In 2007, we collected 385 samples from dogs in Bari, Italy. A total of 183 samples were fecal specimens from household dogs (1–6 months of age) hospitalized with signs of mild to severe gastroenteritis (collection A); 88 were rectal swab specimens from clinically healthy juvenile and adult dogs housed in 4 separate shelters (collection B); and 114 were fecal swab specimens from asymptomatic household dogs (1–6 months of age) receiving routine care at 2 veterinary clinics (collection C).

By using reverse transcription PCR with the broadly reactive consensus primers for caliciviruses, p289/p290 (14), and strain 48–specific primers 493F–526R (Table), we detected vesivirus RNA in 1.1% (2/183), 64.8% (57/88), and 3.5% (4/114) of collection A, B, and C samples, respectively. Partial RNA-dependent RNA polymerase sequences, obtained by using the primers p289/p290 (14), were determined for 10 samples. The sequences shared closest nucleotide identity (90.7%–92.6%) with vesivirus strains 2117, Geel/2008, Allston/2008/USA, and Allston/2009, which were identified as contaminants in CHO cells; the sequences shared 73.6%–74.8% nt identity with CaCV strain 48. We determined the full-length genomic sequence (8,453 nt) of 1 of the canine vesivirus strains, Bari/212/07/ITA, by using consensus primers and 3′ and 5′ RACE protocols (Invitrogen Ltd, Milan, Italy); the sequence was deposited in GenBank (accession no. JN204722). Primers used for virus detection and sequencing are listed in the Table.

Figure 1

Figure 1. Genome organization of vesiviruses (VeVs). The genomic organization and open reading frame (ORF) usage are shown for representative viruses in the main VeV genetic groups: feline calicivirus (FCV) strain F9 (GenBank...

Figure 2

Figure 2. Phylogenetic tree based on the complete amino acid sequence of the capsid protein of vesiviruses (VeVs). The tree was constructed by using a selection of feline calicivirus strains and all of...

The full-length genomic sequence of strain Bari/212/07/ITA shared 89%–90% nt identity with sequences of viruses identified as CHO cell contaminants; it shared only 71.0% nt identity with the prototype CaCV strain 48. Three open reading frames (ORFs) were predicted by sequence analysis of the nucleotide sequence and by comparing results with the genomic organization and ORFs of other vesiviruses (Figure 1). ORF1 was 5,796 nt in length (nt 12–5807) and encoded a 1,921-aa polyprotein. The ORF1 stop codon was followed by 3 nt and then by the ORF2 start codon. ORF2 was 2,079 nt in length (nt 5811–7889) and encoded a 692-aa capsid protein. ORF3 was 405 nt in length (nt 7886–8290) and encoded a 134-aa protein. ORF2 overlapped with ORF3 by 4 nt (Figure 1). The 163-nt 3′ untranslated region of CHO cell–associated strain 2117 was shorter than the 235-nt region in strain 48. The full-length ORF1-encoded polyprotein of Bari/212/07/ITA shared highest amino acid identity (98.5%–98.9%) with CHO cell–associated vesiviruses; it shared 80.5% aa identity with the ORF1-encoded polyprotein of canine strain 48 and <56% aa identity with other vesiviruses. Within the ORF2-encoded protein, the capsid cleavage motive FRAES (aa 155–159) was conserved. As observed, in CHO cell–associated strain 2117 and canine strain 48, a 7-aa insertion (KTIKSQV) was present in the conserved region D. The full-length capsid protein of Bari/212/07/ITA shared 92.6%–92.9% aa identity with the CHO cell–associated vesivirus strains Allston/2009/USA and Allston/2008/USA from the United States, 90.0% aa identity with the CHO cell–associated isolate Geel/2008 from Belgium, 86.3% aa identity with the CHO cell–associated isolate 2117 from Germany (13), and only 70.3% aa identity with the prototype CaCV strain 48 isolated in Japan in 1990 (9). Strain Bari/2012/07/ITA shared <35.8% aa identity with other vesiviruses (Figure 2).

In previous investigations in dogs, strain 48–like vesiviruses were detected in only 2 (1.7%) of 119 samples: a fecal specimen from a dog with signs of enteric disease and a tonsillar swab specimen from a dog with respiratory disease (11). However, serologic investigations suggested that vesiviruses actively circulate in the canine population (11,12). In our virologic investigations, vesivirus RNA was detected in 64.8% (57/88) of dogs housed in 4 shelters but in only 1.1% (2/183) and 3.5% (4/114) of symptomatic and asymptomatic household dogs, respectively. These findings demonstrate that canine vesiviruses can be widespread in some settings or populations (e.g., kennels or shelters) where high population densities create favorable epidemiologic situations for circulation of some microorganisms.

Of interest, partial sequence analysis of the RNA-dependent RNA polymerase fragment of several of the dog-derived strains and full-length genomic sequencing of strain Bari/212/07/ITA showed that these canine vesiviruses were more similar to some vesiviruses found as contaminants of CHO cell cultures than to the prototype CaCV strain 48. Contamination of CHO cells by vesiviruses was documented in 2003 in Germany (13), 2008 and 2009 in the United States, and 2009 in Belgium. The 2008–2009 contamination in the United States was estimated to cost US $100–300 million in lost revenues to the biotech company because production was interrupted to provide adequate sanitation and maintenance of the bioreactors (15). In addition, these contaminations have raised concerns for the potential exposure of humans to these novel viruses because vesiviruses can readily cross the host species barrier (5).

Our findings show that genetically heterogeneous vesiviruses are common in dogs in Italy. Further studies are necessary to understand the ecology of this group of vesiviruses; that is, it must be determined whether vesiviruses circulate in humans or other animal species and, above all, whether they are associated with disease in humans, animals, or both.