Cation-dependent changes in the binding specificity of the platelet receptor GPIIb/IIIa.

The presence of manganese (Mn2+) significantly increases the binding of the platelet surface receptor GPIIb/IIIa to two synthetic peptides Gly-Arg-Gly-Asp-Ser-Pro (GRGDSP) and Leu-Gly-Gly-Ala-Lys-Gln-Ala-Gly-Asp-Val (L10) that contain the recognition sequences RGD and KQAGDV, respectively. This results in an increase in the amount of GPIIb/IIIa adsorbed by GRGDSPK- and L10-Sepharose by 12-20-fold. Additionally, Mn2+ eliminates contaminating platelet vitronectin receptor, alpha v beta 3, which copurifies with GPIIb/IIIa on the peptide affinity columns in the absence of Mn2+. In contrast to this increased peptide binding of GPIIb/IIIa, Mn2+ reduces the binding of GPIIb/IIIa to its macromolecular RGD-containing ligands fibrinogen, fibronectin, and vitronectin. These results could mean that Mn2+ changes the structure of the binding site on GPIIb/IIIa such that it is now better suited to accommodate conformations available to the RGD sequence within short, linear synthetic peptides but not available to the RGD sequences within the natural ligands. To support this hypothesis we tested a conformationally restricted cyclic peptide, cyclic 2,10-GPenGHRGDLRCA, which in competition assays, preferentially inhibits the binding of GPIIb/IIIa to fibrinogen but does not inhibit well the binding of other RGD-dependent integrins, alpha v beta 3 and alpha 5 beta 1 to their respective ligands. In such assays, the presence of Mn2+ dramatically changed the binding specificity of GPIIb/IIIa by shifting the preference of the receptor away from the selective peptide, cyclic 2,10-GPen-GHRGDLRCA toward the nonselective GRGDSP peptide. This shift parallels the Mn2(+)-dependent change of the binding of GPIIb/IIIa to its natural protein ligands.

The presence of manganese (Mn") significantly increases the binding of the platelet surface receptor GPIIb/IIIa to two synthetic peptides Gly-Arg-Gly-Asp-Ser-Pro (GRGDSP) and Leu-Gly-Gly-Ala-Lys-Gln-Ala-Gly-Asp-Val (LlO) that contain the recognition sequences RGD and KQAGDV, respectively. This results in an increase in the amount of GPIIb/IIIa adsorbed by GRGDSPK-and LlO-Sepharose by 12-20fold.
Additionally during hemostasis by binding to one or more of its ligands (i.e. fibrinogen and von Willebrand factor) and forming a platelet plug. GPIIb/IIIa is a member of the integrin family of cell surface receptors (1,2) consisting of two noncovalently associated subunits, GPIIb (olIIb) and GPIIIa (&). Divalent cations such as Ca2+ and M$+ are required for subunit association of GPIIb/IIIa (3-10) and for the binding function of all integrins (2). These cations presumably bind to domains  30,[31][32][33][34]. In addition, GPIIb/IIIa also recognizes a non-RGD sequence on fibrinogen (KQAGDV) that is located at the carboxyl terminus of the fibrinogen y chain (26,35). Synthetic peptides derived from these recognition sequences competitively inhibit GPIIb/IIIafibrinogen interaction by binding to the same or mutually exclusive binding sites on GPIIb/IIIa (26,33,36,37) and inhibit platelet aggregation in vitro and thrombus formation in vivo (31,38).
In this report we demonstrate that MnZ+ dramatically increases the binding activity of the GPIIb/IIIa receptor for synthetic peptides containing the recognition sequences RGD or KQAGDV resulting in higher receptor yields from peptide affinity columns. This finding is in sharp contrast to the reduced binding activity of GPIIb/IIIa for its RGD-containing ligands, fibrinogen, fibronectin, and vitronectin, induced by Mn*+. A possible explanation for this discrepancy may be that the Mn'+ changes the shape or accessibility of the RGD binding site in the receptor so that it is not well suited to bind RGD as it is presented in the natural protein ligands but more highly suited to some of the conformations available to the small synthetic peptides. In support of this idea, we found that a conformationally more restricted cyclic peptide that has improved selectivity for the GPIIb/IIIa did not display a Mn'+-induced increase in binding activity.

RESULTS
Manganese-induced Changes in the Ligand-BindingActivity of GPllb/llla-GPIIb/IIIa was adsorbed from platelet extracts on GRGDSPK-Sepharose in the presence of CaCl* and MgC12 (Ca'+/Mg'+) and eluted from the column with soluble GRGDSP as reported previously (30). Analysis of this receptor preparation on polyacrylamide gels revealed, in addition ' J. Gailit to GPIIb/IIIa, two major contaminating proteins with molecular masses of 155 and 100 kDa under nonreducing conditions (Fig. 1A). The substitution of CaCl* with MnC& in the buffers (MnZ+/Mg2+) resulted in the isolation of significantly higher amounts of GPIIb/IIIa without any detectable protein contaminants (Fig. IA). For both cation conditions (Ca*+/Mg'+ and Mn"/Mg") the column sizes were equal, the GRGDSPK-Sepharose was from the same preparation, and the extracts applied to the columns were of equal volume and from the same extract preparation. A similar manganese-induced increase in the yield of GPIIb/IIIa was observed when extracts of lz51 surface-labeled platelets were used (Fig. 1B). The GPIIIa subunit was consistently labeled more intensely than the GPIIb subunit, as observed by other authors (46). Because the ratio in protein content of GPIIb to GPIIIa was approximately l:l, the different intensity of the bands on autoradi- Peptides containing sequences derived from the carboxyl terminus of the fibrinogen y chain have also been successfully used to isolate GPIIb/IIIa from platelet extracts (37). We used the y chain peptide LGGAKQAGDV (LlO) coupled to Sepharose beads to isolate GPIIb/IIIa in both cation combinations, Ca'+/Mg'+ and Mn'+/Mg'+. Elution with 1 mg/ml LlO peptide resulted in a similar increase in purity and quantity of receptor similiar to that observed with the GRGDSPK-Sepharose when manganese was present. The difference in receptor yield between the two cation conditions was about 12-fold.

Elimination of Vitronectin
Receptor Contamination of GPIIb/IIIa Prepared with Manganese-We employed immunoblotting with the monoclonal anti-IIb antibody, PMI-1(42), and the monoclonal anti-a, antibody 139 to analyze the proteins eluted from the GRGDSPK-Sepharose columns. The immunoblots shown in Fig. 2A show that GPIIb is present regardless of which cations were present during the isolation procedure, suggesting that both cation combinations result in the isolation of the same receptor. The antibody 139 recognized a 155-kDa polypeptide, indicating the presence of platelet N-in GPIIb/IIIa prepared in the presence of Ca*+/Mg'+ as reported previously (47). A mixture of both antibodies more clearly shows the reactivity with two distinct bands. The absence of reactivity with a polyclonal anti-a& antibody (20) indicates that the (Ye is associated with PR (47) and not with p1 (48) (data not shown). Furthermore, vitronectin receptor also copurified with GPIIb/IIIa on the LlO column in Ca'+/ Mg"+, as indicated by the reactivity of antibody 139 with a 155-kDa protein in this preparation as well (Fig. 2B). However, we detected no vitronectin receptor contamination in preparations of GPIIb/IIIa isolated with Mn2+/Mg2+ on either GRGDSPK-or LlO-Sepharose (Fig. 2) A tion of GPIIb/IIIa to the peptide matrices in the presence of manganese is due to an increase in receptor affinity for these peptides. We addressed this possibility by determining, in a receptor liposome assay, the efficacy of soluble GRGDSP peptide in competing with fibrinogen for the binding site on GPIIb/IIIa in the presence and absence of manganese. We found that the binding of both GPIIb/IIIa isolated with Ca'+/ Mg2+ and that isolated with Mn2+/Mg2+ to a fibrinogen substrate was inhibited at much lower concentrations of peptide if manganese was included during the binding assay rather than calcium. Fig. 3 shows the inhibition curves of GPIIb/ IIIa isolated with Ca'+/Mg". These results are in agreement with the possibility that manganese may increase the affinity of GPIIb/IIIa for the GRGDSP peptide.
In contrast to the increase in the apparent affinity for the peptide shown here, Gailit and Ruoslahti (1988) observed that manganese decreased the binding of GPIIb/IIIa to its protein ligands fibrinogen, vitronectin, and fibronectin. This discrepancy was addressed experimentally by testing a series of RGDcontaining peptides with different receptor selectivities. Earlier we showed that constraining the flexibility of synthetic RGD peptides by cyclization results in a change in the receptor selectivity of the peptide (41). Subsequently, we have designed a number of cyclic peptide analogues by substituting the amino acids surrounding the RGD sequence. We hypothesized that if we could design a peptide that would selectively inhibit GPIIb/IIIa binding to fibrinogen and not inhibit CU,~S or c&, its structure should more closely mimic that of the fibrinogen binding site than do the less specific peptides. The selectivity of the peptide analogues was assessed by measuring their ability to inhibit the binding of liposomes containing purified GPIIb/IIIa, placental (Y,& or placental a& to surfaces coated with fibrinogen, vitronectin, or fibronectin, respectively, in the presence of Ca2+/Mg2+. We found that the cyclic peptide, cyclic 2,10-GPen-GHRGDLRCA, was a better inhibitor of GPIIb/IIIa binding to fibrinogen than was the internal standard peptide, GRGDSP, (Fig. 4.4) while it showed considerably less binding to (Y& (Fig. 4B) and LY&, (Fig. 4C) than did GRGDSP. In contrast, we found another cyclic peptide, cyclic 1,7-VRGDSPDG, to be a somewhat selective inhibitor of a&. For this receptor the concentration of cyclic 1,7-VRGDSPDG required for half-maximal inhibition (I&J was about five times lower compared to the internal standard peptide B  4A) and (Y& (Fig. 4C). It is worth noting that all of the peptides tested inhibit or& at a lower concentration than was required for inhibiting the GPIIb/IIIa which probably reflects intrinsic differences in receptor binding assays for these integrins. For example, we know that the concentration of receptor in the liposomes can greatly influence the amount of peptide required for complete inhibition of binding.3 A comparison of the various peptides revealed that GRGDSP and the cyclic 1,7-VRGDSPDG peptide inhibited GPIIb/IIIa binding at much lower concentrations in the presence of Mn*+/Mg'+ than in the presence of Ca*+/M$+. The Mn'+-induced decrease in the ICsO value was about 25-fold for both peptides. The data for GRGDSP are shown in Table I. Moreover, other c&-selective peptides, such as cyclic 2,9-GPenFRGDSFCA, also inhibited GPIIb/IIIa binding at much lower concentrations when Mn'+ was included (data not shown). In contrast, the presence of Mn2+ slightly increased the amount of the GPIIb/IIIa-selective cyclic 2,10-GPen-GHRGDLRCA peptide required to inhibit receptor binding resulting in an apparent reversal of receptor specificity. The degree of selectivity for the two peptides is indicated as a ratio of the ICso for GRGDSP compared to 2,10-GPen-GHRGDLRCA. DISCUSSION We demonstrate that manganese dramatically increases the binding activity of the platelet receptor GPIIb/IIIa to the hexapeptide GRGDSP that contains the receptor recognition sequence RGD. However, manganese has the opposite effect on GPIIb/IIIa binding to its protein ligands, and we show, receptor-selective cyclic RGD peptides, that this discrepancy may be due to the nonselective nature of the GRGDSP peptide.
We first observed that when Mn*+ was included in the buffer during the isolation of GPIIb/IIIa by affinity chromatography on GRGDSPK-Sepharose, a 20-fold higher yield of the receptor could be obtained. This increase was due to an improved adsorption of GPIIb/IIIa to the peptide-sepharose and not due to differences in the efficiency of elution with soluble GRGDSP peptide because additional elution with 10 mM EDTA did not release any more receptor from the columns. The GPIIb/IIIa that was isolated in Mn'+/M$+ appeared to be identical to GPIIb/IIIa isolated in Ca*'/Mg*' in that protein prepared in both ways showed the same electrophoretic mobility on SDS gels, PMI-1 antibody reactivity, and ligand binding specificity in liposome assays (data not shown). Moreover, when soluble GRGDSP peptide was used to inhibit the binding of GPIIb/IIIa-containing liposomes to fibrinogen, the presence of Mn'+/Mg'+ resulted in an approximately 25-fold lower concentration of peptide being required to inhibit receptor binding over that required in the presence of Ca2+/MF.
The liposome data correspond well to the results from the affinity chromatography columns suggesting that the substitution of Ca2+ with Mn2+ in the buffer systems in fact results in an increased receptor binding to the RGD peptide.
The manganese-induced increase in receptor binding to GRGDSP seems to parallel that of another RGD-dependent integrin, human fibronectin receptor (Y& (20,21). Therefore, it may be that Mn'+, as suggested for a& (20), exerts its effects by binding to divalent cation-binding sites on the (Y subunit (GPIIb). However, an involvement of the IIIa subunit cannot be ruled out because Mn2+ might bind to low affinity cation binding sites (49-52) the subunit location of which has not yet been determined.
The platelet vitronectin receptor, a&, copurified on the GRGDSPK-Sepharose column in the presence of Ca2+/MgZ+ (47) and could be detected on autoradiograms and gels stained with Coomassie Blue (Fig. 1). We identified the platelet vitronectin receptor on immunoblots with the monoclonal anti-o, antibody, 139, that was originally raised against placental vitronectin receptor. Recently a novel integrin showing the subunit composition of (Y& has been described (48). To elucidate whether the a, subunit detected in the GPIIb/IIIa preparations is complexed with p1 we analyzed these preparations with a polyclonal anti-c+,& antibody (20). The absence of any reactivity with this antibody indicated the presence of an c&, rather than an ol,& complex. Furthermore, when Mn2+/Mg2+ was used instead of Ca2+/Mg2+ during the purification, the platelet vitronectin receptor did not copurify as suggested by the absence of any anti-a, antibody reactivity on immunoblots (Fig. 2)  increased receptor interaction with this peptide in a manner similar to that of the nonselective GRGDSP. In contrast, the GPIIb/IIIa-selective cyclic 2,10-GPenGHR-GDLRCA peptide inhibited receptor binding slightly less in the presence of Mn2+. The binding of GPIIb/IIIa to the peptides was measured as inhibition of binding to fibrinogen. Because the fibrinogen binding was less efficient in the presence of Mn2+, the binding of the cyclic peptide to GPIIb/IIIa may have decreased even more than this result would suggest. This Mn2+-dependent change in the preference of the receptor for the selective cyclic 2,10-GPenGHRGDLRCA peptide parallels the decrease in the binding to the natural protein ligands in the presence of Mn*+.
These data could explain the observed discrepancy in the effect of Mn2+ on GPIIb/IIIa binding to GRGDSP compared with the natural ligands. Mn2+ might alter the binding site on the receptor in such a way as to decrease its ability to accommodate the natural ligands or a peptide, such as cyclic 2,10-GPenGHRGDLRCA, which closely mimics the RGD structure preferred by the receptor in the presence of calcium. Alternatively, GRGDSP and cyclic 2,10-GPenGHRGDLRCA might bind to separate but related binding sites on GPIIb/ IIIa (36), one being affected by Mn*+ differently than the other. At the present time we are not able to distinguish between these two possibilities.
In either case, the similarity in response to Mn2+ of GRGDSP and LlO peptide binding supports the idea that these two peptides bind to the same site on the receptor. The involvement of divalent cations in altering the apparent affinity of integrins as well as the ligand selectivity will undoubtedly be useful in leading to an understanding of the molecular basis of this interaction.
It has been suggested that the lipid environment may be able to regulate the ligand specificity of placental OI& and this change in specificity has been related to an apparent change in receptor conformation (53). Recently, it has been shown that the C& integrin from platelets has a different ligand specificity than the same integrin from endothelial cells and other cells (58-61). If Mn*+ induces a change in the conformation of integrins, it may shed light on the nature of the binding site in these proteins. These cation-dependent changes in GPIIb/IIIa ligand binding also have implications on the design of assay conditions when GPIIb/IIIa-antagonist are tested in in vitro receptor-binding studies. Finally, the finding that the presence of Mn2+ improves the quantity and quality of GPIIb/IIIa preparations may be very practical for studies requiring large amounts of receptor such as nuclear magnetic resonance analysis and crystallography.