Communication Vol. 266, No. 18, Issue of June 25, pp. 11429-11432,1991 0 1991 by The American Society for Biochemistry and Molecular Biology, Inc.

Printed in U. S. A .

THE JOURNAL OF BIOLOGICAL CHEMISTRY

Labeling of Integrin with 6sco(III) EVIDENCE OF METAL ION COORDINATION SPHERE INVOLVEMENT IN LIGAND BINDING*

(Received for publication, January 22, 1991) Jeffrey W. Smith$ and David A. Cheresh From the Department of Immunology, Research Institute of Scripps Clinic, La Jolla, California 92037

Integrin-mediated cell adhesion to the extracellular matrix is divalent metal ion-dependent; however, a demonstration of the interaction between native inte- grins and divalent metal ions is lacking. Here we pro- vide direct evidence that the vitronectin receptor (VNR) is a metalloprotein. The unique electron shell of Co(II), an ion which we show supports ligand recogni- tion by VNR, enables its oxidative conversion to inert Co(II1). This property facilitated “affinity labeling” of VNR with 68Co(III) by oxidation of the metal ion in situ (Le. in position). An average of 3.5 f 0.5 mol of cobalt were incorporated per mol of VNR. The ability of VNR to bind metal ions was independently confirmed by examining the interaction between VNR and Mn2+ un- der native conditions. The apparent high affinity be- tween VNR and Mn2+ allowed us to observe the specific binding between 64Mn2+ and VNR by equilibrium gel filtration studies. Interestingly, the oxidative incor- poration of Co(II1) into VNR specifically blocked li- gand binding, suggesting that the coordination sphere of metal ion bound to VNR is a critical determinant in integrin-ligand recognition. Furthermore, Mn2+ abol- ished the oxidative affinity labeling of VNR with Co(II1) and consequently blocked the inactivation of VNR by in situ incorporation of Co(II1). Thus, Mn2+ and Co2+ bind to the same or mutually exclusive sites on VNR. These observations provide the first demon- stration that an integrin, specifically VNR, is a metal- loprotein and demonstrate a functional link between the coordination sphere of the bound metal ion and ligand recognition by this receptor.

The integrins are a protein family of cell surface receptors that mediate the adhesion of cells to the extracellular matrix and in some cases are responsible for cell-cell contacts (1-3). Each integrin is composed of noncovalently associated a and

* This study was supported in part by National Institutes of Health Grants CA45726 and CA50286 (to D. A. C.). Portions of this study were presented at the meeting of The American Society for Cell Biology in San Francisco (1988). This is publication number 6699- IMM from the Research Institute of Scripps Clinic, La Jolla, CA. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “aduertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

$ Supported by a fellowship from Cancer Research Institute, New York. To whom correspondence should be addressed: Committee on Vascular Biology (CVB-l), Research Inst. of Scripps Clinic, 10666 N. Torrey Pines Rd., La Jolla, CA 92037. Tel.: 619-554-3411.

/3 subunits, both of which are required for cell surface expres- sion and ligand binding (4 ,5) . At least four lines of functional evidence suggest that divalent metal ions have a role in integrin-ligand interaction: 1) integrin-mediated cell adhesion is divalent metal ion-dependent (1-3); 2) ligand binding to purified integrins is metal-dependent (6-9); 3) altering the type of divalent metal the receptor is exposed to can alter an integrins ligand binding specificity (10); and 4) a mutant form of platelet integrin GPIIb-IIIa’ has been identified which is unable to undergo a metal-dependent conformational change and also fails to bind its ligands (11).

It has been hypothesized that integrins bind metal ions based on indirect structural evidence. The integrin a subunits contain as many as four repeats with significant homology to the helix-loop-helix (HLH) calcium-binding proteins (2, 12). However, the ability of the integrin repeats to actually bind metals is questionable, because they do not conform to the strict HLH motif. In fact, the integrin repeats lack the proper residues required for a-helices which flank each metal binding loop and are ardently conserved in the HLH Ca2+-binding proteins (13). The integrins also contain a nonconservative substitution at residue number 12 within the metal binding loop of the HLH proteins (12). This residue is usually a glutamic acid and is critical for coordination of the metal ion by the HLH proteins (13).

In an effort to gain an understanding of the metal ion dependence of ligand binding to the integrins, we have begun to characterize divalent metal ion interaction with integrin a&, which is also termed the vitronectin receptor (VNR). VNR is noteworthy because it is one of the integrins capable of binding small synthetic peptides containing the RGD se- quence (3 ,4 ,6 ,8 ) . We recently used a photoaffinity derivative of one such peptide to map the ligand binding domain of VNR (8, 14). Similar results were obtained in affinity labeling studies of platelet integrin GPIIb-IIIa (15). The results from these studies demonstrated that ligand recognition by VNR is metal-dependent and that the ligand binding site is proxi- mal to the four putative metal binding sites on the a subunit of VNR. Collectively, these data provided a potential struc- tural and functional link between ligand binding and divalent metal ion interaction with VNR. However, like other studies regarding metallo-regulation of the integrins, our analysis failed to provide a demonstration that VNR was actually capable of binding to divalent metal ions. The objective of the work presented here was to provide such a demonstration and to investigate the role of metals in ligand binding.

EXPERIMENTAL PROCEDURES

Materials-Na’2sI, 54MnC12, and ”CoCI2 (271 Ci/mmol) were pur- chased from Amersham Corp. P-2 gel filtration resin and Chelex 100 resin were obtained from Bio-Rad. Sephadex G-25 and Nonidet P-40 (Surfact-amps P-40) were from Pierce Chemical Co. Analytical grade H p 0 2 and the chloride salts of divalent metal ions were purchased from Mallinkrodt Inc. Radioimmunoassay-grade bovine serum albu- min and ovalbumin were obtained from Sigma. Human placenta was the generous gift of the Department of Obstetrics a t Scripps Memorial Hospital.

’ The abbreviations used are: GPIIb-IIIa, glycoprotein IIB-IIIa; VNR, vitronectin receptor; HLH, helix-loop-helix; SDS-PAGE, so- dium dodecyl-sulfate polyacrylamide gel electrophoresis; SASD, sul- fosuccinimidyl 2-(p-azido-salycylamido)-l,3’-dithiopropionate.

11429

11430 Integrin-Metal Ion Interaction

Protein Purification and Radioiodinatwn-Vitronectin was purified from human plasma by heparin-Sepharose chromatography as de- scribed (16). The purity of vitronectin was >95% as judged by Coomassie staining of SDS-PAGE. Vitronectin was radioiodinated to a specific activity of 70,000 dpm/ng as describedpreviously (17). VNR was purified from human placenta and depleted of endogenously bound divalent metal ions as described previously (8). The concen- tration of VNR was determined with the bicinchoninic acid protein assay from Pierce Chemical Co. and by amino acid analysis.

Solid-phase Ligand Binding Studies-The concentration of diva- lent metal ions required for ligand recognition by VNR was deter- mined by simple modification of the ligand binding assay described previously (18-20). Immobilized VNR was depleted of endogenously bound divalent metal ions by incubation with 5 mM EDTA in binding buffer (50 mM Tris.HC1, pH 7.4, 100 mM NaCl, 1 mg/ml of radio- immunoassay-grade bovine serum albumin) for 20 min at 30 "C. Subsequently the immobilized VNR was replenished with a range of each divalent metal ion in binding buffer for 45 min at 30 "C. Ligand binding was assessed by a 3-h incubation with 1251-vitronectin (20 ng/ well), subsequent washing to remove free ligand, and solubilization of bound 1251-vitronectin with 100 pl of boiling 2 N NaOH to remove bound vitronectin from the microtiter well. Bound vitronectin was then quantified by gamma counting. Nonspecific binding of lZ5I- vitronectin was determined by similar studies performed in the pres- ence of a 300-fold molar excess of either native vitronectin or a synthetic peptide of sequence GRGDSP. Specific binding was typi- cally 90% of the total binding.

The effects of oxidation of the Co(I1). VNR complex on ligand binding were assessed as follows. VNR was immobilized in TiterTek 96-well plates and depleted of endogenously bound metals as de- scribed above. Immobilized VNR was incubated with 2 mM CoC12 in binding buffer for 45 min at 30 "C. Oxidation of the metal-receptor complex was achieved by addition of a range of H202 diluted in binding buffer (21). Oxidation was allowed to proceed for 45 min at 30 "C. The plate was washed to remove HZ02 and then chelated with 5 mM EDTA in the same buffer to remove any remaining Co2+. The plate was washed three times in metal-free binding buffer, and VNR was subsequently replenished with saturating amounts of either CaCI2, MgC12, MnC12, or CoC12 for 45 min. '251-Vitronectin binding was assessed as described above.

Affinity Labeling V N R w i t h 5sCo-Purified VNR was affinity la- beled with T O by introducing cobalt as the labile Co(I1) ion and then converting this to inert Co(II1) by oxidation in situ, i.e. in position (21, 22). Briefly, 10 pl (10 pCi) of 58C0C12 in 0.1 N HCI and 86 pl of 200 Mm Tris-HC1, pH 7.85, were added to purified VNR in 104 pl of 50 mM Tris-HC1, 100 mM NaC1, pH 7.4, 0.1% Nonidet P-40, and 0.4 mM CoC12. The neutral pH of the final mixture was verified by spotting 2 pl to pH-indicating paper. The sample was incubated for 45 min at 37 "C to allow binding between Co2+ and VNR to equilibrate. Co(I1) was converted to Co(II1) by addition of 20 pl of 220 p M HzO, and subsequent incubation for 45 min. The sample was gel-filtered on a 15-ml Sephadex G-25 column equilibrated in binding buffer containing 0.1% Nonidet P-40. Fractions (10 drop, 240 pl) were collected and subjected y counting. The elution position of VNR was determined by analyzing the fractions on SDS-PAGE and in some cases by determining the elution position of lZ5I-VNR. To quantify the mole ratio of cobalt associated with VNR following oxidation, fractions containing VNR were pooled and dialyzed for 18 h against 50 mM Tris-HCL, 100 mM NaCl containing 0.1% Nonidet P-40. The samples were recovered and subjected to y counting to determine the recovery of "Co. The recovery of VNR following gel filtration and dialysis was 43% as determined by the recovery of tracer '"I-VNR in three separate experiments conducted under identical conditions.

Equilibrium Gel Filtration Studies-The binding between purified VNR and 64Mn2C was examined by Hummel-Dreyer gel filtration (23). A 0.7-cc P-2 resin was equilibrated in phosphate-buffered saline, 0.1% Nonidet P-40 containing 10 p~ Mn2+ and 3,800 dpm of 54Mn2+/drop of effluent as a tracer. A 6O-pl sample of EDTA-treated VNR (4 mg/ ml) in the same buffer and containing 27,000 cpm of Iz5I-VNR as a tracer, was applied to the column. One drop fractions were collected and subjected to dual channel y counting on an Pharmacia LKB Biotechnology Inc. 1282 CompuGamma counter. Channels were de- fined so that no counting overlap occurred between lz5I and 54Mn. The samples were also subjected to SDS-PAGE to confirm the elution position of VNR.

RESULTS AND DISCUSSION

VNR Is a Metalloprotein-One objective of this study was to determine whether VNR is a metalloprotein. This exami- nation was prompted by the growing number of studies doc- umenting the involvement of divalent metal ions in integrin- ligand interaction and the corresponding lack of information regarding metal ion binding to native integrins.

Our examination consisted of two approaches, the first involved an examination of VNR interaction with Co(II), which has found wide use as a probe of metalloproteins, because it can be converted from a labile ion to inert Co(II1) by removal of the outer sphere d7 electron by oxidation (21, 24). This technique has been used to localize metals to specific subunits in RNA polymerase (25), to identify an enzymatic intermediate composed of metal-enzyme and substrate in aspartokinase (26), and to define the geometry of the coordi- nation sphere of the bound metal in a zinc finger (27 ) . Since Co2+ has been shown to support ligand binding to other integrins (7), we reasoned that this approach may be useful in examining integrin-metal interaction.

We tested the ability of Co2+, and three other divalent metal ions (Ca2+, M e , and Mn2+) to support ligand binding to VNR (Table I). Co2+ was essentially as effective as Ca2+ in supporting vitronectin binding to VNR (Table I). The appar- ent K d (x200 pM) suggested a relatively low affinity and indicated that the interaction with VNR would be difficult to measure under conditions in which the interaction between metal ion and receptor remained reversible. Therefore, we attempted to "affinity label" VNR by allowing labile 58C0(II) to bind the receptor and subsequently locking the ion into the receptor by treatment with H202. Incorporation of inert Co(II1) can be confirmed by obtaining spectral data (21, 27) or by incorporating radioactive cobaltous ion into the protein (21,22). We opted for the latter approach because the amount of purified VNR is limiting and relatively small amounts of 58C0 can be detected by y counting. VNR was purified and depleted of endogenously bound metals as described above and then incubated with 0.4 mM Co2+ with 68C02+ as a tracer. The mixture was subjected to mild oxidation with H202 and then gel-filtered to separate any complex between VNR and Co(1II) from free cobalt. To quantify cobalt incorporation

TABLE I Divalent metal ion-dependent ligand binding to V N R

The ability of four divalent metal ions to support ligand binding to VNR was assessed either with a solid-phase binding assay, which measured the binding of 1251-vitronectin to purified VNR immobilized in microtiter wells or by photoaffinity labeling the purified receptor in solution with 1251-SASD-GRGDSPK. Vn, vitronectin.

Divalent metal

Ion concentration for half-maximal

Ion concentration for half-maximal

ion Vn binding" RGD recognition* X M

Ca2+ 200 500'; 4,000 Mgl+ 1,000 2,500' Mn2+ 17 16 CO" 200 165

The solid-phase receptor assay was performed as described under "Experimental Procedures." The effect of each divalent metal ion was saturable and the value shown for half-maximal binding is the average of at least two experiments.

*The photoaffinity crosslinking studies to assess the ability of VNR to recognize the RGD sequence were performed as previously described (8). The effects of each metal ion were saturable and the concentration of ion required for half-maximal binding was deter- mined either by gamma counting gel slices corresponding to each subunit or by densitometry scanning of the corresponding autoradi- ographs.

'Values were reported previously in Ref. 8. As reported, RGD recognition was biphasic with respect to Ca2+. The half-maximal concentration of Ca2+ required to elicit each phase is shown.

Integrin-Metal Ion Interaction 11431

samples containing VNR were pooled and dialyzed as an extra precaution. Subsequently, the mole ratio of cobalt to VNR was calculated (Fig. 1). In the absence of oxidation cobalt did not remain associated with VNR. This is expected of a low affinity binding event and, since Ca2+ and Co2+ have similar apparent K d values, is consistent with our inability to detect the interaction between VNR and 4sCa2+ by Hummel-Dreyer gel filtration (not shown). However, oxidation promoted the stable incorporation of an average of 3.5 f 0.5 (n = 4) cobaltous ions into VNR. This is presumably the result of the conversion of bound Co(I1) to Co(II1) (21,22,24). This result is consistent with the incorporation of between three and four cobalt ions into VNR and correlates well with the presence of four putative metal binding sites on the a subunit of the receptor. The complex could not be dissociated by treatment with 5 mM EDTA (not shown). The specificity of the incor- poration was demonstrated by our finding that cobalt did not associate with either ovalbumin or mouse IgG under identical conditions (not shown). Furthermore, an excess of Mn", which had the highest apparent affinity for VNR (Table I), blocked the incorporation of 5 8 C ~ into VNR (Fig. I) , confirm- ing the specificity of the interaction. This also suggested that Mn2+ and Co2+ bind to the same, or mutually exclusive, site(s) on VNR.

Since VNR had a higher apparent affinity for Mn2+ than other ions and it abolished the oxidative incorporation of "Co(II1) into VNR, we sought to confirm the ability of VNR to bind divalent metals by assessing its ability to bind 54Mn2+ with the Hummel-Dreyer gel filtration technique (23). VNR was purified and depleted of endogenously bound metal ions as described under "Experimental Procedures." The receptor was chromatographed on a P-2 resin equilibrated with 10 p~ Mn2+ using a4Mn2+ as a tracer. As shown in Fig. 2, a peak of Mn2+ co-elutes with the purified VNR. A trough of radioac- tivity follows the peak of VNR indicating that VNR depleted the column of Mn2+ during chromatography. The specificity of the interaction was demonstrated by the ability of 5 mM Mn2+ to completely block the association of 54Mn2+ with VNR and the inability of a control protein, mouse monoclonal IgG, to bind s4Mn2+ under identical conditions (not shown). It should be noted that this binding study was performed under

A B C FIG. 1. Affinity labeling of VNR with '*Co2+ . The ability of

VNR to associate with Co'+ was examined by converting Co(II), which was bound to VNR, to inert Co(II1) by oxidation with H202. VNR was incubated with 0.4 mM Co'+ containing 10 pCi of "Co2+ as a tracer for 45 min a t 37 "C. The metal-receptor complex was oxidized by addition of H202. Each sample was chromatographed on a 15-ml G-25 resin, and fractions were collected and subjected to y counting. To quantify the cobalt associated with VNR, fractions containing the receptor were pooled and dialyzed for 18 h to ensure that only tightly bound Co(II1) remained associated with the receptor. Samples were then subjected to y counting to determine the recovery of "CO. The mole ratio of cobalt to VNR was determined following oxidation with H& n = 4 ( A ) , following oxidation in the presence of 5 mM Mn'+, n = 3 ( B ) , and in the absence of oxidant, n = 7 (C). No incorporation of "Co into ovalbumin or mouse IgG was detected under identical conditions.

14

12 -

10 4 8 12 16 20 24 28 32

0

0 8

I 4 6

7

X

4

2

n 0 20 40 60

Fraction

FIG. 2. Purified VNR binds to Mn2+. The ability of purified VNR to bind to Mn2+ was assessed by equilibrium gel filtration. A 0.7-cc P-2 resin was equilibrated in 10 p~ Mn2+ and 3,800 cpm of 54Mn2+/drop of effluent as a tracer. A 60-pl sample of metal-free VNR (4 mg/ml) containing 27,000 cpm of '"I-VNR as a tracer was applied to the column. Fractions were collected and subjected to dual channel y counting to quantify "Mn2+ (W) and '"I-VNR (0). A sample of each fraction was analyzed by SDS-PAGE to confirm the presence of VNR (inset). This is a profile of a representative experiment that was performed five times with nearly identical results.

native conditions and at [Mn2+] that appears to be physiolog- ically relevant from the studies shown in Table I.

A Role for the Coordinution Sphere of Bound Metal in Ligand Binding-Given the strict requirement for metals in integrin- ligand interaction and the proximity of the RGD binding site to the putative metal binding sites, we previously suggested that metal ions may enable ligand binding by participating in a ternary complex between ligand, receptor, and metal ion (14). In this regard, we sought to determine the effect of oxidizing bound Co(I1) to Co(III), which alters the metals coordination sphere (21, 22, 24), on ligand binding to VNR. Immobilized VNR was incubated with Co2+, subjected to oxidation with a range of H202, and then replenished with saturating amounts of either Ca2+, M$+, Mn2+, or Co2+. Our results show that oxidation of the VNR.Co(I1) complex abol- ished vitronectin binding to VNR (Fig. 3A), regardless of the ion used to support ligand binding following the oxidative step. To ensure that the effects of oxidation were specific to cobalt, and not the result of oxidative damage of VNR, iden- tical oxidation was performed in the presence of Mn2+ without effect (Fig. 3A). Oxidation in the presence of Ca2+ and Mg2+ also failed to affect ligand binding (not shown). The inacti- vation of VNR by altering the coordination sphere of bound cobalt indicates that the metal ion coordination sphere is involved in ligand binding. Importantly, the oxidative condi- tions which abolished ligand binding to VNR resulted in the incorporation of an average of 3.5 cobalt ions/receptor (Fig. 1). In conjunction with our previous report (14) which showed that the ligand binding site is proximal to the four putative metal binding sites on the a subunit, the results presented here imply that these sites are closely linked to ligand binding. Our results are also generally consistent with metal partici- pation in a ternary complex with ligand and receptor. How- ever, it is also conceivable that the proper coordination sphere must be maintained in order to keep VNR in an active conformation. Since Co2+ and Mn2+ have been detected in similar ternary complexes (26, 28), our results provide a precedent for further examination of this question.

Evidence That the Divalent Metal Ion Binding Sites on VNR Associated with Ligand Recognition Have Broad Speci- ficity-Our results also indicate that all four of the divalent ions tested may interact with a group of metal binding sites

11432 Integrin-Metal Ion Interaction

140 ligand binding to the receptor (Fig. 3B and Table I). This

m Co2+ bind to the same site(s) on VNR and also shows that 2- 100

120 finding lends further support to the hypothesis that Mn2+ and

the inactivation of VNR is the result of specific binding of Co2+ to VNR rather than nonspecific modification of VNR by the free radical that is generated upon oxidation of each Co2+ ion in the buffer (19). Thus, Mn2+ and Co2+ bind to the same, or, at the very least, mutually exclusive sites on VNR.

i 5 E

L& 40 I Neither Ca2+ or Mg2+ blocked the inactivation of VNR by ? 20 oxidative incorporation of Co(II1). It is possible that these

ions bind to distinct regions of VNR, however, it is likely that 0 we were unable to observe any inhibition, because the appar-

0 1 .o 10 100 ent affinity of these ions for VNR is 5 that of Co2+ (Table I).

120- block the binding of Co2+ and subsequent incorporation as

100" In summary, the results in this report support several .-. .-e--"() salient conclusions: 1) VNR can be affinity labeled with 5 8 C ~

C E metal ion in situ. At saturating concentrations an average of 3.5 mol of cobalt are incorporated per mol of VNR. Impor- tantly, the presence of ligand is not necessary for metal- receptor interaction; 2) MnZ+ binding to VNR can be observed under native conditions and in the same range that this ion supports ligand binding, suggesting that this interaction is biologically important; 3) it is likely that the divalent metal binding sites identified by incorporation of Co(II1) have broad specificity. We conclude that Mn2+ and Co2+ can bind to these

[Mn2+] rnM (log male) sites. 4) Oxidative incorporation of Co(II1) into the divalent metal binding sites of VNR inactivates the receptor, suggest-

FIG. 3. Oxidation of the Co(II).VNR complex inactivates ing that the coordination sphere of the bound metal ions is a VNR. A , the effect of the oxidative conversion of bound CO(II) to critical determinant in ligand binding to VNR. Our future Co(II1) on ligand binding to VNR was examined with the solid-phase efforts will focus on a much needed definition of the location ligand binding assay described under "Experimental Procedures." of the divalent metal ion binding sites on VNR. VNR was immobilized in microtiter wells, treated with EDTA to remove endogenously bound metals, and then incubated with 2 mM REFERENCES Coz+. Subsequently, the immobilized receptor was treated with a range Hynes, R, o, (1987) Cell 48, 549-554 of HzOz. The oxidized VNR was treated with EDTA to remove any 2. Phillips, D. R., Charo, 1. F., Parise, L., and Fitzgerald, L. A. (1988) Blood remaining Co'+ and then incubated with a saturating amount of

Vitronectin binding was quantified as described under "Experimental 5. Buck, C., Shea, E., Duggan, K., and Horwitz, A. F. (1986) J . Cell B i d . 1 0 3 , Procedures." The binding of vitronectin to VNR treated with oxidant in the presence of ~ ~ 2 + rather than co2+ was also examined (0). 6. Pytela, R., Pierschhacher, M. D., Argraves, S., Suzuki, S., and Ruoslahti,

2421-2429

Each data point is expressed as the percentage of maximum binding 7, santoro, S, A, (1986) cell 4 6 , 913-920 E. (1987) Methods. Enzyrnol. 144,475-489

and is the average of triplicate points in which the standard deviation 8. Smith, J. W., and Cheresh, D. A. (1988) J . Bid. Chern. 263 , 18726-18731 was less than 10% of total binding. This is a representative experi- 9. Gailit, J., and Ruoslahti, E. (1988) J. Biol. Chern. 263,12927-12932 merit that was performed eight times with similar results. B, the 10. Kirchhofer, D., Gailit, J., Ruoslahti, E., and Pierschbacher, M. D. (1990)

ability of Mn2+ to inhibit the oxidative inactivation of the VNR. 11. Loftus, J. c . , O'Toole, T. E.? PIOW, E. F., Glass, A,, Frelinger, A. F., and Co(I1) complex was also examined. Immobilized VNR was incubated Ginsberg, M. H. (1990) Scrence 249,915-918 with 2 mM coz+ in conjunction with a range ofMnz+ and then oxidized 12. Fitzgerald, L. A., Poncz, M., Steiner, B., Rall, S. C., Bennett, J. S., and with loo mM H?oz. Of Oxidant* treatment with 13. Strynadka, N. C. J., and James, M. N. G. (1989) Annu. Reu. Biochem. 58 , Phillips, D. R. (1987) BiochernLstry 26,8158-8165

EDTA to remove metal ions, and subsequent incubation of the oxidized receptor with 2 mM Mn2+, the binding of lz51-vitronectin (a) 14. Smith, J. W., and Cheresh, D. A. (1990) J. Biol. Chern. 265,2168-2172

951-998

was assessed as described for A.

[Hz021 r n M (log scale) Therefore, it was not possible to generate experimental con- ditions in which Ca2+ and M e were in sufficient excess to

B

by conversion of Co(I1) to Co(II1) by mild oxidation of the 6 g 80" II

Co(II1). m 5 n

O+ 0 0.1 1 .o 10

either coz+ (m), ~ g 2 ' (A), M ~ Z + (a), or caz+ (+I and I L S I - ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ . 3. Ruoslahti, E., and Pierschhacher, M. D. (1987) Science 238,491-497 71,831-843

4. Cheresh, D. A,, and Spiro, R. C. (1987) J. Biol. Chern. 262,17703-17711

J. Biol. Chern. 265,18525-18530

15. D'Souza, S. E., Ginsberg, M. H., Burke, T. A., and Plow, E. F. (1990) J.

16. Yatahogo, T., Izumi, M., Kashiwaga, H., and Hayashi, M. (1988) Cell Biol. Chern. 2 6 5 , 3440-3446

S t ru t . Funct. 13,281-292 with broad specificity. None of the ions tested supported 17. Fraker, P. J., and Speck, J. c. (1978) Biochern. ~ i ~ p h y s . Res. commun. 80, ligand binding after VNR was inactivated by incorporation of 849-857 co(III) (Fig. 3A) . The simplest explanation of this observation 18. Charo, I . F., Nannizzi, L., Smith, J. W., and Cheresh, D. A. (1990) J. Cell

is that all metals bind to the same sites and the incorporation 19. Smith, d . W., Ruggeri, Z. M., Kunicki, T. J., and Cheresh, D. A. (1990) J. Biol. 1 1 1,2795-2800

Of cO(lll) into these sites prohibits subsequent binding* 20. Smith, J. W., Vestal, D. J., Irwin, S. V., Burke, T. A,, and Cheresh, D. A. Biol. Chem. 2 6 5 , 12267-12271

This hypothesis is supported by our studies which show that Mn2+ completely abolished affinity labeling of VNR with 5 8 C ~ 21. Van Wart, H. E. (1988) Methods EnzYmol. 15% 95-109

(1990) J . B i d . Chern. 265,11008-11013

(Fig. 1). We also tested the ability of Mn2+ to Protect immo- 23. Hummel, J. P., and Dreyer, w. J. (1962) Btochtrn. Bmphys. Acta 6 3 , 530- bilized VNR from inactivation by cobalt oxidation. VNR was 532 incubated with Co2+ and subjected to oxidation with H202 in it: ' w " , U ~ . ! l ~ ~ , ) ~ h ~ m H ~ ~ n ~ O r d , D, C, (1977) ~ i ~ ~ h ~ ~ h ~ ~ 1 6 , the presence of a range of Mn2+. The presence of Mn2+ blocked

(Fig. 3B) . The protective effect of Mn2+ was saturable and 27. Green, L. M., and Berg, J. M. (1989) Proc. Natl. Acad. Sci. U. S. A . 8 6 ,

ing active at a [Mn"] similar to that required to support 3664-3669

22. Wright, J. K., and Takahashi, M. (1976) Btoche.mrstV 15,3704-3710

the inactivation of VNR by oxidative incorporation of ~~(111) 26. Wright, J. K., Feldman, J., and Takahashi, M. (1976) Biochem. Biophys.

concentration-dependent' with one-ha1f Of the VNR remain- 28. Haddy, A. E., Frasch, W. D., and Sharp, R. R. (1989) Biochemistry 2 8 ,

5449-5454

Res. Cornrnun. 7 2 , 1456-1461

4047-4051