAb LM609, but not by the control peptide GRGESP or by normal mouse IgG. In addition to these two agents that activate integrins through inside-out signaling, we observed that Mn 2ϩ (1 mM), which can activate integrins by binding to the extracellular domain and inducing a conformational change (30), also enhanced HUVEC adhesion to CYR61 (data not shown). Together, these results indicate that activation of endothelial cell integrins by PMA, VEGF, or Mn 2ϩ greatly enhances cell adhesion to CYR61 through integrin ␣ v ␤ 3 . The diversity of these agonists indicates that it is activation of integrins, rather than other cellular events, that is responsible for the enhanced cell adhesion. However, a residual level of HUVEC adhesion to CYR61 remained even in the presence of ␣ v ␤ 3 antagonists (e.g. RGD-containing peptides or mAb LM609), again suggesting that another receptor might also contribute to endothelial cell adhesion to CYR61 (27).
In light of the recent finding that fibroblasts and smooth muscle cells adhere to CYR61 through integrin ␣ 6 ␤ 1 (31, 32), we explored the possible involvement of integrin ␣ 6 ␤ 1 in mediating HUVEC adhesion to CYR61. As shown in Fig. 2A, unstimulated HUVEC adhesion to CYR61 was effectively inhibited (ϳ80%) by mAb against the integrin ␣ 6 (GoH3) or ␤ 1 (P4C10) subunit. By contrast, mAb against integrin ␣ v ␤ 3 (LM609) minimally inhibited adhesion (ϳ20%). These results indicate that integrin ␣ 6 ␤ 1 is the primary adhesion receptor for CYR61 in naive HUVECs. Notably, a combination of LM609 and GoH3 or of LM609 and P4C10 completely abrogated cell adhesion to CYR61, indicating that integrins ␣ 6 ␤ 1 and ␣ v ␤ 3 together serve as the adhesion receptors for CYR61 in HUVECs.
In fibroblasts and smooth muscle cells, adhesion to CYR61 requires both integrin ␣ 6 ␤ 1 and heparin sulfate proteoglycans serving as co-receptors, and occupancy of the CYR61 heparinbinding site by soluble heparin abolishes its ability to support cell adhesion (31,32). Likewise, we found that soluble heparin diminished CYR61-supported HUVEC adhesion by ϳ80%, similar to inhibition by mAb GoH3 or P4C10 ( Fig. 2A). Administration of a combination of soluble heparin and GRGDSP peptide, which inhibit cell adhesion mediated through integrins ␣ 6 ␤ 1 and ␣ v ␤ 3 , respectively, abolished adhesion completely. These results further support the conclusion that HUVEC adhesion to CYR61 is mediated through both integrins ␣ 6 ␤ 1 and ␣ v ␤ 3 . Nonetheless, integrin ␣ 6 ␤ 1 functions in cooperation with heparan sulfate proteoglycans to serve as the principal adhesion receptor in unstimulated HUVECs.
Activation of integrin ␣ v ␤ 3 by PMA or VEGF stimulation of HUVECs strongly enhanced the affinity of this integrin for CYR61, rendering ␣ v ␤ 3 , instead of ␣ 6 ␤ 1 , the principal adhesion receptor (Fig. 2, B and C). After activation by either agonist, mAb against integrin ␣ v ␤ 3 (LM609) inhibited cell adhesion by ϳ75%, whereas mAbs against integrin ␣ 6 ␤ 1 (GoH3 and P4C10) inhibited cell adhesion only by 15-20%. The combined administration of LM609 and P4C10 or of GRGDSP peptide and heparin completely abrogated cell adhesion to CYR61. Together, these results show that integrin ␣ 6 ␤ 1 is the principal adhesion receptor for CYR61 in unstimulated HUVECs. Upon stimulation, integrin ␣ v ␤ 3 becomes the major receptor for CYR61, and integrin ␣ 6 ␤ 1 retreats to play an auxiliary role.
HUVEC Migration to CYR61 Is Mediated through Activated Integrin ␣ v ␤ 3 -It was previously shown that CYR61 induces integrin ␣ v ␤ 3 -dependent chemotaxis in human microvascular endothelial cells (13). Inasmuch as integrin ␣ 6 ␤ 1 acts as another receptor for CYR61 in HUVECs as described above, we sought to determine the role of this integrin in HUVEC migration. Cell migration was evaluated using a modified Boyden chamber assay. CYR61 (15 g/ml) was immobilized on the lower surface of the membrane that separated the two chambers, and HUVECs placed in the upper chamber were allowed to migrate through pores of the membrane toward CYR61. As shown in Fig. 3A, unstimulated HUVECs displayed a low level of migration toward CYR61. (The background level of migrated cells was 14 Ϯ 3 cells/field on BSA-coated membrane versus 47 Ϯ 6 cells/field on CYR61-coated membrane.) Either GRGDSP peptide or LM609 blocked CYR61-induced cell migration, suggesting that the cell migration activity is mediated through a low level of activated integrin ␣ v ␤ 3 . However, antiintegrin ␣ 6 (GoH3) and anti-integrin ␤ 1 (P4C10) mAbs or soluble heparin had no effect (Fig. 3B). Stimulation with PMA greatly enhanced HUVEC migration toward CYR61 (285 Ϯ 24 cells/field in PMA-stimulated cells versus 47 Ϯ 6 cells/field in unstimulated cells) (Fig. 3, A and B). This enhanced cell migration was obliterated by integrin ␣ v ␤ 3 antagonists (GRGDSP peptide and LM609), but not by integrin ␣ 6 ␤ 1 antagonists (GoH3 and P4C10), and soluble heparin also had no effect. These data show that HUVEC migration toward CYR61 is dependent upon activated integrin ␣ v ␤ 3 , whereas integrin ␣ 6 ␤ 1 plays no role in this process.
CYR61 Enhances VEGF-induced DNA Synthesis through Activated Integrin ␣ v ␤ 3 -Although CYR61 is not mitogenic by itself, it enhances growth factor-induced DNA synthesis in fibroblasts and endothelial cells (27,33). To understand the mechanism of CYR61 action, we tested the effect of CYR61 on HUVEC mitogenesis. As expected, CYR61 added alone to HU-VECs had no effect on DNA synthesis (Fig. 4), and PMA alone or in combination with CYR61 also did not stimulate DNA synthesis (data not shown). DNA synthesis was induced when HUVECs were stimulated with 5 ng/ml VEGF, which is suboptimal as determined in a dose-response titration (data not shown). However, the presence of CYR61 was able to induce an additional 2-fold enhancement of DNA synthesis over VEGF treatment alone (Fig. 4). This CYR61-enhanced DNA synthesis (but not DNA synthesis induced by VEGF alone) was completely abolished by LM609. By contrast, GoH3 had no effect on CYR61-enhanced or VEGF-induced DNA synthesis. These data show that CYR61 enhances VEGF-induced HUVEC DNA synthesis through ligation to activated integrin ␣ v ␤ 3 , and not integrin ␣ 6 ␤ 1 .
CYR61 Promotes Endothelial Cell Survival through Integrin ␣ v ␤ 3 -A number of angiogenic factors can protect endothelial cells from apoptotic death, although the effect of CYR61 on endothelial cell survival is unknown. To investigate this question, we plated HUVECs on poly-L-lysine, vitronectin, laminin, or CYR61. Cells were either left unstimulated or stimulated with PMA. Additionally, a class of mAbs against integrins called anti-LIBS antibodies recognize ligand-induced binding sites that become exposed when integrins have assumed the activated conformation upon ligand binding. At sufficiently high concentrations, some anti-LIBS antibodies can bind to integrins and force them to assume the activated conformation in the absence of ligand. The integrin ␤ 3 -activating anti-LIBS-6 mAb is one such antibody (34) and was employed as an alternative and specific means to activate ␤ 3 integrins (Fig. 5). After incubation, apoptosis was analyzed using a TUNEL assay. Cells plated on poly-L-lysine were attached through non-integrin-mediated electrostatic interactions, and the majority of these cells (Ͼ90%) underwent apoptosis even in the presence of PMA or anti-LIBS-6 mAb stimulation (Fig. 5). Unstimulated HUVECs attached to vitronectin, a well characterized extracellular matrix ligand capable of conferring anti-apoptotic signals (35,36), showed a minimal apoptotic response (ϳ20%); PMA or anti-LIBS-6 mAb pretreatment of cells did not provide further protection against apoptosis. Previous studies showed that laminin, a ligand of several integrins including ␣ 6 ␤ 1 , failed to sustain endothelial cell survival even in a defined medium containing a high dose of basic fibroblast growth factor (25 ng/ml) due to the inability of laminin to engage integrin receptors that transduce anti-apoptotic signals (37). Consistent with these observations, a large fraction of HUVECs were apoptotic (ϳ65%) when plated on laminin, irrespective of the activation state (Fig. 5). When unstimulated HUVECs adhered to CYR61, a high apoptotic rate (ϳ60%) was observed. By contrast, treatment of HUVECs with PMA or anti-LIBS-6 mAb to activate integrin ␣ v ␤ 3 resulted in significant enhancement of cell survival, thereby reducing the fraction of apoptotic cells to ϳ25%. These results are consistent with the interpretation that in unstimulated HUVECs, cells adhere to CYR61 through integrin ␣ 6 ␤ 1 , which is unable to transduce cell survival signals. When HUVECs are stimulated to activate integrin ␣ v ␤ 3 , however, cells attached to CYR61 are protected from apoptotic death through engagement of integrin ␣ v ␤ 3 .
CYR61 Induces Tubule Formation in HUVECs through Integrins ␣ 6 ␤ 1 and ␣ v ␤ 3 -Although CYR61 has been shown to induce angiogenesis in corneal micropocket implants, its ability to induce tubule formation in endothelial cells has not been demonstrated. We utilized a collagen gel assay to assess the ability of purified CYR61 to induce endothelial cell differenti- ation to form capillary-like tubules. Untreated HUVECs exhibited a minimal differentiation response without apparent tubule formation (Fig. 6A). Cells pretreated with the angiogenic factor basic fibroblast growth factor (5 ng/ml) exhibited sprouting and branching of tubules (data not shown). When formulated into collagen gel, CYR61 induced tubule formation in a dose-dependent manner (Fig. 6, B-E). Whereas mAbs against integrins ␣ 6 (GoH3) and ␤ 1 (P4C10) obliterated CYR61-induced tubule formation, mAb against ␣ v ␤ 3 (LM609) had no inhibitory effect (Fig. 6, F-H). These results support the interpretation that integrin ␣ 6 ␤ 1 is involved in CYR61-induced differentiation of naive HUVECs.
We next investigated the effects of integrin activation on CYR61-induced tubule formation. As shown in Fig. 7A, unstimulated HUVECs in collagen gel did not undergo morphogenesis, and 5 nM PMA-treated cells developed minimal induction of endothelial tubules (Fig. 7C). Gels containing 25 g/ml CYR61 barely induced any tubule formation in unstimulated cells (Fig. 7B), whereas PMA stimulation greatly enhanced the tubule formation response (Fig. 7D). Administration of LM609 inhibited endothelial morphogenesis in PMA-treated HUVECs in both the presence and absence of CYR61. By contrast, GoH3 or control normal mouse IgG had little effect (Fig. 7, E-H) (data not shown). Together, these data show that CYR61 induces endothelial tubule formation through integrin ␣ 6 ␤ 1 in unstimulated HUVECs, but that it induces more extensive endothelial morphogenesis when integrin ␣ v ␤ 3 has been activated. DISCUSSION The primary findings of this study provide new insights into the angiogenic actions of CYR61, an essential regulator of mammalian vascular development. Functionally, we have documented two novel activities of CYR61, establishing its ability to promote vascular endothelial cell survival and tubule formation. Mechanistically, we have identified the cell-surface receptors mediating each of five pro-angiogenic activities of CYR61. In agonist-stimulated endothelial cells in which integrins are activated, CYR61-promoted cell adhesion, migration, survival, growth factor-induced mitogenesis, and endothelial tubule formation are all mediated through integrin ␣ v ␤ 3 . By contrast, CYR61-induced cell adhesion and tubule formation occur through interaction with integrin ␣ 6 ␤ 1 in early passage endothelial cells in which integrins have not been activated. These findings indicate that CYR61 is an activation-dependent ligand FIG. 6. CYR61 induces integrin ␣ 6 ␤ 1 -dependent endothelial tubule formation in unstimulated HUVECs. HUVECs were plated on 24-well plates precoated with type I collagen gels (2 mg/ml) in the absence (A) or presence (B-H) of purified CYR61, and a second layer of gel was overlaid on the attached cells as described under "Materials and Methods." Tubule formation was assessed 16 -20 h thereafter. A, the collagen gel was formulated with CYR61 buffer. B-E, the collagen gels were formulated with 25, 50, 75, and 100 g/ml CYR61, respectively, with endothelial tubules becoming evident at 25-50 g/ml CYR61 (indicated by arrows in B and C). F, the presence of LM609 (40 g/ml) failed to inhibit tubule formation induced by CYR61 (100 g/ml) in the collagen gel. G and H, addition of GoH3 (40 g/ml) and P4C10 (1:50 ascites), respectively, effectively blocked tubule formation. Results are representative of three separate experiments, each performed in duplicate (magnification ϫ100). of integrin ␣ v ␤ 3 and an activation-independent ligand of integrin ␣ 6 ␤ 1 , provide a mechanistic interpretation for certain phenotypes in Cyr61-null mice, and suggest a role for CYR61 in vessel development and maintenance.
For new vessels to form, endothelial cells must migrate, proliferate, and differentiate into tubule structures. Consistent with these requisite cellular processes, CYR61 has been shown to promote endothelial cell adhesion, migration, and proliferation in culture (10) and to induce angiogenesis in corneal implants (13). Data presented herein show that CYR61 also promotes endothelial cell survival (Fig. 5) and formation of endothelial tubules (Figs. 6 and 7). Together, these findings demonstrate that in a purified form, CYR61 can act directly upon endothelial cells to promote each of the requisite cellular steps of angiogenesis. CYR61 is also known to up-regulate the expression of other potent angiogenic inducers such as VEGF-A and VEGF-C in fibroblasts (15). Thus, the angiogenic actions of CYR61 can be both direct and indirect.
During embryonic development, Cyr61 expression is notable in the cardiovascular system, especially in vascular endothelial cells and in smooth muscle cells surrounding the larger vessels (16,33). The importance of CYR61 in vascular development has been established by mutational analysis via gene targeting in mice. Cyr61-null mice suffer undervascularization of the placental labyrinth and loss of vascular integrity in large vessels, resulting in embryonic death (14). In particular, large vessels in Cyr61-null embryos display a disorganized basal lamina with vascular cells undergoing apoptosis. This finding is consistent with the observation that by acting as an adhesion substrate, CYR61 can provide an anti-apoptotic function to endothelial cells (Figs. 5). Enhanced cell survival is critical for endothelial cells in the angiogenic process as they undergo migration, mitogenesis, and differentiation. It is interesting to note that WISP-1, a closely related protein of the CCN family, can promote cell survival through activation of the Akt pathway (38).
CYR61 acts as a non-RGD-containing ligand of integrin receptors, which mediate many of its activities. The utilization of integrins by CYR61 is cell type-and function-specific. For example, cell adhesion to CYR61 in platelets and monocytes is mediated through integrins ␣ IIb ␤ 3 and ␣ M ␤ 2 , respectively, whereas cell adhesion and migration in smooth muscle cells are mediated through integrin ␣ 6 ␤ 1 (32,39,40). Remarkably, CYR61 promotes fibroblast adhesion, migration, and proliferation through integrins ␣ 6 ␤ 1 , ␣ v ␤ 5 , and ␣ v ␤ 3 , respectively (31,41). In this study, we have shown that CYR61 can selectively utilize either integrin ␣ 6 ␤ 1 or ␣ v ␤ 3 in endothelial cells to mediate cell adhesion, depending on the activation state of the cell. Together, these findings demonstrate that CYR61 interacts with distinct integrins in a cell type-, function-, and activation state-specific manner to mediate disparate biological activities.
Integrins are capable of transducing extracellular signals into the cell via ligand binding in outside-in signaling, analogous to signal transduction by growth factor receptors (20,21). It has been recognized that integrins are also capable of insideout signaling, whereby interaction of intracellular molecules with the cytoplasmic domains of the integrin subunits can lead to a conformational change in the extracellular domains, resulting in an "activated" conformation with increased ligand binding affinity (22). Inside-out signaling has been particularly well established for integrins ␤ 2 and ␤ 3 , and these integrins can be activated by either physiological (e.g. growth factors and cytokines) or nonphysiological (e.g. antibodies and Mn 2ϩ ) agents. In this study, we have shown that CYR61 is another example of an activation-dependent ligand of integrin ␣ v ␤ 3 . It is notable that other ligands such as fibrinogen can bind integrin ␣ v ␤ 3 in an activation-independent manner (30). ␤ 1 integrins also undergo activation, as demonstrated by the existence of anti-LIBS antibodies against the ␤ 1 subunit (22). However, activation does not appear to alter the affinity of integrin ␣ 6 ␤ 1 for CYR61.
A wealth of evidence indicates that integrins play important roles in angiogenesis, and in particular, signaling through ␣ v integrins has long been implicated in angiogenic events (3). A number of angiogenic regulators can bind directly to integrins, which mediate at least some of their activities (43). VEGF-and basic fibroblast growth factor-induced angiogenesis has been shown to require integrins ␣ v ␤ 5 and ␣ v ␤ 3 , respectively, in angiogenesis assays using rabbit corneal pocket implants and chick chorioallantoic membrane (3). Furthermore, integrins have been known to associate with growth factor receptors, and cross-talk between the two receptor systems is evident (21,43). Notably, VEGF receptors can associate with ␣ v integrins and can function to activate integrin ␣ v ␤ 3 strongly (28,44). Thus, it is surprising that extensive vasculogenesis, angiogenesis, and organogenesis can occur in mice that lack all ␣ v integrins, indicating that the roles of ␣ v integrins are nonessential in the early phases of mammalian development (45). All ␣ v -null mice develop normally to embryonic day 9.5, although 80% die in mid-gestation due to deficient placentation, and the remaining 20% are born alive, but suffer intracerebral and intestinal hemorrhages. Interestingly, pathological angiogenesis is clearly abnormal in integrin ␤ 3 and ␤ 5 knockout mice, suggesting that molecular requirements may differ for developmental versus pathological angiogenesis (46).
Integrin ␣ 6 ␤ 1 is the principal adhesion receptor for CYR61 in unactivated endothelial cells, and CYR61 induces endothelial tubular networks in both collagen gels and Matrigel (Fig. 6) (data not shown). Matrigel, a basement membrane extract, can support a basal level of endothelial tubule formation without additional angiogenic factors and can promote tumor growth in mice when co-injected with tumor cells (47). Although the signals that promote tubule formation in Matrigel are not well defined, it was found to be dependent upon integrin ␣ 6 ␤ 1 , rather than integrin ␣ v ␤ 3 (48 -50). Given that CYR61 can induce endothelial tubule formation via integrin ␣ 6 ␤ 1 , we explored the possible presence of CCN family members in Matrigel. Interestingly, we found that CTGF can be detected by immunoblotting in the range of 2-5 g/ml, depending on the batch of Matrigel (data not shown). Inasmuch as the cellular and angiogenic activities of CYR61 and CTGF are very similar (39), this concentration of CTGF is likely close to the active range. Thus, endothelial tubule formation observed in Matrigel in the absence of additional factors may be due, at least in part, to the activities of CTGF acting through integrin ␣ 6 ␤ 1 .
Although it is notable that ␤ 1 -null mice fail to develop a vasculature (51), it is unclear which ␤ 1 integrin plays the most critical roles in vascular development. Although the precise role of integrin ␣ 6 ␤ 1 in angiogenesis is currently unknown, our data indicate that integrin ␣ 6 ␤ 1 may participate in the angiogenic process. To date, only a few ligands are known for integrin ␣ 6 ␤ 1 ; these include laminin, CYR61, CTGF, and a collagen fragment known as tumstatin (11,31,52). In addition, the bacterial protein invasin and the sperm surface protein fertilin also bind integrin ␣ 6 ␤ 1 , although these proteins are not expected to play a role in developmental angiogenesis (53,54). Tumstatin is a 28-kDa fragment of the type IV collagen ␣3 chain that inhibits angiogenesis, displays both anti-angiogenic and pro-apoptotic activities, and binds both integrins ␣ 6 ␤ 1 and ␣ v ␤ 3 (52). Two RGD-independent integrin ␣ v ␤ 3 -binding sites in tumstatin confer anti-angiogenic activities, although the role of integrin ␣ 6 ␤ 1 in tumstatin function is unknown (55). Interestingly, whereas short-term stimulation of endothelial cells by angiogenic cytokines such as fibroblast growth factor-␤ leads to acute activation of integrins, prolonged treatment leads to upregulation of integrin ␣ 6 ␤ 1 and down-regulation of integrin ␣ v ␤ 3 (42). Thus, it is tempting to speculate that during the early phase of vessel sprouting, activated endothelial cells may utilize integrin ␣ v ␤ 3 to mediate the angiogenic response. In the late phases of angiogenesis and upon withdrawal of inducing signals, maintenance of vessel integrity may occur via integrin ␣ 6 ␤ 1 . The possibility that integrins ␣ v ␤ 3 and ␣ 6 ␤ 1 may be selectively utilized to play distinct roles in vessel sprouting and maintenance during different phases of angiogenesis is intriguing and merits further investigation.