Requirement for the Synergy Site for Cell Adhesion to Fibronectin Depends on the Activation State of Integrin α5β1

We investigated the influence of the activation state of integrin α5β1 on its dependence on the PHSRN synergy site for binding to RGD in fibronectin. K562 and MV3 cells lacked αvβ3 expression and adhered to fibronectin through α5β1. Mel57 cells adhered through αvβ3 and α5β1. A recombinant fibronectin polypeptide, containing five type III repeats from the central cell binding domain 3Fn6-10, and a mutated polypeptide lacking the synergy site were equally effective in promoting Mel57 adhesion. For K562 and MV3, the mutated polypeptide was not or poorly active compared to the control polypeptide. Expression of αvβ3 in MV3 induced strong adhesion to the mutated polypeptide. TS2/16 stimulatory β1-integrin antibodies or Mn induced α5β1-mediated adhesion of K562 and MV3 to GRGDSP. In the presence of TS2/16 or Mn, α5β1-mediated MV3 adhesion to the mutated polypeptide was equally strong as adhesion to the control polypeptide. Mn or TS2/16 induced weak K562 binding to the mutated polypeptide, and in the presence of a combination of phorbol 12-myristate 13-acetate, Mn, and TS2/16, α5β1-mediated K562 adhesion to the mutated and control polypeptide was equally strong. Our findings demonstrate that requirement for the PHSRN synergy site for α5β1-mediated adhesion to RGD in fibronectin depends on the activation state of the integrin.

Fibronectin (Fn) 1 is an extracellular matrix glycoprotein that functions in cell adhesion and migration in wound healing, embryonic development, and malignant transformation (1,2). The Fn molecule is composed of three types of repeating modules, termed type I, II, and III repeats (3), which are organized into functional domains. Proteolytic cleavage yields several fragments containing domains that promote cell adhesion, including the carboxyl-terminal HepII domain (4), the alternatively spliced type III connecting segment (5), and the central cell binding domain (CCBD).
The CCBD consists of type III repeats, each containing approximately 90 amino acids (6). Cells bind to the CCBD via receptors of the integrin family (7). Integrins are ␣␤ het-erodimeric transmembrane molecules mediating cell-cell adhesion and attachment of cells to the extracellular matrix (8).
The Arg-Gly-Asp (RGD) sequence in the 10th type III repeat (3Fn10) is the key attachment site for binding of these integrins to the CCBD, as demonstrated by inhibition of cell adhesion with synthetic RGD-containing peptides (17,18). Furthermore, two synergistic regions in the CCBD besides RGD have been identified that are required for cell adhesion through ␣IIb␤3 (19) and ␣5␤1 (20 -23). For ␣5␤1 binding to Fn, the synergy region in 3Fn9 is the most important of these two regions (21), and recently, a short amino acid sequence Pro-His-Ser-Arg-Asn (PHSRN) was identified in this repeat that synergistically enhances the cell adhesion promoting activity of the RGD sequence (24). This sequence is also present in an 11-amino acid integrin binding site from 3Fn9 that is recognized by ␣IIb␤3 (25).
Integrins do not always constitutively bind to their ligands with high affinity. Integrin adhesiveness can be stimulated by phorbol esters and other more physiologically relevant agonists (8,26). In addition, antibodies have been described to integrin ␤1 (27)(28)(29)(30), ␤2 (31), and ␤3 (32) subunits, which induce a high affinity state of the integrins. Studies with stimulatory ␤1 antibodies on hematopoietic cells have demonstrated modulation of binding to natural ligands (28 -30), modulation of ligand specificity (33), modulation of binding to different regions in one ligand (34), and modulation of the minimal sequence of a binding site required for adhesion (35).
In the present study, we have investigated the role of the PHSRN synergy site in ␣5␤1and ␣v␤3-mediated cell adhesion to the CCBD in Fn. We show that requirement for the PHSRN synergy site for cell adhesion to the CCBD depends on the integrins expressed and on the activity of the integrins involved.

MATERIALS AND METHODS
Fibronectin, Fragments, and Peptides-Plasma Fn was purchased from Sigma. A 120-kDa chymotryptic Fn fragment containing the CCBD (36,37) was purchased from Life Technologies, Inc. Synthetic peptide Gly-Arg-Gly-Asp-Ser-Pro (GRGDSP) was obtained from the Department of Organic Chemistry, Faculty of Science, University of Nijmegen (The Netherlands) and covalently bound to bovine serum albumin (BSA) as previously described (38).
Production of Recombinant Fibronectin Polypeptides-To avoid the artifactual losses of adhesive activity known to result from adsorbing short polypeptides on substrates (e.g. see Ref. 23), we used recombinant Fn polypeptides containing five type III Fn repeats from 3Fn6 through 3Fn10. The 3Fn6 -10 wild type expression construct was generated based on the T7 phage promotor and a Fn cDNA fragment encoding Fn type III repeat numbers 6 -10 produced using the polymerase chain reaction method (24); the PHSRN sequence was present in repeat 9 and RGD in repeat 10 ( Fig. 2A). The amino-terminal sequence of 3Fn6 starts immediately after an initiation codon for methionine. To create substitution mutants, two complementary oligonucleotides with appropriate sequences were synthesized, annealed, and then cloned between the BamHI and EcoRI sites of 3Fn9. This yielded a mutated polypeptide 3Fn6 -10(SPSDN) where the PHSRN sequence from 3Fn9 was substituted by SPSDN from 3Fn8 (Fig. 2, A and B).
Protein expression was induced by 1 mM isopropyl-1-thio-␤-D-galactopyranoside treatment of Escherichia coli strain BL21 (DE3, pLysS) containing the expression plasmid. The expressed recombinant polypeptides were purified by sequential DEAE and hydroxyapatite column chromatography. The polypeptide was eluted from a DEAE column (DE52, Whatman) using a linear gradient of 0 -0.5 M NaCl in 10 mM sodium phosphate (pH 7.4), 1 mM EDTA, 0.02% sodium azide, applied to a hydroxyapatite column (Bio-Rad), and eluted using a linear gradient from 5 mM sodium phosphate (pH 6.5), 0.4 mM EDTA, 0.02% sodium azide to 250 mM sodium phosphate (pH 6.5), 0.4 mM EDTA, 0.02% sodium azide. The fractions with peak absorbance were evaluated for purity by SDS-polyacrylamide gel electrophoresis, pooled, dialyzed against phosphate-buffered saline without Ca 2ϩ or Mg 2ϩ and with 0.02% sodium azide, and stored at Ϫ80°C.
Cell Lines and Culture Conditions-The human melanoma cell lines used included Mel57 (39) and MV3 (40). The K562 erythroleukemic cell line was provided by Dr. Nancy Hogg. All cell lines were cultured in Dulbecco's modified Eagle's medium (DMEM) (Flow, Irvine, United Kingdom) supplemented with 10% fetal calf serum, penicillin, and streptomycin.
Antibodies Cell Adhesion-Cell adhesion assays were performed as previously described (50). In short, polystyrene microtiter plates (Greiner, Alphen a/d Rijn, The Netherlands) were coated overnight with the appropriate adhesive ligands and blocked for 1 h at 37°C with DMEM containing 0.5% (w/v) BSA. Subsequently, 1 ϫ 10 4 51 Cr-labeled cells in 50 l of DMEM/BSA were added to the wells and incubated for 30 min at 37°C in 5% CO 2 . Unbound cells were removed by washing with DMEM/BSA, bound cells were lysed by detergent, and radioactivity of the lysate was measured in a gamma counter. Results are presented as the mean percentage of cell binding from triplicate wells. For induction of adhesion, radiolabeled cells were preincubated with TS2/16 or 8A2 mAbs for 30 min at 4°C before seeding in the wells or 1 mM MnCl 2 ; or, 100 ng/ml phorbol 12-myristate 13-acetate (PMA) was added to the cells prior to seeding in the wells. For antibody inhibition studies, cells were preincubated with the appropriate mAbs for 30 min at 4°C before seeding into the wells.
Transfection-The full-length cDNA for the integrin ␤3 subunit (51), a kind gift from Dr. Erkki Ruoslahti, was cloned in the polylinker of the mammalian expression vector pBJ1neo (52), kindly provided by Dr. René de Waal-Malefijt. 20 g of this construct was used for stable transfection of MV3 cells according to the calcium phosphate precipitation method (53), using the calcium phosphate transfection system (Life Technologies, Inc.). After 48 h, stably transfected cells were selected by culturing in the presence of 1 mg/ml G418 (Life Technologies, Inc.) for 2 weeks. Cell populations were enriched for ␣v␤3 expression by cell sorting in an Epics Elite flow cytometer using LM609 mAbs. After three cycles of sorting, transfected cell lines contained Ͼ95% ␣v␤3 positive cells. Cells were maintained in culture in medium containing 200 g/ml G418 and were regularly monitored for ␣v␤3 expression.

K562, MV3, and Mel57
Differentially Adhere to the CCBD-We investigated adhesion of K562 human erythroleukemic cells and MV3 and Mel57 human melanoma cells to the CCBD. Of the integrins known to be involved in adhesion to Fn, FIG. 1. Inhibition of adhesion to Fn120 kDa with integrin mAbs. Cells were allowed to adhere to wells coated with 20 g/ml of a 120-kDa fragment of Fn (Fn120 kDa) in the absence (no) or in the presence of inhibitory mAbs to integrin subunits as indicated. Adhesion to BSA was less than 5%. One representative experiment of four is shown.
FIG . 2. Recombinant FN polypeptides. A, schematic representation of a recombinant Fn polypeptide 3Fn6 -10 consisting of five type III Fn repeats from 3Fn6 through 3Fn10 and of a mutated Fn polypeptide 3Fn6 -10(SPSDN) where the region containing PHSRN in 3Fn9 has been substituted by the corresponding region from 3Fn8 (hatched bar). B, in 3Fn6 -10(SPSDN), 16 amino acid residues from 3Fn9 were substituted with the corresponding residues from 3Fn8 to disrupt the PHSRN site in 3Fn9. Boxed sequences are from 3Fn6 -10(SPSDN). The region with italicized nucleotides differs from the original sequence in 3Fn8 for technical reasons to alter restriction sites, but this does not alter the amino acid sequence.  K562  4  4  4  46  7  3  52  3  MV3  4  69  31  62  13  3  107  2  Mel57  3  41  17  23  33  3  33  41 K562 exclusively expressed ␣5␤1 (Table I). MV3 and Mel57 expressed ␣3␤1, ␣4␤1, and ␣5␤1. In addition, Mel57 but not MV3 expressed ␣v␤3. MV3 and Mel57 expressed other ␣v integrins including ␣v␤5 (54) and possibly ␣v␤1 that may bind to Fn. To exclude influences from domains outside the CCBD that are known to have cell adhesive activity (HepII, IIICS), we used a 120-kDa Fn fragment that lacks the heparin-binding domain and the V region but includes the CCBD. K562 adhered weakly to Fn120 kDa, whereas MV3 and Mel57 both adhered strongly (Fig. 1). As expected from the surface expression data, adhesion of K562 was completely blocked by mAbs to ␣5 or ␤1. Even though MV3 expressed several Fn-binding integrins, adhesion was fully blocked by mAbs to ␣5, whereas mAbs to ␣3 or ␣4 or polyclonal anti-␣v had no effect. Adhesion of Mel57 was inhibited approximately 35% by mAbs to ␣5 or ␤1 and about 50% by Adhesion to BSA was less than 3%. One representative experiment of three is shown. mAbs to ␤3 or ␣v␤3 or by polyclonal ␣v. The combination of mAbs to ␣5 and ␣v␤3 completely blocked adhesion of Mel57.
K562 and MV3 Require the PHSRN Synergy Site, Whereas for Mel57 RGD Is Sufficient-To study the mechanism of binding of these cells to the CCBD, we used a recombinant Fn polypeptide containing 3Fn6 -3Fn10 and a mutated polypeptide lacking the recently described PHSRN synergy site (24) (Fig. 2). As shown in Fig. 3, K562 did not adhere to the mutated polypeptide and weakly to the control polypeptide. Only very low adhesion of MV3 cells was observed to the mutated polypeptide, whereas adhesion of MV3 to the control polypeptide was five times higher. In contrast, Mel57 cells adhered strongly to both polypeptides. To investigate if this difference was due to differential expression of ␣v␤3, we transfected MV3 cells with ␤3 cDNA, resulting in ␣v␤3 surface expression (Fig.  4A), and used these cells in adhesion assays. Expression of ␣v␤3 provided MV3 cells with the capacity to adhere to GRGDSP (not shown) and to the mutated polypeptide, and this adhesion could be inhibited by C17 anti-␤3 mAbs (Fig. 4B). A similar level of inhibition was found with LM609 anti-␣v␤3 (not shown).
From these results, we conclude that the differential requirement for the PHSRN synergy site for adhesion to RGD in the CCBD of MV3 versus Mel57 is due to the different binding mechanisms of ␣5␤1 versus ␣v␤3.
Stimulation of ␣5␤1-mediated RGD Binding with ␤1 mAbs, PMA, and Manganese-The fact that K562 did not adhere to the mutated polypeptide whereas MV3 did to a low extent (Fig.  3), even though both cell lines used ␣5␤1, suggested that binding of ␣5␤1 to RGD in 3Fn10 might depend on the activation state of the integrin. To investigate this, we treated both cell lines with 8A2 and TS2/16 stimulatory ␤1 mAbs, with Mn 2ϩ , or with PMA prior to using them in adhesion assays to a GRGDSP peptide. PMA had no effect, 8A2 and TS2/16 induced weak adhesion of K562 to GRGDSP, and treatment of MV3 cells with these mAbs resulted in 25% adhesion to GRGDSP (Fig. 5A). A control ␤1-integrin mAb AJ2 had no effect (not shown). The strong binding of Mel57 was not enhanced by 8A2 or TS2/16. Mn 2ϩ was less effective for K562 but induced adhesion of MV3 cells up to 35%. We performed adhesion inhibition assays to examine whether the effect of 8A2 and TS2/16 was due to activation of ␣5␤1 or to the recruitment of other integrins. Induced adhesion of K562 to GRGDSP in the presence of 8A2 (Fig. 5B) or TS2/16 (not shown) was blocked by mAbs to ␣5 and not by any of the other mAbs. In addition, even though 8A2 and TS2/16 may activate ␣3␤1, ␣4␤1, ␣5␤1, and possibly ␣v␤1 on MV3 cells, induced adhesion of MV3 to GRGDSP was completely blocked by mAbs to ␣5, whereas mAbs to ␣3, ␣4, and ␣v had no effect (Fig. 5B). The fact that the 4B4 anti-␤1 mAb did not inhibit adhesion in the presence of 8A2 is in line with the report that activating and inhibiting antibodies share a common epitope on the ␤1 subunit (55).
Thus, the strength of ␣5␤1 binding to RGD can be increased by Mn 2ϩ and by activating ␤1 antibodies.
Requirement for the PHSRN Synergy Site Depends on the Activation State of ␣5␤1-As stimulatory ␤1 mAbs and Mn 2ϩ induced ␣5␤1-mediated adhesion to GRGDSP, we hypothesized that the activation state of ␣5␤1 determines the requirement for the PHSRN synergy site for cell adhesion to the CCBD. Therefore, we treated K562 and MV3 cells with PMA, TS2/16, or Mn 2ϩ and allowed them to adhere to the mutated and control Fn polypeptides. TS2/16 and, to a lesser extent, Mn 2ϩ induced adhesion of K562 cells to the mutated polypeptide and enhanced adhesion to the control polypeptide (Fig. 6). PMA enhanced adhesion of K562 cells to the control polypeptide but had no effect on adhesion to the mutated polypeptide. For MV3 cells, no effect of PMA was observed, but the low adhesion to the mutated polypeptide was enhanced 5-fold by TS2/16 and Mn 2ϩ , resulting in a level of adhesion that was similar to that observed with the fully active control polypeptide.
The fact that in the presence of TS2/16 or Mn 2ϩ no difference was observed between the mutated and control polypeptide regarding adhesion of MV3 cells, whereas for K562 the mutated polypeptide was still poorly active, could suggest 1) that stimulation of MV3 cells resulted in recruitment of other RGDbinding integrins or 2) that ␣5␤1 on K562 cells was not maximally activated by these agents. To exclude possibility 1, we used mAbs to ␣3, ␣4, ␣5, ␣v, ␤1, ␤3, ␣v␤3, or the combination of these mAbs in the absence of anti-␣5 for inhibition of TS2/ 16-stimulated adhesion of MV3 cells to the mutated polypep- FIG. 5. Stimulation of ␣5␤1-mediated adhesion to GRGDSP. A, Cells were incubated in the absence (no) or in the presence of PMA, 8A2 or TS2/16 stimulatory ␤1 mAbs, or manganese (Mn) and allowed to adhere to wells coated with 20 g/ml BSA-GRGDSP. B, cells that had been previously incubated in the absence (x) or in the presence of 8A2 stimulatory ␤1 mAbs (8A2) were incubated in the absence (no) or in the presence of inhibitory anti-integrin mAbs as indicated and allowed to adhere to wells coated with 20 g/ml BSA-GRGDSP. Adhesion to BSA was less than 5%. One representative experiment of three is shown. tide. Stimulated adhesion was blocked by the anti-␣5 mAb and not by any of the other mAbs or their combination (Fig. 7), suggesting that induction of adhesion to the mutated polypeptide of MV3 by TS2/16 was due to activation of ␣5␤1 and not to recruitment of other integrins.
To investigate possibility 2, we incubated K562 cells with PMA, TS2/16, or Mn 2ϩ and the various combinations and allowed the cells to adhere to the mutated and control polypeptide. In the presence of the combination of TS2/16 and Mn 2ϩ , adhesion to the mutated polypeptide was more than half the level of adhesion to the control polypeptide (Fig. 8). PMA had no effect by itself on adhesion to the mutated polypeptide but enhanced adhesion to the control polypeptide more than 2-fold. Finally, in the presence of the combination of PMA, TS2/16, and Mn 2ϩ , the control and the mutated polypeptide were equally effective in promoting adhesion of K562 cells. This adhesion was blocked by ␣5 mAbs (not shown).
From these results, we conclude that requirement of the PHSRN synergy site for ␣5␤1-mediated adhesion to RGD in the CCBD depends on the activation state of ␣5␤1.

DISCUSSION
In line with earlier reports, we find that ␣v␤3 does not require the PHSRN site. We base this conclusion on 2 observations. First, Mel57 cells express ␣v␤3 and adhere equally well to all molecules tested containing RGD, i.e. GRGDSP, the mutated polypeptide lacking the synergy site 3Fn6 -10(SPSDN), the control polypeptide 3Fn6 -10, and Fn120 kDa. Second, the ␣v␤3 negative MV3 cells do not adhere to RGDcontaining ligands that lack the PHSRN site, and transfection with ␤3 cDNA resulting in ␣v␤3 surface expression leads to binding of these cells to GRGDSP and 3Fn6 -10(SPSDN).
These findings confirm and extend the observations that ␣v␤3 can be retained on an RGD column (56) whereas ␣5␤1 cannot (11). Furthermore, these data are in agreement with the recent report that ␣vand ␣3but not ␣5-containing integrins are bound by a column containing a Fn fragment lacking the synergy region (57). Similarly, it has been reported that ␣IIb␤3 but not ␣v␤3 binding to Fn can be inhibited by an 11-amino acid peptide from 3Fn9 that also contains the PHSRN sequence (25). Thus, RGD is sufficient for binding to Fn through ␣v␤3, whereas ␣5␤1 and ␣IIb␤3 require the synergy region for efficient binding to Fn (24,25).
The major conclusion from this study is that the requirement for the PHSRN synergy site for ␣5␤1-mediated adhesion to the CCBD depends on the activation state of ␣5␤1. This is based on three findings. First, stimulation of K562 cells that express only ␣5␤1, with Mn 2ϩ or stimulatory ␤1-integrin mAbs, in-duces adhesion to GRGDSP and 3Fn6 -10(SPSDN). Second, in the presence of the combination of PMA, TS2/16, and Mn 2ϩ , the mutated and control polypeptide are equally effective in promoting K562 cell adhesion. Third, treatment of MV3 cells with these agents induces adhesion to GRGDSP and enhances adhesion to 3Fn6 -10(SPSDN) to the level of adhesion to 3Fn6 -10, and this effect is completely blocked by antibodies to ␣5 but not by mAbs to ␣3, ␣4, or ␣v or the combination.
Even though the ␣v␤3-negative K562 and MV3 cells express similar levels of ␣5␤1, they differ dramatically in binding to Fn120 kDa through this receptor. The view of cell type-specific regulation of ␣5␤1 affinity proposed by O'Toole et al. (60) suggests that the default low affinity state of the integrin as observed in K562 is switched to a high affinity state in MV3. As a result, MV3 but not K562 cells bind strongly to Fn120 kDa. Our finding that K562 cells bind poorly to Fn120 kDa and that 8A2 increases that adhesion two to three times is in line with earlier findings (61). As expected, Mn 2ϩ and stimulatory ␤1 mAbs do not affect the strong adhesion of MV3 to Fn120 kDa. However, our findings demonstrate that these agents do in fact alter the avidity of ␣5␤1 in MV3 cells but that this change can only be observed in the absence of the PHSRN synergy site. One interpretation of these findings is that intracellular factors (induced by PMA for K562 and factors already present in MV3) can increase the affinity of ␣5␤1 to a level that RGD is recognized in the Fn molecule and that additional extracellular events are required for the final activation of ␣5␤1, leading to full adhesion to RGD in Fn. The synergy site could be involved in the last step by locking the RGD site in the ␣5␤1 binding pocket, and in the presence of TS2/16 or Mn 2ϩ that last step seems to be no longer required. Our finding that PMA enhances K562 adhesion to the control polypeptide, whereas by itself it has no effect on adhesion to the mutated polypeptide, is in line with this idea. Furthermore, the fact that K562 cells in the presence of TS2/16 bind strongly to the control polypeptide without the need for PMA demonstrates that optimal extracellular stimulation (the synergy site plus stimulatory ␤1 mAbs) can abbrogate the need for intracellular activation (PMA).
It is of interest that comparable observations have been reported for ␣4␤1 (35). Even though Jurkat and Ramos cells express an active form of ␣4␤1 in the sense that they are capable of binding to the CS1 domain of Fn, they only bind to a peptide containing the EILDV recognition sequence from CS1 in the presence of stimulatory ␤1 mAbs. The authors suggest that sequences may be present in the NH 2 -terminal portion of CS1 that strengthen ␣4␤1 binding to EILDV, although none have yet been identified. Thus, the presence of sites that synergistically enhance binding of integrins to their recognition sequence might be a general mechanism, and activation by stimulatory ␤1 mAbs and Mn 2ϩ may bypass the dependence on such sites. Our report, however, provides the first example of substitution of the function of a well characterized synergy site by agents that activate the integrins involved.
For leukocytes, stimulatory ␤1 mAbs also increase the affinity of ␣4␤1 for CS1 (35,62) and for VCAM-1 (29), and they can even induce ␣4␤1 binding to the RGDS sequence (34). MV3 cells adhere to CS1 and tumor necrosis factor ␣-stimulated endothelial cells in the absence of stimuli (not shown), indicating the expression of active ␣4␤1 on these cells. For MV3 cells in the absence or presence of stimuli, we do not observe any inhibition of binding to the CCBD with HP2/1 anti-␣4 mAbs, whereas these mAbs inhibit binding to CS1 (not shown). Thus, the reported recognition of RGD by stimulated ␣4␤1 does not play a role in our assay. This difference may be explained by the fact that Ramos cells, as used by Sanchez-Aparicio et al. (34), do not express ␣5␤1. In MV3, the effect of TS2/16 on ␣4␤1 FIG. 7. Inhibition of stimulated adhesion of MV3 to 3Fn6 -10(SPSDN) with integrin mAbs. MV3 cells were incubated in the absence or in the presence of TS2/16 and allowed to adhere to wells coated with 32 g/ml 3Fn6 -10(SPSDN). Inhibitory mAbs to integrin subunits were added as indicated. Adhesion to BSA was less then 3%. One experiment of three is shown.
FIG. 8. Stimulation of adhesion of K562 to recombinant Fn polypeptides. K562 cells were incubated in the absence or in the presence of various combinations of PMA, TS2/16, and Mn 2ϩ as indicated and allowed to adhere to wells coated with 32 g/ml 3Fn6 -10(SPSDN) (dotted bars) or 3Fn6 -10 (filled bars). Adhesion to BSA was less then 4%. One experiment of three is shown. may be masked by the binding to RGD through ␣5␤1. Alternatively, as previously reported (63), stimulatory ␤1 mAbs may selectively activate ␣5␤1 while leaving ␣4␤1 unaffected.
A possible interpretation of our findings may be that the PHSRN synergy site binds to the same epitope as recognized by the stimulatory ␤1 mAbs. It has been previously suggested that the epitope where these mAbs bind may physically interact with extracellular proteins (55). However, the fact that binding of K562 to 3Fn6 -10 can be enhanced in the presence of 8A2 or TS2/16 when PMA is absent indicates that the synergy site and stimulatory mAbs can have additional combined stimulatory effects. Therefore, these data seem to support a model where the synergy site and stimulatory mAbs have different binding sites on ␣5␤1. This is in agreement with the recent report that the mechanism of binding of integrin ␣5␤1 to Fn seems to be through binding of the ␣5 subunit to the synergystic regions and of the ␤1 subunit to RGD (64).
In conclusion, our data demonstrate that ␣5␤1 but not ␣v␤3 requires the PHSRN synergy site for cell adhesion to RGD in the CCBD of Fn but that induction of a high affinity state of ␣5␤1 with PMA, stimulatory mAbs, and/or Mn 2ϩ abrogates this dependence on the PHSRN sequence.