Novel Rickettsia genotypes in ticks in French Guiana, South America

Rickettsia are obligate intracellular bacteria often associated with ticks and best known for causing human diseases (rickettsiosis), including typhus fever and sporadic cases of serious infection. In this study, we conducted a large survey of ticks in French Guiana to understand the overall diversity of Rickettsia in this remote area largely covered by dense rainforests. Out of 819 individuals (22 tick species in six genera), 252 (30.8%) samples were positive for Rickettsia infection. Multilocus typing and phylogenetic analysis identified 19 Rickettsia genotypes, but none was 100% identical to already known Rickettsia species or strains. Among these 19 genotypes, we identified two validated Rickettsia species, Rickettsia amblyommatis (spotted fever group) and Rickettsia bellii (bellii group), and characterized a novel and divergent Rickettsia phylogenetic group, the guiana group. While some tick hosts of these Rickettsia genotypes are among the most common ticks to bite humans in French Guiana, their potential pathogenicity remains entirely unknown. However, we found a strong association between Rickettsia genotypes and their host tick species, suggesting that most of these Rickettsia genotypes may be nonpathogenic forms maintained through transovarial transmission.

Diversity of Rickettsia. The diversity of Rickettsia in French Guiana ticks was examined using sequences from one to five genetic markers (gltA, 16S rRNA, atpA, ompB and coxA). Overall, the examination of this multilocus data set led to the identification of 19 distinct Rickettsia genotypes (FG019a-FG019s hereafter; Table 2), as detailed below.
First, Rickettsia sequences from the gltA gene were taken from a subsample of 92 infected specimens from the 12 infected species (one to 47 specimens per infected species were examined; Table 2). On the basis of DNA sequencing, 12 distinct gltA genotypes with 84.9-99.8% pairwise nucleotide identity were characterized from the 92 specimens examined. Six tick species of the 12 infected species harbored each only one gltA genotype: A. cajennense, A. coelebs, A. longirostre, A. geayi, A. naponense and I. luciae. In each of the six other tick species, two to four distinct gltA genotypes were found. We characterized four gltA genotypes from the seven sequenced A. dissimile specimens (Table 2).
Second, we amplified four additional bacterial markers (16S rRNA, atpA, ompB and coxA) from 44 representative tick samples infected by the 12 Rickettsia gltA genotypes ( Table 2). We then obtained 10 genotypes of 16S rRNA (97.7-99.9% pairwise nucleotide identity), 10 atpA genotypes (85.4-99.8%), 10 ompB genotypes (78.1-99.8%) and eight coxA genotypes (86.1-99.8%). While the 16S rRNA and atpA gene fragments were amplified from the 44 samples, the ompB and coxA were only amplified from 41 and 38 samples, respectively ( Table 2). The diversity at the 16S rRNA, atpA, ompB and coxA gene fragments was consistent with the results inferred from the gltA sequences: Rickettsia infections with distinct gltA sequences have distinct sequences at the other gene markers. No 16S rRNA, atpA, ompB and coxA sequence variation was observed within tick species in which only one gltA genotype was detected (i.e., A. cajennense, A. coelebs, A. longirostre, A. geayi, A. naponense and I. luciae). However, the combined use of these five markers allowed the distinction of additional Rickettsia genetic variation not detected with the single gltA gene sequences. Indeed, while one of the Rickettsia infections of A. dissimile and one of A. humerale shared the same gltA sequence (the gltA sequence type #d in Table 2), their atpA, ompB and coxA (but not 16S rRNA) gene sequences were different, showing that they were thus two distinct Rickettsia genotypes. The examination of gltA, 16S rRNA, atpA, ompB and coxA gene sequences thus led to the identification of 19 Rickettsia genotypes (FG019a-FG019s; Table 2). Only one of these Rickettsia genotypes, FG019c, was shared by several tick species (A. dissimile, A. geayi and A. latepunctatum). Each of the 18 other Rickettsia genotypes (FG019a, FG019b and FG019d-FG019s) was found in only one tick species ( Table 2). phylogeny of Rickettsia. The phylogenetic relationships between the Rickettsia infections were first estimated using the 92 gltA sequences from the 12 infected tick species found in this study, as well as gltA sequences www.nature.com/scientificreports www.nature.com/scientificreports/ from representative Rickettsia species and strains available in GenBank (Fig. 2). The closest relatives of the Rickettsia found in French Guiana were also included in the analyses. No recombination events were detected for the gltA data set using both the RDP and GENCONV methods (all p > 0.23). The ML phylogenetic analysis based on the gltA sequences showed that the 19 Rickettsia genotypes (FG019a-FG019s) found in this study consisted of three distinct groups (Fig. 2 (Fig. 2). It is noteworthy that this strain is also distantly related to the Rickettsia sp. clone Tapirape1 (canadensis group; Fig. 2), which was previously found in A. naponense from Brazil 19 . www.nature.com/scientificreports www.nature.com/scientificreports/ However, none of the Rickettsia genotypes found in this study is closely related to the single species already reported from French Guiana, Candidatus Rickettsia wissemanii (Fig. 2).
A second analysis was performed to refine the intrageneric phylogeny of Rickettsia. For this, we used the Rickettsia 16S rRNA, atpA, ompB and coxA sequences from the 19 Rickettsia genotypes identified in the present work, as well as sequences of representative Rickettsia species and strains available in GenBank. The analysis of single and concatenated gene sequences did not detect significant recombination events in the data set using both RDP and GENCONV methods (all p > 0.08). When the sequences were examined separately for each gene, we obtained the same phylogenetic pattern as observed with the ML analysis based on gltA gene sequences with the partitioning of the 19 Rickettsia genotypes into the same three different groups (i.e., spotted fever, bellii and guiana) (Figs. S1-S4). Indeed, the examination of the16S rRNA and atpA gene sequences of Rickettsia FG019n genotype of A. naponense (neither ompB nor coxA could be amplified from this Rickettsia strain; see Table 2) corroborated the existence of the guiana group: the 16S rRNA, atpA and gltA single-gene phylogenies (Figs. 2, S1 and S2) and the gltA, 16S rRNA and atpA concatenated phylogeny (Fig. 3) showed that the FG019n genotype is highly divergent from all other known Rickettsia groups, species and strains.
Analyses of a multilocus data set (based on the 16S rRNA, gltA and atpA genes) further showed that the nine Rickettsia genotypes (FG019c-g, FG019i, FG019o, FG019q and FG019s) belonging to the bellii group always clustered together with the R. bellii strains previously found in other American tick species, such as A. neumanni (Argentina) and D. variabilis (USA) (Figs. 3 and S1-4). These nine Rickettsia genotypes can therefore be considered as members of the R. bellii species. None of these nine Rickettsia genotypes were 100% identical to already known R. bellii members.
The multilocus data set also showed that the nine Rickettsia genotypes (FG019a, FG019b, FG019h, FG019j-m, FG019p and FG019r) belonging the spotted fever group can be split into two subgroups: , which all clustered with R. amblyommatis on the basis of multilocus analyses (Fig. 3). However, only FG019a and FG019m  www.nature.com/scientificreports www.nature.com/scientificreports/ consistently clustered with R. amblyommatis at each gene marker (Figs. 2 and S1-4), showing that these two genotypes can be considered as members of the R. amblyommatis species. None of these two Rickettsia genotypes were 100% identical to already known R. amblyommatis members, however. The four other Rickettsia genotypes (FG019b, FG019h, FG019l and FG019p) cannot be classified into specific species due to a lack of consensus between the phylogenetic trees (Figs. 2, 3 and S1-4). Indeed, while the 16S rDNA genotype of FG019b is more closely related to R. amblyommatis (Fig. S1), its atpA genotype is more closely related to R. montanensis (Fig. S2). Although these last four Rickettsia genotypes may each represent a novel species, additional gene sequencing is required to determine their precise phylogenetic proximity with other members of the spotted fever group.

Discussion
We found here that Rickettsia infections are common in French Guiana ticks, a pattern also observed among ticks of other South American regions 7,10-13,20 . The incidence of infection varied between tick species of French Guiana: 12 of the 22 tick species examined, including Amblyomma, Haemaphysalis and Ixodes species, harbored Rickettsia, and when present, prevalence ranged from 10 to 100%, with significant variations between sampling localities of some species. We further found that these Rickettsia infections are remarkably diverse. Multilocus strain typing revealed the presence of 19 Rickettsia genotypes of different phylogenetic origins. Of these 19 genotypes, 18 were found each in only one tick species. Remarkably, more than one Rickettsia genotype was found within half of the infected tick species, meaning that this intraspecific variation of infection is common in tick species of French Guiana. The single Rickettsia species known from this region, Candidatus Rickettsia wissemanii 9 , was not detected in the present study. Altogether, this means that at least 20 different Rickettsia genotypes are circulating in ticks in French Guiana.
None of the 19 Rickettsia genotypes we identified in French Guiana had been documented before this study. Multilocus typing showed that 11 of these Rickettsia genotypes can be assigned to two validated Rickettsia species, namely R. amblyommatis (two genotypes) and R. bellii (nine genotypes). These two Rickettsia species are widely present among Central and South American ticks, each infecting more than 10 species 7,11-13,21-26 . Their presence in French Guiana was therefore expected, but the observation of novel genotypes indicates the presence of important geographic variability: R. amblyommatis and R. bellii have probably radiated within their respective regions, including French Guiana, into different genotypes. Overall, this confirms that R. amblyommatis and R. bellii have the widest host range and the broadest geographic distribution among all Rickettsia species reported from South America, as suggested in early studies 7,25 . Besides the R. amblyommatis and R. bellii genotypes, the eight other Rickettsia genotypes are rarer, since they are apparently endemic to French Guiana and cannot be www.nature.com/scientificreports www.nature.com/scientificreports/ assigned to formerly validated species. While a few genotypes remain unclassified within the spotted fever group, we described one novel Rickettsia genotype, which belongs to a novel and divergent group, the guiana group. It is noteworthy that the guiana group has an intermediate phylogenetic position between the spotted fever and bellii Rickettsia groups, since it is more related on the basis of its gltA sequence to the rare species R. mendelii, which was found only in Europe 27,28 .
Most of human pathogenic Rickettsia species are vectored by hard ticks 4,29 . This leads to the question of the pathogenicity of the 19 Rickettsia genotypes we found in French Guiana and the associated health risk. Since none of the 19 Rickettsia genotypes was previously described before this study, no evidence of their pathogenicity currently exists, even for those belonging to validated Rickettsia species. Indeed, although R. amblyommatis and R. bellii are commonly found in the ticks of French Guiana, and more broadly in American ticks 7,[11][12][13][21][22][23][24][25][26] , they have never been found in vertebrate hosts, suggesting that they are nonpathogenicspecies. Interestingly, while the Cayenne tick A. cajennense is one of the most common ticks found in French Guiana, blood-feeding on many different hosts, including humans 8,30,31 , R. amblyommatis (infecting here 25% of the A. cajennense specimens examined) were never detected in humans or animals: while French Guiana is an outermost region of the European Union, with technical and financial resources that close to European countries, no case was notified to date. Another intriguing point is the apparent specificity of Rickettsia genotypes to tick species: 18 out of 19 Rickettsia genotypes were detected in only a single tick species. Even generalist tick species, such as A. cajennense and A. dissimile, feeding on (and sharing) a variety of vertebrate hosts 8,30,31 , did not share the same Rickettsia genotypes. These observations may indicate that at least some of the Rickettsia genotypes in French Guiana are present in ticks but not in vertebrate hosts.
The persistence means of the 19 Rickettsia genotypes remain unknown in French Guiana. As pointed out in a recent study 32 , the current view in rickettsiology has a strong anthropocentric bias and tends to describe all novel Rickettsia species as pathogenic forms. However, most of the novel Rickettsia species or strains discovered in recent years are also found exclusively in arthropods and never in vertebrates [1][2][3]33,34 . In ticks, as for many other arthropods, some Rickettsia are maternally inherited endosymbionts with poorly known effects on tick biology. This is the case for R. buchneri in the black-legged tick I. scapularis 35 , R. peacockii in the American dog tick D. variabilis 36 , and R. vini in the tree-hole tick I. arboricola 16,37 . These nonpathogenic Rickettsia may interact with a variety of tick-borne pathogens 34 , including Anaplasma marginale 38 , Borrelia burgdorferi 39 and also other Rickettsia 40,41 . Indeed, the endosymbiont R. peacockii may possibly hamper the multiplication of the spotted fever agent, R. rickettsii 40 , and may also block transovarial transmission colonization of R. rickettsii, R. montana and R. rhipicephali 40,41 . In French Guiana, further studies are needed to test this hypothesis of endosymbiosis by observing transstadial and transovarial transmission in ticks.
To conclude, this study revealed substantial diversity of Rickettsia, including novel genotypes, specie and group, in ticks in French Guiana. This underlines the need to better document Rickettsia diversity in diverse regions, and more especially in remote regions. A recent meta-analysis suggests that more than 20% of terrestrial arthropods may be infected by Rickettsia, with ticks hosting most of this bacterial diversity 29 , as observed in this study. In many arthropods other than ticks, Rickettsia are nonpathogenic, undergo exclusive maternal transmission to offspring, and may function as both a mutualist and reproductive manipulator 2,3,42,43 . Overall, adaptations of Rickettsia to this diversity of hosts encompass an array of parasitic, but also mutualistic, interactions 1-3 . In French Guiana, the effect of the19 Rickettsia genotypes on human and animal health as well as on tick physiology and reproduction remains to be elucidated.  (Tables 1 and S1). Questing ticks were collected from the vegetation using a drag-flag method over sites covering three types of ecological conditions (periurban, agricultural and natural). Ticks were also directly collected in nests or on hosts (including humans, four domestic animal species and wild animal species; see Table S1). All ticks were stored in 75% ethanol until examination. For each tick specimen, species were formally identified through morphological examination (using dichotomous keys 30,44 ) and DNA sequencing in a previous study 8 .

Detection of Rickettsia.
To avoid external bacterial contaminants, ticks were processed with commercial bleach diluted at 1% for 30 s and then rinsed for 1 min in three successive baths of DNA-free water following a published protocol 45 . For each tick specimen, total DNA was further extracted from whole body using a genomic DNA extraction kit according to the manufacturer's instructions (DNeasy Blood & Tissue extraction kit, Qiagen). The presence of Rickettsia within each DNA template was investigated through high-throughput 16S rDNA sequencing. To this aim, a 251-bp portion of the V4 variable region of the bacterial 16S rDNA was amplified from whole-body DNA samples using the universal forward and reverse primers listed in Table S4. Each PCR product from individual samples was tagged with a unique 35-base barcode using the Nextera Index Kit (Illumina, San Diego, CA, USA). PCR amplifications were performed in duplicates for each sample. PCR reactions were conducted using a Multiplex PCR Kit (Qiagen). Amplified bacterial 16S rDNA products were purified and sequenced on an Illumina MiSeq platform (GenSeq, Montpellier University) and 250-bp end sequence reads were obtained. All bioinformatic analyses were conducted using the pipeline Frogs (https://github.com/geraldinepascal/FROGS) as follows 46 : primers were removed from paired-end sequences with Cutadapt 47 , and these sequences were merged into contigs with FLASH 48 before filtering by length (251 bp ± 10 bp). Chimaeras were removed with VSEARCH 49 , then sequences were clustered using SWARM 50 . We obtained an average number of 29,206 bacterial 16S rDNA reads per tick specimen. Sequences with 97% similarity were clustered together and identified as an operational taxonomic unit (OTU). Each representative OTU sequence was aligned and taxonomically assigned using the Silva database (https://www.arb-silva.de/). To eliminate the possibility of contamination, we included Scientific RepoRtS | (2020) 10:2537 | https://doi.org/10.1038/s41598-020-59488-0 www.nature.com/scientificreports www.nature.com/scientificreports/ four mock DNA extractions under identical conditions using water, buffers and kits utilized for the experimental samples followed by Illumina Miseq analysis of 16S rDNA reads. The negative controls provided only a handful of reads that did not correspond to the bacterial genera found in the tick samples.

Molecular typing of Rickettsia.
A random subset of DNA templates for which Rickettsia reads were obtained through high-throughput 16S rDNA sequencing were used for Rickettsia multilocus typing. These Rickettsia infections were genotyped using independent PCR assays based on gltA, coxA, ompB, atpA and 16S rRNA, using semi-nested or nested PCR assays (Table S2). To prevent possible contamination, different parts of this process were physically separated from one another, in entirely separate rooms. All amplicons were also sequenced to control for false-positive amplifications. Gene features, primers and PCR conditions are detailed in Table S2.
Seminested and nested PCR amplifications were performed as follows: the first PCR run with the external primers was performed in a 10-μL volume containing approximately 20 ng of genomic DNA, 3 mM of each dNTP (Thermo Scientific), 8 mM of MgCl 2 (Roche Diagnostics), 3 μM of each primer, 1 μL of 10× PCR buffer (Roche Diagnostics) and 0.5 U of Taq DNA polymerase (Roche Diagnostics). A 1-μL aliquot of the PCR product from the first reaction was then used as a template for the second round of amplification. The second PCR was performed in a total volume of 25 μL and contained 8 mM of each dNTP (Thermo Scientific), 10 mM of MgCl 2 (ThermoScientific), 7.5 μM of each of the internal primers, 2.5 μL of 10× PCR buffer (Thermo Scientific) and 1.25 U of Taq DNA polymerase (Thermo Scientific). All PCR amplifications were performed under the following conditions: initial denaturation at 93 °C for 3 min, 35 cycles of denaturation (93 °C, 30 s), annealing (Tm = 52-56 °C, depending on primers, 30 s), extension (72 °C, 1 min), and a final extension at 72 °C for 5 min. Known positive and negative individuals were used as controls in each PCR assay. All PCR products were visualized through electrophoresis in a 1.5% agarose gel. Positive PCR products were purified and sequenced in both directions (EUROFINS). The chromatograms were manually inspected and cleaned with CHROMAS LITE (http://www.technelysium.com.au/chromas_lite.html) and sequence alignments were done using CLUSTALW 51 , both implemented in MEGA7. Genotype naming (ie,) was based on the following rationale: the genotype Molecular phylogenetics. The GBLOCKS 52 program with default parameters was used to remove poorly aligned positions and to obtain unambiguous sequence alignments. All sequence alignments were also checked for putative recombinant regions using the RDP3 computer analysis package 53 . Given a set of aligned nucleotide sequences, RDP3 can rapidly analyze these with a range of powerful nonparametric recombination detection methods, including the GENECON 54 and RDP 55 . Phylogenetic relationships were evaluated between Rickettsia strains using gltA, coxA, ompB, atpA and 16S rRNA gene sequences. The evolutionary models most closely fitting the sequence data were determined using Akaike information criterion with the MEGA7 program 56 . Phylogenetic analyses were based on maximum likelihood (ML) analyses. A ML heuristic search, using a starting tree obtained by neighbor-joining, was conducted, and clade robustness was further assessed by bootstrap analysis using 1000 replicates in MEGA7 56 . ethics approval. The use of the genetic resources was declared to the French Ministry of the Environment under reference TREL19028117S/156 and #150401230100, in compliance with the Access and Benefit Sharing procedure implemented by the Loi pour la Reconquête de la Biodiversité. The capture of ticks in the Grand Connétable protected area was authorized by the Prefecture of French Guiana by prefectoral decree R03-2016-09-23-003. All animals were handled in strict accordance with good animal practices as defined by the French code of practice for the care and use of animals for scientific purposes, established by articles R214-87 to R214-137 of the French rural code.

Data availability
Nucleotide sequences of Rickettsia were deposited in the GenBank nucleotide database (Accession numbers: