Interactions of Foot-and-Mouth Disease Virus with Soluble Bovine α_Vβ₃ and α_Vβ₆ Integrins

Abstract

At least four members of the integrin family of receptors, α_Vβ₁, α_Vβ₃, α_Vβ₆, and α_Vβ₈, have been identified as receptors for foot-and-mouth disease virus (FMDV) in vitro. Our investigators have recently shown that the efficiency of receptor usage appears to be related to the viral serotype and may be influenced by structural differences on the viral surface (H. Duque and B. Baxt, J. Virol. 77:2500-2511, 2003). To further examine these differences, we generated soluble α_Vβ₃ and α_Vβ₆ integrins. cDNA plasmids encoding the individual complete integrin α_V, β₃, and β₆ subunits were used to amplify sequences encoding the subunits' signal peptide and ectodomain, resulting in subunits lacking transmembrane and cytoplasmic domains. COS-1 cells were transfected with plasmids encoding the soluble α_V subunit and either the soluble β₃ or β₆ subunit and labeled with [³⁵S]methionine-cysteine. Complete subunit heterodimeric integrins were secreted into the medium, as determined by radioimmunoprecipitation with specific monoclonal and polyclonal antibodies. For the examination of the integrins' biological activities, stable cell lines producing the soluble integrins were generated in HEK 293A cells. In the presence of divalent cations, soluble α_Vβ₆ bound to representatives of type A or O viruses, immobilized on plastic dishes, and significantly inhibited viral replication, as determined by plaque reduction assays. In contrast, soluble α_Vβ₃ was unable to bind to immobilized virus of either serotype; however, virus bound to the immobilized integrin, suggesting that FMDV binding to α_Vβ₃ is a low-affinity interaction. In addition, soluble α_Vβ₃ did not neutralize virus infectivity. Incubation of soluble α_Vβ₆ with labeled type A₁₂ or O₁ resulted in a significant inhibition of virus adsorption to BHK cells, while soluble α_Vβ₃ caused a low (20 to 30%), but consistent, inhibition of virus adsorption. Virus incubated with soluble α_Vβ₆ had a lower sedimentation rate than native virus on sucrose density gradients, but the particles retained all of their structural proteins and still contained bound integrin and, therefore, were not exhibiting characteristics of a picornavirus A particle.

Foot-and-mouth disease virus (FMDV), the etiologic agent of foot-and-mouth disease (FMD), is the type species of the Aphthovirus genus of the Picornaviridae. FMD is a devastating disease of livestock which results in severe economic consequences to affected countries (3, 85). The wide range of species which are affected by the disease, as well as differences in the severity of symptoms between species, has focused attention on host and viral factors involved in viral pathogenesis. In addition, the recent “reemergence” of the disease in developed countries, along with an inability to effectively control the disease in developing countries, highlights the need to explore new technologies involving both vaccines and antiviral agents.

Of the possible host factors involved in picornaviral pathogenesis, the viral receptor plays a major role in both host and tissue tropism (21, 74, 79). FMDV binds to cells through a conserved Arg-Gly-Asp (RGD) sequence located within a flexible external loop between the βG and βH strands (G-H loop) of viral protein 1 (VP1) (11, 23, 49, 57, 59). The RGD tripeptide is a recognition sequence for a number of members of the integrin family of cell surface receptors (77), and FMDV utilizes four members of the α_V subgroup of integrins (α_Vβ₁, α_Vβ₃, α_Vβ₆, and α_Vβ₈) as viral receptors in vitro (16, 20, 39, 41, 43, 64, 65). Integrins are heterodimeric type I membrane proteins, consisting of two subunits (α and β), which dimerize noncovalently at the cell surface (35-37). Of the 24 known integrins, only 8 (α_Vβ₁, α_Vβ₃, α_Vβ₅, α_Vβ₆, α_Vβ₈, α₅β₁, α₈β₁, and α_IIbβ₃) utilize the RGD recognition sequence to interact with their natural ligands (77).

Soluble viral receptors have played a pivotal role in studies of picornavirus early virus-cell interactions due to their ability to interact with virus directly in the absence of the cell membrane. Soluble poliovirus (PV) receptor was first shown to interact with and neutralize PV infectivity in vitro (45, 46). Since these early studies, the structures of receptor-bound PV (14, 28, 31), major and minor group human rhinoviruses (HRV) complexed with intercellular adhesion molecule 1 and the very low density lipoprotein receptor, respectively, have been solved (32, 47, 92). In addition the structures of virus-receptor complexes of swine vesicular disease virus with the coxsackievirus-adenovirus receptor (24), coxsackievirus B3 (CBV3) with the coxsackievirus-adenovirus receptor (29), and echovirus 7 with decay accelerating factor (30) have also been reported. Soluble picornavirus receptors have allowed detailed examination of receptor affinities among different groups of viruses (30, 48, 58, 93) and functional domains of viral receptors (13, 17, 52, 53, 69, 82) and also allowed analysis of picornaviral receptor-dependent conformational alterations that occur during internalization and uncoating (2, 26, 67, 83, 90).

In addition to FMDV, the parechoviruses echovirus 9 and coxsackievirus A9 (CAV9) utilize α_V integrin receptors to infect cells via exposed RGD sequences on the viral capsid (44, 66, 70, 75, 88, 89), and echoviruses 1 and 8 utilize a non-RGD-recognizing integrin (α₂β₁) to infect cells (15). However, even though soluble forms of the human homologs of α_Vβ₃ and α_Vβ₆ have been shown to possess binding activities for their natural ligands (61, 91), there have been no studies of the interactions of the α_V integrin-recognizing picornaviruses with soluble integrin heterodimers. It has been demonstrated that FMDV binds to isolated purified α_Vβ₃ and α₅β₁ integrins (38, 42); however, the latter integrin is not a viral receptor in vitro (65). The recently reported crystallographic three-dimensional structure of soluble human α_Vβ₃ (94, 95) suggests that structural studies of FMDV-integrin interactions are feasible. In addition, since the virus appears to be able to utilize at least four different α_V integrins (see above), it would be useful to examine the affinities of different serotypes and clinical isolates for each of the integrins to determine their roles in infection of susceptible hosts and to determine the role (if any) of these receptors in viral uncoating. In this work, we have generated soluble secreted forms of the bovine α_Vβ₃ and α_Vβ₆ integrins and examined their interactions with representatives of two FMDV serotypes. Surprisingly, we found that even though the viruses can utilize both receptors to mediate infection of cells in culture, they appear to interact very poorly with the soluble secreted α_Vβ₃ integrin, while the α_Vβ₆ integrin is able to bind to virus and neutralize infectivity.

MATERIALS AND METHODS

Viruses, cells, and antibodies.

FMDV type A₁₂ strain 119ab (A₁₂) was derived from the infectious cDNA clone pRMC₃₅ (73). A derivative of this virus with an NP sequence inserted in place of the GVRGDF sequence in the G-H loop (A12-RGD⁻), which can neither infect cells in culture nor cause disease in susceptible animals, has been described previously (59). FMDV type A₂₄ Cruziero (A₂₄) was recovered in vesicular fluid from a foot lesion of a steer experimentally inoculated with virus intradermally into the tongue. The virus was used directly and not passaged in tissue culture. Two type O₁ Campos (O₁Camp) variants, derived from infectious cDNAs containing capsid sequences represented in a vaccine seed stock from Brazil, have been described (78). These viruses are designated O₁C3056H (referred to as vCRM8 in reference 78), a highly cattle-virulent virus which is unable to bind to heparin-Sepharose (78) and utilizes integrin receptors to infect cells (20, 64, 65), and O₁C3056R (referred to as vCRM4 in reference 78), a cattle-avirulent virus which binds to heparin-Sepharose (78) and infects cells expressing heparan sulfate (HS) on the membrane in the absence of integrin receptors (65). A derivative of O₁C3056R with an RGD→KGE mutation in the G-H loop, O₁C3056R-KGE (referred to as vCRM48-KGE in reference 65) was derived by site-directed mutation and has been described elsewhere (65). This virus can also infect cells expressing HS in the absence of integrin receptors (65). A genetically engineered type A₁₂ virus chimera containing the G-H loop from type O₁BFS (A/O) has been described previously (71). COS-1 cells were maintained in Dulbecco's minimal essential medium (MEM) containing 10% fetal bovine serum and an additional 2 mM l-glutamine and 1 mM sodium pyruvate. Human embryonic kidney (HEK) 293A cells were maintained in Dulbecco's MEM containing 10% fetal bovine serum (defined; HyClone) with an additional 2 mM l-glutamine and 1 mM sodium pyruvate. BHK-21 cells were maintained in MEM containing 10% bovine sera and 10% tryptose phosphate broth. BHK3 cells are a derivative of BHK-21 cells which have been adapted to grow in either monolayer or suspension culture. They were obtained from the Empresa Colombiana de Productos Veterinarios in Bogotá, Colombia, and maintained in monolayer culture in the same medium as BHK-21 cells. A stable BHK3 cell line expressing the bovine α_Vβ₆ integrin (BHK3α_Vβ₆) was generated by cotransfecting cDNAs encoding the bovine α_V (64) and β₆ (20) subunits, using FuGENE6 (Roche Molecular Biochemicals) as described elsewhere (20), and selecting cells resistant to G418 and Zeocin (Invitrogen). Cells were cloned by single-cell cloning, and integrin expression was monitored by immunohistochemistry (63).

The monoclonal antibodies (MAbs) LM609 (MAB1976) and 23C6 (CBL544), which react with the α_Vβ₃ heterodimer (19, 34), CSβ6 (MAB2076), which reacts with the β₆ subunit (91), and 10D5 (MAB2077), which reacts with the α_Vβ₆ heterodimer (62), were purchased from Chemicon International Inc. MAbs LM609 and CSβ6 have been shown to cross-react with the bovine homologues of their respective integrins (20, 63, 64) (see Fig. 1, below). A rabbit antiserum prepared against purified human α_Vβ₃ was a gift from Merja Roivainen, National Public Health Institute, Helsinki, Finland.

FIG. 1.

Analysis of soluble integrins secreted from transfected cells. COS-1 cells were transfected with soluble integrin subunits, as denoted in the figure, and labeled with [³⁵S]methionine-cysteine between 24 and 72 h after transfection as described in Materials and Methods. The medium was collected, and the presence of soluble integrins or integrin subunits was determined by RIP, using the antibodies denoted in the figure followed by nonreducing SDS-PAGE and autoradiography. NT, nontransfected cells. The specificities of the antibodies were as follows: rabbit polyclonal anti-α_Vβ₃ antibody (anti-α_Vβ₃); monoclonal anti-α_Vβ₆ antibody (10D5); monoclonal anti-β₆ antibody (CSβ6); monoclonal anti-α_Vβ₃ antibodies (LM609 and 23C6).

Generation of plasmids encoding soluble bovine integrin subunits.

cDNA encoding the signal peptide and ectodomain of the bovine α_V integrin subunit was PCR amplified from the plasmid pBovαVZEO (64) using Pfu polymerase and primers 101 and 102 (Table 1), which insert a stop codon before the coding sequences for the transmembrane and cytoplasmic domains. The resulting amplicon was cloned into pCR-BluntII-TOPO (Invitrogen), and the cloning orientation was determined by sequencing. The amplicon was cut from the plasmid with XhoI and KpnI and cloned into pcDNA3.1(-) (Invitrogen) digested with the same enzymes and containing a neomycin selection marker to produce pBovssαVNEO.

TABLE 1.

Primers used in the cloning of the bovine integrin subunit ectodomains^a

Primer	Sequence (5′→3′)	Subunit	Orientation
101	GGGCTAGCATGGCTTTTCCGCCGCGG	αv	Forward
102	GGGGTACCTTATCAAGGAACAGGCATGGGTGC	αv	Reverse
103	AACTCGAGATGCGAGCGCGGCCGCGG	β3	Forward
104	GCTCTAGATTATCAGTCAGGGCCCTTGGGACAC	β3	Reverse
161	TCTAGACTCGAGCTGAAACGGATGGGGATT	β6	Forward
197	TCTAGATCTCTAATTTGGAGGTTTTGGACA	β6	Reverse

^aInitiating ATGs and stop codons are shown in bold. Restriction sites built into the primers and used during the cloning are underlined. They are Kpnl (primer 102) and Xhol (primers 103 and 161). Other restriction sites used in the cloning correspond to those found in the multiple cloning site of pCR-Bluntll-TOPO (Invitrogen).

An identical strategy was employed to clone the cDNAs encoding the signal peptides and ectodomains of the bovine β₃ subunit from the plasmid pBovβ3ZEO (64), using primers 103 and 104 (Table 1), and the bovine β₆ subunit from the plasmid pBovβ6 (20), using primers 161 and 197 (Table 1). The amplicon encoding the signal peptide and ectodomain of the β₃ subunit was cut from the plasmid with XhoI and PstI and inserted into pcDNA3.1/zeo(-) (Invitrogen) to generate pBovssβ3ZEO. Similarly, the amplicon encoding the signal peptide and ectodomain of the β₆ subunit was cut from the plasmid with XhoI and KpnI and inserted into pcDNA3.1/zeo(-) (Invitrogen) to generate pBovssβ6ZEO.

Production of soluble secreted bovine integrins.

To characterize soluble bovine integrins, the subunits were transiently expressed in COS-1 cells as described previously (20, 64). Twenty-four hours after transfection, the medium was exchanged for MEM containing 1/20 the normal amount of methionine and [³⁵S]methionine-cysteine (50 μCi). After an additional 48 h of incubation, the overlay medium was removed and analyzed for the production of soluble integrins by radioimmunoprecipitation (RIP) and nonreducing sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) using specific anti-integrin polyclonal antibodies and MAbs.

To produce large amounts of unlabeled soluble bovine integrins, stable cell lines were generated expressing either soluble α_Vβ₃ or soluble α_Vβ₆. Briefly, HEK 293A cells were cotransfected with pBOVssαVNEO and either pBOVssβ3ZEO or pBOVssβ6ZEO as described above. After selecting cells resistant to both G418 and Zeocin, colonies were isolated by single-cell cloning and analyzed for soluble integrin production by labeling with [³⁵S]methionine-cysteine and RIP as described above. Cell cultures producing the soluble integrins were seeded into roller bottles. After incubation overnight, the medium was exchanged for serum-free medium and the cells were incubated for an additional 48 h. The medium was removed, clarified by centrifugation, and concentrated by ultrafiltration through Centricon Plus-80 concentrators with a PL-30 membrane (Amicon).

Enzyme-linked immunosorbent assay (ELISA) quantitation of soluble bovine integrins.

Serial dilutions of concentrated culture medium containing soluble integrins were prepared in 0.05 M carbonate-bicarbonate buffer, pH 9.6 (coating buffer; carbonate-bicarbonate buffer capsules; Sigma Chemical Co.). A 100-μl aliquot of each dilution was added to plastic 96-well dishes (Immulon 2HB; Dynatech Laboratories Inc.) and incubated at 4°C overnight. The plates were washed with phosphate-buffered saline (PBS) containing 0.5% Tween 20 (PBS-Tween) and blocked with 100 μl of 1% nonfat dry milk in PBS containing 0.5 mM CaCl₂ and 0.9 mM MgCl₂ (blocking buffer) for 1 h at room temperature. Plates were washed with PBS-Tween, and 100 μl of primary antibody was added to each well. The primary antibody for soluble α_Vβ₃ was a rabbit anti-human α_Vβ₃ antiserum, diluted 1:7,000 in blocking buffer, and the primary antibody for soluble α_Vβ₆ was MAb CSβ6, diluted 1:1,000 in blocking buffer. The plates were incubated for 1 h at room temperature and washed with PBS-Tween, and 100 μl of either horseradish peroxidase (HRP)-conjugated donkey anti-rabbit immunoglobulin G (IgG; Pierce Chemical Co.) or HRP-conjugated goat anti-mouse IgG (Pierce Chemical Co.) (1:1,000 in blocking buffer) was added and incubated for 0.5 h at room temperature. After washing with PBS-Tween, substrate (tetramethyl benzidine; microwell peroxidase substrate system; KPL Chemicals) was added and incubated for 5 to 15 min at room temperature. The reaction was stopped with 1 M H₂SO₄, and the plates were read at 405 nm. Standard curves were generated using purified human soluble α_Vβ₆ (a gift from Dean Sheppard, University of California, San Francisco) or purified human α_Vβ₃ (Chemicon Chemical Co.).

Analysis of soluble integrin interaction with FMDV.

Immulon 2HB 96-well plates were incubated overnight at 4°C with a hyperimmune bovine anti-FMDV antiserum diluted 1:3,000 in coating buffer. The plates were washed with PBS-Tween and blocked with blocking buffer for 1 h at room temperature. Virus (1 μg), diluted in Tris-buffered saline (TBS; 25 mM Tris-HCl [pH 7.5], 150 mM NaCl) containing 0.1 mM CaCl₂, 1 mM MgCl₂, 1 mM MnCl₂, and 0.1% bovine serum albumin (BSA) (TBS+) was added and the plates were incubated overnight at 4°C. The plates were washed with PBS-Tween, soluble α_Vβ₃ or α_Vβ₆ (500 ng) diluted in TBS+ was added, and the plates were incubated for 1 h at room temperature. The presence of bound integrin was determined as described above.

An alternative assay, as described by Jackson et al. (42), was also performed. Briefly, soluble integrin (1 μg/ml) in TBS+ without BSA was bound to Immulon wells overnight at 4°C. After blocking the plates as described above, virus (1 μg) in TBS+ was added and incubated at room temperature for 1 h. Virus binding to the integrin was detected using specific antiviral MAbs.

Neutralization assays.

Virus and soluble integrins were diluted in TBS+ and incubated at 37°C for 1 h. The virus-integrin mixture (250 μl), containing approximately 40 to 60 PFU of virus, was inoculated onto BHK-21 cells in 35-mm tissue culture dishes and incubated for 1 h at 37°C. The inoculum was removed, and the cells were overlayed with 0.6% gum tracacanth in MEM. After 48 h of incubation, the cells were stained with crystal violet-HistoChoice (Amresco Co. Inc.) and plaques were counted. The results are given as the concentration of soluble integrin that reduced the number of plaques by 50% (IC₅₀).

Virus binding assays.

Types A₁₂ and O₁C3056H were grown in BHK-21 cells in the presence of [³⁵S]methionine and purified as described elsewhere (56). The number of virus particles per milliliter was determined spectrophotometrically (4). Labeled virus was diluted in TBS+ and incubated with soluble α_Vβ₃ or α_Vβ₆, diluted in the same buffer, at 37°C for 1 h. Binding assays were performed using BHK3α_Vβ₆ cells, diluted to 2.5 × 10⁶ cells/ml in TBS+, and virus at 3,000 particles/cell as previously described (12).

RESULTS

Analysis of soluble bovine α_Vβ₃ and α_Vβ₆.

COS-1 cells were transfected with the individual truncated α_V, β₃, or β₆ subunit cDNAs or cotransfected with the truncated α_V and either the truncated β₃ or β₆ subunit cDNAs. The cells were labeled between 24 and 72 h after transfection with [³⁵S]methionine-cysteine, and the culture medium was analyzed for the presence of secreted integrins by RIP as described in Materials and Methods. The results are shown in Fig. 1. A rabbit antiserum prepared against human α_Vβ₃ immunoprecipitated proteins of the expected molecular weights from the medium of cells transfected with the truncated α_V, β₃, and α_Vβ₃ subunit cDNAs but did not react with any proteins in the media of cells transfected with the β₆ subunit cDNA alone (Fig. 1a). This antiserum, however, immunoprecipitated both the α_V and β₆ subunits secreted from cells cotransfected with both subunits (Fig. 1a), suggesting that a complete soluble α_Vβ₆ integrin heterodimer was secreted into the medium. In support of this, we utilized the MAb CSβ6, which only reacts with the β₆ subunit (62). This MAb immunoprecipitated a protein of the expected molecular weight in the medium of cells transfected with the β₆ subunit cDNA alone, but it did not react with the truncated α_V subunit (Fig. 1a). The identity of the high-molecular-weight protein immunoprecipitated by this MAb from cells transfected with the β₆ subunit cDNA (Fig. 1a and c) is unknown, but it might be either a highly glycosylated form of the subunit or a dimer. CSβ6, however, immunoprecipitated both the α_V and β₆ subunits secreted from cells cotransfected with both subunit cDNAs, confirming the results with the polyclonal antibody, that the soluble α_Vβ₆ is a heterodimer. The high-molecular-weight protein present in the β₆-transfected cells was not detected in cotransfected cells. The MAb 10D5, which reacts with the complete human α_Vβ₆ integrin (62) and inhibits FMDV infection mediated by the human α_Vβ₆ integrin in vitro (43), did not react with the soluble bovine α_Vβ₆ integrin (Fig. 1a).

MAbs LM609 (19) and 23C6 (34), which react only with the complete integrin heterodimer, were used to analyze the soluble α_Vβ₃ integrin. Both antibodies immunoprecipitated the α_V and β₃ subunits only from the medium of cotransfected cells (Fig. 1b), indicating that α_Vβ₃ is also secreted as an intact heterodimer. The CSβ6 antibody only reacted with secreted β₆ or α_Vβ₆ subunits (Fig. 1a and c) but did not react with either the soluble α_V or α_Vβ₃ (Fig. 1c). Similarly, the 23C6 antibody did not react with soluble α_Vβ₆ (Fig. 1c).

Interactions of soluble bovine integrins with FMDV.

The ability of the soluble integrins to bind to FMDV was analyzed in a capture ELISA-type format, as described in Materials and Methods, and compared with conditioned media from nontransfected HEK 293A cells. We examined integrin binding to type A₁₂ virus, a type A₁₂ virus with RGD deleted (A12-RGD⁻) (59), two variants of type O₁Camp (78), and a type A₁₂ virus containing the G-H loop from type O₁BFS (A/O) (71) (Fig. 2). Soluble α_Vβ₆ bound strongly to types A₁₂, O₁Camp, and A/O (Fig. 2a). The binding was dependent on divalent cations, as it was inhibited by EDTA (data not shown), and also dependent on the RGD sequence, since there was no binding to either a type A₁₂ virus with RGD deleted or an O₁Camp with an RGD→KGE mutation (Fig. 2a). The integrin was able to bind to virus containing the RGD sequence regardless of whether the virus utilized the integrin to mediate infection, as evidenced by the fact that there was no difference in integrin binding between the bovine-virulent O₁Camp variant (O₁C3056H) that required integrin expression to infect cells (65, 78) and the bovine-avirulent variant (O1C3056R) that only utilized membrane-expressed HS to bind and enter cells (65, 78).

FIG. 2.

Analysis of binding of soluble α_Vβ₃ and α_Vβ₆ to FMDV. (a and b) Virus (1 μg) was captured on antibody-coated plastic wells as described in Materials and Methods. Culture medium from nontransfected HEK 293A cells (grey bars) or from transfected cells containing either soluble α_Vβ₆ (a) or α_Vβ₃ (b) (0.5 μg/well; black bars) was incubated with the virus, and binding of integrin was determined using either rabbit anti-α_Vβ₃ (a) or MAb CSβ6 (b) as described in Materials and Methods. (c) Wells were coated with either medium from nontransfected cells (grey bars) or from transfected cells containing soluble α_Vβ₃ (1 μg/ml; black bars) as described in Materials and Methods. Virus (1 μg) was incubated with the integrin, and the binding of virus was determined using an antiviral MAb.

In contrast, soluble α_Vβ₃ surprisingly exhibited little or no binding to any of the viruses used (Fig. 2b), even though the soluble integrin was able to bind to its natural ligand, vitronectin, in a similar assay (results not shown). It has been previously shown by Jackson and colleagues (42) that FMDV was able to interact with purified human α_Vβ₃ containing both the transmembrane and cytoplasmic domains when the integrin was immobilized first and virus was then allowed to bind to it. In the assay shown in Fig. 2b, the soluble integrin was bound to immobilized virus (see Materials and Methods). To examine whether the difference in our results might be due to the assay system, we performed the assay as described by Jackson et al. (42). The results (for type A only), shown in Fig. 2c, indicated that virus can interact with the soluble α_Vβ₃ and that the type of assay, and not the absence of the transmembrane and cytoplasmic domains, seems to account for the differences. A similar result was found for the type O virus (data not shown).

Effect of soluble bovine α_Vβ₃ and α_Vβ₆ on FMDV infectivity.

To determine whether the binding of soluble integrins to virus affects the ability of FMDV to infect cells, we performed plaque reduction neutralization assays as described in Materials and Methods. The results are shown in Table 2 and are represented as the integrin concentration required to inhibit plaque formation by 50%. Soluble α_Vβ₃ was unable to inhibit infectivity of any of the viruses used, a result consistent with its inability to interact with the viruses (Fig. 2b). In contrast, soluble α_Vβ₆ neutralized both type A and O virus infectivity. The IC₅₀ values for types A₂₄, A/O, and O₁C3056H were similar, between 11 and 20 ng/ml; however, about 100-fold more soluble integrin was required to neutralize types A₁₂ and O₁C3056R (Table 2). The specificity of the integrin inhibition was shown by the inability of the soluble α_Vβ₆ to neutralize the O₁C3056R-KGE virus, which can only grow in cells expressing HS but cannot utilize an integrin receptor (65) (Table 2).

TABLE 2.

Neutralization of FMDV infectivity by soluble bovine integrins^a

Virus	IC50 (ng/ml)b
	αvβ3	αvβ6
A12	>5,500	1,325
A24	>5,500	14
A/O	>5,500	11
O1C3056H	>5,500	20
O1C3056R	>5,500	2,600
O1C3056R-KGE	>5,500	>15,500

^aNeutralization was determined by plaque reduction as described in Materials and Methods.

^bThe concentration of integrin which reduced the number of plaques by 50%.

To examine the mechanism of neutralization by the soluble integrins, we examined the binding of radiolabeled types A₁₂ and O₁C3056H to BHK3α_Vβ₆ cells. We utilized this cell line because binding of virus was much improved over that observed in the wild-type BHK3 and BHK-21 cells (data not shown), allowing us to compare virus adsorption under different conditions. The results in Fig. 3 show that the adsorption of both type A₁₂ (Fig. 3a) and O₁C3056H (Fig. 3b) were markedly inhibited by the soluble α_Vβ₆ integrin in a concentration-dependent manner. Surprisingly, we also found that soluble α_Vβ₃ caused a slight, but repeatable, inhibition of binding of both viruses to cells (Fig. 3c and d). When we tried to interfere with the α_Vβ₃ inhibition using the function-blocking MAb LM609, we found that an antibody concentration of 10 μg/ml had no effect on the integrin-induced inhibition, but the antibody itself inhibited binding at higher concentrations (data not shown), probably as a result of interaction with hamster integrins.

FIG. 3.

Effect of soluble integrins on the adsorption of FMDV to cells. FMDV types A₁₂ and O₁C3056H, labeled with [³⁵S]methionine and purified as described in Materials and Methods, were incubated either with soluble α_Vβ₃ or α_Vβ₆ at the concentrations indicated in the figure or with medium from nontransfected HEK 293A cells. Binding assays were performed in BHK3α_Vβ₆ cells at 4°C, using a virus concentration of 3,000 particles/cell, as described in Materials and Methods. (a and b) Soluble α_Vβ₆ incubated with either type A₁₂ (a) or O₁C3056H (b). (c and d) Soluble α_Vβ₃ incubated with either type A₁₂ (c) or O₁C3056H (d).

Analysis of the virus-integrin complex.

The labeled viruses, incubated with soluble α_Vβ₆ and used in the experiment shown in Fig. 3, were analyzed on sucrose density gradients. Both the labeled type A₁₂ (Fig. 4a) and O₁C3056H (Fig. 4b) migrated slower in the gradient than virus incubated with the medium from nontransfected HEK 293A cells. A similar experiment, using viruses incubated with soluble α_Vβ₃, revealed no difference in viral migration in the gradients (data not shown). The viruses in the peak fractions of the gradients, shown in Fig. 4, were analyzed by RIP and PAGE. In both cases, the virus- soluble α_Vβ₆ complexes could not be immunoprecipitated with an anti-FMDV polyclonal serum but were immunoprecipitated with the anti-β₆ antibody, CSβ₆ (data not shown), indicating that the virus still contained bound integrin. In addition, both viruses contained the complete complement of viral proteins, VP1 to -4 (data not shown).

FIG. 4.

Analysis of the virus-soluble α_Vβ₆ complex. The samples used for these binding assays were centrifuged in an SW41 rotor through a 10-to-50% (wt/vol) sucrose gradient in TBS+ at 17,000 rpm for 18 h at 4°C. The gradients were fractionated into 0.4-ml fractions, and 0.1-ml samples were counted. (a) [³⁵S]methionine-labeled type A₁₂ incubated with either medium from nontransfected cells (•) or soluble α_Vβ₆ (6 μg/ml) (○). (b) [³⁵S]methionine-labeled type O₁C3056H incubated with either medium from nontransfected cells (•) or soluble α_Vβ₆ (6 μg/ml) (○). Samples were run on parallel gradients but are represented on the same graph for clarity.

DISCUSSION

FMDV is unique among the Picornaviridae in that it can utilize four different integrin receptors to mediate infection in cell culture (16, 20, 39, 41, 43, 64, 65), yet the roles these individual receptors play in the pathogenesis of the disease are not known. In addition, as a result of tissue culture adaptation, the virus acquires the ability to use HS as a receptor (5-7, 40, 55, 65), although this trait is associated with loss of virulence in susceptible animals (78). Finally, there is evidence for a third, as-yet-unidentified receptor which may be associated with virulence (6, 7, 86, 87, 96). These facts, coupled with a recent study showing that different virus serotypes appear to utilize the integrin receptors with varying efficiencies (20), have complicated studies on early virus-host interactions and viral pathogenesis. Recent outbreaks in developed countries have also highlighted the need to develop antiviral technologies which target specific host or viral factors involved in disease.

In this study, we have generated soluble heterodimers of two of the four known FMDV receptors, α_Vβ₃ and α_Vβ₆, and analyzed their interactions with representatives of serotypes A and O. Analysis of the integrins secreted by transient expression of subunit cDNAs in COS-1 cells showed that, in both cases, the subunits were secreted as complete integrin heterodimers (Fig. 1). Using a MAb that only reacts with the β₆ subunit (91), proteins of the expected size of the soluble α_V and β₆ subunits were immunoprecipitated from the medium of transfected cells (Fig. 1a). Similarly, a polyclonal antibody generated against human α_Vβ₃ reacted with the α_V, but not the β₆ subunit, yet immunoprecipitated both the α_V and β₆ subunits from cotransfected cells (Fig. 1a). We also utilized a MAb which reacts with the complete α_Vβ₆ heterodimer (62) and has been shown to inhibit FMDV infection mediated by the human α_Vβ₆ integrin (43). Interestingly, the antibody did not react with the soluble bovine α_Vβ₆ integrin (Fig. 1a) and was also unable to prevent viral infection mediated by the bovine α_Vβ₆ homolog (data not shown). The medium of cells cotransfected with the soluble α_V and β₃ subunits also contains the integrin as a complete heterodimer, as evidenced by immunoprecipitation of both subunits with two MAbs which only react with the intact integrin (19, 34) (Fig. 1b). The anti-β₆ MAb did not react with the α_Vβ₃ heterodimer, and the anti-α_Vβ₃ MAb did not react with the α_Vβ₆ heterodimer (Fig. 1c), indicating that the correct specific integrins are being synthesized.

To examine the ability of representative viruses to bind to soluble integrins, we utilized a capture ELISA-type assay, using the soluble integrins to bind to immobilized virus on plastic dishes (Fig. 2). Soluble α_Vβ₆ bound specifically to each virus which contained an RGD sequence in the G-H loop (Fig. 2b). The integrin was not able to bind to a type A₁₂ virus with an RGD deletion in the loop or to a type O virus with an RGD→KGE mutation. The latter virus is still capable of replicating in non-integrin-expressing cells, since it has an HS binding site in VP3 (65). In contrast, soluble α_Vβ₃ integrin did not bind to any of the viruses we tested in this assay. This result was surprising, since the A₁₂, A/O, and O₁C3056H viruses have been shown to be able to utilize this integrin to mediate infection (20, 64, 65). The O₁C3056R virus does not require integrins to infect cells (65); however, the presence of the RGD recognition sequence in the G-H loop should have allowed recognition of the α_Vβ₃ integrin, just as it was recognized by the α_Vβ₆ integrin (Fig. 2b).

Since the results with soluble α_Vβ₃ contrast with those reported by Jackson and colleagues (42), who analyzed the binding of each of the seven FMDV serotypes to purified human α_Vβ₃ immobilized on plastic, we performed the assay in a similar format as theirs. The results in Fig. 2c show that virus can interact with the soluble α_Vβ₃ in an RGD-dependent manner. The binding of integrins in solution to immobilized ligands is a function of affinity (84), while the binding of ligands to immobilized integrins is a measure of avidity (80). Thus, these results suggest that α_Vβ₆ can act as a high-affinity receptor for FMDV, while α_Vβ₃ can function as a viral receptor but at much lower affinity. It is interesting to speculate that the role of the high-affinity α_Vβ₆ may be to capture virus and help in the initial replication within the susceptible animal, while the low-affinity α_Vβ₃ plays a role in disseminating the virus within the animal to distant sites of replication.

The interaction of soluble α_Vβ₆ with FMDV results in neutralization of viral infectivity (Table 2). The results clearly show that the integrin is responsible for the neutralization, as the type O virus with KGE in place of the RGD in the G-H loop, which can still replicate by binding to the HS receptor (O₁C3056R-KGE), is not neutralized by the integrin, while the same virus with the RGD sequence in the loop (O₁C3056R) is neutralized by the integrin. The integrin IC₅₀ for this virus, however, is about 100-fold greater than that for the same virus without the HS binding site, which requires that the integrin mediate infection (O₁C3056H). Thus, neutralization of the HS binding virus is probably the result of steric hindrance of the HS binding site by the bound integrin. The finding that the integrin IC₅₀ was about 100-fold greater for the A₁₂ virus than for either the type A₂₄ or A/O virus was surprising and, at this point, we are not sure why this virus should be that much less sensitive to integrin neutralization. One possible explanation is that this virus has been extensively passaged in BHK-21 cells and may have become more adapted to hamster integrins than the other viruses used in this study. We have shown that the type A viruses, in contrast to the type O viruses, utilize the α_Vβ₃ integrin with an equal or greater efficiency than α_Vβ₆, and exchanging the type A₁₂ G-H loop with a type O loop did not affect the efficiency of integrin utilization (20). Yet, the A/O virus was as sensitive to integrin-mediated neutralization as type A₂₄ and more sensitive than the type A₁₂ parent. As expected, the soluble α_Vβ₃ integrin was not able to neutralize any of the viruses tested (Table 2).

The mechanism of soluble α_Vβ₆-mediated neutralization appears to be by inhibiting the binding of virus to cells (Fig. 3a and b) via soluble integrin binding to the receptor binding site on the virus (57, 72). The inhibition of binding of the type A₁₂ (Fig. 3a) or the type O₁C3056H (Fig. 3b) virus by the soluble integrin was about the same, even though the integrin IC₅₀ was quite different for each virus (Table 2). We also found that the soluble α_Vβ₃ integrin caused a low-level inhibition of adsorption of both viruses (Fig. 3c and d). This inhibition appeared to be concentration dependent and contrasted with the results in the virus neutralization assays (Table 2). Thus, it is possible that the low-affinity interaction of the integrin with virus in solution is manifest at the level of virus adsorption.

Interaction of PV with soluble PV receptor results in a conformational alteration of the capsid, generating a particle with a lower sedimentation rate in sucrose gradients and the loss of VP4, called an A particle (2, 26, 90). This particle is identical to the one generated during the natural infection of cells by PV (22, 51). The interaction of soluble intercellular adhesion molecule 1 with major group HRVs 3 and 14 results in quick release of the RNA from the capsid (27, 33, 92, 93), while the interaction of another major group HRV (HRV 16) is more stable and the conformational alterations occur more slowly (47, 68). In contrast, the interaction of the soluble low-density lipoprotein receptor with minor group HRVs does not result in virion conformational change and neutralizes virus infectivity by particle aggregation (54). The main difference between the interactions of PV and major group HRVs with their receptors as opposed to the minor group HRV-receptor interactions is that the receptors of the former group bind into a surface canyon on the virion (14, 28, 47, 92, 93), while receptors of the latter group bind around the fivefold axis of the particle and not in a canyon (32). Like the minor group HRV, integrin receptors interact with FMDV on the surface via the G-H loop (1, 50, 57), and it has been demonstrated that uncoating of FMDV in infected cells proceeds as a single step in acidic endosomes, resulting in the generation of pentameric subunits and RNA, without any apparent intermediates (8-10, 18). The results of this work support these conclusions, since interaction of virus with soluble α_Vβ₆ did not result in the generation of any conformational changes to the virion (Fig. 4). Analysis of the soluble α_Vβ₆-virus complex showed that the virions still retained bound receptor and also had a full complement of viral proteins. The slower sedimentation rate of virus which has been incubated with soluble α_Vβ₆ is probably the result of the particles still retaining bound receptor and not the result of generation of an A particle or other structural intermediate.

A number of antiviral strategies for picornaviruses which target viral attachment and entry events have been developed (see references 60 and 76 and references therein). In addition, the importance of integrins in a number of human pathologies has resulted in the synthesis of small-molecule integrin antagonists and mimetics (25, 81). The results in this work show that soluble receptors can interfere with FMDV infection in vitro, and they provide a basis for further work to analyze both soluble receptors and anti-integrin compounds for use in FMD control strategies.