Competitive Binding of Vascular Cell Adhesion Molecule-1 and the ...

Vol. 269,No. 6, Issue of February 11, pp. 4005-4011, 1994 Printed in U.S.A.

THEJOURNAL OF BIOL~CICAL CHEMISTRY 0 1994 by The American Society for Bioehemiatry and Molecular Biology, h e .

Competitive Bindingof Vascular Cell Adhesion Molecule-1 and the HepII/IIICS Domainof Fibronectin to the Integrin a4Pl* (Received forpublication, June 8, 1993, and in revised form, August 9, 1993)

Rima Makarem, Peter Newham, Janet A. Askari, Linda J. Green, John Clement&, Mark Edwards$, MartinJ. Humphries, and A. Paul Mould5 From the Department of Biochemistry and Molecular Biology, School of Biological Sciences, University of Manchester, Stopford Building, Oxford Road, M13 9PT Manchester, and $British Bio-technology Limited, Watlington Road, Cowley, OX4 5LY Oxford, United Kingdom

immunoglobulin superfamily, is an integral membraneprotein expressed by endothelial cells and a restricted numberof other cell types (Osborn et al., 1989; Rice and Bevilacqua, 1989). Although VCA"1 is normally undetectable on resting endothelium, it can be induced by inflammatory cytokines such as tumor necrosisfactor-a, interleukin-4,andinterleukin-lp (Osborn et al., 1989; Rice and Bevilacqua, 1989; Carlos et al., 1990; Masinovsky et a l . , 1990; Thornhill et al., 1991) and is therefore hypothesized to contributeto leukocyteextravasation ininflammatory conditions suchasrheumatoidarthritis, asthma, and atherosclerosis (Bochner et al., 1991; Cybulsky and Gimbrone, 1991; Laffr6n et al., 1991). VCA"1 also appears to be involved in hematopoietic development and differentiation through itsmediation of B cell interactions with germinalcentersandstem cell-stroma interactionsin bone marrow (Freedmanet al., 1990; Miyake et al.,1991; Clarket al., 1992; Simmons et al., 1992; Teixido et al., 1992). Furthermore, the adhesive function of VCA"1 may be usurped by certain classes of malignant tumorcell to potentiate metastatic spread (Rice and Bevilacqua, 1989; Freedman et al., 1992). The predominant molecular form of V C A " 1 contains seven immunoglobulin modules in its extracellular domain, but the fourth module can be removed from precursor mRNA by alternative splicing to generate a six-module variant (Osborn et al., 1989; Polte et al., 1990, 1991; Cybulskyet al., 1991; Hession et al., 1991). Both forms ofVCAM-1 are recognized by the integrin receptor a 4 p l (Elices et al., 1990). a4pl isexpressed principally by leukocytes (T andB lymphocytes,monocytes, basophils, and eosinophils), but is also functional onderivatives of the embryonic neural crest and developing in muscle (Dufouret al., 1988; Humphries et al., 1988; Hemler et al., 1990; Rosen et al., 1992). The molecular basis of VCA"1 binding to a4pl has not yet been elucidated, but recentdeletion mutagenesis andantibody blocking studies have implicated immunoglobulin modules 1 and 4 as the primary recognition domains (Taichman et al., The adhesion of leukocytes to the vascular endothelium and 1991; Osborn et al., 1992; Vonderheide and Springer, 1992). their subsequent extravasation into tissues critical are steps in a 4 p l also recognizes the extracellular matrix glycoprotein the genesisof an inflammatory response (Springer,1990). Vas- fibronectin (Wayneret al., 1989; Guan and Hynes,1990; Mould cular cell adhesion molecule-1 (VCA"l),I a member of the et al.,1990), and binds in a complex manner to the HepII/IIICS region (Mould and Humphries,1991; Iida et al., 1992). HepII is * These studies were supported by grants from the Wellcome Trust the major proteoglycan-binding domain of fibronectin, while and the Wigan and District Cancer Research Committee (to M. J. H. the IIICS (type I11 connecting segment or V region) is one of and A. P. M.) and by research studentships from the Medical Research three sites subject to alternative splicing within the molecule. Council (to R. M.) and the Science and Engineering Research Council in conjunctionwith the YamanouchiResearch Institute (to P. N.). The Three distinct a4pl-binding sites have been identified within costs of publication of this article were defrayed in part by the payment the HepII/IIICS, and all have been reproduced in synthetic of page charges. This articlemust therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate domain of fibronectin; IIICS,type I11 connecting segment of fibronectin; this fact. PCR, polymerasechain reaction; W120, recombinant fragmentof fibro8 To whom correspondence should be addressed. Tel.: 011-44-61-275- nectin containing type 111 repeats 12-15 and the full-length IIICS re5071; Fax: 011-44-61-275-5082. gion; WO,recombinant fragmentof fibronectin containing only type 111 The abbreviationsused are: (rs)VCA"l, (recombinant so1uble)vas- repeats 12-15; BSA, bovineserum albumin;MES, 4-morpholineethanecular cell adhesion molecule-1; HepII, COOH-terminal heparin-bindingsulfonic acid; ELISA, enzyme-linked immunosorbent assay.

The integrin receptor alp1 binds to two different ligands, the extracellular matrix glycoprotein fibronectin and the endothelial cell surface protein vascular cell adhesion molecule-1 (VCAM-1).Using probes derived from each ligand and a variety of cell adhesion and ligand-receptor binding assays, we have investigated the relationship between the mechanisms of fibronectin and VCA"1 interaction with a4Pl. CS1 peptide, whichrepresents the dominant active site from the HepII/IIICS recognition domainin fibronectin, was foundto inhibit VCAM-1-dependentadhesion in three different assays: MOLT4 T lymphoblastic leukaemia cell attachment to immobilized recombinant soluble VCA"1 (rsVCA"l), MOLT-4 cell attachment to monolayers of VCAM-1-transfected COS-1 cells, and A375-SM melanoma cell spreading on immobilized rsVCAM-1. Half-maximal inhibition required CSl concentrations of 1.7-3.0 mg/ml, some 3-7fold higher than thatneeded to autoinhibit adhesion to CS1-IgG conjugate. Using a more sensitive solid-phase receptor-ligand binding assay, CS1 wasfound to be a potent inhibitor of the binding of rsVCAM-1 to a4P1 (half-maximalinhibition at 13 pg/ml). In agreement with cell-based assays, severalfold lower concentrations of CS1 were required to inhibit binding of recombinant HepII/IIICS regionof fibronectin (half-maximalinhibition at 3 pg/ml).V C A " 1-a4Pl binding was blockednot only by CS1peptide but also by the recombinant HepIYIIICS region of fibronectin. Kinetic analysis of CS1 inhibition of VCA"1 binding revealed that it was directly competitive in nature, indicating that VCA"1 and fibronectin recognize either identical or spatially overlapping binding sites on a4pl. The implications of these results for the future design of VCAM-1 antagonists are discussed.

4005

4006

Integrin-Fibronectin-VCAM-1Interactions

form (Humphries et al., 1986, 1987; Mould and Humphries, 1991). One site (represented by the peptide H1) is found in the HepII region and is therefore expressed in all fibronectin isoforms (Mould and Humphries, 1991); two others (represented by peptides CS1 and CS5) are present in independently spliced segments of the IIICS (Humphries et al., 1986, 1987). The 25mer CS1 peptide dominates the activity of HepII/IIICS and contains the tripeptide Leu-Asp-Val (LDV) as its minimal active site (Komoriya et al., 1991). H1 contains a related motif, Ile-Asp-Ala (IDA), while CS5 incorporates avariant of the prototypic RGD motif, Arg-Glu-Asp-Val (REDV, Humphrieset al., 1986; Mould and Humphries, 1991). In view of the anti-inflammatory potential of VCA"1 antagonists, it isof majorinterest to establish the molecular basis for the interaction of a4pl with its two ligands. Based on specific inhibition of ligand-mediated adhesion by monoclonal antibodies and fibronectin fragments, an early report suggested that the binding sites for the two ligands were distinct (Elices et al., 1990). However, using a series of anti-u4 monoclonal antibodies, a more recent epitope and function mapping study identified a subset of antibodies that had no antifunctional activity, a subset that weakly perturbed fibronectin-mediated adhesion, but not VCAM-1-mediated adhesion, and a subset that were potent inhibitors of adhesion to both fibronectin and VCA"1 (Pulido et al., 1991). Since no antibodies were identified that could selectively inhibit VCAM-1-dependentadhesion, it appears that thebinding sites may be overlapping.To investigate this possibility, we have examined the effect of soluble CS1 peptide and a recombinant formof HepIUIICS on the interaction of VCAM-1 with a4pl in a number of assays. Our results indicate that the VCA"1- and fibronectin-bindingsites within a4pl are spatially close, if not identical, and that binding of one ligand precludes binding ofthe other. These findings could have major implications for the design of specific antiinflammatory drugs targeted at either the a4/3l-VCA"l or a4pl-fibronectin interactions.

EcoRI, ligated into pUC119, and transformed into Escherichia coli JM109. Individual HepIYIIICS clonescontaining each of the five possible variants of the IIICS were identified by restriction analysis and sequenced using the dideoxy chain termination method (Sanger et al., 1977). Inserts were then subcloned into the EcoRI site of a modified version of the pGEX-2T expression vector(Pharmacia, Milton Keynes, U.K.) which includedadditional cloning sites inits polylinker and were used to transform JM109. Recombinant clones were checked forinsert orientation by restriction mapping. Glutathione S-transferase-HepW IIICS fusion proteins were induced and isolated as described by Smith and Johnson (1988). Briefly, 40 ml overnight cultures of JM109 transformed with recombinant plasmids were diluted 1:lO with fresh LB containing 50 p g / d ampicillin and cultured for 1h a t 37 "C. IsopropylP-D-thiogalactoside wasthen added to 0.1 m~ and the culture continued for a further 4 h. Cells werethen centrifuged, lysed in 1/30th volume 1% (w/v) Triton-X100, 0.1% (w/v) SDS, 150 m~ NaC1, 5 m~ EDTA, 2 m~ phenylmethylsulfonyl fluoride, 10pg/ml leupeptin, 50 m~ Tris-HC1, pH 7.4, and sonicated. Extracts were clarifiedby centrifugation at 10,000 x g for 5 min at 4 "C and thenapplied to a glutathione-Sepharose affinity column (Sigma, Poole, U.K.) pre-equilibrated with 150 m~ NaCl, 10 m~ Tris-HC1, pH 7.4.After washing with 150 m~ NaCl, 10 m~ Tris-HC1, pH 7.4, fusion proteins were eluted with 5 m~ glutathione, 50 m~ Tris-HC1, pH 7.4. The glutathione S-transferase portion of the fusion protein was cleaved by addition of 10 unitdml human thrombin (Sigma) for 3 h at room temperature. The cleavage mixture was then applied to a heparinagarose aflinity column (Sigma) pre-equilibrated with 50 m~ NaC1, 10 m~ Tris-HC1, pH 7.4, and recombinant HepII/IIICS protein was eluted with 500 m~ NaCl, 10 m~ Tris-HC1, pH 7.4. Peak fractions were then pooled and dialyzed against 150 IMI NaCl, 10 m~ Tris-HC1, pH 7.4,and stored at -70 "C. Two recombinant variants were used in thestudies in this report; W120 contains type I11 repeats 12-15 and the full-length IIICS region (120 amino acids), while WO contains only type I11 repeats 12-15. Production of Recombinant Soluble VCAM-1-A 1.5-kilobase EcoRIPstI fragment from the 3' end ofVCAM-1cDNA in pCDM8 (British Bio-technology Ltd., Oxford,U.K.)wascloned into M13mp18 and mutagenesis performed according to Kunkel et al. (1987) using the oligonucleotide 5'-CAAGC'ITGCATGCCTGCAGCTAAGGAGAGAUAATAGTC-3'. This converted a glutamate codon immediately before the predicted transmembrane region to a TAG translational stop codon followed by a PstI restriction site. The mutated cDNA was then used to replace the EcoRI-PstI fragment in VCA"l/pCDMS. The complete soluble VCA"1 (SVCAM-1) cDNAwas then excised by XhoI digestion and cloned into the mammalian expression vector pGWlHG (British EXPERIMENTALPROCEDURES Biotechnology Ltd.) which drives expression from a cytomegalovirus Materials-Human plasma fibronectin was purchased from the promoter/enhancer and contains an E. coli xanthine-guanine phosphoBioproducts Laboratory (Elstree, United Kingdom). 110-kDa cell-bind- ribosyltransferase gene which enables selection of stably transfected ing fragment of fibronectin was prepared by thermolysin digestion us- cells by growth with xanthine and mycophenolic acid.Chinese hamster ing established procedures (Zardi et al., 1985); the purity of the frag- ovary cells weretransfected by electroporationwith linearized S V C M ment was >90%as assessed by SDS-polyacrylamide gelelectrophoresis. pGWlHG and clones selectedby growth in Dulbecco's minimal essential The synthetic peptides CS1 (DELPQLVTLPHPNLHGPEILDVPST), medium/Ham's F-12 medium, 10% (v/v) fetal calf serum, 2 m~ glutaCSlscrl (DELPQLVTLPHPNLHGPPSELID), and CS6(LYPHGPmine, 100 unitdml penicillin, 100pg/ml streptomycin sulfate, 300 p g / d GLNPNAST) weresynthesized using Fastmoc chemistry on an Applied xanthine, 6 pg/ml mycophenolic acid.The resulting cell population was Biosystems 431A peptide synthesizer and purified as described previ- subcloned three times by limiting dilution and clones selected for sVously (Humphries et al., 1986, 1987).CS1 was also synthesized with an CAM-1 production by ELISA. SWAM-1 was purified from culture meNH,-terminal cysteine residue to facilitate coupling to IgG as described dium by immunoaffinity chromatography using anti-VCAM-1 mono(Humphries et al., 1987). Rat monoclonal antibodies 16 and 13 (recog- clonal antibody BBIG-V4 as matrix (British Biotechnology Ltd.). nizing the human a5 and /3l integrin subunits, respectively)were gifts Cell Attachment-Cell attachment assays were performedin 96-well from S. K. Akiyama and K. M. Yamada, National Institute for Dental microtiter plates (Costar, Northumbria Biologicals, Cramlington, U.K.). Research, Bethesda, MD.Mouse antihuman a4 integrin monoclonal Wells were coated for60 min at room temperature with 100-pl aliquots antibody HP2/1 was purchased from Serotec (Oxford, U.K.). of adhesion factors diluted with Dulbecco's phosphate-buffered saline, Cloning and Expression of Recombinant HepIIlIIICS Variantsand then siteson the plastic for nonspecific celladhesion were blocked cDNA clones encoding the five different variants of the HepIYIIICS for 30min at room temperature with 100 plof 10 mg/mlheat-denatured region of human fibronectin were synthesized using reverse-transcrip- bovine serum albumin (BSA) (Humphries et al., 1986). The MOLT-4 tase polymerase chain reaction amplification of primary human skin human T lymphoblastic leukaemia cell line was obtained from the Eufibroblast mRNA.Briefly, total mRNA was prepared using the gua- ropean Collection of Animal Cell Cultures (Porton Down, Salisbury, nidinium isothiocyanatdcesium chloridemethod(Sambrooket al., U.K.) and was maintained in WMI-1640 medium with 10%(v/v) fetal 1989).First strandcDNA was then generated by reverse transcription calf serum, 2 m~ L-glutamine, and 50 pdml gentamicin (all from using a 3' oligonucleotideprimer complementary to the end of the 15th GIBCO, Paisley, U.K.).Cells were resuspended to 107/mlin Dulbecco's type I11 repeat of fibronectin and incorporating a terminal EcoRI re- minimal essential medium and allowed to recover at 37 "C for 10 min. striction sequence (5'-TGGAA'ITCAGAC'ITGCCCACGGTAACAAC- 100-pl aliquots of cells were then added to the microtiter wells and CTC-3'). cDNA was purified using Geneclean (Stratech, Luton, U.K.), incubated for 15 min at 37 "C in a humidified atmosphere of 5% (v/v) and amplified by 50 cyclesof the polymerase chain reaction (PCR)using COz. To estimate the reference value for 100% attachment, triplicate without washing, by direct addition of 100 a 5' primer complementary to the start of the 12th type I11 repeat of wells were fixed immediately, human fibronectin which again incorporated an EcoRI restriction se- pl of 5% (w/v)glutaraldehyde for 30min a t mom temperature andwere quence (5'-TGGAA'ITCACTA'ITCCTGCACCAACTGACCTG-3'). PCR stained with crystal violet by a modification of the method of Kueng et cycle conditions were 1 min at 94 "C, 1 min at 45 "C, and 2.5 min at al. (1989). Unboundand loosely bound cellsin experimental wells were 72 "C. The PCR product was purified using Geneclean, digested with removed by shaking and aspiration, and the remaining cells werethen

Integrin-Fibronectin-VCA"1 Interactions fixed as described above.The fixative was aspirated, the wells washed three times with 100 pl of H20, and air-dried. 100 pl of 0.1% (w/v) crystal violet in 200 m~ MES, pH 6, was added to each well and incubated at room temperature for 20 min. Excess dyewas removed bythree washes of 100 pl of H20, and bound dye was solubilizedwith 100 pl of 10% (v/v) acetic acid. The absorbance of each well at 570 nm was then measured using a multiscan ELISA reader (Dynatech, Billingshurst, U.K.). Each sample was assayed in triplicate. Attachment of MOLT4 Cells to VCAM-1-transfected COS-1 CellsCOS-1 cells, an SV40-transformed African Green monkey kidney cell cultured line (provided by J. Cunliffe, Zeneca, Macclesfield, U.K.), were in Dulbecco's minimal essential medium containing 5% (v/v) fetal calf serum, 2 n m L-glutamine, and 50 p g / d gentamicin (all from GIBCO). Cultures were detached with 0.05% (w/v) trypsin, 0.02% (w/v) EDTA and plated at 2 x lo5 celld60-mm dish 24 h prior to transfection. Cells were transfected with full-length VCA"1 cDNA cloned into the eukaryotic expression vector pCDM8 (Seed (1987);obtained from British Biotechnology Ltd.)using the calcium phosphate coprecipitation transfection method(Sambrooket al., 1989). Briefly,500 pl of 280 nm NaC1, 50 nm HEPES, pH 7.1, was mixed with 10 pl of 70 nm sodium phosphate buffer, pH 7.1, and added dropwise, with mixing, to a solution containing 500 pl of 150 Wml pCDM8-VCA"1 and 10 p1 of 2 M CaCl,. ARer incubating for 10 min at room temperature, the suspension was mixed with 6.5 ml of Dulbecco's minimal essential medium, added to 60-mm dishes of COS-1 monolayers,and incubated at 37 "C for6 h. The DNA-calcium phosphate coprecipitate was then removed and the cells shocked with 10% (v/v) demethyl sulfoxide, Dulbecco's minimal essential medium for 1 min. Dimethyl sulfoxidewas then aspirated and the cells washed three to five times with Dulbecco's minimal essential medium. The transfected COS-1 cells were detached 24-h post-transfection using 5 rn EDTA in Hanks' balanced salts solution (GIBCO) and seeded a t 1.5 x lo4 celldwell in microtiter plates precoated with 15 pg/ml fibronectin. To check forthe level of VCA"1 expression, parallel 60-mm dishes of transfected COS-1 cells were detached 48-h post-transfection with 5 rnEDTAin Hanks'balanced salts solution, and analyzed by flow cytometry using monoclonal antibody BBA-6 (mouse anti-human VCA"1; obtained from British Bio-technology Ltd.). Routinely, >30% of cells were found to be positive for VCA"1. 107-108MOLT-4 cells, cultured as described above, were harvested, washed twice, resuspended in 100 pl of RPMI-1640 medium, and incubated with 3 pCi of NaZ5lCrO4(Amersham, U.K.; specific activity 100-500pCi/pg) at 37 "C for 1 h. Cells were then washed twice with WMI-1640 medium containing 10%(v/v) fetal calf serum and resuspended to 6 x lo6 cells/ ml. 48 h post-transfection, 50 pl of Wr-labeled MOLT-4 cells and 50 pl of RPMI-1640 with or without antibodies or peptides were added to preformed COS-1 cell monolayers in microtiter plates, and incubated for 30min at 37 "C in a humidified atmosphere of 5% (v/v)COz. Loosely bound cells were removed by three washes with phosphate-buffered saline, and bound cells were solubilizedwith 100 pl of 0.1 M NaOH for 15 min at room temperature. Aliquots were then counted for radioactivity on a gamma counter. Cell Spreading"A375-SM cells, a human metastatic melanoma cell line (provided by I. J. Fidler, M. D. Anderson Hospital and University of Texas, Houston) were cultured as described (Kozlowskiet al., 1984)in Eagle's minimal essential medium containing 10% (v/v) fetal calf serum, minimal essential medium vitamins, non-essential amino acids, 1 nm sodium pyruvate, and 2 rnglutamine (all from GIBCO).Cell-spreading assays were performedessentially as described previously(Humphries et al., 1986; Mould and Humphries, 1991). Briefly, wells were coated and blocked as described above forthe cell attachment assay. A375-SM cells weredetached with 0.05% (w/v)trypsin, 0.02% (w/v)EDTA, resuspended to 2 x 105/ml in Dulbecco's minimal essential medium, and allowed to recover for 10min at 37 "C. For experiments examining the effects of antibodies or peptides on spreading, 50-pl aliquots of the cell suspension were added to wells together with 50 pl of antibodies or peptides diluted with Dulbecco's phosphate-buffered saline and were then incubated in a humidified atmosphere of 5% (v/v) COz for 90 min at 37 "C. The cells were fixed with 3% (w/v)glutaraldehyde and monitored for the degree of spreading using phase contrast microscopy as described (Humphries et al., 1986). Each point was obtained by counting 300 celldwell from a number of randomly selected fields. No cell spreading was observed on wells coated only with heat-denatured BSA. Solid-phase Receptor-LigandBinding Assay-Lymphocytes were purified from -50 leukocyte concentrates (provided by the Blood Transfusion Services of Liverpool, Lancaster and Manchester, U.K.) by centrifugation on Histopaque 1077 (Sigma). Cells were then washed with phosphate-buffered saline, centrifuged at 400 x g for 10 min, and extracted with 100 ml of 2% (w/v) TritonX-100,5 mg/ml, 150nm NaC1, 1

4007

rn CaCI2, 1 rn MgC12, 2 rn phenylmethylsulfonyl fluoride, 10 pg/ml leupeptin, 25 nm Tris-HC1, pH 7.4, on ice for 30 min. The lymphocyte extract was clarifiedby centrifugation at 40,000 x g for 30min,and the resulting supernatant was precleared by rotary mixing with 4 ml of rat IgG-Sepharose beads for 1h at room temperature. IgG-Sepharose was prepared by coupling at a ratio of 2 mg of IgG to 1ml of CNBr-activated Sepharose (Pharmacia) according to the manufacturer's instructions. The IgG-Sepharose was removed by column filtration, and the filtrate was rotary mixed with 4 ml of monoclonal antibody 13-Sepharosefor 2 h at room temperature. The suspension was then packed into a 1.6-cm diameter column and washed overnight a t 4 "C with 0.1% (w/v) Triton X-100,150 rn NaCl, 1m~ CaC12,1rn MgC12, 25rnTris-HC1, pH 7.4. Bound material was eluted with 0.1% (w/v) TritonX-100, 1 rn CaC12, 1 rn MgCl,, 10 rn sodium acetate, pH 3.2, and 1-ml fractions were collected and neutralized with 0.5 ml 1M Tris-HC1, pH 8.5. 50-pl aliquots were analyzed by SDS-polyacrylamide gelelectrophoresisusing a 7.5% non-reducing resolving gel (Laemmli, 1970) and Coomassie Blue staining and shown t o contain p l integrins of 290% purity. By dotblotting, a4pl was shown to be the most abundant integrin present. Solid-phaseligand-receptorbinding was performed by a modification of the method of Char0 et al. (1990).Ligands (rsVCAM-1or W120) were mixed with equal amounts of ImmunoPure sulfo-N-hydroxysuccinimido biotin (Pierce, Chester, U.K.) and rotary mixed for 30-40 min at room temperature. The mixture was then dialyzed against several changes of Purified 150 n m NaCl, 25m~ Tris-HC1, pH 7.4,t o remove excess biotin. integrins (at a concentration of -0.5 mg/ml) were diluted 1:lOO with phosphate-buffered saline containing divalent cations, and 100-p1 aliquots were added to the wells of a 96-wellELISA plate (Dynatech Immulon).Plates were incubated overnight at 4 "C, and wells werethen blocked with 200 p1 of 5% (w/v)BSA (Sigma), 150 rn NaCl, 0.05% (w/v) NaN3, 10m~ Tris-HC1, pH 7.4, for 1-3h. ELISAexperiments confirmed that a4p1was bound to the wells after these coating steps. Wells were then washed twice with 200 pl of 150 nm NaCl, 1 rn MnC12, 1 nm MgCl,, 25 nm Tris-HC1, pH 7.4(bufferA), and 100-pl aliquots of 0.5-2.0 pg/ml biotinylated ligands diluted in buffer A with 1m g / d BSA, were added, with or without competitor peptides or proteins. The plate was then incubated at 30 "C for 3 h, at which time binding had reached a plateau. Biotinylated ligands were aspirated and the wellswashed three times with buffer A. Bound ligand was quantitated by addition of 100 pl of1:200 ExtrAvidin-peroxidase conjugate (Sigma) in buffer A with 1 mg/ml BSA for 30 min at room temperature. Wells were then washed three times with buffer A, and color was developed using TurboTMB substrate (Pierce) or ABTS (Sigma). The level of nonspecific binding was measured in every experiment by determining the level of binding to wells coated with BSA alone. Control experiments showed that theinteraction of either biotinylated rsVCAM-1 or W120 with the p l integrin preparation in this assay was specific since binding was substantially, and approximately equally, inhibited byEDTA and by anti-functional monoclonal antibodies directed against either the a4 or p l integrin subunits, butnot by an a n t i 4 antibody (Table I). RESULTS

CSl PeptideInhibitsa4pl-mediatedCellAdhesionto VCA"l"T0 assess the relationship between VCA"1 and fibronectin binding to a4p1,we initially investigated the ability of soluble CS1 peptide to interferewith VCAM-1-mediated adhesion. Three different assays were used: two short term assays monitoring theearlyattachmentphase of adhesion (MOLT-4 T lymphoblastic leukaemia cell attachment to plasticimmobilized rsVCA"1 and MOLT-4 attachment toCOS-1 cell monolayers transiently expressing full-length transfected VCAM-l), and a longer term assay monitoring the subsequent morphological changes occurring on adhesion (A375-SM melanoma cell spreading on plastic-immobilized rsVCAM-1). Initial characterization of each assay showed, as expected, that adhesion to VCA"1 was inhibited 290%by the addition of either anti-a4 (HP2/1) or anti-pl (13) monoclonal antibodies to the assay medium, but not by other anti-integrin antibodies (e.g. anti-a5 monoclonal antibody 16; data not shown). In all three assays, CS1 perturbed adhesion in a dose-dependent manner (Fig. 1). The results from both attachment assays were very similar, with half-maximal inhibition of adhesion to VCA"1 occurring at 2-3 mg/ml (Fig. L4 and B ) . Consistent with previous studies reporting a requirement for

4008

Integrin-Fibronectin-VCAM-1 Interactions c

A

-

50

25

0

1

2

3

0

1

2

3

0

1

2

3

CS1 (mghl)

FIG.1. Effect of CS1 peptide on VCAM-1-mediatedcell attachment/spreading. Each panel representsa different assay; A, MOLT-4 cell attachment to immobilized rsVCA"1; B , MOLT-4cell attachment to VCAM-1-transfected COS-1 cells; and C, A375-SMcell spreading on rsVCAM-1. The level of attachmenthpreading to VCA"1 is shown by the closed circles. In A and C the effect of CS1 on CS1-I&-mediated attachmentlspreading was also examined (open circles). Control peptides (C, squares) were either CS6 (A and B ) or CSlscrl (C); both are previously characterized as inactive in cell adhesion assays (Humphrieset al., 1986; Komoriya et al., 1991). Error bars represent S.D. In B , the horizontal line representsthe level of MOLT-4 attachment to untransfected COS-1 cells (attachmentto mock-transfected cellswas indistinguishable from this background level; data not shown).To make comparison easier, the level of adhesion obtained without competitor was taken as 100%. The actual values were as follows:A, rsVCA"131.8 f 0.6%; CS1-I&32.0 f 1.2%;B , 19,140countdmin which is equivalent to 17.1%attachment; C, rsVCAM-190 2 3%; CS1-IgG 782 5%. Inthe experiment which yielded data for C, CS1 was also examined for its ability to perturb spreading on a 110-kDa cell-bindingfragment of fibronectin containingthe RGDS sequence. In accordance withthe recognition of this region by the integrin (r5p1, CS1was found to have no inhibitory activity (78 2 2% spreading at 3 mg/ml on 5 pg/ml 110-kDa fragment).

TABLE I Effect of anti-integrinantibodies and EDTA on t?te binding of biotinylated rsVCAM-1 and HI120 to immobilized p l integrins All antibodies were present at 10 pg/ml; EDTA was used at 5 m M . Results are expressed as percentage binding relative to an untreated control. Errors = standard deviation.

not control peptides, ina dose-dependent manner (Fig. 2). Halfmaximal inhibition of VCA"1 binding was achieved at 13 pg/ml; a 4-fold lower concentration of CS1 was required for half-maximal inhibition of W120 binding. Hence, the solidphase assay appears toreproduce the results obtained incellbased adhesion assays, although with greatly enhanced sensiTreatment rsVCAM-1 binding W120 binding tivity. % Recombinant HepIIIIIICS Blocks Solid-phase VCA"l-a4pl None 100 f 13.8 100 16.6 Interaction-In the samesolid-phase assay, soluble W120 was HPW1 (anti-a4) 24.3 f26.5 1.9 * 10.5 mAbl3 (anti-pl) 25.9 * 4.0 36.7 3.5 also found to inhibit VCA"1 binding (Fig. 3A). As a control, mAbl6 (anti-a5) 88.52 6.1 87.6 * 8.2 soluble WO, which is a related recombinant fibronectin fragEDTA -2.3 * 3.4 12.5 2 8.6 ment lacking the IIICSregion, was unable to significantly inhibit VCAM-1 binding (inset toFig. 3 A ) . These results demonhigher levels of antiadhesive peptides to inhibit attachment strate that the authentic a4pl-bindingdomain of fibronectin, compared with spreading, CS1 was more potent at blocking cell and not just its principalactive site peptide, is able to inhibit spreading (half-maximal inhibition at 1.7 mg/ml; Fig. 1C). In VCAM-1-receptor interactions. Interestingly, unlabeled all cases, control peptides had little or no inhibitory activity. VCA"1 was approximately %fold more active on a molar basis Furthermore, CS1 did not inhibit cell spreading on a 110-kDa at inhibiting bindingthan W120 (Fig. 3, legend 1. In the reverse fragment of fibronectin containing the centralcell-binding domain (Fig. l,legend). These results demonstratethat CS1 pep- experiment (Fig. 3B), VCA"1 was also approximately %fold tide can specifically block adhesion either to a soluble form of more potent for blocking the binding of biotinylated W120 to a 4 p l t h a nunlabeled W120. VCA"1 or to theauthenticproteinintercalatedintothe CS1 Inhibition of VCA"l-ar4pl Binding Is Directly ComCOS-1 cell plasma membrane.Comparisons of the effectiveness of CS1 at blocking adhesion to VCA"1 or to itself (when im- petitive in Nature-CS1 inhibition of VCA"l-a4pl binding could be explained by an allosteric effect of one ligand altering mobilized as an IgG conjugate) revealed that it was 3-7-fold the bindingsite for the other(which may be either competitive less potent on VCA"1 (Fig. 1, A and C ) . This result was or non-competitive), orby a direct competition between thetwo obtained consistently in repeated assays. CSl Peptide Blocks VCA"l-a4pl Interaction in a Solid- ligands for binding to the same or spatially overlapping sites. phase Binding Assay-In order to elucidate the natureof the These possibilities were distinguished by performing kinetic cell-free solid-phase receptorinhibitory effects of the CS1 peptide on VCA"l-a4p1 interac- analyses of CS1 inhibition in the tions, a solid-phasereceptor-ligand bindingassaywas em- ligand bindingassay. Specifically, the effect of constant concenployed. In this assay, equimolar concentrationsof biotinylated trations of CS1 on the dose-dependent binding of rsVCAM-1 to in a doubleligands (rsVCAM-1 or W120, a recombinant fragment of fibro- a4pl wasexamined. These data were represented nectin containing theHepII/IIICS region) were incubated with reciprocal plot (Fig.4); by linear regression analysis eachbinding curve wasfound to intercept at a similar location on t h e y plastic-immobilized a4pl-containingintegrinpreparations, and thelevel of binding was detected using ExtrAvidin peroxi- axis (3.49 0.32 (no peptide), 3.15 T 1.40 (5 pg/ml CSl), and dase. Both binding interactions werefound to be specific since 2.21 T 1.36 (10 pg/ml CSl), i.e. within r l S.D. of each other), demonstrating that CS1 inhibitionis competitive. Using nonthey were sensitive to anti-a4 and anti-pl antibodies and were linear regression analysis (Walmsley and Lowe, 1985), the data divalent cation-dependent (Table I). The bindingof equimolar concentrationsof both biotinylated could be satisfactorily fitted to a model in which CS1 and rsVCAM-1 and W120 to a4pl was blocked by CS1 peptide, but VCA"1 compete for binding to the same site. Furthermore,

Integrin-Fibronectin-VCA"1 Interactions

0

10

20

30

40

4009

50

0 CSl Cdml) FIG.2. Inhibition of the binding rsVCAM-1 and W120to alp1 in solid-phase by CS1. Biotinylated rsVCA"1 (closed circles) or W120 (opencircles)at equimolar concentrationswere allowedt o bind to u4pl in the presence of various concentrations of CS1 peptide or a control peptide (squares). The control peptide ( C ) was CS6. Data are expressed as the mean of six replicate wells. Error bars represent S.D. The results areexpressed relative to the level of binding in theabsence of competitors.

plots of l/rsVCA"l binding versus CS1 concentration were linear (not shown), providing no evidence for allosteric inhibition (Mahler and Cordes, 1966).

50

150

100

200

PROTEIN (nM) T

6 LI L

100

z

80

0

60

0

40

DISCUSSION

The principal findings in this report are(i) CS1, the major a4pl-binding peptide from the HepIYIIICS region of fibronectin, is able to inhibit a4pl-mediated cell adhesion to its second ligand, VCAM-1, in three different assays; (ii) CS1 and a recombinant HepIVIIICS fragment of fibronectin (W120)inhibit a4p1-VCA"1 binding ina solid-phase receptor bindingassay; (iii)CS1 inhibition ofVCA"1 binding is directly competitive in nature, suggesting that the VCA"1 and fibronectin-binding sites on a4pl are eitheridentical or spatially very close. Our initial studies examined the ability of CS1 peptide to interfere withVCAM-1-dependent cell adhesion. CS1 isone of three active a4pl-binding sites within the HepIIiIIICS region of fibronectin, but in peptide form has at least an order of magnitude greater activity in adhesion assays than the other two peptides, and therefore dominates the activity of this domain (Humphries et al., 1987; Mould and Humphries, 1991). CS1 is a comparatively potent peptide and has been demonstrated to perturb HepIIiIIICS-mediated adhesion at concentrations aslow as 50 p~ (Humphries et al., 1987). Three assays were selected to examine the effects of CS1 on VCA"1 function; these examined adhesion to both the authentic membrane-intercalated form of VCA"1 (transiently expressed in COS-1 cells) and to an immobilized recombinant soluble form of the protein. They also examined the ability of VCA"1 to promote early phase attachment and later phase spreading of cells. In each case, CS1 substantiallyinhibited VCAM-1-mediated adhesion. When the ability of CS1 to block adhesion to VCA"1 was compared with its ability to block adhesion to itself (as an immobilized IgG conjugate), it was found to be between 3- and 7-fold less effective on VCAM-1 substrates. This relatively small difference was nonetheless highly reproducible in assays inwhich the level of control adhesion to each ligand was similar. Inhibitor screening in cell-based adhesion assays is always subject to thecriticism that theeffects of the putativeinhibitor might be indirect and might operate on other cellular structures. To discount this possibility, we established a cell-free solid-phase receptor-ligand bindingassay which could measure

20

0 0

10

20

30

40

PROTEIN (nM) FIG.3. Cross-inhibition of the bindingrsVCAM-1 and W120 t o a4Pl.A, effect of W120 protein (open circles) and unlabeled rsVCAM-1 (closed circles) on the binding of biotinylated rsVCAM-1 to u4pl. The concentrations of W120 and rsVCAM-1for 50%inhibition are -110 and 37.5 m, respectively.Inset, comparison of the effect of W120 or WO ( 5 pg/ml) on the binding of biotinylated rsVCAM-1. B , effect of rsVCA"1 (closed circles) and unlabeled W120 (open circles) on the binding of biotinylated W120 binding t o a4pl. The concentrations of W120 and rsVCA"1 for 50% inhibition are -12.5 and 3.9 m, respectively. The results are expressed relative to the level of binding in the absence of competitors. Data are expressed as the mean of six replicate wells. Error bars represent S.D.

the interaction of either rsVCA"1 or a recombinant form of the HepIIiIIICSdomain of fibronectin (called W120)with a4pl. We then examined the ability of CS1 peptide to perturb both interactions. Our resultsconfirmed the conclusions of adhesion assays in that they demonstrated that CS1 could completely block both receptor-ligand interactions. Interestingly, the relative potency of the peptide in blocking binding of the two ligands was similar to that observed in adhesion assays (it was 4-fold less effective at blocking VCAM-1 binding). The solid-phase assay was alsotwo orders of magnitude more sensitive than adhesion assays for detecting inhibition; half-maximal inhibitionof VCA"1 binding to a 4 p l was obtained at a CS1 concentration of 13 pg/ml, compared with 1.7 mg/ml in the spreading assay. This result probably reflects both the high degree of cooperativity of receptor engagement that is required for cell adhesion and the high concentration of ligand on the surface of the microtiter plate wells in cell adhesion assays. It also suggests thatsimilar solid-phase assays may be extremely usefulfor future screening of VCA"1 and fibronectin antagonists.

4010

Integrin-Fibronectin-VCAM-1 Interactions

inhibition of adhesion to VCA"1 may only become apparent at high concentrations of FN-40. The competition between V C A " 1 and fibronectin for bind2o ing to a4pl is reminiscent of other integrin-ligand interactions. Fibrinogen, fibronectin, and von Willebrand factor block the A450 binding of each other to the platelet integrin aIIbp3, possibly because of their common use of the RGD active site motif (Plow et al., 1985. A second active site peptide from fibrinogen, the lo y-chain sequenceLGGAKQAGDV,which bears homology to the NO INHIBITOR LDV active site of CS1, appears to bind to a similar site to RGDS since each peptide competes forthe binding of the other to aIIbp3 (Lam et al., 1987; Santoro and Lawing, 1987;DSouza 0 1.25 2.5 et al., 1990).A comparable result is also found for two a4plbinding peptides from fibronectin, the LDV-containing CS1 and 2 VCAM-1 Qa/ml) the REDV-containing CS5: soluble CS1 blocks CS5-mediated FIG.4. Kinetic analysis of CS1 inhibition ofVCA"l-rr4b1 bind- adhesion and vice versa, and in both cases the inhibition can be ing. The bindingof a rangeof rsVCAM-1 concentrationsto immobilized overcome in a competitive manner by increasing the density of 01401was monitored in the absence or presence of two different concen- substrate-bound ligand (Mould et al., 1991). trations of CS1 peptide (5 and 10 pg/ml). The results are expressed as The ability of fibronectin and VCAM-1 to compete with each double-reciprocals.After linear regression analysis, each curve is seen to intercept within 1 S.D. of each other on they axis (see text for data). other for binding to their common receptor raises important questions about the atomic basis of the interactions. In particular, does VCA"1 contain a CS1-like sequence, or are the two In other experiments, the recombinant W120 protein was ligand active sites different in nature? The first and fourth also found to block solid-phase VCA"l-a4pl binding, demonimmunoglobulinmodules within VCA"1 have beenimplistrating thatthis activity was inherent in thefibronectin molcated as active sites for a4pl binding (Taichman et al., 1991; ecule itself and not just in its active site peptide. In agreement Osborn et al., 1992; Vonderheideand Springer, 1992)and while with the CS1 peptide inhibition results, W120 was severalfold neither contains a sequence that is homologous to the entire more active on a molar basis for blocking the interaction of CS1 sequence, both contain short peptide motifs that are rea4pl with W120 than with VCAM-1. A further advantage of the solid-phase receptor binding as- lated to the LDV active site of CS1. In the future, it will be say is that it is possible to obtain limited kinetic information important to test the possible functional importance of these about the modeof action of inhibitors. By measuring dose- sites with a view to explaining the results in this report. The dependent binding of V C A " 1 in the presence of a series of final implication of this report is that derivatives of the CS1 concentrations of CS1, it was possible to construct double-re- peptide, and its LDV active site, may be effective agents for ciprocal plots that gave insight into the mode of action of the blocking VCA"1 function in vivo. In support of this, CS1 or shorter peptides containing the LDV sequence have been found peptide. These studies revealed that CS1 inhibition of VCA"1 to inhibit the contact hypersensitivity response in mice (Ferwas directly competitive in nature, suggesting that itwas acting by competing forthe VCAM-1-binding site. Since CS1 con- guson et al., 1991)and to block the VCAM-1-mediated interacstitutes only a small fraction of the fibronectin molecule (25 tion of lymphocytes with cultured high endothelial cells (Ager and Humphries, 1990). In the same way that the RGD seamino acids), this implies that the fibronectin and VCA"1binding sites on a4pl are spatially very close and possibly quence has proven to be a lead compound forthe generation of identical. CS1 may be a better inhibitor of fibronectin binding antithrombotic agents, LDV may be useful for the generation of either because there is only partial overlap between the bind- anti-inflammatory agents. ing sites, or because the two ligands bind to the same essenAcknowledgments-We express our thanks to I. J. Fidler for tially the same site but VCA"1 binds with a higher affinity. A375-SM cells, J. Cunliffe for COS-1 cells and for advice on transient Either of these two explanations could also account for why transfection,S. K. Akiyama andK. M. Yamada for providingantibodies, some anti-a4 antibodies are preferential inhibitors of fibronec- and the Blood Transfusion Services at Liverpool,Lancaster,and Manchester for generously providing leukocyte concentrates. We are tin binding (Pulido et al., 1991). Future experiments will be grateful to A. G. Lowe for advice and helpful discussions. directed toward distinguishing between these possibilities. Establishing the relationship between the mechanisms of REFERENCES receptor binding by fibronectin and VCAM-1 is importantsince Ager, A,, and Humpbries, M. J. (1990)Int. Irnrnunol. 2,921-928 it hasimplications for the design of selective inhibitors of each Bochner, B. S., Luscinskas, F. W., Gimbrone, M.A., Newman, W., Sterbinsky, S. A., Derse-Anthony, C. P., Klunk, D., and Schleimer, R. P. (1991)J. Exp. Med. 173, interaction. Early studies suggested that thebinding sites were 1553-1556 distinct since an anti-a4monoclonal antibody, HP1/3,appeared Carlos, T. M., Schwartz, B. R., Kovach, N. L., Yee, E., Rosso, M., Osborn, L., to perturb only VCAM-1-mediated celladhesion. However, deChi-Rosso, G., Newman, B., h b b , R.. and Harlan, J. M. (1990) Blood 76,965970 tailed functional mapping of a series of anti-a4 monoclonal I. F., Nannizzi, L., Smith, J. W., and Cheresh, D.A. (1990) J . Cell Biol. 111, antibodies showed that HP1/3 belongsto the subset of antibod- Charo, 27952800 ies that arepreferential, but not exclusive, inhibitors of fibro- Clark, E.A,, Grabstein, K. H., and Shu, G . L. (1992) J. Irnrnunol. 148,3327-3335 nectin binding to a4pl (Pulido et al., 1991); hence the above Cybulsky, M. I., and Gimbrone, M. A. (1991) Science 251,788-791 M.I., Fries, J. W. U., Williams, A. J.,Sultan, P.,Davis, V.M., Gimbrone, result may have been erroneous. Elices et al. (1990) also re- Cybulsky, M. A,, and Collins, T.(1991)Arn. J. Pathol. 138,815-820 ported that low concentrations of a soluble 40-kDafragment of DSouza, S. E., Ginsberg, M. H., Burke, T. A,, and Plow, E. F. (1990) J. Biol. Chem. 266,344&3446 fibronectin containing the HepII and CS1 regions (FN-40) Dufour, S.,Duband, J.-L., Humphries, M. J., Obara, M., Yamada, K. M., and Thiery, blocked cell binding to FN-40 but had no effect on adhesion to J. P. (1988) EMBO J. 7,2661-2671 VCAM-1. Our observations suggest that this result can be ex- Elices, M. J.,Osborn,L., Takada, Y., Crouse, C., Luhowskyj,S., Hemler, M.E., and Lobb, R. R. (1990) Cell 60,577484 plained by the requirement for higher concentrations of W120 Ferguson, T. A., Mizutani, H., andKupper,T. S. (1991) Proc. Natl. h a d . Sei. (a functionally equivalent protein to FN-40) to block the inter(I. S. A. 88,8072-8076 action of a4pl with VCAM-1, than with W120 itself; hence, Freedman, A. S., Munro, J. M.,Rice, G. E., Bevilacqua,M. P., Morimoto, C.,

i

t /A

P

Zntegrin-Fibronectin-VCA"1 Interactions McIntyre, B. W., Rhynhart, K., Pober, J. S., and Nadler, L.M. (1990)Science 249,1030-1033 Freedman, A. S., Munro, J. M., Morimoto, C., McIntyre, B. W., Rhynhart, K., Lee, N., and Nadler, L. M. (1992)Blood 79,206-212 Guan, J.-L., and Hynes, R. 0.(1990)Cell 80, 53-61 Hemler, M . E., Elices, M. J., Parker,C., and Takada, Y. (1990)Zmmunol. Reu. 114, 45-45 Hession, C., Tizard, R., Vassallo, C., Schiffer,S . B., Goff, D., Moy, P., Chi-Rosm, G., Luhowskyj, S., Lobb, R.. and Osborn, L. (1991)J. Biol. Chem. 268,6682-6685 Humphries, M. J., Akiyama, S . I€, Komoriya, A,, Olden, K, and Yamada, K. M. (1986)J. Cell Biol. 103, 2637-2647 Humphries, M. J., Akiyama, S . K., Komoriya, A,, Olden, K., and Yamada, K.M. (1988)J. Cell Biol. 106, 1289-1297 Humphries, M.J., Komoriya, A,, Akiyama, S. K., Olden, K., and Yamada, K. M. (1987)J. Bwl. Chem. 282,6886-6892 Iida, J., Skubitz, A. P. N., Furcht, L. T., Wayner, E. A., and McCarthy, J. B. (1992) J. Cell Bwl. 118,431-444 Komoriya, A,, Green, L. J., Mervic, M., Yamada, S. S., Yamada, K. M., and Humphries, M. J. (1991)J. Biol. Chem. 266,15075-15079 Kozlowski, J. M., Hart, I. R., Fidler, I. J., and Hanna,N. (1984)J.Natl. CancerZnst. 12,913-917 Kueng, W., Silber, E., and Eppenberger, U. (1989)Anal. Biochem. 182, 1619 Kunkel, T. A,, Roberts, J. D., and Zakour, R. A. (1987)Methods Enzymol. 164, 367-382 Laemmli, U. K. (1970)Nature 227,680485 M 6 n , A, Garcia-Vicuiia. R., Humbria, A,, Postigo, A. A., Corbi, A. L., De Land& zuri, M. O., and, Shchez-Madrid, F. (1991)J. Clin. Znuest. 88,54&552 Lam, S . C.-T., Plow, E. F., Smith, M. A,, Andrieux, A . , Ryckwaert, J . J . , Marguerie, G., and Ginsberg, M. H. (1987)J. Biol. Chem. 262,947-950 Masinovsky, B.,Urdal, D., and Gallatin, W.M. (1990)J. Zmmunol. 146,2886-2895 Mahler, H. R., and Cordes, E. H. (1966)Biological Chemistry, Harper and Row, N Y , pp. 25&253 Miyake, K., Medina, K., Ishihara, K, Kimoto, M., Auerbach, R., and Kincade, P.W. (1991)J. Cell Biol. 114, 557-565 Mould, A. P., and Humphries, M. J. (1991)EMBO J 10,4089-4095 Mould, A. P., Wheldon, L. A., Komoriya, A,, Wayner, E. A,, Yamada, K. M., and Humphries, M. J. (1990)J. Bwl. Chem. 286,40204024 Mould, A. P., Komoriya, A,, Yamada, K. M., and Humphries, M. J. (1991)J. Biol. Chem. 266,35793585

4011

Osborn, L., Hession, C., Tizard, R., Vassallo, C., Luhowskyj,S., Chi-Rosso, G., and Lobb, R. (1989)Cell 59, 1203-1211 Osborn, L., Vassallo, C., and Benjamin, C. D. (1992)J. Exp. Med. 176,99-107 Plow, E. F., McEver, R. P., Coller, B. S., Woods, V. L., and Marguerie, G . A. (1985) Blood 88, 724-727 Polk, T., Newman, W., and Gopal, T.V. (1990)Nucleic Acids Res. 18,5901 Polk, T., Newman, W., Raghunathan, G., and Gopal, T.V. (1991)DNA Cell Biol. 10, 349-357 Pulido, R., Elices, M. J., Campanero, M. R., Osborn, L., Schiffer, S., Garcia-Pardo, A,, Lobb, R., Hemler, M. E., and Sanchez-Madrid, F. (1991)J. Biol. Chem. 266, 10241-10245 Rice, G. E., and Bevilacqua, M. P. (1989)Science 248, 1303-1306 Rosen, G . D.,Sanes, J. R., Lachance, R., Cunningham, J. M., Roman,J.,and Dean, D.C. (1992)Cell 89,1107-1119 Sambmok, T., Fritseh, E. F., and Maniatis, T. (1989)Molecular Cloning: a Lahratory Manual, 2nd Ed.. Cold Spring Harbor Laboratory, Cold Spring Harbor,

NY Sanger, E, Nicklen, S., and Coulsen,A. R. (1977)Proc. Natl. Acad. Sci. U.S . A. 74, 54633467 Santoro, S . A,, and Lawing, W. J. (1987)Cell 48, 867-873 Seed, B. (1987)Nature 329,840-842 Simmons, P. J.,Masinovsky, B., Longenecker, B. M., Berenson, R., lbrok-Storb, B., and Gallatin, W.M. (1992)Blood 80,388-395 Smith, D. B., and Johnson, K S . (1988)Gene (Amst.)67,3140 Springer, T. A. (1990)Nature 346,425-434 Taichman, D.B., Cybulsky, M.I., Djaf€ar, I.. Longenecker, B. M., Teixido, J., Rice, G. E., M o , A,, and Bevilacqua, M.P. (1991)Cell Regul. 2,347455 Teixido, J., Hemler, M. E., Greenberger, J. S., and Anklesaria, P. (1992)J. Clin. Znuest. 90, 3 5 W 6 7 Thornhill,M. H., Wellicome,S . M., Mahiouz, D. L., Lanchbury,J. S . S., Kyan-Aung, U., and Haskard, D. 0.(1991)J. Zmmunol. 146,592-598 Vonderheide, R. H., and Springer, T. A. (1992)J. Exp. Med. 176,1433-1442 Walmsley, A. R., and Lowe, A. G . (1985)Computer Methods Pmg. Biomed. 21, 113-118 Wayner, E. A, Garcia-Pardo,A,, Humphries, M. J., McDonald, J. A,, and Carter, W. G.(1989)J. Cell Biol. 109, 1321-1330 Zardi, L., Carnemolla, B., Balza, E., Borsi, L., Castellani, P., Rocco, M.,and Sin,A. (1985)Eur: J. Biochem. 146,571479