Molecular detection of kobuviruses and recombinant noroviruses in cattle in continental europe

Introduction and Objectives
Noroviruses (NoV) and Kobuviruses (KoV), belong to the family Caliciviridae genus Norovirus and to the family Picornaviridae genus Kobuvirus respectively. Both have a single stranded positive-sense RNA genome. They both infect the gastrointestinal tract of different animal species including human beings. Two NoV and one KoV prototype strains have been already identified in the bovine (Bo) species: Jena virus (JV) and Newbury 2 (NB2) for BoNoV; U1 for BoKoV. Genogroup (G) III gathers all BoNoV strains and is further subdivided into two genotypes where viruses genetically related to JV and NB2 are assigned to the genotype 1 and 2 respectively. Recombination is a common event in NoV and is usually reported near the overlapping region between open reading frame (ORF) 1 (end of the polymerase gene) and ORF2 (beginning of the single capsid protein gene). Two GIII.1/GIII.2 BoNoV recombinant strains have been described including the recombinant strain Bo/NoV/Thirsk10/00/UK (Thirsk10), identified in the year 2000 in Great Britain. To our knowledge, no other genetically related strains have been reported since [1]. Bovine KoV were detected by RT-PCR in stool samples of healthy calves from Japan, in samples from diarrhoeic calves from Thailand [2] and were also identified very recently in Hungary. Bovine NoV prevalence studies performed in different areas have shown the predominance of the GIII.2 genotype but this could reflect a GIII.1 specificity failure in the RT-PCR methods. The aim of this study was to screen cattle stool samples with two primer sets targeting the polymerase and the capsid region. The primer pair targeting the capsid region was designed based on a GIII.1 sequence in order to improve their detection.
Materials and methods
A stool bank (n=300) was created with calf and young stock diarrhoeic samples from five provinces in Belgium (Hainaut, Liège, Namur, Luxembourg, Walloon Brabant) and received from a Belgian diagnostic laboratory through the year 2008. Viral RNA extraction was performed and one step RT-PCR was carried out on 2 µl of each viral RNA extraction using the CBECu-F/R primers (nucleotidic position on JV: 4543-4565 and 5051-5074) and a primer pair, named AMG1-F/R, designed from the JV genomic sequence (F: tgtgggaaggtagtcgcgaca, nucleotidic position on JV: 5012-5032; R: cacatgggggaactgagtggc, 5462-5482). Combined approaches with the CBECu-F and AMG1-R primers, additional internal primers (F2: atgatgccagaggtttcca, position on JV: 4727-4745; R2: gcaaaaatccatgggtcaat, 5193-5211) or CBECu-F and a polyTVN-linker were also carried out on some positive samples. RT-PCR products were directly sequenced twice or cloned before sequencing. Sequencing was carried out at the GIGA facilities of the University of Liège with BigDye terminator kit. Nucleotidic sequences were analysed with the BioEdit software. Nucleotidic similarity with the NCBI genetic database was assessed using the BLAST tool. Phylogenetic inference was performed with the MEGA software. Phylogenetic tree was constructed by neighbour-joining analysis where evolutionary distances were computed using the Maximum Composite Likelihood method. The confidence values of the internal nodes were calculated by performing 1,000 replicate bootstrap values. Genetic recombination was analysed with the Simplot software and the Recombinant Detection Program.
Results
Twenty-eight positive samples were identified in the 300 samples: 24 and 23 BoNoV sequences with the CBECu and AMG1 primer pairs respectively, giving a combined apparent molecular prevalence of 9.33% (CI 95%: [9.27; 9.38%]). Using BLAST, three sequences amplified with CBECu-F/R (BV164, BV362, and BV416) were genetically more related to the GIII.1 JV and Aba Z5/02/HUN sequences and one (BV168) to the recombinant strain Thirsk10. The others were genetically related to GIII.2 BoNoV. All the sequences amplified with AMG1-F/R but one genetically matched with GIII.2 BoNoV. The AMG1-amplicon of the BV416 sample matched with the recombinant strain Thirsk10. A 2410 nucleotide (nt)-large genomic sequence was obtained from BV416 with CBECu-F/TVN linker, which was a recombinant sequence genetically related to the Thirsk10 strain. This result was confirmed by phylogenetic and by Simplot analysis. The potential recombination breakpoint of BV416 was located near or within the ORF1/ORF2 overlapping region depending on the bioinformatic program used. Comparison between its different genomic regions and the JV, Newbury2 and Thirsk10 genomic sequences showed that the polymerase region of BV416 was genetically more related to the GIII.1 than to the recombinant strain. F2/R2 amplicons from BV164 and BV362 were genetically related to GIII.2 and GIII.1 BoNoV respectively. Surprisingly, three amplicons obtained with the combined primer pair CBECu-F/AMG1-R on BoNoV positive samples at the expected molecular weight did not match genetically with BoNoV but did so with different genomic regions of the BoKoV U1 strain (86%, 92% and 93% of nucleotidic identity by BLAST for BV228, 250 and 253 respectively on sequences of about 500-700 nt).
Discussion and conclusions
In this study, very few genotype 1 BoNoV were identified (BV362 was the sole GIII.1 sequence obtained in the ORF1/2 overlapping region), confirming results reported in a previous study on BoNoV infection in the same area [3]. A recombinant status was clarified for BV416. Co-infection with GIII.1 and GIII.2 BoNoV was evidenced in the BV164 sample but could not be excluded in the BV168 sample because an overlapping sequence could not be obtained, although genetic analyses related its CBECu-F/R sequence to the Thirsk10 sequence. These results raise issues about the genetic characterization by primers targeting either the polymerase region or the capsid region. By exclusion of the potential recombination breakpoint, these primers can lead to the misclassification of strains and to the underestimation of circulation of recombinant strains. Multiple alignment and bioinformatic analysis performed with JV, Aba Z5, NB2, Thirk10 and BV416 sequences has suggested a recombination breakpoint for BV416 located near the ORF1/ORF2 overlapping region and one quite similar to those determined for the Thirsk10 strain. Nevertheless the greater similarity of BV416 with the Jena and Aba Z5 viruses in the polymerase region and the exact localization of the recombination breakpoint suggest another origin or genetic evolution than the Thirsk10 strain. The identification, in geographically and temporally different samples, of sequences that could be genetically related to the recombinant Thirsk10 strain suggests at least that Thirsk10-related strains circulate in the north European cattle population. Furthermore, the low detection rate of GIII.1 BoNoV could reflect an evolution of the viral population pattern to the benefit of the Thirsk10-related and genotype 2 strains in the studied region. To date, BoKoV-related sequences have been very rarely identified, and in only three countries (namely Japan, Thailand and Hungary). Their detection in another European country suggests their wider distribution, making them at least emerging bovine viruses in the studied region. In conclusion, prevalence studies on BoNoV using RT-PCR assays, even targeting relatively well conserved genomic regions, need to take into account in their protocols both their high genetic variability and their relative genetic proximity with other viruses, in order to maximize sensitivity and specificity. This study also showed that recombination events could lead to emerging strains in the BoNoV population, as already found for HuNoV. The molecular detection of bovine kobuvirus-related sequences in the studied area extends the distribution of these viruses in Europe.