We analyzed serum and anal swab samples collected during 2013–2015 from 184 domestic dogs and 90 wild coatis in the city of Belo Horizonte (19°55′15′′S, 43°56′16′′W) in the state of Minas Gerais, Brazil (Figure 1). Swab samples of lesions, if present, were also collected. To determine the presence of neutralizing antibodies in serum, we used an orthopoxvirus plaque reduction neutralization test as previously described (12). Serum titers were defined as the highest dilutions that inhibited >70% of virus plaques compared with negative controls.

To detect VACV DNA from serum and anal swab samples, we performed real-time PCR targeting the C11R or A56R gene (12). We directly sequenced A56R fragments in both orientations and in triplicate by using the ABI3130 platform (Applied Biosystems, Waltham, MA, USA). Sequences were aligned with other reference sequences from GenBank by using MEGA 7.0 (http://www.megasoftware.net). Statistical analyses were conducted by using Epi Info software version 7.2.1.0 (https://www.cdc.gov/epiinfo); χ² and Fisher exact tests were applied with significance set at 5%. We also calculated relative odds ratios (ORs) and 95% CIs.

We detected orthopoxvirus neutralizing antibodies in 35 dogs (prevalence rate 19.0%, 95% CI 14.0%–25.5%; titers 100–400 neutralizing units/mL) and in 13 coatis (prevalence rate 14.4%, 95% CI 8.5%–23.3%; titers 100–800 neutralizing units/mL) (Table 1). Univariate analyses indicated significant associations between presence of neutralizing antibodies and the following: male dogs (OR 2.6; p = 0.02), dogs 6–10 years of age (OR 5.2; p = 0.04), coatis captured in 2013 (OR 11.2; p = 0.002), juvenile coatis (<1 y of age) (OR 35; p = 0.001), and adult coatis (>2 y of age) (OR 5.1; p = 0.04).

Samples from all seropositive animals were submitted for quantitative PCR (qPCR) to detect VACV DNA (Table 2). Overall, serum samples from 7 dogs and 6 coatis were positive for the C11R gene; of these, anal swab samples were positive for this gene for 3 dogs and 4 coatis. Samples from the C11R-positive animals were submitted for another qPCR targeting the A56R gene. Serum samples were positive for the A56R gene for 5 dogs and 4 coatis; of these, anal swab samples were positive for A56R for 1 dog and 3 coatis. No lesion swab samples were positive by qPCR for both C11R and A56R genes.

Alignment of the amplified A56R fragments showed high similarity to the homologous gene of VACV isolates from Brazil (Technical Appendix Figure). Furthermore, 5 sequenced samples (from 4 dogs and 1 coatis) showed an 18-nt signature deletion, which is present in sequences of mouse nonvirulent VACV strains from Brazil (group 1 VACV). This deletion was not present in samples from 4 animals (1 dog and 3 coatis), grouping with mouse virulent VACV strains from Brazil (group 2 VACV).

We assessed VACV exposure of 2 interacting species of animal: domestic dogs from an urban area and coatis from a bordering wild area. In contrast to VACV infections, human cowpox virus infections have mostly occurred in urban areas of Europe. Cowpox virus is transmitted to humans mainly by domestic cats that are in contact with rodents, the natural cowpox virus reservoirs (13). However, some authors have hypothesized that domestic dogs could be implicated in the transmission cycle of VACV, acting as a link between the natural reservoirs and humans in urban environments (14,15). Indeed, our molecular findings support exposure and possible VACV infection of these animals, thereby indicating that they are a potential source of VACV exposure for humans in urban areas.

The seroprevalence of orthopoxvirus neutralizing antibodies in dogs in Brazil has been described. Peres et al. found that, along with other farm animals, 22.8% of 114 dogs tested were seropositive for orthopoxviruses (7); this seroprevalence differs from that observed in our study, which was 3.8% lower. In addition, most animals tested by Peres et al. were from rural areas where no bovine vaccinia outbreaks had been officially reported, and 96% of farmers declared that their domestic animals have contact with wild animals (9). These findings indicate that dogs could be exposed to VACV through contact with wild animals, corroborating our hypothesis.

We also detected orthopoxvirus neutralizing antibodies in wild coatis, which is consistent with results of a previous study that described the seroprevalence of orthopoxviruses in procyonids from Mexico (9). Our detection of VACV DNA in anal swab samples from coatis indicate that these animals could act as a source of virus for domestic dogs and humans and serve as a link between wild and urban environments. However, future studies to determine if viable virus is shed are needed to confirm this possibility.

To impart information about the role of domestic animals and wildlife in the natural cycle of VACV, we developed a hypothetical model based on previous studies (2,3,5), which could illustrate the dynamics of VACV circulation in urban areas (Figure 2). Because coatis can circulate in wild environments and surrounding urban areas, they could act as a bridge promoting the transmission of VACV between wild animals (mainly rodents) and dogs or humans. Domestic dogs could transmit VACV directly to humans through close contact or indirectly thought contaminated feces (Figure 2). These data raise questions about VACV circulation in Brazil and open discussions about the role of dogs and coatis in the VACV epidemiologic cycle.

Nucleotide sequences of vaccinia virus A56R gene from domestic dogs and wild coatis, Brazil, 2013–2015, compared with sequences from other orthopoxviruses.