Molecular detection of Rickettsia in fleas from micromammals in Chile

Background Rickettsial diseases are considered important in public health due to their dispersal capacity determined by the particular characteristics of their reservoirs and/or vectors. Among the latter, fleas play an important role, since the vast majority of species parasitize wild and invasive rodents, so their detection is relevant to be able to monitor potential emerging diseases. The aim of this study was to detect, characterize, and compare Rickettsia spp. from the fleas of micromammals in areas with different human population densities in Chile. Methods The presence of Rickettsia spp. was evaluated by standard polymerase chain reaction (PCR) and sequencing in 1315 fleas collected from 1512 micromammals in 29 locations, with different human population densities in Chile. A generalized linear model (GLM) was used to identify the variables that may explain Rickettsia prevalence in fleas. Results DNA of Rickettsia spp. was identified in 13.2% (174 of 1315) of fleas tested. Fifteen flea species were found to be Rickettsia-positive. The prevalence of Rickettsia spp. was higher in winter, semi-arid region and natural areas, and the infection levels in fleas varied between species of flea. The prevalence of Rickettsia among flea species ranged between 0–35.1%. Areas of lower human density showed the highest prevalence of Rickettsia. The phylogenetic tree showed two well-differentiated clades with Rickettsia bellii positioned as basal in one clade. The second clade was subdivided into two subclades of species related to Rickettsia of the spotted fever group. Conclusions To our knowledge, this is the first report of the occurrence and molecular characterization of Rickettsia spp. in 15 flea species of micromammals in Chile. In this study, fleas were detected carrying Rickettsia DNA with zoonotic potential, mainly in villages and natural areas of Chile. Considering that there are differences in the prevalence of Rickettsia in fleas associated with different factors, more investigations are needed to further understand the ecology of Rickettsia in fleas and their implications for human health.

whole-genome sequencing analysis, two groups are recognized within the genus Rickettsia: the major typhus group (MTG) and major spotted fever group (MSFG). The MTG is divided into the typhus group (TG) and ancestral group (AG) and is transmitted by insects. MSFG includes the R. felis group, R. akari group, and the "classical" spotted fever group that includes several species transmitted by mites and hard ticks, of which the most important are R. rickettsii and R. conorii, that cause Rocky Mountain spotted fever and Mediterranean spotted fever, respectively [4]. Since Rickettsia research has focused on species that affect humans, other species have received less attention [5]. Thus, there are several species of rickettsiae identified and are exclusively associated with arthropods. They are without known secondary hosts and associated with other organisms such as herbivorous insects, leeches, amoebas, inclusive algae, and plants, indicating that these are more common than suspected [5,6], and that the effects they could cause in humans when contact is made are unknown.
Worldwide, micromammals, and especially rodents, are the main flea hosts. It is recognized that 74% of known flea species parasitize them; therefore, rodents play a fundamental role in the spread of flea-borne diseases, as various species of rodent fleas can also parasitize humans [7]. In addition to this, many rodent species are capable of inhabiting wild environments and adapting to rural and urban environments, which could favor a continuous gradient of transmission between domestic and wild species, and humans [8,9]. In Chile, despite the great diversity of described fleas (114 species), which mainly parasitize rodents [10,11], a scarce number of studies have detected Rickettsia in fleas [12][13][14][15]. These studies have focused on the molecular detection of pathogens in fleas of domestic mammals, identifying R. felis from cat and dog fleas (Ctenocephalides felis and C. canis) in central (Metropolitan region) and southern Chile (Valdivia) [12][13][14]. Recently, "Candidatus Rickettsia asembonensis", "Candidatus Rickettsia senegalensis", and R. felis, were detected in C. felis from cats in the Easter Island (Rapa Nui) [15]. No studies have shown their presence in rodent fleas. If this adds to the expansion of the human population invading wild areas, the chance of contacting fleas on infected rodents increases. Since, in some places, peri-urban rodents provide a link between wild rodent and human communities, humans are exposed to some zoonotic agents that circulate in these natural ecosystems [16,17].
The aim of this study was to detect, characterize, and compare Rickettsia spp. from the fleas of micromammals in areas with different human population densities in Chile. The findings will provide the baseline for the future surveillance of Rickettsia spp. in Chile.

Sample localities and micromammal-trapping procedures
A total of 1512 micromammals belonging to 18 species (Table 1) were captured during a trapping effort of 11,034 trap/nights from 23 localities (9 cities, 6 villages and 8 natural areas) of the 29 sampled, covering 10 administrative regions in Chile and five bioclimatic regions (hyperarid, arid, semi-arid, sub-humid and hyper-humid), latitude between −20.2167 and −53.1667 (Fig. 1). It was conducted from December 2015 to January 2018, during austral summer (December to February) and austral winter (July and September). These localities were selected based on the following demographic characteristics: (i) city, urban entity that has > 5000 inhabitants; (ii) village, urban entity with a population ranging between 2001-5000 inhabitants, or between 1001-2000 people, where less than 50% of the population that declares having worked, is engaged in primary activities (e.g. livestock, agriculture or fishing) [18]; and (iii) natural area, without human settlement, corresponding to national park (NP; unaltered areas of natural and biological diversity), and national reserves (NR; areas protecting wildlife populations or natural resources).
Micromammals were captured using a Sherman trap (23 × 7.5 × 9 cm, Sherman Co., Tallahassee, USA) and wire-mesh traps (30 × 10 × 11 cm; Forma Ltd., Santiago, Chile) baited with oats. The associated use of both types of traps strongly reduced the likelihood of a species being present but not captured. Each locality was sampled for two consecutive nights. In each sampling locality, the traps were placed in four parallel lines approximately 100 m from each other, and each line was equipped with 50 traps set 10 m apart from each other. Only in cities, traps were used along lines with a 5-10 m inter-trap space, and the traps were placed outside the buildings. The rodents were removed from the traps according to standard techniques [19], and were subsequently anesthetized with ketamine:xilazine (1:1) [20]. Flea samples from rodents were collected by hand or with forceps from the host and placed into sterile cryovials tubes with 95% ethanol. For each rodent, the total number of extracted fleas was recorded (abundance); with these data, the overall mean infection intensity (the number of fleas collected from all species/number of infested hosts), the overall mean abundance of infection (the number of collected fleas from all species/total number of hosts), and prevalence (the proportion of infected hosts) were calculated. The micromammals were identified following Iriarte [21]. Micromammals were released after sampling, except for invasive rodents [black rat (Rattus rattus), Norway rat (Rattus norvegicus), and house mouse (Mus musculus)] that were euthanized by cervical dislocation [19].

DNA extraction and PCR amplification
For DNA extraction, 5 fleas per host were selected, and when the number of fleas per host was less than 5, all the fleas were analyzed. Finally, DNA extraction was performed from 1315 fleas. Each flea was washed and cut between the third and fourth abdominal tergite with a scalpel. DNA was extracted from individual fleas using DNeasy Blood & Tissue Kit (Qiagen, Hilden, Germany) according to the manufacturer's protocols. The incubation time was 5 h; following DNA extraction, the flea's exoskeleton was recovered and stored in 96% ethanol to later mount and identify the flea species.

Phylogenetic and BLAST analyses
All DNA sequences were edited and aligned using the Codon Code Aligner (CodonCode Corporation, Centerville, MA, USA). All sequences generated in this study were compared with those available on GenBank using the BLAST program (see http://www.ncbi.nlm.nih.gov/ BLAST /). A Bayesian probabilities tree was created using MrBayes 3.2 based on gltA 830-bp gene fragment, using Anaplasma phagocytophilum as the outgroup. We used the GTR + G substitution model to reconstruct the tree and 10,000,000 bootstrap trials.

Flea mounting and identification
After DNA extraction, each flea's exoskeleton was recovered and mounted on glass slides using conventional procedures. The fleas were identified using a light microscope, taxonomic keys, and the descriptions of Johnson [24], and Sanchez & Lareschi [25]. Voucher specimens (slides) were catalogued in the Museo de Zoología at Universidad de Concepción (MZUC-UCCC, Concepción, Chile) under the accession numbers 46647-46667.

Statistical analysis
The prevalence (percentage of micromammals parasitized with fleas) and abundance mean (mean number of fleas per host) in species of micromammals was calculated with total of samples of fleas collected (n = 2272), and confidence intervals (95% CI) were calculated, using bootstrap (2000 bootstrap replicates). The prevalence of Rickettsia (percentage of fleas infected with Rickettsia) was calculated based on the PCR results. We used generalized linear models (GLM) with binomial distribution and logit function to identify the variables that may explain Rickettsia prevalence in fleas. The explanatory variables analyzed were bioclimatic regions (hyper-arid; arid; semi-arid; sub-humid; and hyper-humid), location type (city; village and natural area) and season (summer and winter). First, we built a model that included all bioclimatic regions and then we built models for each bioclimatic region. To assess the relationship between the prevalence and sample size, a Spearman correlation analysis was performed. The Chi-square test or Fisher's exact test (if an expected cell count was < 5) was used to evaluate the differences in the prevalence of Rickettsia among species of flea. A P-value < 0.05 was considered statistically significant. The data were analyzed using JMP software ® (SAS Institute Inc., Cary NC, USA).

Results
A total of 2272 fleas were collected from 13 micromammal species, with an overall prevalence of 46.7% (n = 706). The overall mean abundance was 1.5 fleas per host and the overall mean intensity was 3.2 fleas per parasitized host (Table 1). Excluding the species in which < 20 individuals were sampled, the micromammals that presented the highest prevalence of fleas were Loxodontomys micropus (Austral greater mouse, 87.5%) and Octodon degus (Fence degu, 78.3%), and the lowest prevalence was found in R. rattus (29.2%). The abundance and mean intensity were higher in O. degus ( Table 1). The marsupial Thylamys elegans (Llaca mouse-opossum) had a prevalence of fleas of 51.4%. All of the flea species found in T. elegans corresponded to species that were also found in rodents (Table 3). Of all collected fleas, 1315 flea specimens were analyzed, corresponding to 27 species from 15 genera and 8 families ( Table 4). The most abundant flea species were Sphinctopsylla ares (n = 211) and Neotyphloceras chilensis (n = 202; Table 4). The rodents that presented the greatest flea richness were Abrothrix olivacea (olive grass mouse, 17 spp.), R. rattus (14 spp.), A. hirta (long-haired  Table 3). Natural areas were where the largest number of flea species (n = 25) and specimens were collected (n = 784), followed by villages (18 species, 349 specimens) and cities (18 species, 181 specimens). Agastopsylla boxi, Ctenoparia jordani, C. topali, Ectinorus cocyti and Plocopsylla lewisi were exclusive to natural areas (national parks and national reserves). Conversely, Xenopsylla cheopis was only found in one city (Iquique). Neotyphloceras chilensis and S. ares were the dominant species in natural areas (N. chilensis (n = 119); S. ares (n = 151)), and villages (N. chilensis (n = 83); S. ares (n = 50)), while Nosopsyllus fasciatus (n = 37), and C. inopinata (n = 25) were the most frequently collected in cities. Leptopsylla segnis, N. fasciatus, and X. cheopis are synanthropic rodent fleas [26], and were more abundant in cities than in villages and natural areas.

Rickettsiae prevalence on fleas
Fifteen flea species were found to be Rickettsia-positive for the short fragment (401 bp) of the gltA gene, 9 for the long fragment (830 bp) of the gltA gene, 10 for the rpoB gene, and 4 for the sca5 gene ( Table 4). The highest prevalence (13.2%) was detected with the gltA 401-bp gene, followed by the rpoB (5.9%), gltA 830-bp (5.0%) and sca5 (0.5%) genes (Table 4). Among the flea species in which more than 20 individuals were analyzed, the prevalence varied between 0-35.1%. The Neotyphloceras spp. had the highest prevalence of Rickettsia (gltA 401-bp = 29.4%, gltA 830-bp = 9.56%, and rpoB = 11.25%; Table 4). The four fragments (gltA 401-bp, gltA 830-bp, rpoB and sca5) showed significant differences in the prevalence of detected Rickettsia (χ 2 = 193.207, df = 3, P < 0.001), exception for gltA 830-bp and rpoB, which did not show significant differences (χ 2 = 1.934, df = 1, P = 0.164). No association was found between the number of fleas analyzed and the prevalence of Rickettsia detected for any of the genes analyzed (rpoB: ρ = 0.4267, P = 0.12; gltA: ρ = 0.3757, P = 0.18; sca5: ρ = 0.3272, P = 0.35). According to the GLM analysis, the prevalence of Rickettsia infection was significantly higher in the semi-arid region (27.8%). In addition, the overall prevalence was significantly higher in the winter (20.6%) than in the summer (5.3%). The prevalence of Rickettsia was higher in natural areas (15.9%), and cities exhibited a marginally significant lower prevalence (4.97%) compared to the other two location types (village: 11.2%; Table 5). Comparisons between bioclimatic regions showed that in the arid region, the prevalence of Rickettsia was higher in the natural areas and in the winter. While in the semiarid region, the highest prevalence occurred in the winter (73.7%), and the highest prevalence of Rickettsia was detected in the natural areas (77.8%), differentiating from the cities (14.0%). In the sub-humid region, there was no effect of the factors on the prevalence of Rickettsia, whereas in the hyper-humid region, we detected Rickettsia (5.49%) only in the natural areas.

Discussion
To the best of our knowledge, we have provided for the first time evidence for the presence of Rickettsia DNA in 15 flea species identified on wild micromammals and synanthropic rodents in Chile. The prevalence of Rickettsia spp. infections in fleas varied between species of flea, bioclimatic regions, seasons and location type. We found a higher prevalence in winter, the semi-arid region and natural areas. The fleas were characterized as being highly hostopportunistic, occupying various host species [7]. This is confirmed by our study, since of the 27 flea species collected, 19 parasitized more than one species of micromammal. We also highlight the high flea species richness recorded in R. rattus, where 10 of the 14 species identified in this rodent correspond to the flea species identified on native rodents. This rodent was mainly  captured in urban areas; however, we also found it in rural and natural areas, this occurs mainly because these rodents have an omnivore diet and plasticity in their behavior, characteristics that allow them to inhabit a great diversity of environments, adapting successfully to urban, rural and wild environments [30,31]. Rickettsiapositive fleas parasitizing R. rattus in these three areas indicate that this species could play a key role in spreading the disease from wild to urban environments [16,32]. Conversely, we also observed that wild species enter human-occupied environments since they provide shelter and food. Abrothrix olivacea was the most frequently captured wild species in urban and rural areas and had the highest flea richness and the highest number of Rickettsia-positive fleas. This species has been described to have a "random walk" type of dispersal behavior, so it can easily go from wild to domestic environments [33]. These findings are important because these rodent species could act as "bridge hosts" and aid in the spread of the disease [32,34]. On the other hand, in natural areas, the rodent species most frequently captured was A. hirta; this species, like A. olivacea, had a high prevalence of Rickettsia-positive fleas. This rodent decreased its presence in areas with human intervention, which is consistent with the findings reported by Monteverde & Hadora [33], who described that this rodent preferably moves within the wild environment. Rodent populations can act as "source populations" and may be involved in the direct transmission of the pathogen to the target population [34]. The prevalence of Rickettsia spp. infections detected in our study was variable (0-35%), and associated with the identity of the flea species, season, type of locality and bioclimatic area. However, similar differences have been reported in other studies. For example, Radzijevskaja et al. [35] reported different prevalence related to the flea species analyzed (range: 0-43%). Also, Kuo et al. [36] carried out an extensive sampling analyzing the presence of Rickettsia in six species of flea, reporting 0-12.1% of prevalence in the different species of flea analyzed. Furthermore, flea infestations in this study were generally higher during the winter; however, this did not occur in all bioclimatic areas. Other studies have found similar results, attributing this variation to the differences in the seasonal reproductive cycles of the different species of flea [37], which are unknown in most of the species   found in this study. On the other hand, the higher prevalence of Rickettsia in fleas detected in natural areas can be explained by the greater diversity of species of micromammals and, therefore, of fleas. Thus, the differences in the prevalence of infection in the different species of flea, localities, seasons and bioclimatic zones found in our study, reveal the importance of the composition of the community, both fleas, and hosts, in determining the prevalence of Rickettsia in fleas, and therefore in the risk of infection in areas with different human disturbance.
In this study, we found two well-differentiated clades with a high degree of support. Clade R1 is formed by sequences obtained from fleas of the genus Neotyphloceras, collected from rodents Phyllotis darwini, A. olivacea, O. degus, R. rattus, and the marsupial T. elegans from central-north Chile (latitude: − 30° to − 31°S). This clade is related to R. bellii and is described as an ancestral group of Rickettsia [38], and which exhibits some specificity concerning its host [39]. This supports our results, where only bacteria detected in Neotyphloceras were found in this clade. Rickettsia bellii is endosymbiont of hard (Ixodidae) and soft (Argasidae) ticks throughout the American continent [39]. It has been classified as nonpathogenic for animals and humans [40], although seropositive samples have been found in dog blood in Brazil; however, the pathogenic effect is unknown [41]. Experimentally, this bacterium grows easily in mammalian cells. In experimental inoculations in guinea pig and rabbit, it produces, depending on the inoculated dose received, from a mild inflammatory reaction to necrotic scabs a typical symptomatology of other pathogenic rickettsiae [29]. Furthermore, it is capable of producing antibodies in experimental infections in the big-eared opossum Didelphis aurita, but without rickettsemia [42]. These results indicate that some flea species present in wild and synanthropic micromammals could carry a new ancestral genotype of Rickettsia, just like those reported by Song et al. [43] in China from fleas of wild rodents.
The R2 clade was divided into two large groups, R2a and R2b. R2a grouped all of the sequences detected in fleas being extracted from two species of flea, S. ares (Stephanocircidae) and T. rhombus (Rhopalopsyllidae), which were obtained from villages and natural environments through wide latitudinal distribution (latitude of − 35° to − 45°S). This corresponds to the wide distribution of the hosts of infected fleas (A. hirta and A. olivacea). Conversely, R2b was formed by sequences obtained from C. allophyla and C. inopinata belonging to the same family (Hystricopsylidae); both species of flea were collected in wild rodents (A. hirta and A. olivacea) from wild areas (Los Queules NR and Nonguén NR) in the south-central zone of Chile. These sequences are closely related to R. hoogstraalii, R. asembonensis and R. felis, all of which are members of the spotted fever group rickettsiae (SFG) [28,29,38]. The SFG consists of > 30 species that can be found worldwide, most of them with pathogenic effects on humans [44]. Our analysis showed a close relationship with R. hoogstraalii, a widely distributed bacterium that is still unknown for its pathogenicity in humans. This bacterium has been detected in both hard ticks (H. punctata, H. sulcate and H. parva) and soft ticks (Ornithodoros moubata, Carios capensis, C. sawaii and Argas persicus) present in domestic animals, bird nests, vegetation, and human dwellings [3,[45][46][47]. A similar situation occurs with R. asembonensis. It also has a wide distribution worldwide, having been reported in North America and South America, Asia, the Middle East and Europe [48], although it is associated with a greater number of ectoparasites, including fleas, ticks, and mites of domestic and peridomestic animals (C. canis, C. felis, X. cheopis, Pulex irritans, Amblyomma ovale, Rhipicephalus sanguineus, R. microplus and Ornithonysus bacoti) [49][50][51][52][53]. It has also been detected in monkey blood in Malaysia [54] and in dog blood in South Africa [55]. Although these bacteria live in parasitic arthropods close to humans and are closely associated with R. felis, there is no evidence yet of possible infection or pathogenicity [48]. On the other hand, R. felis is an emergent, widely distributed, flea-borne human pathogen, and like R. asembonensis and R. hoogstraalii, is associated with domestic and peridomestic animals and their ectoparasites [56,57]. The main vector is C. felis, although mosquitoes (Anopheles gambiae) have also been detected as competent vectors [58]. Unlike R. asembonensis and R. hoogstraalii, this bacterium is of known pathogenicity causing fever, fatigue, nausea, muscle aches, back pain, headaches, macular rash, joint pain and eschar [49]. Although the BLAST analysis showed a low percentage of similarity with R. felis (sca5 94%), the phylogenetic analysis shows a close relationship with Rickettsia detected in C. allophyla in south-central Chile. Until now, in Chile, only R. felis has been registered in C. felis [12].

Conclusions
To the best of our knowledge, our study reports, for the first time in Chile, the presence of Rickettsia in different species of parasitic fleas of wild micromammals and invasive rodents found in both natural and human environments. Moreover, there is evidence of at least two clades of Rickettsia associated with fleas. These data increase the knowledge of possible Rickettsia vectors/reservoirs in Chile. However, greater efforts should be made to monitor and determine the degree of pathogenicity of the detected rickettsiae.