Genetic Determinants of Altered Virulence of Taiwanese Foot-and-Mouth Disease Virus

Recent FMD outbreak in Taiwan.

In 1997, a devastating and unusual outbreak of FMD, characterized by disease in swine but not in cattle, occurred in Taiwan. The outbreak, which was caused by a serotype O virus (OTai) with an atypical porcinophilic phenotype (6), rapidly developed into a massive epizootic, resulting in cessation of export of all pork products from the country. This outbreak devastated the Taiwanese swine industry (approximately 4 million swine were destroyed) and had a severe impact on the national economy due to costs of control and trade restrictions (estimated at over 6 billion U.S. dollars). During the course of this outbreak, no cattle were reported to have been affected, and virus isolated from infected swine was unable to infect bovine thyroid cells in vitro or to cause typical disease in bovines following intradermal inoculation in the tongue (6).

Genomic regions responsible for host range specificity in vitro.

To define the genetic components of OTai responsible for its porcinophilic phenotype, we have generated and evaluated recombinant viruses with sequences of OTai substituted for sequences of an FMDV strain of high virulence in bovines (25). The engineered chimeric viruses produced for these studies are derived from in vitro-generated RNA transcripts of genome-length cDNAs by using standard techniques (17, 23). Specifically, all chimeric viruses used in these studies were generated by high-efficiency transfection of BHK cells (17), and experiments were performed with low-passage (less than five) stocks of virus produced by high-multiplicity infection of BHK cell cultures. Figure 1 shows schematic diagrams of the genomes of selected chimeric viruses along with photographs of plaques formed by these viruses in monolayer cultures of BHK or bovine kidney (BK) cells. All viruses that formed plaques on BHK cells also formed plaques on both IB-RS-2 cells (a porcine kidney cell line [6]) and porcine kidney cells (data not shown).

The virus whose structure is shown first in Fig. 1, vOTaiCn-An, is derived from a full-length cDNA containing the entire OTai protein-encoding sequence. This chimera contains the A12 S fragment and poly(C) tract found in all of the viruses used in this study (23), with the remainder of the genome [poly(C) tract to poly(A) tail] derived by PCR amplification of cDNA reverse transcribed from OTai virus RNA extracted from lesions of swine infected with a field isolate of OTai (see legend to Fig. 1). As expected, vOTaiCn-An displays the growth restriction of field isolates of OTai on bovine cells (6). Immunostaining of BK cells infected with vOTaiCn-An showed that this virus could establish a limited infection in bovine cells, as indicated by few FMDV antigen-positive cells and no spread of infection to surrounding cells (data not shown). The virus whose structure is shown second in Fig. 1, vO1CamP1, is a chimeric virus previously generated (referred to as vCRM8 in references 1, 20, and 25) by inserting the capsid-encoding region (proteins 1A, 1B, 1C, and 1D) of serotype O1 Campos (from a 1958 bovine isolate from Brazil) in the genetic background of a high-passage serotype A12 virus (from a 1932 bovine isolate from the United Kingdom) (23, 25). vO1CamP1 displays a high level of virulence in cattle (25) and swine (1) and efficiently forms plaques in BHK and BK cell monolayers (Fig. 1). The virus whose structure is shown after vO1CamP1 in Fig. 1, vOTaiP1, encodes the surface-exposed capsid proteins (1B, 1C, and 1D) of OTai in the background of the serotype A12 virus. Interestingly, this virus, whose capsid differs significantly from the South American O1 capsid (Table 1), displays plaquing ability similar to that of vO1CamP1 on both BHK and BK cells (Fig. 1). These findings indicate that the basis of growth restriction in BK cells in vitro was not due to alterations in the capsid, although our previous work has shown a profound influence of FMDV capsid sequences on receptor specificity and tropism in cell culture and virulence in animals (20, 25). Furthermore, results obtained by evaluation of other chimeras failed to show a role for either the OTai Leader proteinase coding region (known to contribute to virus virulence in bovines [2, 18]) or the OTai translation initiation site (the internal ribosome entry site region, well known for its role in determining the virulence of the related picornavirus that causes poliomyelitis [reviewed in reference 28]) in restricting replication in bovine cell cultures (results not shown).

To further map the basis of the altered host range of OTai, a series of viruses were generated to dissect the nonstructural protein-encoding regions (P2 and P3). Sequence data obtained during the construction of these chimeras revealed a significant difference among the 3A coding regions of the OTai, A12, and O1 Kaufbeuren (O1K) viruses (Table 2). The comparisons in Table 2 show that the dramatic differences in the 3A coding regions of the two type O genomes stand in sharp contrast to the modest differences detected between coding regions for 2B, 2C, 3B, 3C, and 3D between these viruses, which are similar to the differences seen between the P2 and P3 coding regions of O1K and the A12 virus.

To test the importance of the 3A region in the bovine growth restriction of OTai, three additional chimeras, shown at the bottom of Fig. 1, were constructed. These three viruses contained either the entire A12 3A coding region or the mutated portion of 3A of OTai (see below), along with some additional regions. In all three chimeras shown at the bottom of Fig. 1, the exchanged regions extend beyond the 3A coding region, and in two of the cases, the first 24 codons of the OTai 3A coding region were not included in the chimeras (the amino acids encoded by A12 and OTai are identical in this 24-codon region) (see Fig. 2). However, in the extensions beyond the borders of 3A, the substituted codons correspond to a small number of well-spaced single amino acid substitutions that are largely conservative in nature (in the NcoI fragment exchanged to form virus vOTaiCn-An/A123A-C [Fig. 1], a total of 3 substitutions in 2C, 5 substitutions in 3BBB, 12 changes in 3C, and 5 changes in 3D were incorporated along with 40 changes clustered in the latter half of 3A; in the EcoRI-EcoRV fragment exchanged to construct vO1CamP1/Tai3A or vOTaiP1/OTai3A [see Fig. 1], 5 changes in 3BBB and 1 change in 3C were incorporated in addition to the 40 changes clustered in the latter half of 3A). Thus, the only significant changes between the last three viruses in Fig. 1 and their corresponding parents are in the 3A coding region. A comparison of the abilities of these three viruses to form plaques on BK cells revealed that substitution of the OTai 3A coding region into vO1CamP1 or vOTaiP1 produced viruses that were unable to form plaques on BK cells, whereas the substitution of the A12 3A into O1TaiCn-An produced a virus capable of forming plaques on these cells (Fig. 1). The plaques formed on BK cells by this latter virus were smaller than those produced by vOTaiP1 and vO1CamP1, suggesting that other regions of the genome contribute to OTai growth restriction on BK cells. Taken together, the data in Fig. 1 demonstrate that the OTai 3A coding region is the primary determinant of the growth restriction of OTai on BK cells.

As pointed out in Table 2, sequence analyses of the genome of OTai revealed a high degree of identity to O1 and A12, except for the 3A coding region. The divergence in 3A sequences among these viruses reflects two dramatic differences in the OTai 3A coding region. The first is a 10-amino-acid deletion corresponding to codons 93 through 102 of the European and South American 3A coding regions, and the second is a large number of mutation located between codons 128 and 147 of the European and South American 3A coding regions. Interestingly, the deletion in the OTai 3A is similar to a deletion found in egg-passaged derivatives of O1 Campos (Fig. 2) and C3 Resende (9). Both of these egg-passaged viruses, which were developed and used as live-attenuated FMD vaccines in bovines in South America (21, 26), exhibited a greatly reduced ability to form plaques on BK cells, and the C3 derivative was reported to have maintained its virulence in swine (26). Evaluation of selected other 3A sequences (C3 Argentina, 1985 (7); SAT-2, Kenya, 1957 [J. W. I. Newman, A. M. Q. King, and S. Ortlep, personal communication]; and sequences obtained from PCR-amplified cDNAs of several other serotypes or subtypes of FMDV [Asia1, Lebanon, 1983; O11, Indonesia, 1962]) showed that none of these isolates contained the deleted or mutated segments identified in OTai.

Genomic regions responsible for host range specificity in vivo.

In vivo studies were conducted to establish the role of OTai 3A as a determinant of bovine virulence. vO1CamP1, which encodes the typical, full-length 3A of serotype A12, has been shown in previous studies to very efficiently cause vesicle formation on bovine tongues following intradermal inoculation, with a 50% infectious dose of 5 to 50 PFU (25). Furthermore, the animals inoculated with vO1CamP1 developed a systemic infection (25). A similar intradermal inoculation study performed with vOTaiCn-An in bovine 07 showed a significantly different outcome (Fig. 3A). In this case, lesions were detected only at doses of 750,000 and 75,000 PFU, and the quality of the lesions was significantly different than that observed with vO1CamP1. Specifically, the lesions caused by the high doses of vOTaiCn-An did not progress into sores denuded of epidermis, which were characteristic of the vesicles formed with all inoculations of over 50 PFU of vO1CamP1 (or other cattle-virulent FMDV). Moreover, animal 07 did not display lameness or vesicles on any feet during the 3-week observation period following inoculation. Thus, our results with the genetically engineered form of OTai are consistent with those obtained with the field isolate (6).

Inoculation of bovine 08 with vO1TaiCn-An/A123A-C demonstrated that the addition of the full-length, wild-type form of 3A to this OTai derivative produced a bovine-virulent virus (Fig. 3B). Forty-eight hours after tongue inoculation with this virus, vesicles were detected at all doses greater than 6 PFU, showing that this virus was highly efficient in establishing infectious foci in the epidermis of the tongue (Fig. 3B). Interestingly, bovine 08 did not display other signs of systemic disease, including pedal lesions or lameness, over the 3-week observation period.

Addition of the OTai 3A coding sequence to vO1CamP1 was capable of attenuating the infection caused by this virus. Bovine 210 was inoculated with graded dilutions of vO1CamP1/OTai3A and observed for 3 weeks. At 48 h, lesions were detected at the 260-PFU dose, but these lesions consisted of white swollen vesicles rather than the severe erosions detected with vO1CamP1 and vO1TaiCn-An/A123A-C (Fig. 3C). However, the following day, erosions appeared (although only at doses greater than 260 PFU), and this animal showed a systemic disease, characterized by spread of the infection to the feet. Infection of a second bovine with dilutions of vO1CamP1/OTai3A produced similar signs of disease (results not shown).

To confirm that our observations were not significantly influenced by individual responses of animals to inoculation, a single bovine was inoculated with two doses each of four different viruses. Bovine 101 was inoculated with vO1CamP1, vO1CamP1/OTai3A, vOTaiP1, and vOTaiP1/OTai3A (Fig. 3D) at doses of 40,000 or 40 PFU. Both vO1CamP1 and vOTaiP1 were able to cause severe lesions at high and low doses (Fig. 3D). vO1CamP1/OTai3A was able to cause swelling at the high dose, but no lesions were detected at the low dose. vOTaiP1/OTai3A was unable to cause signs of infection at either dose (Fig. 3D). Taken together, these in vivo data demonstrate that the mutated 3A coding region of OTai plays a major role in the attenuation of this virus in bovines.

Although FMDV can be distinguished from other members of the Picornaviridae by a much longer 3A protein (153 amino acids versus 87 for poliovirus), all picornavirus 3A proteins contain a 15- to 20-amino-acid hydrophobic domain approximately 60 to 80 residues from their N termini that is thought to function in binding the viral RNA replication machinery to cellular membranes (29). Furthermore, the interaction of 3A with cellular membranes has been shown to cause a cytopathic effect, prevent surface expression of proteins, and to inhibit protein secretion from 3A-expressing cells (3, 4). Thus, changes in the interaction of 3A with the host cell could affect the outcome of infection by directly altering RNA replication or by altering the ability of the cell or the host to mount a response to infection at the level of cytokine secretion or display of viral antigens in the context of the major histocompatibility complex. In the case of hepatitis A virus, changes in 3A have been documented during adaptation to cell culture (10, 11, 15, 19), and for poliovirus there is molecular genetic evidence that changes in 3A alter host range in vitro (14).

The activities of 3A in virus-infected cells, and their implication in alteration of host range of other picornaviruses, support a role for OTai 3A in the altered virulence of OTai. Moreover, the changes we have detected in OTai, including a deletion similar to those previously reported for egg-adapted FMDVs (9), suggest that under some circumstances altered 3A genes could have a selective advantage. Determining if these changes contribute to the high level of virulence of OTai in swine, and discovering the time of appearance of these changes in field isolates of FMDV, will contribute to our understanding of how a bovine-attenuated FMDV emerged to cause this recent devastating outbreak in swine.