Monoclonal antibody 9EG7 defines a novel beta 1 integrin epitope induced by soluble ligand and manganese, but inhibited by calcium.

The monoclonal antibody 9EG7 has been previously found to recognize an epitope induced by manganese on the integrin β1 chain (Lenter, M., Uhlig, H., Hamann, A., Jeno, P., Imhof, B., and Vestweber, D.(1993) Proc. Natl. Acad. Sci. U. S. A. 90, 9051-9055). Here we show that treatment of β1 integrins with manganese or soluble integrin ligands (e.g. fibronectin and RGD peptide) induced the 9EG7 epitope. This epitope was also induced upon EGTA treatment to remove calcium, and the addition of calcium inhibited 9EG7 epitope induction by manganese or by ligand. Further emphasizing the importance of the 9EG7 epitope, the 9EG7 antibody itself stimulated adhesion mediated by multiple β1 integrins, and conversely, ligands for α2β1, α3β1, α4β1, and α5β1 all stimulated 9EG7 expression. Together these results support a model whereby (i) calcium inhibits β1 integrin function because it prevents the appearance of a conformation favorable to ligand binding and (ii) manganese enhances β1 integrin function because it induces the same favorable conformation that is induced by adding ligand, or removing calcium. Notably, other β1-stimulating agents (magnesium and mAb TS2/16) did not induce 9EG7 expression unless ligand was also present. Thus, although 9EG7 may reliably detect the ligand-bound conformation of β1 integrins, its expression does not always correlate with integrin “activation.” Finally, mouse/chicken β1 chimeric molecules were used to map the 9EG7 epitope to β1 residues 495-602 within the cysteine-rich region, and antibody cross-blocking studies showed that the 9EG7 epitope is distinct from all previously defined human β1 epitopes.

The study of variable integrin activation states has been facilitated by the use of monoclonal antibodies that selectively recognize distinct integrin conformations. Notably, the expression of several integrin epitopes is regulated by cell triggering with various agonists, divalent cation gain or loss, ligand binding, or combinations of these events (9, 11, 12, 14 -19). Also, several antibodies can stimulate integrin adhesive functions, presumably by stabilizing an active conformation (4, 12, 15, 20 -26). Interestingly, some antibodies not only recognize epitopes induced by ligand, but also stimulate integrin function themselves, whereas other antibodies apparently have only one or the other of these properties.
The mechanism for ligand binding and activation of ␤ 1 integrins has not been well studied, partly because few antibodies that specifically recognize activated or ligand-bound forms have been available. One possible "activation-specific" antibody (called 15/7) has been found to selectively recognize ␤ 1 on activated T cell subsets in vivo (27). Also, two recently described antibodies (9EG7 and SG/7) define ␤ 1 neoepitopes induced by divalent cations. The former recognizes an epitope up-regulated in response to Mn 2ϩ treatment (28), whereas the latter defines a ␤ 1 epitope induced by either Mn 2ϩ or Ca 2ϩ , but not Mg 2ϩ (29). However, the association of these epitopes with integrin functions has not been extensively studied. Also, there are two antibodies that recognize ligand-induced ␤ 1 epitopes (30,31). In other studies of ␤ 1 integrins, the inhibitory effects of Ca 2ϩ (32)(33)(34)(35)(36), and the stimulatory effects of manganese (34,35,(37)(38)(39) have often been noted, but few mechanistic insights have emerged.
Because little is known regarding ␤ 1 integrin conformational changes, we have utilized the 9EG7 mAb 1 to study events accompanying ␤ 1 integrin activation and ligand binding. We define "activation" as an increase in the potential of an integrin to bind ligand, and/or to mediate the more complex function of cell adhesion. We wish to clearly distinguish "activation" as a distinct phase that precedes "ligand binding." Here we have found that the 9EG7 antibody defines a ␤ 1 conformation of fundamental importance because (i) the epitope is negatively regulated by calcium, (ii) it is induced by manganese, (iii) it is induced by the binding of all ␤ 1 integrin ligands tested, and (iv) the 9EG7 antibody itself stimulates all ␤ 1 adhesive functions tested. Finally, we mapped the 9EG7 epitope to a site within ␤ 1 , distinct from that recognized by all other known anti-human ␤ 1 antibodies, and not previously shown to be regulated by divalent cations and ligand binding.
Cell Adhesion Assay-K562 cells were maintained in RPMI 1640 medium supplemented with 10% fetal bovine serum. Cell attachment to fibronectin was carried out as described previously (48). Briefly, fibronectin (5 g/ml) was coated onto 96-well microtiter plates (Flow Laboratories), and incubated overnight at 4°C. Then 0.1% heat-denatured BSA (hBSA) was added (45 min) to block nonspecific adhesion. Cells were labeled by incubation with the fluorescent dye BCECF-AM (Molecular Probes, OR) for 30 min and then washed sequentially with divalent cation-free PBS, 1 mM EDTA in PBS, and TBS (Tris-buffered saline: 24 mM Tris-HCl, pH 7.4, 137 mM NaCl, 2.7 mM KCl), before resuspending in TBS containing 0.1% hBSA and 2 mM glucose (assay buffer). Cells (5 ϫ 10 4 /well) were then added to each well in triplicate in the presence of various concentrations of divalent cations, mAb TS2/16, or PMA. After incubation for 25 min at 37°C, the plate was washed three times with TBS to remove unbound cells. Cells remaining attached to the plate were analyzed using a fluorescence analyzer (Cyto-Fluor 2300, Millipore Co.). After subtraction of background cell binding to hBSA-coated wells, values for cells bound/mm 2 were calculated as described (49).
Flow Cytometry-For flow cytometry, cells were washed once with TBS, and preincubated on ice for 10 min with 5% BSA in TBS containing 0.02% sodium azide (washing buffer) and, then where indicated, with divalent cations or other stimuli. Aliquots of 3 ϫ 10 5 cells were then incubated for 45 min on ice with primary antibodies (ascites at a final dilution of 1:250 or purified mAbs at a final concentration of 1 g/ml). Cells were washed three times with washing buffer and incubated for 30 min on ice with fluorescein isothiocyanate-conjugated goat anti-mouse (Life Technologies, Inc.) or anti-rat IgG (Sigma). Cells were washed three times and analyzed using a FACScan machine (Becton Dickinson, Oxnard, CA).
Immunoprecipitation-The mouse/chicken ␤ 1 integrin chimeras (MC3 and MC5) were expressed in NIH-3T3 cells as described previously (50). These cells were labeled with ImmunoPure NHS-LC-biotin (Pierce) and then lysed with extraction buffer (1% Nonidet P-40 in PBS containing 2 mM MgCl 2 , 1 mM phenylmethylsulfonyl fluoride, 2 g/ml aprotinin, and 2 g/ml leupeptin). After immunodepletion to remove proteins binding nonspecifically to protein A-Sepharose beads, mouse anti-chicken ␤ 1 antibodies were added and immune complexes were collected on protein A beads that had been preadsorbed with rabbit anti-mouse sera. Rabbit anti-␤ 5 immune complexes were directly collected on protein A beads. After four washings with extraction buffer, proteins were eluted, resolved by 7% SDS-polyacrylamide gel electrophoresis, tranferred to nitrocellulose, incubated with Extravidin-HRP and finally detected with the Renaissance chemiluminescent kit (Du-Pont NEN). For reprecipitation experiments, biotinylated ␤ 1 integrin MC3 and MC5 complexes were precipitated using the appropriate antichicken mAb, and then eluted at 70°C for 10 min in 1% Nonidet P-40 detergent. The eluates were then reprecipitated with mAb 9EG7 that had been preadsorbed onto protein A beads, and analyzed on 7% SDSpolyacrylamide gel electrophoresis as above.

Stimulation of Cell Adhesion by the 9EG7 mAb-
The observation that mAb 9EG7 detects a Mn 2ϩ -dependent ␤ 1 integrin conformation (28) prompted us to evaluate whether this conformation is functionally relevant for cell adhesion. For these experiments we used the erythroleukemia cell line K562, expressing the ␣ 5 ␤ 1 fibronectin receptor but no other ␤ 1 integrins. In the absence of any added divalent cations, K562 cells did not adhere to immobilized fibronectin, and the addition of the 9EG7 mAb was unable to induce adhesion (Fig. 1). However, in the presence of intermediate levels of Mn 2ϩ (0.01-0.05 mM), mAb 9EG7 stimulated strong dose-dependent adhesion (Fig. 1). The effect of 9EG7 was more evident as Mn 2ϩ was increased from 0.01 to 0.05 mM. However at higher Mn 2ϩ levels (Ͼ0.1 mM), Mn 2ϩ itself already had a maximal stimulatory effect, such that the effect of 9EG7 became less evident. In all cases, K562 cell adhesion to fibronectin was completely inhibited by the anti-␣ 5 mAb A5-PUJ2 (data not shown), consistent with mediation by ␣ 5 ␤ 1 . In another experiment, ␣ 3 -transfected K562 cells showed ␣ 3 -dependent adhesion to kalinin-containing matrix that was also stimulated by mAb 9EG7 (data not shown). These results suggest that the 9EG7 mAb may recognize and stabilize a functionally relevant integrin conformation. In this respect, 9EG7 resembles several other anti-␤ 1 antibodies that stimulate integrin adhesive function (4,23,24).
9EG7 Epitope Induction by Mn 2ϩ , but Not by Other Integrinstimulating Agents-We next compared the ability of various integrin-activating stimuli to induce the epitope. Treatment with Mn 2ϩ yielded a dose-dependent increase in 9EG7 expression from nearly 0% (relative to total ␤ 1 ), up to more than 30%, with maximal expression at 5.0 mM Mn 2ϩ and half-maximal at 1 mM ( Fig. 2A). Representative flow cytometry profiles are shown in Fig. 3 (upper panels). Notably, upon Mn 2ϩ stimulation, the amount of 9EG7 bound/cell was increased, relative to total ␤ 1 on K562 cells, with no evidence for cellular subpopulations. In a separate experiment, stimulation with Mn 2ϩ also caused increased cell adhesion to fibronectin, with maximal adhesion at Ն5.0 mM and half-maximal at 0.02-0.03 mM Mn 2ϩ ( Fig. 2A). Again, adhesion was completely inhibited by mAb A5-PUJ2 (data not shown). In contrast to Mn 2ϩ , Mg 2ϩ did not induce the 9EG7 epitope on K562 cells, even though Mg 2ϩ did induce ␣ 5 ␤ 1 -dependent cell adhesion to immobilized fibronectin, with maximal adhesion at ϳ4 mM and half-maximal at 1-2 mM Mg 2ϩ (Fig. 2B). Neither expression of the 9EG7 epitope nor adhesion to fibronectin was induced by Ca 2ϩ over a wide range of concentrations (0.01-25 mM; data not shown).
Similar to Mg 2ϩ , the stimulatory antibody mAb TS2/16 ( Fig.  2C) and the phorbol ester PMA (Fig. 2D) induced little or no 9EG7 expression, but did promote cell adhesion to fibronectin in a dose-dependent manner. Because integrin-mediated adhesion requires divalent cations, 1 mM MgCl 2 and 1 mM CaCl 2 were added during the adhesion assays and were also present during the analysis of 9EG7 epitope expression (Fig. 2, C and D). If MgCl 2 and CaCl 2 were omitted, TS2/16 and PMA still failed to induce 9EG7 epitope expression (data not shown).
Induction of the 9EG7 Epitope by Soluble Ligand-Next we tested the effects of soluble ligands on 9EG7 epitope expression. As seen in representative flow cytometry profiles (Fig. 3, lower panels) and in Fig. 4B, the ligand-mimetic peptide GRGDSP, but not the control peptide GRGESP, caused increased 9EG7 expression relative to total ␤ 1 . Also, the 9EG7 epitope was consistently induced by intact fibronectin at concentrations greater than 100 nM (Fig. 4A), and also by soluble collagen, kalinin, VCAM-1, and CS1 peptide on K562 transfectants expressing ␣ 2 ␤ 1 , ␣ 3 ␤ 1 , and ␣ 4 ␤ 1 , respectively (data not shown). To facilitate ligand binding, these experiments were carried out in the presence of 1 mM MgCl 2 and 1 mM CaCl 2 . Also, it was essential to utilize serum-free conditions, to avoid complications due to serum fibronectin. We conclude that both ligand occupation and Mn 2ϩ induce a similar ␤ 1 integrin conformation, recognized by 9EG7.
Induction of the 9EG7 Epitope by Soluble Fibronectin Is Potentiated by Mg 2ϩ and mAb TS2/16, but Not by PMA-For those agents (Mg 2ϩ , TS2/16, PMA) able to stimulate integrin adhesive function but not 9EG7 expression (Fig. 2), further experiments were carried out, testing their ability to synergize with a suboptimal dose of soluble fibronectin. As indicated in Table I, fibronectin (100 nM), Mg 2ϩ (5 mM), or TS2/16 (3 g/ml) added by themselves induced minimal 9EG7 expression. However, in the presence of 100 nM fibronectin, both MgCl 2 (5 mM) and TS2/16 (3 g/ml) each stimulated a marked increase in the 9EG7 epitope above the basal level. Also, the stimulatory effect of Mn 2ϩ , obvious even in the absence of fibronectin, was more pronounced when fibronectin was present. In sharp contrast, a relatively high dose of PMA did not induce 9EG7 expression, even in the presence of fibronectin.
Inverse Correlation between the Presence of Calcium and 9EG7 Expression-During preliminary experiments aimed at testing divalent cation effects on 9EG7 expression, we found that preincubation with EDTA itself caused an increase in the 9EG7 epitope. As shown in Fig. 5, incubation of K562 cells with EDTA yielded a dose-dependent increase in 9EG7 expression. At EDTA concentrations higher than 0.5 mM, the percentage of total ␤ 1 expressing the 9EG7 epitope was 25-30%, a level comparable to the maximal level induced by Mn 2ϩ . Preincubation with EGTA yielded essentially the same result as EDTA (Fig. 5), indicating that removal of Ca 2ϩ rather than Mg 2ϩ was most critical for causing 9EG7 epitope expression. Notably, the effect of EDTA was temperature-dependent (inset), since the high level of 9EG7 induction seen at 37°C was decreased at 20°C and negligible at 4°C.
Also, the effect of EGTA on 9EG7 epitope expression was fully reversible. Incubation of K562 cells with 2.5 mM EGTA (for 30 min at 37°C) caused elevated 9EG7 expression (from 5.2% up to 32.6%). Continued incubation for another 30 min in 2.5 mM EGTA, or after addition of buffer that slightly diluted the EGTA (to 2.3 mM), did not markedly alter 9EG7 expression. The experiments represented in panels C and D were carried out in the presence of 1 mM MgCl 2 and 1 mM CaCl 2 . Cells were stained with 9EG7 or mAb 13 and analyzed by flow cytometry to determine the percent of 9EG7 expression relative to total ␤ 1 stained by mAb 13 (left y axes). Bound TS2/16 did not interfere with detection of 9EG7 because they bind to nonoverlapping epitopes (see Table II and "Discussion"), and we used a fluorescein isothiocyanate-conjugated second antibody specific for rat (9EG7) but not mouse (TS2/16) primary antibodies. Cell adhesion was measured (in the absence of 9EG7) as indicated under "Materials and Methods" (right y axes). In contrast, subsequent addition of Ca 2ϩ (to 12.5 mM, for 30 min at 37°C) lowered 9EG7 expression back to a basal level (4.1% relative to total ␤ 1 , Fig. 6A). Remarkably, simply removing EGTA by washing cells and resuspending them in PBS for an additional 30 min also resulted in a loss of much of the 9EG7 epitope (down to 9.5%). Because this occurred even when the PBS was pretreated with Chelex 100 resin, we suspect that the calcium responsible for this effect is derived from the cell, rather than buffer contamination. This result is consistent with very tight binding of Ca 2ϩ to ␤ 1 integrins (␣ 5 ␤ 1 ) on K562 cells, as also suggested from Fig. 5.
Notably, 9EG7 expression was also elevated to a roughly similar extent upon preincubation with either 5 mM MnCl 2 or 25 M GRGDSP peptide (Fig. 6, B and C). Again, this expression was nearly completely reversed upon the addition of 12.5 mM Ca 2ϩ . When 2.5 mM EGTA was added subsequent to treatment with either 5 mM MnCl 2 or 25 M GRGDSP peptide, no further increase in the 9EG7 epitope was observed. These results suggest that Ca 2ϩ had already been depleted due to incubation with 5 mM MnCl 2 or 25 M GRGDSP peptide.
Detailed Ca 2ϩ titrations confirmed that the addition of Ca 2ϩ (at Ͼ0.1 mM) could reverse both the stimulatory effects of 5 mM Mn 2ϩ (Fig. 7A) and 25 M GRGDSP ligand (Fig. 7B). At lower doses of Mn 2ϩ and GRGDSP, the percent inhibition by Ca 2ϩ was even more pronounced (data not shown). In contrast, Mg 2ϩ had only a very minor inhibitory effect on Mn 2ϩ stimulation, and in fact, Mg 2ϩ exerted a slight to moderate stimulatory effect above that seen with 25 M GRGDSP alone.
Analysis of the 9EG7 Epitope-The majority of anti-␤ 1 antibodies, including both function enhancers (e.g. TS2/16) and blockers (e.g. mAb 13), have been mapped to a common epitope that includes residues 207-218 (42). Thus as expected, prein-cubation with mAb 13 did block TS2/16 binding, and similarly it blocked the binding of another interesting antibody, mAb 15/7 (27), that had not previously been mapped. Also as expected, mAb 13 did not block binding of three other antibodies (LM534, LM442, K20), that define two distinct epitopes near the ␤ 1 cysteine-rich region (Table II). In comparison, mAb 9EG7 did not block the common epitope seen by TS2/16, mAb 13, 15/7, SG/7, and many other anti-␤ 1 antibodies not listed here, and also did not block the less common epitopes defined by K20, LM534, or LM442 binding.
Because 9EG7 did not overlap strongly with any previously defined epitope, precise localization required a chimeric ␤ 1 mapping approach. For this, previously described mouse/ chicken chimeras (50) were utilized, since 9EG7 recognizes mouse ␤ 1 (28) but fails to recognize chicken ␤ 1 (data not shown). As shown in a reprecipitation experiment (Fig. 8), the mouse/chicken MC3 chimera was isolated from NIH-3T3 cells by the CSAT mAb (lane 1), and then reprecipitated using the 9EG7 mAb (lane 6), but not by a negative control antibody (lane 5). In contrast, 9EG7 did not recognize the MC5 chimera (lane FIG. 5. Effect of EDTA and EGTA on 9EG7 epitope expression. K562 cells were washed in PBS and then resuspended in TBS (ϩ5% BSA and 0.02% NaN 3 ) containing either EDTA or EGTA. After 30 min at 37°C cells were analyzed by flow cytometry to determine the percent of 9EG7 relative to total ␤ 1 expression. In parallel experiments cells were incubated with 2.5 mM EDTA for 30 min at the indicated temperatures (inset).

FIG. 4. Effect of fibronectin and RGD peptide on 9EG7 expression.
K562 cells were incubated (in the presence of 1 mM MgCl 2 and 1 mM CaCl 2 ) with either soluble human fibronectin for 30 min at 37°C (panel A) or GRGDSP and GRGESP peptides for 5 min at room temperature (panel B). The percent of 9EG7 relative to total ␤ 1 expression was determined by flow cytometry.  2 and 4), no chimeric ␤ 1 was present in the ␤ 5 lanes. Also, control experiments using polyclonal anti-␤ 1 sera confirmed that ␤ 1 integrins were indeed available for reprecipitation (lanes 7 and 9). As indicated in the schematic diagram (Fig. 8C), these results are consistent with 9EG7 binding to ␤ 1 between residues 495 and 602.

DISCUSSION
Here we have characterized a novel ␤ 1 integrin epitope that is highly relevant to ligand binding and adhesive functions. The 9EG7 epitope was induced by all ␤ 1 integrin ligands tested (GRGDSP peptide, fibronectin, soluble collagen, kalinin, VCAM-1, and CS1 peptide) and the antibody itself could stimulate the adhesive function of all ␤ 1 integrins tested, most likely by stabilizing a conformation favorable to ligand binding. In contrast to the previous report (28), we saw no evidence for blocking of ␤ 1 integrin-mediated adhesion.
Calcium and Ligand Effects-Our results suggest that constitutive expression of the 9EG7 epitope on ␤ 1 integrins is prevented by very tightly bound calcium. First, neither EGTA or EDTA was effective in promoting 9EG7 expression unless K562 cell temperature was elevated to 37°C. Second, elevated 9EG7 epitope expression was readily lost upon removal of EDTA or EGTA, presumably due to trace amounts of calcium derived from the cell itself. Assuming that calcium released from the cell could reach as high as 1-10 M, the presence of 300 M EGTA (enough to induce half-maximal 9EG7 expression) would chelate all but 1-10 nM free calcium (51). Thus, the apparent K i for calcium inhibition of 9EG7 binding (in the absence of Mn 2ϩ or ligand) could be as low as 1-10 nM.
Notably, binding of ligand to ␤ 1 integrin also triggered the appearance of the 9EG7 epitope. The addition of excess Ca 2ϩ could reverse this effect, but under these conditions, levels in the mM range were required. We assume that ligand binding is inhibited by this excess Ca 2ϩ because (i) this is far in excess of the amount needed to prevent 9EG7 binding in the absence of ligand, and (ii) binding of 9EG7 itself is not appreciably increased in affinity (data not shown). In this regard, there is recent biophysical evidence for a mechanism whereby bound ligand can displace ␤ 3 integrin cations (52) from a site (aa 118 -131) that is very well conserved in ␤ 1 (aa 129 -142), and also required for ligand binding to ␤ 1 integrins (53).
The inhibitory effect of Ca 2ϩ on several ␤ 1 integrin functions (32)(33)(34)(35)(36)54) and on some ␤ 3 integrin functions (55) has been well established. Now we have the fundamental new insight that Ca 2ϩ but not Mg 2ϩ may act largely by obstructing the appearance of a conformation (defined by 9EG7) that is favorable for ligand binding. In this regard, it is likely that the enhanced cell migration associated with a lower Ca 2ϩ /Mg 2ϩ ratio in wound fluid (56) probably involves a more favorable ␤ 1 ligand-binding conformation such as described here.
Mn 2ϩ Stimulation Effects-The induction of the 9EG7 epitope by Mn 2ϩ confirms results noted earlier (28). Just as for soluble ligand, the effects of Mn 2ϩ were almost completely reversed by the addition of Ca 2ϩ . Also, the addition of EGTA following induction by Mn 2ϩ or soluble ligand caused no further 9EG7 induction. Experiments analyzing Ca 2ϩ inhibition effects on Mn 2ϩ -induced 9EG7 expression yielded results consistent with non-competitive inhibition (data not shown). Thus, the interaction between Mn and Ca may be complex, with separate sites involved, as suggested for ␣ V ␤ 3 (55). In this regard, Mn 2ϩ but not Ca 2ϩ supported ␣ 5 ␤ 1 integrin clustering in a ligand-independent fashion. 2,3 It has been well established that Mn 2ϩ is a strong stimulator of ␤ 1 integrin function (34,35,(37)(38)(39). Now we gain a new insight into this activity since Mn 2ϩ not only supports ligand binding, but also, by itself, stabilizes an epitope favorable to ligand binding, and causes a diminished inhibitory effect of Ca 2ϩ . In contrast, Mg 2ϩ is able to support ligand binding, but does not appear to have these other activities. It has been suggested elsewhere (55) that Mn 2ϩ could reach 1-12 M in many tissues, and up to 50 M upon bone resorption. These levels approach the point at which Mn 2ϩ begins to stimulate ␣ 5 ␤ 1 -dependent cell adhesion and 9EG7 expression (i.e. see Fig. 2A).
Is 9EG7 a Detector of Activated Integrins?-While the 9EG7 antibody is a reliable detector of ligand binding, and itself can stimulate cell adhesion, the 9EG7 epitope does not closely correlate with integrin activation. To avoid confusion regarding the term activation, we have defined it here as the increased potential of an integrin to bind ligand or mediate adhesion. By this definition, 9EG7 does not define an activation epitope as previously suggested (28). First, agents such as EDTA and EGTA inhibit ligand binding, but nonetheless stimulated increased 9EG7 expression. Second, agents such as Mg 2ϩ and TS2/16 did not directly induce 9EG7 expression themselves (in the absence of ligand), but could facilitate 9EG7 expression indirectly, by enhancing ligand binding. This divergence between the effects of Mn 2ϩ and soluble ligand, compared to Mg 2ϩ and TS2/16, had not previously been recognized.
Rather than defining an activation epitope, the 9EG7 epitope can be better characterized as a "ligand-induced binding site" or "LIBS," such as have been described for ␤ 3 integrins (11, 57).
However, even the term LIBS is only partially correct, since in the absence of ligand, both the addition of Mn 2ϩ and removal of Ca 2ϩ also induce 9EG7 expression (see below).
Although the phorbol ester PMA stimulated adhesive activity, it did not directly stimulate 9EG7 expression, nor did it synergize with soluble fibronectin to give increased 9EG7 expression. We are not sure why PMA was previously found to cause increased 9EG7 expression (28), except that possibly peripheral blood T cells could differ from K562 cells in this regard. Nonetheless, our results are consistent with previous findings that PMA could stimulate adhesion without increasing ␤ 1 integrin ligand binding affinity (59,60). Rather than altering affinity, PMA could alter receptor clustering as previously reported (61), thus leading to enhanced avidity for ligand. Consonant with this, reduced clustering of ␣ 4 ␤ 1 integrin did not change 9EG7 expression, but caused markedly diminished cell adhesion. 3 Likewise, ␣ chain cytoplasmic tail deletion did not change Mn 2ϩ -inducible 9EG7 expression, despite causing diminished adhesive function (62).
All of these results reinforce the idea that there are at least three ways to increase the functional potential of ␤ 1 integrins. First, agents such as Mn 2ϩ increase ligand binding potential as well as 9EG7 expression; second, agents such as Mg 2ϩ and TS2/16 increase ligand binding potential without directly inducing 9EG7 expression; and third, agents such as PMA can stimulate cell adhesion without either stimulating ligand binding or inducing 9EG7 epitope expression.
In preliminary data, we have found that increasing doses of Mn 2ϩ or GRGDSP peptide uniformly increased the number of 9EG7 binding sites (relative to total ␤ 1 ), without appreciably altering the apparent K d (ϳ10 nM) for 9EG7 binding to ␤ 1 (data not shown). This result is consistent with each individual receptor existing in either a "ϩ" or "Ϫ" state with regard to the 9EG7 epitope. Notably, we have never observed 9EG7 expression to reach more than 40 -50% relative to the total ␤ 1 expression, when measured either by flow cytometry or by immunoprecipitation from solution (data not shown). In this regard, several other antibodies that report ligand-induced or activated integrin conformations also fail to bind to more than 50% of the integrins (9,17,55). As yet, no adequate explanation for this widely observed phenomenon has been offered. Possibly, it may be an intrinsic property of integrins to exist in multiple conformations, with it being very difficult to shift equilibrium completely toward the conformation resembling the ligand-bound state.
Location of the 9EG7 Epitope-Most anti-human ␤ 1 inhibitory and stimulatory antibodies (42), the Mn 2ϩ -inducible anti-␤ 1 antibody 15/7 (27), the Mn 2ϩ -and Ca 2ϩ -inducible mAb SG/7 (29), and the ligand-inducible 8A2 (31) and 12G10 (30) antibodies all map to a narrow region including ␤ 1 residues 207-218 ( a K562 cells were pretreated for 10 min at 20°C in the presence of 5 mM Mn 2ϩ with either control buffer, or mAb 9EG7 or mAb 13 (both rat antibodies) at 10 g/ml. Cells were then incubated with saturating concentrations of the indicated mouse anti-␤ 1 mAbs for 45 min on ice and subsequently stained with FITC-conjugated anti-mouse second antibody that did not crossreact with the rat mAb 13 and 9EG7 primary antibodies.
FIG. 8. Epitope mapping for mAb 9EG7. A, ␤ 1 chimeras were precipitated from NIH-3T3 cells using anti-chicken ␤ 1 antibodies CSAT (lane 1) and W1B10 (lane 3), respectively. As a control, mouse ␤ 5 was precipitated from the same lysates (lanes 2 and 4). B, reprecipitation of material precipitated by CSAT (lanes 5-7) or by W1B10 (lanes 8 and 9) was carried out using either negative control mAb P3 (lane 5), mAb 9EG7 (lanes 6 and 8) or an antiserum to the cytoplasmic domain of ␤ 1 (lanes 7 and 9). Note that the diminished intensity of the ␤ 1 bands seen by the anti-␤ 1 tail serum is likely due to competition by a pool of immature ␤ subunit that is not surface labeled. C, a schematic diagram of the MC3 and MC5 ␤ 1 chimeras is shown. region, thus becomes perhaps the first known cation-regulatable and/or ligand-inducible epitope to map elsewhere in ␤ 1 . Furthermore, although 9EG7 mapped to an area in the cysteine-rich domain partly overlapping the region containing the K20 and LM442/LM534 sites, 9EG7 did not cross-block those antibodies, further indicating that it recognizes a novel human ␤ 1 epitope. Comparison of mouse (64), human (65), and chicken (66) ␤ 1 sequences within the 495-602 region reveals only 6 positions (all between 577-599) where chicken ␤ 1 , which is not recognized by 9EG7, shows non-conservative differences from both the mouse and human sequences, which are recognized by 9EG7. However, 9EG7 binding to that subregion remains to be formally proven by further mutagenesis experiments. Also, it remains to be determined whether the anti-chicken antibodies TASC, and G, mapping to the 495-602 region (50), recognize the chicken equivalent of the 9EG7 epitope. Binding of neither the TASC nor G antibodies has yet been reported to be regulated by divalent cations or ligand, although the TASC mAb does induce ␤ 1 adhesive function (4).
Together these results emphasize that conformational changes due to ligand and divalent cation binding events are not limited to a small area within the primary structure. Rather, our results help to demonstrate that profound changes occur throughout the integrin molecule upon ligand binding and/or cation manipulation. This point is additionally reinforced by several anti-integrin ␣ chain epitopes that also are both induced by ligand, and regulated by divalent cations (e.g. Refs. 11,18,69,70).
In conclusion, the 9EG7 antibody defines a novel epitope near the cysteine-rich region of the ␤ 1 integrin, now known for the first time to be conformationally involved in both ligand and cation regulation. As such, 9EG7 has been exceptionally useful. For example, it has allowed us to demonstrate that a fundamental role for Ca 2ϩ may be to counterbalance those agents (including ligand) and Mn 2ϩ that stimulate 9EG7 expression. In addition, it has allowed us to elucidate three cat-egories of function-activating stimuli. Finally, the appearance of the 9EG7 epitope not only provides a useful tool for studying the regulation of ␤ 1 integrin function, but also it implies that ligand binding may expose this epitope, which could potentially play a role in interactions with other integrins or other proteins, and ultimately in outside-in signaling.