Recombinant Canine Coronaviruses in Dogs, Europe

Subtype IIb originates from recombination with porcine transmissible gastroenteritis virus.

C oronaviruses (CoVs) (order Nidovirales, family Coronaviridae) are exceptionally prone to genetic evolution through accumulation of point mutations in genes encoding for structural and nonstructural proteins and homologous recombination among members of the same antigenic group (1). CoVs are organized by antigenic group. The fi rst group is subdivided into subgroups 1a and 1b. Subgroup 1a includes highly related viruses (i.e., porcine transmissible gastroenteritis virus [TGEV] and its derivative porcine respiratory coronavirus [PRCoV], feline coronaviruses [FCoVs], and canine coronaviruses [CCoVs]) (2). According to a proposal by the Coronavirus Study Group of the International Committee of Taxonomy of Viruses, and given the virus' close genetic relatedness (i.e., >96% aa identity in the key replicase 1ab domains), TGEV, FCoV, and CCoV should not be considered separate viruses. Instead, they should be considered host range variants of the same species (3).
CCoVs exemplify the genetic evolution and complexity of CoVs. To date, 2 CCoV genotypes are known, CCoV-I and CCoV-II (4); they share up to 96% of nucleotide identity in the viral genome (Lorusso et al., unpub. data) but are highly divergent in the spike protein gene (5). In addition, CCoV-I displays a novel open reading frame (ORF3) that encodes for a putative glycosylated protein, which is likely secreted from the infected cells (6). The 2 CCoV genotypes are commonly detected in the feces of dogs with diarrhea and often simultaneously infect the same dog (7). Both CCoV genotypes have been associated with mild clinical signs in pups, although hypervirulent strains have been reported to cause severe, fatal enteritis (8-10); a pantropic variant was responsible for systemic disease in natural and experimental conditions (11)(12)(13).
It has been postulated that TGEV originated from CCoV-II through cross-species transmission, which is supported by the high genetic relatedness between the 2 viruses and by the presence of ORF3 remnants in CCoV-II and TGEV genomes (6). More recently, novel CCoV-II strains, which likely originated from a double recombination event with TGEV, occurring in the 5′ end of the spike protein gene, have been isolated (14,15). Accordingly, genotype II has been further divided into 2 subtypes, CCoV-IIa and CCoV-IIb, including extant and TGEV-like CCoVs, respectively (14).
To assess the distribution of the TGEV-like CCoVs in canine populations of different geographic areas of Europe, we used a subtype IIb-specifi c reverse transcription-PCR (RT-PCR) assay. In addition, we evaluated the genetic relationship among the identifi ed strains by using sequence and phylogenetic analyses.

RNA Extraction
Specimens were homogenized (10% wt/vol) in Dulbecco modifi ed Eagle medium and subsequently clarifi ed by centrifuging at 2,500 × g for 10 min. For RNA extraction, 140 μL of the supernatants were then used by means of QIAamp Viral RNA Mini Kit (QIAGEN S.p.A., Milan, Italy); according to the manufacturer's protocol, RNA templates were stored at -70°C until use.

Development of RT-PCRs Specifi c for Classical and TGEV-like CCoVs
Considering the high divergence observed in the 5′ end of the spike gene between classical (subtype IIa) and TGEVlike (subtype IIb) CCoVs, specifi c CCoV-IIa and CCoV-IIb gel-based RT-PCRs were developed. Primer 20179 (sense, 5′-GGCTCTATCACATAACTCAGTCCTAG-3′) binds a conserved region at the 3′ end of ORF1b and was recruited from a previous study (13), whereas antisense primers INS-R-dg (5′-GCTGTAACATAKTCRTCATTCCAC-3′) and 174-268 (5′-CAACATGTAACCTTTGTCTGTGATC TGC-3′) target regions at the 5′ end of the spike protein gene of feline CoV-II (FCoV-II)/classical CCoV-II and TGEV/TGEV-like CCoV, respectively. Separate RT-PCRs with primers 20179/INS-R (CCoV-IIa) or 20179/174-268 (CCoV-IIb) were conducted by using SuperScript One- Step RT-PCR for Long Templates (Invitrogen S.R.L.), according to the manufacturer's instructions. The following thermal protocol was used: reverse transcription at 50°C for 30 min, inactivation of Superscript II RT at 94°C for 2 min, 40 cycles of 94°C for 30 s, 55°C for 30 s, 68°C for 30 s, and fi nal extension at 68°C for 10 min. The PCR products were detected by using electrophoresis through a 1.5% agarose gel and examination under UV light after ethidium bromide staining.

RT-PCR for Amplifi cation of the 3′ End of the Spike Protein Gene of CCoV-II
To rule out any potential infection by true TGEV strains and to confi rm the recombinant origin of the TGEVlike CCoVs, we submitted 20 samples that were positive for CCoV-IIb and negative for CCoV-I to RT-PCR amplifi cation of the 3′ end of the spike protein gene of CCoV-II (7). Primers S5 (5′-TGCATTTGTGTCTCAGACTT-3′) and S6 (5′-CCAAGGCCATTTTACATAAG-3′) were used in the RT-PCR, performed according to the protocol described for CCoV subtyping.

Sequence and Phylogenic Analyses
The RT-PCR products obtained with primer pairs 20179/174-268 and S5/S6 from 26 samples having positive CCoV-IIb-specifi c assay results and representative of the different geographic areas were subjected to direct sequencing at the BaseClear B.V. (Leiden, the Netherlands). The sequences were manually edited and analyzed by using BioEdit software (22) and National Center for Biotechnology Information (www.ncbi.nlm.nih.gov) and European Molecular Biology Laboratory (www.ebi.ac.uk) analysis tools. Phylogenetic and molecular evolutionary analyses were conducted by using Mega 4.1, beta (23). Phylogenetic trees on the basis of partial 5′ (339-nt) and 3′ (520-nt) ends of the spike protein gene were elaborated by using parsimony and neighbor-joining methods, which supplied statistical support with bootstrapping >1,000 replicates. Group-2 CoV canine respiratory CoV-240/05 (24) was used as an outgroup. The nucleotide sequences of the analyzed CCoV-IIb strains were deposited in GenBank under accession nos.

CCoV Genotype and Subtype Distribution
The geographic distribution of the CCoV types and subtypes is reported in Table 1. Genotype-specifi c amplifi cation assays showed that 93 (18.86%) of 493 CCoV-positive samples were positive for CCoV-I, and 218 (44.22%) were positive for CCoV-II. In addition, 182 samples (36.92%) were positive for both genotypes. Both genotypes were found to circulate in most European countries that had been sampled, and an overall prevalence of CCoV-II was found in all countries except Greece, where the 2 genotypes were detected approximately to the same extent.
By using the developed TGEV-like RT-PCR, we found that 78 (19.50%) of 400 samples containing CCoV-II strains, alone or in combination with CCoV-I, were positive for CCoV-IIb. The remaining 322 CCoV-II strains were positive for subtype IIa, whereas mixed infections

RT-PCR of ORF7a/7b Region
After RT-PCR with primer pair N3SN/R3AS, CCoV and FCoV reference strains yielded an amplicon >1,000 bp, with the exception of TGEV-like CCoV 341/05, which gave a 929-bp product as a consequence of a 154-nt deletion in ORF7b (14). In contrast, a 367-bp product was obtained from TGEV-Purdue, as previously described (19). This pattern of amplifi cation agreed with the absence of ORF7b in TGEV (25). All 20 samples that were positive according to the CCoV-IIb-specifi c assay and negative according to the CCoV-I TaqMan assay were confi rmed to contain true CCoV strains because they yielded RT-PCR products considerably larger than the 367-bp amplicon obtained from TGEV.

Sequence and Phylogenetic Analyses
We selected the following TGEV-like strains for sequence analysis of the 5′ and 3′ ends of the spike protein gene: 12 strains from Italy, 8 from the United Kingdom, and 5 from Hungary, plus the single strains from Greece, Romania, and Sweden. All RT-PCR products were sequenced except those obtained from the samples from Greece and Sweden, which yielded weak bands despite the considerable viral RNA titers in the original fecal samples (5.02 × 10 4 and 5.76 × 10 6 RNA copies/μL of template, respectively).
In contrast, analysis of the 3′ end of the spike gene showed nucleotide identities of 87.1%-93.1% to TGEV and of 90.8%-99.0% to CCoV-IIa. The best identities among CCoV-IIa isolates were to strain Insavc-1, which has been proposed as intermediate virus between CCoV and TGEV (26) and to the more recent pantropic strain CB/05 (11) in the 5′ and 3′ ends, respectively.
At the phylogenetic level, the sequenced strains were grouped in the same cluster with TGEV and prototype CCoV-IIb strains at the 5′ end of the spike protein gene, displaying an obvious distance to both CCoV-IIa/FCoV-II and CCoV-I/FCoV-I (Figure, panel A). The TGEV-like CCoVs from the United Kingdom formed a unique clade, whereas the strains detected in eastern Europe were mixed with CCoV-IIb viruses from Italy. The old TGEV-like strain, UCD1, clustered with TGEV isolates. At the 3′ end of the same gene, subtype IIa and IIb strains segregated together and were separated from the FCoVs and the porcine CoVs TGEV and PRCoV (Figure, panel B). In addition, the strains from the United Kingdom were again grouped in a separate subcluster.

Discussion
CoVs are exceptionally prone to variability through accumulation of point mutations and recombination events. A CCoV strain displaying close relatedness to porcine CoVs in the N-terminus of the spike protein, which is related to porcine CoVs, was isolated ≈20 years ago, but the subsequent molecular characterization was restricted to the 5′ end of the spike gene (19). Additional TGEV-like CCoVs were reported more recently in Italy (14) and the United Kingdom (15). However, full biological and molecular characterization was carried out only for the isolates from Italy, showing that the TGEV-like strains are likely recombinant with TGEV at the level of the 5′ end of the spike gene (14). Experimental infection of CCoV-seronegative beagle pups showed that TGEV-like (i.e., subtype IIb) CCoV induces clinical signs resembling those of classical (i.e., subtype IIa) CCoVs, that is, mild diarrhea for a few days. (14) Unlike pantropic CCoV (11)(12)(13), TGEV-like CCoV was not able to spread systemically. Of the 4 recombinant strains detected in Italy, 2 had originated from eastern Europe, but at the phylogenetic level they were mixed with strains from Italy. The prevalence of this CCoV subtype in the canine population has not been determined in previous studies.
Our epidemiologic investigation assessed the distribution of CCoV-IIb in the dog population of Europe. Approximately 50% of the analyzed samples were positive for CCoV, showing the presence of CCoV-I or CCoV-II. Mixed infections caused by both genotypes were detected in <40% of the CCoV-positive samples, considerably lower than previously reported percentages (7,17). Approximately 20% of the CCoV-II-positive samples contained TGEV-like strains (Table 1). However, the prevalence of this CCoV subtype differed by geographic origin of the samples; the highest detection rates (96.87%) were observed in Hungary. The recombinant origin of all strains characterized by RT-PCR was confi rmed by sequence analysis of 5′ and 3′ ends of the spike gene and by RT-PCR of the ORF7a/7b region. The selected 26 TGEV-like strains were related to prototype strains from Italy (14) in both the 5′ and the 3′ ends of the spike gene. A comparison with the prototype UK strains reported by Erles and Brownlie (15) was not possible, however, because the unique spike sequence deposited in the GenBank is located more downstream of the gene with respect to the sequences that we obtained.
On the basis of the spike gene sequences, the strains from Italy and eastern Europe were closely related, whereas the strains from the United Kingdom were more genetically distant (Table 3). This pattern of segregation was confi rmed by phylogenic analysis, which showed that viruses detected in the United Kingdom formed a separate cluster with respect to the samples from Italy, Hungary, and Romania ( Figure). The genetic relatedness between the TGEV-like strains from Italy and those from eastern Europe may be accounted for by extensive dog importation to Italy. In addition, dog exchange between eastern Europe and Italy has been associated with the reemergence of canine infectious hepatitis (27) and the spread of the arctic lineage of canine distemper virus (28).
In the 5′ end of the spike gene, the old TGEV-like strain, UCD1, was found to be genetically more related to true TGEV isolates than to recent CCoV-IIb strains, thereby accounting for recombination events occurring at different times. With the exception of strain UCD1, analysis of archival samples found the oldest TGEV-like strain in 2001, about 4 years before this CCoV subtype was reported in Italy (14).
Our study confi rms that recombinant CCoVs are effectively circulating in dogs in different European countries. Considering the genetic distance in the spike protein, this circulation questions the effi cacy of vaccines, which are based on classical (CCoV-IIa) strains, against the emerging TGEV-like (CCoV-IIb) viruses. Only vaccination trials and subsequent challenges by TGEV-like strains might assess whether the poor cross-reactivity between CCoV-IIa and CCoV-IIb observed in a previous study (14) might affect the immune response of dogs against the recombinant viruses.