Systematic replacement of amino acid residues within an Arg-Gly-Asp-containing loop of foot-and-mouth disease virus and effect on cell recognition.

The conserved Arg-Gly-Asp (RGD) motif found in a hypervariable, mobile antigenic loop of foot-and-mouth disease virus (FMDV) is critically involved in virus attachment to cells by binding to an integrin, probably related to alphavbeta3. Here we describe (i) the synthesis of 241 15-mer peptides, which represent this loop of FMDV (isolate C-S8c1) and single variants in which each amino acid residue was replaced by 16 others and (ii) the inhibitory activity of these peptides on the ability of FMDV C-S8c1 to recognize and infect susceptible cells. This approach has allowed a first detailed evaluation of the specificity of each residue within a RGD-containing protein loop on cell recognition. The results indicate that, in addition to the exquisitely specific RGD triplet, two highly conserved Leu residues located at positions +1 and +4 downstream of the RGD and, to a lesser extent, the residue at position +2 are the only critical and specific determinants within the loop in promoting cell recognition of a viral ligand. The results support the proposal that, in spite of their involvement in antibody recognition, RGD and other FMDV loop residues are remarkably conserved because of their essential role in cell recognition.

FMDV, a highly variable RNA virus of the Picornaviridae family, causes one of the economically most important diseases of farm animals (reviews in Refs. 30 and 31). FMDV is able to infect cells via a conserved RGD motif, which is recognized by one or several integrins probably related to ␣ v ␤ 3 , the vitronectin receptor (15). This RGD triplet is also involved in antibody recognition of FMDV by being a critical part of a major, hypervariable antigenic site (site A) (32,33). Site A is located in a long, solvent-exposed, and flexible loop (the G-H loop of capsid protein VP1), which could not be positioned in the structures determined for native virions (34,35). A structure for the G-H loop was, however, elucidated upon chemical reduction of a serotype O virion (36). A similar structure has been determined for the G-H loop of a serotype C (isolate C-S8c1) virus, as reproduced by a 15-mer peptide (A15) complexed with the Fab fragment of an antivirus neutralizing antibody (33) (see Fig. 1). In both structures, the RGD motif adopts a very similar conformation, which resembles those found in other integrin ligands. Short RGD-containing peptides inhibited weakly (IC 50 in the millimolar range) the infectivity and cell attachment of FMDV (12,13,16). Remarkably, long peptides (16 -19 amino acids), which reproduced the G-H loop of serotype C viruses, inhibited very effectively FMDV infectivity (IC 50 about 0.8 M) and attachment to cells (16). The relevance of the RGD motif in cell recognition of FMDV has been confirmed using site-directed mutagenesis (14).
In addition to the RGD motif, some flanking residues play a role, either directly or most probably through modulation of the conformation of the RGD triplet, in the specific recognition of some ligands by different integrins (1,2,18,37,38). Recently, Rieder et al. (39) constructed four site-directed mutants of FMDV of serotype A, which differed in one or several VP1 residues located close to the C terminus of the RGD triplet. Some of these mutants showed differences in attachment to cells and infectivity. However, systematic amino acid replacement studies, aimed at establishing the specificity of each residue around the RGD motif on cell recognition of FMDV or any other RGD-containing ligand, are lacking. The high efficiency of type C long peptides in inhibiting cell recognition of FMDV renders them useful reagents for such studies. In this report we describe (i) the synthesis of 241 15-mer peptides, which reproduce site A of FMDV C-S8c1 and single variants of this sequence, and (ii) the systematic evaluation of the inhibitory effect of each variant peptide on FMDV infectivity. The results indicate that, in addition to the RGD triplet, two highly conserved Leu residues located at positions ϩ1 and ϩ4 downstream of the RGD motif and the residue at position ϩ2 are the only critical and specific determinants within site A in promoting cell recognition of a type C FMDV.

EXPERIMENTAL PROCEDURES
Virus and Cell Lines-FMDV C-S8c1 is a plaque-purified virus clone derived from C-S8 (C 1 Santa Pau-Sp/70), a representative isolate of serotype C viruses from Europe (40). The BHK-21 cell line was obtained from the ATCC.
Peptide Synthesis-A group of 240 analogs of the reference peptide A15 (YTASARGDLAHLTTT) was prepared by systematic single residue replacements at every position with 16 of the genetically coded amino acids (all except Cys, Met, and Trp (37)). The peptides were synthesized as C-terminal carboxamides by simultaneous solid phase procedures on PAL-MBHA resins (41) using Fmoc (N-(9-fluorenyl)methoxycarbonyl) chemistry in an Abimed AMS422 instrument. Peptide resins were deprotected and cleaved by treatment with trifluoroacetic acid/triethylsilane/water (95:2.5:2.5, 2 h, room temperature). The peptides were isolated by precipitation with cold diisopropyl ether and centrifugation and then taken up in 0.1 M acetic acid and lyophilized. The peptides used showed contents Ͼ80% in the target sequence, as judged by analytical high pressure liquid chromatography. The identity of the synthetic peptides was further confirmed by amino acid analysis and electrospray mass spectrometry, which were consistent with the expected compositions. Stock solutions of peptides in phosphate-buffered saline (PBS, complete) and soluble peptide concentration were rechecked by amino acid analysis. The pH of the stock solutions was also checked and, if necessary, adjusted to neutrality.
Cyclization of FMDV Peptides-The head-to-tail cyclic version of the reference peptide, cyclo-(YTASARGDLAHLTTT), was synthesized by solution cyclization of a suitable, fully protected precursor (42). The synthesis of peptides A21 (YTASARGDLAHLTTTHARHLP) and A22Cys (TTCTASARGDLAHLTTTHACHL) and cyclization via a disulfide bond of the latter peptide have been described (43).

Inhibition of Virus Infectivity by Synthetic Peptides-
The assays were carried out essentially as described (16). Briefly, cell monolayers were washed with cell culture medium (Dulbecco's modified Eagle's medium) and incubated for 45 min at room temperature with PBS as a control or with several 10-fold dilutions of A15 or variant peptides in PBS. The final concentrations of peptide ranged from 0.1 M to 1 mM, depending on the variant tested. A defined amount of FMDV C-S8c1 diluted in PBS was then added (about 40 -100 plaque-forming units), and the cells were further incubated for 45 min at room temperature. After washing with Dulbecco's modified Eagle's medium, the cells were covered with an agar overlay and incubated for 36 h at 37°C. The monolayers were fixed and stained and the virus plaques counted. The IC 50 value is the concentration of peptide that produces a 50% inhibition in the number of plaque-forming units. A relative IC 50 was determined for each variant peptide by dividing its IC 50 value by the IC 50 of the homologous (A15) peptide. Each experiment was carried out in duplicate. Because the different concentrations of each peptide were tested in entirely independent experiments, the dose-response curves obtained for the peptides further confirmed that no gross experimental errors occurred. Reproducibility of the assay was checked by including in each experiment several concentrations of peptide A15 as a positive control and a high concentration (100 M) of an inactive variant (A15 with Glu substituted for Asp-143) (16) as a negative control. To provide another independent confirmation of the values obtained, an additional series of experiments was carried out in duplicate using the approximate IC 50 of each peptide obtained in the first series.

Inhibition of FMDV Infectivity by Linear and Cyclic Synthetic Peptides, Which Represent Antigenic Site A-Peptide A15
precisely reproduces antigenic site A (VP1 residues 136 -150) of FMDV C-S8c1 with regard to reactivity with monoclonal and polyclonal anti-virus antibodies (Refs. 44 and 45, and references therein). Because of the availability of the three-dimensional structure of peptide A15 complexed to an antibody ( Fig.  1), we have chosen this peptide as a model of the G-H loop to study the role of amino acid residues within this segment in determining the ability of FMDV to recognize and infect susceptible cells. Consistently with our previous results using related peptides of similar length (16), A15 very efficiently inhibited the infectivity of the homologous FMDV C-S8c1 (IC 50 ϭ 0.5 M) ( Fig. 2A).
The ability of short RGD-containing synthetic peptides to inhibit binding of protein ligands to integrins is sometimes significantly modulated by cyclization (e.g. Refs. 37 and 46). We have tested the effect of cyclization via an amide bond linking the N-and C-terminal residues of A15 on the inhibitory capacity of the peptide in virus infectivity assays (Fig. 2B). Interestingly, such head-to-tail cyclization led to complete loss of in- hibitory activity. This result suggests that an adequate conformation of the RGD motif is critical for cell recognition of the FMDV loop and that this particular cyclization prevented adoption of such conformation by the peptide. The same cyclization also abolished reactivity of the peptide with anti-FMDV antibodies. 2 In contrast, disulfide cyclization of a related peptide (peptide A22Cys) had no effect on its antigenicity (43) and on its inhibitory activity in infectivity assays (Fig. 2B).  The capacity of every possible singly substituted peptide representing variant forms of the G-H loop (residues 136 -150, including the RGD) to inhibit FMDV C-S8c1 infectivity was compared with that of the homologous A15 peptide. The only exceptions were Cys, Met, and Trp, which were not included to avoid ambiguity in interpreting the results due to oxidation (37). The results obtained using the complete replacement set are summarized in Figs. 3 and 4 and described below.
Evaluation of the Effect of Single Amino Acid Replacements within the RGD Triplet on the Capacity of Peptide A15 to Inhibit FMDV Infectivity-Systematic replacements at the RGD triplet (Figs. 3 and 4) clearly showed that almost any single substitution, either of Arg-141 or Asp-143, increased the IC 50 by much more than 2000-fold relative to that of the homologous peptide. Exceptions included replacement of Arg-141 by Ser, Gly, Lys, or Gln and of Asp-143 by Asn (relative IC 50 about 1000 -2000). In contrast, most replacements of Gly-142 gave relative IC 50 values from 160 (Tyr, Asn) to about 1000 (Leu, Ile, Val, and Phe); replacement of Gly-142 by Thr or Ser had a comparatively minor effect (relative IC 50 ϭ 10 -20). Thus, Gly is not as specific as Arg and Asp within this RGD motif. Figs. 3 and 4 also show that Tyr-136, one of the two site A residues that are absolutely conserved in FMDV (the other being Asp-143 at the RGD triplet), can be replaced by any other residue without affecting at all the inhibitory effect of the A15 peptide. This suggests that conservation of Tyr-136 among FMDVs is not due to any role in cell recognition (see "Discussion").

Evaluation of the Effect of Single Amino Acid Replacements within Segments Contiguous to the RGD Triplet on the Capacity of Peptide A15 to Inhibit FMDV Infectivity-Data in
Most single replacements at the two hypervariable stretches found within antigenic site A of FMDV type C (residues 137-140 and 148 -150; Fig. 5) had no significant effect on inhibition of FMDV infectivity by the peptide (Figs. 3 and 4). Only 10 out of 112 tested replacements had some negative effect (relative IC 50 between 6 and 12). Surprisingly, replacement of Ala-138 by Tyr, but not by any other residue, had a drastic negative effect (relative IC 50 about 200) on the inhibitory capacity of the peptide.
In contrast to the high tolerance for replacements within the hypervariable stretches, three of the four residues following the RGD triplet (LAHL, positions 144 -147; Fig. 5), were not generally replaceable in this assay (Figs. 3 and 4). In particular, most replacements of residues Leu-144 or Leu-147, which are highly conserved in most FMDV serotypes and also in other RGD-containing viruses ( Fig. 5; see "Discussion"), drastically decreased the inhibitory effect of the A15 peptide. Replacements of Leu-144 by most residues led to relative IC 50 values from about 40 to higher than 2200. Some chemically conservative substitutions (Ile, Val), and also Gln had less of an effect, with a relative IC 50 of about 15. The only replacements accepted at this position were the basic residues Lys and Arg. Leu-147 was not replaceable by most residues (relative IC 50 from about 30 to 900). Ala and Tyr and, again, Arg and Lys had a significant but not dramatic negative effect at this position (relative IC 50 from 10 to 25). Only some chemically conservative substitutions (Ile, Val, and Phe) were acceptable. Half of the replacements of Ala-145 also led to substantial negative effects, which were, however, generally milder (relative IC 50 from 6 to 65) than those at positions 144 and 147. In contrast, replacements of His-146, located within the same stretch, had either a minor effect or no significant effect at all. DISCUSSION The efficient, specific inhibitory effect on FMDV infectivity of A15 and related peptides is primarily due to inhibition of virus binding to cells (16). The use of long peptides permitted the systematic probing of the effect of replacements along most of the length of the RGD-containing protein loop under study. This FMDV loop is considered to act as an essentially selfcontained unit, which is loosely connected to the rest of the viral capsid (36,45,47). For example, single and multiple substituted site A peptides faithfully mimic reactivity with monoclonal antibodies and antigenic specificity of the homologous type C viruses (44,48). 3 Thus, functional studies with variant peptides representing this loop receive further validation. However, some caution is needed in interpreting the results; the effect of multiple substitutions on FMDV site A antigenicity is not necessarily additive, and it depends on the sequence context (49); thus, double or multiple replacements may have unexpected effects also on cell recognition.
Any replacement within the RGD triplet, especially of Arg and Asp, decreased dramatically recognition of peptide A15 by the FMDV receptor. The protruding Arg and Asp side chains of this RGD motif have been shown to directly interact with residues in the concave paratope of an antivirus neutralizing antibody (33), and they may similarly interact with the cell receptor. The presence of any side chain at the central position of this triplet could sterically interfere with receptor binding and/or impair adoption of an adequate RGD conformation. The much greater effect of the methyl group in RAD, relative to those of the larger side chains in RSD, RTD, RND, and RYD, and the most drastic effect of Pro in RPD support the latter possibility. A role of the conformation adopted by this RGD motif in receptor recognition is supported by the drastic effect of cyclization of peptide A15 on its inhibitory capacity. It may be noticed also that the functionality of SGD, GGD, KGD, and RGN, though very low, was higher than that of any other peptide substituted at Arg or Asp. This may have some signif- icance since unique FMDVs harboring the sequences RSGD (50) or RGGD (51) have been isolated, and the sequences KGD (52), NGR (the opposite of RGN (24,26)), and RYD (53) have been found involved in binding to some integrins.
In addition to RGD, three of the four amino acids within a helical stretch contiguous to this triplet (residues ϩ1 to ϩ4 downstream of the RGD; Fig. 1) participate in cell recognition. These four residues are highly conserved in serotype C FMDV (51), despite their critical interactions with most site A neutralizing antibodies (32,33,44,49), 4 which probably exert a strong selective pressure for variation (48, 51, 54 -58). Some of the few antigenic variants containing replacements within this segment showed decreased fitness (58). We proposed that FMDV of serotype C is highly intolerant of amino acid substitutions not only at the RGD but also within the contiguous helical segment because the entire stretch is involved in cell recognition (33,45). Based on a study with four site-directed mutants of a serotype A FMDV, Rieder et al. (39) also suggested a role for the helical stretch in attachment to cells. Leu residues at positions ϩ1 and ϩ4 downstream of an active RGD motif are also highly conserved in most FMDV serotypes (59, 60) (Fig. 5), in coxsackievirus A9, and in echovirus 22. This led Chang et al. (61) to propose a role for these residues in the function of the RGD of all these viruses, which may share a common receptor (9,15). The results described here provide experimental evidence for the above proposals. Replacement of the residue immediately following the RGD triplet also influenced binding of other RGD-containing ligands to some integrins (e.g. Refs. 1, 2, and 37).
In the FMDV loop peptide, Leu ϩ1 and Leu ϩ4 are located in the same face of the short helix contiguous to the RGD turn (Fig. 1). Though substitution of these Leu residues by other hydrophobic residues had, on average, a lesser effect than that of residues with a polar side chain, replaceability did not correlate well with similarities in hydrophobicity, size, and/or charge between the parent and the substituted amino acid (Fig.  4). These observations may favor an indirect role on receptor recognition, which may also be supported by ongoing molecular modeling of peptide variants and circular dichroism experiments.
In our assay, the only accepted replacements of Leu at position ϩ1 were the basic residues Arg and Lys (Met, Cys, and Trp could not be tested; see "Results"). This may be highly significant since, in addition to Leu, the only other residues found with some frequency at these positions in FMDV were Arg (present in most SAT2 and Asia1 isolates analyzed) and Met (in some type A and SAT3 isolates) (Fig. 5). The only accepted replacements of Leu at position ϩ4 were Phe, Ile, and Val and, accordingly, the very few type C isolates that lacked Leu at position 147 and had Ile or Val instead (Fig. 5); of more than 170 isolates sequenced only 6 (all of them of type A) showed some other residue at this particular position. Modulation of receptor recognition of poliovirus by variable residues within antigenic regions at the capsid surface has been very recently found (62). Similarly, the involvement of a somewhat variable stretch both in antibody and cell recognition of FMDV raises the possibilities that some FMDV variants could be recognized by different integrins and that immune pressure may occasionally lead to a switch in the putative integrin recognized by the virus. Early experiments indeed suggested that some FMDV of different serotypes could bind to different receptors (63).
Tyr-136 and the hypervariable stretches flanking the RGD-LAHL segment within antigenic site A of FMDV type C do not apparently play a major role in determining cell recognition. However, Tyr-136 is absolutely conserved in FMDV, and only a few different residues are found in each position within the hypervariable regions (Fig. 5). Thus, restrictions to variation unrelated to cell recognition may also be operating on this major antigenic viral loop.