Foot-and-Mouth Disease Virus Forms a Highly Stable, EDTA-Resistant Complex with Its Principal Receptor, Integrin αvβ6: Implications for Infectiousness

Abstract

The initial stage of foot-and-mouth disease virus (FMDV) infection is virus binding to cell surface integrins via the RGD motif in the GH loop of the VP1 capsid protein. As for all ligand/integrin interactions, the initial contact between FMDV and its integrin receptors is cation dependent and hence inhibited by EDTA. We have investigated this binding process with RGD-containing peptides derived from the VP1 capsid protein of FMDV and discovered that, upon binding, some of these peptides form highly stable, EDTA-resistant associations with integrin αvβ6. Peptides containing specific substitutions show that this stable binding is dependent on a helical structure immediately C terminal to the RGD and, specifically, two leucine residues at positions RGD +1 and RGD +4. These observations have a biological consequence, as we show further that stable, EDTA-resistant binding to αvβ6 is a property also exhibited by FMDV particles. Thus, the integrin-binding loop of FMDV appears to have evolved to form very stable complexes with the principal receptor of FMDV, integrin αvβ6. An ability to induce such stable complexes with its cellular receptor is likely to contribute significantly to the high infectiousness of FMDV.

Foot-and-mouth disease virus (FMDV) is the type species of the genus Aphthovirus within the family Picornaviridae and the etiological agent of foot-and-mouth disease, a severe vesicular condition affecting a large number of artiodactyls, including domesticated ruminants and pigs (1, 34). Presently, the virus is endemic in many parts of the world, including South America, Africa, and Asia (19). Foot-and-mouth disease is highly contagious and difficult to control as FMDV has a wide host range (see above) and a rapid replication cycle, small amounts of virus can initiate infection, and infected animals excrete high levels of virus. In addition, multiple modes of transmission have been recognized, including airborne spread, sometimes over long distances, including overseas (1, 10, 12, 31).

Field isolates of FMDV use integrins to initiate infection (14, 15, 29). The integrin family of cell adhesion receptors are a conserved series of αβ heterodimers, which bind in a divalent cation-dependent manner to ligands through recognition of short motifs that usually include one of the acidic residues glutamate (E) or aspartate (D) (13). Examples of such motifs include arginine-glycine-aspartate (RGD) or leucine-aspartate-valine (LDV), and short peptides containing these motifs can interact similarly with integrins (13). Recognition of RGD-containing proteins can proceed in a stepwise manner where the initial RGD binding is enhanced by a second stabilizing interaction involving so-called synergy sites on the ligand (2, 21, 23). The concept of a synergy site was first described for binding of α5β1 to fibronectin (Fn). Thus, high-affinity binding of Fn to α5β1 requires the RGD motif located on the 10th type III domain of Fn and a second synergy site in the 9th type III domain (23). Similarly, the large extracellular matrix protein vitronectin binds integrin αvβ3 in a stepwise manner; initial binding is RGD and cation dependent and reversible but can proceed to form an essentially nondissociable complex (30). In contrast, a 15-mer RGD peptide derived from vitronectin binds reversibly to this integrin (30). Formation of nondissociable complexes between integrins and RGD-containing ligands has also been observed for fibrinogen and several snake venom disintegrins. In general, these observations have been made with the integrins αvβ3 and αIIbβ3 (21, 25, 28, 30, 33).

FMDV contains a highly conserved RGD motif located on a surface-exposed and conformationally flexible loop (the GH loop of capsid protein VP1) and has been reported to infect cells via the RGD-binding integrins αvβ1, αvβ3, αvβ6, and αvβ8 (3, 14, 16, 18). Given the strong expression of αvβ6 in the epithelia targeted by FMDV, this integrin is widely believed to be the most relevant receptor in vivo (5, 27). The integrin αvβ6 has been shown to recognize the extended motif RGDLXXL (where X is any amino acid), which is highly conserved in FMDV (6, 8, 20, 24). Recently we have shown that the RGD +1 and +4 leucines are necessary to enhance the potency of FMDV-derived peptides as anti-αvβ6 antagonists in virus/integrin binding experiments (6). Crystallographic analyses of FMDV and FMDV-derived peptides in complex with anti-FMDV Fab fragments have shown that the GH loop of VP1 consists of a short region of β-strand followed by the RGD tripeptide in an open conformation immediately preceding a helix (35, 36). Recently we have shown that this structure is preserved by a VP1 GH loop peptide derived from type O FMDV (O-20mer) (Table 1) whereas an equivalent peptide derived from type C FMDV (C-20mer) did not form a stable helix (9). In this study the presence of the helix was linked to the inhibitory potential of the peptide, as the type O peptide was a more potent inhibitor than the helix-deficient type C peptide in αvβ6-mediated cell adhesion assays (9). The role for the helix was further supported by showing that a variant of the type O-20mer peptide containing just two d isomers of valine did not form a helix C terminal to the RGD and also performed poorly as an inhibitor of αvβ6-mediated binding events (9).

TABLE 1.

Peptide sequences^a

Virus	Peptide	Sequence
O1BFS	O-20mer	NAVPNLRGDLQVLAQKVART
	O-17mer	--VPNLRGDLQVLAQKVAR-
	RGE	--VPNLRGELQVLAQKVAR-
	AAD	--VPNLAADLQVLAQKVAR-
	L8A	--VPNLRGDAQVLAQKVAR-
	L11A	--VPNLRGDLQVAAQKVAR-
	L8-11A	--VPNLRGDAQVAAQKVAR-
	Q9A	--VPNLRGDLAVLAQKVAR-
	O-20merDV1217b	NAVPNLRGDLQVdLAQKVdART
C-S8c1	C-20mer	YTASARGDLAHLTTTHARHL
	C-17mer	-TASARGDLAHLTTTHAR--

^aPeptide sequences were derived from the VP1 GH loop of FMDV O1BFS (22) or C-S8c1 (24). O-20mer, O-17mer, C-20mer, and C-17mer are based on the wild-type loop sequence. Amino acid substitutions are shown in boldface type.

^bO-20merDV1217 is a variant of O-20mer containing the d isomers of valine at positions 12 and 17.

In this report we have investigated further the role of the helical structure and the non-RGD residues in binding of FMDV-derived peptides to αvβ6. Using resistance to EDTA-induced dissociation as an indicator of highly stable binding, we show that upon binding, peptides derived from the VP1 GH loop of type O FMDV form EDTA-stable associations with αvβ6 whereas equivalent peptides from type C FMDV fail to do so. Using peptides with specific amino acid substitutions, we have identified the helix and RGD +1 and RGD +4 leucines as key for stable binding. These observations have a biological corollary since we further show that the FMDV virions also form stable, EDTA-resistant complexes with cellular αvβ6. Formation of these highly stable complexes is likely to contribute to the unusually high infectivity of FMDV by increasing the proportion of attached virions and, therefore, the likelihood of virus internalization and infection.

MATERIALS AND METHODS

Viruses, cell lines, and antibodies.

Preparation of FMDV working stocks and virus purification were carried out as described previously (6). The generation of primary bovine thyroid (pBTY) cell cultures and the A375Pβ6 and SW480-αvβ6 cell lines and their respective integrin expression profiles have been described previously (6, 9, 18, 37). The anti-αvβ6 monoclonal antibodies (MAb) 6.8G6 and 71C5 were gifts from Paul Weinreb and Shelia Violette (Biogen Idec, Cambridge, MA) and have been described previously (38). The anti-αvβ6 MAb 10D5 was obtained from Chemicon International.

rsαvβ6 and peptides.

Recombinant soluble αvβ6 (rsαvβ6) (37) lacking the αv and β6 cytoplasmic domains was purified from CHOβ6 cells (a gift from D. Sheppard, University of California San Francisco) with an immunoaffinity column consisting of the anti-αv MAb L230 conjugated to agarose beads by using a Perbio Carbolink kit according to the manufacturer's instructions. High-performance liquid chromatography-purified peptides (Table 1) based on the VP1 GH loop of FMDV O1BFS (22) and C-S8c1 (24) were generated at the peptide synthesis facilities at Cancer Research UK or Institute for Animal Health Compton, United Kingdom, by using standard procedures. Some peptides were biotinylated on the N terminus.

Solid-phase binding assay.

Ninety-six-well plates (Immulon IB; ThermoLabSystems, MA) were coated overnight at 4°C with 0.8 μg/ml rsαvβ6 (100 μl/well) in Tris-buffered saline (TBS) (20 mM Tris, pH 7.5,150 mM NaCl) supplemented with divalent cations (1 mM CaCl₂ and 0.5 mM MgCl₂) (TBS-CaMg). Plates were washed three times with TBS-CaMg and nonspecific binding sites blocked with 200 μl/well TBS-CaMg containing 1% casein (block buffer). Biotinylated peptides (1 nM; 100 μl/well) were allowed to bind for the stated time in block buffer or in block buffer without divalent cations but supplemented with 20 mM EDTA. Wells were washed as described above, and integrin-bound biotinylated peptide was detected with 50 μl/well ExtrAvidin-horseradish peroxidase (Sigma-Aldrich), diluted 1:500 in block buffer. Signal was developed using a TMB+ system (DAKOcytomation Ltd., Ely, United Kingdom), and an Opsys MRX microplate reader (Dynex Technologies, Worthing, United Kingdom) fitted with Revelation Quicklink software (Dynex Technologies) was used to quantify the absorbance at 450 nm, with a reference wavelength of 570 nm.

Flow cytometry. (i) Assessment of αvβ6-specific peptide binding.

Cells were resuspended in TBS-CaMg supplemented with 0.1% bovine serum albumin (BSA), 0.1% sodium azide, and 10 μg/ml MAb 6.8G6 (38) or control mouse immunoglobulin G1 (IgG1) for 1 h at 4°C. Biotinylated peptides were then added for a further 0.5 h at 4°C. The cells were washed and cell-bound peptide detected by incubation with 10 μg/ml polyclonal rabbit antibiotin (Abcam, Cambridge, United Kingdom) for 0.5 h at 4°C. Note that a rabbit antibody was used in order to avoid cross-reactivity with the mouse antibody (6.8G6) used to block the αvβ6 ligand binding site. Cells were washed twice and incubated with 10 μg/ml R-phycoerythrin-conjugated goat anti-rabbit (Molecular Probes) for 0.5 h at 4°C before two further washes and fluorescence analysis (FL2-H) on an LSR-1 flow cytometer (Becton Dickinson) fitted with CellQuest software. For each experiment, 10,000 events were counted and cell debris/apoptotic cells gated out using a plot of forward scatter versus side scatter.

(ii) Analysis of cation-dependent peptide binding.

Cells in suspension were washed with TBS, 20 mM EDTA, 0.1% BSA, and 0.1% sodium azide and divided into three populations (A, B, and C). Population A was washed a further three times in the same buffer. Divalent cations were added to populations B and C by washing the cells three times with TBS-CaMg, 0.1% BSA, and 0.1% sodium azide. Anti-integrin antibody (10 μg/ml) or biotinylated peptide was added to population A in TBS, 20 mM EDTA, 0.1% BSA, and 0.1% sodium azide or to populations B and C in TBS-CaMg, 0.1% BSA, and 0.1% sodium azide at 4°C for 0.5 h.

Populations A and B were then washed three times in TBS, 20 mM EDTA, 0.1% BSA, and 0.1% sodium azide and population C in TBS-CaMg, 0.1% BSA, and 0.1% sodium azide, and the populations were maintained in these buffers for the remainder of the experiment. Biotinylated peptides were detected using sequential incubations (for 0.5 h at 4°C) with mouse antibiotin antibody (10 μg/ml; Jackson ImmunoResearch) and goat anti-mouse IgG conjugated to AlexaFluor-488 (10 μg/ml; Molecular Probes, Invitrogen, United Kingdom). Anti-integrin antibody was detected using only the goat anti-mouse IgG AlexaFluor-488-conjugated antibody. After two further washes in the appropriate buffer, fluorescence analysis (FL-1H) was carried out as described above.

(iii) Analysis of ligand binding to SW480-αvβ6 and pBTY cells.

Purified FMDV O1Kcad2 was used in all virus binding assays. Virus binding to SW480-αvβ6 and pBTY cells was determined using the anti-FMDV MAb D9 and a goat anti-mouse IgG conjugated to R-phycoerythrin (Southern Biotechnology Associates) by flow cytometry using a FACSCalibur instrument (Becton Dickinson), counting 10,000 cells per sample as previously described (6). Background fluorescence was determined by omitting either the virus or the primary anti-FMDV antibody from the assay, and both conditions gave nearly identical results. Unless stated, all virus/peptide binding steps were carried out in TBS-CaMg supplemented with 2% goat serum and 3% BSA.

In the experiment shown in Fig. 4, peptides were bound to the cells at 4°C for 0.5 h prior to addition of virus (10 μg/ml) for a further 0.5 h, and the peptides remained present throughout the virus binding step. At this point, the cells were washed and cell-bound virus detected as described above.

FIG. 4.

The O-20mer peptide inhibits FMDV binding to αvβ6. pBTY cells were treated with peptides (O-20mer, O-17mer, C-17mer, and L11A) (Table 1) for 0.5 h on ice prior to the addition of virus (10 μg/ml) for a further 0.5 h. The cells were washed, and cell-bound virus was detected by flow cytometry. (A) Representative histogram using the O-17mer as the competitor. The white histogram shows virus binding to the cells in the absence of peptide competition. The shaded histogram shows virus binding to the cells pretreated with the O-17mer at 1 μM. The black histogram shows the background mean fluorescence intensity for the assay (see Materials and Methods). Note that the shaded histogram and the black histogram virtually overlay. (B) Virus binding as the percentage of binding in the absence of peptide competition. Data are shown for the O-20mer (broken line, filled triangles), the O-17mer (broken line, filled squares), the C-17mer (solid line, open squares), and the L11A (solid line, open triangles) peptides. The means and standard deviations of triplicate samples from one of two identical experiments with similar results are shown. Note the O-17mer and the O-20mer peptides inhibit virus binding with nearly identical IC₅₀. Similarly, the L11A and C-17mer peptides inhibit virus binding with nearly identical IC₅₀.

In the experiment shown in Fig. 5, peptides were bound to the test cells for 0.5 h on ice, whereas control cells were treated in the absence of peptide. Excess unbound peptide was removed by washing with TBS-CaMg. The cells were then incubated with TBS containing 50 mM EDTA for 0.5 h on ice and then washed with TBS-CaMg to restore divalent cations. The cells were then incubated with FMDV (10 μg/ml) for 0.5 h on ice and cell-bound virus detected by flow cytometry as described above.

FIG. 5.

The type O FMDV 17-mer forms a stable, EDTA-resistant complex with αvβ6. Peptides (O-17mer, L8A, L11A, Q9A, and RGE [Table 1]; 1 mM) were bound to αvβ6 on pBTY and SW480-αvβ6 cells for 0.5 h on ice. Control cells were treated in the absence of peptide. Excess unbound peptide was removed by washing, and the cells were incubated in the presence of EDTA for 0.5 h on ice. The cells were then washed to restore divalent cations and incubated with FMDV (10 μg/ml) for a further 0.5 h on ice. Cell-bound virus was then detected by flow cytometry. Virus binding is shown as the percentage of binding to the control cells. The means and standard deviations of triplicate samples from one of two identical experiments with similar results are shown.

In the experiment shown in Fig. 6, biotinylated peptides were bound to cells for 0.5 h at 4°C in TBS-CaMg. Excess, unbound peptide was removed by washing with the same buffer. Control cells were analyzed for peptide binding at this stage. Test cells were incubated for a further 0.5 h on ice with either TBS-CaMg or TBS containing 50 mM EDTA and then washed with TBS-CaMg to restore divalent cations. Peptide that remained bound to the cells after these treatments was detected by flow cytometry using the mouse antibiotin antibody (as described above) and a goat anti-mouse IgG R-phycoerythrin conjugate (Southern Biotechnology Associates). Background fluorescence was determined by omitting the peptide from the assay.

FIG. 6.

Stable binding of the type O FMDV 17-mer to αvβ6 is dependent on the RGD +1 and RGD +4 leucine residues. Biotinylated peptide (1 μM) O-20mer, L8A, or L11A was bound to αvβ6 on SW480-αvβ6 and pBTY cells for 0.5 h on ice. Excess, unbound peptide was removed by washing. Control cells were analyzed at this stage for peptide binding by flow cytometry (black bars). Test cells were incubated in the presence of divalent cations (white bars) or EDTA (gray bars) for a further 0.5 h on ice. Peptide binding was then determined as described above and is shown as the percentage of binding to the control cells. The means and standard deviations of triplicate samples from one of two identical experiments with similar results are shown.

In the experiment shown in Fig. 7, FMDV (10 μg/ml) was bound to the cells for 0.5 h at 4°C in TBS-CaMg. Excess, unbound virus was removed by washing with the same buffer. Control cells were analyzed at this stage for virus binding. The test cells were then incubated with either TBS-CaMg or TBS containing 50 mM EDTA or with peptides in TBS-CaMg for a further 0.5 h at 4°C. The cells were then washed with TBS-CaMg to restore divalent cations and cell-bound virus detected by flow cytometry as described above.

FIG. 7.

EDTA-resistant binding of FMDV to cellular αvβ6. (A) FMDV (10 μg/ml) was bound to αvβ6 on SW480-αvβ6 or pBTY cells for 0.5 h at 4°C. Excess, unbound virus was removed by washing. Control cells were analyzed at this stage for virus binding by flow cytometry. Test cells were then incubated in the presence of either divalent cations (TBS-CaMg) or EDTA (EDTA) for a further 0.5 h at 4°C. The cells were then washed to restore divalent cations and cell-bound virus determined as described above. Virus binding is shown as the percentage of binding to the control cells. (B and C) FMDV (10 μg/ml) was bound to αvβ6 on SW480-αvβ6 or pBTY cells for 0.5 h at 4°C. Excess, unbound virus was removed by washing. Control cells were analyzed at this stage for virus binding by flow cytometry. Test cells were then incubated in the presence of peptides (Table 1) (open bars, 100 μM; filled bars, 10 μM) for a further 0.5 h at 4°C. The cells were then washed and cell-bound virus determined as described above. The means and standard deviations of triplicate samples from one of two identical experiments with similar results are shown.

Enzyme-linked immunospot assay.

In the experiment shown in Fig. 8, FMDV O1Kcad2 or C-S8c1 (multiplicity of infection [MOI] of ∼10) was bound to pBTY monolayers in 96-well plates for 0.5 h at 4°C. Excess, unbound virus was then removed by washing with Dulbecco's modification of Eagle's medium. At this stage, control cells were processed through the assay (see below). Test cells were incubated with either TBS-CaMg or TBS containing 50 mM EDTA for a further 0.5 h at 4°C. The cells were then washed with cell culture media and infection initiated by warming to 37°C. After 5 h, the cells were fixed with paraformaldehyde and the number of infected cells determined using MAb 2C2 (which recognizes the FMDV 3A protein, a marker for virus replication) in an enzyme-linked immunospot assay as described previously (4).

FIG. 8.

Effect of EDTA treatment on the infectivity of FMDV. FMDV O1Kcad2 or C-S8c1 was bound to pBTY monolayers (MOI of ∼10) in 96-well plates for 0.5 h at 4°C. Excess, unbound virus was removed by washing. Control cells were analyzed for infection at this stage (as described below). Test cells were incubated in the presence of either divalent cations (TBS-CaMg) or EDTA (EDTA) for a further 0.5 h at 4°C. The cells were then washed with cell culture media, and infection was initiated by warming to 37°C. After 5 h, the cells were fixed with paraformaldehyde and the number of infected cells was determined using an enzyme-linked immunospot assay. For the test cells, infection is shown as a percentage of infection relative to that for the control cells. The means and standard deviations of triplicate samples from one of two experiments with similar results are shown.

RESULTS

FMDV-derived peptides develop EDTA-stable binding to recombinant αvβ6.

Nondissociable binding of biotinylated 20-mer peptides, derived from the sequences of FMDV O1BFS (O-20mer) and C-S8c1 (C-20mer) (Table 1), was initially assessed using immobilized rsαvβ6 in a solid-phase binding assay. Figure 1 shows that in the presence of divalent cations both the O-20mer (Fig. 1A) and the C-20mer (Fig. 1B) peptide bound to rsαvβ6 in a time-dependent manner, with an observable increase in binding at 90 min compared with that at 30 min (Fig. 1A and B, bars 1 to 3). Note that the scale on the y axis indicates that the O-20mer bound much better than the C-20mer. The initial contact between FMDV and its integrin receptors is cation dependent and hence inhibited by EDTA; consistent with these observations, binding of the O-20mer and the C-20mer peptides to αvβ6 was abolished by coincubation with EDTA (Fig. 1A and B, bar 4), confirming that peptide binding is also cation dependent and likely to be RGD mediated. We also investigated the effect of delayed addition of EDTA on binding of the O-20mer and the C-20mer peptides. Peptides were allowed to bind for 30 min in the presence of divalent cations before the addition of EDTA for a further 30 or 60 min (Fig. 1A and B, bars 5 and 6). The delayed addition of EDTA did not dissociate existing peptide/integrin complexes but prevented further binding of the biotinylated peptide (Fig. 1A and B, compare bars 1 with bars 5 and 6). The same results were observed using a 1,000-fold excess of unbiotinylated homologous peptide as the competitor (data not shown). Thus, on integrin binding, the O-20mer and C-20mer peptides appear to form an EDTA-resistant complex with rsαvβ6.

FIG. 1.

FMDV-derived peptides develop EDTA-stable binding to recombinant αvβ6. Biotinylated peptide (1 nM) O-20mer, C-20mer, or O-20merDV1217 was bound to immobilized recombinant αvβ6. Bars 1 to 3 show peptide binding for 30, 60, and 90 min, respectively. Bar 4 shows peptide binding for 90 min in the presence of 20 mM EDTA. Bars 5 and 6 show peptide binding for 30 min prior to the addition of EDTA and binding continued for a further 30 min (bar 5) or 60 min (bar 6). Bar 7 shows data for a scrambled peptide. Bar 8 shows peptide binding in the absence of αvβ6. Data represent the means and standard deviations of quadruplicate wells and are representative of two independent assays with similar results.

Next, we investigated whether formation of the helix immediately following the RGD (see the introduction) is necessary for EDTA-resistant binding by using a variant of the O-20mer peptide in which valine-12 and valine-17 were substituted by the corresponding d isomers (O-20merDV1217) (Table 1). Previously we have shown that these substitutions abolish the helical structure and reduce the inhibitory potency of the O-20mer in αvβ6-mediated cell adhesion assays (9). Similarly to the C-20mer, the O-20merDV1217 peptide showed greatly reduced binding to rsαvβ6 compared with that of the O-20mer (Fig. 1A and C, bars 1 to 3). However, the introduction of the d isomers of valine at positions 12 and 17 did not alter the cation sensitivity of the peptide, as the initial binding still required divalent cations (Fig. 1C, bar 4) and existing peptide/integrin complexes were resistant to EDTA-induced disassociation (Fig. 1C, bars 5 and 6).

EDTA-resistant binding of FMDV-derived peptides to cellular αvβ6.

Recombinant truncated integrin adsorbed to plastic may not represent adequately the physiological situation; therefore, we sought to confirm the above-described results by using wild-type αvβ6 expressed on a cell surface. For these studies we used the A375Pβ6 cell line. A375Pβ6 cells express other RGD-binding integrins (α5β1, αvβ3, αvβ5, and αvβ8) in addition to αvβ6 (9). Therefore, we confirmed the αvβ6 specificity of the peptides by using the αvβ6-blocking antibody (MAb 6.8G6) (Fig. 2). Biotinylated peptides were bound in the presence or absence of MAb 6.8G6, and peptide binding was detected by flow cytometry. These experiments confirmed that the O-20mer (Fig. 2B), C-20mer (Fig. 2C), and O-20merDV1217 (Fig. 2D) peptides bind A375Pβ6 cells and that this binding is mediated by αvβ6, as peptide binding was inhibited by preincubation of the cells with MAb 6.8G6 but not by the control IgG. Note that in order to obtain comparable binding to A375Pβ6 cells, it was necessary to use a 1,000-fold higher concentration of C-20mer and O-20merDV1217 than of O-20mer.

FIG. 2.

αvβ6 specificity of the FMDV peptides. A375Pβ6 cells were preincubated with MAb 6.8G6 (mouse anti-αvβ6, gray histogram) or control mouse IgG (white histogram) before exposure to biotinylated peptide O-20mer (0.001 μM), C-20mer (10 μM), or O-20merDV1217 (10 μM) in the presence of divalent cations. Control cells (A) were not exposed to peptide. The cells were then washed and cell-bound peptide detected by flow cytometry. Black histograms, representing cells preincubated with IgG and not exposed to peptide, are shown in each panel as a reference. Data are representative of three independent experiments with similar results.

Next, we investigated the ability of the peptides to form EDTA-stable complexes with αvβ6 on A375Pβ6 cells. Figure 3 confirms that binding of O-20mer (used at 0.01 μM) (Fig. 3D), C-20mer (used at 10 μM) (Fig. 3E), and O-20merDV1217 (used at 10 μM) (Fig. 3F) to αvβ6 on A375Pβ6 cells is cation dependent, as peptide binding was detected in the presence of divalent cations but not on coincubation with EDTA. Peptide binding in the presence of divalent cations followed by subsequent washing of the cells extensively in EDTA is also shown (Fig. 3). Figure 3D shows that binding of O-20mer was unaffected by this treatment, whereas binding of C-20mer (Fig. 3E) was abolished. In contrast to C-20mer, binding of O-20merDV1217 to αvβ6 on A375Pβ6 cells was partially resistant to EDTA treatment (Fig. 3F). Binding of control antibody (MAb 10D5) (Fig. 3C), which is known to recognize αvβ6 in a cation-dependent manner (38), was abrogated by coaddition or postbinding addition of EDTA; however, the non-cation-dependent anti-αvβ6 (MAb 71C5) (Fig. 3B) bound under all conditions tested. Thus, EDTA-stable binding of O-20mer was observed for both cellular and rsαvβ6, whereas EDTA-stable binding of C-20mer occurred only with the recombinant integrin. Binding of the O-20merDV1217 peptide was stable to postattachment EDTA washing with rsαvβ6 but displayed a weak resistance to this treatment with cellular αvβ6. Thus, the time-dependent formation of EDTA-resistant complexes with αvβ6 is dependent primarily upon the amino acid sequence of the peptide but appears to be enhanced by a helical secondary structure C terminal to the RGD motif.

FIG. 3.

EDTA-stable binding of FMDV-derived peptides to cellular αvβ6. A375Pβ6 cells were exposed to control antibody, cation-independent anti-αvβ6 MAb 71C5, cation-dependent anti-αvβ6 MAb 10D5, or biotinylated peptide O-20mer (0.01 μM), C-20mer (10 μM), or O-20merDV1217 (10 μM), and cell-bound antibody/peptide was detected by flow cytometry. The red histogram shows binding in the presence of divalent cations. The blue histogram shows binding in the presence of EDTA. The green histogram shows initial binding in the presence of divalent cations followed by washing in EDTA and maintenance of the cells in EDTA for the remainder of the experiment. Data are representative of two experiments with similar results.

Inhibition of FMDV binding to cellular αvβ6 by virus-derived peptides.

The above-described studies have shown that the FMDV type O peptide (O-20mer) forms a stable, EDTA-resistant complex on binding cellular αvβ6, whereas an equivalent peptide derived from type C virus (C-20mer) does not. These results are consistent with our previous observations showing that C-20mer is a poor inhibitor of αvβ6-mediated cell adhesion compared with O-20mer (9). Previously, we have shown that a truncated version of O-20mer (O-17mer) (Table 1) is a potent inhibitor of FMDV binding to αvβ6 on pBTY and SW480-αvβ6 cells (6, 18, 37). On these cells, αvβ6 is expressed abundantly on the cell surfaces and functions as the major receptor for FMDV attachment and infection (6). Therefore, we compared the abilities of the O-20mer and O-17mer to inhibit αvβ6-mediated FMDV binding to pBTY cells. A version of the O-17mer containing a leucine (L)-to-alanine (A) substitution at the RGD +4 site (L11A) (Table 1) and a 17-mer version of the type C-derived peptide (C-17mer) (Table 1) were included for comparison. The L11A peptide has been shown previously to be a relatively poor inhibitor of FMDV binding to αvβ6 compared with the O-17mer (6). Figure 4 shows that the O-20mer and O-17mer peptides inhibit virus binding to αvβ6 on pBTY cells in a dose-dependent manner with nearly identical 50% inhibitory concentrations (IC₅₀), whereas the C-17mer was a relatively poor inhibitor, being equivalent to L11A. These data show that the full inhibitory potency of the type O peptides is contained within the 17-mer sequence and strongly suggest that the O-17mer would also form a stable complex with αvβ6. In addition, the relatively poor inhibitory potency of the C-17mer compared with those of the O-17mer and O-20mer peptides supports our conclusion that peptides derived from type O FMDV form more stable complexes with αvβ6 than corresponding peptides derived from the VP1 GH loop sequence of type C viruses.

The type O FMDV 17-mer forms a stable complex with αvβ6.

The above-described results show that the O-20mer and O-17mer inhibit FMDV binding to αvβ6 with nearly identical potencies. Next, we confirmed whether the O-17mer also forms a stable, EDTA-resistant complex with αvβ6. Similarly to the L11A peptide, we have shown that a version of the O-17mer containing an L-to-A substitution at the RGD +1 site (peptide L8A) (Table 1) is a poor inhibitor of αvβ6-mediated FMDV binding compared with the O-17mer, whereas alanine substitutions at any other non-RGD residue did not alter the inhibitory potency of the peptide (6). These observations suggest that the L8A and L11A peptides may not be able to form stable complexes with the integrin. Therefore, we also investigated the abilities of the L8A and L11A peptides to form stable, EDTA-resistant complexes with αvβ6. A version of the O-17mer containing an A substitution at position 9 (Q9A) (Table 1) was also included, as this peptide has been shown to be functionally equivalent to the O-17mer in αvβ6-mediated, virus-binding peptide competition experiments (6). For this study, we used both pBTY and SW480-αvβ6 cells. Peptides were added to the cells at 1 mM (Fig. 5), as at this concentration the L8A and L11A peptides have been shown to inhibit FMDV binding to αvβ6 on SW480-αvβ6 and pBTY cells by >95%, indicating that integrin saturation had been achieved (Fig. 4) (6). After peptide binding, the cells were washed extensively with EDTA. Virus binding was then used to indicate whether the peptide had been displaced from the integrin by the EDTA treatment. Figure 5 shows that a low level of virus binding was observed, compared with the level for the control cells (i.e., no peptide treatment), when the cells had been pretreated with the O-17mer, indicating that the peptide remained bound to the integrin and was not displaced by the EDTA treatment. Similar results were obtained using the Q9A peptide in place of the O-17mer, indicating that these two peptides are functionally equivalent. The inhibition was RGD dependent, as virus binding was not inhibited by pretreatment of the cells with the biologically inactive RGE version of the O-17mer (Table 1 and Fig. 5). Similarly to the RGE peptide, virus binding was not inhibited by pretreatment of the cells with the L8A and L11A peptides, indicating that these peptides had been displaced from the integrin. These experiments show that the O-17mer forms a stable, EDTA-resistant complex with αvβ6.

Stable binding of the O-17mer peptide to αvβ6 is dependent upon the leucine residues at the RGD +1 and RGD +4 sites.

To confirm the role of the RGD +1 and RGD +4 leucines in stable binding, we investigated binding of biotinylated versions of the L8A and L11A peptides to αvβ6 on SW480-αvβ6 and pBTY cells. The biotinylated version of O-20mer was included as a control for stable binding. Peptides were bound to the cells in the presence of divalent cations. The cells were subsequently washed to remove excess unbound peptide and incubated in the presence of either divalent cations or EDTA. Figure 6A shows that for pBTY cells the majority of O-20mer remained bound to αvβ6 following postbinding incubation with either divalent cations or EDTA, indicating that stable, EDTA-resistant complexes had been formed. In contrast, the majority of the L8A peptide (Fig. 6B) had dissociated from the integrin during the incubation step and slightly more peptide dissociation was observed for cells incubated with EDTA. In contrast to the L8A peptide, ∼60% of the L11A peptide (Fig. 6C) remained bound to αvβ6 on pBTY cells following postbinding incubations with either divalent cations or EDTA. Similar observations were made using SW480-αvβ6 cells (Fig. 6), with the exception that more dissociation of the L11A peptide was observed than with pBTY cells. These observations show that stable binding of the FMDV peptides to αvβ6 is dependent upon the leucine residues at the RGD +1 and RGD +4 sites and that the RGD +1 leucine appears to make a more significant contribution to the stability of the peptide/integrin complex. In addition, these data appear to show that the L8A and L11A peptides form slightly more stable complexes with bovine αvβ6 (expressed on pBTY cells) than with its human counterpart (expressed on SW480β6 cells).

EDTA-resistant binding of FMDV to cellular αvβ6.

The above-described studies have shown that peptides (O-20mer and O-17mer) derived from the VP1 GH loop of type O FMDV form stable, EDTA-resistant complexes with αvβ6. Next, we investigated whether this property was shared by the intact virus. For these studies, we used FMDV O1Kcad2, which has a VP1 GH loop sequence identical to that of the O-20mer. Virus was bound to αvβ6 on SW480-αvβ6 and pBTY cells and subsequently incubated in the presence of either divalent cations or EDTA (Fig. 7A) or with peptides in the presence of divalent cations (Fig. 7B and C). Virus that remained bound to the cells after these treatments was detected by flow cytometry. Figure 7A shows that, for both SW480-αvβ6 and pBTY cells, postbinding EDTA treatment did not dissociate virus from αvβ6, indicating that stable complexes had been formed. Figure 7B and C show that after virus had been bound treatment of the cells with the O-17mer (Fig. 7, RGD) or Q9A peptide displaced virus from αvβ6. This effect was RGD dependent, as control peptides lacking the RGD motif (RGE and AAD) (Table 1) did not displace virus from preexisting virus/αvβ6 complexes (Fig. 7B and C). Similarly, peptides containing an L-to-A substitution at the RGD +1 site (L8A and L8-11A) (Table 1) did not displace virus from the integrin. The L11A peptide containing a single L-to-A substitution at the RGD +4 site displayed an intermediate ability to displace virus from preformed virus/integrin complexes. These observations support the above-described conclusion that the RGD +1 leucine makes a more significant contribution than the RGD +4 leucine to the stability of the peptide/integrin complex.

In parallel experiments, FMDV (O1Kcad2 [Fig. 8A] or C-S8c1 [Fig. 8B]; MOI of ∼10) was bound to pBTY cell monolayers at 4°C in the absence or presence of EDTA. Cells where virus was bound in the absence of EDTA were washed and incubated in the presence of either divalent cations or EDTA (Fig. 8). After these treatments, the cells were washed and placed at 37°C to initiate infection. Five hours after infection had begun, the cells were fixed with paraformaldehyde and the number of infected cells quantified using an enzyme-linked immunospot assay. Consistent with the requirement for divalent cations for the initial virus binding to αvβ6, inclusion of EDTA in the virus binding step reduced infection by >98% (data not shown). Figure 8A shows that incubation of the cells with either divalent cations or EDTA after the virus (O1Kcad2) binding step did not reduce significantly the number of infected cells compared with the number of control cells that received no postbinding wash. These results are consistent with those shown in Fig. 7A, which shows that virus binding to cellular αvβ6 is not reduced by postbinding EDTA treatment, and confirm that FMDV type O virions form stable, EDTA-resistant complexes with αvβ6 and that cell-associated virus remained infectious after EDTA treatment. Figure 8B shows that the same results were obtained using FMDV C-S8c1, which exhibited a similar resistance to EDTA-induced suppression of infection of pBTY cells. These data appear to contrast with those using the C-20mer, which has its sequence derived from C-S8c1 and was removed from cellular αvβ6 by EDTA treatment (Fig. 3E) (see Discussion).

DISCUSSION

Here we report the discovery and characterization of a novel interaction between FMDV-derived peptides (O-20mer and O-17mer) and the FMDV receptor integrin αvβ6. The initial peptide/integrin interaction is RGD mediated and cation dependent and, hence, inhibited by EDTA. However, once formed, the peptide/integrin complex is rapidly stabilized and becomes resistant to EDTA-mediated dissociation. Our results show that stabilization of the complex is dependent on the leucine residues which occupy the RGD +1 and +4 positions and the helical structure immediately C terminal to the RGD. The key role played by the RGD +1 and +4 leucines in stabilizing the complex explains our previous observations that the inhibitory potency of FMDV-derived peptide antagonists of αvβ6 carrying amino acid substitutions at these sites is greatly reduced (6). Importantly, the helix region enhances peptide binding by several orders of magnitude, as shown by the fact that the helix-deficient variant peptide O-20merDV1217 (9) requires ∼1,000-fold-higher concentrations than parental O-20mer for peptide binding to be detected (Fig. 1 and 3). These data suggest that the LXXL motif synergizes with the RGD, generating a binding affinity much greater than would be expected by either interaction alone. Here we propose that the role of the helix is to maintain the functional synergism between the RGD and synergy site leucines. A role for the helix in the synergy interaction is supported by our data showing that the O-20merDV1217 peptide (which does not form a stable helix C terminal to the RGD) has a reduced ability to form stable complexes with αvβ6, despite this peptide retaining the RGD +1 and RGD +4 leucines (Fig. 1C and 2). The role of the helix in the synergy interaction is also supported by our data obtained using the C-20mer. Similarly to the O-20merDV1217, we have shown previously that the C-20mer does not form a stable helix C terminal to the RGD and is a poor inhibitor of αvβ6-mediated cell adhesion (9). In the current study we have shown (i) that the C-20mer does not form stable, EDTA-resistant complexes with cellular αvβ6 (Fig. 3) and (ii) that a type C 17-mer peptide (the C-17mer) is a poor inhibitor of αvβ6-mediated FMDV binding to αvβ6, being equivalent to the L8A peptide which lacks the RGD +1 leucine and hence a synergy site (Fig. 4). These observations can be explained by the inability of the C-S8c1-derived peptide to form a stable helix and therefore a functional synergy site.

Owing to their positions within the C-terminal helix, in three-dimensional space, the RGD +1 and +4 leucines are adjacent amino acids and form a hydrophobic patch on the outer face of the helix, where they are positioned favorably for integrin binding (6, 9). Studies using saturation transfer difference nuclear magnetic resonance have shown that it is highly likely that the RGD +1 and RGD +4 leucines bind physically to αvβ6, since on peptide binding these residues are the closest to the αvβ6 surface (9). Here we propose that binding of the LXXL leucines results in the switch to EDTA-resistant binding; however, we do not know if the postintegrin binding insensitivity to EDTA results from the stability of the complex being less dependent on the coordinating divalent cation that links the integrin to the ligand or whether it is buried upon complex formation.

Most FMDVs have an L at RGD +1; however, a number of isolates can have either methionine (M) or arginine (R) at this site. We have shown previously that FMDV-derived peptides containing an RGDM motif are functionally equivalent to peptides containing an RGDL motif to inhibit virus binding to αvβ6 in peptide competition studies (6). These same studies showed that RGDR-containing peptides have a reduced specificity for αvβ6. These observations suggest that M could substitute for L in the synergy site whereas an R at the RGD +1 position may in some way reduce the affinity of the synergy interaction. Further studies are required to determine whether this is the case.

The above-described conclusions are based on data using monomeric peptides. To investigate whether EDTA-stable binding is possible with FMD virions, we tested the stability of virus binding to αvβ6 on pBTY cells. These studies showed that FMDV O1Kcad2 can promote EDTA-resistant binding to cellular αvβ6. Interestingly, we made the same observation using FMDV C-S8c1 (from which the sequence of the C-20mer is derived) as this virus also appeared to form stable complexes with αvβ6 on pBTY cells. These results seem unexpected given that the C-20mer peptide cannot form a stable helix or bind in an EDTA-stable manner to αvβ6 on cells. However, it is likely that a helix is formed in the virus when the ends of the GH loop are constrained in the context of VP1. This explanation is plausible since a helical structure is stabilized in FMDV C-S8c1-derived peptides when in complex with anti-FMDV Fab fragments (35, 36). An alternative reason to explain the apparent EDTA resistance of the FMDV C-S8c1/αvβ6 complex on cell surfaces may be linked to the ability of single viruses to ligate multiple integrins simultaneously and thus promote integrin aggregation, which would be expected to activate the integrins and allow them to generate maximal integrin-dependent signaling and activity, as described previously (26). Since aggregation of integrins does enhance their activation state, this would enhance the stability of virus binding. This may also explain why the C-S8c1-derived C-20mer peptide could bind stably to recombinant αvβ6 but not cellular αvβ6. It is generally accepted that inactive integrins have a bent stereochemical form, with their ligand binding site adjacent to the cell membrane (32). Thus, the inability to bind ligand may in part be due to restricted access of ligands for the binding site. Recombinant integrins would have no such obstruction and are thus likely to be constitutively active. In addition, since the α-subunit cytoplasmic tail regulates the integrin activation state (7) its absence in rsαvβ6 reinforces the likelihood that the soluble integrin is constitutively active. Therefore, EDTA-stable binding of FMDV peptides that lack a fully functional synergy site may be possible if the αvβ6 integrins can be activated fully.

We have also tested the ability of FMDV to form EDTA-resistant complexes with αvβ3. Unfortunately we have not been able to detect virus binding to cellular αvβ3 even though we used a number of different αvβ3-expressing cell lines (6). Instead, for this investigation we used immobilized αvβ3 (Chemicon International) in a solid-phase assay. This assay has been used previously to demonstrate authentic RGD-mediated and cation-dependent binding of FMDV to αvβ3 (11, 17). Treatment with EDTA after virus binding displaced >90% of the virus from the integrin (data not shown), suggesting that the ability of FMDV to form stable complexes does not extend to integrin αvβ3.

In conclusion, these studies have shown that a helical structure C terminal to the RGD and two leucine residues of this helix (RGD +1 and +4 leucines) form a synergy site for RGD-mediated binding to αvβ6. Our results are consistent with a stepwise model of ligand binding in which the initial RGD-mediated interaction is rapidly stabilized and enhanced by further interactions involving the synergy site. Consequently, the virus/integrin complex becomes resistant to EDTA. This model explains the poor inhibitory effect on αvβ6-mediated virus binding and infection of FMDV-derived peptides that lack either an intact RGD motif or leucines at RGD +1 and +4 (6). Our observations that FMDV-derived peptides form stable complexes with the FMDV receptor αvβ6 have biological significance since we show that native virus particles also form these highly stable, EDTA-resistant complexes. Foot-and-mouth disease is one of the most contagious diseases of mammals. In part, this may be explained by the ability of the virus to be transmitted over long distances and the rapid dissemination of virus throughout an infected host (see the introduction). However, the current study confirms that the integrin-binding loop of FMDV is highly adapted for binding to αvβ6 and strongly supports the conclusion that the virus has evolved to use this integrin as its principle receptor in vivo (27). Formation of highly stable complexes with αvβ6 is likely to contribute significantly to the high infectivity of FMDV by increasing the proportion of attached virions and, therefore, the likelihood of virus internalization and infection. These data have increased our understanding of how ligands bind to αvβ6 and will greatly improve the design of specific antagonists for this integrin, some of which may eventually be useful in controlling FMDV infection.