Cross‐species transmission of retroviruses among domestic and wild felids in human‐occupied landscapes in Chile

Abstract Human transformation of natural habitats facilitates pathogen transmission between domestic and wild species. The guigna (Leopardus guigna), a small felid found in Chile, has experienced habitat loss and an increased probability of contact with domestic cats. Here, we describe the interspecific transmission of feline leukemia virus (FeLV) and feline immunodeficiency virus (FIV) between domestic cats and guignas and assess its correlation with human landscape perturbation. Blood and tissue samples from 102 free‐ranging guignas and 262 domestic cats were collected and analyzed by PCR and sequencing. Guigna and domestic cat FeLV and FIV prevalence were very similar. Phylogenetic analysis showed guigna FeLV and FIV sequences are positioned within worldwide domestic cat virus clades with high nucleotide similarity. Guigna FeLV infection was significantly associated with fragmented landscapes with resident domestic cats. There was little evidence of clinical signs of disease in guignas. Our results contribute to the understanding of the implications of landscape perturbation and emerging diseases.


| INTRODUC TI ON
Emerging infectious diseases are a serious threat to global biodiversity (Smith et al., 2006). Human-induced land-use and ecological changes of natural habitats have been proposed to be major drivers of pathogen emergence (Daszak et al., 2001;Dobson & Foufopoulos, 2001;Murray & Daszak, 2013). There are numerous examples of how anthropized landscapes can facilitate pathogen transfer among domestic animals and wildlife (Foley et al., 2013;Millán et al., 2016;Murray & Daszak, 2013;Riley et al., 2004).
RNA viruses that are transmitted through direct contact are the most-frequently cited to spread by interspecific host jumping (Woolhouse et al., 2005). They often display high rates of nucleotide substitution and thus have higher capacity to adapt to new hosts (Jones et al., 2008). Feline leukemia virus (FeLV) and feline immunodeficiency virus (FIV) are RNA viruses that cause two of the most common infectious diseases affecting domestic cats (Luria et al., 2004;O'Brien et al., 2012). Both of these felid-specific retroviruses are horizontally transmitted through saliva or other body fluids (Munro et al., 2014). Transmission occurs mostly through fighting, especially in male animals as they are more aggressive, but also by grooming and sharing of food. Vertical transmission occurs occasionally (Pan et al., 2018). Viremia is persistent in approximately one third of FeLV-exposed cats; it results in severe clinical disease including immunosuppression, anemia and/or neoplasia (Mullins & Hoover, 1990). The main consequence of FIV in the infected organism is immunosuppression, but it can also produce neoplasia, blood dyscrasias due to myelosuppression and neurological disorders (Hartmann, 2012).
Both FeLV and FIV have worldwide distribution, with prevalence in domestic cats ranging from 3%-30% for FeLV and 0%-50% for FIV (Gleich et al., 2009). The spatial and temporal dynamics of these pathogens differed regionally in the United States, with higher FIV prevalence observed in the southern and eastern U.S., in contrast to higher prevalence of FeLV infections in the western U.S. (Chhetri et al., 2013). This pattern suggests that spatial risk factors can vary geographically, perhaps in response to variations in specific virus strains or rates of vaccination. However, there was no evidence of changes in positivity rates in FIV or FeLV during the studied period (Chhetri et al., 2013).
Most FeLV infections in non-domestic felids appear to be self-limiting (Sleeman et al., 2001). However, outbreaks linked to mortality have been reported (Iberian lynx; Meli et al., 2009), suggesting that infection can be pathogenic. Possible interspecific transmission of FIV between domestic and wild felids has been described Mora et al., 2015;Nishimura et al., 1999;Troyer et al., 2005).
Environmental and behavioral barriers may hinder interspecific FIV and FeLV transmission under natural conditions . However, human landscape perturbation may facilitate interspecies transmission by increasing contact probability between domestic cats and wild felids. Deforestation and human encroachment into wild habitats are increasing in Chile (Echeverria et al., 2008;Wilson et al., 2005). These perturbations can have a major impact on wild species adapted to the dense temperate rainforests of southern South America, such as the guigna (Sanderson et al., 2002).
Guignas have a restricted distribution in central and southern Chile (30°-48°S) and southwestern Argentina (39°-46°S west of 70°W; Napolitano et al., 2014). Estimates of plausible lower and upper bounds for the total number of mature individuals range from 6000 to 92,000. Four of the six geographic groups were estimated to have at least 1000 mature individuals (lower bound; Napolitano, Gálvez, et al., 2015). Habitat loss and fragmentation have restricted them to forest fragments surrounded by a human matrix with domestic animals (Gálvez et al., 2013;Sanderson et al., 2002). Although guignas can adapt to perturbed landscapes by using vegetation corridors, increased fragmentation has been associated with wider dispersal and reduced levels of neutral genetic diversity (Napolitano, Díaz, et al., 2015), followed by increased probabilities of encountering domestic carnivores and cross-species pathogen transmission. Decreased adaptive genetic diversity may also increase susceptibility to infectious diseases, resulting in epidemics, high mortality and the potential of local extinctions (Iberian lynx [L. pardinus], Meli et al., 2010; lion (P. leo) Roelke-Parker et al., 1996). Guignas are solitary, pairing only when mating (Sanderson et al., 2002), and thus may be less likely to sustain new infectious agents in their populations. Therefore, sympatric carnivores serving as reservoir hosts may help maintain a relatively high pathogen population, increasing transmission and disease risks (Millán et al., 2009). Guignas are currently classified by the IUCN Red List as Vulnerable (Napolitano, Gálvez, et al., 2015).
Based on the above information, we conducted an extensive molecular survey for FeLV and FIV infection in guignas across their entire distribution range in Chile and in sympatric populations of

| Study area
The study area included four bioclimatic regions in central and southern Chile (33°S-46°S) encompassing the entire current distribution range of the guigna in Chile (Mediterranean area, rainytemperate area, Chiloé Island, and oceanic cold-temperate area; Napolitano, Gálvez, et al., 2015;Figure 1). Study sites included a gradient of different landscape types, ranging from continuous pristine native forest with no human presence to human-dominated landscapes with high human density and fragments of remnant forest surrounded by a matrix of agriculture, livestock activities and domestic cats and dogs.

| Sample collection
Between 2008 and 2018, 102 free-ranging guignas were sampled using either tomahawk-like live traps (n = 52) or after being received at wildlife rescue and rehabilitation centers (WRRC; n = 8). Whole blood samples were collected from these 60 animals, and buffy coats were extracted. We also sampled 42 animals that had been roadkilled or euthanized at WRRC. Bone marrow, Peyer's patches and small intestine samples were collected from these during complete necropsies. Portions of these tissues were archived and subsamples obtained. All the tissues and organs mentioned are target tissues for FeLV and FIV. All live captures and tissue collection were conducted using established methods (Napolitano, Díaz, et al., 2015) following bioethical and animal welfare guidelines and protocols and with prior permission from the Chilean Agriculture and Livestock Service (SAG) and relevant Animal Ethics Committees.
For each guigna, sex, age range (estimated from dentition), physical condition, and clinical signs of disease (Crowe, 1975) were assessed by a veterinarian; date and GPS location of collection/capture were recorded for each animal. A total of 38 females and 64 males, 63 adults and 16 juveniles were sampled (ages of 23 animals were unknown).
We collected whole blood samples from 254 free-roaming, nonferal "outside" domestic cats inhabiting rural communities throughout the distribution of the guigna in Chile; buffy coats of each sample were extracted. In addition, bone marrow, Peyer's patches, and small intestine samples were collected during 8 complete necropsies from animals F I G U R E 1 Map of study area showing the overall prevalence of feline leukemia virus and feline immunodeficiency virus in guignas (white color) and domestic cats (shaded black) as well as percent relative prevalence of both viruses in guignas and domestic cats from the different study areas (prevalence is expressed with respect to the sample sizes recorded in each study area) found dead (roadkills) or euthanized at veterinary clinics. Samples were stored frozen at −20°C until PCR analysis. The sex, age class, date, and location were recorded for each individual. A total of 129 females and 133 males, 226 adults, and 36 juveniles were sampled, none of which had been previously vaccinated or neutered.

| Genetic analysis
Genomic DNA was extracted using a commercial kit (DNeasy Blood & Tissue kit; Qiagen ® ). DNA amplification was performed by nested PCR in a BIO-RAD T100 thermal cycler. We conducted MHC class II gene PCR as previously described (Castro-Prieto et al., 2010) as an internal control to confirm quality of extracted DNA. All samples were validated and subsequently analyzed for FeLV and FIV. We amplified a 291-base-pair (bp) fragment of proviral DNA from the FIV gag genomic region and a 211-bp fragment from the FeLV U3 LTR genomic region (unique region which distinguishes endogenous from exogenous FeLV; Miyazawa & Jarrett, 1997) using external and internal primers and conditions described in Mora et al. (2015). All positive samples were tested to amplify longer fragments for FeLV, a 700-bp fragment of the FeLV U3 region (Miyazawa & Jarrett, 1997) and a 468-bp segment of the U3 region of the LTR of FeLV-A (Cattori et al., 2006). PCR of domestic cat samples and guigna samples was run separately to avoid cross contamination. Ultrapure water was used as a negative control. Positive controls were FIV and FeLV proviral DNA from domestic cats previously diagnosed by PCR and confirmed by nucleotide sequencing. Positive controls were checked against sequences obtained to rule out contamination products and sequences from study animals. All sequences distinct from the positive controls were included in phylogenetic analysis and sequence comparisons. PCR products were separated by electrophoresis in 2% agarose gels stained with SYPRO red protein gel, visualized with a transilluminator. Forward and reverse strands were sequenced by the Sanger method at Macrogen Inc.
Multiple sequence alignments were conducted using the CLUSTAL W algorithm (Geneious®). Phylogenetic trees were constructed based on Bayesian and maximum-likelihood methods. The best model of evolution was selected with jModelTest2 (version 2.1.6; Darriba et al., 2012), under the Akaike Information Criterion.
A ntST network was generated using the median joining method implemented in popart (Leigh & Bryant, 2015). Phylogeographic structure of FeLV and FIV in guignas and domestic cats was assessed comparing G ST and N ST coefficients in Permut (Pons & Petit, 1995 1000 permutations of N ST ). Genetic structure in guigna and domestic cat host species was estimated using pairwise Phi st tests in Arlequin (Excoffier & Lischer, 2010;1000 permutations) and the nearest-neighbor statistic S nn in DnaSP v5 (Librado & Rozas, 2009). Newly identified FeLV and FIV sequences were submitted to the GenBank database under the accession numbers indicated in Table S1.

| Landscape analyses
To describe landscape features associated with FeLV and FIV infection in guigna, for each sample location we generated a circular buffer area with QuantumGIS 2.14 ® , corresponding to the mean home range of the species (male = 446 ha; female = 170 ha; Dunstone et al., 2002;Sanderson et al., 2002;Schüttler et al., 2017). Within each buffer area, which differed for males and females based on their average home ranges, we described or quantified five landscape variables: (i) percentage of vegetation cover (Hansen et al., 2013), (ii) number of houses, (iii) distance to the nearest house, (iv) land use (fragmented landscape or continuous forest) and (v) bioclimatic region: Mediterranean region, rainy-temperate region, Chiloé Island (rainy-temperate to oceanic cold-temperate transition) region, and oceanic cold-temperate region (INE, 2018). Land use (variable 4) is a qualitative variable with two categories. We defined continuous landscape as a buffer area composed only of continuous vegetation, which may include roads (as functional connectivity for guignas is not limited by roads; Gálvez et al., 2013Gálvez et al., , 2018Sanderson et al., 2002). We defined fragmented landscape as a buffer area including human settlements, agricultural land and/or fragments of forest surrounded by a matrix of human activities. GIS layers were obtained from the Ministerio de Bienes Nacionales website. QGIS 2.14 ® software was used to extract the attribute values of landscape variables for spatial analysis. We tested the collinearity among predictors by running bivariate Pearson correlations among pairs of variables. No statistically significant correlations were found.
To address spatial autocorrelation in our data, we conducted a Global Moran's I test using ArcGIS Pro. We obtained non-significant results (Moran's index = 0.38, z-score = 0.46, p-value = 0.64) suggesting there is no pattern of data spatial clustering.

| Assessment of clinical signs of disease
We measured hematological, biochemical and histological parameters, and physical examination findings blind to viral PCR status.
For hematological and biochemical analysis, whole blood preserved in EDTA of 20 guignas and serum samples of 19 guignas were tested.
Hematological parameters analyzed included erythrocyte count (RBC), white blood cell count, hemoglobin concentration, mean cell volume, mean corpuscular hemoglobin concentration and hematocrit determination using the Abacus Junior Vet Analyzer (Diatron ® ).
For histopathological analysis, we collected samples during complete necropsies of 32 animals. Histological evaluation was performed on selected formalin-fixed paraffin-embedded tissues, sectioned at 5 μm and stained with hematoxylin eosin. Tissues were evaluated by a board-certified pathologist blind to viral PCR status, geographic location, and other demographic information. Lesions were scored as present or absent and compared between PCRpositive and PCR-negative guignas. Immunohistochemistry using antibodies cross-reactive to B (CD79a) and T (CD3) lymphocytes was performed following the manufacturer's protocol (Biocare Medical).
Lymph node architecture was evaluated for number of follicles, presence and number of secondary follicles, lymphoid expansion into the paracortex, presence of lymphocytes within medullary cords and the presence of intrafollicular hyalinization, with results compared between positive and negative guignas.

| Statistical analysis
Spatial and biological (host age and sex) independent variables and FeLV and FIV infection (binary response variable) were assessed using multivariate logistic regression analyses, calculating crude and adjusted odds ratios (ORs) with 95% confidence intervals (CIs).
Goodness of fit models were assessed using the Hosmer-Lemeshow test and residuals analysis. Differences in infection prevalence between domestic cat and guigna, and between biogeographic regions, as well as comparisons of hematological and biochemical parameters of infected and non-infected guignas, were analyzed by Mann-Whitney U tests and Kruskal-Wallis tests. All statistical analyses were performed in R software with a significance level of p < 0.05.

| Prevalence and genetic diversity
Feline leukemia virus-infected guignas were detected in all bioclimatic areas (Table S2). FeLV-infected domestic cats were detected in all rural communities studied across bioclimatic areas (Table S3). In contrast, FIV-infected guignas were detected in only two of the four bioclimatic areas, while FIV-infected domestic cats were identified in all bioclimatic areas (Tables S2,S3).
Feline leukemia virus DNA was detected in 20.6% of guignas and 20.2% of domestic cats, and FIV was detected in 3.0% of guignas and 3.0% of domestic cats. All FIV-infected guignas were also coinfected with FeLV; FIV/FeLV co-infections occurred in 1.14% of domestic cats (Tables S2,S3). There was no statistically significant difference in FeLV and FIV prevalence between the two host species

| Exposure risk factor analysis
Guignas were significantly (9.6×) more likely to be exposed to FeLV in fragmented landscapes (95% CI = 2.37-65.32; p = 0.005; adjusted odds ratio = 9.6; Table S2) and male guignas were 3.6 times more likely to be infected with FeLV than females (95% CI = 1.18-12.42; p = 0.03; adjusted odds ratio=3.6; Hosmer-Lemeshow test for goodness of fit model, p = 0.99). No other statistically significant differences were observed for FeLV in guignas comparing ages, bioclimatic areas or for the other spatial variables, percentage of vegetation cover, number of houses, distance to the nearest house, or land use category (fragmented landscape or continuous forest). The three FIV-positive guignas (also coinfected with FeLV), two adult males and one young female, were all found in fragmented landscapes.
However, no statistical analysis for the spatial and biological variables studied was possible for FIV due to the low observed prevalence. A significantly lower FeLV prevalence was observed among domestic cats in the rainy-temperate area (95% CI = 0.12-0.60; p = 0.01) and oceanic cold-temperate area (95% CI = 0.14-0.81: p = 0.02). No statistically significant differences for domestic cat FeLV or FIV infection or coinfection were observed with respect to sex or age (95% CI = 0.55-1.91; p = 0.9 and 95% CI = 0.35-2.29; p = 0.9, respectively; Table S3).

| Serology and clinical signs in guigna
Serum samples from four and seven FIV and FeLV PCR-positive guignas, respectively, were analyzed and found negative using INgezim and IDEXX kits (Table 1)

| D ISCUSS I ON
The present study represents one of the largest surveys of retroviral infection in a wild feline and sympatric rural free-roaming domestic cats. We surveyed guignas from across their complete distribution range in Chile, documenting for the first time that they are infected with FeLV and FIV throughout most of their range. Estimates of prevalence of FeLV and FIV for guignas and domestic cats were similar to those found in a previous pilot study on Chiloé Island (Mora et al., 2015). Previous records of retroviral prevalence in domestic cats in Chile ranged from 4%-15% for FIV and 4%-10% for FeLV (Bilbao, 2008;Troncoso et al., 2013). Feline immunodeficiency virus is a species-specific infection, and cross-species transmission is highly restricted by interspecies transmission barriers. However, although rare, a few cross-species transmissions have been reported: An FIV strain shared by both bobcats (Lynx rufus) and pumas (Puma concolor) occupying the same habitat in Florida and California (Franklin et al., 2007;Lee et al., 2014), a free-ranging leopard cat that acquired FIVfca from a domestic cat (Nishimura et al., 1999), FIVfca in a captive puma in Argentina and lion FIV clade A sequences (FIVpleA) amplified from a snow leopard F I G U R E 3 Maximum-likelihood tree of 211 bp fragment of FeLV U3LTR genomic region for guignas and domestic cats. Sequences from this study are highlighted (red = domestic cat NtST, green = domestic cat/guigna shared NtST, blue = guigna NtST). Numbers in sequence names correspond to the ntST described in this study (H1, H2, H3, H4, H5, H9, H10, H13, H14, H15, H16). GenBank accession numbers in parentheses. Bootstrap values ≥70 are shown at the nodes of the tree. Lgu, Leopardus guigna; Fca, Felis silvestris catus; Pco, Puma concolor (Panthera uncia) and a tiger (Panthera tigris) in Asian zoos Troyer et al., 2005).
Viruses isolated from different species seem to group more by geographic region of the host than in groupings concordant with the phylogenetic relationships of the host species . Cross-species transmission among bobcats and pumas in the USA is believed to have been facilitated by both species occupying the same habitat. A somewhat similar situation is apparent in the guigna and rural domestic cats sampled in this study, where animals may come into direct contact facilitated by guignas approaching households occasionally when preying on fowl (Sanderson et al., 2002), or domestic cats roaming freely up to 2 km away from the household into the forest (López-Jara et al.,   O'Brien et al., 2012). Higher exposure to pathogens in wild carnivore populations and disease emergence have been reported in human-dominated landscapes (Alexander & Appel, 1994;Cleaveland et al., 2000;Laurenson et al., 1998;Millán et al., 2016;Riley et al., 2004;Sillero-Zubiri et al., 1996).
In Chilean fragmented landscapes with domestic cat presence, where guignas increase their dispersal (Napolitano, Díaz, et al., 2015) and occasional poultry attacks by guignas within human settlements occur, increased aggressive domestic cat-guigna encounters may enhance opportunities for novel pathogen exposure and spillover (Mora et al., 2015;Sanderson et al., 2002). Considering that FIV and FeLV are shed in high concentrations in saliva and that the major mode of transmission is through bites (Sykes et al., 2014a(Sykes et al., , 2014b, these aggressive encounters may facilitate interspecific transmission. Once the species barrier has been crossed, it is possible for the virus to propagate among individuals of the new host species , followed by a level of interspecific encounters sufficient to allow virus transfer, establishment and adaptation (Engering et al., 2013;Parrish et al., 2008).
Male guignas had higher probability of FeLV infection than females; this is also the case for domestic cats due to more aggressive behavior (Hartmann, 2006;Munro et al., 2014). In most solitary mammals, females are philopatric and males disperse from their natal area, fiercely fighting for territory acquisition and access to females (Moyer et al., 2006;Prugnolle & De Meeûs, 2002;Ratnayeke et al., 2002;Waser & Jones, 1983). Although guignas are solitary felids, less likely to acquire an infectious agent through direct contact, male guignas are expected to play a role in intraspecific transmission of pathogens. Transmission of infection to females could follow from mating activities, when the tomcat bites the female (Meli et al., 2010).

Feline leukemia virus-infected guignas and FeLV and FIV-
infected domestic cats were present in all bioclimatic areas studied.
However, FIV-infected guignas were detected in only two of these (rainy-temperate region and Chiloé Island), perhaps because our more limited sampling in these bioclimatic areas was insufficient to detect the very low prevalence of FIV. Given that both pathogens are transmitted by direct contact, we would not expect bioclimatic variables to influence distribution and/or prevalence directly. However, higher rates observed in central Chile (Mediterranean area), where over 50% of Chile's population resides (Napolitano, Gálvez, et al., 2015), may be linked to higher domestic cat density (no official domestic cat census is available). Habitat changes in several parts of the rainy-temperate region, including extensive and intensive native forest habitat loss and fragmentation are probably also contributing to higher occupancy of humans and domestic cats (Echeverria et al., 2008;Sanderson et al., 2002;Wilson et al., 2005).
This study represents the first comprehensive evaluation of whether FeLV and FIV are causing disease in guignas. We found a low frequency of clinical signs (compatible with FeLV infection, but not specific) and no histopathological evidence of disease in infected guignas. However, our sample sizes were relatively small, and many cats died due to other acute causes such as vehicular trauma, precluding normal infection and disease course. Proviral DNA detected by molecular tests is integrated in the genome and could potentially reactivate and/or recombine with other viral subtypes, leading to emerging diseases and posing future threats for guignas, including potential population extinctions and impacting the species' long-term viability.
Feline immunodeficiency virus infection is usually endemic (Troyer et al., 2005). Because of a long history of co-evolution, felids are generally well adapted to their presence if ecosystems are stable and they develop immunity, so disease is not usually observed (Carpenter & O'Brien, 1995;Olmsted et al., 1992;Packer et al., 1999;Troyer et al., 2005). Evidence of morbidity and mortality caused by FIV infection have rarely been documented in free-ranging wild felids; it does not appear to have an immunosuppressive effect (Pedersen & Barlough, 1991;Roelke et al., 2009) or typically affects only individual animals rather than populations (Troyer et al., 2005).
In contrast, FeLV does not appear to be endemic for most free-ranging populations of wild felines, except for European wildcats (Felis silvestris; Millán et al., 2009). Therefore, it is likely that most free-ranging felid species have not developed resistance to this virus and may develop fatal disease due to its immunosuppressive and carcinogenic effects Marker et al., 2003). and serology negative (Hartmann, 2012). Regressive infection could potentially reactivate into progressive infection (Hartmann, 2012).
However, with our limited sampling we cannot calculate the proportion of regressive versus progressive infection (if any) in guignas.
Future studies should aim to sequence whole viral genomes to provide more complete evidence.
Predicting and preventing epidemics in wild felids will require a comprehensive understanding of the ecology and pathogenesis of infectious diseases. Only with continuous research and monitoring, accurate diagnostic tools and strong predictive models will we be able to prevent disease from emerging as a significant factor in wild felid survival. Managers and national authorities should consider limiting access of domestic dogs and cats to natural areas in combination with the expansion of disease management programs throughout Chile, including vaccination programs (as are already being implemented by the Ministry of the Environment in some areas), responsible pet ownership promotion, increased law enforcement of existing regulations and by facilitating veterinary care in rural landscapes. This study is the first comprehensive assessment of FeLV and FIV infection in guignas from across their entire distribution in Chile, enhancing our knowledge on pathogen transmission at the wildlife-domestic interface.

ACK N OWLED G EM ENTS
We gratefully acknowledge local inhabitants of rural communi-