Positively Charged Residues at the Five-Fold Symmetry Axis of Cell Culture-Adapted Foot-and-Mouth Disease Virus Permit Novel Receptor Interactions

Abstract

Field isolates of foot-and-mouth disease virus (FMDV) have a restricted cell tropism which is limited by the need for certain RGD-dependent integrin receptors. In contrast, cell culture-adapted viruses use heparan sulfate (HS) or other unidentified molecules as receptors to initiate infection. Here, we report several novel findings resulting from cell culture adaptation of FMDV. In cell culture, a virus with the capsid of the A/Turkey/2/2006 field isolate gained the ability to infect CHO and HS-deficient CHO cells as a result of a single glutamine (Q)-to-lysine (K) substitution at VP1-110 (VP1-^Q110^K). Using site-directed mutagenesis, the introduction of lysine at this same site also resulted in an acquired ability to infect CHO cells by type O and Asia-1 FMDV. However, this ability appeared to require a second positively charged residue at VP1-109. CHO cells express two RGD-binding integrins (α5β1 and αvβ5) that, although not used by FMDV, have the potential to be used as receptors; however, viruses with the VP1-^Q110^K substitution did not use these integrins. In contrast, the VP1-^Q110^K substitution appeared to result in enhanced interactions with αvβ6, which allowed a virus with KGE in place of the normal RGD integrin-binding motif to use αvβ6 as a receptor. Thus, our results confirmed the existence of nonintegrin, non-HS receptors for FMDV on CHO cells and revealed a novel, non-RGD-dependent use of αvβ6 as a receptor. The introduction of lysine at VP1-110 may allow for cell culture adaptation of FMDV by design, which may prove useful for vaccine manufacture when cell culture adaptation proves intractable.

INTRODUCTION

Foot-and-mouth disease (FMD) is endemic in many regions of the world and is one of the most widespread, epizootic transboundary animal diseases, affecting many species of wildlife and livestock, such as cattle, sheep, goats, and pigs. The significant economic losses that result from FMD are due to the high morbidity of infected animals and stringent trade restrictions imposed on affected countries (1). Vaccination plays a major role in controlling FMD, either to lessen the effects of an outbreak in FMD-free countries or for control and eradication in regions where it is endemic. The etiological agent of FMD, foot-and-mouth disease virus (FMDV), exists as seven distinct serotypes (O, A, C, Asia-1, and the Southern African Territories [SAT] serotypes SAT-1, SAT-2, and SAT-3). Within each serotype, a large number of antigenic variants exist (2). Intraserotype diversity is driven by a high mutation rate during replication that is caused by an error-prone viral RNA-dependent RNA polymerase (3) and thus complicates efforts to control disease by vaccination due to incomplete protection between some antigenic variants (4). Hence, the most effective vaccines closely match the outbreak virus, which can necessitate the development of new vaccine strains. The current vaccines are inactivated virus preparations grown in large-scale cell culture. Therefore, the production of a new vaccine is critically dependent upon adaptation of viruses from the field for growth in cell culture, which can prove problematical for some viruses.

Foot-and-mouth disease virus is the type species of the Aphthovirus genus of the Picornaviridae, a family of nonenveloped, single-stranded positive-sense RNA viruses. The viral capsid is formed by 60 copies each of four structural proteins (VP1 to VP4) arranged in icosahedral symmetry. The outer capsid surfaces are formed by VP1, which surrounds the five-fold symmetry axis, and VP2 and VP3, which alternate around the three-fold axis (5). VP4 is myristoylated and located inside the capsid and is thought to play an essential role in the final stage of assembly and in endosomal membrane penetration by the viral RNA (6, 7). In vivo, FMDV has a strong tropism for epithelial cells, which is in part due to the epithelial cell-restricted expression of integrin αvβ6, which is the principal receptor used by field viruses to initiate infection (8–12). Integrin binding is mediated by a highly conserved arginine-glycine-aspartic acid (RGD) motif located at the apex of a structurally disordered loop (the GH loop of VP1). The integrin specificity of FMDV has been the subject of several studies, and three other RGD-dependent integrins (αvβ1, αvβ3, and αvβ8) have also been reported to be receptors for field strains of the virus (13–15); however, the role of these integrins in pathogenesis is unclear, and we have found that αvβ3 is a poor receptor for FMDV in vitro (16). Furthermore, despite recognizing their ligands via the RGD motif, two other RGD-dependent integrins (αvβ5 and α5β1) do not appear to serve as receptors for FMDV (17). This may be in part due to the residues that flank the viral RGD that are known to influence integrin-ligand interactions (10). Structural analyses of FMDV and FMDV-derived peptides have shown that the integrin-binding loop consists of a short region of a β-strand followed by the RGD, which is in turn is followed by a helical structure (16, 18–22). Typically, native ligands for αvβ6 have leucine (L) or methionine (M) at the RGD +1 site and leucine or isoleucine (I) at the RGD +4 site (16, 23, 24). FMDV may be highly adapted to use αvβ6 as a receptor, as it has a similar conserved sequence (L, M, or arginine at the RGD +1 site and L or I at the RGD +4 site) following the RGD. This region is known to be important for binding to αvβ6, as ligands that lack a complete RGD have been shown to bind αvβ6 via a DLXXL motif (where X indicates any amino acid) (24), and we have shown that alanine substitutions at either the RGD +1 or +4 site reduces the potency of FMDV-derived peptides as anti-αvβ6 antagonists (16). The integrity of the helix after the RGD is also important for binding to αvβ6, as it maintains the RGD +1 and RGD +4 residues in orientations accessible for direct interactions with the integrin (18, 25). These observations suggest that the helix and the identity of the residues at the RGD +1 and +4 sites play important roles in defining the integrin specificity of FMDV.

A major driving force for cell culture adaptation of FMDV is that the availability of receptors and passage through cultured cells often results in the selection of variants with altered receptor preferences (5). For example, cell culture growth often selects for viruses that use heparan sulfate (HS) as a receptor; HS can initiate infection via an integrin-independent process (26–33). As a result, cell culture-adapted viruses have an increased virulence and expanded host range for cultured cells. This has led to HS-binding viruses being characterized by their ability to infect CHO cells, which lack all of the known integrin receptors of FMDV, combined with an inability of these viruses to infect HS-deficient CHO cells (30).

Most information regarding HS binding has come from studies with type O FMDV. The HS-binding site is formed by a shallow depression in the center of the biological protomer and accommodates four or five sugar residues that make multiple contacts with all three outer capsid proteins (34). Remarkably, most of this structure is conserved in field viruses, and the switch to HS binding arises from only one or two residue changes at the center of the HS-binding site that result in a net gain in positive charge (33, 34). The most important change is at VP3-56, which is typically histidine (H) in field viruses and switches to arginine (R) in cell culture-adapted strains. The R is important for HS binding, as it allows ionic interactions with two sulfate groups. HS binding has also been demonstrated for other FMDV serotypes. In type C FMDV, the capsid residues that contribute to HS binding appear to be different from those of type O viruses and have not been precisely mapped, although the acquisition of a positive residue at VP3-173 has been strongly implicated to play a major role (26, 27). Cell culture-adapted SAT viruses have also been reported to use HS as a receptor. As for the type C viruses, the capsid residues that bind HS are different from those of type O viruses and have been reported to cluster around the five-fold symmetry axis of the virion (31, 32). For type A viruses, the mechanism of cell culture adaption is less clear; we found that FMDV A10₆₁ has an HS-binding site that is structurally similar to that of type O FMDV (28), whereas Rieder et al. found that cell culture adaption of FMDV A12 selected for variants with residue changes near the RGD motif, which suggested that the virus may have altered its integrin specificity (35). Other nonintegrin, non-HS receptors can also mediate FMDV infection, as type O and type C viruses have been described that lack the RGD motif and that infect HS-deficient cells by using an as-yet-unidentified receptor (26, 27, 36). For type C viruses, the site on the capsid associated with this phenotype has not been identified, while for type O viruses VP1 residues that lie close to the five-fold symmetry axis have been implicated.

We previously described an FMDV infectious copy plasmid (pO1K-A) that carries genes for capsid proteins VP1, VP2, and VP3 of the A/Turkey/2/2006 field strain combined with VP4 and the nonstructural proteins of a cell culture-adapted type O virus, O₁Kaufbeuren B64 (O1KB64) (37). Virus (O1KA-A/BTY1) recovered from this plasmid caused normal signs of FMD in cattle and pigs (37, 38). Here, we studied cell culture adaption of O1K-A/BTY1 and identified a single amino acid substitution (glutamine [Q] to lysine [K]) at VP1-110 that allows for integrin- and HS-independent infection of CHO cells. The introduction of K at this same site into recombinant viruses with capsids derived from field isolates of type O and Asia-1 FMDV also allowed infection of CHO cells. The only proviso is that CHO cell infection appeared to also require a second positively charged residue (K or R) at VP1-109. However, with primary bovine thyroid cells (pBTY), which express αvβ6, the presence of lysine at VP1-110 appeared to result in enhanced interactions with αvβ6 that allowed virus with KGE in place of the integrin-binding RGD to use αvβ6 as a receptor.

MATERIALS AND METHODS

Cells.

BHK cells were cultured in Dulbecco's modified Eagle's medium, and CHO and heparan sulfate-deficient CHO cells (pgsD-677 and pgsA-745) were cultured in Ham's F-12, with each medium supplemented with 10% fetal calf serum (FCS), 20 mM glutamine, penicillin (100 SI units/ml), and streptomycin (100 μg/ml). Primary bovine thyroid (pBTY) cells were prepared and cultivated as described previously (39).

Peptides, antibodies, and reagents.

The FMDV 17-mer peptide (VPNLRGDLQVLAQKVAR) and the control RGE version were synthesized at the Pirbright Institute. The FMDV 12-mer peptide (VPNLRGDLQVLA) and the control RGE version were synthesized at the Oxford Centre for Molecular Science, Oxford, United Kingdom. GRGDSP and GRGESP were purchased from Anaspec. Monoclonal antibody (MAb) P1F6 (mouse anti-αvβ5) and MAb 23C6 (mouse anti-αvβ3) were from Chemicon. MAb 6.8G6 (mouse anti-αvβ6) was a gift from Biogen (40). MAb PB1 (mouse anti-hamster α5β1) was from the Developmental Studies Hybridoma Bank (University of Iowa) and purified with protein A (Pierce) according to the manufacturer's instructions. MAb 2C2, which recognizes the FMDV 3A protein, was a gift from Emiliana Brocchi (IZS, Brescia, Italy) (41).

Infectious copy plasmids and rescue of infectious copy-derived viruses.

Construction of infectious copy plasmids O1K-A and O1K-OUK has been described in detail previously (37). These plasmids are based on the infectious copy plasmid pT7S3, which carries a cDNA copy of the full-length viral RNA for FMDV O1K/B64. In plasmids O1K-A and O1K-OUK, the coding regions for VP2 (1B), VP3 (1C), VP1 (1D), and 2A of O1K/B64 have been replaced by the corresponding coding sequences of FMDV A/Turkey/2/2006 and O/UKG/34/2001, respectively. A stock of FMDV Asia-1 BAR/9/2009 was obtained from the FAO World Reference Laboratory for FMD at the Pirbright Institute, Pirbright, United Kingdom, and passaged once through pBTY cells. After a second round of infection of pBTY cells, RNA was extracted using TRIzol (Invitrogen) as recommended by the manufacturer. Single-stranded cDNA, PCR, and construction of infectious copy plasmid O1K-Asia were carried out essentially as described for O1K-A and O1K-OUK, and the appropriate oligonucleotide primers used are listed in Table 1.

Table 1.

Primer sequences for cDNA synthesis, mutagenesis, PCR, or sequencing

Virus	Name	5′–3′ sequence	Assaya
A-Turkey/2006	A-VP1-Q110K-m1	CCGCCTACCACAAGAAGCCATTTACGAG	M
	A-VP1-Q110K-m2	CTCGTAAATGGCTTCTTGTGGTAGGCGG	M
	A-VP1-K109Q-m1	CCGCCTACCACCAGAAGCCATTTACGAG	M
	A-VP1-K109Q-m2	CTCGTAAATGGCTTCTGGTGGTAGGCG	M
	A-VP1-K109A-m1	CCGCCTACCACGCAAAGCCATTTACGAG	M
	A-VP1-K109A-m2	CTCGTAAATGGCTTTGCGTGGTAGGCG	M
	A-KGEm1	TACAACTGGTAATGGCAGAAAAGGTGAACTGGGGCCTCTTGCGGCGCG	M
	A-KGEm2	CGCGCCGCAAGAGGCCCCAGTTCACCTTTTCTGCCATTACCAGTTGTA	M
	AVP3F	GCCGGTTGCTTGTACAGAC	S
	AVP3R	GTCTGTACAAGCAACCGGC	S
OUKG	O-VP1-A110K-m1	ACCAATCCAACGGCTTACCACAAGAAACCGCTCACCCGGCTTGCACTGCCT	M
	O-VP1-A110K-m2	AGGCAGTGCAAGCCGGGTGAGCGGTTTCTTGTGGTAAGCCGTTGGATTGGT	M
	O-Seq1	GGTGAGTTCCCTTCTAAG	S
	O-Seq2	GCGAGCGTCAACTGGCA	S
Asia-1/Bar9/2009	Asia-cDNA	GACTGGGTCCTTTGACCGTGCT	cDNA
	Asia/Bar/NheI	CGCTGCTAGCTGACAAGAAAACAGAAGAGACA	PCR
	Asia/Bar/ApaI	AGAAGAAGGGCCCAGGGTTGGAC	PCR
	AsiaVP1-Q110K-m1	AACCCAACTGCCTACCAGAAGAAGCCCATCACCCGCCTGGCG	M
	AsiaVP1-Q110K-m2	CGCCAGGCGGGTGATGGGCTTCTTCTGGTAGGCAGTTGGGTT	M
	Asia-1seq1	CCAACTGACTCTTTTCC	S
	Asia-1seq2	CAGAGACTACAAGTGTGCA	S
All viruses	O1A	AACAACTACTACATGCAGC	S
	O2B	CTCCTGCATCTGGTTGATGG	S

^aAssay abbreviations: cDNA, cDNA synthesis; M, mutagenesis; S, sequencing.

Infectious copy plasmids based on O1K-A and O1K-OUK were linearized by digestion with HpaI, and full-length RNA transcripts were made with T7 RNA polymerase using an Ambion Megascript kit as described by the manufacturer. Due to the presence of multiple HpaI sites, RNA was synthesized from plasmids based on O1K-Asia without the linearization step. The RNAs were checked by using agarose gel electrophoresis and then introduced into BHK cells by electroporation, essentially as described previously (42). Electroporated BHK cells were incubated for 24 h and were harvested following freezing. In each case, a subsequent passage of the virus onto BHK or pBTY cells was performed, and the appearance of cytopathic effect (CPE) was monitored. Mutations were introduced into infectious copy plasmids by using a QuikChange Lightning site-directed mutagenesis kit (Agilent Technologies, CA) as described by the manufacturer and the appropriate primer pairs as listed in Table 1.

Virus passage.

Cell monolayers (∼90% confluent) in 25-cm² flasks were infected with FMDV. At 24 to 48 h postinfection (i.e., when extensive CPE was seen), the cells were freeze-thawed, and the lysate was clarified and stored at −80°C. Each infection used 1 ml of the previous infection in a total volume of 5 ml of virus growth medium (normal cell culture medium with 1% FCS).

Plaque assay.

Subconfluent cell monolayers were incubated with serial dilutions of FMDV samples for 15 min. The cells were then overlaid with 4 ml of Eagle's overlay (0.6% indubiose, 5% tryptone phosphate broth, 1% FCS in Eagle's medium). At 48 h postinfection, the cells were fixed and stained with 4% formaldehyde and methylene blue.

Quantification of infection.

The assay to quantify infection has been described in detail previously (17). Briefly, cells in 96-well tissue culture plates were grown until approximately 90% confluent. The cells were incubated for 1 h at 37°C with FMDV at a multiplicity of infection (MOI) of 0.3 PFU/cell. The monolayers were washed and incubated with serum-free medium at 37°C for a further 3 h. Infection was stopped, and the cells were fixed with 4% paraformaldehyde. The cells were permeabilized with 0.1% Triton and incubated with blocking buffer (10 mM Tris-HCl [pH 7.5], 140 mM NaCl, 1 mM CaCl₂, 0.5 mM MgCl₂, 10% normal goat serum, 1% fish gelatin). Infected cells were identified by sequential incubation with MAb 2C2, a biotinylated goat anti-mouse IgG secondary antibody (Southern Biotechnologies) and streptavidin-conjugated alkaline phosphatase (Caltag Laboratories), each for 1 h at room temperature, followed by alkaline phosphatase substrate (Bio-Rad) for 10 min. The cells were then washed with distilled water and allowed to air dry. Infected cells stained dark blue and were counted using an enzyme-linked immunosorbent spot assay plate reader (Zeiss). Nonspecific labeling was determined by either omitting MAb 2C2 or performing the assay with mock-infected cells. For competition experiments, peptides or antibodies were added to the cells for 0.5 h (peptides) or 1 h (antibodies) prior to the addition of virus and remained present during the virus incubation step. Each experiment was carried out with triplicate samples for each condition.

vRNA extraction, PCR, and DNA sequencing.

Viral RNA (vRNA) was extracted to obtain 0.5 ml of infected cell lysate by using TRIzol and resuspended in 20 μl RNase-free water as per the manufacturer's instructions (Invitrogen). Single-stranded DNA was synthesized from RNA templates by using the Invitrogen first-strand cDNA synthesis kit and primer O2B as per the instructions provided. PCR was performed using Kod polymerase (Novagen) and primers O1A and O2B. The cycling conditions were the following: 95°C for 2 min; 30 cycles of 95°C for 20 s, 52.6°C for 15 s, and 70°C for 1 min; a final incubation at 70°C for 1 min. PCR products were analyzed on standard agarose gels, extracted using the Illustra GFX PCR DNA and gel band purification kit (GE Healthcare), and eluted in 30 μl of water. PCR products were sequenced using the BigDye Terminator v3.1 cycle sequencing kit (Applied Biosystems) and the Applied Biosystems 3730 DNA analyzer.

RESULTS

Lysine at VP1-110 permits heparan sulfate-independent FMDV infection of CHO cells.

Previously, we described a recombinant virus (O1K-A/BTY1) recovered from an infectious copy plasmid (pO1K-A) by transfection of in vitro-transcribed vRNA into BHK cells and one passage through pBTY cells (37). As expected for a virus with a capsid derived from a field isolate, O1K-A/BTY1 was infectious for pBTY and BHK cells but was not infectious for CHO cells (Table 2), as they lack all of the known integrin receptors of FMDV. Similarly, after one further passage through BHK cells, this virus (O1K-A/BTY1/BHK1) was not able to infect CHO cells. After three further passes through BHK cells, the resulting virus (O1K-A/BTY1/BHK4) had gained the ability to infect CHO cells, although the titer was ∼100-fold lower than on BHK cells (Table 2), indicating a lower efficiency of infection of CHO cells. The sequence of the entire capsid encoding region of O1K-A/BTY1/BHK4 was determined and revealed only one residue difference from the parental virus. In A/Turkey 2/2006, O1K-A/BTY1, and O1K-A/BTY1/BHK1, glutamine (Q) occupies VP1-110, while in O1K-A/BTY1/BHK4, lysine (K) is found at this site. To confirm that this substitution confers the ability to infect CHO cells, we engineered the Q-to-K change into VP1-110 (VP1-^Q110^K) of pO1K-A. After one passage through BHK cells, the rescued virus (O1K-A/VP1-^Q110^K/BHK1) had an identical capsid sequence as O1K-A/BTY1/BHK4 and was able to infect pBTY, BHK, and CHO cells. After two more passes through BHK cells, the resulting virus (O1K-A/VP1-^Q110^K/BHK3) retained K at VP1-110 and the ability to infect CHO cells (data not shown).

Table 2.

Titers of infectious copy-derived viruses (A/Turkey/2/2006 capsid) with K/Q, K/K, or R/K at VP1 109/110

Infectious copy plasmid and virus	VP1 109/110a	Titer of virus in cell typeb
		pBTY	BHK	CHO	CHO-677	CHO-745
pO1K-A	K/Q
O1K-A/BTY1	K/Q	3.5 × 106	2.5 × 105	0	0	0
O1K-A/BTY1/BHK1	K/Q	2.5 ×107	4 × 106	0	0	0
O1K-A/BTY1/BHK4	K/K	ND	1.9 × 107	1.7 × 105	2.0 × 105	2.0 × 105
pO1K-A/VP1-Q110K	K/K
O1K-A/VP1-Q110K/BHK1	K/K	3.0 × 106	1.0 × 107	2.0 × 105	2.0 × 104	2.0 × 104
pO1K-A/VP1-109AK110	A/K
O1K-A/VP1-109AK110/BHK2	A/K	ND	1.5 × 105	0	ND	ND
O1K-A/VP1-109AK110/BHK6	A/K	ND	CPE	0	ND	ND
pO1K-A/VP1-109QK110	Q/K
O1K-A/VP1-109QK110/BHK3	Q/K	ND	2.5 × 106	0	ND	ND
O1K-A/VP1-109QK110/BHK6	R/K	ND	4.0 × 107	1.0 × 104	1.0 × 103	ND

^aResidues at VP1 109 and 110.

^b0, no CPE; ND, not done; CPE, cytopathic effect seen but virus titer not determined.

A net gain of positive charge on the capsid is associated with HS binding (see above). Therefore, we investigated whether viruses with K at VP1-110 also required HS receptors to initiate infection in CHO cells that were deficient in expression of HS (CHO-677) or expression of both HS and chondroitin sulfate (CS) (CHO-745). Consistent with the inability to infect CHO cells, viruses with Q at VP1-110 (O1K-A/BTY1 and O1K-A/BTY1/BHK1) did not infect CHO-677 or CHO-745 cells, whereas viruses with K at this position (O1K-A/BTY1/BHK4 and O1K-A/VP1-^Q110^K/BHK1) were able to infect both cell lines (Table 2). We verified that CHO-677 and CHO-745 cells lacked HS by showing that they were not susceptible to infection by FMDV O1BFS/1860 (data not shown), a virus shown previously to rely solely on HS as the receptor for infection of CHO cells (16, 30). These observations showed that the acquisition of K at VP1-110 allows for HS-independent infection of CHO cells. We also tested the ability of O1K-A/BTY1/BHK4 to infect CHO-lec2 cells, which are deficient in sialic acid. These cells were also susceptible to infection (titer on CHO-lec2 cells, 1.8 × 10⁵) by a virus with K at VP1-110.

Lysine at VP1-110 permits integrin-independent FMDV infection of CHO cells.

Heparan sulfate is used as a receptor by certain cell culture strains of FMDV that lack the need for the viral RGD or cellular integrins (see above). Therefore, it is possible that K at VP1-110 also creates an integrin-independent cell attachment site. Alternatively, K at VP1-110 could serve to enhance integrin-mediated infection. The ability to infect CHO cells suggests the former, as these cells do not express any of the known integrin receptors of FMDV. However, CHO cells express two other RGD-binding integrins (αvβ5 and α5β1) that are not normally used by FMDV (30, 43–45), and it is possible that the acquisition of K at VP1-110 may enhance use of these integrins as receptors. To investigate this possibility, we carried out competition experiments using RGD-containing peptides, a short GRGDSP peptide and two longer peptides (a 12-mer and 17-mer) that have sequences derived from the FMDV integrin-binding loop (and their control RGE counterparts), along with function-blocking antibodies to α5β1 (PB1) (46) and αvβ5 (P1F6) (40). The peptides have been shown to block binding to a number of RGD-dependent integrins, including αvβ1, αvβ3, αvβ5, αvβ6, αvβ8, and α5β1 (10, 14, 15, 47–50), and the antibodies are known to be cross-reactive for hamster integrins (45, 46). CHO cells were pretreated with these reagents prior to incubation with virus (O1K-A/VP1-^Q110^K/BHK1) for 1 h at 37°C. The cells were washed to remove extracellular virus, and infection continued for a further 3 h. The cells were fixed and permeabilized, and infected cells were identified and quantified as described in Materials and Methods. Figure 1 shows that the peptides (Fig. 1A) (compared to the corresponding control RGE peptides) and anti-integrin antibodies (Fig. 1B) did not inhibit infection of CHO cells, thereby showing that viruses with K at VP1-110 do not use α5β1 or αvβ5 as a receptor.

Fig 1.

(A) CHO cells in 96-well plates were mock treated or treated with peptides for 0.5 h and then infected with FMDV O1K-A/VP1-^Q110^K/BHK1 for 1 h at an MOI of 0.3 in the absence or presence of peptide (17-mer, 12-mer, GRGDSP [6-mer], or counterpart RGE control peptides). (B) CHO cells in 96-well plates were mock treated or treated with antibodies (20 μg/ml) for 1 h and then infected with FMDV O1K-A/VP1-110K/BHK3 for 1 h at an MOI of 0.3 in the absence or presence of antibody (PIF6, αvβ5; PB1, α5β1). At the end of each experiment (panels A and B), the monolayers were washed and infection was continued for a further 3 h. Infection was stopped, the cells were fixed and permeabilized, and infected cells were identified as described in Materials and Methods. The results were normalized to the mock-treated controls. Shown are the means ± standard deviations for triplicate wells. Each panel shows one experiment representative of two conducted, each giving similar results. Statistical significance was analyzed using an unpaired one-tailed t test with a P value less that 0.05 taken to be significant. NS, not significant.

To confirm that infection of CHO cells by viruses with K at VP1-110 does not require integrins, we attempted to generate a mutant virus with K at VP1-110 and KGE in place of RGD. The KGE sequence is commonly used as a negative control for RGD-dependent interactions, and recombinant FMDV with capsids derived from field isolates are rendered noninfectious by the KGE-for-RGD substitution (26, 27, 43, 44, 51). The KGE virus was passaged twice through BHK cells (O1K-A/VP1-^Q110^K/KGE/BHK2) without visible signs of CPE. However, subsequent passage on BHK cells resulted in extensive CPE. The capsid-coding sequences of the BHK3 and BHK4 viruses (O1K-A/VP1-^Q110^K/KGE/BHK3 and O1K-A/VP1-^Q110^K/KGE/BHK4) were determined. Both viruses showed only a single residue change compared to the input virus, one that resulted in an E-to-D change to create KGD at the RGD site. Therefore, to obtain a virus with KGE, we passaged the O1K-A/VP1-^Q110^K/KGE/BHK2 stock (which did not show signs of CPE on BHK cells; see above) through CHO cells. The first passage (O1K-A/VP1-^Q110^K/KGE/BHK2/CHO1) resulted in partial CPE at 48 h postinfection, while the next passage (O1K-A/VP1-^Q110^K/KGE/BHK2/CHO2) resulted in complete CPE at 24 h. The capsid-coding sequence of this virus (O1K-A/VP1-^Q110^K/KGE/BHK2/CHO2) was determined, and we found that both the KGE motif and the K at VP1-110 had been retained and that no other residue changes were introduced during cell culture passage. The KGE virus infected pBTY, BHK, CHO, and HS-deficient CHO-677 cells (Table 3). Furthermore, infection of CHO cells was not inhibited by the GRGDSP peptide or the FMDV 17-mer peptides (data not shown). Together, the above results confirm that the introduction of K at VP1-110 creates a virus that does not require HS, the VP1 RGD motif, or αvβ5 or α5β1 integrins for infection of CHO cells.

Table 3.

Titers of infectious copy-derived viruses (A/Turkey/2/2006 capsid) with KK or KQ at VP1 109/110 and either KGD or KGE

Infectious copy plasmid and virus	VP1 109/110a	RGDb	Titer of virus in cell typec
			pBTY	BHK	CHO	CHO-677
pO1K-A/VP1-Q110K/KGE	K/K	KGE
O1K-A/VP1-Q110K/KGE/BHK2/CHO2	K/K	KGE	CPE	1.5 × 106	3.0 × 104	2.0 × 104
O1K-A/VP1-Q110K/KGD	K/K	KGD	CPE	3.0 × 106	9.0 × 103	5.0 × 103
pO1K-A/KGE	K/Q	KGE
O1K-A/KGD	K/Q	KGD	CPE	8.5 × 107	0	0

^aResidues at VP1 109 and 110.

^bResidues at the RGD motif site.

^c0, no CPE; CPE, cytopathic effect seen but virus titer not determined.

A positively charged residue at VP1-109 is also required for infection of CHO cells.

Figure 2 shows the positions of the residues at VP1-110. There is no structural information for the A/Turkey/2/2006 virus; therefore, the figure shows the location of VP1-110 modeled onto the structure of a related type A FMDV, A₁₀61 (28). This shows that VP1-110 is surface exposed and lies in a depression at the five-fold symmetry axis (Fig. 2). Interestingly, in A/Turkey/2/2006 the adjacent residue at VP1-109 is also exposed and occupied by K. Thus, the acquisition of K at VP1-110 creates a dense cluster of 10 positively charged residues that surround the pore at the five-fold symmetry axis. To determine if infection of CHO cells arises as a consequence of the VP1-^Q110^K substitution alone, or if the K at VP1-109 also contributes to this phenotype, we generated two further infectious copy plasmids with either A or Q at VP1-109 followed by K at VP1-110 (pO1K-A/VP1-¹⁰⁹AK¹¹⁰ and pO1K-A/VP1-¹⁰⁹QK¹¹⁰, respectively). Viruses rescued from these plasmids caused extensive CPE in BHK cells by the second or third passage, respectively (O1K-A/VP1-¹⁰⁹AK¹¹⁰/BHK2 and O1K-A/VP1-¹⁰⁹QK¹¹⁰/BHK3). The sequence of the capsid-coding region was determined and showed that the VP1 ¹⁰⁹AK¹¹⁰ or ¹⁰⁹QK¹¹⁰ sequences were faithfully retained and that no further residue changes were introduced during cell culture passage. Although these viruses were able to infect BHK cells, they were not infectious for CHO cells (Table 2). These observations suggest that a K is also required at VP1-109 for CHO cell infection. To support this conclusion, the above viruses were further passed through BHK cells to create BHK6 stocks. The virus with ¹⁰⁹AK¹¹⁰ was still noninfectious for CHO cells and retained the ¹⁰⁹AK¹¹⁰ motif, while the virus with ¹⁰⁹QK¹¹⁰ had gained the ability to infect CHO cells (Table 2) and showed a single residue change of Q to R at VP1-109, thereby creating an RK motif at VP1 109 and 110. Thus, it appears that R or K at VP1-109 in combination with K at VP1-110 is required for infection of CHO cells. In type O FMDV, a histidine-to-R switch at VP3-56 essentially creates an HS-binding site. Therefore, it is possible that the acquisition of R at VP1-109 may also create such a site. However, we eliminated this possibility by showing that the virus with ¹⁰⁹RK¹¹⁰ could infect CHO-677 cells, thereby showing that infection of CHO cells was independent of HS (Table 2).

Fig 2.

(A) The crystallographic structure of a pentamer of FMDV A10₆₁. The residues at VP1-109 and VP1-110 are highlighted in blue and red, respectively. Note that in A10₆₁ the residues at VP1-109 and VP1-110 are K and A, respectively. Also shown is a closer view of the pore at the five-fold symmetry axis, looking down at the outer surface of the pentamer (B) and from a side-on view (C). The residues at these sites are solvent exposed and line the depression surrounding the pore.

Lysine substitution at VP1-110 permits CHO cell infection by Asia-1 and type O FMDV.

The above results showed that when occupied by positively charged residues, VP1 109 and 110 create a dense patch of positive charge at the five-fold symmetry axis that appears to allow infection of CHO cells by a type A FMDV. Therefore, we investigated if the introduction of K at the same sites also conferred the ability to infect CHO cells upon infection with type O or Asia-1 FMDV. For this, we used infectious copy plasmids pO1K-OUK (37) and pO1K-Asia (see Materials and Methods), which encode the capsid-coding region (VP1, VP2, and VP3) of either the type O 2001 United Kingdom outbreak strain (O/UKG/34/2001) or an Asia-1 field isolate (Asia-1/BAR/9/2009), respectively, in the genetic background of O1K/B64. Similar to A/Turkey/2/2006, both parental viruses O/UKG34 and Asia-1/BAR/9/2009 and the viruses derived from the corresponding infectious copy plasmids pO1K-OUK and pO1K-Asia have K at VP1-109; thus, a K substitution at VP1-110 would create a ¹⁰⁹KK¹¹⁰ motif. Lysine substitutions were introduced at VP1-110 in pO1K-OUK and pO1K-Asia, generating pO1K-Asia/VP1-^Q110^K and pO1K-OUK/VP1-^A110^K, respectively. Viruses were rescued from the parental plasmids (pO1K-Asia and pO1K-OUK) and from pO1K-Asia/VP1-^Q110^K and pO1K-OUK/VP1-^A110^K. The Asia-1 viruses were propagated through BHK cells. As cell culture growth of type O FMDV can select viruses that bind HS, the type O viruses were initially passaged through pBTY cells (because they express αvβ6) and then BHK cells. The sequence of the capsid-coding region of the resulting viruses (O1K-Asia/BHK1, O1K-Asia/VP1-^Q110^K/BHK1, O1K-OUK/BTY2/BHK3, and O1K-OUK/VP1-^A110^K/BTY2/BHK2) were determined and showed that for all viruses the amino acid sequence encoded by the input vRNA was faithfully retained. The viruses with wild-type (wt) capsids (O1K-Asia/BHK1 and O1K-OUK/BTY2/BHK3) were not infectious for CHO cells, whereas the viruses with K at VP1-110 (O1K-Asia/VP1-^Q110^K/BHK1 and O1K-OUK/VP1-^A110^K/BTY2/BHK2) had gained the ability to infect CHO cells (Table 4). Thus, it appears that in addition to type A FMDV, the ability to infect CHO cells is transferable to type O and Asia-1 FMDV by the creation of a KK motif at VP1 109 and 110.

Table 4.

Titers of infectious copy-derived viruses (Asia-1 BAR/9/2009 and O/UKG/34/2001 capsids) with K/Q, K/A, or K/K, at VP1 109/110

Infectious copy plasmid and virus	VP1 109/110a	Titer of virus in cell typeb
		BHK	CHO
pO1K-Asia	K/Q
O1K-Asia/BHK-1	K/Q	2.5 × 107	0
pO1K-Asia/VP1-Q110K	K/K
O1K-Asia/VP1-Q110K/BHK-1	K/K	1.5 × 107	1 × 104
pO1K-OUK	K/A
O1K-OUK/BTY2/BHK3	K/A	4 × 106	0
pO1K-OUK/VP1-A110K	K/K
O1K-OUK/VP1-A110K/BTY2/BHK2	K/K	2 × 106	7.5 × 104

^aResidues at VP1 109 and 110.

^b0, no CPE.

Influence of the VP1-^Q110^K substitution on use of integrin αvβ6.

The KGD virus recovered from O1K-A/VP1-^Q110^K/KGE/BHK3 was named O1K-A/VP1-^Q110^K/KGD, and as expected it was infectious for BHK, CHO, and CHO-677 cells (Table 3). We also attempted to recover a virus with the RGD-to-KGE change by using pO1K-A (which has the wt Q at VP1-110) on BHK cells. The first and second passages did not show signs of CPE, whereas the third and fourth passages resulted in extensive CPE within 24 h. The capsid-coding sequences of the BHK3 and BHK4 viruses were determined. Similar to the virus with K at VP1-110 (see above), the BHK3 virus showed only a single residue difference, which resulted in a partial reversion at the RGD site to create a KGD motif. This virus was named O1K-A/KGD, and as expected for a virus with Q at VP1-110, it was not infectious for CHO or HS-deficient CHO-677 cells (Table 3). The virus recovered at the fourth passage through BHK cells showed a further residue change (K to R) that restored the RGD domain. As described above, we recovered virus with KGE in place of the RGD when K was present at VP1-110 when we used CHO cells. Presumably, this was made possible by the lack of selective pressure to restore the RGD due to the absence of appropriate integrin receptors on CHO cells. Therefore, we also attempted to recover virus with KGE and Q at VP1-110 by using O1K-A/KGE/BHK2 (which did not show signs of CPE on BHK cells) and CHO cells. However, after four passes we did not see CPE. This was repeated with the same outcome, and infectious virus was not recovered. These observations are consistent with previous reports that replacement of the RGD by KGE renders viruses with capsids of FMDV field isolates noninfectious (26, 27, 43, 44, 51). Together, the above observations show that we could recover infectious virus with KGD at the RGD site when either Q or K were present at VP1-110, but we could only recover a KGE virus if VP1-110 was occupied by K.

Next, we investigated the receptors that mediate infection by the KGD and KGE viruses. The observation that the viruses with KGD or KGE infect BHK cells to a higher titer than CHO or CHO-677 cells suggests they may retain the ability to use integrins as receptors despite not having a complete canonical RGD integrin-binding motif. To determine if this is the case, we could not use BHK cells, as cross-reactive antibodies for hamster integrins are limited and the integrins expressed on BHK cells have not been established. Instead we used pBTY cells, as FMDV is known to use αvβ6 as the major receptor to initiate infection (16). Primary BTY cells were incubated with RGD-containing peptides and then infected with FMDV (with RGD, KGD, or KGE) in the continued presence of the peptide for 1 h. The cells were washed to remove peptide and extracellular virus, and after a further 3 h, infection was quantified as described in Materials and Methods. Consistent with our previous observations (16), Fig. 3A shows that the FMDV 17-mer peptide inhibited infection by the virus with the wt capsid (O1K-A/BTY1/BHK1, which has an RGD and Q at VP1-110). The virus with an RGD motif and K at VP1-110 (O1K-A/VP1-^Q110^K/BHK1) was also inhibited by the peptide. However, infection by the VP1-^Q110^K-substituted virus was inhibited less efficiently than virus with the wt capsid. In addition, Fig. 3A shows that infection by the viruses with KGD in place of the RGD (with either Q or K at VP1-110) was inhibited by the peptide, suggesting that these viruses also use an RGD-dependent integrin as a receptor. We also found that the peptide was a very efficient inhibitor of infection by the virus with KGE (O1K-A/VP1-^Q110^K/KGE/BHK2/CHO2). At peptide concentrations of >0.25 μM, infection by the KGE virus was completely inhibited.

Fig 3.

(A) pBTY cells in 96-well plates were mock treated or treated with the FMDV 17-mer peptide (or RGE control) for 0.5 h and then infected with FMDV (O1K-A/BTY1/BHK1, O1K-A/KGD, O1K-A/VP1-^Q110^K/BHK1, O1K-A/VP1-^Q110^K/KGD, or O1K-A/VP1-^Q110^K/KGE/BHK2/CHO2) for 1 h at an MOI of 0.3 in the absence or presence of peptide. (B) pBTY cells in 96-well plates were mock treated or treated with antibodies at the indicated concentrations for 1 h and then infected with O1K-A/BTY1/BHK1, O1K-A/KGD, O1K-A/VP1-^Q110^K/BHK1, or O1K-A/VP1-^Q110^K/KGD for 1 h at an MOI of 0.3 in the absence or presence of antibody (6.8G6, αvβ6; 23C6, αvβ3). (C) pBTY cells in 96-well plates were mock treated or treated with MAb 6.8G6 for 1 h and then infected with O1K-A/BTY1/BHK1 or O1K-A/VP1-^Q110^K/KGE/BHK2/CHO2 for 1 h at an MOI of 0.3 in the absence or presence of antibody. At the end of each experiment, the monolayers were washed and infection was continued for a further 3 h. Infection was stopped, the cells fixed and permeabilized, and infected cells were identified as described in Materials and Methods. The results were normalized to the results with the mock-treated controls. Shown are the means ± standard deviations for triplicate wells. Each panel shows one experiment representative of two conducted, each giving similar results. Statistical significance was analyzed using an unpaired one-tailed t test with a P value less than 0.05 taken to be significant. *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001.

The above observations showed an apparent reduced sensitivity to peptide competition by the virus with an RGD motif and the VP1-^Q110^K substitution. This finding could result from the ability of the virus to use alternative receptors due to the presence of the K at VP1-110. Alternatively, this resistance could be explained if the K at VP1-110 had a positive influence on FMDV-integrin interactions. If the former were the case, the KGE, which also has the VP1-^Q110^K substitution, would be expected to be similarly resistant to the peptide. However, under the conditions tested, infection by these viruses was completely inhibited, suggesting that the resistance to peptide competition results from a more efficient use of integrin receptors. Together, these results indicate that the primary receptor used by the VP1-^Q110^K-substituted virus (O1K-A/VP1-^Q110^K/BHK1) to infect pBTY cells is most likely an integrin and that the VP1-^Q110^K substitution has a positive influence on integrin use.

We showed previously that αvβ6 is the primary integrin receptor used by FMDV to infect pBTY cells (16). Therefore, the above observations suggest that viruses with KGD or KGE may also use αvβ6 as a receptor to initiate infection. To investigate this, we carried out further competition experiments using function-blocking antibodies (in place of the peptides) that were cross-reactive for bovine integrins αvβ6 and αvβ3 (Fig. 3B). Similar to the peptide study above, viruses with an intact RGD and either Q or K at VP1-110 were included in these studies. An antibody to αvβ6, but not one to αvβ3, inhibited infection of pBTY cells by the viruses with an intact RGD. As seen with the peptide, infection by the VP1-^Q110^K-substituted virus (O1K-A/VP1-^Q110^K/BHK1) was inhibited less efficiently by the αvβ6 antibody than infection by the virus with Q at this site (O1K-A/BTY1/BHK1). This observation supports our conclusion that K at VP1-110 appears to enhance the ability of FMDV to use αvβ6 as a receptor. The antibody against αvβ6, but not that against αvβ3, also inhibited infection by viruses with KGD or KGE, which confirmed that they use αvβ6 as the receptor. Infection by the virus with KGE was inhibited efficiently (>95%) by the antibody to αvβ6 when used at 20 μg/ml (data not shown). To confirm this observation, Fig. 3C shows the results of a further competition experiment in which we used a range of concentrations of the αvβ6 antibody. These results showed that infection by the KGE virus was inhibited by the anti-αvβ6 antibody, but not by the anti-αvβ3 antibody. Thus, when K is present at VP1-110, FMDV with KGE appears to retain the ability to use αvβ6 as its receptor on pBTY cells, albeit with an apparent reduced affinity.

DISCUSSION

Adaptation to cell culture is an essential first step in the development of new vaccines, and for FMDV this often involves selection of variants with altered receptor specificity that are no longer dependent upon integrins for infection. This is reflected by a strong predisposition toward selection of variants that bind HS and infect CHO cells. However, cell culture-adapted viruses have also been reported that use receptors that are neither integrin nor HS (26, 27, 36). Here, we have shown that a single amino acid substitution at VP1-110 (VP1-^Q110^K) allows for HS-independent infection of CHO cells. This was initially shown when we used a recombinant virus with the capsid of the A/Turkey/2/2006 field isolate, and it was subsequently confirmed when we used viruses with capsids derived from field isolates of type O and Asia-1 FMDV. CHO cells express two RGD-binding integrins (α5β1 and αvβ5) that are not normally used by FMDV as receptors. Here we have shown that viruses with the VP1-^Q110^K substitution do not acquire the ability to use these integrins as receptors, thereby confirming that infection of CHO cells does not require integrins. However, despite showing that infection of CHO cells by the VP1-^Q110^K-substituted virus was integrin independent, the primary receptor on pBTY cells was shown to be the integrin αvβ6, as at a low MOI, the infection by the VP1-^Q110^K-substituted virus was inhibited to near completion by an RGD peptide (Fig. 3). The VP1-^Q110^K substitution also appeared to allow for enhanced interactions with integrin αvβ6, which allowed a virus with KGE in place of the normal RGD integrin-binding motif to use this integrin as a receptor.

The A/turkey/2/2006 virus naturally has K at VP1-109. Consequently, the introduction of K at the adjacent residue creates a dense patch of positive charge that surrounds the five-fold symmetry axis. Interestingly, the ability to infect CHO cells was lost if the K at VP1-109 was replaced by A, despite retaining the K at VP1-110, showing that both of the K residues at VP1 109 and 110 are required for infection of CHO cells. However, when the K at VP1-109 was replaced by Q (creating ¹⁰⁹Q/K¹¹⁰), virus infectious for CHO cells was recovered, but this virus had acquired an R in place of the Q at VP1-109, thereby creating a ¹⁰⁹R/K¹¹⁰ motif. Thus, the ability to infect CHO cells appears to require positively charged residues at both VP1 109 and 110 and is tolerant to either R or K at VP1-109. The attachment receptor on CHO cells for the VP1-^Q110^K-substituted virus is not known. However, the clustering of positive residues at the five-fold symmetry axis could mediate cell attachment through interactions with negatively charged molecules at the cell surface. The major negatively charged moieties at the cell surface are sialic acid and glycosaminoglycans (GAGs) such as HS and CS. However, our results suggest that none of these molecules serves as a virus attachment receptor, as the VP1-^Q110^K-substituted virus was infectious for CHO-677 cells, which lack HS, CHO-745 cells, which lack HS and CS, and CHO-lec2 cells, which lack sialic acid. However, given the inherent tolerance to either K or R at VP1-109, it is possible that the receptor interaction mediated by the residues at VP1 109 and 110 lacks specificity and that more than one type of negatively charged molecule could be used; thus, a lack of only one type of molecule may not be sufficient to prevent infection.

As stated above, the VP1-^Q110^K substitution appeared to enhance virus interactions with αvβ6. The main evidence to support this conclusion came from the observations that αvβ6 antagonists (RGD peptide and anti-αvβ6 antibody) were less effective inhibitors of infection by the VP1-^Q110^K-substituted virus than virus with the wt capsid (i.e., with Q at VP1-110) and from the observations that a virus with KGE at the RGD site could use αvβ6 as a receptor if K was present at VP1-110. The VP1-^Q110^K substitution could lead to enhanced interactions with αvβ6 if the presence of K altered the conformation of the VP1 GH loop (the integrin-binding loop) to one more favorable for integrin binding. However, we think this unlikely, as viruses with the VP1-^Q110^K substitution did not gain the ability to use αvβ5 or α5β1 as receptors on CHO cells. Alternatively, cell attachment of the VP1-^Q110^K-substituted virus could serve to effectively concentrate virus at the cell surface, making recruitment and subsequent virus endocytosis by αvβ6 more likely. This could explain why αvβ6 is the predominant receptor on pBTY cells and the relatively low infectivity of VP1-^Q110^K-substituted viruses for CHO cells (Table 2), because in the absence of αvβ6, viruses held at the cell surface may be internalized inefficiently.

Our results show that viruses with KGD in place of the RGD can use αvβ6 as a receptor. It is highly likely that this results from binding of the KGD motif at the normal RGD-binding site on integrin. This conclusion is supported by the observations that infection of pBTY cells with the KGD viruses was inhibited by antagonists that specifically target the RGD-binding site on the integrin, i.e., an RGD-containing peptide and a function-blocking antibody that acts as a ligand mimetic due to an RGD motif in complementarity determining region H3 (40). The conclusion that the KGD viruses interact with αvβ6 at the RGD-binding site is further supported by reports that ligands containing a DLXXL motif can bind αvβ6 in the absence of a full RGD (24). The integrin-binding loop of FMDV A/Turkey/2/2006 includes the sequence RGDLGPL, and therefore the KGD viruses could use αvβ6 due to the presence of the DLGPL sequence. Other viable FMDVs have been described with mutations in the RGD (28, 52, 53), and some of these retain the D of the RGD motif and the conserved residues normally found after the RGD (see the introduction). For most of these viruses, integrin binding has not been investigated; however, a type A virus described by Rieder et al. (52) with SGD was shown to preferentially infect cells transfected to express αvβ6, implying that this integrin may be used as a receptor. The SGD virus used in these studies had an M (and not L) immediately after the RGD, which suggests that the DMXXL motif may also bind αvβ6. This conclusion is in agreement with our studies, which have shown that peptides with L, M, or R immediately after the RGD are equally effective as αvβ6 antagonists (16).

Our results also showed that a virus with KGE in place of the RGD can use αvβ6 as a receptor. This was somewhat unexpected, as previous observations showed that FMDV is rendered noninfectious by the switch from RGD to KGE (26, 27, 43, 44, 51). However, these observations were made using viruses with capsids derived from field isolates, which rely solely on integrins for infection. Indeed, our inability to recover infectious virus with KGE in the context of the wt capsid (i.e., when VP1 110 is Q) is consistent with these observations. Thus, it appears that the presence of K at VP1-110 facilitates both recovery of the KGE virus using CHO cells and the ability to use αvβ6 to initiate infection of pBTY cells. Previously, we proposed a two-step model for FMDV binding to αvβ6, in which the initial interaction by the RGD was rapidly stabilized by a second synergistic interaction (16, 18). This synergistic interaction is dependent upon the residues at the RGD +1 and RGD +4 sites and the integrity of the helical structure immediately after the RGD that orientates the RGD +1 and RGD +4 residues into positions available for integrin recognition (16, 18, 25). Based upon the arguments presented above for the KGD viruses, we believe that KGE viruses are also likely to bind at the normal RGD attachment site on αvβ6, as infection was also inhibited by the RGD-containing peptide and anti-αvβ6 antibody; however, the binding affinity for αvβ6 would be expected to be lowered, as infection by the KGE virus was shown to be exquisitely sensitive to competition by these reagents. A type O FMDV with KGE in place of the RGD has been described previously (36). This virus also had mutations in VP1 that included the acquisition of aromatic amino acids at positions 108 and 174 and positively charged residues at positions 83 and 172 that mapped close to the five-fold symmetry axis; however, this virus was shown not to use integrin receptors, as infection was resistant to an RGD-containing peptide. More recent studies using SAT serotype FMDVs have described viruses that acquired positively charged residues at VP1 110 and 112 during cell culture adaptation (31, 32). However, in contrast to our results, these changes were reported to create an HS-binding site. Similar observations have also been reported for coxsackievirus A9 and some other members of the human enterovirus B species, as the presence of positively charged residues at VP1-132 has been shown to create an HS-binding site at the five-fold symmetry axis that allows for an expanded tropism for cultured cells (54).

In conclusion, we have identified a single amino acid change (VP1-110, glutamine to lysine) at the five-fold symmetry axis of the FMDV capsid that allows for integrin- and HS-independent infection of CHO cells. The same change resulted in enhanced interactions with αvβ6 in cells expressing this integrin, which allowed a virus with KGE in place of the integrin-binding RGD to use αvβ6 as a receptor. The introduction of positively charged residues at VP1 109 and 110 may allow for cell culture adaptation of FMDV by design, which may prove useful for vaccine manufacture when cell culture adaptation proves intractable.