Journal of General Virology (2006), 87, 1509–1519
DOI 10.1099/vir.0.81757-0
Evolution of Bovine herpesvirus 4: recombination and transmission between African buffalo and cattle Benjamin Dewals,1 Muriel Thirion,1 Nicolas Markine-Goriaynoff,1 Laurent Gillet,1 Katalin de Fays,1 Fre´de´ric Minner,1 Virginie Daix,1 Paul M. Sharp2 and Alain Vanderplasschen1 1
Immunology–Vaccinology (B43b), Department of Infectious and Parasitic Diseases, Faculty of Veterinary Medicine, University of Lie`ge, B-4000 Lie`ge, Belgium
Correspondence Alain Vanderplasschen
2
[email protected]
Institute of Genetics, University of Nottingham, Queens Medical Centre, Nottingham NG7 2UH, UK
Received 9 December 2005 Accepted 17 February 2006
Bovine herpesvirus 4 (BoHV-4) has been isolated from cattle throughout the world, but virological and serological studies have suggested that the African buffalo is also a natural host for this virus. It has previously been found that the Bo17 gene of BoHV-4 was acquired from an ancestor of the African buffalo, probably around 1?5 million years ago. Analysis of the variation of the Bo17 gene sequence among BoHV-4 strains suggested a relatively ancient transmission of BoHV-4 from the buffalo to the Bos primigenius lineage, followed by a host-dependent split between zebu and taurine BoHV-4 strains. In the present study, the evolutionary history of BoHV-4 was investigated by analysis of five gene sequences from each of nine strains representative of the viral species: three isolated from African buffalo in Kenya and six from cattle from Europe, North America and India. No two gene sequences had the same evolutionary tree, indicating that recombination has occurred between divergent lineages; six recombination events were delineated for these sequences. Nevertheless, exchange has been infrequent enough that a clonal evolutionary history of the strains could be discerned, upon which the recombination events were superimposed. The dates of divergence among BoHV-4 lineages were estimated from synonymous nucleotide-substitution rates. The inferred evolutionary history suggests that African buffalo were the original natural reservoir of BoHV-4 and that there have been at least three independent transmissions from buffalo to cattle, probably via intermediate hosts and – at least in the case of North American strains – within the last 500 years.
INTRODUCTION Bovine herpesvirus 4 (BoHV-4) is a gammaherpesvirus that has been isolated throughout the world from healthy cattle, as well as those exhibiting a variety of diseases (Li et al., 2005; Markine-Goriaynoff et al., 2003b; Thiry et al., 1992). The widespread distribution of BoHV-4 in cattle (Bos taurus and Bos indicus) populations justified the conclusion that the bovine species is the natural host; hence the nomenclature of the virus. However, BoHV-4 has also been detected in a variety of other ruminants (Todd & Storz, 1983; Van Opdenbosch et al., 1986). In particular, BoHV-4 infection is very common among wild African buffalo The GenBank/EMBL/DDBJ accession numbers for the nucleotide sequences determined in this work are AY847305–AY847311 and AY847322–AY847348. Supplementary tables with primer details and GenBank accession numbers for the BoHV-4 sequenced regions are available in JGV Online.
0008-1757 G 2006 SGM
Printed in Great Britain
(Syncerus caffer). An early survey performed mainly within one reserve in Kenya revealed that BoHV-4 can be isolated from the blood of healthy wild African buffaloes with a high frequency (25 %) and that almost all (94 %) animals tested had antibodies against BoHV-4 (Rossiter et al., 1989). More recently, our seroprevalence study of 400 sera from wild African buffaloes from numerous locations in eastern and southern Africa revealed that, independent of their geographical origin, members of this species exhibit a seroprevalence of anti-BoHV-4 antibodies of around 70 % (Dewals et al., 2005), a rate much higher than is generally observed in cattle populations (16–18 %) (Wellenberg et al., 1999). These observations suggest that African buffalo may be the original host species of BoHV-4 and that other species, including domestic cattle, may have acquired the virus more recently. An independent observation supports the role of the African buffalo as the original host species of BoHV-4. 1509
B. Dewals and others
The Bo17 gene of BoHV-4 is the only viral gene known to date that encodes a homologue of the cellular core 2 b-1,6-Nacetylglucosaminyltransferase-mucin type (Vanderplasschen et al., 2000). Our phylogenetic study demonstrated that the Bo17 gene is related significantly more closely to the cellular homologue from African buffalo than to that from cattle, indicating that the viral gene was acquired from a recent ancestor of the African buffalo (Markine-Goriaynoff et al., 2003a). Furthermore, the gene-acquisition event was dated at around 1?5 million years ago, long after the separation of the Bos and Syncerus lineages and long before the domestication of cattle. The sequences of the Bo17 gene from nine BoHV-4 strains, three from African buffalo and six from cattle, were split into two clades according to the species of origin (Markine-Goriaynoff et al., 2003a), consistent with restriction-profile analyses showing a clear differentiation between BoHV-4 strains isolated from the two species (Bublot et al., 1991; Markine-Goriaynoff et al., 2003a; Rossiter et al., 1989). For Bo17, the divergence between the two viral lineages was estimated to have occurred about 0?7 million years ago, suggesting a relatively ancient transmission of BoHV-4 from the ancestor of the African buffalo to Bos primigenius, the wild ancestor of domesticated cattle. Among the cattle strains, the one isolated from zebu (B. indicus) formed an outgroup to those from the taurine cattle (B. taurus) lineage and the split was estimated to have occurred at about the same time as the divergence of the ancestors of B. taurus and B. indicus, around 0?2–0?3 million years ago, consistent with a host-dependent split between zebu and taurine BoHV-4 strains. Thus, whilst these data indicated that BoHV-4 has been co-evolving with the African buffalo for at least the last 1?5 million years, they also pointed to infection of cattle since long before they were domesticated, around 10 000 years ago. These interpretations of the evolutionary history of BoHV-4 must be treated with caution because they rely on the analysis of a single gene, Bo17. Recombination has been reported in numerous species of herpesviruses (Burrows et al., 1996; Norberg et al., 2004; Poole et al., 1999; Thiry et al., 2005; Walling & Raab-Traub, 1994). The consequence of recombination is that different genes may have different
evolutionary histories. Therefore, we set out to investigate the phylogenetic relationships among the nine BoHV-4 strains that were used in our former study (MarkineGoriaynoff et al., 2003a) by characterizing four more gene sequences from three different regions distributed across the genome. The results demonstrate that BoHV-4 evolution has indeed involved recombination among divergent strains and that the evolutionary history of the Bo17 gene is not typical of that of the genome as a whole. These data necessitate a new interpretation of the origins of the various lineages of BoHV-4 distributed worldwide among cattle.
METHODS Virus strains. Nine strains representative of the BoHV-4 species were used (Table 1). The complete sequence of strain 66-p-347 has been published previously (Zimmermann et al., 2001). Sequences from the other eight strains were obtained in this study. Amplification and sequencing. Three regions of the BoHV-4
genome – open reading frame (ORF) 16 (encoding a v-Bcl-2 homologue), Bo10 (encoding a putative glycoprotein unique to BoHV-4) and ORF 71 (encoding a v-FLIP homologue)–ORF 73 (encoding a LANA homologue) (ORF 71 and ORF 73 plus the intervening intergenic region) – were amplified by using PCR. The positions of these regions within the genome are shown in Fig. 1 and the primers used for the amplifications are given in Supplementary Table S1 (available in JGV Online). PCR products were subsequently purified (Wizard PCR Preps DNA purification system; Promega) and sequenced by primer walking using a BigDye Terminator v3.1 cycle sequencing kit (Applied Biosystems). An ABI Prism 3730 DNA Analyser (Applied Biosystems) was used for analysis of samples. sequence accession numbers. The sequences reported in this paper have been deposited in GenBank. The accession numbers of these sequences, together with those of previously published sequences included in the analysis, are listed in Supplementary Table S2 (available in JGV Online).
Nucleotide
Sequence analysis. Sequences were aligned by using CLUSTAL W
(Thompson et al., 1994), with minor manual adjustment. The numbers of non-synonymous (KA) and synonymous (KS) substitutions per site were estimated by the method of Li (1993). Phylogenetic trees were estimated from nucleotide sequences by the neighbourjoining (NJ) method (Saitou & Nei, 1987), using distances corrected by Kimura’s two-parameter method and 1000 bootstrap
Table 1. BoHV-4 strains used in this study Strain* V. test LVR 140 MOVAR DN599 66-p-347 M40 108, 130, Buf.
Host species
Country of isolation
Restriction profile
Reference source
Cattle (Bos taurus) Cattle (Bos taurus) Cattle (Bos taurus) Cattle (Bos taurus) Cattle (Bos taurus) Zebu (Bos indicus) African buffalo (Syncerus caffer)
Belgium Belgium Hungary USA USA India Kenya
MOVAR-like MOVAR-like MOVAR-like DN599-like DN599-like Unclassifiable Unclassifiable
Thiry et al. (1981) Wellemans et al. (1983) Bartha et al. (1966) Mohanty et al. (1971) Storz (1968) Moreno-Lopez et al. (1989) Rossiter et al. (1989)
*The BoHV-4 strains used in this study are representative of the BoHV-4 species. They were isolated in different countries and from various host species. They belong to the three groups defined based on restriction profile. 1510
Journal of General Virology 87
Evolution of BoHV-4
Fig. 1. Regions of the BoHV-4 genome analysed in this study. The structure of the BoHV-4 genome is shown, with the positions and orientations of the five genes within the long unique region (LUR) and the lengths of the sequences in strain 66-p-347.
replicates, as implemented in CLUSTAL W. Phylogenetic trees were also estimated by the maximum-likelihood (ML) method, implemented using DNAML from the PHYLIP package (J. Felsenstein; http:// evolution.genetics.washington.edu/phylip.html). No significant differences were found between the NJ and ML trees; the results from the NJ analyses are presented. For most of the sequences analysed, there was no suitable outgroup available and, so, each tree was rooted at its midpoint; for the one case where an outgroup was available (the Bo17 gene), midpoint rooting yielded the same result as outgroup rooting (Markine-Goriaynoff et al., 2003a). Observed substitution rates were consistent with midpoint rooting for the other genes. Recombination was inferred from topological differences between the trees for different genes and was not dependent on the position of the root (except in one case, discussed specifically). Mosaicism of sequences due to recombination was investigated by analysis of phylogenetically informative sites; the statistical significance of putative recombination breakpoints was assessed by a permutation test (Hatwell & Sharp, 2000). The protein encoded by the Bo10 gene was found to contain a highly variable, Ser-rich region that was difficult to align with confidence. This region, corresponding to codons 38–94 in the 66-p-347 Bo10 sequence, was excluded from the comparative analyses.
RESULTS To investigate the diversity and evolution of BoHV-4 strains isolated from African buffalo and cattle, nine strains representative of the BoHV-4 species (Table 1) were compared for four different regions distributed across the genome (Fig. 1). Three strains were isolated from buffalo in Kenya, five from taurine cattle from Europe or North America and one from zebu cattle in India; these nine strains are representative of the three groups of BoHV-4 strains defined on the basis of restriction profiles. From left to right, the genomic regions selected for this study were ORF 16, the Bo10 ORF, ORF 71–ORF 73 and the previously characterized Bo17 ORF. The ORF 71–ORF 73 region includes two genes and the region between them (IG 71–73). A putative 74 codon ORF, called Bo13, overlapping ORF 71 and ORF 73 and thus spanning the IG 71–73 region, has been described in the genome of strain 66-p-347 (Zimmermann et al., 2001). However, we found an insertion of 1 nt within IG 71–73, resulting in a frameshift, in six of the nine viral strains. Also, in a comparison of strains 66-p-347 and http://vir.sgmjournals.org
M40 (the latter lacks the frameshift), there was an excess of non-synonymous over synonymous substitutions. These observations suggest that this ORF does not function as a gene. In each region, the three European strains (V. test, MOVAR and LVR 140) were very similar, differing by no more than 2 nt and showing overall diversity (summed across regions) of 60 % are shown. Asterisks highlight recombination events described in the text. Horizontal branches are drawn to scale. Bar, 0?01 nucleotide substitution per site.
the trees from individual sequences indicate the impact of occasional inter-strain recombination events. To facilitate interpretation of the differences among the trees, the extent of synonymous nucleotide substitution (KS) between pairs of strains was calculated for each gene. Synonymoussubstitution rates are expected to be similar for different genes and, so, differences among genes in KS values for the same pair of strains may reflect differences in the time since common ancestry, due to recombination. 1512
The ORF 16 tree differed from the concatenated tree in that the topological positions of the 130 and Buf. strains switched (Fig. 2b). The extent of synonymous substitution between Buf. and 108 was lower than for any other gene (Table 2), suggesting that the ancestor of the Buf. strain acquired DNA by recombination from a recent ancestor of the 108 strain. The Bo10 tree differed from the concatenated tree in two, more subtle, ways. First, the North American strains were very close to the European strains, with KS values less than half those seen for other genes, suggesting that a recombination event occurred between the ancestors of the two groups. Second, midpoint rooting placed the 130, 108 and Buf. strains together with M40 on one side of the root. Although an alternative rooting with the same topology as the concatenated tree is possible, the KS values for Bo10 between M40 and the buffalo strains are lower than those for other genes (except Bo17) and lower than those for Bo10 between M40 and the taurine strains. This suggests the occurrence of a recombination event between the ancestor of the buffalo strains and an ancestor of the M40 strain. However, the KS values do not provide a clear indication of the direction of the exchange in this case. In the ORF 71 tree, the North American strains clustered within the African buffalo clade (Fig. 2b), suggesting that the ancestor of the North American strains acquired the ORF 71 sequence from an ancestor of the 108 and 130 strains. In the ORF 73 tree, the three buffalo strains did not form a monophyletic clade. This topology could be explained as the result of an acquisition of sequence by an ancestor of the 108 and 130 strains from the common ancestor of the European and North American strains, but further investigation of this region (see below) revealed a complex mosaic of evolutionary histories within this gene. Finally, the Bo17 tree differed from all others in placing the M40 strain close to the clade of taurine cattle viruses, and Bo17 KS values between M40 and the five taurine viruses were unusually low. This suggests that an ancestor of the M40 strain acquired sequence encompassing the Bo17 ORF from a common ancestor of the North American and European strains. The three groups of strains defined previously from genome-restriction profiles comprised (i) the European strains, (ii) the North American strains and (iii) others, termed ‘unclassifiable’ (Markine-Goriaynoff et al., 2003a). It is now apparent that the unclassifiable strains are not a true group, as they are much more heterogeneous than the two other groups, are not monophyletic and have complex relationships due to interlineage recombination (Fig. 2). The ORF 71–73 region of the BoHV-4 genome contains recombination breakpoints The three contiguous genomic regions ORF 71, IG 71–73 and ORF 73 yielded three different trees (Fig. 2), suggesting the presence of (at least) two recombination breakpoints within this ORF 71–73 region. As recombination need not occur at the boundaries of genes, the distribution of informative sites within this region was examined in order Journal of General Virology 87
http://vir.sgmjournals.org
Table 2. Synonymous-substitution rates (KS) among BoHV-4 sequences Pairwise comparison
Estimated time of divergence (years)d
ORF 16
Bo10
ORF 71
ORF 73
Bo17
Average
Corrected averageD
Mean corrected average
0?0778 0?0778 0?0778 0?0671 0?0671 0?0886 0?0889 0?0887
0?0809 0?0809 0?0809 0?0809 0?0809 0?0633 0?0633 0?0543
0?0876 0?0878 0?0876 0?1003 0?1003 0?1012 0?1012 0?0954
0?0800 0?0800 0?0800 0?0838 0?0922 0?0754 0?0838 0?0636
0?0234 0?0234 0?0256 0?0166 0?0166 0?0568 0?0734 0?0570
0?0699 0?0700 0?0704 0?0697 0?0714 0?0771 0?0821 0?0718
0?0818 0?0819 0?0818 0?0837 0?0865 0?0884 0?0913 0?0826
0?0848
848 000
0?0666 0?0852 0?0758 0?0666 0?0852 0?0758 0?0666 0?0852 0?0758 0?0558 0?0743 0?0650 0?0558 0?0743 0?0650
0?0849 0?0849 0?0604 0?0849 0?0849 0?0604 0?0849 0?0849 0?0604 0?0903 0?0903 0?0603 0?0903 0?0903 0?0603
0?0714 0?0714 0?0768 0?0714 0?0714 0?0770 0?0714 0?0714 0?0768 0?0204 0?0204 0?0154 0?0204 0?0204 0?0154
0?0422 0?0504 0?0447 0?0422 0?0504 0?0447 0?0422 0?0504 0?0447 0?0382 0?0464 0?0486 0?0463 0?0383 0?0534
0?0817 0?0989 0?0820 0?0817 0?0989 0?0820 0?0842 0?1014 0?0845 0?0697 0?0865 0?0700 0?0697 0?0865 0?0700
0?0694 0?0782 0?0679 0?0694 0?0782 0?0680 0?0699 0?0787 0?0684 0?0549 0?0636 0?0519 0?0565 0?0620 0?0528
0?0732 0?0852 0?0698 0?0732 0?0852 0?0699 0?0741 0?0860 0?0705 0?0628 0?0804 0?0612 0?0628 0?0804 0?0628
0?0732
732 000
0?0190 0?0190 0?0190 0?0190 0?0190 0?0190
0?0089 0?0089 0?0089 0?0089 0?0089 0?0089
0?0707 0?0707 0?0708 0?0708 0?0707 0?0707
0?0342 0?0421 0?0342 0?0421 0?0342 0?0421
0?0200 0?0200 0?0200 0?0200 0?0222 0?0222
0?0306 0?0321 0?0306 0?0322 0?0310 0?0326
0?0244 0?0270 0?0244 0?0270 0?0251 0?0278
0?0260
260 000
0?0084 0?0083 0?0169
0?0251 0?0251 0?0000
0?0259 0?0259 0?0000
0?0408 0?0488 0?0077
0?0245 0?0406 0?0245
0?0249 0?0297 0?0098
0?0252 0?0305 0?0098
0?0279
279 000
0?0098
98 000
1513
*KS values were calculated according to the method of Li (1993). DCorrected KS averages were calculated after exclusion of KS values generated by pairwise comparisons of sequences resulting from recombination events. The latter are presented in bold. dTime of divergence was estimated as follows: [mean corrected average/2(561028)] (Markine-Goriaynoff et al., 2003a).
Evolution of BoHV-4
M40 vs other strains M40 vs MOVAR M40 vs LVR 140 M40 vs V. test M40 vs 66-p-347 M40 vs DN599 M40 vs 130 M40 vs 108 M40 vs Buf. European and American strains vs African strains MOVAR vs 130 MOVAR vs 108 MOVAR vs Buf. LVR 140 vs 130 LVR 140 vs 108 LVR 140 vs Buf. V. test vs 130 V. test vs 108 V. test vs Buf. 66-p-347 vs 130 66-p-347 vs 108 66-p-347 vs Buf. DN599 vs 130 DN599 vs 108 DN599 vs Buf. North American strains vs European strains MOVAR vs 66-p-347 MOVAR vs DN599 LVR 140 vs 66-p-347 LVR 140 vs DN599 V. test vs 66-p-347 V. test vs DN599 Among African strains 130 vs Buf. 108 vs Buf. 130 vs 108
KS*
B. Dewals and others
to localize the putative breakpoints. The three European strains were considered as a single taxon, as were the two North American strains and two of the buffalo strains (108 and 130); thus, together with strains Buf. and M40, there were five taxa in the analysis, which may be connected by 15 possible unrooted tree topologies (Fig. 3). There were 24 phylogenetically informative sites, including a tetranucleotide at position 976–979 (GTCT in Buf. and M40; ACAA in the other strains) that we considered as a single site (Table 3). Each informative site was consistent with, i.e. would require a single mutation on, three of the 15 possible topologies; any of the 12 other topologies would require two mutations at the same site on different branches of the tree. Within ORF 71, there were five (of seven) informative sites consistent with topology 12, the tree found for ORF 71 (Fig. 2b); across the rest of the region, there was only one other such site, at the far end of ORF 73. Over the intergenic region and ORF 73, 12 (of 17) informative sites were consistent with topology 1, the tree found for ORF 73 (Fig. 2b). In a permutation test, this array of informative sites for topologies 1 and 12 was significantly non-random (P