Examination of soluble integrin resistant mutants of foot-and-mouth disease virus

Background Foot-and-mouth disease virus (FMDV) initiates infection via recognition of one of at least four cell-surface integrin molecules αvβ1, αvβ3, αvβ6, or αvβ8 by a highly conserved Arg-Gly-Asp (RGD) amino acid sequence motif located in the G-H loop of VP1. Within the animal host, the αvβ6 interaction is believed to be the most relevant. Sub-neutralizing levels of soluble secreted αvβ6 (ssαvβ6) was used as a selective pressure during passages in vitro to explore the plasticity of that interaction. Results Genetically stable soluble integrin resistant (SIR) FMDV mutants derived from A24 Cruzeiro were selected after just 3 passages in cell culture in the presence of sub-neutralizing levels of ssαvβ6. SIR mutants were characterized by: replication on selective cell lines, plaque morphology, relative sensitivity to ssαvβ6 neutralization, relative ability to utilize αvβ6 for infection, as well as sequence and structural changes. All SIR mutants maintained an affinity for αvβ6. Some developed the ability to attach to cells expressing heparan sulfate (HS) proteoglycan, while others appear to have developed affinity for a still unknown third receptor. Two classes of SIR mutants were selected that were highly or moderately resistant to neutralization by ssαvβ6. Highly resistant mutants displayed a G145D substitution (RGD to RDD), while moderately resistant viruses exhibited a L150P/R substitution at the conserved RGD + 4 position. VP1 G-H loop homology models for the A-type SIR mutants illustrated potential structural changes within the integrin-binding motif by these 2 groups of mutations. Treatment of O1 Campos with ssαvβ6 resulted in 3 SIR mutants with a positively charged VP3 mutation allowing for HS binding. Conclusions These findings illustrate how FMDV particles rapidly gain resistance to soluble receptor prophylactic measures in vitro. Two different serotypes developed distinct capsid mutations to circumvent the presence of sub-neutralizing levels of the soluble cognate receptor, all of which resulted in a modified receptor tropism that expanded the cell types susceptible to FMDV. The identification of some of these adaptive mutations in known FMDV isolates suggests these findings have implications beyond the cell culture system explored in these studies.

Previously, soluble α v β 3 and α v β 6 lacking the transmembrane and cytoplasmic tail domains were shown to still function as FMDV receptors [25]. Pre-treatment of serotype A and O FMDV particles with soluble secreted bovine α v β 3 and α v β 6 prior to application on permissive cell lines was investigated as an antiviral therapy. Interestingly, only soluble α v β 6 limited FMDV attachment to host cells by competing for receptor binding sites on virus particles. Soluble α v β 3 exhibited a low affinity interaction with the virus particles and failed to attach to FMDV in the same manner as α v β 6 with no significant effect on infectivity. It remains to be determined whether binding to blocking molecules such as soluble receptor, will impact FMDV interaction with the cellmembrane receptor or affect viral growth.
Here, we conducted in vitro experiments to evaluate the selective pressure exerted by soluble receptor protein on FMDV attachment and examined the evolution of virushost cell interactions. Studies conducted with poliovirus and its cognate receptor [26,27] showed that surface and internal capsid residues regulate attachment to the receptor and conformational change of the virus. Here, subneutralizing levels of soluble secreted bovine α v β 6 (ssα v β 6 ) were used to develop soluble receptor resistant mutants of FMDV A24 Cruzeiro. Of the 4 α v -integrins used by FMDV for host cell attachment, the β 6 heterodimer was selected on the basis that ssα v β 6 most substantially impeded FMDV infection [25]. Additionally, the β 6 heterodimer was shown to be most responsible for the tissue tropism of FMDV in cattle [28]. Following 3 successive passages of FMDV (serotype A and serotype O) pre-treated and co-incubated with ssα v β 6 on the LFBK cell line, which is permissive to infection by all 7 serotypes of FMDV ( Figure 1, Table 1) [29], virus was isolated that persisted despite the presence of soluble integrin (SI). The A-type isolates exhibited mutations in the normally conserved RGD motif or just outside of it within the G-H loop of VP1, while the O-type mutants displayed changes in VP3 and residues proximal to the VP1 RGD motif. SI resistant (SIR) FMDV mutants were further characterized for altered receptor tropism, relative sensitivity to neutralization by SI, as well as sequence and structural alterations in the G-H loop.

Selection of FMDV serotype A SIR mutants
To examine the adaptability of serotype A FMDV (represented by A24 Cruzeiro) to the presence of soluble receptor during infection, we employed ssα v β 6 , which is preferentially exploited by serotype A FMDV for host cell attachment [7,25]. As previously described, ectodomains of α v and β 6 integrin subunits were secreted from stably transfected cells, which were properly folded and capable of heterodimerization (data not shown) [25]. The bovine kidney LFBK cell line was selected for these studies on the basis that it is susceptible to infection by all 7 FMDV serotypes, similar to primary bovine kidney cell culture and [29]. Western blot analysis confirmed that LFBK cells express both β 3 and β 6 integrins ( Figure 1, Table 1). As expected, CHO 677 cells [30,38] did not express β 3 and β 6 integrins [21,[30][31][32][33][39][40][41]. While BHK-21 cells expressed β 3 integrin, we were unable to detect the presence of β 6 integrin. As an additional control, RNA Helicase A (RHA), known to be expressed in all 3 cell lines, was also detected. By extension, these findings allowed for the potential that both α v β 3 and α v β 6 could be found on the surface of LFBK cells, which is consistent with repeated Figure 1 Integrin expression profile. BHK-21, CHO 677, and LFBK cell lysates were examined by Western blot probing with: anti-β 3 integrin (ITGβ3), anti-β 6 integrin (ITGβ6), and anti-RHA. Equivalent loading between lanes was confirmed by probing with anti-tubulin-α. Molecular weight ladder bands are indicated. experiments showing the susceptibility of this cell line to non-HS-adapted field isolates of FMDV (data not shown). Wild-type (WT) FMDV A24 Cruzeiro was preincubated with 10 μg/mL ssα v β 6 (sub-neutralizing concentration for A24 Cruzeiro, Additional file 1: Figure S1) at 37°C for 1 hour prior to application of the coated virus on LFBK cells. As depicted in Figure 2A, LFBK cells were infected with ssα v β 6 coated WT FMDV A24 Cruzeiro at a MOI of 1 at 37°C for 24 hours (P1). Thirty-five plaques were isolated and used for a second round (P2) of infection with ssα v β 6 . Only 17 of the 35 infections demonstrated cytopathic effects (CPE) at 24 hpi. Isolates from 17 P2 infections were individually used for a third selection (P3) with ssα v β 6 . This resulted in only 4 productive infections derived from A24 Cruzeiro numbered: 15, 23, 42, and 45. The 4 soluble integrin resistant (SIR) serotype A (A-type) isolates (designated A-SIR #15, A-SIR #23, A-SIR #42, and A-SIR #45) were subject to a final selection with ssα v β 6 , where by 24 hpi A-type SIR isolates 15, 23, and 45 produced 100% CPE and isolate 42 produced 50% CPE.
The A-type SIR mutants diverged with respect to their plaque morphologies on different cell lines. WT and A-SIR mutants maintained relatively large plaque sizes on LFBK cells ( Figure 2B, top panel). On CHO K1 monolayers, A-SIRs #42 and #45 showed reduced plaque size relative to LFBK cells ( Figure 2B, lower panel). Comparative viral growth on LFBK and CHO K1 cells suggests some A-SIRs might utilize HS. To examine this further, the growth analysis was expanded to include the CHO 677 cell line ( Figure 1, Table 1) [39,40,44]. In addition, the IBRS2 cell line with a unique receptor profile where FMDV infection is initiated via α v β 8 [9,16,42], was also examined ( Table 1). IBRS2 cells were permissive to infection by WT and A-SIR viruses, growing to titers comparable to LFBK cells (Table 2). Interestingly, A-SIR #42 productively infected CHO 677 cells ( Figure 2B, Table 2). SIR #42 grew to roughly equivalent titers on CHO 677 and CHO K1 cells ( Figure 2B, Table 2). Viral growth on CHO 677 cells suggested this A-SIR mutant adapted to utilize an as yet unidentified and uncharacterized third FMDV receptor [21,41]. It was inferred that growth of A-SIR #45 on CHO K1 cells but not CHO 677 cells was indicative that this virus exploited HS. Different growth patterns for each A-SIR mutant on different cell lines also suggested these viruses are genetically distinct. These findings illustrated the rapid adaptability (after only 3 passages) of serotype A FMDV to overcome interference with host cell attachment.
A-SIR mutants exhibit reduced sensitivity to SI neutralization Next, the relative sensitivities of A-SIR mutants to ssα v β 6 neutralization were examined relative to their WT counterpart. Each virus was pre-incubated with ssα v β 6 at gradually increasing concentrations (5,10,20,30, and 40 μg/mL), applied to LFBK monolayers, and evaluated for reductions in virus titer. As shown in Figure 2C, the four A-SIR mutants could be divided into 2 classes with respect to sensitivity to neutralization by ssα v β 6 : a highly resistant (#15 and #23, Class I) and a moderately resistant class (#42 and #45, Class II). A-SIRs #15 and #23 grew to titers 200-1000 fold higher than WT with increasing ssα v β 6 , while mutants #42 and #45 exhibited titers 20-100 fold higher than WT ( Figure 2C). Cumulatively, these findings reinforced the supposition that preand co-incubation of ssα v β 6 acts as a selective pressure forcing FMDV particles to quickly select for resistance to the treatment.
A-SIR mutants continue to utilize α v β 6 Since the A-SIR mutants exhibited reduced sensitivity to ssα v β 6 neutralization, we investigated whether they exhibit an altered integrin preference: shifting from α v β 6 to another integrin or a non-integrin receptor. The capability of A-SIRs to infect cells expressing distinct integrins was tested using a previously established transient transfection-infection assay [45]. COS-1 cells do not support FMDV infection and lack integrins used by FMDV (Table 1) [7]. These cells were co-transfected with plasmids encoding the full-length α v -integrin subunit and 1 of 4 different full-length β subunits (β 1 , β 3 , β 5 , or β 6 ). Expression was confirmed by immunocytochemical staining ( Figure 3A). Of note, the antibody used to detect the α v β 1 only recognizes the β 1 subunit; as such, positive staining may be reflective of other β 1 containing integrin heterodimers. WT virus and A-SIR mutants were incubated on the cells transiently expressing integrin with 35 S-methionine. After 24 hours, levels of viral protein synthesis were evaluated by radioimmunoprecipitation (RIP) ( Figure 3B). Based on the sets where FMDV proteins (3D, VP0, VP1-3) were detected, it could be inferred as to which integrin heterodimers was/were used by WT and A-SIR viruses to gain entry to COS-1 cells for replication. Consistent with previous reports, viral protein synthesis was detected for A24 Cruzeiro WT in cells expressing α v β 1 , α v β 3 and α v β 6 ( Figure 3B) [7,25]. A-SIRs #15 and #23 were only amplified on α v β 6 expressing cells. A-SIRs #42 and #45 infected cells displaying α v β 1 and α v β 6 . However, the signal for virus proteins was noticeably lower for A-SIR #42 and #45 relative to WT, though we cannot exclude the possibility that this was partially due to differences in transfection efficiencies. Besides WT, A-SIR #45 was the only other virus able to infect via α v β 3 . Interestingly, although resistant to neutralization by ssα v β 6 , all 4 A-SIR viruses replicated on α v β 6 expressing COS-1 cells. Thus, it was inferred that all 4 A-SIR mutants maintained sufficient affinity for α v β 6 to continue to infect cells via this receptor.
To reinforce the supposition that the A-SIR viruses retained the ability to bind α v β 6 , CHO 677 cells, which could not permit FMDV infection except for A-SIR #42, were transiently transfected with α v β 6 as was done for COS-1 cells (data not shown). The expression of α v β 6 restored infectivity of WT and A-SIR viruses, with titers within 0.5 to 1 log less than on LFBK cells (Table 2), with A-SIR #45 attaining a titer 1 log higher than both class I SIRs. Notably, A-SIR #42 achieved titers 2 logs higher than on untransfected CHO 677 cells, suggesting α v β 6 is the preferred surface receptor for this virus.

Amino acid substitutions detected in A-SIR mutants
To further distinguish A-SIR viruses from their parental counterpart, A-SIR mutants were examined for alterations in their nucleotide and amino acid sequences relative to WT (Accession #AY593768). No amino acid substitutions were detected in VP4, VP2, or VP3, with a single silent mutation detected in VP3. Interestingly, the 2 A-type SIR classes exhibited 2 corresponding types of mutations in VP1 summarized in Table 3 and Figure 4. Class I A-SIRs that were highly resistant to ssα v β 6 (#15 and #23) displayed a G145D amino acid substitution in the RGD motif considered essential for integrin interaction. The V154A substitution distinguished A-SIR #23 from #15. Class II A-type SIRs moderately resistant to ssα v β 6 (#42 and #45) exhibited substitutions at the RGD + 4 position with L150P for #42 and L150R for #45. Moreover, A-SIR #42 displayed 2 additional amino acid alterations upstream of the RGD motif: E95K and S96L. Previous reports have described the helical segment immediately C-terminal to the RGD motif as contributing to the interaction between FMDV and its cognate integrin receptor, specifically, amino acid positions RGD + 1 and RGD + 4. The RGD + 1 and RGD + 4 side chains face out like those of the RGD motif and are frequently occupied by leucine residues [8,14,[46][47][48][49]. While proline substitution at the RGD + 4 position was previously detected in the A5 Westerwald FRG/58 field isolate ( [50] b; [51]) and serially passaged populations of a serotype C FMDV strain [21], the arginine substitution at this position appears unprecedented. Cumulatively, the P1 sequences support the findings that 2 distinct classes of SIR mutants were derived from A24 Cruzeiro: where each class exhibited distinct forms of amino acid substitutions localized to the VP1 G-H loop.

Structural predictions of A-SIR mutant modifications
After defining the amino acid changes within VP1 in the A-SIR mutants (Figure 4), we explored how those changes might affect the overall structure of the receptor-binding site in the G-H loop. The VP1 crystal structure has been solved for FMDV O1/BFS 1860/UK/ 67 (Accession 1FOD) [52]. Using the coordinates of 1FOD as a template, homology structures were generated of WT A24 Cruzeiro VP1 as well as those of the 4 A-SIR mutants using the Geno3D algorithm [53]. For each virus, 10 different "best fit" homology models were generated for VP1. Representative images of the G-H loop region were given particular scrutiny ( Figure 5).
The RGD to RDD substitution detected in class I A-SIR mutants #15 and #23 placed a bulky negatively charged residue into the center of the integrin binding site, which can be seen projecting outward in the homology structures generated. The introduction of D145 "spread out" the adjacent side chains within the RGD motif, increasing the distance between the R144 and D146 side chains relative to WT ( Figure 5). In all 10 of the predicted tertiary structures, the amino acid alteration failed to effect any significant change in the backbone structure of this region of VP1.
The previously undetected L150R substitution at the RGD + 4 position of class II A-SIR #45 replaces a linear non-polar amino acid side chain with a positively charged residue. Interestingly, Geno3D failed to predict a consensus structure for the A-SIR #45 G-H loop. Class II A-SIR mutant #42 exhibited a L150P substitution at RGD + 4, which was previously identified in FMDV A5 Westerwald FRG/58 [50,51]. The homology model of the A-SIR #42 G-H loop revealed the proline residue introducing a kink that "compressed" the RGD motif with the three side chains in much closer proximity than WT ( Figure 5). This structural prediction was essentially the opposite of what was produced for the class I A-SIR mutants.
In addition to L150P, A-SIR #42 also displayed 2 substitutions upstream of the RGD: E95K and S96L, which are proximal to the VP1-VP3 interface. Using Geno3D, ribbon diagrams were generated of capsid protomers from WT and A-SIRs to determine if the protein interface was disrupted (Additional file 2: Figure S2). A short helical domain observed at the VP1-VP3 interface in the WT sequence was notably absent with the E95K/S96L substitution, replaced with a loop structure. With L150P previously detected in A5 Westerwald, E95K/S96L may be more significant to the adaptation of A_SIR #42. We concluded that the mutations elucidated in the A-SIR    mutants produce hypothetical structural alterations likely to affect cellular recognition.

O1 Campos derived SIR mutants circumvent ssα v β 6 by a different mechanism
A-SIR mutants derived from A24 Cruzeiro overcame the selective pressure introduced by ssα v β 6 by compensatory mutations targeted to the RGD motif or the RGD + 4 position within the VP1 G-H loop. In an effort to determine if a similar strategy would be exploited by a different FMDV serotype, the experiment was repeated whereby O1 Campos was pre-treated and passaged three times in the continued presence of 10 μg/mL ssα v β 6 (also a sub-neutralizing concentration for O1 Campos, data not shown). Three O-type SIR viruses were recovered designated O-SIR #1, O-SIR #9, and O-SIR #46 ( Figure 6A). All three O-SIRs produced similar small plaque forming units on LFBK cells relative to their WT progenitor ( Figure 6B, top panel). Unlike the WT virus, the O-SIRs productively infected CHO K1 cells ( Figure 6B, bottom panel), achieving titers approximately 1 log lower than on LFBK cells (  Figure 6B, right panel and Table 4). When CHO 677 cells transiently expressing α v β 6 were substituted in this assay, infectivity was restored for all O-SIR mutants and O1 Campos WT, with titers equivalent to or exceeding those on LFBK cells (Table 4). This particular finding was consistent with what was observed for the A-SIR viruses, where affinity for α v β 6 was maintained. Additionally, both O1 Campos and the O-SIR mutants demonstrated titers 1-2 logs higher on IBRS2 cells relative to LFBK cells (Table 4), which suggested that these viruses infect via α v β 8 with greater efficiency. Further analysis revealed that all three O-SIRs were uniformly 2 logs more resistant to neutralization by ssα v β 6 up to 20 μg of ssα v β 6 ( Figure 6C). However, while this resistance was maintained by O-SIR #1 and O-SIR #9 above 20 μg ssα v β 6 , O-SIR #46 was inhibited by ssα v β 6 at concentrations above 20 μg. Initial examinations of the reliance of the O-SIRs on different integrin heterodimers revealed the presence of FMDV proteins in untransfected COS-1 cells infected with the O-SIR mutants, similar to the pattern observed for O-SIR infected LFBK cells ( Figure 6D). No increase in viral protein synthesis was observed for cells transiently expressing integrins (data not shown). Like CHO K1 cells, COS-1 cells express HS, thus it was inferred that the O-SIR mutants had circumvented ssα v β 6 neutralization by adapting to utilize HS for host cell attachment.
Given the disparities in plaque morphology and sensitivity to soluble integrin neutralization, it was expected that the O-type SIR mutants would be genetically distinct. In contrast to A-SIRs, compensatory mutations    were not confined to VP1, but were also identified within VP3 ( Figure 6E). The mutations detected in the O-SIRs were consistent with those previously found in HS-adapted serotype O FMDV passaged multiple times in cell culture with a signature H56R substitution in VP3 [22,23,54].

Discussion
Distinct lineages of RNA viruses are frequently referred to as "quasi-species" due to their inherent mutability [47,55]. Their rapid adaptability stems from the lack of "proofreading" function in the RNA-dependent RNA polymerases responsible for amplification of RNA viral genomes. The poliovirus RNA polymerase delivers approximately 1 error in 2200 bases, which is significant for a virus genome of approximately 7500 nucleotides [56]. Constant variation with each duplicated RNA viral population represents a stumbling block in the development of anti-viral therapies. A previous study explored using soluble receptor treatments to impair FMDV infection [25,49]. Here, we investigated the effect of subneutralizing levels of soluble secreted α v β 6 (ssα v β 6 ) on FMDV replication in cells and documented the rapid emergence of resistant FMDV populations, how they circumvented the selective pressure of ssα v β 6 , and the strategies deployed as they adapted.
Previous studies have investigated the effect of multiple passaging in cell culture on FMDV acquisition of alternative receptors, attenuated phenotypes, and for the dispensability of the RGD motif [21,49,57]. For instance, following 100 passages of a serotype C FMDV (C-S8c1) in cell culture, the virus was able to enter cells by an integrin-independent manner and multiple mutations were identified in the variant genomes [21]. For serotype A12 virus, it has been shown that mutations obtained during adaptation of field (bovine) isolates to cell culture localize downstream of the RGD and those mutations appear to alter the affinity for the integrin receptor [49].
Here, four SI resistant (SIR) FMD viruses derived from A24 Cruzeiro and three derived from O1 Campos were rapidly selected after only 3 rounds of cell culture selection in the continued presence of ssα v β 6 . The A-SIR mutants could be separated into 2 classes that were either highly or moderately resistant to neutralization by ssα v β 6 . Highly resistant A-SIRs (Class I) did not select HS as a secondary receptor and maintained α v β 6 affinity. These variants displayed a G145D substitution in the highly conserved VP1 RGD motif. The moderately resistant A-SIRs (Class II) also retained α v β 6 affinity and appeared to either utilize HS or an as yet unidentified and uncharacterized third FMDV receptor. Moreover, similar to A24 Cruzeiro, all 4 A-SIR mutants were amplified on IBRS2 cells, indicating that the A-SIRs have maintained the ability to infect and replicate on cells expressing an abundance of α v β 8 , which can also function as a FMDV receptor [9,16,42].

Class I A-SIR Mutants
A-SIR mutants appear to have circumvented ssα v β 6 treatment by 2 different routes. Substitution of an aspartic acid (G145D) in the RGD motif ( Figure 4) rendered A-SIR #15 and 23 (Class I) highly resistant to ssα v β 6 ( Figure 2C) while paradoxically maintaining use of α v β 6 for infection ( Figure 3B). Notably, class I A-SIRs did not adapt to utilize HS. Interestingly, this RDD mutation in the cell receptor binding site has been detected in a field strain of Asia1 virus (Asia1/JS/CHA/05) after just 2 passages, one in vivo and one in vitro [57]. RDD Asia1 variants showed no change in their ability to replicate in established cell lines, nor did it alter clinical onset of disease in animals tested [57]. While the environmental pressure that selected for RDD Asia1 variants is unknown, it suggests that this mutation might be a natural adaptation mechanism that could manifest across different FMDV serotypes. This hypothesis is supported by the observation that natural outbreaks of serotype A FMDV featuring the RDD amino acid substitution have occurred in Argentina on two occasions: A25-Arg/59 (GenBank #AY593769) and A25-Arg/ 61 (GenBank #AY593789).
The likely reason for the selection of a modified RGD was also investigated using homology models of the VP1 G-H loop structure and position of the RDD side chains ( Figure 5). The data showed an additional negative charge provided by G145D widened the gap between the side chains of R144 and D146. Potentially, the widening of the RDD side chains or the introduction of an additional negatively charged side chain might diminish the affinity of the motif for the integrin by sterically complicating the binding of ssα v β 6 , which is not anchored to a membrane, to the virus particle during pre-treatment in solution. Correspondingly, by reducing the affinity but not ablating the interaction with α v β 6 ( Figure 3B, Table 2), unoccupied RDD sites on the virus particle after pre-treatment and subsequent co-incubation with ssα v β 6 likely allow infection of host cells via α v β 6 .

Class II A-SIR Mutants
In contrast, moderately resistant Class II A-SIRs left the RGD intact, but substituted a key amino acid at the RGD + 4 position [8,14,[46][47][48], with the conserved leucine replaced with proline for A-SIR #42 and arginine for A-SIR #45. L150P was previously identified in FMDV A5 Westerwald FRG/58 [50,51] as well as a multiply passaged serotype C FMDV [21,58]. However, L150R appears unprecedented. Interestingly, while both viruses replicated in CHO K1 cells, only A-SIR #42 infected CHO 677 cells. However, both class II A-SIRs exhibited preference for α v β 6 over alternative surface molecules ( Figure 2, Table 2). Given L150P was previously detected in the WT field strain A5 Westerwald [51], it will be interesting to explore in the future whether this amino acid alteration was solely responsible for shifting receptor tropism to allow infection of cells devoid of FMDV integrin receptors and HS [21,30,38,41]. However, two additional mutations detected in VP1, E95K and S96L, might also contribute or be exclusively responsible for the A-SIR #42 phenotype.
Homology modeling was also employed to illustrate how Class II amino acid substitutions affected both the G-H loop ( Figure 5) and the capsid protomer (Additional file 2: Figure S2). In contrast to Class I A-SIRs, L150P appears to compress the side chains projecting from the RGD motif, where R144 and D146 are in closer proximity than in A24 Cruzeiro or Class I A-SIRs ( Figure 5). The unique E95K/ S96L substitutions localized to the VP1-VP3 interface, potentially altering capsid protomer stability (Additional file 2: Figure S2). Modeling algorithms failed to generate a consensus structure for the G-H loop of A-SIR #45, where multiple potential side chain angles for L150R altered the shape and presentation of the RGD motif (data not shown).

O-SIR mutants
When the approach used to generate the A-SIRs was applied to O1 Campos, this resulted in three distinctive O-SIR mutants that bind to HS and that carried a signature H56R mutation. This particular mutation has been observed in attenuated HS-adapted FMDVs [22,23,54]. Similar to their A-type counterparts, the O-SIRs also could be amplified on IBRS2 cells suggesting that the selective pressure of co-incubation with ssα v β 6 did not abrogate the affinity for α v β 8 . However, unlike A24 Cruzeiro and the A-SIRs, O1 Campos and the O-type SIR mutants amplified 2-3 logs more in IBRS2 cells relative to LFBK cells, thus it could be inferred that serotype O FMDV may have a preference for α v β 8 as a receptor. As such it would be intriguing to explore the effects of successive passages of O1 Campos in the presence of ssα v β 8 in the future. Lastly, it is interesting to note that the mutations accumulated by the O-SIRs were not in the VP1 RGD motif, but rather peripheral to it and at locations in VP3, separate from the primary site of attachment to the host cell.

Conclusions
Together the results of this study have shed light on the plasticity of FMDV serotype A and O interactions with its primary cognate receptor: α v β 6 . Amino acid substitutions detected in SIR mutants were rapidly selected to overcome soluble receptor neutralization. Thus, the tolerance of the VP1 G-H loop to amino acid substitutions plays an essential role in cell receptor adaptability of FMDV. Interestingly, mutations identified in the current study have been also identified in FMDV field isolates causing outbreaks suggesting that similar selective pressures may exist in the natural host environment. The mode and relative speed with which FMDV was able to adapt to a selective pressure, represented here by ssα v β 6 , highlights both the evolutionary advantage of highly mutable RNA viruses and the challenges of designing effective antiviral therapies against these pathogens.

Radio-immunoprecipitation (RIP)
Virus-infected cells grown overnight in the presence of 35 S-methionine were lysed with 1% Triton X100. Aliquots of each sample were precipitated with 20% trichloroacetic acid (TCA) to determine the counts per minute (cpm). Lysates were mixed with Protein A/G agarose beads, tumbled 15 minutes, centrifuged at 1500 rpm 10 minutes, and the supernatants collected. Recovered supernatant was tumbled with indicated antibodies (6S6, LM609, 15 F11, and CSβ6) 1 hour at 4°C. The mixture was tumbled overnight at 4°C with fresh Protein A/G agarose. Afterwards, the agarose was washed 3 times with NET/NP40 buffer (NaCl, EDTA, Tris, Nonidet-P40). Finally, the agarose was boiled in sample buffer without β-mercaptoethanol, pelleted, the supernatants separated by SDS-PAGE, and the gel examined by autoradiography.

Virus titer assay
One hour post-adsorption, the inoculum was removed, and cells washed in a mild acid solution followed by virus growth media (VGM, DMEM containing L-glutamine). VGM was then added and cells incubated 24 hours at 37°C. Afterwards, virus-infected cells were harvested and titers determined by plaque assay as previously described [60]. Plates were fixed, stained with crystal violet (0.3% in Histochoice; Amresco, Solon, OH), and plaques counted. Values calculated for number of plaque-forming units (PFUs) per milliliter (mL) were plotted using Microsoft Excel (Microsoft Corporation, Redmond, WA). Assays were performed in triplicate.

Sequencing of the FMDV P1 Region
The P1 region in twenty FMDV isolates were sequenced from PCR product with sequencing primers providing at least 3X coverage across 3,000 base pairs. PCR products were purified with QIAQuick spin columns (Qiagen) in accordance with the manufacturer's instructions. Sequencing reaction mixtures (10 μl) contained 2.5 μM primer, 20 ng of PCR product, and 0.75 μl of Big Dye (Applied Biosystems) in molecular biology-grade water. The sequence cycling conditions were 30 s of preincubation at 85°C; 25 cycles of 10 s at 96°C, 5 s at 50°C, and 4 min at 60°C; and a 10-min 60°C final extension. The sequencing reaction mixtures were purified with Agencourt's CleanSEQ system in accordance with the manufacturer's instructions (Beckman Coulter).