Feline Leukemia Virus Frequently Spills Over from Domestic Cats to North American Pumas

ABSTRACT

Feline leukemia virus (FeLV) is a gammaretrovirus with horizontally transmitted and endogenous forms. Domestic cats are the primary reservoir species, but FeLV outbreaks in endangered Florida panthers and Iberian lynxes have resulted in mortalities. To assess prevalence and interspecific/intraspecific transmission, we conducted an extensive survey and phylogenetic analysis of FeLV infection in free-ranging pumas (n = 641) and bobcats (n = 212) and shelter domestic cats (n = 304). Samples were collected from coincident habitats across the United States between 1985 and 2018. FeLV infection was detected in 3.12% of the puma samples, 0.47% of the bobcat samples, and 6.25% of the domestic cat samples analyzed. Puma prevalence varied by location, with Florida having the highest rate of infection. FeLV env sequences revealed variation among isolates, and we identified two distinct clades. Both progressive and regressive infections were identified in cats and pumas. Based on the time and location of sampling and phylogenetic analysis, we inferred 3 spillover events between domestic cats and pumas; 3 puma-to-puma transmissions in Florida were inferred. An additional 14 infections in pumas likely represented spillover events following contact with reservoir host domestic cat populations. Our data provide evidence that FeLV transmission from domestic cats to pumas occurs widely across the United States, and puma-to-puma transmission may occur in genetically and geographically constrained populations.

IMPORTANCE Feline leukemia virus (FeLV) is a retrovirus that primarily affects domestic cats. Close interactions with domestic cats, including predation, can lead to the interspecific transmission of the virus to pumas, bobcats, or other feline species. Some infected individuals develop progressive infections, which are associated with clinical signs of disease and can result in mortality. Therefore, outbreaks of FeLV in wildlife, including the North American puma and the endangered Florida panther, are of high conservation concern. This work provides a greater understanding of the dynamics of the transmission of FeLV between domestic cats and wild felids and presents evidence of multiple spillover events and infections in all sampled populations. These findings highlight the concern for pathogen spillover from domestic animals to wildlife but also identify an opportunity to understand viral evolution following cross-species transmissions more broadly.

INTRODUCTION

Pathogen spillover from domestic animals to wildlife can cause devastating disease outbreaks that threaten the persistence of wildlife populations (1–5). The emergence of infectious diseases has the greatest effects on small, fragmented populations of animals and can result in the extirpation of endangered populations (4, 6, 7). Thus, an understanding of interspecific disease transmission can guide conservation efforts by informing management practices (e.g., vaccination guidelines and translocation) and revealing parameters that contribute to spillover risk.

The endangered Florida panther (Puma concolor coryi), a subspecies of the North American puma (Puma concolor), offers an example of a population that is at increased risk of extinction from disease spillover from domestic animals. Due to habitat fragmentation, declines in prey, and unregulated hunting during the 19th and 20th centuries (8), Florida panthers had become completely isolated in South Florida on 5% of their historic range (9) and declined to a minimum of 20 to 30 individuals by the early 1990s (10). The lack of gene flow with conspecifics, along with the reduced size of the population, led to inbreeding depression (8, 11), resulting in a multitude of negative impacts on various panther demographic, morphological, and biomedical parameters. The Florida panther appeared to be headed toward imminent extinction, which ultimately led to the initiation of a genetic rescue program via the release of eight female pumas from Texas into an occupied panther range in South Florida in 1995 in the hopes of alleviating the negative impacts of inbreeding and improving genetic variation. The introduction of Texas pumas increased the density of panthers in Florida and the survival of offspring (8, 12). Even with the benefits accrued to the population via genetic rescue (13), panthers remain endangered and at risk for various stochastic events, including intra- and interspecific disease transmission. For example, following interactions with feline immunodeficiency virus (FIV)-infected bobcats (Lynx rufus) in Florida, panthers have become infected with the bobcat-adapted strain of FIV (FIV_lru), eventually leading to intraspecific transmission within the panther population (14). Florida panthers have also been infected with pseudorabies virus (suid alphaherpesvirus 1) following predation of infected feral swine (Sus scrofa), which is a significant cause of death in Florida panthers (15). Importantly, outbreaks of feline leukemia virus (FeLV) have resulted in Florida panther mortalities (1, 16). Domestic cats (Felis catus) are the primary reservoir of FeLV (17), and sequence analyses of FeLV-infected panthers have implicated domestic cats as the origin of infections (1, 18). Transmission to panthers is hypothesized to occur during predation events (14, 17), and FeLV has been shown to infect other felid species, including Iberian lynxes (Lynx pardinus) (19) and captive jaguarundis (Puma yagouaroundi) (20), among others. Therefore, FeLV is of concern for the conservation of wild felids, including the Florida panther and other puma populations throughout North America.

Although FeLV infections have been well documented in the Florida panther (1, 16, 18), few other puma populations have been closely evaluated for FeLV incidence. We therefore tested pumas from diverse populations across North America (California, Colorado, and Florida) to determine the prevalence of the virus in wild populations. Puma populations in southern California, specifically the coastal region south of Los Angeles, are fragmented and exhibit low genetic diversity, similar to the Florida panther population (21, 22). In contrast, populations in eastern California and western Colorado exhibit high levels of genetic diversity and decreased levels of habitat fragmentation (21, 23). North American bobcats (Lynx rufus) have also been shown to have close interactions with both domestic cats and pumas, resulting in the transfer of pathogens, including FeLV (14, 24–27). Thus, we surveyed bobcats from overlapping regions for FeLV to compare prevalences in both free-ranging nondomestic felid species. Additionally, we assessed FeLV from free-ranging domestic cats to relate it to puma and bobcat FeLV prevalences and genotypes. Our study populations thus include an endangered, heavily managed, inbred puma population (Florida panthers); an urban population affected by habitat fragmentation that is not subject to hunting, with subregions of genetic restriction (California coastal populations); an exurban region impacted by increasing urbanization (Colorado Front Range); and more natural regions where the degree of inbreeding is low, one where regulated harvest is conducted (Colorado Western Slope) and one where hunting is prohibited (California Sierra Nevada Mountains). We additionally included two suspected positive samples collected from pumas in the Northern Cascade Mountains of Washington State to assess the FeLV genotype of an additional population experiencing an outbreak. This work therefore represents a large-scale survey of FeLV spillover from domestic cats to pumas across diverse habitats and populations.

RESULTS

FeLV prevalence.

We identified FeLV-positive samples using both quantitative PCR (qPCR) and Sanger sequencing in all three study areas (Fig. 1). We ultimately detected 20 positive pumas (20/641 [3.12%]), 19 positive domestic cats (19/304 [6.25%]), and 1 positive bobcat (1/212 [0.47%]) (Table 1). We found that the prevalences differed significantly among species (chi-square value, 12.98; P < 0.002). In each state, domestic cats represented the highest prevalence (California, 11/154 [7.14%]; Colorado, 5/109 [4.59%]; Florida, 3/41 [7.32%]), and bobcats accounted for the lowest prevalence (California, 0/89 [0%]; Colorado, 0/52 [0%]; Florida, 1/71 [1.41%]), as only a single bobcat from Florida was FeLV positive. The prevalence in pumas was intermediate to those in bobcats and domestic cats in every state (California, 5/228 [2.19%]; Colorado, 2/165 [1.21%]; Florida, 13/248 [5.24%]) (Table 1). Infected pumas represented a higher proportion of infections in Florida than in other states, and we found a significant difference among FeLV prevalences by state (chi-square value, 6.33; P < 0.05). Additionally, we found a slightly higher prevalence of FeLV in Colorado pumas in the Front Range (1/71 [1.41%]) than in pumas in the Western Slope region (1/94 [1.06%]), but the difference was not significant (chi-square value, 0.04, 1 degree of freedom; P > 0.20). We did not detect a significant difference in domestic cat FeLV prevalences by state (chi-square value, 0.80; P > 0.20). The values that we report in Table 1 were for samples with recovered FeLV sequences or samples that were positive on two independent qPCR runs but were not able to be amplified via sequencing.

FIG 1.

Geographic distribution of pumas sampled for feline leukemia virus (FeLV) testing in California (CA) (A), Colorado (CO) (B), and Florida (C). Positive samples are indicated with red (California), green (Colorado), and blue (Florida) triangles. Negative samples are indicated with gray triangles. Gray shading indicates urban regions.

TABLE 1.

Feline leukemia virus is present in pumas from California, Colorado, and Florida^a

Felid group	California		Colorado		Florida		Total
	No. of positive animals/total no. of animals tested	Prevalence (%) (95% CI)	No. of positive animals/total no. of animals tested	Prevalence (%) (95% CI)	No. of positive animals/total no. of animals tested	Prevalence (%) (95% CI)	No. of positive animals/total no. of animals tested	Prevalence (%) (95% CI)
Pumas	5/228	2.19 (0.94–5.03)	2/165	1.21 (0.33–4.31)	13/248	5.24 (3.09–8.76)	20/641	3.12 (2.03–4.77)
Domestic cats	11/154	7.14 (4.03–12.34)	5/109	4.59 (1.98–10.29)	3/41	7.32 (2.52–19.43)	19/304	6.25 (4.04–9.55)
Bobcats	0/89	0 (0.00–4.14)	0/52	0 (0.00–6.89)	1/71	1.41 (0.25–7.56)	1/212	0.47 (0.00–2.62)

^aThe prevalence varies significantly among species (chi-square value, 12.98; P < 0.002), with the highest prevalence in domestic cats and the lowest prevalence in bobcats, regardless of the state. There is also a significant association among puma prevalences by state (chi-square value, 6.33; P < 0.05), with a higher prevalence in Florida. Additionally, Florida has the highest prevalence of FeLV across all felid groups. No significant difference was found among domestic cat populations across states (chi-square value, 0.80; P > 0.20). CI, confidence interval.

FeLV proviral load.

Domestic cats had higher FeLV proviral loads than pumas (domestic cat average proviral load, 10^5.7 copies per 10⁶ cells; puma average proviral load, 10^5.03 copies per 10⁶ cells [P = 0.051]) (Fig. 2A), similar to work reported in a previous study (28). Based on previously developed proviral load criteria (28–30), most domestic cats (75%) and pumas (81%) included in our study demonstrate progressive infections associated with high FeLV proviral loads detected; however, we identified cats and pumas (5 domestic cats and 4 pumas) with proviral loads of less than 10⁴ per million cells, representative of regressive infection (31) (Fig. 2). FeLV proviral loads spanned approximately 6 logs in both species (Fig. 2).

FIG 2.

FeLV proviral loads vary by 6 orders of magnitude in naturally infected domestic cats and pumas. Puma samples are illustrated with dark-colored circles, and domestic cat samples are indicated with translucent squares. Blue samples originated from Florida, red samples originated from California, and green samples originated from Colorado. The dashed lines indicate the approximate boundary between progressive and regressive infections; most infections identified were progressive (proviral load of >10⁴ copies per million cells). (A) The mean proviral load was higher in domestic cats than in pumas (P = 0.051 by a Mann-Whitney test). The bimodal distribution of proviral loads indicates progressive (high proviral load) and regressive (low proviral load) infections in both domestic cats and pumas. (B) Proviral loads were similar between two FeLV clades identified in 26 samples that were sequenced (P = 0.182 by a Mann-Whitney test). (C) Florida panthers and Washington pumas had higher viral loads than California and Colorado pumas. Statistical significance was not determined due to the low number of positive samples in Colorado and Washington (n = 2). (D) Domestic cat proviral loads were similar across geographic sites (P = 0.54 by a Kruskal-Wallis test).

Phylogenetic analysis.

A phylogenetic analysis was conducted on 30 FeLV sequences (15 from cats, 14 from pumas, and 1 from a bobcat) (Fig. 3). In several instances, founder FeLV sequences in domestic cats could be linked to FeLV in pumas. We paired data on sequence similarity to geographic and temporal data to estimate the instances of FeLV spillover from domestic cats to pumas (n = 3) versus puma-to-puma transmission (n = 3). An additional 14 pumas were not directly linked to a sequenced domestic cat, but transmission is likely to be from domestic cats as they are known to be the primary reservoir, and these individuals represented single puma infections in areas where contemporary domestic cat FeLV sequences were not obtained (Table 2).

FIG 3.

Sequence analysis supports FeLV spillover from domestic cats to pumas across the United States. Maximum likelihood trees were constructed from the 2-kb segment of the env gene sequenced from 30 animals (15 domestic cats, 14 pumas, and 1 bobcat). Florida samples are indicated in blue, California samples are in red, and Colorado samples are in green. Sequences from two FeLV-positive pumas from Washington are shown in black. Pumas are indicated with boldface type, domestic cats are indicated with italic type, and one bobcat is indicated with plain text. Brackets indicate subclades and are color-coded by state, with the exception of the subclade found only in pumas, which is in black. (A) Two distinct clades of FeLV were documented in both domestic cats and pumas, with samples from California, Colorado, and Florida identified in each clade. Washington isolates fell within clade 1. Nonsignificant nodes are maintained within the tree for transparency and to demonstrate the extent of genotypic similarity among the samples. (B) Additional FeLV subtypes (FeLV-B, FeLV-C, FeLV-D, FeLV-T, and FeLV-TG35) are genetically distinct from the isolates in this study. Clade 1 isolates cluster with FeLV-FAIDS, a pathogenic isolate obtained from domestic cats in 1988 (51).

TABLE 2.

FeLV spillover from domestic cats occurs across the country^a

U.S. state	Puma ID	Sequence similarity to sequenced domestic cats	Sequence similarity to FeLV-positive pumas	Spatial proximity to FeLV-positive pumas (puma ID[s])	Sampling date (mo/yr)	Likely origin of transmission
Florida	x2270	Yes (and bobcat)	Yes	Yes (x2272)	8/2014	Presumed domestic cat and then puma to puma
	x2272	Yes (and bobcat)	Yes	Yes (x2270)	4/2015
	x2004	No	No	No	12/2010	Presumed domestic cat
	x1948	No	Yes	Yes (x1755, x1955)	1/2004	Presumed domestic cat and then puma to puma
	x1755	No	Yes	Yes (x1948, x1955)	5/2003
	x1955	No	Yes	Yes (x1948, x1755)	7/2004
	x1740	NA	NA	No	10/1985	Presumed domestic cat
	x1751	NA	NA	No	6/1995	Presumed domestic cat
	x1833	NA	NA	No	6/1988	Presumed domestic cat
	x1842	NA	NA	No	1/1990	Presumed domestic cat
	x2271	NA	NA	No	12/2014	Presumed domestic cat
	x2273	NA	NA	No GPS data	4/2016	Presumed domestic cat
	x2274	NA	NA	No GPS data	2/2016	Presumed domestic cat
California	x2881	No	No	No	8/2017	Presumed domestic cat
	x2874	Yes	No	No	3/2018	Domestic cat
	x109R1	Yes	No	No	11/2008	Domestic cat
	x394R2	No	No	No	10/2009	Presumed domestic cat
	x2849	NA	NA	No	3/2018	Presumed domestic cat
Colorado	x1131	Yes	No	No	2/2010	Domestic cat
	x424	No	No	No	12/2008	Presumed domestic cat

^aDark-gray-shaded boxes indicate puma-to-puma transmission following initial infection from a domestic cat (n = 3), determined by sequence similarity to other puma isolates, proximity in the time of sampling, and spatial proximity (GPS data from the location of sampling). Light-gray-shaded boxes indicate domestic cat-to-puma transmission (n = 3), determined by sequence similarity to domestic cat samples obtained from the same region. Boxes with no shading indicate samples that have an unknown origin but that are thought to have originated from domestic cats (n = 14). These include puma samples that were not sequenced (indicated by NA [not applicable] in sequence similarity columns) and puma sequences that were not closely related to a sequenced domestic cat sample.

Two distinct clades of FeLV were identified (Fig. 3A), which were distinguished by numerous nonsynonymous amino acid changes in the env gene. The majority of the amino acid changes are present in the portion of the env gene that codes for the surface protein (gp70) and are identified within the receptor binding domain of gp70 (32–34). Many of the amino acid changes cluster within the variable regions of the receptor binding domain that are associated with receptor binding (variable region A [VRA], variable region B [VRB], and the proline-rich region [PRR]) (33) (Fig. 4). Mutations in this part of the env gene are often associated with different subtypes of FeLV that bind to different receptors (1, 32, 35, 36). However, both clades cluster only with FeLV subtype A (FeLV-A) (Fig. 3B).

FIG 4.

Numerous amino acid changes found in variable regions of the FeLV receptor binding domain. The receptor binding domain is separated from the neutralization domain by the proline-rich region (PRR), and the gray-shaded regions indicate approximate variable regions of the env gene (variable region A [VRA] variable region B [VRB] and the PRR) that have been associated with receptor binding (33). (A) Amino acids that are unique to clade 1 (relative to clade 2) (shown by red vertical lines). (B) Amino acids that are specific to the Colorado subclade of clade 1 (relative to the rest of clade 1) (shown by blue vertical lines). (C) Amino acids that are specific to the California subclade of clade 2 (relative to the rest of clade 2) (shown by orange vertical lines). (D) Amino acids that are specific to the Florida subclade of clade 2 (relative to the rest of clade 2) (shown by purple vertical lines). (E) Amino acids that are specific to the subclade of clade 2 found only in puma samples (relative to the rest of clade 2) (shown by 1 green vertical line at the 5′ end of VRA). LTR, long terminal repeat.

Clade 1 primarily represents animals sampled from the Western United States, with the exception of one domestic cat sampled in Florida (Fig. 3). The linear regression analyses that we performed using TempEst (37) did not indicate any temporal association between unique sequences (R² = 0.34), and we did not document a significant difference in proviral loads between clades (Fig. 2).

DISCUSSION

We have documented interspecific FeLV transmission between domestic cats and pumas in three states, indicating that outbreaks of FeLV are a potential risk to puma populations (Fig. 3A). Most infections in pumas occur as a result of transmission from domestic cats in all geographic areas analyzed (Table 2). This work demonstrates disease spillover between domestic animals and wildlife in previously unrecognized regions and illustrates the importance of the continued monitoring of transmissible pathogens in wildlife.

We found the highest prevalence of FeLV in domestic cat populations (total prevalence, 6.25%) (Table 1). Our phylogenetic analysis emphasizes the role of domestic cats in the spillover of FeLV across the United States in that domestic cat FeLV sequences cluster closely with puma sequences in every region studied. Although some puma samples (for example, samples from pumas x2881, x394R2, and x424) do not demonstrate sequence similarity to FeLV isolates from geographically and temporally proximate domestic cats, we infer that domestic cats are the origin of infection as they are the known primary reservoir. The lack of sequence similarity between the isolates from domestic cats and the ones from these pumas likely reflects the incomplete sampling of domestic cats representing the incomplete detection of circulating viral diversity. As with other viruses infecting pumas and domestic cats, it is probable that infection is unidirectional, from domestic cats to pumas, as domestic cats rarely survive interactions with pumas (3). The fact that FeLV in pumas is not exclusively phylogeographically clustered (Fig. 3A) also suggests that the observed phylogenetic structure is likely governed primarily by FeLV transmission from domestic cats, which move large distances with people.

Pumas living in proximity to human developments have more interactions with domestic animals, and a higher percentage of their diets consists of domestic cats, which increases the likelihood of transmission of FeLV (38, 39). Exurban development is expanding into puma habitats and increasing the number of puma populations in proximity to humans (21, 40, 41). Disease outbreaks can threaten puma populations, as seen in past FeLV infections in Florida panthers, where progressive/persistent infections were frequently associated with clinical disease, and the death of these animals occurred quickly after viremia was discovered (16). Current populations of pumas in North America, including Florida panthers and puma populations in southern California, exhibit decreased genetic diversity due to reduced gene flow between populations and inbreeding (12, 21). These populations are at an increased risk of extinction due to stochastic events, including outbreaks of disease. For this reason, the prospects for the transmission of FeLV are of high importance for threatened or endangered populations, especially those for which the urban-wildlife interface is growing (12, 21, 42).

The clustering of FeLV sequences recovered from Florida panthers coupled with the time and location of capture supports our conclusion that puma-to-puma transmission occurred at least 3 times in Florida panthers, consistent with previous studies (Table 2) (1, 16). The onward transmission in Florida but not other regions could be the result of several factors, including increased contact due to the small and isolated population, the reduced genetic diversity of the population, and intensive population monitoring (12). While pumas are known to be solitary animals, previous work has shown that pumas with overlapping territories exhibit a higher tolerance for other pumas, leading to increased social interactions (43). Thus, the restricted area in Florida could be increasing the interactions between animals and the transmission of FeLV. Additionally, the panther population size has increased by at least 5-fold since the genetic restoration was initiated in 1995 (13). The increased number of panthers in the restricted habitat would also increase the probabilities of intraspecific interactions and viral transmission. Alternatively, we documented a higher prevalence and a higher proviral load in Florida panthers than in California and Colorado pumas (Table 1 and Fig. 2), indicative of progressive infections, which would have increased viral shedding and subsequent transmission to other individuals via saliva during social interactions or fighting (17, 44). Florida panthers have historically exhibited extremely low levels of genetic variation in comparison to other puma populations (12), which may contribute to a higher likelihood of progressive infection. Additional research into the role of puma genetic diversity in FeLV infections would provide insight into the risk of disease to puma populations with low genetic diversity and more generally would explore the link between host genetic diversity and susceptibility to viral infections.

Proviral loads in both cats and pumas suggest that the majority of documented infections are consistent with progressive disease associated with higher viremia and higher likelihoods of clinical disease and death (Fig. 2) (29, 31). We found fewer low-titer infections in domestic cats than would be predicted based on the known viral disease course measured under idealized, laboratory-based conditions (30). It is likely that we were biased in measuring high-titer, progressive infections due to the types of samples that were available. Puma samples were collected either opportunistically from carcasses or during field captures, which limited the ability to maintain a cold chain. In addition, some samples were collected and stored for 35 years prior to this analysis. Since DNA degradation has commonly occurred in these samples, we would disproportionately underdiagnose regressive infections where less viral DNA is present. Furthermore, sample degradation may have resulted in reduced assay sensitivity; therefore, we may have underestimated the true number of positive samples in each region. Poor sample quality also contributed to our inability to amplify full env genes for phylogenetic analysis in 6 of 20 FeLV-positive animals (confirmed in at least two independent triplicate qPCRs). Regardless of these technical considerations, the significant proportion of progressive FeLV infections in pumas indicates the importance for the conservation of the species.

A single bobcat from Florida was infected with FeLV. This is the first report of FeLV in a free-ranging bobcat, although other populations have been tested (45). Previous studies have indicated that bobcats are more likely than pumas to persist in urban environments and therefore have an increased potential for exposure to domestic cats (25, 46). The fewer bobcat FeLV diagnoses may indicate the less frequent predation on domestic cats by bobcats than by pumas (25, 26), which may result from domestic cat avoidance of bobcats and other mesocarnivores (46, 47). Previous work also demonstrated that pumas are exposed to numerous pathogens more frequently than bobcats (2), possibly due to a higher probability of lifetime exposure to prey-associated pathogens due to higher cumulative exposure to prey (2, 48–50).

Our phylogenetic analysis reveals two distinct clades of FeLV characterized by numerous amino acid changes throughout the env gene (Fig. 4). These clades were previously described by Watanabe et al. (36) as part of a subset of samples that were found only in Europe and North and South America. Samples from each species and state are present in each clade, and they did not cluster temporally (Fig. 3A). In addition, we found no significant difference in the proviral loads between the two clades, suggesting that one viral variant is not better able to replicate in pumas or domestic cats than the other (Fig. 2). Isolates in clade 1 (genotype III/1 [36]) share many single nucleotide polymorphisms with a replication-defective variant of FeLV, FeLV-feline AIDS (FeLV-FAIDS) (Fig. 4B), previously reported to result in immunodeficiency and high morbidity (51). Further research is required to assess the potential of unique receptor binding, the possibility of altered virulence associated with one clade versus the other, and the relevance for domestic cats and other felid species.

Phylogenetic analysis demonstrates moderate phylogeographic clustering (Fig. 3B). Clade 1 is composed predominantly of samples from the Western United States, with the exception of one domestic cat from Florida. No Florida panther samples are found in clade 1, suggesting that the majority of panthers from the Eastern United States are infected with FeLV of clade 2. There are also clear geographic subclades with closely related samples found in Colorado (clade 1), California (clade 2), and Florida (clade 2). However, geography is not the primary factor affecting phylogeny, as indicated by samples that are closely genetically related to samples from other parts of the United States (Fig. 3A). Additionally, there is a subclade of clade 2 composed only of samples isolated from pumas and panthers. This is likely to represent incomplete domestic cat sampling, and it is probable that additional domestic cat sequences would be closely linked to the puma samples in this subclade. Although the puma and panther sequences are closely related, it is unlikely that this represents direct transmission between pumas and panthers due to the geographic separation of the distinct populations. Florida panthers are the only puma population that persists in the Eastern United States (12); thus, it is more likely that the sequence similarities between puma and panther FeLV isolates represent incomplete domestic cat sampling.

With increasing interactions among humans, domestic animals, and wild populations, the transmission of infectious diseases among populations is likely to increase in frequency. We have previously shown the transmission of feline immunodeficiency virus (FIV) from bobcats to pumas (14), feline foamy virus (FFV) (also known as feline spumavirus) from domestic cats to pumas (3), and Mycoplasma haemominutum from domestic cats to bobcats and pumas (26). While feline foamy virus (FFV) cross-species transmission has resulted in high infectivity but low pathology in the recipient host, bobcat FIV has been documented to be poorly adapted to pumas and was transmitted within pumas only in Florida (50). Here, we have demonstrated that FeLV of domestic cats represents a source of spillover of a pathogenic virus into wild felid populations, with variable outcomes depending on the population-level features of the recipient host. This work illustrates the need for the continued monitoring of the interspecific transmission of pathogenic viruses among distinct populations and further studies of pathogen-host interactions that lead to susceptibility, transmission, and disease outcomes.

MATERIALS AND METHODS

Ethics statement.

Puma and bobcat samples were collected as part of ongoing studies at the California Department of Fish and Wildlife (CDFW) between 1998 and 2018, Colorado Parks and Wildlife (CPW) between 2005 and 2014, and the Florida Fish and Wildlife Conservation Commission (FWC) between 1985 and 2016 and were provided to Colorado State University (CSU) for diagnostic evaluation. Domestic cat samples were collected by collaborating shelters and sent to CSU during the same period. Blood and tissue samples from these studies have been archived and reused for many unique analyses. The CSU, CDFW, CPW, and FWC Institutional Animal Care and Use Committees reviewed and approved this work prior to initiation. This work was performed in accordance with the U.S. Department of Agriculture Animal Welfare Act and the Guide for the Care and Use of Laboratory Animals (52). The CSU public health assurance number is D16-00345. CSU is accredited by AAALAC International.

Sample sources.

We used blood and tissue samples from 641 free-ranging pumas in California, Colorado, and Florida and 2 free-ranging pumas in Washington. All samples were provided by state agencies after being collected opportunistically or via other ongoing research in each state. GPS coordinates were provided for the collection location for each animal (2, 25). We used ArcGIS software to generate maps with sites of sample collection based on the GPS coordinates for collection locations. In California, 228 samples were collected across the state from 1998 to 2018, and most were from documented mortalities. In Colorado, 165 samples were taken from 2005 to 2014, primarily from puma populations within the Front Range and the Western Slope. These animals were opportunistically sampled and released during a study investigating puma-human interactions, including the impacts of hunting (53–55). A total of 248 panther samples were collected in Florida from 1985 to 2015 from ongoing studies, routine monitoring, and documented mortalities. The samples were from across the range of the Florida panther, with a concentration of samples from the Big Cypress National Preserve (1). Two additional samples were received from suspected positive pumas collected in the Northern Cascades in Washington State in 2007 for diagnostic analysis. These samples were FeLV positive and therefore were included in the sequencing analysis to assess the genotypic relationship of FeLV to isolates from other puma populations, but they were omitted from the prevalence data since population-level surveillance data and other demographic information were unavailable. Bioarchived bobcat samples (n = 212) from California, Colorado, and Florida previously collected for pathogen analysis using documented methods (2, 56) were also tested. For pumas and bobcats, whole blood was typically sampled during capture-and-release studies, while tissue samples (spleen, liver, kidney, lymph node, and thymus) were collected and analyzed from postmortem samples. Whole-blood samples from archived free-ranging domestic cats were collected from shelters in each state (n = 304) to serve as direct comparators for prevalences and FeLV sequences in each region (2). In Florida, three samples from domestic cats known to be positive for FeLV were received for sequencing from shelters and veterinary clinics. These were included in our phylogenetic analysis but were omitted from our prevalence data to achieve a more representative sample.

DNA extraction.

Blood sample preparations used for extraction included isolated white blood cells, peripheral blood mononuclear cells (PBMCs), buffy coat samples, and/or clot samples in addition to whole blood. For DNA extraction from blood samples, we processed 100 μL of the whole blood or clot per sample, and we used 50 μL of white blood cells, PBMCs, and buffy coat samples for processing, due to the higher DNA concentration in the latter types of samples. The subsequent reagent quantities were made proportional to equalize the volume of each sample during processing. We processed the whole-blood and tissue samples using a DNeasy blood and tissue kit (Qiagen, Inc., Valencia, CA, USA). We modified the Qiagen protocol for blood samples at the first incubation step by incubating the samples for up to 2 h. To increase the DNA yield from the samples, we also changed the final elution step by heating 50 μL of buffer EB to 55°C prior to eluting the DNA twice in the same tube after allowing the EB buffer to incubate on the column at room temperature for 4 min.

Tissue samples included the spleen, liver, kidney, and thymus. We modified the DNA extraction protocol for tissue samples by homogenizing approximately 20 mg of the tissue using a FastPrep instrument after adding 450 μL of buffer ATL and 50 μL of proteinase K to each sample. We then centrifuged each sample for 10 min at 14,000 × g to pellet the debris and transferred 200 μL of the supernatant to a microcentrifuge tube. After adding 200 μL of both buffer AL and 100% ethanol to each sample and mixing by vortexing, we conducted the remainder of the protocol as recommended by the manufacturer (Qiagen, Inc., Valencia, CA, USA). For the final elution step, we heated 50 μL of buffer EB to 55°C and incubated the sample at room temperature for 4 min before centrifugation at 8,000 rpm for 1 min. We repeated this step to increase the yield of DNA. Following DNA extractions, we used a NanoDrop 8000 spectrophotometer (Thermo Fisher Scientific, USA) to quantify the DNA in each sample.

FeLV real-time qPCR sample screening.

We evaluated 641 puma samples, 304 domestic cat samples, and 212 bobcat samples for FeLV provirus using a real-time quantitative PCR assay developed previously by Torres et al. (29). This assay specifically amplifies exogenous FeLV sequences (i.e., “FeLV-A”) and does not detect endogenous FeLV present in domestic cats. For each sample, we used 12.5 μL of iTaq universal probe supermix (Bio-Rad, USA), 1 μL of FeLV-A forward (F) primer 5′-AGTTCGACCTTCCGCCTCAT-3′ (10 μM), 1 μL of FeLV-A reverse (R) primer 5′-AGAAAGCGCGCGTACAGAAG-3′ (10 μM), 1 μL of FeLV-A probe 5′-TAAACTAACCAATCCCCATGCCTC-3′ (2 μM), 7.5 μL of water, and 2 μL of the DNA template. We ran DNA samples in triplicate. Negative controls (DNA from domestic cats known to be negative), positive controls (DNA from domestic cats known to be naturally infected with FeLV), and no-template controls (purified water) were tested in each independent run. We used a CFX Connect real-time PCR detection system (Bio-Rad, USA) to perform the reactions under the following cycling conditions: 95°C for 3 min followed by 40 cycles with the following conditions: 95°C for 5 s and 60°C for 15 s. Samples were considered positive only if all three replicates were positive, with a minimum limit of detection of 10 copies per reaction. If two samples were positive within a single run, we reran the samples, and the sample was considered positive if all three replicates were positive in the subsequent reaction. We attempted to sequence each sample that was qPCR positive. If sequencing failed, additional qPCR analysis was conducted, and samples were considered positive but unsequenced if all three replicates demonstrated positive results. We determined the prevalence of FeLV in each state and species based on the number of positive samples compared to the total number of tested individuals. Confidence intervals were calculated according to the method of Wilson (57).

qPCR quantification of FeLV proviral loads.

To quantify proviral copy numbers in each sample, we normalized each assay using the feline CCR5 gene specific to each felid species. qPCR for FeLV reactions and CCR5 and standard curves were conducted on the same plate (58). On one half of a 96-well plate, we used the protocol outlined above for the FeLV-A qPCR assay. On the other half of the plate, we used the same samples but performed CCR5 qPCR. Each sample contained 12.5 μL of iTaq universal probe supermix (Bio-Rad, USA), 2 μL of puma or domestic cat CCR5 F primer (2.5 μM), 5 μL of puma or domestic cat CCR5 R primer (2.5 μM), 2 μL of puma or domestic cat CCR5 probe (2.5 μM), 1.5 μL of water, and 2 μL of the DNA template. Puma and domestic cat CCR5 primer sequences were described previously by Chiu and VandeWoude (28). Standard curves were generated using custom synthetic nucleotides (gBlocks; IDT), which contained oligonucleotide sequences of FeLV-A and CCR5 targeted by FeLV and puma CCR5 primers (28). We ran all reactions on a CFX Connect real-time PCR detection system (Bio-Rad, USA) under the following cycling conditions: 95°C for 3 min and 40 cycles at 95°C for 5 s and 60°C for 15 s. We quantified the proviral load in each sample as previously described by using SQ values that were calculated using the standard curve for each qPCR plate (31). Since P. concolor does not harbor endogenous FeLV, and FeLV-A primers do not detect endogenous FeLV, all positive results indicate infection with exogenous FeLV (1).

Isolation of provirus from positive samples using PCR.

We used nested PCR to recover proviral sequences for Sanger sequencing. We targeted the env gene region of the FeLV sample as this region exhibits the most variation and allows the characterization of FeLV subtypes (1, 35, 59). Four sets of primers were used to amplify overlapping amplicons, as illustrated in Table 3.

TABLE 3.

Primer sets used to amplify FeLV env^a

env segment	Step	Sequence (5′→3′)		Annealing temp (°C); elongation time (min)	Approximate amplicon size (bp)
		Forward primer	Reverse primer
2 kb	1	GCGGGTCCATTATCTGAACCCAATACC	GAGCCTGGAGACTGCTGGTA	61.1; 3	3,000
	2	CGCCTCACATGTAAAGGCTGCA	TAGCTGGCTAAGTTTTGGGGTAGG	61.1; 2	2,000
Tile 1	1	ACGGAGTTGCTGCTTGGATC	CTAGGCCCCGTCTGTTTTGT	54.3; 2	1,750
	2	CCCCAGCTCAGACGATCCAT	GCTATGCCCCCTACAGTGAG	54.3; 1.5	1,525
Tile 2	1	CCTCAGGCAATGGGACCAAA	CCTTTCTCTGGGGACTAAATGGA	60.3; 1.5	1,300
	2	AAACCCCCATCCCGACAATC	TTATCTGTTGGTACTGTTGGGT	52.4; 1.25	1,200
500 bp		CTCAGGCAATGGGACCAAA	CTCAGGCAATGGGACCAAA	59.3; 0.5	500

^aPrimers for tile 1, tile 2, and the 500-bp section of env were designed to amplify short segments of FeLV to increase the diagnostic sensitivity for degraded samples. Unique exogenous FeLV env sequences were targeted to avoid the amplification of domestic cat endogenous FeLV (enFeLV), permitting the amplification of FeLV from both puma and domestic cat samples.

We performed each nested PCR using 10 μL of Kapa Hotstart HiFi polymerase (Kapa Biosystems, USA), 1 μL of both the forward and reverse primers (see Table 3 for primer specifications), 4 μL of water, and 2 μL of the DNA template per sample. We performed the second step of each nested PCR in the same manner, using 2 μL of the PCR product obtained from the first reaction for all subsequent reactions except for the amplification of the env region tile 1, which was performed using 1 μL of DNA and PCR products from each reaction and 5 μL of water. For PCR targeting the 500-bp section of env, we used 12.5 μL of Kapa Taq polymerase (Kapa Biosystems, USA), 1 μL of both the forward and reverse primers (Table 3), 8.5 μL of water, and 2 μL of the DNA template per sample. The cycling conditions were as follows: 95°C for 3 min followed by 35 cycles with the following conditions: 98°C for 20 s, the variant annealing temperature (Table 3) for 15 s, 72°C for a time specific for the expected length of the product (Table 3), followed by 72°C for 6 min. We ran each reaction on a C1000 Touch thermal cycler (Bio-Rad, USA). We resolved PCR products on a 0.7% agarose gel using electrophoresis for 30 min at 80 V. To prepare samples for cloning, we excised bands of the correct size for each PCR from the gel and purified the PCR products using a MEGAquick-spin total fragment DNA purification kit (iNtRON Biotechnology, South Korea).

Recovery of proviral sequences and cloning.

We cloned the purified PCR products into Escherichia coli XL2-Blue ultracompetent cells (Agilent Technologies, USA) from pJET 1.2 blunt vectors using the CloneJET PCR cloning kit (Thermo Fisher Scientific, USA). To ensure that each bacterial colony contained a PCR product of the correct size, we used screening PCR on each colony. Each reaction mixture contained 5 μL of Kapa Taq polymerase (Kapa Biosystems, USA), 4 μL of water, and 0.5 μL of each primer used in the second step of the nested PCR (Table 3). The cycling conditions were as follows: 95°C for 3 min; 35 cycles of 95°C for 30 s, 60°C for 30 s, and 72°C for 2 min; and 72°C for 5 min. For plasmids containing a PCR product of the correct size, we isolated and purified the plasmids using the DNA-spin plasmid purification kit (iNtRON Biotechnology, South Korea). The plasmids were Sanger sequenced, and we verified and assembled the genetic sequences using Geneious Prime (Biomatters Ltd., New Zealand). We compiled the data sets for the env genes together with sequences obtained from GenBank.

Construction of phylogenetic trees.

We used RDP4 software (60) to create a recombination-free env gene nucleotide data set for the construction of a maximum likelihood phylogenetic tree using samples sequenced in this data set, including domestic cat and puma samples from California, Colorado, and Florida (Fig. 3A). We constructed an additional maximum likelihood phylogenetic tree with additional sequences of other FeLV subtypes, including FeLV-B, FeLV-C, FeLV-D, FeLV-T, FeLV-TG35, and FeLV-FAIDS, a replication-defective variant of FeLV-A (Fig. 3B). Friend murine leukemia virus (GenBank accession number NC_001362) was included as an outgroup in both trees. To find the best-fit substitution models, we used MEGA X. A TN93+I model was used to produce the maximum likelihood trees in Fig. 3A, and a TN93+G model with 5 discrete gamma categories was used to produce maximum likelihood trees with the additional FeLV subtypes (Fig. 3B). We performed 1,000 iterations to calculate bootstrap values. We performed a linear regression of the root-to-tip divergences of the maximum likelihood tree using the TempEst program to assess the temporal relationships among the FeLV env sequences (37).

Statistical analysis.

We estimated the statistical significance of FeLV prevalence using a chi-squared test with 2 degrees of freedom (61). We tested for significance by comparing prevalences across all puma populations, all domestic cat populations, and each species. We calculated the statistical significance of FeLV proviral loads by using a Mann-Whitney test (62) to compare the proviral loads between domestic cats and pumas and between clade 1 and clade 2. We did not calculate the significance of proviral loads between pumas in different states due to the inadequate sample size of positive pumas in Colorado and Washington (n = 2). The significance of proviral loads between domestic cats in different states was calculated using a Kruskal-Wallis test (63).

Data availability.

All new FeLV sequences in the phylogenetic tree have been uploaded to the NCBI GenBank database with accession numbers ON995415 to ON995434. Some sequences in the phylogenetic tree in this study were sequenced by a coauthor in a previous study. The NCBI GenBank accession numbers for these samples are as follows: MH116004 for sample x2653, MF681664 for x1613, MH116005 for x2655, MF681668 for x2272, MF681667 for x2270, MF681665 for x2004, MF681672 for x1948, MG020273 for x1955, and MG020272 for x1755. All sample metadata, which include the sample identification numbers, demographic data, and results of FeLV testing for each sample, are available at Dryad at https://doi.org/10.5061/dryad.9cnp5hqn4.