**Eco-Epidemiological Profile and Molecular Characterization of Simian Foamy Virus in a Recently-Captured Invasive Population of** *Leontopithecus chrysomelas* **(Golden-Headed Lion Tamarin) in Rio de Janeiro, Brazil**

**Thamiris S. Miranda 1, Cláudia P. Muniz 1,2, Silvia B. Moreira 3, Marina G. Bueno 4,5, Maria Cecília M. Kierul**ff **4,6, Camila V. Molina 4, José L. Catão-Dias 7, Alcides Pissinatti 3, Marcelo A. Soares 1,2 and André F. Santos 1,\***


Received: 6 August 2019; Accepted: 29 September 2019; Published: 10 October 2019

**Abstract:** Simian foamy viruses (SFV) infect a wide range of Old World and Neotropical primates (NP). Unlike Old World primates, little is known about the diversity and prevalence of SFV in NP, mainly from a free-living population. Phylogenetic analyses have shown that SFV coevolved with their hosts. However, viral strains infecting *Leontopithecus chrysomelas* did not behave as expected for this hypothesis. The purpose of this study was to determine the eco-epidemiological profile and molecular characterization of SFV in a recently captured invasive population of *L. chrysomelas* located in Niteroi/RJ using buccal swab as an alternative collection method. A prevalence of 34.8% (32/92) and a mean viral load of 4.7 log copies of SFV/106 cells were observed. With respect to time since capture, SFV prevalence was significantly higher in the group of animals sampled over 6 months after capture (55.2%) than in those more recently captured (25.4%) (*p* = 0.005). Infected solitary animals can contribute to SFV transmission between different groups in the population. SFV strains formed two distinct clades within the SFV infecting the Cebidae family. This is the first study to use buccal swabs as a tool to study SFV diversity and prevalence in a recently free-living NP population upon recent capture.

**Keywords:** spumavirus; viral prevalence; epidemiology; Neotropical primates; free-living primates; Brazil

#### **1. Introduction**

Simian foamy viruses (SFV) are complex retroviruses that naturally infect a wide range of non-human primates, including Neotropical primates (NP) [1–3]. Phylogenetic analyses have indicated that SFV coevolved with nonhuman primates for at least 60 million years [4], contributing to the lack of pathogenicity observed in these animals [5]. In NP, the prevalence is generally higher in animals in captivity (45–51%) compared to animals in the wild (14–30%) [2,3]. Although SFV has been described in at least 23 species of NP [3], there are only five complete genomes sequenced [6–9], a small number considering the high diversity of NP, distributed in at least 176 species, 17–21 genera and three to five families according to distinct classification systems [10,11]. Although the first NP SFV has been identified over four decades ago in cell cultures of spider monkey (*Ateles sp*.) saliva [12], very little is known about the distribution, prevalence, and genetic variability of SFVs that infect this group, and most studies have been conducted with captive animals. Although there are two studies reporting SFV prevalence in free-ranging NP, they are restricted to a limited number of available specimens and species [2,3]. Therefore, the epidemiological profile of SFV for a given NP species and/or genera in the wild is at least an inaccurate estimate since it has never been evaluated at the population level.

*Leontopithecus chrysomelas* (golden-headed lion tamarin) is a small size NP belonging to the Cebidae family [13] categorized as *EN-Endangered* by The International Union for Conservation of Nature (IUCN) [14] and it is endemic in the south of Bahia state, Brazil [15]. However, a few *L. chrysomelas* individuals have been introduced into an urban Atlantic Forest fragment in Niterói city (Rio de Janeiro state, Brazil) by a private collector in the mid-90s, being considered as an exotic invasive species in this region [16]. This invasive population have had close contact with humans and domestic animals, entering at human houses and being fed by them, increasing the risk of virus transmission in both directions [17–20]. Moreover, the few introduced animals reproduced, becoming hundreds of animals, estimated in excess of 700 in late 2015 [21], and could be a threat to the local golden lion tamarin (*Leontopithecus rosalia*), an endangered species endemic to Rio de Janeiro state, with risks of disease transmission [14–16], competition by habitat and hybridization [16]. For those reasons, many *L. chrysomelas* family groups were captured as part of a conservation project to remove this introduced species, administered and conducted by the non-governmental organization Pri-Matas Institute since 2012. The captured animals were kept in quarantine at *Centro de Primatologia do Rio de Janeiro* (CPRJ; Guapimirim, RJ, Brazil) and between 2012 and 2013 some were translocated to an area in southern Bahia without *L. chrysomelas* and others groups were maintained in captivity [16].

Yet retroviruses have a close phylogenetic relationship with their hosts [22], the dynamics of infection can be influenced by ecological and behavioral factors, impacting their prevalence, virus–host interactions, within- and between-species transmission [23,24], and also transmission to the surrounding human population [25–27]. SFV transmission occurs mainly through bites and grooming [28]. Therefore, social behaviors that increase contact between individuals may potentiate the likelihood of SFV transmission, impacting SFV prevalence rates. In NP, there are different complex social and behavioral structures [29,30]; however, little is known about how these structures and anthropogenic actions impact the viral ecology of SFV in free-living animals. In a phylogenetic analysis early study by our group, the only SFV sequences obtained from a *L. chrysomelas* and a *L. rosalia* did not cluster to form a single clade for *Leontopithecus* [3]. Here, for the first time in NP, a large number of recently-captured *L. chrysomelas* specimens were analyzed, allowing us to deepen our knowledge on SFV prevalence, circulating viral genetic diversity, and how social behaviors and the environment may influence SFV transmission in this population.

#### **2. Materials and Methods**

#### *2.1. Study Population and Ethics Statement*

Buccal swab samples were collected from of 92 *L. chrysomelas* captured in Niterói city (Rio de Janeiro state, Brazil) in the period from December 2014 to September 2017. The captured specimens were distributed in 29 family groups and 4 were found to be solitary (captured alone). Each *L. chrysomelas* family group and solitary specimen was kept separated after their capture and have their material collected from two to fourteen months. Specimens were classified as two groups, those collected from two to six months after their capture and those collected from seven to fourteen months after their capture. The sexual maturity was classified according to size, dentition and weight of the specimens and these data, as well as gender, geographic location of capture and family groups were made available. The material and information were collected by the CPRJ and Pri-Matas veterinarians.

All procedures were conducted in full compliance with Federal permits issued by the Brazilian Ministry of the Environment (SISBIO 30939-12) and samples were collected following the national guidelines and provisions of IBAMA (Instituto Brasileiro do Meio Ambiente e dos Recursos Naturais Renováveis, Brazil; permanent license number 11375–1). The project was approved by the Ethics Committee on the Use of Animals (CEUA/CCS) of Universidade Federal do Rio de Janeiro, under the reference number 037-14.

#### *2.2. Sample Collection, Processing and Confirmation of Genomic DNA Integrity*

Buccal swabs were collected using sterile cotton swabs with a plastic shaft that was then placed into a sterile tube containing 500 μL of saline solution (0.9% NaCl), transported to the Genetics Department of Universidade Federal do Rio de Janeiro on ice and were stored at −80 ◦C until processing. Genomic DNA (gDNA) was extracted from buccal swab samples using the PureLink®Genomic DNA kit (ThermoFisher Scientific, Grand Island, NY, USA) according to the manufacturer's specifications. After extraction, samples had their contaminants (PCR inhibitors) removed using the OneStep™ PCR Inhibitor Removal Kit (Zymo Research, Irvine, CA, USA). Shortly thereafter, samples were quantified using Nanodrop and stored at −20 ◦C. The integrity of the gDNA for PCR analysis was checked by PCR amplification of a mitochondrial constitutive gene (*cytB*) as previously described [1]. All DNA samples testing positive for *cytB* sequences were further considered suitable for SFV PCR detection.

#### *2.3. Detection and Quantification of SFV*

To detect NP SFV proviral DNA, we first performed a screening semi-nested PCR for short integrase sequences of 192 bp using generic primers and standard PCR conditions as previously described [1] as a diagnostic PCR test for NP SFV using 12 ng of buccal gDNA. In addition to the conventional diagnostic PCR for SFV detection, a real-time PCR assay was also performed to detect and quantify SFV viral copies in buccal swab samples, targeting a 124 bp region of the *pol* gene as previously described [31]. Primers and probes were designed using an alignment of available *pol* sequences from NP SFV, including representatives from all three NP families [1]. Briefly, one forward and one reverse primer were used (QSIP4Nmod (for) 5 -TGC ATT CCG ATC AAG GAT CAG C-3 and QSIR1Nmod2 (rev) 5 - TTC CTT TCC ACY WTY CCA CTA CT-3 ), with the probe DIAPR2 5 -FAM-TGG GGI TGG TAA GGA GTA CTG WAT TCC A-SpC6-3 . Following a 10 min incubation at 95 ◦C to activate Taq polymerase, a three-step PCR was performed at 95 ◦C for 15 sec, 50 ◦C for 15 sec, and 62 ◦C for 15 sec for 55 cycles using the 7500 Real-Time PCR platform (Applied Biosystems, Foster City, CA, USA). The sensitivity of the assay was 100 copies of SFV/reaction, as determined in [31].

To normalize the amount of diploid cells per reaction, the mean number of housekeeping gene ribonuclease P/MRP 30 kDa subunit (*RPP30*) copies of five *L. chrysomelas* swab samples (284 copies/ng) was used, as described previously [31]. Thus, since each cell has two copies of the RPP30 gene, the mean used was 142 cells/ng of DNA in buccal swab samples of *L. chrysomelas*.

#### *2.4. Amplification of a Larger SFV Fragment from the Cebidae Family*

For the positive samples in at least one of the SFV detection tests (diagnostic PCR or real-time PCR), one additional PCR was carried out to amplify a larger SFV subgenomic region for phylogenetic analysis. Despite the controversy about the number of NP families, the *Leontopithecus* genera was classified as part of the Cebidae family in the present report [11]. Primers were then designed using a conserved region of *pol* in an alignment of two SFV complete genomes representing the Cebidae family available at GenBank from marmoset and yellow-breasted capuchin (accession numbers GU356395 and KP143760, respectively) [7,9] and *pol* sequences generated by the diagnostic PCR test in this study as describe above. Briefly, the new nested PCR was performed using primers: (1◦ Round: *pol*5474 5 GCCAAACATGAGAAAGGATG 3 and *pol*5960 5 TACCACTTTGTAGGTCTTCC 3 with annealing temperature of 53.4 ◦C) and (2◦ Round: *pol*5500 5 GTCATATCCGTAYGTGCAAAC 3 and *pol*5878 5 CTTTGGGGGTGGTAAGG 3 with annealing temperature of 56 ◦C), amplifying a 378 bp fragment.

In addition, primer combinations were tested in the 2◦ round above to analyze the amplification efficiency of the viral fragments. The *pol*5474 *and pol*5878 primers (annealing temperature of 54 ◦C), and *pol*5500 with *pol*5960 (annealing temperature of 56 ◦C) were combined, amplifying fragments of 404 bp and 460 bp, respectively. PCR were performed with an initial temperature of 94 ◦C, followed by 35 cycles of 94 ◦C for 30 seconds, primer annealing at temperatures varying according to the combination of primers for 30 sec and 72 ◦C for 90 sec. Products were sequenced by the Sanger method using and ABI 3130 XL automatic sequencer (Applied Biosystems) and primers of the second round.

#### *2.5. Sequence Analysis*

Generated sequences were submitted to the BLASTn tool (http://blast.ncbi.nlm.nih.gov) for similarity analysis with SFV sequences deposited at Genbank. SFV sequences were edited using the SeqMan program v.7.0 (DNASTAR, Madison, WI, USA) and aligned with NP SFV reference sequences deposited in GenBank using the BioEdit Sequence Alignment Editor v.7.0.4 [32]. From the alignment, a phylogenetic tree was generated with Mega7 [33], using the maximum likelihood method and the Tamura 3-parameter correction model with discrete gamma rate variation. The bootstrap method was used with 1000 replicates to estimate the reliability of the phylogenetic clusters. Values above 70% were considered significant [3]. A similarity analysis between SFVlcm strains in the *pol* sequence was conducted using the Maximum Composite Likelihood model. The analysis involved 13 nucleotide sequences. All positions containing gaps and missing data were stripped. There was a total of 263 nucleotide positions in the final dataset. Evolutionary analyses were conducted in MEGA7. The geographical distribution of the *L. chrysomelas* family groups was plotted using the coordinates of family groups using program RStudio [34] using package Leaflet OpenStreetMap® contributors under license Open Data Commons Open Database License (ODbL).

#### *2.6. Statistical Analyses*

To understand better the epidemiology of SFV in *L. chrysomelas,* differences in SFV prevalence were evaluated with Chi Square trend analysis for the following categories: males and females; and infants, juveniles and adults. We also divided the familiar groups according to the time elapsed between capture and collection: one group between 1 and 6 months of captivity time (*n* = 63) and another group between 7 and 14 months in captivity (*n* = 29). After logarithmic transformation of SFV viral load data, *T*-tests were performed to test for associations between viral loads and all the characteristics mentioned above. For low sample numbers (*n* < 20), the Fisher Exact Test was used to evaluate epidemiological prevalence between groups with different captivity times and the efficiency of amplification by the diagnostic PCR between the two strains of SFVlcm described in this study. For all these tests, *p*-values ≤ 0.05 were considered significant.

#### *2.7. Data Availability*

All SFV sequences generated herein have been deposited at GenBank with the accession numbers MN178627 to MN178637.

#### **3. Results**

#### *3.1. Population Profile*

Samples of 29 golden-headed lion tamarin family groups were collected in this study. The mean number of specimens captured was eight per group, ranging from three to 12, while the mean number of specimens collected was three by group, ranging from one to seven animals. Of the 92 animals collected at CPRJ, we found a higher proportion of males (56.5%) than females (43.5%) (Table 1). With respect to sexual maturity, the specimens analyzed were constituted largely by adults (*n* = 45; 49%), followed by juveniles (*n* = 32; 35%) and infants (*n* = 15; 16%). All gDNA samples extracted from buccal swabs were positive for the constitutive mitochondrial *cytB* gene, and therefore were considered suitable to the molecular tests for SFV detection and quantification.

**Table 1.** Comparison of simian foamy virus (SFV) prevalence estimates by quantitative PCR (qPCR) and conventional PCR (cPCR) in relation to sex and the sexual maturity of *L. chrysomelas.*


#### *3.2. SFV Molecular Detection and Quantification*

A sample was considered positive for SFV infection when it tested positive in either one of the two molecular tests used (conventional diagnostic PCR and/or qPCR). Using this criterion, 15 samples (16.3%) were positive by diagnostic PCR and 28 (30.4%) were positive by qPCR (Table 1). When comparing the two assays, the results were 70% concordant. Of the discordant results, qPCR was more sensitive (17%) than the conventional diagnostic PCR (4%) (*p* = 0.006). Thus, 32/92 (34.8%) of the animals were considered infected with SFV (Table 1). Females and males presented similar SFV prevalence (40%) and (30.7%), respectively (*p* = 0.483). Regarding sexual maturity, no statistical difference was observed in the prevalence of SFV infection between different groups (*p* = 0.502) (Table 1). By grouping infants with juvenile specimens and comparing with adults, the SFV prevalence between immature and mature animals was very similar, 17/47 (36.2%) and 15/45 (33.3%), respectively (*p* = 0.946).

We sought to address whether the low sensitivity of the conventional diagnostic PCR was related to the lower number of SFV DNA copies of the negative samples for diagnostic PCR, but positive for qPCR, but there was no correlation between those conditions (*p* = 0.175). Among the 28 samples that had detectable SFV DNA VL, after normalization with the mean RPP30 copies in buccal swab cells, the mean VL was 4.7 log copies of SFV/106 cells, ranging from 3.47 to 5.98 log copies/106 cells. No differences were found between oral SFV DNA VL of males (*n* = 14) and females (*n* = 14) (mean of 4.7 log and 4.8 log copies/10<sup>6</sup> cells, respectively; *p* = 0.735). Regarding sexual maturity, the mean DNA VL were also similar between the different groups: infants (*n* = 4), juveniles (*n* = 10) and adults (*n* = 14) with means of 4.7, 4.6 and 4.7 log copies/106 cells, respectively (*p* = 0.254).

With respect to time in captivity, SFV prevalence was lower in animals kept in captivity within 1–6 months (25.4%; 16/63) than in animals that stay in captivity more than seven months (55.2%; 16/29) (*p* = 0.005) (Table 2). The SFV prevalence among females also differed in the two groups with 32% (9/28) in the former group and 66.6% (8/12) in the latter (*p* = 0.042). The same was observed to male infections, with 20% (7/35) in the shorter captivity group and 47.1% (8/17) in the longer captivity group (*p* = 0.043). The SFV prevalence was also higher in the longer captivity group among all different age

groups. However, only in juveniles had SFV prevalence reached a borderline statistical significance (*p* = 0.055; Table 2).


**Table 2.** Comparison of SFV prevalence in relation to sex and sexual maturity of groups classified according to captivity time at Centro de Primatologia do Rio de Janeiro (CPRJ).

#### *3.3. Phylogenetic Analysis and Similarity of SFV from L. Chrysomelas*

To perform a phylogenetic analysis and to infer the evolutionary history of the SFV that infect this population of *L. chrysomelas*, it was necessary to amplify larger PCR fragments. Of the 32 SFV previously positive samples, three samples amplified a 378 bp fragment with the 5500 and 5878 primer combination (see Methods); another six samples amplified a 404 bp fragment with the 5474 and 5878 primer combination and two additional samples amplified a 460 bp fragment with the 5500 and 5960 combination of primers tested in the second round PCR. Only two samples amplified for two different primer combinations (5500 and 5878; 5474 and 5878). However, for one of the samples (specimen 780), the two primer combinations amplified two distinct variants (Figure 1). In total, 10 animals amplified for larger region of *pol*. Due to a short sequence overlap of our generated sequences with the SFV *pol* sequences available from Genbank, the phylogenetic analysis was limited only to the five complete NP SFV genomes available in the literature. The analysis suggests there are two distinct lineages of SFV co-circulating in the population of *L. chrysomelas* analyzed; a major lineage, herein named SFVlcm-1 (described in red; Figure 1), formed a single clade that branches out of the other SFVs infecting the Cebidae family, and another lineage (SFVlcm-2; described in blue), formed a clade with SFV infecting *Sapajus xanthosternos* and *Callithrix jacchus* (Figure 1). As expected, both strains clustered within the viruses infecting the Cebidae family. When analyzing the PCR amplification efficiency of NP SFV between the two strains found, we observe that among SFVlcm-1 strain only 25% (2/8) amplified by the conventional diagnostic PCR, whereas SFVlcm-2 strain had 100% (2/2) of the strain PCR-amplified (*p* = 0.520).

The pairwise distance analysis showed that the sequences within each strain are similar to each other, with an average divergence of 1% within strain 1 and of 2.6% within strain 2. When comparing SFVlcm-1 to -2, the mean divergence between them was 11%, higher than when compared strain 1 to sequences of other representatives of the Cebidae family, 8.5% and 8.8% for *Sapajus* and *Callithrix*, respectively (Table 3).

#### *3.4. Evaluation of SFV Transmission among Groups of L. chrysomelas*

To investigate the eco-epidemiological profile of the SFV infection among the *L. chrysomelas* groups, a map was plotted using the GPS coordinates obtained during specimens' captures in the forest area of Niteroi city to analyze SFV distribution (Figure 2). The viral distribution among the family groups was widely disseminated in the population. The SFVlcm-1 strain was present in three spatially separated groups and the SFVlcm-2 strain was limited to a single group. All groups belonged to the central forest area (Figure 2).

**Figure 1.** Platyrrhini SFV: phylogeny tree inferred using maximum likelihood analysis with a fragment of viral polymerase (360 bp). New sequences generated in the current study are marked in red (cluster SFVlcm-1) and in blue (cluster SFVlcm-2), all deposited at GenBank under the accession numbers MN178627 to MN178637. Bootstrap support was determined using 1000 nonparametric resampling replicates and values ≥ 70% are provided at nodes.

**Table 3.** Evolutionary divergence estimates between SFV sequences from Cebidae.


The number of nucleotide substitutions per site between sequences is shown. Standard error estimates are shown above the diagonal. Analyses were conducted using the Maximum Composite Likelihood model. The analysis involved 13 nucleotide sequences. Codon positions included were 1st + 2nd + 3rd. All positions containing gaps and missing data were stripped. There were a total of 263 nucleotide positions in the final dataset. Evolutionary analyses were conducted in MEGA7. The colors represent the viral strains: in red SFVlcm -1; blue SFVlcm-2 and gray the complete genomes of SFVsxa and SFVcja.

Interestingly, we observed that of the four solitary animals, three were infected by SFV (Figure 2). Three were males and one was female, with an SFV prevalence of SFV of 67% (2/3) and 100% (1/1), respectively. All the solitary animals were adults. These data suggest that errant males and females can contribute to the spread of SFV infection within this free-ranging primate population.

**Figure 2.** Eco-epidemiology of SFV in the *L. chrysomelas* family groups in the city of Niteroi/RJ. The location of each *L. chrysomelas* family group is represented by circles and of solitary animals by triangle. The red color represents infected animals or groups (when at least one animal is infected in the group), while the blue color represents the uninfected animals measured by conventional diagnostic PCR and/or quantitative PCR. Gray halos around the circles depict the presence of the SFVlcm-1 strain, while the yellow halo represents the SFVlcm-2 strain. The absence of halos indicates lack of amplification of the larger pol fragment, not allowing the classification in SFVlcm-1 or 2.

#### **4. Discussion**

The study of SFV in NP can be specially challenging due to the difficult access to free-living primates and limited volumes of blood that can be collected, since many specimens have a small size [29]. Moreover, of the 176 species of NP that circulate in Brazil, many are threatened to extinction [35]. To detect the SFV provirus, a high mass of genomic DNA (250–500 ng) is necessary from peripheral blood mononuclear cells [1], since blood cells are a recognized site of foamy virus latency [36]. Therefore, the use of buccal swabs is an important tool for SFV detection, since it preserves the animal's health and provides a higher viral load since the oral mucosa is a major SFV replication site [28,31]. Thus, alternative stress-relieving methods, such as buccal swab, are attractive sample sources for the study of SFV, mainly in small primates threatened to extinction.

In the more recently-captured subgroup studied here, an SFV prevalence of 25.4% was observed, similar to the one found in previous studies with free-living primates (14–30%) [2]. However, the SFV prevalence of the subgroup with longer time in captivity was much higher (55.2%), in agreement to the observed in captive Peruvian and Brazilian primates (45–51%) [2,3]. This increase in SFV prevalence

among animals kept in captivity for longer occurred for both sexes and mainly among juveniles. Although this population of *L. chrysomelas* lived in a restricted fragment of Atlantic forest, favoring the contact between groups, the captivity environment clearly contributed to increased transmission of SFV. It is known that an environment that does not promote the welfare and interest of the animal can generate stress, which can be reflected in behavioral changes such as increased aggressiveness [37], but likely also in the susceptibility to infectious agents. Transmission of SFV can happen through blood transfusion [38], maternal milk [39] and mainly by biting and grooming [5]. Thus, stressful environments can collaborate for a higher dissemination of SFV between captive animals.

The area where these animals have occupied is very fragmented (Figure 2), and some areas are very close to urban areas, where many of these animals were seen close to household waste to feed [18]. The proximity between non-human primates and humans can contribute to a risk of SFV zoonotic transmission to the latter [25–27]. Although until now SFV is not known to cause disease in its natural hosts [5], the association of SFV infection with mild anemia was observed in humans [40]. Little yet is known about the transmissibility of SFV from NP to humans and their consequences, but, previous work by our group has shown prevalence rates of SFV zoonotic transmission to primate handlers using serological assays [25]. We are currently working with primate handler samples to deepen our knowledge of NP SFV zoonotic transmission.

When conventional diagnostic PCR was standardized, there were only three complete NP SFV genomes available at Genbank, and the sensitivity of the assay was measured at 100% in seven NP genera studied (*Cebus, Alouatta, Callithrix, Aotus, Ateles, Saimiri, Cacajao* and *Pithecia*) [1]. However, sensitivity drops too much for detecting SFV from other genera such as *Leontopithecus*. Therefore, a quantitative PCR was developed using all available NP SFV *pol* sequences as references for degenerate primer design that amplifies a smaller and more conserved *pol* gene region. This assay was able to detect SFV in two additional species of *Leontopithecus*, and in one species each of *Callimico* and *Saguinus* previously found to be SFV-negative using the conventional diagnostic PCR assay [31]. As demonstrated previously [31], the qPCR was shown to be more sensitive than the conventional diagnostic PCR for detection of NP SFV, When correlating the number of SFV copies with the sensitivity of conventional diagnostic PCR, similar to what has been observed for feline FV [41], no association was found. These results suggest that the false negatives in the conventional PCR may be due to a high genetic heterogeneity of NP SFV sequences at primer locations determined previously to be 41% in the virus *pol* region [1].

SFV DNA VL comprises both the integrated virus (provirus) and the genomic DNA of the virus particle, since SFV can produce both DNA and RNA particles [5]. Little is known about the standards of the DNA VL in the oral mucosa of NP. A recent study [31] found a mean viral load of 4.7 log SFV copies/106 cells among 23 NP specimens of 12 different species in captivity, including four *L. chrysomelas* specimens [31], similar to that has been found in this study. However, when comparing the VL of only four *L. chrysomelas* and one *L. rosalia* quantifiable for SFV of the previous study (range 2.9–7.3 log SFV copies/106 cells), the variation was much higher than the one observed in this study (standard deviation 0.62), which can be explained by differences in sample size. In addition, no association was found regarding the viral load and the sex of the animal, also as observed in the previous study. Finally, also as in the previous study [31], we could not observe any age-related viral load trends in buccal swab samples. These results differ from those reported for rhesus macaques, in which viral load increases with age in the oral cavity of the animals [36]. However, it should be noted that we quantified VL DNA instead of VL RNA, as reported in the rhesus study, and that may explain such lack of correlation observed here. Another important issue is that Liu et al. [24] tested 173 fecal samples from wild chimpanzees (including 87 SFVcpz RNA-positive samples), and none of them detected viral DNA. DNA genome particle production may not reflect replication *in vivo*, and may represent an *in vitro* artifact when using tumor cells with high dNTP levels. Thus, it is unclear whether the viral detection tests (conventional diagnostic PCR and quantitative PCR) in this study are also detecting viral DNA but only proviral DNA.

When comparing the impact of different demographic factors on SFV prevalence, either in the population or in subgroups according to the captivity time, we observed that, as described by others [1–3], the animal sex does not seem to influence SFV acquisition. Yet it has been reported that SFV prevalence increases with age [3,39], no such correlation was observed here. This homogenization in the prevalence between the age groups can be explained, at least in part, by the social behavior of *Leontopithecus*. Group members do social grooming, all members of the family groups help to carry the offspring of the alpha couple and in nature (or captivity), all individuals sleep together, often in the hollows of trees [42], intensifying the contact between them and consequently the chance of SFV transmission. Another interesting ecological characteristic of many NP like *L. chrysomelas,* upon reaching maturity and especially males, is to leave their groups to form a new family group to avoid consanguinity [42]. Interestingly, of the four solitary animals, three were infected, even though they were kept isolated after months at CPRJ since their capture in the wild, showing that these animals may contribute to the dissemination of SFV in the population by entering into existing groups or forming new groups. Our results demonstrate for the first time a new SFV transmission dynamics in primates, on an ecological scale, highlighting the importance of molecular and ecological virology studies in free-living primates.

The use of primers for PCR amplification of larger fragments of the SFV LTR-*gag* region and the *pol* region from previous studies showed a low efficiency to amplify SFV from *L. chrysomelas* [1,3], indicating that SFVlcm can harbor a high nucleotide heterogeneity, at least in the region of primer annealing. To amplify larger DNA fragments, we developed new PCR primers only using sequences of representative SFV genomes infecting primates of Cebidae family to increase specificity. However, the new PCR amplified only 24% of samples diagnosed as SFV-positive, suggesting that there may be more variants circulating in the population, requiring more sensitive techniques, such as shotgun next-generation sequencing, to amplify the complete genomes of these viruses [6].

Phylogenetic analysis showed that, unlike a previous study [3] where a SFV sequence of *L. chrysomelas* clustered with SFV infecting Pitheciidae family members, all SFV sequences from *L. chrysomelas* here in generated grouped into the Cebidae family, which is expected according to the co-speciation hypothesis [4]. However, two distinct lineages of SFVlcm were observed. The most frequent, which formed a separate clade, was named SFVlcm-1, while the other, SFVlcm-2, formed a clade with SFVsxa and SFVcja, infecting *Sapajus xanthosternos* and *Callithrix jacchus*, respectively (Figure 1). The nucleotide divergence between the two strains was 11%, although this refers to a small fragment of the viral *pol* gene, which is conserved among SFVs. Since an earlier study has reported the occurrence of cross-species transmission between *L. chrysomelas*, *Sapajus xanthosthernos* and Pitheciidae species [3], cross-species transmission between species may have occurred in the Atlantic forest fragment where *L. chrysomelas* lives, where there are reports of other PN species, such as *Callithrix jacchus*. Another possibility could indicate that these variants came from cross-species transmission events prior to the arrival of the specimens in Niteroi/RJ, since in the endemic area of Bahia there are also other species of primates, as black-tufted marmoset (*Callithrix penicillata*) and yellow-breasted capuchin monkey (*Sapajus xanthosthernos*). However, as the phylogenetic inference of the two lineages was limited to only five complete NP SFV genomes available in the literature, of only two NP families, and was based on short *pol* sequences, we cannot assess whether the SFVlcm-2 clade was derived from a recombination event between SFVlcm-1 and an SFV from another species, as it has been already described in SFV-Infected Old World monkeys [43]. These alternative scenarios turn the understanding of the complete evolutionary history of the SFV infecting *L. chrysomelas* a difficult task at the moment. This issue will only be clarified with the amplification of larger SFV sequences or complete genomes from *L. chrysomelas* derived from the native population of Bahia and other NP SFV representatives.

In conclusion, we have demonstrated here for the first time an increase in the SFV prevalence of recently-captive *L. chrysomelas*, including the characterization of two novel SFV strains, SFVlcm-1 and -2, by using oral swab as an efficient alternative non-invasive method. We also present new ecological dynamics of SFV transmission from infected solitary animals that dispersed to form new groups or joined existing groups. Further studies are needed to fully characterize the SFV variants in this species, only preliminarily described here, which will improve our understanding of retroviral infections in the Platyrrhini parvorder, covering all primates of the Americas.

**Author Contributions:** Conceptualization, A.F.S., M.A.S., A.P.; methodology, T.S.M., C.P.M., S.B.M., M.G.B.; validation, S.B.M., M.G.B., M.C.K.; formal analysis, T.S.M., C.V.M., M.A.S., A.F.S.; investigation, T.S.M., C.V.M., M.S., A.F.S.; resources, J.L.C.-D., M.A.S., A.F.S.; data curation, T.S.M., C.V.M., S.B.M., M.G.B., M.C.M.K.; writing—original draft preparation, T.S.M., M.A.S., A.F.S.; writing—review and editing, T.S.M., S.B.M., M.G.B., M.C.M.K., A.P., M.A.S., A.F.S.; supervision, M.A.S., A.F.S.; project administration, M.A.S., A.F.S.; funding acquisition, M.C.M.K., J.L.C.-D., A.P., M.A.S., A.F.S.

**Funding:** This study was supported by the Brazilian Research Council (CNPq; grant 312903/2017-0 to A.F.S.) and by the Rio de Janeiro State Science Foundation (FAPERJ; grant E-26/112.647/2012 to MAS and grant E-26/202.738/2018 to A.F.S.).

**Acknowledgments:** We are grateful to Pri-Matas team and CPRJ/INEA collaborators who assisted in providing samples and biogeographical data. We would also like to thank all the institutions and organizations which provided financial support for the Tamarins Translocation Project including the Fundação Grupo O Boticario, the Lion Tamarin of Brazil Fund, the Primate Action Fund, the Margot Marsh Foundation, The Mohamed bin Zayed Species Conservation Fund, Câmara de Compensação Ambiental/ Secretaria de Estado do Ambiente e Sustentabilidade - SEAS (RBO Energia e Porto Sudeste, the Tropical Forest Conservation Act/ Fundo Brasileiro para Biodiversidade (TFCA/FUNBIO)—Rio de Janeiro.

**Conflicts of Interest:** We declare no conflicts of financial or personal interests.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **Infection with Foamy Virus in Wild Ruminants—Evidence for a New Virus Reservoir?**

#### **Magdalena Materniak-Kornas 1,\*, Martin Löchelt 2, Jerzy Rola <sup>3</sup> and Jacek Ku ´zmak <sup>1</sup>**


Received: 18 October 2019; Accepted: 31 December 2019; Published: 3 January 2020

**Abstract:** Foamy viruses (FVs) are widely distributed and infect many animal species including non-human primates, horses, cattle, and cats. Several reports also suggest that other species can be FV hosts. Since most of such studies involved livestock or companion animals, we aimed to test blood samples from wild ruminants for the presence of FV-specific antibodies and, subsequently, genetic material. Out of 269 serum samples tested by ELISA with the bovine foamy virus (BFV) Gag and Bet antigens, 23 sera showed increased reactivity to at least one of them. High reactive sera represented 30% of bison samples and 7.5% of deer specimens. Eleven of the ELISA-positives were also strongly positive in immunoblot analyses. The peripheral blood DNA of seroreactive animals was tested by semi-nested PCR. The specific 275 bp fragment of the *pol* gene was amplified only in one sample collected from a red deer and the analysis of its sequence showed the highest homology for European BFV isolates. Such results may suggest the existence of a new FV reservoir in bison as well as in deer populations. Whether the origin of such infections stems from a new FV or is the result of BFV inter-species transmission remains to be clarified.

**Keywords:** foamy viruses; BFV; wild ruminants; European bison; red deer; roe deer; fallow deer; seroreactivity; inter-species transmission

#### **1. Introduction**

Foamy viruses (FVs), also known as spumaviruses, are the least known subfamily of *Retroviridae*[1]. Some features of their replication pathway and complex genomic organization distinguish them from other retroviruses [2,3]. Infections with FVs are persistent with sustained antibody response against viral antigens and the presence of viral DNA in leukocytes [4]. The most likely routes of FV transmission are via the transfer of blood and saliva and social interactions [3,5–7]. Over the last 60 years, FVs have been isolated and described in different species of non-human primates (Simian FVs (SFVs)) [8], as well as in cattle (Bovine FV (BFV), in the past also called bovine syncytial virus (BSV)) [9,10], cats (Feline FV (FFV)) and horses (Equine FV (EFV)) [3,11]. Several other non-primate FVs have been reported as having been isolated or simply described in sea lions, leopards, sheep, goats, hamsters, and American bison on the basis of cross-antigenicity with known FV, specific cytopathic effects or electron microscopy analyses [10,12–16]. Although FVs can be commonly isolated from infected animals, no disease has been associated with infections and, therefore, FVs are recognized as apathogenic on their own [17,18]. This lack of pathogenicity contrasts strongly with the cytopathic effects seen in vitro in infected cell cultures, with the appearance of "foamy-like" syncytia [17,19]. Based on the detection of diverse SFVs in simian-exposed humans, many studies have been focused on the inter-species

transmission of FVs from simian and non-simian FVs [18,19]. While infections of humans by FVs from different simians and non-human primates are well evidenced, little is presently known about the possibility of such inter-species transmission caused by FVs of live-stock animals. Since BFV is highly prevalent within cattle populations [3,7,20], special attention should be paid to the possible involvement of BFV in inter-species transmission, especially regarding free-ranging wild ruminants. This is a very important and pertinent issue, owing to increasing human impact on the environment, globalization, and the establishment of breeding of some wild ruminants posing new threats including the uncontrolled transmission of infectious agents into wildlife [21,22]. There are many examples of highly prevalent life-stock viral pathogens crossing species barriers into wild ruminants, including bovine respiratory viruses like parainfluenza virus (BPIV-3), bovine adenovirus (BAdV), or bovine respiratory syncytial virus (BRSV) infecting European bison (*Bison bonasus*) in Poland [23]. The most important alphaherpesvirus, bovine herpesvirus 1 (BoHV) have also been reported to infect almost 40% of cervids in Poland [24], and a low percentage of the bison population [25]. Inter-species infections with ruminant retroviruses have been also reported previously: Bovine leukemia virus (BLV) infections have occasionally been described in European bison [25] or alpaca (*Vicugna pacos*) [26], while small ruminant lentiviruses (SRLV) infections have been found in Rocky Mountain goats (*Oreamnos americanus*) [27], Passirian goat in northen Italy [28] and recently in red deer (*Cervus elaphus*) and muflon (*Ovis aries musimon*) in Spain [29]. All reported cases are most likely due to the spill-over from domestic animals, acquired similarly to the well documented case of SRLV infection of endangered wild ibex (*Capra ibex*) in the French Alps, which was probably a result of sharing grazing grounds with a small herd of heavily infected goats [30].

The goal of the current study was the detection of antibodies and genetic material of BFV or a related FV in blood samples collected from free-ranging wild ruminants in Poland in order to address questions related to inter-species transmissions and altered pathogenicity in the new host or as part of a changed virome/microbiome.

#### **2. Materials and Methods**

#### *2.1. Animal Samples*

The samples used in this study came from 269 wild ruminants (suborder: *Ruminantia*, within the order of even-toed ungulates, *Artiodactyla*). Out of those, 256 samples were collected from cervids (family of *Cervidae*) including red deer (*Cervus elaphus*, n = 134), roe deer (*Capreolus capreolus*, n = 103), or fallow deer (*Dama dama*, n = 19), and 13 from free-ranging bovides (family of *Bovidae*), the highly endangered European bison (*Bison bonasus*). The serum samples of European bison and 18 fallow deer had been deposited as archival samples in the Departments of Biochemistry and Virology, NVRI, respectively. Whole blood samples were collected from the main vein or aorta of red deer and roe deer mainly as blood clots during seasonal hunting. All specimens were collected during the 2009/2010 hunting season. Blood clots were squeezed through sterile gauze and centrifuged for 15 min. at 3000 rpm. Obtained supernatants were collected for serological testing and pellets containing blood cells were washed twice with PBS and frozen at −70 ◦C until DNA preparation.

#### *2.2. DNA Preparation*

Total DNA was extracted from pelleted blood cells using the DNeasy Blood & Tissue Kit (Qiagen, Hilden, Germany) following the manufacturer's instructions. DNA concentrations and the 260 nm/280 nm ratio were measured spectrophotometrically using GeneQuant (GE Healthcare, Warsaw, Poland) and stored at −20 ◦C until use. The DNA quality of selected samples was tested using capillary electrophoresis with highly sensitive gel (Fragment Analyser, Agilent).

#### *2.3. Antibody Detection*

GST (glutathione S-transferase) capture ELISAs were performed to examine the antibody response to BFV proteins in sera of wild ruminants using a well-established and validated generic GST-ELISA for domestic cattle as described by Romen and co-workers [7]. In short, 96-well microtiter plates (Thermo Labsystems, Dreieich, Germany) were coated with glutathione casein, blocked with 0.2% (*w*/*v*) casein and 0.05% (*v*/*v*) Tween 20 in PBS (blocking buffer), and then incubated with cleared *E. coli* lysates at a concentration of 0.25 μg/μL (total lysate in blocking buffer) containing the GST-tag or GST-X-tag fusion proteins (X = BFV-Gag, BFV-Bet, or BFV-Env). For pre-absorption of GST-binding antibodies, all sera were incubated at a dilution of 1:100 in a blocking buffer containing 2 μg/μL total lysate of a GST-tag expressing *E. coli* culture prior to application on the coated plates. After pre-absorption serum samples were incubated for 1 h at RT in the coated ELISA plate wells, washed, and incubated for 1 h at RT with Protein G—peroxidase conjugate (Sigma, 1:10,000 dilution). Protein G has a broad binding capacity for ruminant IgG [31]. TMB (Tetramethylbenzidine, Sigma, Poznan, Poland) was added as a substrate. For each serum, the absorbance of the GST-tag was determined and subtracted from the absorbance with the GST-X-tag protein to calculate the specific reactivity against the BFV antigens. Optical density (OD) measurements were done in duplicates and antibody levels were expressed as average net OD. As positive and negative internal controls, the pool of serum samples from five BFV naturally infected cows and five uninfected animals, diagnosed by GST-ELISA and PCR tests [32], were used at 1:100 dilutions.

Due to the lack of positive and negative controls from wild ruminants, cut-off values were calculated from the ELISA results for BFV Gag and Bet antigens obtained for cervids and European bison, excluding 3 outliers of bison origin and 9 outliers of cervid origin. These criteria resulted in 10 samples from European bison and 247 from cervids. The calculation was done in two ways: a less stringent cut-off was calculated as 1 × (mean + 3SD) and a highly stringent one as 2 × (mean + 3SD), which provided two cut-off values for each antigen, both methods are commonly used for diagnostic ELISA tests where defined negative and positive controls from the species tested are not available [33].

#### *2.4. Western Blotting Analysis*

Cf2Th cells (canine fetal thymus cells, Cat. No. 90110521, European Collection of Authenticated Cell Cultures (ECACC), UK) were co-cultured with BFV100-infected Cf2Th cells in a proportion of 10 to 1 and grown in DMEM, supplemented with 10% fetal bovine serum in the 5% CO2 atmosphere. Three days after infection, when the cytopathic effect appeared, cells were lysed using a CHAPS buffer (0.5 M EDTA, 1 M Tris HCL pH 8.8, 100 mM NaCl, 0.5 M CHAPS, 0.5 M sodium deoxycholate; Sigma, Poznan, Poland). Uninfected Cf2Th cells were grown under the same condition. Of the total cell lysates, 10 μg of infected and uninfected control cells were separated by SDS-PAGE and served as the antigen for western blotting analyses (WB) [4]. Wild ruminant sera were used at 1:100 dilutions (*v*/*v* in 3% bovine albumin, 0.01% Tween 20, PBS) and Protein G–peroxidase conjugate (Sigma, Poznan, Poland) at 1:10,000 dilution. As positive and negative controls, the pools of serum samples from five BFV naturally infected cows and five uninfected animals, diagnosed by GST-ELISA and PCR tests [32], were used at 1:100 dilutions. ECL Plus reagents (GE Healthcare, Warsaw, Poland) were used for the detection of specifically bound antibodies.

#### *2.5. FVs DNA Detection, Cloning, and Sequencing*

Semi-nested polymerase chain reaction (PCR) was performed using genomic DNA from blood cells of 16 red and roe deer selected based on high reactivity in ELISA and the availability of high DNA quality, which was confirmed using a capillary system with highly sensitive gel (Fragment Analyser, Agilent). A set of external primers: BFVpolF1: TGGGAAAACCAGGTCGGACATC, BFVpolR: TACGACATCTGCTGTAAACAATGC, FFVpolF1: TGGGGAGAATCAGGTGGGTCATA and FFVpolR: TACAACATCTCCAGTAAACAACCC, EFVpolF1: TGGGAAAATCAAGTGGGACATA, EFVpolR: TACAACATCTGCAGTAAATAAGGC and internal primers: BFVpolF2: ATGGACGCTGGAGGATGGTGTTAGAC, FFVpolF2: ATGGTCGCTGGAGAATGGTACTGGAC and EFVpolF2: ATGGACGATGGAGAATGGTACTGGAT (Genomed, Warsaw, Poland) were designed to 100% match the corresponding part of the *pol* gene encoding for reverse transcriptase of all known non-primates FVs, i.e., BFV (GenBank accession no.: NC\_001831.1 ), FFV (GenBank accession no.: NC\_039242.1), and EFV (GenBank accession no.: AF201902.1). The first amplification included 2 U of DyNazyme DNA polymerase (Thermo Fisher Scientific), 1× PCR buffer with 1.5 mM MgCl2, 0.2 μM of each primer, 0.4 mM of dNTP-mix (Thermo Fisher Scientific) and 1μg of genomic DNA. The temperature profile was as follows: initial denaturation at 94 ◦C for 3 min, denaturation at 94 ◦C for 45 s, annealing at 54 ◦C for 45 s, elongation at 72 ◦C for 1 min, and final elongation at 72 ◦C for 5 min. Semi-nested amplification was completed in similar conditions using 1/10 volume of the first PCR as a template. The expected size of the amplicon was 275 bp.

Another semi-nested PCR was performed using DNA from the blood of one selected red deer. The following primers (Genomed, Warsaw, Poland) were used to amplify the sequence located in the BFV *gag* gene (1139-1381 nt): Gag-1: GACGCAACAAACCAACCAC; Gag-2: GTTCTTGTCCGTATCGTTGTG [34] and BFVpolR: TACGACATCTGCTGTAAACAATGC. The first amplification was performed in the same conditions as described for the BFV *pol* reaction, but included Gag-1 and BFVpolR primers, while the semi-nested PCR was performed as described previously with Gag-1 and Gag-2 primers [4], but using 1/10 volume of first PCR as a template. The expected size of the semi-nested PCR product was 243 bp.

Nested PCR for BFV LTR-derived sequences was performed using genomic DNA from blood cells of selected deer (see earlier in this section). The following primers (MWG Biotech) within the BFV LTR region, mostly covering the U3 region and the beginning of the R region, were used in this study: BFV-LTR-1\_S: TTACTTGCCCGGAGGATTGG, BFV-LTR-1\_AS: TAGTGATCTGGAAGGTAAGC, BFV-LTR-2\_S: CTTATGGATGGAGCCTTATGG, BFV-LTR-2\_AS: CTTACCACAGCCTGGAAGTC. All primers were designed to 100% match all known BFV sequences (GenBank accession no.: U94514, GI 9629644, GI 22947830). The first reaction of the amplification included 2.5 U of *Taq* DNA polymerase (ThermoFisher Scientific), 1× PCR buffer with 1.5 mM MgCl2, 0.2 μM of each primer, 0.1 mM of dNTP-mix (ThermoFisher Scientific), and 1μg of genomic DNA. The first PCRs were performed with the following temperature profile: initial denaturation at 94 ◦C for 3 min, denaturation at 94 ◦C for 45 s, annealing at 52 ◦C for 45 s, elongation at 72 ◦C for 2 min, and final elongation at 72 ◦C for 10 min. Nested amplification was done at similar conditions, with the exception of the annealing step where the temperature was 54 ◦C for 45 s and using 1/10 volume of the first PCR as a template. The expected size of the amplicon was 874 bp.

The resulting amplicons were analyzed on 1% agarose gels, cloned into the pCR2.1-TOPO vector (Invitrogen), and sequenced from both sides according to the method of Sanger [35] by Genomed, Warsaw, Poland for the *pol* sequence and GATC (Konstanz, Germany) for the LTR sequences.

For bioinformatics analyses, primer sequences were removed from the cloned amplicons, which were aligned to the reference sequences of non-primate FVs available in GenBank (EFV—LC\_381046.1, AF201902.1, NC\_002201.1, BFV-US—NC\_001831.1, BFV-3026—AY134750.1, BFV-100—JX307861.1, and BFV Riems—JX307862.1, as well as all available isolates of FFV—accession numbers are listed on the phylogenetic tree in Figure 5) using the Geneious alignment module within the Geneious Pro 5.3 software (Biomatters Ltd., Auckland, New Zealand). The alignment was submitted to the MEGA 6.0 version for the best model selection measured by the the Bayesian information criterion (BIC) and the corrected Akaike information criterion (AICc). According to the results Tamura 3-parameter with Gamma distribution [36] substitution model was applied in MEGA 6.0 to infer a phylogenetic tree using maximun likelihood method. The statistical confidence limits of the phylogram topologies were assessed with 1000 bootstrap replicates. The sequences obtained in this study were deposited

in the GenBank database with the following accession numbers: MN630606-MN630611 for *pol* and MN630602–MN630605 for LTR sequences.

#### *2.6. Statistical Analysis*

Scatter plot analyses were performed to calculate the linear correlations of the net OD values obtained for Gag and Bet antigens in ELISA tests. Calculations and the generation of graphs were done using STATISTICA ver. 10 (StatSoft, part of Dell Software, USA).

#### **3. Results**

#### *3.1. Serological Screening of Wild Ruminants Samples*

The study included 269 serum samples from wild ruminants collected in different parts of Poland (Figure 1). The samples originated from cervids such as red deer (*Cervus elaphus*, n = 134), roe deer (*Capreolus capreolus*, n = 103), and fallow deer (*Dama dama*, n = 19), as well as free-ranging bovides, the highly endangered European bison (*Bison bonasus*, n = 13). Serum samples were assayed for the reactivity toward BFV Gag, Bet, and Env-SU antigens using GST-capture ELISA as previously described [7]. The observed net OD values for the Gag antigen ranged between 0.001–1.246 and between 0.001–1.339 for Bet in cervid samples. The overall reactivity of bison samples was lower and ranged between 0.056–0.660 for Gag and 0.087–0.590 for Bet. The reactivity to the Env-SU antigen was very low, comparable with the background, and was therefore not further considered as the GST-tagged Env has also been shown in previous studies to be of low diagnostic value [7].

**Figure 1.** Geographic distribution of samples collected from wild deer in Poland. Blue dots show the areas of the samples collection which correspond to closest cities, n—the number of samples collected in the particular areas. F—no. of fallow deer, J—no. of red deer, S—no. of roe deer; green rings—the major regions of dairy cow production in big farms; orange ring—regions with moderate dairy cow production, mainly in small family farms.

Due to the differences of overall reactivity, scatter plot analyses of the net OD values for Gag and Bet antigens were separately performed for European bison (Figure 2a) and the different deer species (Figure 2b), which showed that the correlation between the seroreactivity to both antigens was strong in the population of bison, while in deer, it was weaker.

Cut-off values were calculated for Gag and Bet separately, using two methods commonly used in serodiagnostics for the different tests and animal populations (see Materials and Methods, grey and red dashed lines) (Figure 2). This approach distinguished between very high reactive sera, as well as those assessed as inconclusive with reactivity in the grey intermediate zone between the two independently calculated cut-off values (Figure 2).

**Figure 2.** Distribution of Gag and Bet serological reactivity presented as scatter plots: (**a**) serum samples of bison origin, (**b**) serum samples of cervid origin. Each point represents a data pair of an individual serum of the following origins: blue dot—European bison, red dot—fallow deer, blue square—red deer, and yellow triangle—roe deer. Dashed lines, grey and red, indicate the cut-off values calculated in two ways grey—cut-off = 1 × (mean + 3SD), red—cut-off = 2 × (mean + 3SD). The grey boxes between the dashed lines indicate the grey zones of inconclusive ELISA results. The upper right sector shows double-positive, the lower-left double-negative, the lower-right sector displays sera positive for Bet only, and the upper left sector represents sera positive for Gag exclusively. The correlation coefficient (r) and *p*-value (p) are indicated in the graphs.

This analysis indicated that seven samples showed clearly high reactivity in comparison to other samples only in deer populations, while the reactivity of another 16 samples was in the intermediate grey zones. Out of these 23 samples with high or intermediate reactivity, four were from European bison, representing 30% of bison samples tested. The remaining 19 samples represent 7.4% of deer specimens tested, including six from roe deer and 13 from red deer origin. Interestingly, all highly reactive deer samples came from six locations, with seven samples (7/256, 2.7%) from Gorzów Wielkopolski in the western part of Poland, and six samples (6/256, 2.3%) from Nidzica in the north of the country (Figure 1, Table 1). Three samples from bison and three from red deer showed high reactivity to both, Gag and Bet antigens. One bison and six deer samples reacted strongly with Gag only (4 from red deer and 2 from roe deer), while ten samples were reactive to Bet antigen (6 from red deer and 4 from roe deer).


**Table 1.** Summary of Gag and Bet ELISA, western blotting (WB), and PCR results.

n.t.—not tested, n.a.—not available. Areas of sample collection are indicated according to Figure 1.

To verify the ELISA results, 15 samples with reactivity to at least one BFV antigen were subsequently tested by western blotting assay (WB) with lysates of BFV100-infected Cf2Th cells, used as an antigen. Uninfected Cf2Th cells were used as a control antigen. Specific reactivity against Gag results in the presence of a double band at about 60/58 kDa and, for Bet, a single band at approximately 46 kDa is characteristic (own unpublished study). Out of the 15 samples tested, eleven samples showed strong reactivity to the BFV cellular antigen (Figure 3) including three ELISA-positive samples from bison, while three samples from deer showed only very faint bands in WB, marked by +/-, in Table 1. The pattern of bison sera WB reactivity was very similar to the positive control, which contained pooled sera of naturally BFV-infected cows. In contrast, the pattern of deer reactivity was slightly

different, especially for Bet reactivity, which was, in most cases, clearly weaker and even completely missing in animals 125/9 and 125/10. One sample, no. 98/4 showed no reactivity in WB.

**Figure 3.** Detection of foamy virus (FV)-specific antibodies by immunoblotting analysis with a cellular antigen in representative serum samples of deer and European bison; (−) lane with uninfected Cf2Th cells lysate as antigen, (+) lane with Cf2Th/BFV100 cells lysate as antigen; P—BFV positive control serum, N—BFV negative serum.

#### *3.2. Detection of FV DNA*

We then aimed to detect FV-specific DNA in the total DNA extracted from the blood cells of ruminants with high reactivity towards BFV antigens, as determined by serology. A set of specific PCR primers was designed to match a conserved region of the *pol* gene encoding the BFV, EFV, and FFV reverse transcriptase. DNA extracted from blood cells of animals with serum reactivity towards BFV antigens was used as the template in semi-nested PCRs. Specific amplification was obtained for only one sample collected from red deer no. 113/13 (Figure 4). The 275 bp PCR product was cloned and sequenced. Alignment of sequences of six clones with all sequences of BFV, EFV, and FFV isolates available in GenBank showed the highest homology with the Polish BFV100 isolate and German BFV Riems, both representing the European clade of BFV [37] (97% identity for clone 0 and 100% for the other four clones) (Figure 5). Such a high similarity is comparable to the homology of FFV sequences of domestic cat and mountain lion origin.

**Figure 4.** Electrophoretic analysis of semi-nested PCR amplification products. M1, Gene Ruler 1 kb Plus DNA Ladder; M2, GeneRuler Low Range DNA Ladder, Fermentas, **P**, positive control (blood DNA of calf experimentally inoculated with BFV), N, negative reaction control, samples no. 113/11–113-16.

**Figure 5.** Phylogenetic tree inferred from the 275 bp amplicons of the *pol* region from deer 113/13 and extracted from the sequences of all BFV, EFV, and FFV isolates (the majority of the sequences) available in GenBank. The origin of FV isolates has been indicated as follows: BFVbta, cow (*Bos taurus*); EFVeca, horse (*Equus caballus*); FFVfca, cat (*Felis catus*); FFVpco, mountain lion (*Puma concolor*). All analyzed sequences were devoid of primer sequences. The tree was generated with MEGA 6.0 software by maximum likelihood method and presented with condensed branches.

Furthermore, a set of pan-BFV specific primers was used to amplify part of the LTR sequences from red deer no. 113/13. The LTR PCR amplicons were cloned, sequenced, and aligned to the respective sequences of BFV isolates available in GenBank (Figure 6). Unfortunately, no successful amplification was achieved when primers specific for BFV *gag* were used.

Both phylogenetic analyses (Figures 5 and 7) clearly show that the new sequences from European red deer are most closely related to the European clade of known BFV isolates [37]. In addition, LTR clones showed the highest similarity to the Polish BFV100 isolate (99.2% to 99.8% identity for analyzed clones) (Figure 6). The overall similarity for both sequences was very high and most changes were single transitions or transversion. Interestingly a deletion of two nucleotides at the LTR position 803/804 nt, the beginning of the R region of the LTR, was present in all clones from red deer, which is not present in any of the European clade BFV isolates.

**Figure 6.** Alignment of the LTR clones sequences derived from red deer no. 113/13 with respective sequences extracted from BFV isolates available in GenBank.

**Figure 7.** Phylogenetic tree inferred from the 874 bp sequence of the LTR region amplified from deer no. 113/13 and extracted from sequences of all BFV isolates available in GenBank. All analyzed sequences were devoid of primer sequences. The tree was generated with MEGA 6.0 software by maximum likelihood method. Bar—nucleotide substitutions per site.

#### **4. Discussion**

Most known FVs are highly prevalent in their hosts. For instance, the prevalence of BFV among cattle ranges between 7% and 50% of cattle worldwide, while in Poland it reaches over 30% (see for summary [10]). Transmission of BFV is suggested to occur through saliva, and, therefore, there is likely to be a risk of its spread through shared grazing areas or direct social or environmental contacts [5,6]. Serological investigation of sera from wild ruminants, presented in this study using BFV Gag and Bet antigens, clearly showed that about 8.5% of the sera reacted with at least one BFV antigen in a generic ELISA system. None of the tested samples reacted with the BFV Env antigen, which has already been shown to be an antigen with low diagnostic value in BFV-infected cattle [7,38]. Interestingly, about one fourth (3/13) of European bison sera analyzed specifically reacted with both BFV Gag and Bet in two independent serological assays. In contrast, only 9.7% of red deer and 5.8% of roe deer serum samples tested with the Gag and Bet ELISA and the WB based on the BFV-infected cells showed reactivity and it was often either against Bet or Gag but rarely (1.1%) against both antigens together. None of the fallow deer serum samples reacted with Gag or Bet antigens, which may be because of the low number of samples tested in comparison to red or roe deer. Highly BFV-related, conserved non-primate FV *pol* and LTR sequences were detected by diagnostic PCR in blood cell samples from one red deer, corresponding analyses could not be performed for bison due to the lack of DNA samples.

If the strongly BFV-related sequences isolated from the BFV-reactive red deer 113/13 are also representative for the other deer scored seropositive, this would indicate that these animals have probably been infected with BFV derived from domestic cattle. Since contact between deer and grazing cattle is possible under the farming conditions in Poland, we assume that inter-species transmission from cattle to red and roe deer and possibly other deer species occurs at a low but consistent level similar to the transmission of SFVs to humans highly exposed to non-human primates [18,39–41]. In simian-to-human transmissions, aggressive behavior, biting, and blood exchange are the major route of infection, this is different in ruminants. In BFV, free and mixed grazing of cattle and deer in rural areas may lead to BFV transmission as has been well documented for small ruminant lentiviruses in France by Erhouma and others [30], where close contact between wild ibex and goats at pastures led to small ruminant lentiviruses infections in wild animals, or for BoHV-1 which was detected in cervid populations in Germany and in Poland [24,42]. BFV transmission through saliva has been described as the most frequent route of transmission in natural conditions [3,10]; however, our own studies showed that the isolation of BFV from saliva is possible only from a small percentage of animals [32]. Unpublished data also showed that, in contrast to SFV or FFV, BFV viral load in the saliva of naturally and experimentally infected cows is very variable, from quite high to undetectable in some BFV-positive animals, but the explanation for this phenomenon is still unclear. However, a similar scenario in the wild may limit the possibility of BFV transmission to wildlife animals.

The low correlation of Bet and Gag reactivity in both deer species can also be taken as an indication for individual inter-species transmission events. These interspecies transmissions may result in attenuated or abortive replication, which could lead to incomplete or variable immune recognition of BFV antigens and thus a deviant pattern of immune responses compared to BFV in productively infected cattle [43]. In addition, the abrogation of the immune response after the transgression of the species barrier and adaptation to the new host has been reported for some viruses [43,44] and may explain the immune response pattern seen here in deer. Alternatively, wild-ranging red, roe, and fallow deer may be also exposed to currently unknown BFV-like viruses from other ruminants leading to unexpected and variant patterns of immune reactivity. This point may also explain the high number of PCR negative deer samples in contrast to the results of serological tests. The genetic differences of BFV-like viruses from different species could affect primer binding, leading to a lack of amplification, but, on the other hand, a very low number of BFV copies (below the sensitivity of PCR reaction reaching 10 copies) in peripheral blood of infected animals or the presence of any PCR inhibitors could also lead to negative PCR results. However, such discordance between the presence of BFV-specific antibodies and viral DNA may also result from the natural clearance of the productive BFV infection in heterologous hosts as was observed by Morin and co-workers [44] in experimental infection of calves with caprine arthritis-encephalitis virus.

The high prevalence (3 out of 13 animals tested) and overall tight correlation of reactivity against BFV Gag and Bet in European bison (see ELISA and WB data, Figures 2 and 3) may indicate that a currently unknown BFV-like virus is endemic in bison. This may be similar to the isolation of a retrovirus similar to the bovine foamy virus from American bison, reported years ago by Amborski and co-workers [12]. While such a bison-specific FV is serologically related but distinct (comparably low signal intensities in the Gag and Bet ELISAs, [7,38]), it may also be genetically distinct from the BFV of cattle. In contrast to deer, the habitats of bison and domestic cattle do not overlap frequently since bison are kept in nature reserves, clearly reducing the chance of exposure to BFV from cattle.

European bison are an endangered large European animal that was close to extinction at the beginning of the 20th century. Tight protection programs, especially in Poland and other countries [45], have led to the recovery of this iconic species at the price of severe inbreeding. Loss of genetic diversity within a species increases the susceptibility to infection with many pathogens, including bacteria, viruses, and parasites [23,46,47]. Under such conditions, even innocuous infectious agents like BFV may gain pathogenic potential, either on their own or as part of the microbiome/virome. This indicates a strong rationale for further investigations into old and emerging ruminant FVs. In addition, since only low numbers of bison have been studied here, further research that includes higher numbers of samples is warranted. These new studies should also go beyond serology and PCR-mediated DNA detection and should include the analysis of oral samples and milk for virus detection and isolation and maybe even encompass novel DNA detection methods based on deep sequencing.

In summary, we demonstrate clear evidence that free-ranging ruminants are exposed to and infected by FVs closely related to BFV. However, more studies are required to know whether the highly endangered, inbred, and iconic bison and the different deer species are infected by species-specific, novel FVs or by BFV of cattle origin crossing species barriers.

**Author Contributions:** M.M.-K. was involved in the study design, performed research and data analysis, and wrote the manuscript; M.L. was involved in the study design and wrote the manuscript; J.K. designed and supervised the study and wrote the manuscript; J.R. was involved in samples collection and preparation. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Acknowledgments:** The authors would like to thank Martha Krumbach (DKFZ, Heidelberg) for critically reading the manuscript and the veterinarians and hunters for their assistance in the sample collection.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **Isolation of an Equine Foamy Virus and Sero-Epidemiology of the Viral Infection in Horses in Japan**

#### **Rikio Kirisawa 1,\*, Yuko Toishi 2, Hiromitsu Hashimoto <sup>3</sup> and Nobuo Tsunoda <sup>2</sup>**


Received: 10 June 2019; Accepted: 3 July 2019; Published: 5 July 2019

**Abstract:** An equine foamy virus (EFV) was isolated for the first time in Japan from peripheral blood mononuclear cells of a broodmare that showed wobbler syndrome after surgery for intestinal volvulus and the isolate was designated as EFVeca\_LM. Complete nucleotide sequences of EFVeca\_LM were determined. Nucleotide sequence analysis of the long terminal repeat (LTR) region, *gag, pol, env, tas,* and *bel2* genes revealed that EFVeca\_LM and the EFV reference strain had 97.2% to 99.1% identities. For a sero-epidemiological survey, indirect immunofluorescent antibody tests were carried out using EFVeca\_LM-infected cells as an antigen against 166 sera of horses in five farms collected in 2001 to 2002 and 293 sera of horses in eight farms collected in 2014 to 2016 in Hokkaido, Japan. All of the farms had EFV antibody-positive horses, and average positive rates were 24.6% in sera obtained in 2001 to 2002 and 25.6% in sera obtained in 2014 to 2016 from broodmare farms. The positive rate in a stallion farm (Farm A) in 2002 was 10.7%, and the positive rates in two stallion farms, Farms A and B, in 2015 were 40.9% and 13.3%, respectively. The results suggested that EFV infection is maintained widely in horses in Japan.

**Keywords:** equine foamy virus; isolation; Japan; sero-epidemiology; spumaretrovirus

#### **1. Introduction**

Foamy viruses (FVs) belong to the subfamily *Spumaretrovirinae* within the family *Retroviridae* [1]. FVs have been isolated from a wide range of mammals, including nonhuman primates [2–5], cats [6], cows [7], horses [8] and bats [9], and it has been shown that they establish lifelong infection [10,11]. FV infections have not been shown to be associated with any defined disease [1,12]. The non-pathogenicity of FVs is an essential factor for the development of a foamy viral vector in gene therapy [13]. The FV genome consists of genes encoding canonical retroviral Gag, Pol and Env proteins and a regulatory protein Tas and an accessory protein Bet [1].

The prevalence of simian FVs has been studied in detail, but there have been few studies on the prevalence of other animal FVs [14]. The prevalence of feline FV (FFV) in domestic cats and wild cats was reported to range from about 30% to 100% depending on sex, age, and the geographical region [14–22]. The prevalence of bovine FV (BFV) infection in cattle was reported to range from 7% to 45% [23–26]. The prevalence of equine FV (EFV) in horses has not been reported.

In 2000, equine foamy virus was isolated for the first time from blood samples of naturally infected healthy horses after co-cultivation of phytohemagglutinin (PHA)-activated lymphocytes derived from sero-positive horses with permissive human U373-MG cells and hamster BHK21 cells [8]. Nucleotide sequence analysis revealed that EFV is phylogenetically close to non-primate FVs, especially BFV. There has been no further isolation of EFV since the first isolation in 2000.

In this report, the first isolation of EFV in Japan (the second isolation of EFV in the world) in primary horse kidney cells co-cultured with fresh peripheral blood mononuclear cells (PBMC) from a broodmare showing wobbler syndrome after surgery for intestinal volvulus and the molecular characterization of the isolated virus are described. The results of a serological survey using the Japanese EFV isolate in thoroughbred horses in Japan are also described.

#### **2. Materials and Methods**

#### *2.1. Cell Cultures and Virus Isolation*

Primary horse fetal kidney (HFK) cells were prepared according to the standard method from a fetal kidney that was obtained from a euthanized pregnant mare due to the judgment of a poor prognosis for a forelimb fracture, and the cells were cultured in MEM supplemented with 10% fetal calf serum (FCS) as the growth medium at 37 ◦C. A blood sample from a horse (Horse A) that exhibited wobbler syndrome the day after a surgical operation for intestinal volvulus in an equine hospital, not in our medical center of Rakuno Gakuen University, as veterinary medicine was collected in heparin-containing tubes on October 1, 2001, and peripheral blood mononuclear cells (PBMC) were isolated by Ficoll-Paque gradients (density of 1.077 g/mL). The PBMC were co-cultured with HFK cells in culture dishes (35 mm in diameter) in the growth medium for virus isolation at 37 ◦C under a 5% CO2 atmosphere. The culture medium was removed the next day, fresh MEM supplemented with 4% FCS as a maintenance medium was added, and the cells were cultured at 37 ◦C. The cultured cells were observed daily and the maintenance medium was changed at 4-day intervals until the appearance of a cytopathic effect (CPE). A CPE was observed 10 days after the start of cultivation and the HFK cells showing a CPE were detached by trypsin-EDTA solution and harvested as virus-infected single cells 4 days after the appearance of a CPE. The infected cells were stored at −80 ◦C using CELLBANKER I (Takara Bio Inc., Kusatsu, Shiga, Japan) as the cryopreservation medium. Serum of Horse A was collected on October 1, 2001. Horse A was euthanized due to the judgment of a poor prognosis about 1 week after the operation. We also used conserved sera of Horse A that had been stocked monthly in our laboratory from January 2000.

#### *2.2. DNA Extraction*

Total DNA was extracted from HFK cells showing a CPE (about 80% of the cells) as described previously [27].

#### *2.3. Polymerase Chain Reaction (PCR) for Detection of the Equine Foamy Virus Genome*

To detect EFV DNA in cells showing a CPE, PCR assays targeting LTR were carried out using the Expand High Fidelity PCR system (Roche Diagnostic GmbH, Mannheim, Germany) and primers listed in Table 1, which were designed on the basis of the complete EFV sequence (GenBank AF201902) by using DNASIS Pro (Hitachi Software Engineering Co., Ltd., Tokyo, Japan). PCR amplification was carried out as described previously [28] under the following conditions: an initial denaturation step of 94 ◦C for 5 min, 35 cycles of 94 ◦C for 30 s, 55 ◦C for 30 s, 72 ◦C for 1 min and 30 s, and a final extension step of 72 ◦C for 5 min. The PCR products were purified and sequenced as described below.


**Table 1.** Primers used in PCR amplification.


**Table 1.** *Cont.*

<sup>1</sup> Location at the complete nucleotide sequence of EFV (Genbank AF201902).

#### *2.4. Restriction Enzyme Digestion and Southern Blot Hybridization*

The DNA extracted from cells showing a CPE was digested to completion with restriction endonuclease *Bam*HI under conditions recommended by the manufacturer (Takara Bio Inc., Tokyo, Japan). The digested fragments were separated by electrophoresis in 0.7% agarose gels in Tris-acetate-EDTA buffer (40 mM Tris-acetate, 1 mM EDTA, pH 8.0) and were transferred to nitrocellulose filters (0.45 μm, Schleicher & Schuell, Dassel, Germany) according to the method by Southern [29]. The filters were pre-hybridized for 2 h and then hybridized for 14 h with a probe of the isolated viral LTR labeled with the non-isotopic reagent digoxigenin-dUTP [27]. An enzyme immunoassay kit (Roche Diagnostics, Basel, Switzerland) was used for detecting hybridized fragments. For a DNA molecular weight marker, *Hin*dIII-digested lambda DNA labeled with digoxigenin-dUTP (Roche Diagnostics) was used.

#### *2.5. Sequence and Phylogenetic Analyses*

Regions coding for Gag, Pol, Env, Tas and Bet were amplified by PCR with the Expand High Fidelity PCR system (Roche Diagnostics) and each of the specific primers listed in Tables 1 and 2. The PCR products were purified by Chroma spin columns (Clontech Laboratories, Inc., Mountain View, CA, USA) or a High Pure PCR Product Purification kit (Roche Diagnostic) and used for sequencing. Sequencing was conducted in Hokkaido System Science Co. Ltd. (Sapporo, Japan) using specific primers and walking primers. Sequence analyses were conducted by DNASIS Pro (Hitachi Software Engineering Co., Ltd., Tokyo, Japan). Phylogenetic analysis of the nucleotide sequences was conducted by using MEGA7 software with 1000 bootstrap replicates of the neighbor-joining method [30]. Evolutionary distances were estimated according to the Kimura 2-parameter method [31]. The DDBJ accession number assigned to the complete sequence of the analyzed isolate is LC381046.


**Table 2.** PCR-amplified regions.

**Table 2.** *Cont.*


<sup>1</sup> Region at the complete nucleotide sequence of EFV (Genbank AF201902).

#### *2.6. Serum Samples*

Sera obtained from horses in 10 farms (Farms A to J) in Hokkaido in Japan were used for a sero-epidemiological survey. Farms A and B were stud farms, and the others were breeding farms. In 2001 to 2002, sera were collected from 28 stallions in Farm A on June 7, 2002; 72 mares in Farm C on June 13, 2002; 25 mares in Farm D on April 17, 2001; and 29 mares in Farm E and 12 mares in Farm F on June 30, 2002. In 2014 to 2016, sera were collected from 44 stallions in Farm A on June 30, 2015; 15 stallions in Farm B on May 15, 2015; 107 mares in Farm C on June 30, 2015; 25 mares in Farm D on January 26, 2015; 30 mares in Farm G on March 11, 2015; 39 mares in Farm H on January 17, 2014; 22 mares in Farm I on January 28, 2016; and 11 mares in Farm J on October 31, 2016. For the broodmare from which EFV was isolated (Horse A, Farm C), serum collected on October 1, 2001 (date of onset of wobbler syndrome) and sera collected on February 14, 2001; October 11, 2000; and January 26, 2000 were used. All of the serum samples were initially sent to our laboratory to test for equine herpesvirus 1 (EHV-1) infection. The serum separation procedure was as follows. A blood sample from each horse was collected in a plain tube and was allowed to clot by leaving it at room temperature. The clot was removed by centrifugation at 1000× g for 10 min and the resulting supernatant, designated as a serum, was transferred to a clean polypropylene micro tube. After inactivation of the complement at 56 ◦C for 30 min, the serum was used for a serological test of EHV-1 infection. After the test, the serum was stored at −20 ◦C.

#### *2.7. Indirect Immunofluorescence Assay (IFA) for Detection of Antibodies to EFV*

EFV-infected HFK cells were detached by trypsin-EDTA solution and washed three times with phosphate-buffered saline (PBS, pH 7.4) by centrifugation at 200× *g* for 5 min. The cells were re-suspended in a small volume in PBS and smeared on a 15-well multitest slide glass (MP Biomedicals, LLC, Solon, OH, USA) and then fixed in 100% acetone on ice for 30 min. Equine sera were diluted serially from 1:20 to 320 and incubated with the fixed cells at 37 ◦C for 30 min in an incubation chamber. The cells were washed three times with PBS. After drying the cells at room temperature, the cells were incubated with 1:80 diluted fluorescein isothiocyanate conjugated goat-anti horse IgG (Jackson Immuno Research Inc., West Grove, PA, USA) at 37 ◦C for 30 min. After a final wash, infected cells were visualized under a fluorescence microscope (Olympus, Tokyo, Japan). An IFA titer of 20 or greater was regarded as positive. Uninfected HFK cells were used as control cells.

#### **3. Results**

#### *3.1. Virus Isolation and Identification*

Initially, we suspected a neurological type of equine herpesvirus 1 infection in a horse showing clinical symptoms of wobbler syndrome. However, co-cultivation of PBMC from the affected horse and HFK cells did not show any herpes viral-like CPE after incubation for 1 week. At 10 days after cultivation, small syncytia were observed and they gradually increased in size, and many vacuoles were observed in the syncytia. These morphological changes resembled the foamy CPE in response to FV infection [1]. The isolate was designated as EFVeca\_LM. EFVeca\_LM was highly cell-associated in HFK cells, and cell-free virus was not released in culture supernatants. Cell-free virus was also not obtained after three cycles of freeze-and-thawing of EFVeca\_LM-infected HFK cells. By PCR amplification using primers for the LTR region of EFV, identical products with the same estimated size, approximately 1450 bp, were observed in agarose gel electrophoresis. The nucleotide sequence identity of the amplified products without primer sequences was 98.2% (1381/1407) against that of EFV (Figure 1). In Southern blot analysis, the LTR probe detected a 12-kbp fragment in DNAs from virus-infected cells without restriction enzyme digestion and 7.1-kbp and 1.4-kbp fragments in *Bam*HI-digested DNAs (Figure 2A). By analogy to the EFV genome, the 12-kbp fragment represents unintegrated linear viral genomic DNA and the 7.1-kbp and 1.4-kbp fragments correspond to about 60% of the region of the genome from the 5' end and the LTR region located at the 3' end, respectively (Figure 2B,C). We could not detect the integrated viral DNA in host chromosomal DNA in Figure 2A. Possible reasons were that copy numbers of the integrated DNA were extremely low compared to those of unintegrated viral DNA and that sensitivity of our Southern blot analysis was insufficient for detecting the integrated viral DNA. The same results were obtained by Tobaly-Tapiero et al. [8].


**Figure 1.** Comparison of the nucleotide sequences of LTR regions in the Japanese isolate EFVeca\_LM and reference EFV. Identical nucleotides are indicated by dots. Numbers on the left and right sides are the nucleotide positions of EFV complete genome sequence (AF201902).

**Figure 2.** (**A**) Southern blot analysis of total DNA from HFK cells infected with the isolate. The LTR region was used as a probe. Lane 1: molecular weight marker, lambda DNA *Hin*dIII digest labeled with digoxigenin-dUTP, Lane 2: uncut DNA, Lane 3: *Bam*HI-digested DNA. (**B**) EFV genomic structure. (**C**) *Bam*HI restriction map of the EFV genome. Fragments of thick lines were detected by the LTR probe in lane 3 in (**A**).

Altogether, the isolated virus, EFVeca\_LM, was identified as equine foamy virus belonging to the *Spumaretrovirinae* subfamily.

EFV IFA titers of Horse A are shown in Table 3. EFV antibody already existed in the serum collected at the time of onset of wobbler syndrome. In three conserved sera, EFV antibodies were also detected and showed almost the same IFA titers as that in serum collected at the time of onset of wobbler syndrome. The titers of all of the tested sera against uninfected control HFK cells were less than 20.


**Table 3.** EFV antibody titers determined by IFA tests in Horse A.

<sup>1</sup> Date of onset of wobbler syndrome.

#### *3.2. Sequence Analysis*

The provirus DNA of EFVeca\_LM was completely sequenced and found to be 12,034 bp. The complete provirus genomic sequence of EFVeca\_LM was submitted to DDBJ under the accession number LC381046. The complete genomic sequences of EFVeca\_LM were compared to those of EFV (GenBank AF201902) (Table 4). The LTR nucleotide sequences of EFVeca\_LM and EFV showed 98.2% identity. The nucleotide sequence and amino acid sequence of the *gag* gene showed 98.6% and 99.1% identities, respectively. The nucleotide sequence and amino acid sequence of the *pol* gene showed 98.6% and 99.1% identities, respectively. The nucleotide sequence and amino acid sequence of the

*env* gene showed 98.3% and 98.5% identities, respectively. The nucleotide sequence and amino acid sequence of the *tas* gene showed 99.1% and 100.0% identities, respectively. The nucleotide sequence and amino acid sequence of the *bel2* gene showed 97.2% and 97.3% identities, respectively.


**Table 4.** Identities of nucleotide sequences and amino acid sequences of an isolated virus and EFV.

<sup>1</sup> Number in parenthesis is the size of the nucleotide sequence; <sup>2</sup> Number in parenthesis is the size of the amino acid sequence.

Phylogenetic comparisons of the full-length provirus genome of EFVeca\_LM and those of various animal foamy virus isolates revealed that EFVeca\_LM belonged to the same clade as EFV (Figure 3).

**Figure 3.** Phylogenetic analysis of the Japanese isolate EFVeca\_LM (LM LC301046) and other foamy virus strains. The phylogenetic tree was generated using complete nucleotide genome sequences. EFVeca\_LM is indicated by a closed circle. Bootstrap values less than 50% are not shown on the corresponding nodes. BFV: bovine foamy virus, FFV: feline foamy virus, SFVocr: simian foamy virus Otolemur crassicaudatus, SFVpsc: simian foamy virus Pan troglodytes schweinfurthii.

#### *3.3. Sero-Epizootiology*

To examine the prevalence of EFV antibodies in Japanese horses, we conducted IFA tests using EFVeca\_LM and a total of 166 sera obtained from one stallion farm (Farm A) and four broodmare farms (Farms C to F) in 2001 to 2002 and a total of 293 sera obtained from two stallion farms (Farms A and B) and six broodmare farms (Farms C, D, G to J) in 2014 to 2016 (Table 5). The titers of all of the tested sera against uninfected control HFK cells were less than 20. All of the farms had EFV antibody-positive horses. The positive rates in sera obtained from broodmare farms in 2001 to 2002 ranged from 20.7% to 28.0% (average: 24.6%), and the positive rates in sera obtained from broodmare farms in 2014 to 2016 ranged from 12.8% to 35.5% (average: 25.6%). The positive rate in sera obtained from a stallion farm (Farm A) in 2002 was 10.7% and the positive rates in sera obtained from two stallion farms, Farm A and Farm B, in 2015, were 40.9% and 13.3%, respectively. Average ages of broodmares for which sera were tested and broodmares for which sera were antibody-positive in 2001 to 2002 were

9.8 years and 10.3 years, respectively. The average ages of broodmares for which sera were tested and broodmares for which sera were antibody-positive in 2014 to 2016 were 10.7 years and 12.0 years, respectively. The average ages of stallions in Farm A for which sera were tested in 2002 and in 2015 were 9.1 years and 11.3 years, respectively. The average ages of stallions in Farm A for which sera were antibody-positive in 2002 and in 2015 were 8.0 years and 13.7 years, respectively. Positive rates in sera obtained from Farms A and C in 2015 were higher than those in sera obtained in 2002, and there was a relationship between the positive rate and aging in Farms A and C in 2015 (Table 6). In Farm A, the positive rate in sera obtained in 2015 from stallions aged 15 to 24 years (83.3%) was clearly higher than the positive rates for other age groups (27.3% in stallions aged 4 to 9 years and 20.0% in stallions aged 10 to 14 years). In Farm C, the positive rates in sera obtained in 2015 from broodmares aged 10 to 14 years old (42.0%) and broodmares aged 15 to 24 years (56.3%) were clearly higher than the positive rate for broodmares aged 4 to 9 years (19.5%). In Farm C in 2002, the positive rates in sera were similar for the three age groups. The average ages of horses reared in Farms A and C had increased from 9.1 years to 11.3 years and from 9.4 years to 10.6 years in 2015, respectively. In Farm A, the same four horses were reared both in 2002 and 2015, and two horses were EFV antibody-positive in 2002 (one 4-year-old horse and one 10-year-old horse) and also antibody-positive in 2015 (one 17-year-old horse and one 23-year-old horse). The remaining two horses did not possess EFV antibody in 2002 (one 9-year-old horse and one 10-year-old horse) but possessed EFV antibody in 2015. In Farm C, there was no same horse reared both in 2002 and 2015.


**Table 5.** Prevalence of EFV antibody in horses in Hokkaido in Japan.

<sup>1</sup> Number in parenthesis is the average age (years) of antibody-positive horses; <sup>2</sup> Number in parenthesis is the average age of tested horses; <sup>3</sup> Number in parenthesis is antibody-positive rate; <sup>4</sup> Positive sera/ tested sera.


**Table 6.** Relationship between average ages of EFV antibody-positive horses and EFV antibody-positive rates.

<sup>1</sup> Number in parenthesis is antibody-positive rate; <sup>2</sup> Average age of antibody-positive horses/ average age of antibody-negative horses/ average age of total tested horses; <sup>3</sup> One positive horse also had the antibody in 2002 (4-year-old horse). The remaining two horses were not reared in 2002; <sup>4</sup> This horse was reared in 2002 (9 years old) and was antibody-negative in 2002; <sup>5</sup> Both horses were reared in 2002 (10 years old) and one horse was antibody-positive in 2002; <sup>6</sup> These horses were not reared in 2002.

#### **4. Discussion**

In this study, we isolated an EFV strain for the first time in Japan from PBMC obtained from a horse that showed symptoms of wobbler syndrome after surgery for intestinal volvulus. In general, spumaretroviruses have no pathogenicity in animals. Although pathogenicity of EFV has not been clearly demonstrated, the EFV we isolated and the symptoms observed in the horse might have no relationship. Concerning EFV isolation from PBMC, Tobaly-Tapiero et al. [8,32] reported that it took 4 weeks to isolate EFV in a highly FV-permissive adherent cell line (either human U373-MG cells or hamster BHK21 cells) after pre-cultivation of PBMC with the mitogenic lectin PHA-P for 2 days. On the other hand, we were able to isolate EFV from mitogen-untreated PBMC in HFK after co-cultivation for 10 days. Recently, we isolated another EFV from PBMC obtained from an EFV antibody-positive horse in HFK after co-cultivation for 10 days. These results suggested that isolation of EFV from PBMC could be conducted by co-cultivation with HFK without pre-cultivation of PBMC with mitogens. However, in our system, cell-free EFV was not released into the supernatant of EFV-infected HFK cell culture. Furthermore, cell-free EFV was not produced after three cycles of freezing and thawing of infected cells in culture medium. BFV is also highly cell-associated and spreads mainly through cell-to-cell transmission [33–36]. Most primate foamy viruses budded from intracellular membranes and cell-free viruses were produced following three cycles of freezing and thawing of infected cells [32]. However, Tobaly-Tapiero et al. [32] obtained cell-free EFV in the culture supernatant of EFV-infected

human U373-MG cells without a freezing and thawing procedure. In ref. [32], it was reported that this phenomenon might be due to the lack of a dilysine motif in the C-terminus of the primate Env glycoprotein [37,38]. Our EFV isolate also lacked the dilysine motif. Therefore, the process of EFV replication in cells might be different depending on the cell type. We plan to propagate our EHV isolate in human U373-MG cells to confirm the results obtained by Tobaly-Tapiero et al. [32].

Nucleotide sequence data for EFVeca\_LM showed high identities to those for prototype EFV (97.2% to 99.1% in various coding regions). The FFV clade clusters with a sequence identity of about 94% to 99% in partial *gag* and *pol* genes [14,39]. However, in the partial *env* gene, FFVs were divided into two distinct genotypes corresponding to two distinct serotypes [39–42]. In BFVs, phylogenetic analysis of complete genomic sequences revealed two clades, the European clade and non-European clade [39]. A recent Japanese BFV isolate belonged to the non-European clade based on results of phylogenetic analysis of partial *env* gene sequences [36]. Therefore, EFV might also be divided into two or more clades if more EFV isolates are obtained worldwide.

Since sero-prevalence of the EFV antibody in broodmares in Japan was about 25% in both the periods 2001 to 2002 and 2014 and 2016, it is thought that EFV infection might persist in almost a constant percentage of horse populations. This is the first report on sero-epidemiology in horses, though preliminary studies in horses in Poland showed the presence of provirus nucleic acid of EFV in about 15% of the tested animals [43]. In our study, the EFV-positive rate increased in an age-dependent manner, especially in Farms A and C in 2015. Furthermore, the average ages of horses reared in both farms in 2015 were slightly increased compared to those in 2002. In Farm A, two horses were sero-converted between 2002 and 2015. The mechanism of EFV transmission is not known, but it is likely that the longer the time spent in the same farm, the greater is the chance to become sero-positive [23]. A significant interaction between age and sero-positivity to BFV has also been reported and it was suggested that this phenomenon is due to horizontal transmission [23]. Furthermore, since the EFV-antibody positive rate in stallions of Farm A in 2015 was the highest among all farms in both periods, sexual transmission might have occurred in the breeding season. In any case, it is thought that most infections result from horizontal transmission.

Fortunately, we have sera from Farms A and C that have been stocked monthly for about 18 years in our laboratory, and we plan to conduct a detailed serological survey to determine the epidemiology of EFV infection in horses. Furthermore, we plan to isolate other EFVs from sero-positive horses and examine the molecular epidemiology of EFV infection in horses in Japan.

**Author Contributions:** Conceptualization, R.K.; Methodology, R.K.; Formal Analysis, R.K.; Resources, Y.T., H.H., N.T.; Writing-Original Draft Preparation, R.K.; Writing-Review & Editing, Y.T., N.T.; Supervision, R.K.

**Funding:** This research received no external funding.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

#### *Article*
