Comparative Genomics of Foot-and-Mouth Disease Virus

Abstract

Here we present complete genome sequences, including a comparative analysis, of 103 isolates of foot-and-mouth disease virus (FMDV) representing all seven serotypes and including the first complete sequences of the SAT1 and SAT3 genomes. The data reveal novel highly conserved genomic regions, indicating functional constraints for variability as well as novel viral genomic motifs with likely biological relevance. Previously undescribed invariant motifs were identified in the 5′ and 3′ untranslated regions (UTR), as was tolerance for insertions/deletions in the 5′ UTR. Fifty-eight percent of the amino acids encoded by FMDV isolates are invariant, suggesting that these residues are critical for virus biology. Novel, conserved sequence motifs with likely functional significance were identified within proteins L^pro, 1B, 1D, and 3C. An analysis of the complete FMDV genomes indicated phylogenetic incongruities between different genomic regions which were suggestive of interserotypic recombination. Additionally, a novel SAT virus lineage containing nonstructural protein-encoding regions distinct from other SAT and Euroasiatic lineages was identified. Insights into viral RNA sequence conservation and variability and genetic diversity in nature will likely impact our understanding of FMDV infections, host range, and transmission.

Foot-and-mouth disease (FMD) is an acute, systemic disease of domestic and wild cloven-hooved animal species and is caused by Foot-and-mouth disease virus (FMDV), the prototype member of the genus Aphthovirus of the family Picornaviridae. FMDV occurs as seven distinct serotypes (Euroasiatic serotypes A, O, C, and Asia1 and South African Territories [SAT] serotypes SAT1, SAT2, and SAT3) and multiple subtypes reflecting significant genetic variability. The highly contagious nature of FMDV and the associated productivity losses make it a primary animal health concern worldwide. Effective vaccines and stringent control measures have enabled FMD eradication in most developed countries, which maintain unvaccinated, seronegative herds in compliance with strict international trade policies. However, the disease remains enzootic in many regions of the world, posing a serious problem for commercial trade with FMD-free countries.

FMDV is a small nonenveloped virus with a pseudo T=3 icosahedral capsid made up of 60 copies each of four structural proteins. The capsid surrounds an 8.4-kilobase, positive-sense, single-stranded RNA genome which is covalently bound at its 5′ end to the small viral protein 3B (or VPg) and is polyadenylated at its 3′ end. Upon virus entry into a cell, the viral genome is rapidly translated into a polyprotein which is co- and posttranslationally cleaved by viral proteinases into several partially cleaved, likely functional, intermediates and ultimately into 12 mature proteins (87, 97).

The FMDV genome organization is similar to that of other picornaviruses, including a large single open reading frame (ORF) flanked by highly structured 5′ and 3′ untranslated regions (5′ UTR and 3′ UTR, respectively) (Fig. 1A). The 5′ UTR consists of, from the 5′ end, a 350- to 380-nucleotide (nt) “short” (S) fragment, a 100- to 420-nt poly(C) tract (90% C), and the approximately 700-nt 5′ terminus of the genomic “long” (L) fragment, which contains three or four tandemly repeated pseudoknots, a stem-loop cis-acting replication element (cre), and a type II internal ribosome entry site (IRES) (69). The FMDV 5′ UTR plays important roles in cap-independent translation initiation of the viral polyprotein and in viral genome replication (64). The 3′ UTR is about 90 nt long and is thought to contain cis-acting elements required for efficient genome replication (4).

FIG. 1.

FMDV UTR variability. (A) FMDV S and L genomic fragments showing the locations and predicted secondary structures of UTRs. Cn, poly(C) tract; AUG, start codons; pAn, poly(A); I, II, and III, S fragment, IRES, and 3′ UTR, respectively. (B) Rates of substitution per nt between positions 1 and 1218 and positions 8277 and 8441 in a genomic ClustalW alignment. Arrows with numbers indicate positions with high rates of nt substitution. (C) 5′ UTR nt conservation between positions 1 and 1153 for the 103 FMDV isolates examined in this study. Motifs previously defined as conserved between picornavirus IRESs are underlined. Boxes indicate primers used for amplification. Asterisks indicate nt positions tolerant to insertions/deletions in two or more isolates. (D) Nucleotide conservation within S fragment (I), IRES (II), and 3′ UTR (III) predicted secondary structures. Arrows indicate regions used for PCR amplification. UTR regions from the isolate C5Argentina/69 were used to generate secondary structure predictions and plots with mfold and Squiggles. Black circles indicate invariant nt among the 103 isolates presented here; gray circles represent nt that are conserved in at least 100 of 103 isolates (96%); gray shading indicates residues comprising previously described motifs. Nucleotide position numbers correspond to the consensus sequence generated by a ClustalW alignment of the 103 isolates.

The four capsid proteins, 1A, 1B, 1C, and 1D (also known as VP4, VP2, VP3, and VP1, respectively), are encoded by the N-terminal half of the ORF, and with the exception of 1A, which is excluded from the virion surface, are involved in antigenicity and binding to a subset of RGD-dependent integrins and heparan sulfate proteoglycan receptors on the cell surface (reviewed in reference 45). Nonstructural proteins represent about two-thirds of the ORF-encoded proteins and include L^pro, 2A, 2B, 2C, 3A, 3B, 3C^pro, and 3D^pol (93). FMDV polyprotein processing is mediated by L^pro, 3C^pro, and 2A. L^pro is a papain-like protease that, in addition to excising itself from the polyprotein, cleaves the cellular translation initiation factor eIF4G, resulting in a shutoff of host cap-dependent translation (65). 3C^pro, a member of the trypsin family of serine proteinases, performs all but three of the cleavages leading to mature viral proteins and also cleaves host cell proteins (111). FMDV 2A mediates autocleavage at its C terminus, apparently by inducing a ribosomal skip during polyprotein synthesis (19, 20). Although the functions of the FMDV 2B and 2C proteins are unknown, preliminary work suggests that, similar to those of other picornaviruses, they localize to endoplasmic reticulum (ER)-derived vesicles, the sites of viral genome replication (106). 3A is thought to be a multifunctional integral membrane protein that enhances viral RNA synthesis by 3D^pol and stimulates cleavage of the 3CD precursor (91). FMDV encodes three nonidentical copies of genome-linked 3B, a protein required for viral RNA replication (26). The 3D gene encodes the viral RNA-dependent RNA polymerase, and it and 3A colocalize with ER membrane-associated replication complexes (80, 92).

Despite a basic understanding of many aspects of picornaviral biology, much information regarding FMDV UTR, protein, and protein precursor functions and roles in virulence, host range, and virus transmission remains poorly understood. Comparative genomic analysis of a large number of FMDV genomes may allow the identification of highly conserved regulatory or coding regions which are critical for aspects of virus biology. To date, the few comparative analyses of full-length FMDV genomes have relied on intratypic (serotype O) or intertypic (serotypes O, A, and C) comparisons of a small number of isolates and serotypes (70, 86). In addition, complete genomic sequences of some serotypes are not available (SAT1 and SAT3) or, for others, only represent highly cell culture-adapted isolates (serotypes A and C). Studies involving large numbers of virus isolates have largely focused on genomic regions encoding structural proteins associated with serotype specificity for phylogenetic purposes (reviewed in reference 55).

For this work, we obtained and analyzed 103 complete genome sequences representing all FMDV serotypes, including the previously unavailable SAT1 and SAT3 genomes. Our analyses identified novel highly conserved genomic regions indicating functional constraints for variability, novel viral genomic motifs with likely biological relevance, and previously undescribed virus lineages.

MATERIALS AND METHODS

FMDV isolates.

One hundred three FMDV isolates were obtained from the Animal Plant Health Inspection Service, United States Department of Agriculture (Table 1). Except for recent isolates from Argentina, the United Kingdom, and South Korea (2000-2001), most isolates belonged to the Animal Plant Health Inspection Service archive collection and represented a broad temporal and geographical distribution (Table 1). Although certain archive isolates have been adapted to guinea pigs or passaged in cell cultures, isolates obtained from their natural hosts were well represented. For some analyses, complete genome or whole polyprotein FMDV sequences currently available in GenBank release 141 were considered and included (Table 1).

TABLE 1.

FMDV sequences

Sequence no.c	Isolate namea	Country, year of isolation	Passage historyb	Accession no.
1	A25 Argentina/59	Argentina, 1959	GP, 1 LK	AY593769
2	A Canefa 1/61	Argentina, 1961	BOV 2 LK	AY593789
3	A24 Argentina/65	Argentina, 1965	BOV, 3 LK	AY593767
4	A26 Argentina/66*	Argentina, 1966	8 GP	AY593770
5	A Argentina/00	Argentina, 2000	BOV	AY593782
6	A Argentina/01 Areco	Argentina, 2001	BOV	AY593783
7	A ARGp55/01	Argentina, 2001	BHK-21 passage 55	AY593784
8	A ARGp64/01	Argentina, 2001	BHK-21, passage 64	AY593785
9	A Trenquelauquen/01	Argentina, 2001	BOV	AY593786
10	A General Lopez/01	Argentina, 2001	BOV	AY593790
11	A Uruguay/01	Uruguay, 2001	BOV	AY593801
12	A Uruguay/98	Uruguay, 2001	BHK-21	AY593802
13	A Bage/77*	Brazil, 1977	2 BTT	AY593787
14	A Brazil/79	Brazil, 1979	1 BHK-21, 1 IBRS-2	AY593788
15	A Venceslau/76	Brazil, 1976	5 BHK-21, BTT	AY593803
16	A13 Brazil /70	Brazil, 1970	7 GP	AY593753
17	A16 Belem/59	Brazil, 1959	BOV, 1 LK	AY593756
18	A17 Aguarulbos/67	Brazil, 1967	3 BOV, 3 LK	AY593757
19	A24 Cruzeiro/55	Brazil, 1955	5 BHK-21, 5 IBRS-2	AY593768
20	A Sabana/85	Colombia, 1985	1 BTT, 1 BHK-21, 1 IBRS-2	AY593794
21	A27 Colombia/67	Colombia, 1967	10 GP	AY593771
22	A29 Peru/69	Peru, 1969	GP, 1 LK	AY593773
23	A18 Zulia/67	Venezuela, 1967	1 BK, 10 GP, 2 LK	AY593758
24	A32 Venezuela/70	Venezuela, 1970	1 BOV, LK	AY593775
25	A10 Holland/42	Holland, 1942	1 IBRS-2	AY593751
26	A12 Valle Strain 119*	Great Britain, 1932	77 BOV	AY593752
27	A1 Bayern/Bavaria	Germany, 1971	GP, 1 IBRS-2	AY593759
28	A3 Mecklenburg/68	Germany, 1968	1 BOV, 1 LK	AY593776
29	A5 Westerwald/51*	West Germany, 1951	ET, BTT	AY593781
30	A4 WG/42	West Germany	4 BTT, 4 BHK-21	AY593777
31	A4 WG/72	West Germany, 1972	1 BOV, 1 LK	AY593779
32	A5 Allier/60	France, 1960	3 GP, 4 BT, 4 BHK-21	AY593780
33	A8 Parma/62	Italy, 1962	BOV, 1 LK	AY593792
34	A4 Spain*	Spain	ET, 1 BTT	AY593778
35	A2 Spain/69	Spain, 1969	1 LK	AY593774
36	A14 Spain/59	Spain, 1959	BOV, 1 LK	AY593754
37	A Philippines/75	Philippines, 1975	4 GP, 2 LK	AY593793
38	A15 Thailand/60	Thailand, 1960	1 BOV, 3 LK	AY593755
39	A22 Turkey/65*	Turkey, 1965	2 BTT	AY593765
40	A28 Turkey/72	Turkey, 1972	1 LK	AY593772
41	A20 USSR 1/64	USSR, 1964	2 LK	AY593760
42	A Iran/1/98	Iran, 1998	1 BOV	AY593791
43	A22 Iraq 24/64	Iraq, 1964	4 BTT, 1 LK	AY593762
44	A22 Iraq 26/64*	Iraq, 1964	3 BOV	AY593763
45	A22 Iraq/70	Iraq, 1970	2 LK	AY593764
46	A21 Kenya/64	Kenya, 1964	9 GP	AY593761
47	A23 Kenya 6/65	Kenya, 1965	2 LK	AY593766
48	Asia 1 Leb 1/83*	Kfar Kela/Lebanon, 1983	2 PK, 2 BTT	AY593799
49	Asia 1 Leb 2/83*	Lebanon, 1983	2 BOV	AY593800
50	Asia 1 Leb 3/83*	Southern Lebanon, 1983	2 PK, 2 BTT	AY593798
51	Asia 1/1 PAK/54	Pakistan, 1954	6 BTT, 2 LK	AY593795
52	Asia 1/2 ISRL 3/63	Israel, 1963	BOV, 7 GP	AY593796
53	Asia 1/3 Kimron	Israel, 1963	BOV, 9 GP	AY593797
54	C Waldman strain 149*	Great Britain, 1970	BTT	AY593810
55	C1 Noville/65	Switzerland, 1965	2 PI, BHK-21	AY593804
56	C1 Oberbayern/60*	Germany, 1960	BHK-21	AY593805
57	C3 Indaial/71	Brazil, 1971	2 BHK-21	AY593806
58	C3 Resende/55	Brazil, 1955	1 LK	AY593807
59	C4 Tierra del Fuego/66	Tierra del Fuego, Argentina, 1966	1 LK	AY593808
60	C5 Argentina/69	Argentina, 1969	5 BHK-21, 5 GP	AY593809
61	O Rey Iran/66	Iran, 1966	BTT, BK, 1 LK	AY593834
62	O Uruguay/63	Uruguay, 1963	BK, 2 LK	AY593837
63	O UK 2001*	United Kingdom, 2001	BVF	AY593836
64	O UK2001-ED	Plum Island, 2002	1 BHK-21	AY593831
65	O UK2001-FB	Plum Island, 2002	2 BHK-21	AY593832
66	O1 Argentina/65	Argentina, 1965	2 BTT, 1 LK	AY593814
67	O1 BFS18/67	United Kingdom	1 LK	AY593815
68	O1 BFS46/67	United Kingdom	2 BHK-21	AY593816
69	O1 Brugge	Belgium, 1973	1 LK	AY593817
70	O1 Campos/58	Brazil, 1958	1 BTT	AY593818
71	O1 Campos/94	Plum Island	BHK-21	AY593819
72	O1 Canefa/64*	Argentina, 1964	2 BTT	AY593820
73	O1 Caseros/67*	Argentina, 1967	4 BTT	AY593821
74	O1 M11*	Unknown, 1953	BTT	AY593822
75	O1 Manisa/69*	Turkey, 1969	BVF	AY593823
76	O1 SKR/00*	S. Korea, 2000	Swine epithelium	AY593824
77	O1 Vallee/39*	Pirbright, 1953	Bovine epithelium	AY593825
78	O10 Philippines	Philippines/58	1 LK	AY593811
79	O10 Philippines 2/58*	Philippines/58	BTY, BK, BOV	AY593812
80	O11 Indonesia/62	Indonesia, 1962	BTT, 2 LK	AY593813
81	O2 Brescia/47	Italy, 1947	1 LK	AY593826
82	O3 Venezuela/71	Venezuela, 1971	BTT, 1 LK	AY593827
83	O5 India/62*	India, 1962	1 BK, 1 BTT	AY593828
84	O6 Pirbright/65	United Kingdom /65	4 GP	AY593829
85	O7 Poland/59	Poland, 1959	1 LK	AY593830
86	O Taiwan/97	Taiwan, 1997	Swine epithelium	AY593835
87	O Penghu/99	Taiwan, 1999	Swine epithelium	AY593833
88	SAT 1 BOT 1/68	Botswana (BEC), 1968	3 BTY, 5 BHK-21	AY593845
89	SAT 1 Rhod5/66*	Rhodesia, 1966	1 BTT	AY593846
90	SAT 1/1 BEC	Botswana (BEC), 1970	2 LK	AY593838
91	SAT 1/20 RV 11/37	Pirbright, WRL, 1970	3 IBRS-2	AY593839
92	SAT 1/3 SWA1/49	Unknown, 1970	4 IBRS-2	AY593840
93	SAT 1/4 SR2/58	Rhodesia, 1970	3 IBRS-2	AY593841
94	SAT 1/5 SA/61	Unknown, 1972	3 IBRS-2	AY593842
95	SAT 1/6 SWA40/61	Unknown, 1970	3 IBRS-2	AY593843
96	SAT 1/7 ISRL 4/62	Israel, 1962	3 IBRS-2	AY593844
97	SAT 2/1 Rhod/48	Rhodesia, 1948	3 IBRS-2	AY593847
98	SAT 2/2 106/67	Unknown, 1970	2 IBRS-2	AY593848
99	SAT 2/3 Kenya 11/60	Kenya, 1960	3 IBRS-2	AY593849
100	SAT 3/2 SA57/59	Unknown, 1969	1 LK	AY593850
101	SAT 3/3 BEC20/61	Botswana (BEC), 1961	2 LK	AY593851
102	SAT 3/3 Kenya 11/60	Kenya, 1960	4 IBRS-2	AY593852
103	SAT 3/4 BEC 1/65	Botswana (BEC), 1965	1 LK	AY593853
104	O/JPN/2000	Japan	BEF, 4 BK	AB079061
105	O strain Chu-Pei	Taiwan	Swine	AF026168
106	OTau-YuanTW97	Taiwan	Unknown	AF154271
107	O1 Geshure	Israel	Unknown	AF189157
108	O1 Yunlin	Taiwan	Swine	AF308157
109	O/SK 2000	South Korea	Unknown	AF377945
				AY312586
	O/SK 2002	South Korea	Unknown	AY312588
110	O Tibet /1/99	China	Unknown	AF506822
111	O Akesu/58	China	BOV	AF511039
112	SAT 2 ZIM/7/83	Zimbabwe, clone	Vaccine strain, cloned	AF540910
113	A12-strain 119	United Kingdom	Tissue culture, cloned	M10975
114	Asia 1 IND 63/72	India, vaccine	Vaccine strain, unknown	AY304994
115	A22/550 Azerbaijan 65	USSR	Unknown	X74812
116	O-HKN/2002	China	Unknown	AY317098
117	O strain OMIII	Attenuated from strain Akesu/58	Unknown	AY359854
118	Asia1 YNBS/58	China	Unknown	AY390432
119	C rp99	C-S8c1 derivative	Persistently infected BHK-21 cells, passage 99	FAN133358
120	C rp146	C-S8c1 derivative	Persistently infected BHK-21 cells, passage 146	FAN133359
121	C-s8c1	Spain (Olot)	BHK-21	FDI133357
122	SAT 2 KEN/3/57	Kenya	Unknown	FDI251473
123	O1Campos	Brazil	Tissue culture, cloned	FDI320488
124	C3Arg85	Argentina	BHK-21	FMV7347
125	C3Arg85 Rb-15	C3Arg85 derivative	Partially resistant BHK-Rb cells	FMV7572
126	TAW/2/99 TC	Taiwan	BEF, 3 BHK-21, 1 BTY, 2 BHK-21	AJ539136
127	TAW/2/99 BOV	Taiwan	BEF, 3BHK-21, 1BTY, 2BHK-21, 1 BOV, 2 swine, 1 BOV	AJ539137
128	Tibet/CHA/99	China	1 BOV, 3 BHK-21	AJ539138
129	O/SK 2000	South Korea	BOV,1 BHK-21	AJ539139
130	SAR/19/2000	South Africa	BOV, 2 PK, 1 IBRS-2, 1 BHK-21	AJ539140
131	UKG/35/2001	United Kingdom	2 swine	AJ539141
132	A10-61	Argentina	Unknown, cloned	PIFMDV1
133	O1-Kaufbeuren/66	Germany	BHK-21, cloned	PIFMDV

^a*, RNA was extracted directly from tissue or vesicular fluid.

^bBHK, BHK21 baby hamster kidney cells; LK, PK, and BK, lamb, porcine, and bovine kidney cells, respectively; BTY, bovine thyroid cells; IBRS-2, porcine kidney epithelial cell line; ET, epithelial tissue; BTT, bovine tongue tissue; GP, guinea pig vesicular fluid; BOV, bovine; BVF, bovine vesicular fluid; BEF, bovine esophageal fluid.

^cSequence numbers 104 to 133 are FMDV sequences from GenBank.

FMDV RNA isolation, RT-PCR amplification, and sequencing.

Full-length FMDV genome sequences were obtained by reverse transcription (RT) of the complete viral genomic RNA followed by redundant amplification and sequencing of overlapping cDNA fragments spanning the entire viral genome. The total RNA was extracted directly from 140 μl of vesicular fluid, the supernatants of infected cell cultures, or 0.1 g of epithelial tissue from vesicular lesions by the use of RNeasy Mini kits (QIAGEN) and was then resuspended in 40 μl of double-distilled H₂O.

Complete viral genomes were reverse transcribed to cDNAs by use of a Super SMART PCR cDNA synthesis kit (Clontech), using the supplied modified oligo(dT) according to the manufacturer's instructions. After spin column purification, the full-length double-stranded cDNA from each isolate was amplified in eight identical long-distance PCRs (2 μl cDNA in a 50-μl reaction volume; 20 cycles of 95°C for 15 s, 65°C for 15 s, and 68°C for 2 min) using primers supplied in the kit. Next, 21 overlapping 1.2-kb PCR amplicons spanning the whole FMDV genome were generated in 96-well plates (2.5 μl of template from each of the eight double-stranded cDNA replicates in 25-μl reactions using the primers described in Table S1 in the supplemental material). Adjacent amplicons overlapped for >50% of their lengths. To prevent truncation by subsequent rounds of PCR, we artificially created poly(C) sequences by substituting flanking sequences with specific primers (F1R and F2F) containing six C's or G's.

Amplified products of the correct size were purified (DNA filtration system; Eppendorf) and resuspended in 40 μl of sterile double-distilled H₂O per well. All amplicon replicates (eight per FMDV subgenomic fragment per isolate) were subjected to direct sequencing (2.5 μl of template per reaction) using position-specific forward and reverse primers (see Table S2 in the supplemental material), Big Dye Terminator cycle sequencing kits (Applied Biosystems), and a PRISM 3730xl automated DNA sequencer (Applied Biosystems). To overcome virus variability and PCR/sequencing errors, we sequenced each amplicon in multiple reactions with both forward and reverse primers, specifically selected by serotype (see Table S2 in the supplemental material). Since each amplicon overlapped its neighbor >50%, the multiple direct sequencing reactions resulted in a redundancy rate ranging from 13 to 48 and averaged 28 sequence events per nucleotide.

Direct DNA sequencing of amplicons derived from a given FMDV isolate yielded a master sequence representing the most probable nt for each position of the sequence. This approach prevented analyses of minor sequence variants, polymerase misincorporation errors, and sequencing ambiguities through multiple independent cDNA synthesis, PCR amplification, and direct sequencing events. Due to the quasispecies nature of FMDV populations, polymorphisms were detected at some nt positions. Nevertheless, all positions could be unambiguously assigned to a single nt due to the high degree of redundancy generated by the sequencing strategy. The following sequence ambiguity code was used: K (T/G), M (A/C), R (A/G), S (C/G), W (A/T), Y (C/T), B (C/T/G), D (A/T/G), H (A/C/T), V (A/C/G), and N (A/C/G/T).

Sequence analysis.

Bases were called from chromatogram traces produced with Phred (25), which also produced a quality file containing a predicted probability of error at each base position. Viral sequences were assembled with the Phrap (http://www.phrap.org) and CAP3 assemblers (43). Gap closure was performed as described previously (3). Multiple sequence alignments were performed with the ClustalW (1.7) and Dialign (2.2) computer programs (73, 107). The positions of 5′ UTR sequences presented in Results refer to the consensus sequence generated by ClustalW alignment of this region for the 103 isolates, with some manual editing. Nucleotide substitution analysis was carried out by use of the DISTREE (1.2) (103), DNArates (1.1) (105), ALISTAT (16), and PRETTY (GCG 10 software package [16]) programs. Analyses of codons and synonymous/nonsynonymous (syn/nonsyn) substitution ratios were calculated by using SNAP (77), CodonW (http://www.molbiol.ox.ac.uk/cu/), and codeml (PAML3.14 package), which was also used for statistical evaluations of heterogeneous selection pressures at amino acid sites (116). For protein analysis, the PRETTY program was used. Searches for motifs and/or signal sequences were performed with the MOTIFS (GCG package), HMM (58), pfscan (http://www.isrec.isb-sib.ch/ftp-server/pftools/pft2.2/pft2.2.tar.Z), and Blocks (36) programs of the PROSITE, Pfam, and Blocks databases. Transmembrane protein segment predictions were performed by using Memsat (50), Tmpred (41), Toppred (114), Psort (76), and Saps (11). Secondary structure predictions were performed for proteins by using GOR secondary structure prediction (29) and Pratt (48, 49) and for RNAs by using mfold (46, 117), rnafold (40), and Squiggles (83). Pseudoknot analysis was conducted with pknots (96).

Phylogenetic analysis was performed on aligned genomic and subgenomic regions of FMDV by utilizing the neighbor-joining and split decomposition methods as implemented in the Phylo_win (28) and/or SplitsTree 4.0 (http://www-ab.informatik.uni-tuebingen.de/software/jsplits/welcome_en.html) (44) software package and by utilizing maximum likelihood as implemented in Puzzle (104) and dnaml in the Phylip package (http://evolution.genetics.washington.edu/phylip.html) (27). Individual protein-coding regions were used to screen for incongruent tree topologies suggestive of genomic recombination, and split decomposition analysis (7) of genomic and subgenomic regions was utilized to graphically screen for reticulated branching patterns which were also suggestive of recombination. Similarity plots and bootscanning analysis (100), which compare a given query sequence to several reference sequences via incremental sliding sequence windows to yield corrected similarity values and bootstrap resampling frequencies, respectively, were performed as implemented in the SimPlot, v. 2.5, package (63), utilizing default settings and a window size of 400 nt. As a further test for potential recombination, Sawyer's run test (102), as implemented in GENECONV, v. 1.81 (http://www.math.wustl.edu/∼sawyer), was used on Dialign-aligned FMDV polyprotein-encoding regions, using default settings.

Nucleotide sequence accession numbers.

The GenBank accession numbers for the genome sequences of FMDV isolates sequenced for this study are listed in Table 1.

RESULTS AND DISCUSSION

With the exception of the poly(C) tract, full-length genomic sequences from 103 FMDV isolates representing all serotypes, including isolates from serotypes not previously fully sequenced (SAT1 and SAT3), were obtained and analyzed. Given that the structure and dynamics of FMDV populations in nature are complex and the relatively small number of SAT2 and SAT3 genomes presented here, it is possible that additional sequence data from field isolates will alter the status of residues and nucleotides presented below as invariant.

FMDV genomes ranged from 8,046 to 8,215 nt, consistent with previously reported genome lengths (86, 108). Although small insertions/deletions were observed in the coding region affecting L^pro, 1B, 1C, 1D, and 3A, most variability in genome size was due to insertions/deletions in the UTRs.

5′ UTR.

The approximately 1,300-base 5′ UTR plays important roles in FMDV replication. However, the specific contribution of S fragment, poly(C) tract, and L fragment pseudoknots to FMDV biology is unknown. The cardiovirus poly(C) tract has been shown to affect virulence (23), while shortening of the FMDV poly(C) tract affected virus growth in vitro but not virulence in a suckling mouse model (95). The cre is essential for picornaviral replication and contains a conserved AAACA motif which in poliovirus functions as a template for 3D^pol-mediated uridylylation of 3B (75). Mutation of the FMDV cre stem region reduced virus RNA replication (68). The approximately 500-nt FMDV IRES, which is responsible for cap-independent polyprotein translation (60), is predicted to contain four structural domains (Fig. 1D, panel II, domains 2 to 5) that interact with cellular factors involved in host translation initiation (94). Domain 2 binds the polypyrimidine tract-binding protein at a UUUC motif. Domain 3 contains at least three loop structures that are conserved in all picornaviruses, including a GNRA motif essential for the maintenance of the IRES tertiary structure and potentially involved in long-range RNA-RNA interactions. The Y-shaped domain 4 and domain 5 interact with the host translational factors eIF4B and eIF4G (65, 99). A 22-nt spacer for ribosomal recognition separates the IRES from the first of two polyprotein start codons, and an additional 84 nt separates the first start codon from the second. Picornavirus utilization of the second start codon seems to be affected by its adjacent sequence and by sequences in the IRES (82). Unlike poliovirus, FMDV utilizes the second start codon twice as often as the first (66).

The analysis in this study of 5′ UTRs from 103 FMDV isolates showed that only 12% and 33% of nt were invariant in the 5′ UTR S and L fragments, respectively (Fig. 1B and C; Table 2). In pairwise comparisons, however, FMDV isolates averaged 80% and 85% nt identity for the S and L fragments, respectively, indicating a high degree of conservation between isolates. This suggests that although the FMDV 5′ UTR can tolerate changes at most positions, selective pressures clearly favor overall conservation, likely for the maintenance of a functional secondary structure. Relatively high rates of substitution (>10) in the 5′ UTR L fragment were observed at positions 513, 558, 615, 684, 696, and 1144 (Fig. 1B).

TABLE 2.

FMDV genome variability^a

UTR or genome region	No. of nt positions aligned	No. of invariant nt	% Invariant nt	Ts/Tv rate (%)	Average pairwise identity (%)	Minimum pairwise identity (%)
5′ UTR	1,150	303	26	2.89	83	64
S-UTR	393	48	12	3.93	80	51
L-UTR	757	255	33	2.63	85	70
3′ UTR	161	15	9	3.93	82	63
ORF	7,076	3,287	46	2.44	84	73
Lpro	606	227	37	3.23	80	61
1A	255	143	56	2.53	84	73
1B	660	243	37	1.75	75	62
1C	675	235	35	1.84	75	58
1D	693	147	21	1.55	68	48
2A	48	23	48	2.64	89	73
2B	462	282	61	6.67	91	81
2C	953	527	55	5.14	89	78
3A	459	160	35	3.75	85	68
3B	216	96	44	5.51	89	69
3C	639	380	59	4.12	89	77
3D	1,410	824	58	5.53	91	83

^aEstimates were based on ClustalW (UTR data) or Dialign (ORF data) alignments of the 103 FMDV isolates examined. The number of invariant nt for each genomic region relative to the total number of positions was used to estimate the percentage of invariant nt. The ratio of transitions to transversions (Ts/Tv) was obtained by using Distree and Kimura 2 parameter correction with gamma distributions. ALISTAT was used to obtain the average % pairwise identity and the minimum pairwise identity. The 5′ UTR includes the S and L fragments 5′ of the first AUG. The S-UTR includes nt from position 1 to the poly(C) tract. The L-UTR includes nt from the end of the poly(C) tract to the first AUG. The 3′ UTR includes nt from the polyprotein stop codon to the poly(A) tail.

5′ UTR S fragment.

S fragment lengths averaged 322 and 373 nt for SAT and Euroasiatic isolates, respectively, and spanned positions 1 to 393 in the 5′ UTR alignment (Fig. 1B). SAT viruses contained insertions/deletions of 1 to 3 nt, with most occurring at positions 120 to 160 and 200 to 300, including SAT-specific nt present at positions 38, 39, 67, 95, 250, 350, 366, and 375. Euroasiatic isolates contained insertions/deletions of 1 to 5 nt in length (with the exception of isolates C Waldman strain 149, A Canefa 1/61, and A25 Argentina/59, which contained a 76-nt deletion between positions 153 and 228) and included Euroasiatic-specific nt present at positions 86, 123, 124, 144, 145, 155, 156, 157, 174, 181 to 191, 197 to 219, 240, 256, 277, 292 to 294, 300 to 306, and 324 (Table 3). Notably, SAT1/7 Isrl 4/62 and SAT2-3 Kenya 11/60 appeared to have the most complete S fragment sequences of those sampled, lacking both SAT and Euroasiatic-specific deletions, and contained nt at positions 142, 143, 239, 290, 291, and 307 which were absent from other viruses.

TABLE 3.

Selected FMDV 5′ UTR insertions/deletions

Isolate or fragmenta	nt positions
	Ins	Dels
S
A1 Bayern/Bavaria		80-82
A3 Mecklenburg/68		80-82
A Canefa 1/61		114
A25 Argentina/59		114
Asia 1/1 PAK/54		114
O11 Indonesia/62		114
SAT 1/7 ISRL 4/62	142, 143
SAT 2/3 Kenya 11/60	142, 143
C Waldman strain 149*		153-228
A25 Argentina/59		153-228
A Canefa 1/61		153-228
A24	270
O1 Argentina	270
SAT 1/7 ISRL 4/62	290, 291
SAT 2/3 Kenya 11/60	290, 291
A Argentina/01 Areco		290-293
A ARGp55/01		290-293
A ARGp64/01		290-293
A General Lopez/01		290-293
A Trenquelauquen/01		290-293
A Uruguay/01		290-293
A Uruguay/98		290-293
O UK 2001		292
O UK 2001-ED		292
O UK 2001-FB		292
O1 SKR/00		292
A22 Azerbaijan	320
A22 Iraq24/64	320
A22 Iraq26/64	320
A22 Turkey/65	320
A28 Turkey/72	320
A22 Turkey/65		364-366
A28 Turkey/72		364-366
L
A28 Turkey/72		418-478
O1 Vallee/39		418-478
O5 India/62*		418-478
O Akesu/58		418-478
O Akesu/58 vaccine		418-501
C1 Noville/65		418-478
C1 Oberbayern/60*		418-478
C-s8c1 Spain		418-478
C-Perst.inf.BHKp99		418-478
C-Perst.inf.BHKp146		418-478
SAT 1/7 ISRL 4/62		418-487
A24 Cruzeiro/55		427-487
A21 Kenya/64		427-487
A20 USSR 1/64		427-487
O11 Indonesia/62		427-487
SAT 2/3 Kenya 11/60		427-487
SAT 1/20 RV 11/37		427-487
SAT 1 Rhod5/66*		427-487
SAT 2/1 Rhod/48		427-487
SAT 2/2 106/67		427-487
SAT 3-4 BEC 1/65		427-487
A5 Westerwald/51*	427-443
O Penghu/99		460-502
O Yunlin Tw97		460-502
O HKN/2002		460-502
O Taiwan/97		460-502
A26 Argentina/66*		419-523
A16 Belem/59		419-523
O11 Indonesia/62		523-546
Asia1-IND 63/72		554-591
A3 Mecklenburg/68		554-591
A4 Spain*		554-591
A2 Spain/69		554-591
A4 WG/72		554-591
A1 Bayern/Bavaria		554-591
A10 Holland/42		554-591
O1 M11		554-591
O2 Brescia/47		554-591
O3 Venezuela/71		554-591

^a*, isolates from which RNA was extracted directly from tissue or vesicular fluid.

Although the overall nt substitution rates in the S fragment were low, specific regions of variability were observed, and only 19 nt were invariant, excluding the regions used to prime amplification (Fig. 1C and D). S fragment sequences generally grouped into SAT and Euroasiatic lineages, which averaged only 50% nt identity with each other; however, little or no correlation with other serotypic groupings was observed (data not shown). Notably, several isolates of disparate serotypes contained very similar S fragment sequences, including the C Waldman strain 149 and A12 Valle strain 119 S fragments, which shared 98% nt identity.

In most cases, secondary structure analyses of S fragments predicted a single stem-loop structure similar to that previously proposed (12, 24). However, the S fragment sequences of 16% of the isolates folded into alternative structures within similar free energy levels (e.g., strains A22 Turkey/65, A24 Argentina 65, and Asia 1 Leb 83) (data not shown).

5′ UTR L fragment.

The 5′ UTR region of the L fragment ranged from 604 to 751 nt long. Remarkably, apparently unrelated virus isolates contained identical deletions in the L fragment (Table 3). SAT and Euroasiatic-specific L fragment insertions/deletions were identified, as were isolate-specific insertions/deletions, including an 18-nt insertion located 28 nt downstream of the poly(C) tract in strain A5 Westerwald/51.

The pseudoknot region between positions 403 and 600 was highly tolerant to changes, with no invariant nt located within the first 200 positions and insertions/deletions tolerated at each position (Fig. 1C). Previously undescribed invariant nt and motifs located between the pseudoknot region and the IRES included AGAAWYGGGACGU (positions 617 to 629), GCRCACGWAACGCGC (positions 632 to 646), and ACAAAC (positions 668 to 673) within the cre.

FMDV IRES sequences (positions 640 to 1151) showed 70 to 100% nt identity in pairwise comparisons, 47% invariant nt, and numerous invariant motifs in the predicted secondary structure domains (Fig. 1C and D). Domain 2 contained the polypyrimidine tract-binding protein motif UUUC and three previously undescribed motifs at the base, bulge (invariant GGUCUWGAG motif), and apex regions of the stem-loop. Domain 3 contained GYRA (corresponding to the conserved picornaviral GNRA motif) and CRAAA (except in isolates O1Argentina/65 and OAkesu/58) motifs in loops A and B, respectively. While the picornavirus domain 3 loop D motif ACCC was present in 99/103 FMDV isolates, the loop 3C motif ACAC was not conserved in FMDV. Domain 3 also contained novel highly conserved motifs, including the UCGUMGCGGAGCA (positions 823 to 835) motif and GRUACUGGUA and GRGACUGGUA motifs (positions 965 to 974), specific for Euroasiatic and SAT serotypes, respectively, and CUGGWGRCAGGCUAAGGAUGCCCU (positions 983 to 1006), predicted to form a prominent bulge at the base of the stem (Fig. 1C and D). Domain 4 was highly conserved, with two novel invariant motifs (GAUCUGAG [positions 1039 to 1046] and UUAAAAG [positions 1080 to 1087]) predicted to form prominent bulges in the secondary structure. In domain 5, 19 of the 21 nt were invariant (Fig. 1C and D). The conservation of these motifs suggests biological significance. Notably, the 22-nt region between the IRES and the first AUG was highly variable, possibly underlying FMDV's preferential use of the second start codon (9, 66). In summary, the FMDV 5′ UTR, although tolerant for nt substitutions and insertions/deletions, showed significant conservation, especially in the IRES, where novel sequence motifs were identified.

Polyprotein region. (i) Nucleotide sequence and codon analysis.

The nucleotide variability for the 103 FMDV polyprotein (ORF) and individual protein regions is summarized in Table 2. Overall, 46% of all nt were invariant, with 73% nt identity between the least similar pair of sequences. An average transition/transversion (Ts/Tv) rate of 2.4 and syn/nonsyn substitution ratio of 2.1 were observed. As expected, the region encoding 1D was the most variable, exhibiting the lowest percentage of invariant nt (21%), Ts/Tv rate (1.55), and syn/nonsyn ratio (1.03) of all regions. In fact, excluding the highly conserved 1A (VP4) protein, regions encoding structural proteins had significantly lower Ts/Tv rates, nonsyn substitutions/site, and syn/nonsyn ratios than the rest of the genome. Although the substitution rates given here are averages for each protein-coding region, specific regions or residues within each protein were observed to contain higher or lower substitution rates (Fig. 2B and C).

FIG. 2.

FMDV coding region variability. (A) Schematic diagram of FMDV polyprotein coding region showing the positions of, from top to bottom, protein-encoding regions (open boxes), cleavage intermediates, and mature proteins (lines). (B) Graphic representation of rates of nt substitution per site as calculated with DNArates software. (C) Graphic representation of nonsynonymous substitutions per site as calculated with SNAP software.

In general, regions encoding FMDV nonstructural proteins were highly conserved between isolates, exhibiting higher percentages of invariant residues, higher Ts/Tv rates and syn/nonsyn substitution ratios, and lower overall numbers of nonsyn changes/site (Table 2; Fig. 2C). The most conserved FMDV regions were those encoding 2B and 3C, which exhibited the highest percentage of invariant nt (61% and 59%, respectively) and amino acids (76%) in the genome. Notably, the nonstructural proteins L^pro, 3A, and 3B exhibited variability comparable to that of the structural proteins 1B, 1C, and 1D (Table 4), suggesting that these proteins are subjected to selective pressures distinct from those on the other nonstructural proteins. A comparison of CODEML ω rates between different substitution models further indicated that four of these proteins (L^pro, 1D, 3A, and 3B) may indeed undergo diversifying selection (Table 4) (116). Notably, this was observed for relatively few codons (1.5% to 7.6%), with fewer highly significant ones (P > 0.9) tending to cluster within each protein (positions 19, 20, 22, 23, and 82 in L^pro; positions 45, 48, 142, 143, 144, and 146 in 1D; positions 44, 132, 135, 136, and 144 in 3A; positions 4 and 11 in 3B1; and positions 17, 18, and 19 in 3B2). Interestingly, the variable capsid proteins 1B and 1C did not appear in this analysis to undergo diversifying selection (Table 4).

TABLE 4.

FMDV genome variability in encoded amino acid sequences^a

Genome region	No. of aa positions aligned	No. of invariant aa	% Invariant aa	SNAP result			CODEML result
				Syn	Nonsyn	Syn/nonsyn	ω	t	PS
ORF	2,360	1,363	58	756	362	2.08	ND	ND	ND
Lpro	202	89	44	70	45	1.55	0.27	20.8	Y
1A	84	68	81	33	5	6.6	0.01	22.5	N
1B	221	103	47	100	63	1.6	0.06	34.9	N
1C	225	87	39	99	63	1.6	0.08	32.9	N
1D	232	56	24	102	99	1.03	0.16	35.8	Y
2A	17	11	65	5	0.52	9.6	ND	ND	ND
2B	153	117	76	35	7	5	0.06	8	N
2C	318	228	72	83	18	4.6	0.06	10.5	N
3A	153	57	37	44	21	2.1	0.16	12.7	Y
3B	72	36	50	17	6	2.8	0.33	8.2	Y
3C	213	163	76	55	14	3.9	0.04	11.3	N
3D	471	348	74	113	20	5.6	0.06	8.4	N

^aThe synonymous/nonsynonymous (syn/nonsyn) substitutions were estimated by use of the Nei and Gojobori algorithm as implemented in SNAP. Estimates for diversifying positive selection were obtained by using codeml. ω, nonsyn/syn ratio averaged over sites (d_N/d_S); t, maximum-likelihood tree length measured by the number of nt substitutions along the tree per codon; PS, positive selection; Y, yes; N, no; ND, not done.

(ii) Amino acid sequence analysis.

Fifty-eight percent of all amino acids (aa) were invariant among the FMDV isolates analyzed here, with the percentage of amino acid conservation for each viral protein correlating with nt conservation in the corresponding genomic region (Fig. 3; Tables 2 and 4). Positions at putative protein cleavage sites showed various degrees of sequence conservation, with Euroasiatic lineages being more conserved than SATs. Residues at the 1A/1B, 2A/2B, 2B/2C, 2C/3A, and 3B/3C cleavage sites exhibited high degrees of conservation (Table 5).

FIG. 3.

FMDV invariant amino acids (single letter code). The consensus sequence is based on the comparison of 103 FMDV isolates conducted for this study, with the exception of the 1D region, which also included all 1D sequences available in GenBank (release 141).

TABLE 5.

FMDV polyprotein cleavage site variability^a

Junc- tion	Common amino acids for serotype
	ALL (n = 133)									SAT (n = 16)									A (n = 46)									O (n = 23)									C (n = 7)									Asia (n = 6)
L/1A	R	K	L	K	*	G	A	G	Q	L	F	L	R	*	G	A	G	Q	R	K	L	K	*	G	A	G	Q	R	K	L	K	*	G	A	G	Q	R	K	L	K	*	G	A	G	Q	K	R	L	K	*	G	A	G
	K46	R46	V3	R36					H3	A1	Y2	V2	K2					H1	K9	R10							H1	K8	R7		R2						Q2			R1
	L15	F15								T1									Q4	N1
	T1	Y3																	W1
	Q6	N1
	A2
	W1
1A/1B	A	L	L	A	*	D	K	K	T	A	L	L	A	*	D	K	K	T	A	L	L	A	*	D	K	K	T	A	L	L	A	*	D	K	K	T	A	L	L	A	*	D	K	K	T	A	L	L	A	*	D	K	K
	T1																											T1
1B/1C	P	S	K	E	*	G	I	F	P	P	A	K	Q	*	G	I	L	P	P	S	K	E	*	G	I	F	P	P	S	K	E	*	G	I	F	P	P	S	K	E	*	G	I	F	P	P	S	K	E	*	G	I	V
		A10		Q17			V1	V21			S3		E2				I6					Q1			V10
		V2						I7			V2						V2
		G2						L7			G1						F1
		M1									M1
1C/1D	P	R	T	Q	*	T	T	S	T	P	S	R	Q	*	T	T	S	A	P	R	Q	Q	*	T	T	A	T	A	R	T	Q	*	T	T	S	A	A	R	Q	Q	*	T	T	T	T	A	R	Q	Q	*	T	T	T
	A67	S8	Q27	E14				A43	A51		V3	T2						V2			P10					T7	V7			A9	E10					T6			R2					A3	A1			R2
	T1	V5	R26					T23	V9		A2	A2									S10					S4	A3			R5						P1
		A2	A16						P2		T2										T7									V1
		T2	P14						L1		C1										A3
		C1	S12						S1												L1
			V1																		R1
			L1
1D/2A	P	A	K	Q	*	L	L	N	F	P	A	K	Q	*	L	C	N	F	P	A	K	Q	*	L	L	N	F	P	V	K	Q	*	T	L	N	F	P	A	K	Q	*	L	L	N	F	P	E	K	Q	*	V	L	N
	V5	G1	R1			S1	C12	S1	C1	V3	E6				M3	L1	S1	C1		E8	R3								A2				L7					E1					S2								L1
		E25				T19	S3				D2					A1					G1								T2																						T1
		T2				A2	A1				V1					S1					T1																														A1
		S1				V4	F1									F1					V1
		V38				M3
		K1
		D2
2A/2B*	S	N	P	G	*	P	F	F	F	S	N	P	G	*	P	F	F	F	S	N	P	G	*	P	F	F	F	S	N	P	G	*	P	F	F	F	S	N	P	G	*	P	F	F	F	S	N	P	G	*	P	F	F
	P2		L1	R1			L2	S1								L2			P1
2B/2C	A	E	K	Q	*	L	K	A	R	A	E	K	Q	*	L	K	A	R	A	E	K	Q	*	L	K	A	R	A	E	K	Q	*	L	K	A	R	A	E	K	Q	*	L	K	A	R	A	E	K	Q	*	L	K	A
							R3																		R1																		R1
2C/3A	I	F	K	Q	*	I	S	I	P	I	F	K	Q	*	I	S	I	P	I	F	K	Q	*	I	S	I	P	I	F	K	Q	*	I	S	I	P	I	F	K	Q	*	I	S	I	P	I	F	K	Q	*	I	S	I
				L1				L2									L2																		T1																		V1
								T3
								V1
3A/3B	P	Q	A	E	*	G	P	Y	A	P	Q	A	E	*	G	P	Y	A	P	Q	A	E	*	G	P	Y	A	P	Q	A	E	*	G	P	Y	A	P	Q	A	E	*	G	P	Y	A	P	Q	A	E	*	G	P	Y
		R9	T2						T24		K1							T1		R3	T1						T3									T4									T1
		H5							S2		H1									H4																V1
		K1							V1
3B/3C	I	V	T	E	*	S	G	A	P	I	V	T	E	*	S	G	C	P	I	V	T	E	*	S	G	A	P	I	V		E	*	S	G	A	P	I	V	T	E	*	S	G	A	P	I	V	T	E	*	S	G	A
						T1		C15							T1		A2
3C/3D	P	H	H	E	*	G	L	I	V	P	H	T	E	*	G	L	V	V	P	H	H	E	*	G	L	I	V	P	H	H	E	*	G	L	I	V	P	H	H	E	*	G	L	I	V	P	H	H	E	*	G	L	I
		Q2	T13					V16	L1			H2					I2	L1									I1								V1
									I2			N1

^aIndividual columns and rows represent FMDV serotypes and protein junctions, respectively. The first line in each row indicates the most frequent amino acid for each position, with alternative amino acids listed below and subscripts indicating the number of isolates with that particular residue. The total number (n) of isolates compared within each serotype is indicated at the top. *, protein cleavage point.

Leader proteinase.

FMDV L^pro is expressed as long (Lab) and short (Lb) isoforms that result from alternative usage of either the first or second start codon, respectively, with the sequence between the two AUGs predicted to form a stable hairpin structure crucial to IRES activity (39, 66). L^pro isoforms have indistinguishable activities and specificities and have been proposed to play a role in virulence that may involve the regulation of host interferon responses (89).

L^pro lacked insertions/deletions, with the exception of two additional aa in Euroasiatic lineages (positions 22 and 23). Each methionine start codon was invariant, indicating that both L^pro isoforms are significant for aspects of FMDV biology. Consistent with the previously described hypervariability in the region flanked by the two AUGs, only one residue (C6) was invariant in this region (113). Only 44% of the residues in Lb were invariant, making it much less conserved than was previously reported (30). Although substitutions were concentrated in terminal regions of L^pro, the predicted secondary structure in these regions remained relatively unaltered (data not shown).

A detailed analysis of the L^pro/1A junction resulted in a more ambiguous cleavage motif than was previously described (Table 5) (113). Only GAGXS at the 1A N terminus and the previously described basic residue (K/R) required for cleavage at the L^pro C terminus were invariant. Residues required for L^pro catalytic activity (C52, H149, and D165) (32, 59), suggested to be involved in L^pro autocatalysis (E77) (89), and important for eIF4G cleavage (H110 and H139) (90) were invariant, except for an H110D substitution in all SAT viruses, an E77Q substitution in O6 Pirbright/65, and an E77K substitution in A Phillipines/75 (Table 6). The present FMDV analysis revealed that only 43 of the 65 residues previously identified as conserved between FMDV O1 Kaufbeuren/66 L^pro and the aphthovirus equine rhinitis A virus L^pro were invariant, with 35 of the 43 residues concentrated in three distinct regions (positions 44 to 63, 95 to 110, and 133 to 185) (38) (Table 6). Additional, previously undescribed invariant FMDV L^pro motifs are shown in Table 6.

TABLE 6.

Amino acid conservation in FMDV and ERAV L^pro

Position no.a	1	6	30, 31	37	40	42	44-47	51-55	58-63	66, 67	73	75, 76	80
FMDVb	M	C	ME	G	K	F	SRPN	NCWLN	QLFRY	EP	Y	SP	T
FMDV/ERAVc	M							NCWLN	QL				T
Position no.	87	91	93	95	98-102	104	106	108-110	112	114, 115	117, 118	121-124	126	130
FMDV	L	T	L	L	GGPPA	V	W	IK(H/D)	L	TG	GT	RPSE	C	G
FMDV/ERAV		T		L	G PP	V					GT	P		G
Position no.	133-143	145, 146	149-153	156	159, 160	162	164-166	169-173	175	178, 179	183-185	187-189
FMDV	M(C/T)LADFHAGIF	KG	HAVFA	T	GW	A	DDE	YPWTP	P	VL	VPYD	EP
FMDV/ERAV	L DF A	K	HAVF	T	GW		DD	YP TP		VL	PYD

^aNumbers indicate L^pro amino acid positions from the first polyprotein AUG of Dialign-aligned FMDV polyproteins.

^bInvariant L^pro residues among 133 FMDV isolates. The proteolytic active sites C52 and H149 are shown in bold.

^cInvariant L^pro residues between FMDV and ERAV sequences (38).

These data indicate that despite a relatively low overall L^pro aa sequence conservation among the FMDV isolates examined here, functional elements such as the catalytic and cleavage sites and additional, previously undescribed motifs are invariant. The variability observed at the L^pro N and C termini may affect specific virus-host interactions, including ribosomal recognition of alternative start codons.

Structural proteins.

FMDV structural proteins are involved in capsid assembly and stability, virus binding to target cells, and antigenic specificity, influencing significant aspects of virus infection and immunity (45). The high level of variability in FMDV external capsid proteins observed here likely reflects the selective pressures on them.

(i) 1A (VP4).

1A was the most conserved FMDV protein, with 81% of the aa being invariant (Fig. 3; Table 4), including the N-terminal myristylation site and a previously identified heterotypic swine and bovine T-cell epitope (1A positions 20 to 35) (10). Interestingly, the Q73 residue in the SAT viruses distinguishes them from Euroasiatic lineages. Residues potentially specific for SAT2 and SAT3 (I76) and for SAT1 (V80) were also identified.

(ii) 1B (VP2).

1B, a protein of 218 or 219 aa, plays a critical role in virion structural stability and maturation (14). N-terminal regions of 1B contain previously undescribed motifs. For example, DKKTEETTLLEDRILTTRNGHT(T/I)STTQSSVG and DKKTEETT(L/H)LEDRI(L/M/V)TT(S/R)H(G/N)TTTSTTQSSVG are conserved among Euroasiatic and SAT virus isolates, respectively. Three previously identified T-cell epitopes (positions 48 to 68, 114 to 132, and 179 to 187) were conserved (88). Notably, only the N-terminal half of each epitope was invariant, while their C-terminal regions were highly variable.

(iii) 1C (VP3).

1C, a protein of 219 to 221 aa, contains important conformational neutralizing epitopes and makes significant contributions to capsid stability (62). Among the FMDV isolates examined here, 1C was highly variable, with only 39% of the aa being invariant. Most amino acid substitutions were concentrated in four regions (positions 55 to 88, 130 to 140, 176 to 186, and 196 to 208), with insertions/deletions present in the first two regions and a previously described T-cell epitope present in the second region (88). The 1B/1C cleavage site was nearly invariant in Euroasiatic lineages but much less conserved among SAT isolates (Table 5).

(iv) 1D (VP1).

1D is the most studied FMDV protein due to its significance for virus attachment and entry, protective immunity, and serotype specificity (45). 1D ranges in size from 217 to 221 aa, with insertions/deletions contained in regions 140-150 and 166-170. Overall, only 26% of the 1D residues are invariant. The invariant residues and motifs are shown in Fig. 3. Given the available data and the proposed significance of the conserved RGD motif for virus reception and pathogenesis (71, 72), it is notable that complete integrin-binding RGD motifs were lacking in 9 of the 103 FMDV isolates examined here (GGD in C Waldman strain 149 and C4 Tierra del Fuego/66, TGD in C5 Argentina/69, RGE in Asia1/2 Isrl 3/63 and A21 Kenya/64, RDD in A25 Argentina/59 and A Canefa 1/61, HGD in Asia1/3 Kimrom, and PGD in A27 Colombia/67) and in additional sequences present in GenBank (RGN in Asia1 Pak/1/54 and O/Syria/1/87, RSG in A A/IND/110/99, RGE in O KEN/5/95, KGD in O PAK/1/94 and O/IRQ/26/2000, SGD in O Akesu/58, and IGD in O3/Venezuela/51), with substitutions tolerable at any one of the three residues.

Exposure of the 1D N terminus, a region which includes a 10-aa motif which is conserved in several other picornaviruses, is critical for poliovirus entry into the cell (13). The FMDV 1D sequences studied here lack this N-terminal motif, suggesting that some aspects of viral entry may differ from those of other picornaviruses.

Structural implications of capsid protein amino acid conservation.

Picornaviral 1A is cleaved from the 1AB (VP0) precursor by unknown mechanisms, and this cleavage is required for virion maturation and infectivity (6, 74). During poliovirus entry into host cells, 1A is involved in receptor-mediated capsid conformational changes resulting in membrane ion channel formation and the release of viral RNA into the cytoplasm (reviewed in reference 42). Several FMDV residues which are known or suspected to affect 1AB cleavage based on their homology to other picornaviral 1ABs were identified in the FMDV isolates examined here. Notably, a number of these residues were located in previously unidentified invariant sequence motifs, suggesting that the critical function of these residues may be contextual. Specifically, these included 1B residues H145, P144, and L83 and 1C residues G39, F41 (replaced by Y in three isolates), and A50, which is included in the conserved motif LDVAEACPT (positions 45 to 53) (8). Additional residues contributing to the 1AB cleavage pocket include 1D P204, contained in the motif RMKRAE(T/L)YCPR (positions 195 to 205), and 1B V32 (I in SAT2 viruses), T33, and Y36, included in the motif TTSTTQSSVG(V/I)T(Y/F)GY (positions 22 to 36). C-terminal to this last motif lies a putative serotype sequence signature (positions 36 to 47) confirmed for 173 available 1B sequences (YSTXEDHXXGPN in A viruses, YXTXEDFVXGPN in O viruses, YATXEDXXGPN in C viruses, YXVXEDAVSGPN in Asia1 viruses, YAXXDXFLPGPN in SAT1 viruses, YADXDSFRXGPN in SAT2 viruses, and YXSADRFLPGPN in SAT3 viruses) and the motif GPNT(S/N)GLEXRVxQAER(F/Y)(F/Y)K (positions 45 to 63), which is present in all FMDV isolates examined here.

An H-rich region at the 1B/1C interphase is proposed to mediate 1B-1C hydrogen bonding and is likely involved in virion sensitivity to acid, a characteristic with implications for virus stability and transmission (2). 1B residues H21 (position 19 in SAT viruses), H145, H157, and H174 were invariant in all FMDV isolates studied here, while H87P and H168Y substitutions were present in serotype C and Euroasiatic viruses, respectively. In addition, 1C contained five invariant H residues, at positions 86 (84 in SAT viruses), 109, 146, 149, and 198 (196 in SAT viruses).

The N termini of five 1C molecules make up the β annulus at the axis of virion fivefold symmetry, thus contributing to capsid stability (1). The 53 N-terminal residues of 1C, including P4, which is possibly involved in binding to the 3C^pro substrate-binding site, and C7, a residue that is invariant in Euroasiatic isolates and involved in disulfide bonding between 1C N termini, were invariant within serotypes (1). However, all SAT viruses demonstrated an unexpected nonconservative C7V substitution, suggesting that SAT virus capsids may exhibit distinct physical properties.

Nonstructural proteins. (i) 2A.

FMDV 2A is an 18-aa peptide similar to the C terminus of cardiovirus 2A, a protein which induces a modification of the cellular translation apparatus resulting in 2A release (19, 20). The FMDV 2A proteins examined here averaged an 89% amino acid identity. Fourteen residues were identified as ≥98% invariant, including the DVEXNPG motif, which is essential for encephalomyocarditis virus 2A activity (21). The high level of 2A conservation likely reflects structural and functional constraints associated with the small size of the protein.

(ii) 2B.

FMDV 2B localizes to sites of virus genome replication in ER-derived vesicles (106). The 2B protein in other picornaviruses enhances membrane permeability, blocks protein secretory pathways, suppresses apoptotic responses by affecting intracellular Ca²⁺ homeostasis, and is implicated in virus-induced cytopathic effects (17, 47, 112).

FMDV 2B, a protein of 154 aa, contains 117 invariant residues (Table 4), with amino acid substitutions limited to only one or two alternate residues per site. A previously undescribed, conserved transmembrane domain was predicted between positions 120 and 140, suggesting that 2B is an integral membrane protein and consistent with the proposed localization of 2B to ER-derived vesicles (106).

(iii) 2C.

FMDV 2C is homologous to poliovirus 2C, an ATPase affecting the initiation of minus-strand RNA synthesis and whose precursor, 2BC, induces vesicle formation in the cytoplasm (53). FMDV 2C localizes to membrane-associated virus-replicating complexes (106).

FMDV 2C is 318 aa long and contains 72% invariant residues, including those of the putative ATP/GTP binding domain (positions 110 to 116, 160 to 163, and 243 to 246) (15). Amino acid substitutions unique to individual SAT serotypes were identified between positions 33 and 92.

(iv) 3A.

FMDV 3A has been implicated in virus virulence and host range, similar to the 3A proteins of other picornaviruses (33, 61). Deletions in 3A have been associated with FMDV attenuation in cattle and with the porcinophilic phenotype of O Taiwan/97 (54). However, evidence suggests that other viral genetic determinants are necessary for these phenotypes (54, 81).

FMDV 3A ranges from 143 to 153 aa, with insertions/deletions preferentially occurring at positions 70 to 110 and 130 to 150, regions identified here as containing previously undescribed variability. In fact, 3A is one of the most variable proteins encoded by FMDV, with fewer invariant aa (37%) than the variable capsid proteins 1B and 1C and the highly variable L^pro and contrasting with the high degree of conservation previously reported for a limited number of PanAsia 3A sequences (54). As described above, 3A contains residues predicted to undergo positive selection, including the Q44 residue, at which a Q44R mutation was previously associated with a pathogenic phenotype in guinea pigs (78). Although the Q44R mutation was present in several guinea pig-adapted isolates examined here, this mutation was absent from other isolates adapted to guinea pigs and present in still others that were not adapted to guinea pigs, suggesting that alternative mutations may underlie this particular phenotype. A previously described transmembrane domain (positions 60 through 76) (54) which tolerated conservative amino acid substitutions at all positions except invariant residues L64, L68, A70, and I72 was also variable in this study. Our data revealed limited variability in the potential 3A T-cell epitope that was previously identified between positions 21 and 35 (10). The variable nature of 3A suggests that it may be highly informative for epidemiologic or forensic purposes. Additionally, the likely role for 3A in virulence and host range suggests that interactions with host factors underlie 3A's variability and the diversifying selection predicted to act upon it.

(v) 3B.

FMDV is the only picornavirus to encode multiple 3B proteins (3B1, 23 aa; 3B2, 24 aa; and 3B3, 24 aa), whose homologue in poliovirus primes genomic RNA synthesis during virus replication (26, 85). In our study, 3B1, 3B2, and 3B3 were present in all FMDV isolates. The motif GPYXGP (except for the substitution GPYXRP in Sat 1/20 Rv 11/37), which contains a Y residue homologous to the poliovirus Y3 residue involved in phosphodiester linkage of 3B to the 5′ end of the viral genome, was invariant in the N terminus of each protein (5) (Table 7). The 3B C-terminal regions contained more amino acid variability, including the majority of observed nonconservative substitutions and the fewest invariant residues. Notably, 3B3 was highly conserved in all isolates examined, supporting previous experimental evidence indicating that only this 3B isoform is essential for virus viability (26, 84). 3B1 and 3B2 were more variable, and in fact, contained residues predicted here to undergo diversifying selection (3B1 residues 4 and 11 and 3B2 residues 17, 18, and 19). These data, in conjunction with experimental data indicating that the deletion of 3B1 and 3B2 may affect FMDV virulence and host range (84), suggest that, similar to that of 3A, 3B1 and 3B2 protein variability reflects host range-specific functions.

TABLE 7.

FMDV 3B (VPg) variability

Serotype	Sequencea
3B1	G	P	Y	A*	G	P	L	E	R	Q	K*	P	L	K	V	R	A	K	L	P	Q	Q		E
				T23			M3				Q14			R2	L14	K38	T5	E4	P1		R6	A14
				S2			V1				R19						V2	R1	F1		K4	H5
											I1											R1
3B2	G	P	Y	A	G	P	M	E	R	Q	K	P	L	K	V	K	A*	K*	A*	P	V	V	K	E
				V3	R1		L23	D1	K10		Q18			R3	L15	R4	V29	R21	L15	S1		A16	R1
																	T4		T3			I4
																	E1		N/P/V1
3B3	G	P	Y	E	G	P	V	K	K	P	V	A	L	K	V	K	A	K	N	L	I	V	T	E
				D1			M1	R1	R1		A1				L1	R1	T2	R1	A15	P15
																				M3

^aThe most frequent amino acid for each position is shown in the top row for each serotype; alternative amino acids and the number of isolates with that particular residue (subscripts) are listed below. SAT serotype-specific residues are shown in italics. Asterisks indicate positions predicted to undergo diversifying selection by codeml analysis.

(vi) 3C.

FMDV 3C^pro is a 213-amino-acid proteinase involved in the cleavage of most viral precursor proteins and of host factors associated with translation (eIF4A and eIF4G) and transcription (histone H3). A previous mutational analysis of 3C^pro identified the catalytic triad residues C163, H46, and D84 and residues critical for proper protein folding, i.e., D84 and Y136, all of which were invariant in the isolates examined here (34). A previously undescribed invariant motif (VKGQDMLSDAALMVLH) and a predicted transmembrane domain were identified at positions 76 to 91 and 27 to 44, respectively. A high degree of overall 3C^pro conservation indicates a limited tolerance for mutations, likely due to significant structural/functional constraints. Despite the 3C^pro conservation, 36 SAT-specific residues were identified, suggesting that distinct selective pressures exist for the two virus groups.

(vii) 3D.

The 469-amino-acid viral RNA-dependent RNA polymerase 3D^pol, responsible for generating minus- and plus-sense genomic RNA, is one of the most conserved FMDV proteins. Our analysis indicated that although it is conserved, 3D^pol is more tolerant of substitutions than was previously reported (56), as our results extended the proportion of variable residues from 8.6% to 26%. D245, N307, and G295, which are essential for maintaining the functional integrity of the picornaviral polymerases (56), were invariant in FMDV 3D^pol, along with other residues described as being highly conserved among all RNA-dependent RNA polymerases (31), including the NTP-binding residues G337, D338, and D339 (115). The previously described picornaviral polymerase peptide motifs KDELR, PSG, YGGD, and FLKR were conserved among all FMDV isolates analyzed (18, 51). Finally, the three previously described hypervariable, hydrophobic antigenic regions of 3D^pol (aa 1 to 12, 64 to 76, and 143 to 153) were also variable in all virus isolates examined here (31).

3′ UTR.

The picornaviral 3′ UTR binds several viral and host proteins and is believed to contain structural cis-acting elements required for negative-strand RNA synthesis (4). Removal of the terminal poly(A) tract or mutagenesis of structural elements abrogates the infectivity of FMDV infectious clones (98). FMDV 3′ UTR sequences stimulate virus-specific, IRES-dependent translation (67) and likely affect other aspects of viral replication, including genome circularization (37).

The FMDV 3′ UTRs were highly variable in length among the isolates, ranging from 85 to 101 nt. A secondary structure analysis of the FMDV 3′ UTR confirmed the Y shape which is also predicted for other picornaviral 3′ UTRs, suggesting that the structure plays an important role in the 3′ UTR function (Fig. 1D and data not shown). A previously unidentified motif was located at the vertex of the Y structure between positions 37 and 61 (Fig. 1C and D). In some cases, however, structural features which could conceivably affect the efficiency of ribosome/RNA dissociation and translation were observed (for example, the exceptionally long stem of A10 Holland/42 and the short stem of O2 Brescia/47) (data not shown).

FMDV phylogenetic and recombination analysis.

To date, phylogenetic analyses have been performed largely on FMDV sequences from the 1D coding region. These analyses have permitted the discrimination among serotypically related FMDV isolates (55). Split decomposition, a clustering technique used previously on 1D sequences to suggest the occurrence of quasispecies evolution in Euroasiatic FMDV (22), was used to examine the complete genome sequences of all 103 isolates described here (Fig. 4A). The results indicated that complex phylogenetic relationships exist between members of different serotypes, including relationships between the A24 Argentina/65 and European/South American O1 viruses; between A12 Valle strain 119 and C Waldman strain 149; between the O1 M11, O2 Brescia/47, and O3 Venezuela/71 viruses and European serotype A viruses; between 05India/62 and Asia1 viruses from Israel; among six SAT1 and three SAT3 viruses; and most notably, between SAT1/7 Isrl 4/62 and SAT2/3 Kenya 11/60 and other SAT1 and SAT2 viruses (Fig. 4A and data not shown).

FIG. 4.

FMDV phylogenetic and recombination analysis. (A) One hundred three FMDV genomes were aligned with ClustalW and visually screened for reticulated branching patterns by split decomposition cluster analysis, using SplitsTree with Hamming distances. Numbers at terminal nodes correspond to the isolates presented in Table 1, arcs and labels indicate serotype-specific groups, similar colors indicate isolates exhibiting obvious interserotypic relationships, red stars indicate SAT viruses with unique nonstructural protein sequences, and bars represent estimated distances. Similar results were obtained with alternative corrections for multiple substitution (data not shown). Subgenomic split decomposition analysis was conducted on structural protein (1A to 1D) (B) and nonstructural protein (L^pro and 2A to 3D) (C) coding regions aligned with Dialign. (D to G) Similarity (upper box) and bootscanning (lower box) plots were generated with a query isolate against reference isolates aligned with Dialign, indicating corrected percent similarities and bootstrap frequencies as percentages of permuted trees, respectively. Polyprotein schematics are shown above the similarity plots. Query isolates are indicated at the left within similarity plots and included the following: (D) SAT 1/7 Isrl 4/62 (no. 96), (E) A24 Argentina/65 (no. 3), (F) O1 M11 (no. 74), and (G) A10 Holland/42 (no. 25). Reference isolates are indicated individually or as groups in color, as labeled at the right within similarity plots, and included the following: (D) red, SAT 2/3 Kenya 11/60 (no. 99); green, SAT1 viruses (no. 88 to 95); blue, SAT2 and SAT3 viruses (no. 97 and 98 and 100 to 103, respectively); black, A viruses (no. 30, 38, and 46), O viruses (no. 80 and 86), C viruses (no. 57), and Asia1 viruses (no. 51); (E) green, A14 Spain/59 (no. 36); orange, A29 Peru/69 (no. 22); blue, A viruses (no. 2, 10, 18, 19, 22, 25, 27, 30, 38, 39, 42, 46, and 47); red, O1 Campos/58 (no. 70); gray, O viruses (no. 74, 78, 80, 83, and 87); (F) red, A10 Holland/42 (no. 25); pink, A1 Bayern/Bavaria (no. 27); green, A4 WG/72 (no. 30); blue, A viruses (no. 2, 10, 18, 19, 22, 38, 39, 42, 46, 47); black O1 Campos/58 (no. 70); orange, O1 Vallee/39 (no. 77); gray, O viruses (no. 63, 75, 77, 78, 80, 83, 87); (G) red, A12 Valle Strain 119 (no. 26); pink, A1 Bayern/Bavaria (no. 27); orange, A3 Mecklenburg/68 (no. 28); green, A4 WG/72 (no. 30); blue, A viruses (no. 2, 3, 5, 10, 13, 17, 19, 21, 22, 29, 32, 33, 36 to 39, 41, 46, and 47). Aligned polyprotein-encoding nt positions are indicated below the bootscan plots. Colored bars within similarity plot boxes indicate fragments identified by GENECONV as having significant global permutation values (P < 0.05) shared between the query sequence and particular reference sequences.

The data from analyses of individual protein-encoding regions indicated that incongruent tree topologies exist for different genomic regions (data not shown). An analysis of 1B, 1C, 1D, or all of the structural protein-encoding regions together indicated few obvious phylogenetic incongruities, placing viruses into serotypic, temporal, and/or geographic groups (Fig. 4B and data not shown). In general, analyses of the entire structural protein-encoding region improved bootstrap/confidence values relative to 1D analysis alone (data not shown). Conversely, analyses of the 5′ UTR and regions encoding individual or all nonstructural proteins failed to produce the serotype-inclusive groups observed with capsid protein sequences, often yielding unresolved, star-like topologies among diverse serotypes (Fig. 4C and data not shown). Consistent with the complex relationships illustrated in Fig. 4A, analyses of nonstructural protein-encoding regions indicated groupings of serotypically disparate viruses, which extended previous but limited data (54, 110, 113) (Fig. 4C). Similarity plots and bootscanning analysis further indicated that several of these interserotypically related viruses contained nonstructural protein-encoding regions most similar to those of viruses with divergent capsid protein-encoding sequences, suggesting the occurrence of intertypic recombination during the history of these lineages (Fig. 4D to F and data not shown). Examples of potential intratypic recombination were also noted, more obviously in nonstructural protein-encoding regions but also in structural protein-encoding regions, as has been suggested previously (Fig. 4F and G and data not shown) (35, 109).

Taken together, these data suggest that FMDV capsid protein sequences may undergo intertypic recombination infrequently relative to that in other genomic regions, which conceivably undergo complex recombination events and fail to display serotype-specific phylogenetic relationships. These observations are consistent with and extend previous reports of FMDV recombination, which has been inferred from incongruent L^pro, 3C^pro, and 1D topologies and observed (and suggested to predominate) in C-terminal genomic regions (52, 57, 101, 113). Similar observations have recently been made for enterovirus genomes, which were suggested to undergo relatively extensive recombination in nonstructural gene regions while generally maintaining serospecific monophyly (79). These observations, including the inability to reliably define viral relationships based on nonstructural protein-encoding sequences and data suggestive of recombination, raise interesting questions about FMDV genome evolution in nature and the relative contribution of recombination to the generation of FMDV genetic and population diversity.

Notably, the analysis described here identified a novel SAT lineage represented by SAT1-7 Isrl 4/62 and SAT2-3 Kenya 11/60, which contain nonstructural protein-encoding regions that are divergent from each other but are also clearly distinct from those of the other SAT and Euroasiatic lineages presented here (Fig. 4C). The distinct nature of nonstructural proteins from these structurally/serotypically SAT-like viruses was supported by a previously available genome sequence similar to that of SAT2-3 Kenya (SAT2 Ken/3/57; 97% overall nucleotide identity) (Table 1) and by previous data indicating that the SAT2 Ken/3/57 L^pro- and 3C-encoding regions may be distinct from those of other SATs (102). Taken together, these results suggest that more FMDV genome diversity may exist in nature than is currently indicated by serology or 1D sequence analysis.