Two Functional Active Conformations of the Integrin α2β1, Depending on Activation Condition and Cell Type*

For several integrins, the existence of multiple conformational states has been studied intensively. For the integrin α2β1, a major collagen receptor on platelets and other cell types, however, no such experimental data were available thus far. Recently, our group has developed a monoclonal antibody IAC-1 sensitive to the molecular conformation of α2β1 because it only binds to the activated state of α2β1 on platelets, induced upon inside-out signaling. By investigating IAC-1 binding in combination with collagen binding after inside-out stimulation and outside manipulation, we demonstrated the existence of three different conformations of α2β1 on platelets and Chinese hamster ovary cells as follows: (i) a nonactivated, resting state with no collagen nor IAC-1 binding; (ii) an intermediate state, induced by outside manipulation, with collagen but no IAC-1 binding; and (iii) a fully activated state, induced after inside-out stimulation, with both collagen and IAC-1 binding. Moreover, these different conformational states of α2β1 are dependent on the cell type where α2β1 is expressed, as IAC-1 binding to peripheral blood mononuclear cells and Jurkat cells could also be induced by outside manipulation, in contrast to platelets and α2β1-expressing Chinese hamster ovary cells. Finally, we revealed a functional relevance for these different conformational states because the conformation of α2β1, induced after outside manipulation, resulted in significantly more cell spreading on coated collagen compared with nonactivated or inside-out stimulated cells.

Integrins belong to a family of heterodimeric (␣␤) surface receptors that play a major role in cell-cell and cell-matrix interactions (1). Each subunit typically contains a relatively large extracellular domain, a single-pass transmembrane domain, and a short cytoplasmic tail (2). Integrins can become activated and transduce signals across the plasma membrane in both directions; intracellular binding of cellular components to the integrin cytoplasmic domains activates the ligand binding capacity of the extracellular domain (insideout signaling), whereas ligand binding to the extracellular domain or other stimuli acting at the outside of the integrin (outside manipulation) can also lead to conformational alterations that then again can result in the activation of intracellular pathways (outside-in signaling) (3). Thus, integrin receptors are able to switch from an inactive (low affinity) to an active (high affinity) state, and vice versa, in response to binding events taking place at both the ligand binding area and the cytoplasmic domains. Generally, it is accepted that intersubunit associations at the cytoplasmic and transmembrane domains maintain integrins in the low affinity state and that interruption of this association results in activation of the integrin (4,5). In particular, for several integrins like ␣IIb␤3, ␣2␤1, ␣L␤2, and ␣M␤2, it has been demonstrated that a conserved cytoplasmic GFFXX sequence of the ␣ subunit modulates inside-out activation by acting as a hinge, keeping the integrin in an inactive conformation (6 -8).
Significant progress is currently made in demonstrating that integrins contain a number of flexible joints and connections, leading to a broad spectrum of possible conformational states with distinct effects on the overall ligand binding affinity of the receptor (9). For the I domain lacking integrins ␣V␤3 and ␣IIb␤3, three different conformational states have been demonstrated by electron microscopy and crystal structures as follows: (i) a closed conformation (resting state), not able to bind its natural ligand; (ii) an extended conformation with closed headpiece (low affinity state); and (iii) an extended conformation with open headpiece (high affinity state) (10 -13). These two extended, active conformations are both favored by outside-in and inside-out signaling and can bind their natural ligand. The integrins ␣4␤1 and ␣5␤1 also exhibit multiple affinity states on the cell surface depending on the activation condition, similar to that described for ␣V␤3 and aIIb␤3 (14,15). For the I domain containing integrins ␣L␤2 and ␣M␤2, the existence of such conformational states has been demonstrated as well. In the study by Jin et al. (16), molecular dynamic simulations, with pulling forces applied to the ␣7 helix of the I domain, were used to reproduce intermediate and open conformations of the integrin ␣L subunit, in agreement with the crystal structures described previously (17).
The integrin ␣2␤1 (also designated VLA-2, GPIaIIa, or CD49b/ CD29) is an I domain-containing integrin, expressed on a variety of cell types including platelets, white blood cells, endothelial cells, and fibroblasts (18 -20). This integrin acts mainly as a collagen receptor on platelets (21), and on other cell types it can serve both as a collagen and laminin receptor (22). On platelets, ␣2␤1 plays an important role in primary hemostasis by binding to collagen in the subendothelium exposed after vessel injury (23). On white blood cells, ␣2␤1 is thought to be involved in tissue injury and inflammation situations (18,19). More recently, ␣2␤1 on activated T lymphocytes and Jurkat cells was found to influence Fas-induced apoptosis (24).
For ␣2␤1, studies on the existence of different conformational states have been hampered, mainly because until recently no antibodies were available that specifically recognize the activated or ligand-bound forms of ␣2␤1. Our group developed a monoclonal antibody (mAb), 3 IAC-1, that recognizes an epitope within the ␣2 I domain on platelets, hidden in the resting state but accessible when platelets are stimulated with agonists such as thrombin, ADP, or the snake venom convulxin (25). Moreover, because this antibody does not interfere with platelet collagen binding, it defines a new class of antibodies that is distinct from those belonging to the "ligand-induced binding sites" and the "ligand mimetic" groups.
The objective of this study was to use IAC-1 to provide further insight into the conformational states of ␣2␤1 after inside-out and outside manipulations and to do this on four different cell types (platelets, ␣2␤1-expressing CHO cells, PBMC, and Jurkat cells). Moreover, the functional relevance of these different conformational states of ␣2␤1 was evaluated.
Monoclonal antibodies and soluble human collagen type I (type VIII, Sigma) were labeled with FITC according to the manufacturer's instructions (Pierce). The FITC labeling of collagen was performed in 50 mM borate buffer, pH 8.5, after which the excess of fluorescent dye was removed by overnight dialysis against PBS. As a consequence of this treatment, the resulting collagen-FITC consists of reconstituted collagen fibrils.
Platelets and peripheral blood mononuclear cells (PBMC) were isolated from the blood of healthy volunteers who did not use acetylsalicylic acid for the last 10 days. Blood was taken on acid/citrate/dextrose (10:1.5 v/v), and platelets were washed by several centrifugation steps essentially as described previously (27). Platelets were finally resuspended at a concentration of 3 ϫ 10 5 cells/l in Hepes/Tyrode (13.7 mM NaHCO 3 ; 0.2 mM KCl; 0.1 mM MgCl; 0.03 mM Na 2 HPO 4 ; 1.2 mM NaHCO 3 ; 0.5 mM Hepes; 0.01% (w/v) glucose; pH 7.4) containing 0.3% BSA. For the isolation of PBMC, blood was taken on low molecular weight heparin (250 units/ml final, Clexane, Aventis Pharma, Brussel, Belgium) and separated on Ficoll-Paque (Amersham Biosciences) following the manufacturer's instructions. PBMC were finally resuspended at a concentration of 5 ϫ 10 5 cells/ml in MEM (Invitrogen).
For the activation experiments, ␣2␤1-expressing CHO cells and Jurkat cells were harvested, washed with XL buffer (13.7 mM NaCl; 0.5 mM KCl; 0.1 mM MgCl 2 ; 0.5 mM Hepes; 0.01% (w/v) glucose; pH 7.4), and resuspended at 5 ϫ 10 5 cells/ml, except where indicated otherwise. Platelets and PBMC were resuspended in their respective buffer or medium as described above. Reaction mixtures of cells (50 l in a total of 100 l) and an agonist were incubated for 30 min at room temperature. The agonists Mn 2ϩ and Co 2ϩ were used at a final concentration of 1 mM DTT at a final concentration of 10 mM. The peptide Ac-KRFYV-VMWK-NH 2 and the cell-permeable peptides were used at a final concentration of 50 and 200 M, respectively. Where indicated, platelets were also incubated with the antagonist PGE 1 at 5 M final concentration. After washing twice with an excess of their respective buffer, cells were used for further experimentation.

Peptide Design
All peptides (see  Sequences and analytical data of the synthetic peptides Sequences are given in the single-letter amino acid code. Peptides were synthesized and purified as described under "Materials and Methods." Ac stands for acetylated. high performance liquid chromatography (RP-HPLC; Jasco Co.), and all peptides were characterized by electrospray ionization-mass spectrometry (Bruker Daltonics Esquire 3000 plus) and amino acid analysis. Analytical data of peptides showed more than 95% purity. The agonist peptide Ac-KRFYVVMWK-NH 2 , based on the binding domain of thrombospondin-1 to the CD47 receptor, was used to stimulate ␣2␤1 on different cell types via inside-out signaling (29,30). Cellpermeable chimeric peptides were designed by placing the hydrophobic delivery sequence VTVLALGALAGVGVG to the human ␣2 cytoplasmic tail sequence KLGFFKR, as described by Wang et al. (31). The control peptide, consisting of the non-sense sequence TKSKYNE, was used as a negative control (31).

Flow Cytometry
Cells were incubated in the presence or absence of an activating agent for 30 min as described above. After washing, cells were incubated with either 10 g/ml IAC-1-FITC or 40 g/ml collagen-FITC (final concentration) for 30 min at room temperature. Afterward, cells were washed in their respective buffer or medium and finally resuspended in 400 l of PBS, supplemented with 7-aminoactinomycin (Molecular Probes, Eugene, OR) diluted 1:400. When an indirect staining was used, cells were incubated with 10 g/ml of primary antibody for 30 min at room temperature, followed by incubation with a polyclonal rabbit-antimouse IgG-FITC (diluted 1:100) for 30 min at room temperature. Cells were immediately analyzed by flow cytometry (Beckman Coulter, Miami, FL), and the mean fluorescence intensity (MFI) was determined (10,000 cells for each condition).

Cell Adhesion Assays
Cell Adhesion to Collagen Under Static Conditions-Wells of a microtiter plate (Greiner Bio-One GmbH, Frickenhauser, Germany) were coated with human collagen type I (25 g/ml in PBS) overnight at 4°C. Wells were blocked with 3% milk powder in PBS for 2 h at room temperature. ␣2␤1-Expressing CHO cells or Jurkat cells were preincubated for 30 min in the presence or the absence of an activating agent. After washing with XL buffer, a dilution series of activated and nonactivated cells was added to the wells and incubated for 0, 5, 10, 20, 30, 60, and 120 min at room temperature. Cell adhesion was determined by measuring endogenous phosphatase activity colorimetrically at 405 nm, after lysis of the cells with Triton X-100, using p-nitrophenyl phosphate (Merck) as a substrate (32). Wells were washed six times after coating and blocking and three times elsewhere. All assays were performed in duplicate and repeated at least three times.
Cell Attachment and Spreading Assay-Cell attachment and spreading was studied as described previously with modifications (33). Coverslips (18 ϫ 18 mm, Menzel-Glaser, Braunschweig, Germany) were coated with human collagen type I (25 g/ml in PBS) overnight at 4°C. Coverslips were blocked with 1% BSA and 0.1% glucose in Hepes/Tyrode buffer for 30 min at room temperature. Here the ␣2␤1-expressing CHO cells and Jurkat cells were used at a concentration of 6 ϫ 10 5 cells/ml in Iscove's medium and were preincubated for 30 min in the presence or the absence of an activating agent. Coverslips were washed twice with Hepes/Tyrode buffer before the addition of 300 l of cell suspension and incubated for either 10 or 60 min at room temperature. Coverslips were thoroughly washed to remove unbound cells and then fixed with 2% paraformaldehyde in PBS for 30 min at room temperature. After washing, coverslips were stained with Mayer's hematoxylin (2 min), followed by Fuchsin Ponceau (5 min), and mounted with DPX Mountant (Canemenco and Marivac Inc., Quebec, Canada). Adhesion to BSA-coated coverslips and adhesion of CHO-dhfr ϩ cells to collagen-coated coverslips, used as negative controls, never exceeded 0.2-1.0% of the cell input.
Double Immunofluorescence Labeling of Actin and ␣2␤1-Four-well LabTek cover glasses (Nunc A/S, Roskilde, Denmark) were coated with human collagen type I and blocked as described above. The ␣2␤1-expressing CHO cells, at 6 ϫ 10 5 cells/ml in Iscove's medium, were preincubated for 30 min in the presence or absence of an activating agent, added to the cover glasses, and incubated for 60 min at room temperature. After washing, cells were fixed with 2% paraformaldehyde for 20 min at 37°C, followed by permeabilization in a 0.1% solution of Triton X-100 for 5 min at room temperature. Cells were incubated with 200 nM phalloidin/Texas Red to stain actin and 10 g/ml IAC-1-FITC to stain activated ␣2␤1 for 1 h at 37°C. As a positive control, cells were first incubated with 5 g/ml AK7 for 1 h, followed by polyclonal rabbit anti-mouse IgG-FITC (diluted 1:50) for 1 h. As a negative control, 5 g/ml MOPC-21-FITC was used. After washing, cells were mounted with Prolong Gold antifade with 4,6-diamidino-2-phenylindole (Molecular Probes).
Analysis was performed with an inverted Nikon Eclipse TE200 microscope connected to an Image Analyzer (Lucia, Laboratory Imaging Ltd., Analis, Namur, Belgium). Quantitative results were obtained by determining the number of adherent cells by light microscopy, and fluorescence microscopy was used to visualize actin and ␣2␤1, using standard emission and excitation filters.
The number of adherent cells per coverslip was calculated as follows: (number of adherent cells at time point x/number of cells added per coverslip) ϫ 100. Adherent cells are defined as cells that are immobilized to the substrate, independently of cell morphology. Cell morphol- ogy was used to distinguish between attached and spread cells. Attached cells are defined as cells with a rounded morphology, whereas spread cells had a clear elongated morphology with clearly visible cytoplasm around the nucleus (34).
Statistical Analysis-Student's t test for paired data were used to test statistically significant differences. Data given are the mean Ϯ S.E.

Inside-out Signaling and Outside Manipulation Result in Different
Conformational States of ␣2␤1-We demonstrated previously that inside-out activation of human platelets by convulxin, for example, resulted in exposure of both the collagen-binding sites in ␣2␤1 and a neo-epitope of the receptor, recognized by the mAb IAC-1 (25). Here we extended this observation by manipulating the inside of the cell by using a cell-permeable peptide containing the KLGFFKR sequence of the human ␣2 cytoplasmic tail sequence (31). Addition of this peptide induced activation of ␣2␤1 on the platelet membrane as demonstrated by a significant increase in collagen binding (p Ͻ 0.05) (Fig. 1). This binding was specific because of the following: (i) addition of the control non-sense peptide had no effect on collagen binding (Fig. 1), and (ii) addition of the inhibitory anti ␣2-I domain mAb 15D7 could reverse the collagen binding (data not shown). The inside manipulation by the cellpermeable peptide also resulted in a significant increase in IAC-1 binding (p Ͻ 0.05) (Fig. 1). Addition of this peptide did not result in an up-regulation of ␣2␤1, because no increase in MFI was observed when staining with the control mAbs Gi9 and AK-7, nor was a significant binding of the unrelated mouse anti-human IgG1 MOPC-21 observed (data not shown). Besides, addition of this peptide resulted in an overall activation of platelets as determined by P-selectin exposure (Fig. 2). However, addition of PGE1 before addition of the peptide still induced collagen and IAC-1 binding without the exposure of P-selectin (Fig. 2), demonstrating that addition of the peptide per se induces conformational changes of ␣2␤1, independent of general cell activation.
Most interestingly, by using this cell-permeable peptide we could for the first time induce inside-out activation of ␣2␤1 on CHO cells. Indeed, the activated receptor on CHO cells bound both collagen and IAC-1 (Fig. 1). By using this heterologous cell line, we could therefore further prove the specificity of IAC-1 for activated ␣2␤1.
Next, we investigated the binding of collagen and IAC-1 to platelets and ␣2␤1-expressing CHO cells after outside manipulation with divalent cations (Mn 2ϩ and Co 2ϩ ), the reducing agent DTT or TS2/16, an activating anti-␤1 antibody, and found that all resulted in a significant increase in collagen binding, indicating that ␣2␤1 was activated (p Ͻ 0.05, Fig. 3). Most interestingly, this outside manipulation was not accompanied by an increase in IAC-1 binding (Fig. 3). The different outside manipulations did not result in an up-regulation of ␣2␤1 on the cells, as determined by Gi9 or AK-7 binding, nor was P-selectin exposed (data not shown). Hence, because these results demonstrate that the neo-epitope of IAC-1 is not exposed after outside manipulation in contrast to inside-out signaling, we presumed that at least two different active conformations of ␣2␤1 exist.
The Different Conformational States of ␣2␤1 Regulate Adhesion and Spreading to Collagen-To evaluate whether these two different active conformations have a functional role in cell adhesion, the adhesion of ␣2␤1-expressing CHO cells to coated collagen type I was studied under static conditions.  First, the time course of cell adhesion to coated collagen type I was determined. Adhesion of nonactivated ␣2␤1-expressing CHO cells to collagen increased with time and reached a maximum at around 60 min (Fig. 4A). When the cells were stimulated with either Mn 2ϩ (outside) or the cell-permeable ␣2-cytosolic peptide H-VTVLALGALAGV-GVGKLGFFKR-NH 2 (inside-out), cell adherence was significantly increased (p Ͻ 0.05, Fig. 4A). Most interestingly, no significant difference in adhesion of ␣2␤1-expressing CHO cells was observed between outside and inside-out-stimulated cells, indicating that adhesion of the two active conformations of ␣2␤1 to coated collagen occurs with the same efficiency (Fig. 4A).
Next, the effect of outside and inside-out stimulation on cell spreading was evaluated as described before (33). Therefore, ␣2␤1-expressing CHO cells, stimulated or not, were allowed to adhere to collagen-coated coverslips for 10 or 60 min, after which the cells were fixed and stained. Both outside and inside-out manipulation gave similar adhesion results as observed above (data not shown). Moreover, the adhesion to collagen was ␣2␤1-specific because (i) CHO-dhfr ϩ cells did not adhere to collagen, and (ii) preincubation with the inhibitory mAb 15D7 almost completely abolished adhesion (data not shown). When the adherent cells were analyzed in more detail, an interesting observation was made. At 10 min, almost all adherent cells were simply attached to collagen, independent of activation, with 94 Ϯ 1.3% of attached nonactivated cells and 98 Ϯ 2.1% attached cells after outside and 96 Ϯ 1.9% after inside-out stimulation, respectively. At 60 min, however, a significant proportion of the adherent ␣2␤1-expressing CHO cells stimulated with Mn 2ϩ (outside) obtained a spread morphology in contrast to nonactivated and inside-out stimulated ␣2␤1-expressing CHO cells (p Ͻ 0.05; Fig. 4B). These results indicate that the different active conformational states, induced after inside-out and outside manipulation, have a functional relevance in adhesion and spreading. Finally, we could demonstrate by fluorescent images that spread ␣2␤1-expressing CHO cells, obtained after outside manipulation, did not stained positive for IAC-1 in contrast to attached ␣2␤1-expressing CHO cells, obtained after inside-out manipulation (Fig. 5). Staining of ␣2␤1-expressing CHO cells with the unrelated mouse anti-human IgG MOPC-21 did not result in a fluorescence signal (data not shown), and staining with the anti-a2 I domain mAb AK7 always resulted in a positive FITC signal, independent of stimulation (Fig. 5).
The Different Conformational States of ␣2␤1 Are Cell Typedependent-Because the collagen receptor ␣2␤1 is also present on other cell types, we included both PBMC and Jurkat cells to substantiate our discovery of several active conformational states of ␣2␤1.
We started by inducing inside-out signaling on these cell types with the synthetic peptide Ac-KRFYVVMWK-NH 2 , known to activate indirectly ␣2␤1 on platelets and smooth muscle cells and ␣V␤3 on white blood cells by binding to the thrombospondin-1 receptor CD47 (29,30,35). As expected, incubation with Ac-KRFYVVMWK-NH 2 resulted in a significant increase of both collagen and IAC-1 binding on platelets from 1.8 Ϯ 0.8 to 29.7 Ϯ 1.9% and from 1.5 Ϯ 0.9 to 24.5 Ϯ 1.5%, respectively (p Ͻ 0.05, Fig. 6A). Moreover, addition of Ac-KRFYV-VMWK-NH 2 to PBMC was also able to induce both collagen and IAC-1 binding (Fig. 6), demonstrating that CD47 on white blood cells cannot only modulate ␤3 integrins but also the ␤1 integrin ␣2␤1. This peptide, however, was not able to activate ␣2␤1 on Jurkat cells because no increase in collagen nor IAC-1 binding was observed (Fig. 6). This is  Inside-out or outside stimulated ␣2␤1-expressing CHO cells were allowed to adhere to reconstituted human collagen type I coated cover glasses for 60 min. After fixation and permeabilization, cells were incubated with phalloidin-Texas Red to stain actin and IAC-1-FITC to stain activated ␣2␤1. Incubation of stimulated cells with AK7, followed by rabbit-anti-mouse IgG-FITC, was used as a positive control to stain ␣2␤1 Representative images are shown of attached cells (inside-out) and spread cells (outside). Bar, 10 m.

Two Functional Conformations of the Integrin ␣2␤1
NOVEMBER 4, 2005 • VOLUME 280 • NUMBER 44 consistent with previous reports where it has been suggested that the function of CD47, although highly expressed on Jurkat cells (36), appears to be integrin-independent (37,38).
Next, we also used the cell-permeable ␣2-cytosolic peptide H-VTV-LALGALAGVGVGKLGFFKR-NH 2 to activate ␣2␤1 on PBMC and Jurkat cells via inside-out signaling. For both cell types, this activation resulted in a significant increase in collagen and IAC-1 binding (p Ͻ 0.05; Fig. 7), similar to that described for platelets and ␣2␤1-expressing CHO cells. Furthermore, the inside-out stimulation by peptides resulted in MAC-1 activation on PBMC, similar to the P-selectin exposure described for platelets (data not shown).
Just as for platelets and ␣2␤1-expressing CHO cells, we stimulated PBMC and Jurkat cells via outside manipulation. Here we only used divalent cations for activation because TS2/16 had no effect on these cell types, and addition of DTT resulted in a marked down-regulation of ␣2␤1, as determined by AK-7 binding (data not shown). The latter is consistent with previous reports describing that DTT can affect integrin expression on lymphocytes (39). Activation of ␣2␤1 on PBMC and Jurkat cells by the divalent cations Mn 2ϩ and Co 2ϩ resulted in a significant increase in collagen binding, indicating that ␣2␤1 was activated (p Ͻ 0.05; Fig. 8). Most surprisingly, and in striking contrast to platelets and ␣2␤1-expressing CHO cells, this outside activation of ␣2␤1 was accompanied by a significant increase in IAC-1 binding (p Ͻ 0.05; Fig. 8). Next, in line with ␣2␤1-expressing CHO cells, a time course of adhesion to coated collagen type I was determined for the Jurkat cells. Also, a significantly increased cell adhesion after stimulation was observed, with no difference between outside and inside-out manipulation (Fig. 9A). Evaluation of the cell morphology of the adherent Jurkat cells revealed at every time point and after every stimulation merely attached cells (Fig.  9B). Taken together, these results demonstrated that the different conformational states of ␣2␤1 are cell type-dependent and confirmed their functional relevance in adhesion and spreading.

DISCUSSION
During the last years, much progress has been made in unraveling the complexity of the broad spectrum of possible conformational states in integrins (9,40,41). Collective observations from electron microscopy, crystal structures, mutational introduction of disulfides, and other functional studies suggest that integrins may assume at least three different activation states, represented by different quaternary conformations as follows: (i) a closed conformation (resting state), not able to bind its natural ligand; (ii) an extended conformation with closed headpiece (low affinity state); and (iii) an extended conformation with open headpiece (high affinity state) (10,(15)(16)(17)42). These two extended, active conformations are both favored by outside-in and inside-out signaling and can bind their natural ligand.
For the integrin ␣2␤1, a major collagen receptor on platelets, no such experimental data were reported thus far, mainly because until recently, no antibodies were available that specifically recognize activated or ligand-bound forms. We have developed mAb IAC-1, sensitive to the molecular conformation of ␣2␤1 because it only binds to the activated form of ␣2␤1 on platelets, induced via inside-out signaling (25). The aim of the this study was to use this mAb IAC-1 to evaluate the effects of different inside-out and outside manipulations on the conformational state of ␣2␤1.
First, the effect of inside-out and outside manipulations of ␣2␤1 on platelets and CHO cells was investigated. For inside-out signaling, we designed a cell-permeable peptide containing the ␣2-cytoplasmic tail sequence KLGFFKR, known to be important for regulating inside-out activation of several integrins (6,31,43). Addition of this peptide resulted in activation of ␣2␤1, as observed by a significant increase in collagen binding, and also exposed the neo-epitope of IAC-1. For outside manipulation, three different agents were used known to activate integrins as follows: the divalent cations Mn 2ϩ and Co 2ϩ (44 -46); the reducing agent DTT, which breaks two disulfide bonds within the cysteine-rich domain of the ␤-subunit (47,48); and the activating anti-␤1 antibody TS2/16 (45). Manipulation of platelets and ␣2␤1-expressing CHO cells with those three activating agents resulted in a significant collagen binding, indicating that ␣2␤1 on these cells became activated. Most interestingly, this collagen binding was not accompanied by IAC-1 binding, implying that the neo-epitope of IAC-1 is not exposed after outside manipulation. Hence, based on these data it can be suggested that ␣2␤1 can be present in three different conformational states as follows: (i) the nonactivated, resting state with no collagen nor IAC-1 binding; (ii) an intermediate state, induced after outside manipulation, with collagen but no IAC-1 binding; and (iii) a fully activated state, induced after inside-out stimulation, with both collagen and IAC-1 binding. In addition, we hypothesize that these conformational states of ␣2␤1 resemble the closed conformation, the extended conformation with closed headpiece, and the extended conformation with open headpiece, as already described for several integrins. However, more detailed research using electron microscopy and crystal structures is necessary to confirm this hypothesis.
For ␣L␤2, the mAb MEM148 has been described that recognizes an epitope not expressed in the resting integrin state but exposed upon Mg 2ϩ /EGTA treatment (49). Just recently, the epitope of this mAb has been mapped to a critical proline residue located on the inner face of the ␤2 hybrid domain (50). In this study, ␣L␤2-expressing MOLT-4 cells could bind MEM148 after Mg 2ϩ /EGTA treatment and were able to bind the ligand ICAM-1 as expected, but most interestingly, treatment with the activator phorbol ester 12-myristate 13-acetate (PMA) was not able to induce detectable expression of the MEM148 epitope on these cells. This was rather unexpected because PMA is known to promote ␣L␤2 ligand binding (51). So it appears that the mAb MEM148 recognizes an epitope that become accessible after outside manipulation (by the divalent cation magnesium) but not after inside-out stimulation (by PMA). The specificity of this mAb for different conformational changes in the integrin is in line with our present data; however, we observed no binding of IAC-1 after outside manipulation but only after inside-out stimulation. Still, we could demonstrate also that our mAb IAC-1 can distinguish between conformational states of ␣2␤1 induced after different agonistic stimulations.
Next, we tried to determine whether these different conformational states would influence the function of ␣2␤1 and therefore allow ␣2␤1expressing CHO cells, stimulated with an inside-out or outside agonist, to adhere to coated collagen for various periods of time. Although both activation ways clearly enhanced the adhesion of CHO cells to coated collagen as compared with resting cells, no significant difference in adhesion was observed between cells treated either way. More detailed investigation of this adhesion, however, revealed some interesting observations. First, ␣2␤1-expressing CHO cells manipulated with the outside stimulating agent Mn 2ϩ spread significantly more on coated collagen than nonactivated CHO cells. This is consistent with a study demonstrating that DTT stimulation (outside) is required to induce cell spreading of aIIb␤3-expressing CHO cells on coated von Willebrand factor (33). Second, and rather unexpected, we observed that ␣2␤1expressing CHO cells manipulated with Mn 2ϩ spread significantly more than ␣2␤1-expressing CHO cells activated via inside-out signaling. There is some evidence for other integrins that certain conformational states are likely to be more suited for a transient adhesion like cell spreading, rolling, and migration, whereas other conformations favor a more stable adhesion. For example, it has been demonstrated that the integrin ␣L␤2 can support such a transient adhesion of leukocytes (namely rolling), together with selectins and ␣4 integrins, during inflammation in vivo (52). More recently, this observation could be linked to a certain conformational state of ␣L␤2. It was shown that an intermediate conformation of ␣L␤2, consisting of the extended state with closed headpiece, was responsible for a transient adhesion of leu- kocytes, whereas the fully open conformation with open headpiece appeared to be required for firm adhesion (53). Along the lines of these studies on ␣L␤2, we now assume that the intermediate conformational state of ␣2␤1 induced after outside manipulation is in a more suitable orientation to induce spreading on collagen than the fully activated state induced after inside-out stimulation. However, we still need a better insight in the physiological role of these conformational states of ␣2␤1, especially in regard to the involvement of the different conformations of ␣2␤1 during platelet thrombus formation in vivo.
Ultimately, we could link the flow cytometric observations on IAC-1 binding after the different stimulations to the observed effects on cell adhesion, because we could observe IAC-1 binding on attached cells after inside-out stimulation but no IAC-1 binding to spread cells after outside manipulation. The fact that cell adhesion was observed without IAC-1 binding has already been described previously for platelet adhesion to collagen under flow conditions (54).
Apart from studying the effect of outside manipulation on platelets and ␣2␤1-expressing CHO cells, we also performed experiments with PBMC and Jurkat cells. Also, inside-out signaling of ␣2␤1 always resulted in both collagen and IAC-1 binding. Most unexpectedly, outside manipulation with divalent cations not only resulted in collagen binding but was accompanied by an increase in IAC-1 binding, which is in contrast to what was seen upon outside manipulation of platelets or ␣2␤1-expressing CHO cells. Although this discrepancy in IAC-1 binding after outside manipulation between platelets and ␣2␤1-expressing CHO cells on the one hand and PBMC and Jurkat cells on the other hand remains to be clarified, it is known that integrins can display different ligand binding properties and functions, depending on the cell type on which the integrin is expressed (55). Similar to our findings with IAC-1, a very recent study reported that the exposure of the neo-epitope of MEM148 (already mentioned above) could vary after stimulation, depending on the cell type where the integrin is expressed (50). This study demonstrated that MEM148 could bind to ␣L␤2-expressing MOLT-4 cells after Mg 2ϩ /EGTA stimulation but not after PMA treatment (50), which is in contrast to experiments performed previously, where ␣L␤2-expressing myeloid cells did bind MEM148 after PMA stimulation (56).
In conclusion, the data presented here show that at least two different functionally active conformational states of ␣2␤1 exist depending on the way the integrin is stimulated and on the cell type where ␣2␤1 is expressed, and furthermore that these different conformations correlate with the different activities of the integrin.