Animal Sample Collection

A 1-year-old castrated male dog that had been kenneled for 3 weeks was brought to the University of California, Davis (UC Davis), Veterinary Medical Teaching Hospital for evaluation of progressive vomiting and diarrhea with hematochezia. Despite initial supportive therapy at the referring veterinarian, clinical signs worsened; at UC Davis, the dog was treated for hypovolemic shock. Because of suspected disseminated intravascular coagulation and a poor overall prognosis, the owner elected to have the dog euthanized and granted permission for routine necropsy, which was performed at the Anatomic Pathology Service of the UC Davis School of Veterinary Medicine. The clinical and postmortem workups for infectious causes in this case included negative test results for infectious causes of enteric disease, such as canine parvovirus, canine enteric coronavirus, Salmonella spp., canine distemper virus, Campylobacter spp., Clostridium perfringens enterotoxin A gene, Cryptosporidium spp., and Giardia spp. Histologic results showed extensive fibrinoid vascular necrosis, thrombosis, and hemorrhage throughout the gastrointestinal tract and kidneys, as well as granulomatous lymphadentitis of the mesenteric lymph nodes. Special stains of histologic specimens revealed no detectable bacteria or other infectious agents. Liver tissue samples were collected, stored in whirl-pack bags, and frozen at −80°C until further processing.

Sample Preparation and Nucleic Acid Extraction

A liver tissue sample (≈25 mg) were immersed in 1 mL cold Hank’s balanced saline solution and disrupted with a tissue homogenizer for 30 sec on ice. The resulting homogenates were placed on dry ice for 5 min and then thawed at room temperature (18). Freezing and thawing were repeated twice. Samples were clarified by centrifugation at 10,000 × g for 3 min; the supernatants were then filtered and underwent nuclease treatment as described (19). Viral nucleic acids were extracted by using the QIAamp Viral RNA Mini Kit (QIAGEN, Valencia, CA, USA) and stored at −80°C.

Library Preparation and Sequencing

Viral nucleic acid libraries were prepared as described (19). The library of single-stranded DNA fragments was sequenced by using the Genome Sequencer FLX Instrument (Roche, Indianapolis, IN, USA).

Sequence Data Analysis

The pyrosequencing reads were sorted and trimmed as described (19). Trimmed reads from each sample were assembled de novo by using the MIRA assembly program (20), with a criterion of >95% identity over 35 bp. The assembled sequences and singlets >100 bp were compared with the GenBank nonredundant nucleotide and protein databases (www.ncbi.nlm.nih.gov/genbank) by using BLASTn and BLASTx, respectively (http://blast.ncbi.nlm.nih.gov/Blast.cgi). Potential viral sequences with significant hits (E-value <0.001) to known virus sequences were identified.

Genome Sequencing and Analyses

PCR and Sanger sequencing were used to confirm the presence of virus genome sequences assembled from deep sequencing reads. Inverse PCR was then used to amplify the genome of target circoviruses with primers based on the sequences obtained by deep sequencing. Virus genome sequences obtained were deposited in GenBank (accession nos. KC241982–KC241984). Putative open reading frames (ORFs) with coding capacity >100 aa were predicted by Vector NTI Advance 11 (Invitrogen, Carlsbad, CA, USA). The stem-loop structure was predicted by using Mfold (21).

Phylogenetic Analysis

Phylogenetic analyses based on aligned amino acid sequences from full-length replicate (Rep) proteins were generated by using the neighbor-joining method in MEGA4 (22), using amino acid p-distances with 1,000 bootstrap replicates. Other tree-building methods, including maximum parsimony and maximum likelihood, were used to confirm the topology of the neighbor-joining tree.

Prevalence Study of DogCV in Sample Cohorts

Real-time PCR using 2 primers and a conventional hydrolysis probe with a 5-primer 6-FAM and 3-primer TAMRA label and TaqMan Universal PCR Master Mix (Applied Biosystems, Foster City, CA, USA) was used to detect DogCV in DNA extracts from 3 dog sample cohorts: 1) fecal samples from 204 healthy dogs; 2) fecal samples from 168 dogs with diarrhea; and 3) blood samples from 480 dogs with thrombocytopenia and neutropenia, fever of unknown origin, or past tick bite. Primer pairs and probes for canine circovirus are given in the Table. Total nucleic acid was extracted by using the Corbett X-Tractor Gene platform (QIAGEN). Real-time PCR was conducted by using the real-time PCR instrument LightCycler 480 (Roche, Indianapolis, IN, USA) under these conditions: 50°C for 2 min, then 95°C for 10 min, followed by 40 cycles of 95°C for 15 sec and 60°C for 1 min. Synthetic DNA fragments (≈150 bp) of the corresponding regions were used to produce a standard curve and an analytical sensitivity of 10 molecules.

ISH Analysis

A fourth sample cohort consisted of tissue samples from 21 necropsy cases of dogs whose clinical signs or microscopic lesions matched the sentinel animal (i.e., hemorrhagic diarrhea, vasculitis, and/or granulomatous disease); these samples were selected from the tissue archives of Anatomic Pathology at the UC Davis Veterinary Medical Teaching Hospital. Control tissues were obtained from 5 dogs whose cause of death was unrelated to vascular disease. Tissue sections chosen for analysis were, in part, case dependent because the presence of vasculitis or inflammation among these cases was not limited to a single tissue type. Spleen, lymph node, jejunum, and ileum were examined in all cases; other examined tissues were kidney, brain, adrenal gland, pancreas, duodenum, heart, and lung. Tissues from the sentinel dog were included in this analysis.

Tissue sections were mounted on 3-aminopropyltriethoxylane–coated slides (Fisher Scientific, Pittsburgh, PA, USA). A 25-nt oligomer (CTCAGACAGAGACACCGTTGCTATG) complementary to a segment of ORF2 (capsid) was 3′-end labeled by the addition of a single digoxigenin-II-dideoxy undine (Eurofins MWG Operon, Huntsville, AL, USA). A manual capillary-action work station (Fisher Scientific) was used to perform colorimetric DNA in situ hybridization. Tissue sections were deparaffinized and digested by incubation at 37°C for 10 min in 0.25% pepsin in 1× Tris-buffered saline (pH 2.0); pepsin activity was stopped by a 5-min incubation at 105°C. Nucleic acid was denatured by 5-min incubation at 105°C in 100% formamide. Tissue sections were then incubated at 37°C in 10 µmol of the digoxigenin-labeled probe in hybridization buffer (22.5% deionized formamide, 7.5% chondroitin sulfate, 5× saline sodium citrate, 0.25% blocking reagent, and 50 mmol phosphate buffer). Sections were incubated in an antidigoxigenin antibody solution (500:1 dilution) containing 2.5 mL buffer 1 with 5 μL of antidigoxigenin Fab fragments conjugated with alkaline phosphatase (750 U/mL) (Roche). Sections were then washed and developed according to manufacturer instructions before counterstain with 1% fast-green FCF for 5 min. Slides were mounted with ImmunoHistoMount (Immunobioscience, Mukilteo, WA, USA) and coverslipped with SHUR/Mount (Triangle Biomedical Sciences, Durham, NC, USA). No background hybridization was seen when replicate tissue sections were incubated with an unrelated digoxigenin-labeled probe with similar guanine-cytosine content or when matched tissue from unaffected dogs were incubated with the DogCV-specific probe (23–25).

Electron Microscopy

Selected pieces of formalin-fixed lymph node tissue from the sentinel dog were post fixed in 2.0% glutaraldehyde and then routinely processed and embedded in epoxy resin (Eponate12 kit; Ted Pella, Inc., Redding, CA, USA). Selected thick sections were stained with toluidine blue, as described (26). Ultrathin sections from select areas of the lymph node were examined by using a Zeiss (Gottingen, Germany) 906E transmission electron microscope.

Metagenomic Identification of Canine Circovirus

Of ≈10,000 sequence reads, 5 contigs and singlets composed of 52 sequence reads from the liver tissue had significant similarity to the Rep protein of circoviruses (E-value <1⁻¹⁰). Two bocavirus sequences were also detected. Because circoviruses have a circular genome, the full viral genome was then amplified by inverse nested PCR and the amplicon sequenced by primer walking. The assembled genome was named DogCV strain UCD1 (DogCV-UCD1).

Genome Analysis and Phylogenetic Relationship

Figure 1

Figure 1. . . A) Genome organization of dog circovirus (DogCV) and porcine circovirus 2 (PCV2). B) Phylogenetic analysis of DogCV strains (UCD1–3, isolated from tissue, feces, and blood respectively) based on the...

The complete circular genome of DogCV-UCD1 was 2,063 nt (GenBank accession no. KC241982). Analysis of the genome sequence showed characteristics typical of circoviruses, including an ambisense organization with 2 major inversely arranged ORFs encoding the putative replication-associated (Rep, 303 aa), and capsid (Cap, 270 aa) proteins. A characteristic stem-loop structure with a conserved nonanucleotide motif (5′-TAGTATTAC-3′, similar to the consensus of bird and pig circoviruses) was also found in the 5′-intergenic region (135 nt, between the start codons of the 2 major ORFs). The 3′-intergenic region of DogCV-UCD1 between the stop codons of the 2 major ORFs was 203 nt (Figure 1, panel A) (27).

The complete genome of another DogCV strain (DogCV-UCD2, GenBank accession no. KC241984) was amplified and sequenced from a fecal sample from a shelter-housed dog that had vomiting and diarrhea. This strain shared 95% overall genome nucleotide identity with DogCV-UCD1, and both strains showed 96%–97% nucleotide identity to the recently reported canine circovirus isolate 214 from blood (17).

The putative Rep proteins of DogCV-UCD1 showed 42%–54% amino acid identity to the Rep proteins of porcine and avian circoviruses, with the closest identity to PCV1. The DogCV-UCD1 capsid showed <30% amino acid identity to known circovirus capsids. Sequence alignment of the putative Rep protein of DogCV-UCD1 with those of known species in the genus Circovirus identified several highly conserved amino acid motifs, including WWDGY, DDFYGW, and DRYP. Motifs associated with rolling circle replication (FTLNN, TPHLQG, and CSK) and the dNTP binding (GXGKS) were also identified. The N terminal region of the Cap protein was highly basic and arginine-rich, as is typical for circoviruses. A phylogenetic analysis of the complete Rep protein of DogCV strains and all known circoviruses was performed, with chicken anemia virus (genus Gyrovirus) as the outgroup (Figure 1, panel B). The phylogenetic tree showed that DogCV strains grouped with PCV1 and PCV2, forming a distinct clade of mammalian circovirus, whereas circoviruses affecting birds clustered separately.

Prevalence Study

DogCV was detected by real-time PCR in fecal samples from 14/204 healthy dogs and 19/168 dogs with diarrhea; the difference in prevalence was not significant (11.3% vs. 6.9%; p>0.05 by χ² test). Of the 19 dogs with diarrhea who had DogCV detected in fecal samples, 13 (68%) were co-infected with >1 other pathogens, including canine enteric coronavirus, Cryptosporidium spp., C. perfringens α toxin, Giardia spp., Salmonella spp., Campylobacter jejuni, and Campylobacter coli (tested by PCR).

The prevalence of DogCV in blood samples from the cohort of dogs with thrombocytopenia and neutropenia, fever of unknown origin, or past tick bite was 3.3% (16/480), similar to that reported for canine serum samples (2.9%, 6/205) (17). The partial Rep and/or Cap protein regions (≈350 bp) were amplified from 11/16 samples. All showed >96% nucleotide identity, except 1 amplicon, which had <90% nucleotide distance to DogCV-UCD1 and -UCD2. We sequenced the complete genome of this virus (DogCV-UCD3; GenBank accession no. KC241983); it showed 91%–92% amino acid identity of the complete Rep and Cap proteins to DogCV-UCD1 and -UCD2 and to the published canine circovirus (CaCV-1 strain NY214; GenBank accession no. JQ821392) (17).

ISH Analysis and Pathologic Findings in Positive Cases

To establish tissue distribution and investigate whether DogCV contributes to canine disease, we developed and validated an ISH oligomeric probe and examined the sentinel dog and dogs from 21 suspected, retrospective cases that included >2 of these 3 signs: vasculitis, hemorrhage, or granulomatous disease. A wide spectrum of affected tissues was represented in this group; matching tissues were also examined from 5 control dogs in which these signs were not present. Samples from the sentinel dog (dog 1) and 3 other dogs (dogs 2–4) were positive for DogCV by ISH analysis. All other tissue samples from control dogs were negative by ISH analysis. Clinical signs, gross and histologic findings, and distribution of virus DNA as determined by ISH were used to examine a possible causal role for DogCV.

Dog 1 was a male beagle who had acute onset of vomiting and hemorrhagic diarrhea. Dog 2 was a 5-year-old, female, spayed Boston terrier who had vomiting and diarrhea. Dog 3 was a 1-year-old, female, spayed boxer with a 5-day history of lameness and progressive tetraparesis. Dog 4 was a 2-year-old Greyhound found dead with bicavitary hemorrhage; blood smear and PCR showed this dog was co-infected with Babesia conradae.

Figure 2

Figure 2. . . Organ tissues from sentinel dog (dog 1) and 2 other dogs (dogs 2 and 3) that were positive by in situ hybridization (ISH) analysis for dog circovirus (DogCV). A)...

Gross examination revealed consistent lesions among these 4 dogs, including lymphadenopathy and hemorrhage. In dogs 1 and 2, the hemorrhage was associated with the gastrointestinal tract (Figure 2, panels A, C); dog 2 had additional multifocal to coalescing regions of hemorrhage in the kidneys (Figure 2, panel B). Gross evidence of hemorrhage in dog 3 was restricted to the ventral surface of the brain along the basilar artery overlying the medulla; dog 4 had bicavitary hemorrhage. The common histologic lesion in all dogs was fibrinonecrotizing vasculitis, although the distribution of affected vessels and the amount of associated hemorrhage varied. In dog 1, segments of inflamed or necrotic vessels were seen in the intestine (multiple segments), urinary bladder, liver, spleen, and lungs. In dog 2, vasculitis was limited to the intestine and spleen, and in dog 3, vasculitis was in kidneys, intestine (Figure 2, panel G), heart, liver, spleen, and meninges. In dog 4, only a few vessels were affected, at the corticomedullary junction in the kidneys. For all dogs, histiocytic drainage or granulomatous lymphadenitis were seen in Peyer’s patches (Figure 2, panels C, E) and in >1 lymph node. In addition, in dogs 2 and 3, multiple lymph nodes were severely necrotic. All dogs had microscopic lesions in the kidneys, but intensity and character varied widely. Dogs 1 and 2 had tubular necrosis with little inflammation, dog 3 had a severe granulomatous interstitial nephritis, and dog 4 had multifocal hemorrhage and minimal lymphocytic, plasmacytic nephritis. Dogs 1 and 3 had multifocal pancreatitis and adrenalitis. No multinucleate giant cells, common in pigs infected with PCV2, were seen.

By ISH analysis, abundant cytoplasmic viral nucleic acid was detected in macrophages within germinal centers and subcapsular and medullary sinuses of the mesenteric lymph nodes from all 4 dogs (Figure 2, panel H), mandibular lymph nodes from 2 dogs, ileal Peyer’s patches from all 4 dogs (Figures 2, panels D, F), ellipsoids (terminal arterioles) of the spleen for 1 dog, and germinal centers in splenic white pulp for 2 dogs. Although the reticular network was abundantly positive in dog 1 (Figure 2, panel H), the common pattern was localization to the centers of lymphoid or Peyer’s patch follicles (Figure 2, panels D, F), corresponding to the morphology and distribution of dendritic cells. Rare nuclei of elongate cells (presumed endothelium) lining small vessels of the adrenal cortex or intestinal lamina propria suprajacent to the Peyer’s patches were positive in 2 dogs. No nucleic acid was detected by ISH analysis in other tissues, regardless of the character or intensity of the inflammation. None of the 5 control dogs showed DogCV DNA by ISH.

Transmission Electron Microscopy Analysis

Figure 3

Figure 3. . . Lymph node from sentinel dog from which dog circovirus was identified. A) Toluidine blue stain shows multiple macrophages within the medullary sinus contain vacuoles and discrete, oblong to round,...

Ultrastructural analysis of the mesenteric lymph node from dog 1 revealed macrophages laden with large numbers of intracytoplasmic inclusions (Figure 3, panel A). Inclusions were round, oblong, or irregular; were <0.5 µm; and often clustered (up to 25 per cell) within the cytoplasm, (Figure 3, panel B). Most inclusions were granular and electron-dense and had a distinct periphery but were nondelineated by membrane. Some inclusions contained paracrystalline arrays of icosahedral virions that were 9–11 µm in diameter (Figure 3, panel C).