Canine vector-borne pathogens in rural dogs in Chile: molecular survey and co-infection patterns

Background: Canine vector-borne pathogens (CVBP) comprises a relevant and globally distributed group of disease agents. The aim of this study is to determine de co-occurrence of the most relevant CVBP of veterinary and zoonotic interest, in free-ranging, owned, rural dogs of central Chile, and to evaluate risk factors and potential “hidden” hematological alterations associated to pathogen coinfection by two or more pathogens. Methods: Nine groups of canine vector-borne pathogens (CVBP) were molecularly investigated in 111 free-ranging, owned rural dogs in the Metropolitan Region of Chile. Results: At least one pathogen was detected in 75% of the dogs. The most prevalent agent was Anaplasma platys (36%), followed by Candidatus Mycoplasma haematoparvum (CMhp; 31%), Mycoplasma haemocanis (Mhc; 28%), Trypanosoma cruzi (17%), Leishmania spp. (4.5%) and Acanthocheilonema reconditum (1%). DNA of Ehrlichia spp., Rickettsia spp., Bartonella spp., Piroplasmida and Hepatozoon spp. was not detected. Thirty-eight dogs (34%) were coinfected, either by two (n=20), three (n=7), or four agents (n=1). The most common co-infection pattern was CMhp – Mhc (n=14). CMhp was involved in 71%, Mhc in 58%, and A. platys in 50% of the co-infections. Prevalence of A. platys was higher in juvenile than in adult dogs, whereas the opposite was found for CMhp and Mhc. Adult dogs had five times more probabilities of being coinfected than young animals. Dogs positive for A. platys were infested by a larger number of Rhipicephalus sanguineus sensu lato ticks than uninfected individuals. At clinical evaluation, most of the animals were considered healthy, with only eight dogs (7%) presenting pale mucous membranes. Co-infected animals showed higher white blood cell count, segmented neutrophil count GGT levels than non-co-infected

4 endemic area for Chagas' disease, with 1507 human cases notified in 2017 [22]. Since 1999, Chile has been recognized as a country that eliminated the domestic cycle of Trypanosoma cruzi [22].
Antibodies against T. cruzi were detected in 7% of periurban dogs in the Metropolitan Region [23] though no molecular evidence has been provided. While the main vector of the domestic cycle, Triatoma infestans has been controlled, the role of the sylvatic vector, Mepraia spinolai, is becoming increasingly important in periurban areas of Santiago [24,25]. Acanthocheilonema sp. was firstly detected in a dog [26], and later on reported with a prevalence of 17% with the Knott test in the Metropolitan Region [27]. Circulating microfilariae of Dirofilaria repens, Acanthocheilonema reconditum and Acanthocheilonema dracunculoides were also detected in 22% of the screened rural dogs from Lampa (Metropolitan Region) (López et al., 2012b). In South America, Leishmania infantum (syn. L. chagasi) is the most important etiological agent of canine visceral leishmaniosis, while Leishmania brazilensis mostly causes cutaneous forms in dogs [29]. There are no previous reports in Chile of Leishmania spp. Similarly, Hepatozoon spp. and piroplasmid parasites have not been investigated in Chile, despite the presence of suitable vectors (e.g. R. sanguineus) and the report of Hepatozoon canis in whole region of South America [30,31] and of Hepatozoon felis or Hepatozoon americanum, in wild canids in Argentina [32,33]. Canine babesiosis in South America is caused by the intraerythrocytic protozoan parasites Babesia vogeli and B. gibsoni [34,35].
Untreated animals are useful sentinels for vector and pathogen environmental pressure [36]. Dog population in Chile was estimated in 4.059.200 individuals [37], and owned free-roaming dogs are common in Chile [38]. Owned free-ranging dogs (i.e., characterized by the lack of continuous direct supervision and irresponsible ownership) are considered the intermediate stage between wellmanaged pet with movement restriction and feral dogs without human control and management [39].
Prophylactic measures such as antiparasitic treatments are infrequently applied to rural dogs by their owners [13]. Free-ranging dog lifestyle is considered an important factor in disease epidemiology [40]. In fact, outdoor and/ or hunting lifestyle has been associated with higher exposure to some CVBP when comparing with indoor and pet lifestyle [41,42]. Therefore, due to the above-mentioned factors, it is likely that the actual burden of CVBP in the Chilean dog population has been underestimated. The 5 aim of this study is to determine de co-occurrence of nine groups of the most relevant CVBP in freeranging, owned, rural dogs of central Chile, and to evaluate risk factors and potential "hidden" hematological alterations associated to pathogen co-infection by two or more pathogens.

Study area and dog sampling
The study was conducted in the Metropolitan Region of Chile (33°26′16″S, 70°39′01″W), which holds the highest human population density of the country (461.8 habs/km 2 ) [43]. This region has a typical Mediterranean climate, with dry, hot summers and rainy winters. Mean annual temperature is around 14.7ºC (minimum and maximum absolute annual temperature= -1.1 and 37.3 ºC) and annual precipitation is around 243.3 mm [44]. Typical vegetation is sclerophyllous scrub and forest [44].
From 2016 to 2018, 111 free-roaming rural dogs were sampled and examined. All sampled animals were free-ranging (without permanent confinement). Age (based on tooth eruption) and sex of dog were recorded, and general clinical sign examination carried out. Dogs were inspected for ticks for 5 minutes, and the arthropods collected were preserved in 95% ethanol. Blood obtained from the cephalic vein was collected in two separated EDTA tubes and a further tube with serum separator.
The serum was removed after centrifugation and frozen at -20ºC until biochemistry analysis.
Hematological analyses were performed on whole blood and the remaining sample frozen at -20° until molecular analysis. Trypanosoma cruzi was detected and quantified by real-time PCR following the protocols described by Yefi-Quinteros et al. (2018). Leishmania spp. DNA was screened by qPCR using the protocol described by Francino et al. (2006) and positive samples were further analyzed by cPCR using primers and run protocol previously described (Cortes et al., 2004) for sequencing purposes. Filaroids were screened by cPCR as described by Casiraghi et al. (2001). To avoid cross-contamination, DNA extraction, mixing of DNA-free PCR reagents and the addition of the template DNA were carried out in separated areas with separate equipment and solutions. PCR products were visualized on a 2% agarose electrophoresis gel, and later purified and sequenced by Sanger technique. Obtained sequences were then compared with those available in GenBank by BLAST analyses (http://www.ncbi.nlm.nih.gov/blast). Collected ticks were identified using morphological criteria [52].

Hematology and serum chemistry
The following hematological parameters were analyzed through manual and automatic cell counter

Data analysis
Confidence intervals for prevalence were calculated using the "EpiR" package (Stevenson et al. 2013) of R software (R Development Core Team 2013). Differences in the occurrence of pathogens, the existence of coinfection and the number of pathogens per host depending on the dog's sex (male/female) and age (young/adult) were evaluated. For A. platys, Mhc and CMhp, the prevalence and abundance of R. sanguineus were also analyzed as independent variables. Generalized linear mixed models (GLMMs) were used to study the binary variables (i.e., pathogen occurrence = absence/presence; pathogen coinfection = no coinfected/ coinfected), count data (number pathogens), and fixed and random effects. GLMMs handle non-normal data by using link functions and exponential family distributions and incorporate random effects [53]. Study zone (Andean hillside, central valley and coastal hillside) was included as random effect. GLMMs were analyzed using the "lme4" package [54] of R software [55] with binomial error (logit-link function) and Poisson error (loglink function). The best model was selected using the "dredge" function from the "MuMIn" package, which generates, given a full model, a subset of models and selects the best model that best fit the data, based on Akaike information criterion corrected to sample size (AICc). The overall fit of best model was assessed by residuals analysis and comparison with the null model (with an intercept and random effects only), using the likelihood ratio test. Individuals with information on all factors were included in the models. In the case that a category of the independent variables had not any positive animal, the evaluation of that variable was carried out by Fisher's exact test. In that case, that variable was removed from the full model of GLMM analysis. Differences in hematological and biochemistry values were tested using Student's t or Mann Whitney U depending of data distribution.
Initially, differences between adult and young dogs were evaluated. In case of not finding significant differences between ages, these were pooled to assess the association between parameter and coinfection status, and otherwise were analyzed separately. All statistical analyses were carried out using R software [55].

Pathogen occurrence and co-infection patterns
At least one CVBP was detected in 75% of the dogs (Table 1)  (n=22) and A. platys in 50% of them (n=19).

Risk factor analysis
The probability of being infected by A. platys was four times higher (OR=4. 13 Figure 3). The prevalence and abundance of R. sanguineus were not related to the presence of any other agent. Adult dogs had five times more probabilities of being coinfected than young ones (OR=4.9, 95%CI= 1.4-17.8) (Figure 2). The best GLMM (with Poisson error) for the number of pathogens detected per dog was the null model. Hence, no studied factor was associated with the number of detected pathogens.

Clinical, hematological and biochemical findings
Most of the animals were considered apparently healthy in the clinical exploration. Only eight of the dogs (7.2%) presented pale mucous membranes, without differences between co-infected and non-coinfected animals (Fisher's p=1). Co-infected animals showed higher white blood cell count (WBC) (t= 2.01, p<0.05) and segmented neutrophil count (t=2.46, p<0.05) and GGT levels (U= 583.5, p<0.03) (Table 2; Figure 4).

Discussion
The present study is the most extensive study ever in Chile, including the most relevant CVBP in dogs.
We found a widespread occurrence of CVBP in rural, free-ranging dogs of central Chile, with three quarters of the dogs positive to at least one pathogen. Previous studies in other parts of the world showed that rural dogs are frequently exposed to or infected by different vector-borne pathogens 9 [56][57][58]. The outdoor activity of these dogs exposes them to a range of vectors, which is, together with the absence of antiparasitic prophylactic management, the most likely reasons for our findings.
In this sense, higher rates of exposure or infection were found in rural dogs when compared with their urban counterparts [59][60][61][62]. In the Metropolitan Region of Chile, the prevalence of R. sanguineus and Ct. canis was indeed higher in rural than in urban dogs [63].
Anaplasma platys was the only Anaplasmataceae confirmed in this study. This picture is similar to that reported in other geographical areas where R. sanguineus sensu lato is the only tick species infecting dogs (e.g., Latrofa et al. 2014;Otranto et al. 2019). The absence of E. canis, a bacterium that was reported in northern Chile [11], is not surprising considering that only the tropical lineage of R. sanguineus s.l. is able to transmit E. canis [66], and that in central Chile (where our study was conducted), only the temperate lineage has been previously reported [12,67]. Accordingly, the seroprevalence of 69% using and A. phagocytophilum-based IFI commercial kit registered in dogs from Santiago ( [15] probably corresponded to A. platys, due to the serologic cross-reaction among Anaplasmaceae species [68]. The observed prevalence in our study can be considered high when compared with other studies in rural dogs. It is close to the prevalences reported by Brown et al. [69] in free-roaming dogs associated with remote Aboriginal communities in Australia, but higher than other studies carried out in Brazil [60], Ivory Coast [70], Kenya [70] and Uganda [56]. The higher probability of A. platys infections in young dogs in our study was already been recorded in a previous study in Africa [70], most likely due to a primary exposure of juvenile individuals to the pathogen [71,72] and might be related to the lower levels CD8 T lymphocytes found in young dogs [73], which has a role in clearance of rickettsial infections [74]. In agreement with our results, other studies found that dogs infested with R. sanguineus were more likely to be infected with or exposed to Anaplasma than uninfested dogs [60,75,76]. Overall, our results suggest that the risk of infection with A. platy is more associated with the tick abundance than just the presence of the tick.
The overall prevalence of hemoplasmas in in our study was similar to the prevalence observed in rural dogs in southern Chile [21], and in rural or free-ranging dogs worldwide, such as Australia [77] and Brazil [78]. Rural environments were suggested to be a risk factor for hemoplasma infections when compared with urban environments [21,79], and, accordingly, in free-ranging compared with domestic pet dogs [80]. We found similar prevalence for both hemoplasma species, coinciding with the findings by Soto et al. [21], although these authors did not detect any co-infection. The observed higher infection rate in older dogs may be explained by and increased probability of exposure throughout life and/or by the characteristic long-term bacteremia of hemoplasma infection [81,82].
Lack of hemoplasma clearance was reported in infection follow-up studies [83,84].
To the best of our knowledge, this survey represents the first molecular detection of T. cruzi in dogs in the Metropolitan Region of Chile, although the presence of parasitized dogs in this region was known [23,85]. The only previous study in the country was conducted in two regions in the North of Chile [86]. It is known that dogs are competent hosts of importance in the cycle of T. cruzi in endemic areas [87,88]. Despite of this, there have been few molecular studies in rural dogs in the Americas [89][90][91].
Rural dogs have been suggested as a bridge between the domestic and sylvatic transmission cycles [89], and this can be the case in our study area, where all of the studied dogs live outdoor and some of them accompany muleteers in areas where triatomines abound [25]. Further studies should aim to characterize the genetic diversity of T. cruzi in the region.
The presence of Leishmania spp. is herein reported for the first time in Chile, though it is endemic in some neighboring areas of Argentina [29,92,93]. Only one of the five Leishmania-positive dogs were also infected by T. cruzi, ruling out the possibility of an undesirable amplification of Trypanosoma kDNA using this protocol. Sand flies are present in Chile [94], but none of the species has known vectorial capacity. If confirmed, our findings would markedly extend the distribution of Leishmania spp., though the species identity has not been ascertained. Due to the importance of this parasite for veterinary medicine and public health, the proper characterization of the cases should be further investigated.
A higher prevalence of microfilariae of A. reconditum was detected at knott test in previous studies conducted in Chile [27,28] and this could be due to the number of animals tested or to the lower occurrence of specific vectors (i.e., fleas and lice) in the sampled animals. Ctenocephalides felis, in particular, is considered a well stablished intermediated host [95][96][97]. The globally distributed Ctenocephalides felis is also a common parasite of dogs in the Metropolitan Region of Chile [13,63] although were found at low prevalence in the dogs included in this study and lice were not found at all (unpublished results).
Absence or a very low number of Rickettsia and Bartonella DNA amplification in dogs reported here has been previously observed elsewhere [98,99], though a high prevalence of the pathogen has been diagnosed in ectoparasites vectors [17,100]. In absence of acute infections, low blood bacterial levels could be observed for these pathogens, being detect only by cPCR [101,102].
Pirosplasmid and Hepatozoon spp. DNA was not detected during this survey, likely due to the geographical isolation of Chile from the rest of the continent. Nevertheless, the arrival of imported cases of dogs infected with H. canis, B. vogeli and/or B. gibsoni must be considered a risk for the Chilean canine population, also considering the widespread presence of R. sanguineus in the country and the circulation of dogs among countries.
A third of the studied dogs were co-infected with two or more pathogens. Co-infection is considered frequent in CVBD-endemic areas, especially in dogs living environments with high vector density and without antiparasitic treatment [8]. Interestingly, although A. platys was the most prevalent agent in our study, was not the pathogen most commonly associated with co-infections in dogs, in contrast with previous studies in areas where R. sanguineus is prevalent [71] in areas. In our case, hemoplasma species were common in cases of coinfection, and multiple infections has indeed been considered as risk factor for hemoplasma infection [80,103]. Higher WBC and segmented neutrophil level were found in co-infected animals. No consistent leukogram abnormalities have been associated with canine hemoplasmosis or anaplasmosis [21,81,104]. However, increased leukogram values have been associated with T. cruzi infections [105]. On the other hand, higher GGT values were found in co-infected animals. Anyway, almost all the GGT values were into the range of the reference values [106]. It is possible that our findings may be explained by absence of acute stages of infection. Chronically infected dogs usually present low bacteremia or parasitemia [107]. Thus, dogs with chronic or "hidden" infections used to be apparently healthy with absent or minor hematological abnormalities [8,72]. For example, most of the cases of canine haemoplasmosis used to be chronic asymptomatic infections and infected dogs seemed unable to clear the infection [82]. Therefore, as suggested before, co-infection complicates the diagnosis based on clinical examination and hematological and biochemistry abnormalities alone [108].

Conclusion
Rural free-ranging dogs of central Chile are infected or parasitized by a range of agents of veterinary and zoonotic interest, being a remarkable number of dogs infected by more than one pathogen. It is important to remark that those clinically healthy but infected dogs could be acting as subclinical carriers of different CVBP, possibly contributing to the spreading of some of these pathogens to potential vectors and among their owners, other dogs, or protected wild carnivores.

Acknowledgements
We wish to thank to Andrea Chirife, Carla Tagini, Claudia Ulloa, Manuel Lepe, Violeta Barrera for assistance during dogs sampling; to Carla Barría, Giada Annoscia and Rossella Panarese for the essential help during molecular analysis. Graphical abstract was created by Biorender.com.

Availability of data and materials
The data supporting the conclusions of this article are included within the article. Raw data used or analyzed during the present study are available from the corresponding author upon reasonable request.

Funding
This study was funded by Fondecyt Regular 1161593.

Ethics declarations
Ethics approval and consent to participate This study was approved by the authorities in bioethics from Universidad Andres Bello under authorization 08/2016. Dog owners signed an informed consent form before samples were taken.

Consent for publication
Not applicable.

Competing interests
The authors declare that they have no competing interests.     Abundance of Rhipicephalus sanguineus depending on the Anaplasma platys infection status.

Figure 4
Significant differences in total leukocyte count, segmented neutrophil count and gammaglutamyl transferase (GGT) depending in the co-infection status. Black lines indicate de mean and orange lines the maximum and minimum reference values based on Thrall et al. [106].

Supplementary Files
This is a list of supplementary files associated with this preprint. Click to download.
Graph Abstract.tif