Ligand binding to integrin α_vβ₃requires tyrosine 178 in the α_v subunit

Integrins are a large family of αβ heterodimeric adhesion receptors that is often subdivided into groups based on 8 known integrin β subunits.1 The β₃-integrin subfamily is composed of α_vβ₃, originally identified as the vitronectin receptor, and α_IIbβ₃, a platelet-specific receptor for fibrinogen and von Willebrand factor. α_IIbβ₃ plays a crucial role in platelet aggregation, normal hemostasis, and pathological thrombus formation.2 On the other hand, α_vβ₃ is expressed in a number of tissues, including platelets, endothelial cells, vascular smooth muscle cells, and osteoclasts, and it plays a key role in angiogenesis and bone resorption.3,4 

The α_IIb and α_v subunits are homologous and 36% identical in primary amino acid sequence,5 and most ligands that bind to α_IIbβ₃, including fibrinogen, von Willebrand factor, and vitronectin, also bind to α_vβ₃. However, there are some distinctive features between these 2 integrins.3 First, the α_IIb subunit has been found only in combination with β₃, whereas α_v can associate with at least 5 β subunits (β₁, β₃, β₅, β₆, and β₈).6 Second, some ligands, such as osteopontin, matrix metalloproteinase-2, and adenovirus penton base, bind to α_vβ₃ but not to α_IIbβ₃. Third, treatment of α_IIbβ₃ with ethylenediamine tetraacetic acid (EDTA) at 37°C dissociates the complex into its individual subunits; α_vβ₃ remains a heterodimer. Finally, the ligand-binding function of α_vβ₃, but not α_IIbβ₃, is suppressed by calcium (Ca²⁺).7 

Generally, all integrins require divalent cations for ligand recognition, and multiple residues important for ligand binding have been identified on both α and β subunits.8 The N-terminal region of integrin α subunits has 7 repeats of homologous sequences of about 60 amino acid residues. Some integrin α subunits (eg, α₂, α_L, and α_M) contain an inserted domain of about 200 amino acids residues (the I-domain) between the second and the third repeats in the α subunit, which is critically involved in ligand binding.9,10 The crystal structure of the I-domain has been determined, and the metal ion-dependent adhesion site (MIDAS) motif that contributes to cation binding, as well as ligand binding, has been clarified.11Interestingly, a MIDAS-like motif essential for the ligand-binding function was also identified in integrin β subunits.12,13 On the other hand, integrin α subunits, such as α_v, α_IIb, and α₄, do not have the I-domain.

The structural basis for the interaction between non–I-domain α subunits and their ligands remains elusive. For the α_IIbsubunit, peptide cross-linking studies have shown that the histamine-histamine-leucine-glycine-glycine-alanine-lysine-glutamine-alanine-glycine-aspartic acid-valine (HHLGGAKQAGDV) sequence derived from the COOH terminus of the γ-chain of fibrinogen interacts with residues 294-314, encompassing the second putative calcium-binding domain of α_IIb.14 However, recent characterization of molecular defects in Glanzmann thrombasthenia and mutagenesis studies demonstrated several discontinuous residues important for ligand binding: proline (P)145, D163, L183, G184, tyrosine (Y)189, Y190, phenylalanine (F)191, G193, and D224.15-19 Springer has proposed that the 7 N-terminal sequence repeats of integrin α subunits are folded into a β-propeller domain.20 The proposed domains contain seven 4-stranded β-sheets (W1-W7) arranged in a torus around a pseudosymmetry axis. Interestingly, the discontinuous residues identified in α_IIb as important for ligand binding were located in the regions predicted to adopt a β-turn structure on the upper face of the β-propeller model: P145 within the W3 4-1 loop; L183, G184, Y189, Y190, F191, and G193 within the predicted W3 2-3 loop; and D224 within the predicted W4 4-1 loop. Furthermore, the analysis of a Japanese variant of Glanzmann thrombasthenia, KO, and alanine-scanning mutagenesis have shown that D163 in the W3 4-1 loop (cystein [C]146-C167) is essential for ligand binding.16 In contrast, the single available peptide cross-linking study of α_vβ₃ showed that 2 distinct linear regions in α_v, residues 139-167 and 312-349, cross-link to an arganine-glycine-aspartic acid (RGD) peptide.21 

In this study, we took advantage of the data regarding ligand-binding sites in α_IIb and investigated the role in ligand binding of the predicted loops between repeats 2 and 3 (W3 4-1 loop) and within repeat 3 (W3 2-3 loop) in α_v. By performing alanine-scanning mutagenesis of recombinant α_vβ₃ expressed in 293 cells, we demonstrate a critical role for Y178 within the predicted W3 2-3 loop of α_v in ligand binding. In contrast to α_IIb, however, no mutations in the W3 4-1 loop affect ligand binding to α_vβ₃, thereby implying key structural differences in the adhesive ligand-binding sites of the 2 β₃ integrins.

The following monoclonal antibodies (mAbs) were used in the study: LM609,22 a murine mAb specific for α_vβ₃, and LM142,23 a mAb specific for α_v (gift from Dr David Cheresh, The Scripps Research Institute, La Jolla, CA); AP5,24 a mAb specific for β₃ (gift from Dr Thomas Kunicki, The Scripps Research Institute); anti-LIBS1 (anti–ligand-induced binding site 1) mAb (specific for β₃)25 and anti-LIBS6 mAb (specific for β₃)26 (Dr Mark Ginsberg, The Scripps Research Institute); and AP3 (specific for β₃)27 (Dr Peter Newman, The Blood Center of Southeastern Wisconsin, Milwaukee, WI). WOW-1 Fab,28 a monovalent ligand-mimetic mAb specific for activated α_vβ₃, was created by replacing the heavy chain hypervariable region 3 (H-CDR3) of PAC-1 Fab, a ligand-mimetic mAb specific for activated α_IIbβ₃,29 with a single integrin-binding domain of adenovirus penton base. We also used the RGDW peptide (gift from Dr Jiro Seki, Fujisawa Pharmaceutical Co., Osaka, Japan).30 

Wild-type (WT) α_v complementary DNA (cDNA) (gift from Dr David Cheresh, The Scripps Research Institute) and WT β₃ cDNA (gift from Dr Gilbert White, University of North Carolina, Chapel Hill, NC) were cloned into mammalian expression vector pcDNA3 (Invitrogen Corp, San Diego, CA). To introduce single alanine substitutions into α_v, overlap extension polymerase chain reaction (PCR) was carried out as previously described.16For example, to generate the Y178→A (Y178A) α_v mutant, we synthesized mismatched sense primer α_v178A-s, 5′-GGTCCTGGTAGCTTTGCATGGCAAGGTCAGC-3′ (sense, nucleotides [nt] 648-678; mismatched sequences underlined) and antisense primer α_v178A-as, 5′-GCTGACCTTGCCATGCAAAGCTACCAGGACC-3′ (antisense, nt 678-648; mismatched sequences underlined), which were constructed based on the published sequence.31 PCR was performed by using α_v cDNA as a template and primers α_v258-s, 5′-GCAAACACCACCCAGCC-3′ (sense, nt 258-274) and α_v178A-as, or primers α_v178A-s and α_v952-as, 5′-GCAGCCATCTGCTCGCCAG-3′ (antisense, nt 952-934).

The 2 individually amplified PCR products were mixed and used as a template for PCR using primers α_v258-s and α_v952-as. The amplified PCR products were digested with PpuMI and AflIII. The fragments digested with AflIII and XbaI were isolated from the full-length α_v cDNA cloned into pcDNA3. These 2 fragments were introduced together into the pcDNA3 that had been digested with PpuMI and XbaI. The nucleotide sequences of the inserts were confirmed by sequence analysis. In a selected experiment, the C142-C155 loop in α_v was swapped with the corresponding sequence of α_IIb C146-C167. The WT or mutant α_vconstruct was cotransfected into 293 cells with WT β₃construct by the calcium-phosphate method as previously described.32 The cells were cultured in Dulbecco's modified Eagle's medium (DMEM) with 10% heat-inactivated fetal calf serum (FCS) and analyzed 2 days after transfection.

Immunoprecipitation was performed as previously described with slight modification.33 In brief, cells were surface-labeled with sulfo-NHS-biotin (Pierce, Rockford, IL) and lysed in a buffer containing 1% Triton X-100, 25 mM Tris-HCl (tris[hydroxymethyl] aminomethane–hydrochloride), 100 mM sodium chloride (NaCl) (pH 7.4), 0.1 mg/mL leupeptin, 4 μg/mL pepstatin A, 1 mM phenylmethylsulfonyl fluoride, and 10 mM benzamide. Then 200 μg protein from each sample was immunoprecipitated with the mAb LM609, and precipitated bands were identified on Western blots using peroxidase-conjugated avidin.

Fibrinogen (Kabi, Stockholm, Sweden) was labeled with fluorescein isothiocyanate (FITC) as previously described34 and stored at −80°C until use. FITC-fibrinogen binding to 293 cells was assessed by flow cytometry as described.33 Briefly, 50-μL aliquots of 1.5 × 10⁵ washed cells in Ca²⁺-free–Tyrode–HEPES (4-(2-Hydroxyethyl)-1-piperazineethanesulfonic acid) buffer containing 1 mM magnesium dichloride (MgCl₂) were incubated with mAb LM142, specific for α_v (5 μg/mL), for 30 minutes on ice. After washing, 0.5 mM manganese dichloride (MnCl₂) was added into the cell suspension to induce a high-affinity state of α_vβ₃. Cells were then incubated with 150 μg/mL FITC-fibrinogen in the presence or absence of 1 mM RGDW peptide and phycoerythrin (PE)-conjugated antimouse immunoglobulin G (IgG) (1:5 dilution) for 25 minutes at 22°C and then incubated with propidium iodine (PI) (Sigma Chemical Co., St Louis, MO) for 5 minutes at 22°C. After washing, fibrinogen binding (FL1) was analyzed on the gated subset of single high α_vβ₃-expressing (FL2) live cells (PI-negative, FL3). Specific fibrinogen binding was defined as that inhibited by 1 mM RGDW peptide.

For WOW-1 Fab binding to α_vβ₃, cells in Ca²⁺-free–Tyrode–HEPES buffer containing 1 mM MgCl₂ were incubated with 5 μg/mL WOW-1 Fab in the presence of 0.5 mM MnCl₂ for 30 minutes at 22°C. After washing, cells were incubated with 5 μg/mL Alexa-conjugated goat antimouse IgG F(ab′)₂ (Molecular Probes, Eugene, OR) for 25 minutes on ice and then incubated with PI for 5 minutes at 22°C. After washing, WOW-1 binding (FL1) was analyzed on the gated subset of single, living cells. WOW-1 binding to untransfected 293 cells was routinely taken as a measure of nonspecific binding because this value was similar to that obtained for WOW-1 binding to α_vβ₃ transfected cells in the presence of 1 mM RGDW.

For the induction of LIBS on α_vβ₃, washed cells in Tyrode-HEPES buffer containing 1 mM MgCl₂ and 1 mM calcium dichloride (CaCl₂) were incubated with 1 mM RGDW for 30 minutes at 22°C. The cells were then incubated for 30 minutes with AP5, anti-LIBS1, or anti-LIBS6 at a final concentration of 5 μg/mL. After washing, cells were incubated with FITC-conjugated goat F(ab′)₂ antimouse IgG for 25 minutes. The mixtures were incubated with PI for an additional 5 minutes and washed, and mAb binding was analyzed on the gated subset of single, living cells.

Adhesion assays were performed as described by Faull et al.35 Wells of 96-well microtiter plates were coated with up to 1 μg fibrinogen per well in 100 μL phosphate-buffered saline (PBS) and incubated at 4°C overnight. After washing with PBS, wells were blocked with PBS containing 1% bovine serum albumin (BSA) (Sigma) for 90 minutes at 22°C. To determine background adhesion, control wells were coated with 1% BSA. Cells were washed twice with PBS and resuspended in DMEM containing 0.1% BSA at a concentration of 1 × 10⁶ cells per mL. Then, 100-μL aliquots of cell suspension were added to wells in triplicate. The plate was incubated in a humidified 37°C incubator for 60 minutes. After washing with PBS, the adherent cells were checked by visual inspection, and adhesion was quantified by measuring endogenous cellular acid phosphatase activity in an enzyme-linked immunosorbent assay.

Recent studies have demonstrated that several residues within the C146-C167 loop and the G184-G193 loop in the α_IIbsubunit, which correspond to the W3 4-1 loop and W3 2-3 loop in the proposed β-propeller model, respectively, play a critical role in ligand binding. Figure 1 shows the corresponding regions (C142-C155 and G172-G181) in the α_vsubunit and the residues we replaced with alanine. The WT or mutant α_v construct was transiently cotransfected into 293 cells with the WT β₃ construct, and α_vβ₃ surface expression was examined by flow cytometry using the α_vβ₃complex-specific mAb LM609. As shown in Figure2A, the surface expression of mutant α_vβ₃ was 70% to approximately 123% of WT α_vβ₃. Similar results were obtained when mAb LM142 was used to quantify α_v and mAb AP3 was used to quantify β₃ (data not shown).

Because 293 cells normally express α_vβ₁ but not α_vβ₃,36 we were concerned that endogenous α_v might associate with transfected β₃ and contribute to the total α_vβ₃ expressed on these cells. However, by comparing α_vβ₃ transfectants to β₃ transfectants, we found that endogenous α_v contributed no more than approximately 17% to the total α_vβ₃ expressed. Immunoprecipitation experiments employing LM609 further showed that the expression of endogenous α_v associated with transfected β₃ in 293 cells is low, and that the surface expression levels of mutant F177A, Y178A, and W179Aα_vβ₃ are comparable to those of WT α_vβ₃ (Figure 2B). These results indicate that transfected α_v, not endogenous α_v, was the major contributor toward α_vβ₃expression in 293 cells. Moreover, none of the alanine-scanning mutants of α_v adversely affected surface expression of α_vβ₃.

To analyze the ligand-binding function of each mutant α_vβ₃, we initially examined the binding of FITC-conjugated soluble fibrinogen to α_vβ₃. Because α_vβ₃ expressed on 293 cells is present in a low-affinity state and does not bind soluble ligands, cells were incubated with 0.5 mM MnCl₂, which induces a high-affinity state of integrins by a direct effect on the extracellular domain (Figure3A).37 To avoid even a minimal contribution of endogenous α_v in 293 cells to fibrinogen binding, we selectively analyzed the subset of transfectants expressing high levels of exogenous α_vβ₃monitored by LM142. As shown in Figure 3B, a D119Y mutation within the ligand-binding site of β₃ abolished fibrinogen binding to α_vβ₃ as expected,38 confirming the specificity of the ligand-binding assay in this system. Moreover, a Y178A mutation within the predicted G172-G181 loop of α_vabolished fibrinogen binding, and both F177A and W179A mutations adjacent to the Y178 also moderately impaired binding. However, none of the mutations within the C142-C155 loop of α_v (S144A, Q145A, D146A, D148A, D150A, Q152A, and G153A) disturbed fibrinogen binding to the receptor.

To further examine the ligand-binding function of mutant α_vβ₃, we next examined the binding of WOW-1 Fab, a monovalent ligand-mimetic anti-α_vβ₃antibody. The Y178A mutation in α_v, as well as the D119Y mutation in β₃, markedly inhibited WOW-1 binding to MnCl₂-treated cells. On the other hand, swapping of the C142-C155 region of α_v with the corresponding region of α_IIb (C146-C167) did not show a marked inhibition of WOW-1 Fab binding. Because WOW-1 Fab is sensitive to changes in α_vβ₃ affinity rather than avidity,28 these results suggest that the Y178A mutation disrupts the conformation of the ligand-binding pocket in α_vβ₃.

Because the integrin activator MnCl₂ may have additional effects on cells, we cotransfected the WT or the mutant α_v with the T562Nβ₃ mutant, which constitutively activates β₃ integrins and eliminates the need for MnCl₂.33 As shown in Figure4, T562Nβ₃ augmented fibrinogen binding when complexed with the WT α_v, but it failed to induce fibrinogen binding when complexed with the Y178Aα_v. These results indicate that Y178 in the α_v subunit is critical for soluble ligand binding to α_vβ₃.

To further clarify the effects of the Y178A mutation on the structure of α_vβ₃, the reactivities of several mAbs (AP5, LIBS1, and LIBS6) specific for LIBS on β₃ were examined in the absence of ligands. These LIBS epitopes are believed to be outside the ligand-binding pocket of the receptor. As shown in Figure 5A, there was no apparent difference in the reactivities of these mAbs between WT α_vβ₃ and Y178Aα_vβ₃, suggesting that this mutation does not grossly alter the conformation of α_vβ₃. The binding of activation-independent ligands, such as RGD peptides to α_vβ₃, has been shown to induce LIBS expression on the receptor.39Indeed, 1 mM RGDW increased the binding of all 3 LIBS mAbs to WT α_vβ₃. However, it failed to induce LIBS expression on the Y178Aα_vβ₃ mutant. These data suggest that the Y178A mutation in α_v disturbs the binding of small and macromolecular RGD ligands to α_vβ₃.

Because immobilized vitronectin and fibrinogen are activation-independent ligands for α_vβ₃, we further examined cell adhesion to immobilized ligands in the absence of integrin activation. Parent 293 cells showed marked adhesion to vitronectin probably via endogenous α_vβ₁(data not shown),36 whereas they showed only modest adhesion to fibrinogen even at a concentration of 10 μg/mL. Therefore, we examined the effect of the Y178Aα_vβ₃ on the cell adhesion to immobilized fibrinogen but not to vitronectin. Transfection of both WT α_v and β₃ markedly increased the adhesion of the transfectants; transfection of WT β₃ alone induced only a modest increase in adhesion at concentrations of 2.5 and 5 μg/mL (Figure 6). As compared with the WT β₃ transfectant, Y178Aα_vβ₃ failed to increase cell adhesion to immobilized fibrinogen. However, the adhesion of F177A and W179Aα_vβ₃ mutants was only slightly impaired, especially at a relatively low concentration of fibrinogen (1.25 μg/mL). Thus, the effect of Y178A on binding of α_vβ₃ to soluble fibrinogen is also observed with immobilized fibrinogen.

The aim of this study is to reveal the structural basis for the interaction between α_vβ₃ and its ligands, especially ligand-binding sites in the α_v subunit of α_vβ₃. In non–I-domain integrin α subunits, particularly the α_IIb subunit, several ligand-binding sites have been identified by the characterization of naturally occurring mutations in patients with Glanzmann thrombasthenia as well as mutagenesis analyses.15-19 However, ligand-binding regions of α_v remain elusive. To clarify the critical regions for ligand binding in the α_vsubunit, we focused on the predicted W3 4-1 loop (C142-C155) and W3 2-3 loop (G172-G181) and investigated their role in ligand binding by alanine-scanning mutagenesis. The results demonstrate that Y178 in the W3 2-3 loop in α_v is one of the critical residues for ligand binding. In sharp contrast to α_IIb, none of the mutations in the predicted W3 4-1 loop impaired ligand binding, which suggests that there are key structural differences in the adhesive ligand-binding sites of α_vβ₃ and α_IIbβ₃. The differences in the locations of the ligand-binding sites between α_v and α_IIb in the β-propeller model are summarized and illustrated in Figure 7.

There is mounting evidence that the predicted loops between N-terminal repeats 2 and 3 (W3 4-1 loop) and within repeat 3 (W3 2-3 loop) are important for ligand binding in non–I-domain αsubunits.18,40-43 Previous studies have shown that the N-terminal one-third of the α subunit regulates ligand-recognition specificity of β₃ integrins44 and that residues 139-167 in α_v corresponding to the W3 4-1 loop are a cross-linking site for RGD peptides.21 However, the α_v Y178 identified in this study is located in the predicted W3 2-3 loop, and none of the alanine substitutions within the W3 4-1 loop examined impaired ligand binding. The inhibition of ligand binding by alanine substitutions at the residues adjacent to Y178 (F177 and W179) and the failure of the induction of β₃ LIBS with RGDW, even at high concentrations, provide further support for the critical role of Y178 in ligand binding. Although the W3 4-1 loop and W3 2-3 loop are separated in the primary structure, the proposed propeller model suggests that these regions are close to each other in 3-dimensional space and may explain the apparent discrepancy between the previous cross-linking study and the present study (Figure 7).

Interestingly, many critical residues identified in non–I-domain α subunits have aromatic side chains, suggesting that in addition to oxygenated residues such as D163 in α_IIb,16 aromatic residues are important for ligand binding.45 The corresponding residues to the α_v Y178 in α_IIb(Y190),18 α₄ (Y187),42,43 and α₅ (F187),43 and the adjacent residues Y186 and W188 in α₃40 have been demonstrated to be critical for ligand binding, although the role of W3 2-3 loop in α₃ is still controversial.41 In contrast, the role of W3 4-1 loop appears to be different between integrin α subunits. D163 in α_IIb and threonine (T)162 and G163 in α₃, which are located in the predicted W3 4-1 loop, appear critical for ligand binding, whereas the W3 4-1 loop in α₄ and in α_v (this study) do not. More recently it has been demonstrated that the replacement of the W3 4-1 loop in α_v with the corresponding loop in α₅ did not disturb ligand binding but changed ligand-recognition specificity.46 Our swapping mutagenesis of the W3 4-1 loop of α_v with the corresponding region of α_IIb did not abolish WOW-1 Fab binding, suggesting that the W3 4-1 loop is not critical for ligand-recognition specificity between α_vβ₃ and α_IIbβ₃.

WOW-1, a monovalent ligand-mimetic antibody, was created by replacing the H-CDR3 of PAC-1 Fab with a single integrin-binding domain of the adenovirus penton base, a viral coat protein that consists of 5 subunits, each containing an integrin-binding RGD motif. Penton base is known to facilitate adenovirus internalization through α_vintegrins, particularly α_vβ₃ and α_vβ₅.47 WOW-1–like penton base recognizes that activated state of α_vβ₃ and α_vβ₅. Using mAb B5-IVF2 specific for β₅, we have determined that 293 cells express β₅ as well as α_v (S.H. and Y.T., unpublished data, June 1999). Thus, WOW-1 Fab might bind to α_vβ₃-transfected 293 cells through α_vβ₅ as well as α_vβ₃. Nonetheless, the bulk of WOW-1 Fab binding to transfected cells appeared to be through α_vβ₃ because antibody binding was largely abolished by the Y178A substitution in α_v or the D119Y substitution in β₃. Integrin activation encompasses at least 2 events: (1) modulation of receptor affinity through conformational changes and (2) modulation of receptor avidity through facilitation of lateral diffusion and/or clustering of heterodimers. The binding of monovalent ligand WOW-1 Fab to α_vβ₃ likely reflects affinity modulation, whereas the binding of multivalent ligand, such as fibrinogen, likely reflects both affinity and avidity modulation.28 

Our data with WOW-1 Fab suggest that the Y178A substitution disturbs integrin affinity modulation rather than avidity modulation. In addition to ligand binding, cell adhesion to immobilized ligand can be strongly influenced by post–ligand-binding events. Indeed, in spite of the moderate inhibition of soluble fibrinogen binding by F177A as well as W179A substitution, these substitutions induced only a modest inhibition of cell adhesion to immobilized fibrinogen. One could argue the possibility that the Y178A substitution in α_v may disturb conformational changes from resting to activated states of integrin because of the failure of the induction of β₃LIBS with RGDW.48 However, this possibility is unlikely because the Y178A substitution in α_v completely abolished the interaction with immobilized fibrinogen, an activation-independent ligand. Thus, this residue is likely involved in direct contact with the ligand. It would be interesting to know whether the Y178A substitution may affect divalent cation binding. Recently, employing human-to-mouse chimeras, Puzon-McLaughlin et al49localized binding sites for ligand-mimetic murine mAbs against α_IIbβ₃ and demonstrated that the involvement of several discontinuous sites in both α_IIband β₃ is unique to ligand-mimetic antibodies. Involvement of certain residues in both α_v (Y178) and β₃ (D119) subunits in WOW-1 binding is consistent with their data, and these residues may participate in a ligand-binding pocket in α_vβ₃.

The α_vβ₃ is expressed in a number of cell types: endothelial cells, arterial smooth muscle cells, platelets, subpopulation of leukocytes, osteoclasts, and tumor cells. The α_vβ₃ is involved in cell adhesion, proliferation, and migration and has been shown to play a crucial role in tumor angiogenesis, intimal hyperplasia after arterial injury, wound healing, and osteoporosis in the adult organism. Recently, human clinical trials are in progress to evaluate the effects of the humanized anti-α_vβ₃ mAb in patients with late-stage cancer.50 The present results provide new information concerning the interaction between RGD ligands and the non–I-domain α_v subunit as well as key structural differences between α_vβ₃ and α_IIbβ₃ with regard to ligand binding. This information may facilitate the development of novel antagonists specific for α_vβ₃.

We thank Dr David Cheresh for mAbs LM142 and LM609 and the vector containing WT α_v cDNA; Dr Mark Ginsberg for mAbs anti-LIBS1 and anti-LIBS6; Dr Thomas Kunicki for a mAb AP5; Dr Peter Newman for a mAb AP3; Dr Gilbert White for the vector containing WT β₃ cDNA; Dr Martin Hemler for a mAb B5-IVF2; and Dr Jiro Seki for RGDW peptide.