Detection and genotypes of Toxoplasma gondii DNA in feces of domestic cats in Colombia

The high prevalence of Toxoplasma gondii in the human population in Colombia has been linked to the existence of a high density of urban stray cats, exposing the whole population to a high density of oocysts. The goal of this study was to determine the DNA prevalence of T. gondii by conventional PCR and to phylogenetically analyze ROP18 sequences from positive samples in domestic cat (Felis catus) fecal samples in the city of Armenia, Quindío. Fecal samples from 140 cats were collected from 10 districts around the city. Samples were concentrated using Ritchie’s method and analyzed through optical microscopy. Concentrates were used for DNA extraction followed by nested PCR amplification for T. gondii gene B1. PCR for ROP18 was performed on all B1 positive samples; the ROP18 sequences obtained were related to the Archetype I Brazilian and Chinese strains. No oocysts were detected by optical microscopy; however, 17.8% (25/140) B1 and 24% (6/25) ROP18 PCR-positive samples were detected. Phylogenetic analyses showed that isolates clustered into a single group. We assessed whether associations existed between T. gondii positive fecal samples and survey variables such as cat healthcare and socioeconomic characteristics of owners, but no statistically significant associations were found. The presence of T. gondii in cat feces is an important factor contributing to the high prevalence in the human population of this city.


Introduction
Toxoplasma gondii is an obligate intracellular parasite with worldwide distribution inducing toxoplasmosis and infecting humans via warm-blooded animals. Wild and domestic felids are the only known definitive hosts with the ability to shed oocysts in their feces. Common pathways of infection include oocyst-contaminated water, soil, and food; tissue cysts in undercooked or raw meat; and congenital transmission [13,20,37,43]. Oocysts are the environmentally resistant form of the parasite and play a key role in transmission to new hosts and ecosystems, generating the need to study humans alongside domestic and wild animal populations [43]. The large number of oocysts shed during primary-infection by felids could lead to extensive environmental contamination, which can infect a high number of intermediate species, such as humans, mice or birds. [26]. High rainfall rates can facilitate survival of oocysts for months, explaining why regions with higher precipitation show higher prevalence compared to arid regions, which show far lower rates of infection in the population living in these areas [1,16,25].
In Colombia, the prevalence in the human population varies between 30% and 60% [6], and this high prevalence has been linked to the existence of a high density of urban stray cats, exposing people to an elevated density of oocysts [10]. This high density of free-ranging domestic cats can explain why cat ownership in homes does not increase the risk for T. gondii in surveys in some cities in Colombia [29]. A study in Armenia city in 1998 found 89.3% seroprevalence in 28 domestic cats and detected a 66.6% shedding prevalence of T. gondii-like oocysts by microscopy in fecal samples from 18 cats [30]. In 2006, another study found a seroprevalence of 84.8% in 33 stray cats, but no oocysts were identified by microscopy in feces collected from the rectum from these cats [10]. Although training helps with identification of T. gondii-like oocysts, morphological structure along cannot confirm that the oocysts visualized are actually T. gondii, as Hammondia oocysts for example look identical [11]. Consequently, molecular detection-based methods, like PCR, can be an alternative and complementary method to microscopy to identify cats infected with T. gondii. As a consequence, the objective of this study was to determine the prevalence of T. gondii DNA in cat fecal samples by conventional PCR and to analyze the ROP18 gene from positive samples in Armenia, Quindío, Colombia.

Methodology Sample
Armenia is the capital of the department of Quindío in Colombia's central mountain range, at an altitude of 1480 m above sea level. Armenia has 301,226 inhabitants, according to the projections of the Administrative Department for National Statistics (DANE, 2018). Armenia is divided into 10 administrative zones, called communes. The number of cats to sample was estimated according to the rabies vaccination program statistics reported in Armenia, with a population of 15,015 owned cats [28]. Therefore, considering inferences with prevalence with 95% confidence, an accepted error margin of 8%, and using an expected prevalence of 66.6% based on previous studies, at least 132 owned cats needed to be sampled according to Epiinfo v.7.2.3.1 software. The residences were selected at random in the 10 communes of the city of Armenia.

Questionnaire
The owners were interviewed about the following aspects: number of stray cats observed 500 m around the residence; commune; age and sex of domestic cats; kind of food that the domestic cats eat; water source for domestic cats to drink; last deworming (recently: 30 days ago or less; not long since: 30-90 days; long ago: more than 90 days); and hunting habits.

Microscopic analysis
The cat feces samples were processed with the Ritchie technique [35], approximately 5 g of fecal sample from each cat were diluted in 15 mL of 0.9% sterile saline solution and centrifuged for 5 min at 1600 Âg. This washing procedure was repeated three times. We discarded the supernatant, and resuspended the pellet using 5 mL of sterile saline solution. Then, we added 5 mL of 10% formalin solution and 3 mL of 99% diethyl ether (Sigma, USA) to each tube. We sealed and rigorously shook the tube to bring the diethyl ether into contact with all parts of the sediment and performed a new centrifugation at 1000 Âg for 2 min. After this, 30 lL of sediment were scanned in triplicate using Olympus microscopy with 40Â objective lens using 1% parasitological lugol (Químicos Albor, Colombia) as a contrast visual. The microorganisms were identified by three observers and confirmed with the collaboration of Dr. Fidel Angel Nuñez from the Tropical Medicine Institute Pedro Kouri, Havana, Cuba.

Toxoplasma gondii DNA extraction procedure and PCR detection
The resultant pellet from the Ritchie technique was used to obtain DNA from cat feces. The pellets were washed four times in 2 mL tubes with PBS and centrifuged at 4500 Âg for 10 min. The supernatant was discarded and 600 lL of DNAzol (Invitrogen, USA), 10 lL of isoamyl alcohol (Fisher Scientific, USA) and 0.3 g of zirconium silicate beads with 0.5 mm diameter (BioSpec, USA) were added. Afterwards, the tubes were shaken five times in a Mini Bead Beater (BioSpec, USA) to maximum speed for 1 minute and 1 minute in ice [41]. Later, a Wizard Genomic DNA extraction kit (Promega, USA) was used for nuclear lysis and purification, following the manufacturer's protocol. In order to detect DNA, the T. gondii B1 repetitive fragment (GenBank accession number AF179871) was used in an amplification nested PCR method [5]. All B1 positive samples were sequenced to confirm that the amplified product was in fact T. gondii, because it is possible to amplify DNA from other organisms [27]. As has been previously described [31,41,45], the primers for the first PCR were Toxo N1 5 0 -GGAACTGCATCCGTTCATGAG-3 0 and Toxo C1 5 0 -TCTTTAAAGCGTTCGTGGTC-3 0 to obtain a 193-bp fragment. The second PCR was performed with the primers Toxo N2 5 0 -TGCATAGGTTGCCAGTCACTG-3 0 and Toxo C2 5 0 -GGCGACCAATCTGCGAATACACC-3 0 to obtain a 96bp fragment. The first amplification protocol consisted of one initial stage of denaturation for 5 min at 94°C, followed by 40 cycles of amplification, and 1 cycle consisting of 1 min at 94°C for DNA denaturation, 1 min of annealing at 53°C, and 1 min of extension at 72°C. Subsequently, an additional step of 10 min of final extension at 72°C was performed. The second PCR with the product of the first amplification, consisted of one initial stage of denaturation for 5 min at 94°C, followed by 14 cycles of amplification, and 1 cycle consisting of 1 min at 94°C for DNA denaturation, 1 min of annealing at 53°C, and extension at 72°C for 30 s. Then, an additional step of 10 min of final extension at 72°C was carried out. The ROP18 T. gondii sequence (GenBank accession number JX045319.1) was detected in B1 PCR-positive samples. To detect this sequence, we used forward primer ROP18S 5 0 -GACCGTCTTTCAAGAGGAGGA-3 0 and reverse primer ROP18R 5 0 -ACGCTGGTGAGAGGTGCAC-3 0 to obtain a 514-bp fragment. The amplified protocol consisted of one initial stage of denaturation for 3 min at 94°C followed by 35 cycles of 30 s at 94°C for DNA denaturation, 45 s at 60°C for annealing, and 30 s at 72°C for extension. Then, 5 min at 72°C for final extension was performed. All primers were synthetized by Invitrogen Corporation (USA). Finally, a 1.5% agarose electrophoresis gel was used to analyze PCR products. The positive control was DNA from the T. gondii control RH strain, and the negative control was distilled water in the presence of primers.

Statistical analysis
As the dependent variable was binary (presence vs. absence of T. gondii DNA in fecal samples), a Chi-squared test was used to assess relationships with other variables including demographic, behavioral and cat care. A p value < 0.05 for the significance level was employed. OR values were calculated with a 95% confidence interval. The Statgraphics Centurion v.17 software was used.

Sequencing and phylogenetic analyses
PCR-positive products for ROP18 were gel-purified from low-melt agarose gels, followed by recovery using a Wizard PCR SV and PCR clean up system kit (Promega, USA). Sequencing was done under BigDye Ò terminator cycling conditions by using the normal automatic service by Macrogen (South Korea) in a 3730XL DNA sequencer with the same primers as the PCR amplifications. BLAST (http://blast.ncbi.nml. nih.gov/Blast.cgi) search on the GenBank database with all ROP18 sequences was performed, verifying that they belong to T. gondii, as well as to identify closely related genotyped strains. We explored strain relatedness in more detail through phylogenetic analyses, including ROP18 sequences available in GenBank and ToxoDB. In a previous study, T. gondii ROP16 sequences from human and meat samples obtained in the same study area showed a high degree of genetic divergence and clustered with highly virulent strains, and we expected the ROP18 sequences we obtained to show the same pattern within the phylogeny [2]. Sequences were aligned with T. gondii sequences available in GenBank and ToxoDB (Supp. Table 1), restricting the number of sequences per strain to one (except for specific strains with sequences in both GenBank and Tox-oDB). We included ROP18 sequences isolated from patients with ocular toxoplasmosis in the same study area [38]. Alignment was performed with MAFFT v7.187 [21], using the auto routine and default settings. Most of downloaded sequences corresponded to the complete gene (1663-1671 bp), depending on the strain [15], and sequences obtained here represented only a partial region. Therefore, we decided not to trim the resulting alignment, as regions dominated by gaps, insertions or deletions usually contain phylogenetically useful information, and the percentage of identity obtained in BLAST suggested that our sequences were sufficiently informative.
The evolutionary model and partition scheme that best suited the aligned sequences were determined using PARTITION-FINDER v 1.1.1 [22], according to the corrected Akaike information criterion (AICc). The optimal partitioning was the minimum partitioning scheme, and the resulting model was K80 [39]. These parameters were implemented in the Bayesian analyses executed in BEAST 1.8.3 [8]. We assumed a constant population size to estimate the patterns of speciation across the phylogeny, and a strict clock; we kept priors at the default. The number of Markov chain Monte Carlo (MCMC) iterations was set to 100 million, with 25% of trees obtained as burning, retaining trees each 1000 steps. Two independent chains were run and we checked effective sample size (ESS) in Tracer v1.6 [8]. The maximum clade credibility (MCC) tree was estimated in TreeAnnotator [8], discarding 25% of initial trees. We also investigated alternative phylogenetic grouping, constructing an unrooted network (split network) with the Neighbor-Net method implemented in SplitsTree4 [18]. We estimated a haplotype network, with equal weighting on transversions/transitions, without considering gaps/missing and removing invariable sites, and the median-joining network algorithm [3] implemented in PopART [24].

Prevalence of Toxoplasma gondii by microscopy and PCR in cat fecal samples
Between November 2014 and August 2015, in the communes selected, 521 residences were visited. In 217 of these houses (41.6%), there was at least one domestic cat, and in 140 of the 217 houses (64.5% of visited houses with at least one domestic cat), it was possible to collect fecal samples with permission from the owners. We found via optical microscopy, Ancylostoma spp. in 6/140 samples (4.3%), Toxocara spp. in 6/140 samples (4.3%), Toxascaris spp. in 3/140 samples (2.1%), and Hymenolepis spp. in 4/140 samples (2.8%) (Fig. 1). Toxoplasma gondii-like oocysts were not observed.
Of the 140 collected samples, 25 (17.85%) were positive for T. gondii DNA by B1 PCR and confirmed through sequencing according to the results obtained from Macrogen (South Korea). Prevalence among the communes varied between 0% and 37.8% (Table 1), but these differences did not attain statistical significance (v 2 test p = 0.27).

Phylogenetic analysis of ROP18 sequences
We assayed ROP18 amplification on B1 PCR-positive samples and found 6/25 (24%) ROP18-positive samples. The Bayesian tree retained (Fig. 1A) clarified the phylogenetic position of T. gondii isolates obtained from cats in the present study, relative to the other T. gondii strains analyzed. Our sequences clustered into a single clade with Brazilian and Chinese strains (Fig. 1, Clade C). Despite overall support for most of the nodes, we obtained strong support (PP < 0.99) for three major clades: the first clade (A) consisted of Archetype III strains, a second clade (B) grouped the Chinese strains I and II [6] and the Archetype II strains [27], and a third clade (C) consisted of Colombian, Brazilian and Chinese strains (III, according to Gao et al. [17]), and a genotype previously identified and named Archetype I [16]. All six ROP18 sequences obtained in the present study grouped into clade C, which included ROP18 sequences from mouse-virulent strains (GT1 and RH) and ocular toxoplasmosis strains (ROP18YM) (Supp. Fig. 1).
Major clades were also identified by the unrooted network (Fig. 1B). The haplotype network constructed (Supp. Fig. 2) showed that three ROP18 sequences obtained from cat feces grouped into a single haplotype, which also contained the Archetype I (RH and GT1) and Chinese III (TgCkjs and TgCts-d1) strains, as well as Brazilian strains. Additionally, the remaining three sequences were found to represent unique haplotypes, separated by one to three mutation steps from the aforementioned haplotype.

Discussion
Oocyst shedding by the domestic cat (Felis catus) is epidemiologically important and shedding needs to be investigated because (1) it is suspected that oocyst density is correlated with high prevalence in some communities, (2) oocysts can lead to outbreaks of acute human toxoplasmosis, and (3) they are probably significantly responsible for infection in animals, which could be animals for human consumption [40]. Likewise, felids are highly susceptible and a single bradyzoite from cyst tissue can be enough to produce infection [9].   In this study, it was not possible to detect T. gondii oocysts using the optical microscopy technique in domestic cat feces. The same occurred in an earlier study in Armenia [10]. One possible explanation for the lack of evidence of oocysts in microscopy examination is that, at a given point in time, only 1% of cats could be shedding oocysts [10] and the power of the sample that we calculated could fail to detect such a low prevalence of oocyst shedding. The low frequency of oocysts observed by optical microscopy could also be explained by the fact that we evaluated owned domestic cats that often eat inside houses. The situation could have been different if we had tested wild cats with a natural pray diet. Furthermore, it is important to note that identification may be difficult due to the morphological similarities between T. gondii and other coccidian parasites that can be shed in cat feces. Additionally, oocyst shedding by cats occurs over short periods of time, with primary infection mainly in juvenile cats [7,19]. Other gastrointestinal parasites were found, showing that the frequency of domestic cat feces samples with helminths was lower compared to previous reports in the same city [12]. This could indicate improved deworming habits by cat owners in the city of Armenia.
In contrast with the absence of oocysts via microscopy, PCR detected 17% of cats with the presence of T. gondii DNA in cat feces samples. The prevalence of T. gondii, according to the DNA frequency, in this study was 17.85% (25/140). This is low in comparison with other studies, like the one performed in Portugal, where 20.5% prevalence was found in domestic cats feces samples using the B1 sequence in PCR [14]. However, our study found a higher prevalence compared to that observed in a Korean study, where prevalence was 4.5% [20], and a study in Switzerland, where prevalence was 0.4% [4]. This is the first study in Colombia detecting T. gondii DNA from cats feces samples, but the frequency of positive samples may be lower compared to previous studies with oocyst prevalence reported at 66.6% [30] and seroprevalence at 84.8%-89.3% [10,30]. These differences could be associated with the methodology and cats sampled. These PCR results must be interpreted with caution since the T. gondii PCR-positives are not always related to oocysts. This is because the presence of T. gondii DNA could be due to the presence of bradyzoites of infected prey consumed by a cat [32]. In addition, the PCR methods overestimate exposure to infective parasites because they detect all populations of infectious and noninfectious parasites alive or dead [36]. The utility of PCR detection on cat feces samples is that it offers information on the maximum occurrence and, after sequencing for PCR products, results can be used for genetic analysis of the strains circulating in a geographic region. We did not find statistically significant relationships between the presence of T. gondii DNA and the variables obtained from the questionnaire. However, other studies [23,42] have indicated that cats were probably exposed to the infection from an early age and later, many times throughout their lives, through predation and hunting [43].
Lower prevalence of T. gondii in cat feces in some countries could be related to greater access of domestic cats to human food with a low risk of infection [43]. In contrast, high prevalence could be explained by the access of domestic cats to undercooked meat provided by their human owners or by hunting [44].
Previous studies on the genetic diversity of T. gondii strains infecting cats in Colombia have shown that parasite populations are highly diverse [10]. Phylogenetic and haplotype analyses including ROP18 sequences obtained here also found high genetic diversity, despite being obtained from closely located sampling sites. Interestingly, their phylogenetic relationships confirmed the presence of virulent ROP18 alleles circulating in cats, as reported for patients with ocular toxoplasmosis in the same city [38] (Supp. Fig. 1). However, fine-scale genotyping is imperative to reach definitive conclusions, facilitating the evaluation of the virulence of cat T. gondii strains, and their relationships with ocular toxoplasmosis strains. The clades found with recent phylogenetic analyses based on a concatenated matrix that included the ROP18 gene [15,33,34]. In conclusion, this study confirmed a lack of microscopic detection of oocysts in domestic cats in Armenia (Colombia), but that it is possible to detect and analyze genetic diversity by PCR analysis. In this case, the ROP18 gene was found to be an indicator of virulence of T. gondii found in cat feces samples.

Conflict of interest
The authors declare that they have no conflict of interest.

Supplementary materials
Supplementary materials are available at https://www. parasite-journal.org/10.1051/parasite/2020023/olm Supplementary Figure 1. Polymorphic nucleotide sites at the ROP18 polymorphic region for 6 isolates of T. gondii from cat feces (plus RH control strain), 4 isolates of T. gondii from ocular toxoplasmosis [38], and three sequences representing three archetypal lineages detected for T. gondii strains (GT1, ME49 and VEG). Consensus sequence was determined by the percentage of nucleotides common to all sequences. Sites in grey indicate similarity with the consensus sequence. Supplementary Figure 2. Median-joining network for T. gondii haplotypes of the rhoptry protein 18 gene. The size of the circles represents the haplotype frequency and the colors represent the country to which the strain belongs. The number of transversal lines connecting haplotypes represents the number of substitutions between them. Sequences found in cats in the present study are shown with numbers (6_3_3; 5_1_6; 3_1_6; 3_1_8; 4_3_2; 4_1_8).
Supplementary Table 1. Accession numbers and country of T. gondii ROP18 gene sequences downloaded from the GenBank database, and later included in our phylogenetic analyses.