Seleção de alvos genômicos para a filogenia de vírus da diarreia viral bovina (BVDV) e identificação de novo subtipo de BVDV-2

Abstract

Bovine viral diarrhea virus 1 (BVDV-1) and BVDV-2, classified within the genus Pestivirus, family Flaviviridae, are important pathogens of cattle worldwide. Pestiviruses are enveloped viruses and contain a positive-sense, single-stranded RNA molecule of approximately 12.3 kilobases as the genome. Up to date, 21 subtypes of BVDV-1 (a-u) and 4 of BVDV-2 (a-d) have been described. Although this genetic variability, BVDV isolates/strains have been frequently subtyped by phylogenetic analysis of the partial sequence of the 5' untranslated region (5'UTR) and/or coding regions such as Npro and E2. These analyses, however, can generate erroneous results, or with low statistical support, hindering the knowledge about the real diversity and circulation of BVDV subtypes. The best alternative to circumvent this obstacle would be to subtype the viruses by analyzing the complete genome (CG), a costly and unfeasible strategy for large-scale use. Thus, the first study investigated the most suitable genomic targets for subtyping of BVDV-1 and BVDV-2, comparing subtyping based on individual gene/region analysis and that based on whole genome (GC) analysis. The study was performed with 91 (BVDV-1) and 85 (BVDV-2) GC available in the GenBank database. The viruses were subtyped by analyzing their GC as well as the coding regions of the individual genes and the untranslated regions (complete 3' and 5'UTRs and partial 5'UTRs). Also, the geodesic distance between the tree generated by GC (reference) analysis and those generated by genomic region/UTR analyses were calculated. In general, analyses based on 3'UTR and 5'UTR showed the least reliable subtyping compared to GC analysis. For BVDV1, the phylogeny based on C, Erns, E1, E2, p7, NS2, NS3, NS4B, NS5A and NS5B was equivalent to that of GC. Regarding the BVDV-2 coding region, there was at least one noncompliance compared to the GC analysis, in all targets analyzed. After calculating the geodesic distance between GC and the coding region/UTRs trees, NS4B (for BVDV-1) and NS5A (BVDV-2) showed topology and edge lengths closer to the trees generated by GC analysis. In addition, 14 sequences of BVDV-2 could not be classified as subtype a, b or c. A second, more detailed study was then performed to investigate whether these sequences would represent a distinct BVDV-2 subtype. Initially, an "equivalence test" between phylogenetic analyses based on 85 complete/near-complete genomes (CNCGs) and their respective open reading frames (ORFs) proved that ORFs can be reliably used for BVDV-2 phylogeny. This made it possible to increase the data set to 139 sequences. Among these, seven sequences that could not be classified as BVDV-2 a-d formed a distinct cluster in all the phylogenetic trees analyzed. This cluster was suggested as a new subtype, called BVDV-2e. The BVDV-2e sequences also showed 44 amino acid changes compared to BVDV-2a-c. In conclusion, it is suggested that the NS4B and NS5A regions can be used as genomic targets for subtyping BVDV-1 and BVDV-2, respectively. In addition, a cluster of BVDV-2 was identified that may represent a new subtype (BVDV-2e), which could be investigated in future epidemiological and phylogenetic studies of BVDV.