Diversity of spotted fever group rickettsiae and their association with host ticks in Japan

Spotted fever group (SFG) rickettsiae are obligate intracellular Gram-negative bacteria mainly associated with ticks. In Japan, several hundred cases of Japanese spotted fever, caused by Rickettsia japonica, are reported annually. Other Rickettsia species are also known to exist in ixodid ticks; however, their phylogenetic position and pathogenic potential are poorly understood. We conducted a nationwide cross-sectional survey on questing ticks to understand the overall diversity of SFG rickettsiae in Japan. Out of 2,189 individuals (19 tick species in 4 genera), 373 (17.0%) samples were positive for Rickettsia spp. as ascertained by real-time PCR amplification of the citrate synthase gene (gltA). Conventional PCR and sequencing analyses of gltA indicated the presence of 15 different genotypes of SFG rickettsiae. Based on the analysis of five additional genes, we characterised five Rickettsia species; R. asiatica, R. helvetica, R. monacensis (formerly reported as Rickettsia sp. In56 in Japan), R. tamurae, and Candidatus R. tarasevichiae and several unclassified SFG rickettsiae. We also found a strong association between rickettsial genotypes and their host tick species, while there was little association between rickettsial genotypes and their geographical origins. These observations suggested that most of the SFG rickettsiae have a limited host range and are maintained in certain tick species in the natural environment.

lice, fleas, and mites. TG is composed of Rickettsia typhi and R. prowazekii, while TRG is composed of R. akari, R. australis, and R. felis. Among the tick-borne rickettsiae, AG includes R. bellii and R. canadensis. More than 25 species of tick-borne rickettsiae that have been validated so far belong to SFG. Furthermore, the members of SFG rickettsiae have been increasing as many new species have been proposed recently [3][4][5][6][7] .
In Japan, R. japonica was the first SFG Rickettsia discovered in 1984 as the causative agent of Japanese spotted fever (JSF) 8,9 . Since then, several other SFG rickettsiae, namely R. heilongjiangensis, R. helvetica, and R. tamurae have been recognised as etiological agents of human diseases [10][11][12] . SFG rickettsiae with unknown pathogenicity, such as R. asiatica and Candidatus R. tarasevichiae, have also been reported 13,14 . In addition, several studies conducted in Japan have documented the presence of other Rickettsia species/genotypes in animals and questing ticks [15][16][17] . However, in most cases, only single or a limited number of genes have been analysed, making it difficult to generate an overview of the genetic diversity of SFG rickettsiae, since multiple gene sequencing are recommended in the classification of rickettsial isolates 18 .
The relationship between SFG rickettsiae and their vector tick species has been studied previously. It is evident that some SFG rickettsiae, such as R. rickettsii, are associated with several different tick species in different genera, while others, such as R. conorii, are linked to specific tick species 19 . In Japan, R. japonica is considered to be in the former group since it has been recorded from wide range of tick species including Dermacentor taiwanensis, Haemaphysalis hystricis, H. cornigera, H. longicornis, H. flava, H. formosensis, H. megaspinosa, and Ixodes ovatus 20 . On the other hand, vector tick species of other rickettsiae, such as R. asiatica and R. heilongjiangensis, which are respectively transmitted by I. ovatus and H. concinna, seem to be limited 11,13 .
The aim of the present study was to understand the overall diversity of SFG rickettsiae and their vector tick species in Japan. By collecting questing ticks at more than 100 different sampling sites across Japan, a nationwide cross-sectional study for SFG rickettsiae was conducted. The samples included 19 different tick species covering most of the commonly found species in Japan. Our results indicate that there exist more SFG rickettsiae genotypes than previously known. The information on the relationship between SFG rickettsiae and vector ticks is useful for further characterisation of each rickettsiael member in more detail.

Results
Detection of sFG rickettsiae by real-time pCR for gltA. Out  successfully sequenced, which resulted in 15 different gltA genotypes ( Fig. 1 and Table 1). In the present study, the gltA genotype is defined as a gltA sequence type that is different from the others even by a single nucleotide.

Multiple genes sequencing.
To further characterise Rickettsia spp. based on five other genes; outer membrane protein A gene (ompA), outer membrane protein B gene (ompB), 17-kDa common antigen gene (htrA), surface cell antigen-4 (sca4), and 16S ribosomal RNA gene (16S rRNA), PCR analyses were conducted on the selected samples of each gltA genotype. A total of 57 samples were employed in this analysis. We selected more than two samples from each genotype except for G7 which was found in only one sample ( Table 3). The samples with higher rickettsial burden were selected based on the results of gltA real-time PCR. The mean rickettsial burden in the template DNA ranged from 2.3E + 2 to 2.1E + 4 copies/μL ( Table 3). The htrA gene was successfully amplified and sequenced for all gltA genotypes. Although 16S rRNA PCR gave amplicons in all gltA genotypes, the following sequencing analysis revealed that rickettsial 16S rRNA gene sequences were obtained in only 12 gltA genotypes. The ompB, ompA and sca4 genes were amplified and sequenced in 11, five and six different gltA genotypes, respectively. All genes were successfully sequenced in two gltA genotypes (G6 and G7). Four genes were successfully amplified in six gltA genotypes (G1, G2, G5, G8, G10, and G11), and three genes were amplified in four gltA genotypes (G3, G4, G9, and G13). Only the htrA gene was amplified in three gltA genotypes (G12, G14, and G15) ( Table 3). The sequencing analysis of the amplified products revealed that there were no sequence differences in any of the genes in the samples with the same gltA genotypes. The sequence types obtained from each gltA genotype were different from each other. The sequence identity with the closest Rickettsia species are provided in Supplementary Table S2.
Species classification of SFG rickettsiae. Phylogenetic trees inferred from ompA, ompB, sca4, htrA, and 16S rRNA analysis are shown in Fig. 2. G4 and G5 formed a distinct cluster with R. helvetica in all trees when sequences were available and thus were identified as R. helvetica. Being supported by more than three trees, G13, G10, G8, and G3 were identified as R. asiatica, R. monacensis (former Rickettsia sp. In56), and R. tamurae, and Candidatus R. tarasevichiae, respectively (Fig. 2). The other nine gltA genotypes could not be classified into specific species due to a lack of consensus between the trees and/or absence of sequences from previously validated rickettsial species in the same phylogenetic clusters.

Discussion
The present study included a total of 2,189 individual ticks collected at 114 different sampling sites in six regions of Japan for the screening of SFG rickettsiae. Our nationwide sampling enabled us to collect as many as 19   (Kagoshima, Nagasaki, and Okinawa prefectures) parts of Japan, were positive for SFG rickettsiae 16 . Another nationwide survey conducted in 5 prefectures (Chiba, Hokkaido, Kochi, Tokushima, and Toyama prefectures) including JSF endemic areas reported an overall positive rate for SFG rickettsiae to be 25.8% (186 out of 722) in 10 different tick species 21 .
We determined partial sequences of the gltA gene of SFG rickettsiae by conventional PCR, which was previously designed to characterise SFG rickettsiae in Japan 16 . Based on the sequences of the gltA gene obtained from 352 ticks, the SFG rickettsiae detected in the present study were provisionally divided into 15 genotypes. In the molecular classification of SFG rickettsiae, the analysis of multiple genes commonly used by other researchers is a prerequisite 18 . Therefore, we attempted to obtain the sequences of five additional genes, ompA, ompB, htrA, sca4, and 16S rRNA. These efforts lead to the identification of four validated rickettsial species, namely R. asiatica, R. helvetica, R. monacensis, and R. tamurae, and the provisional species Candidatus R. tarasevichiae (Fig. 2).
Prior to this study, there was no official report of the presence of R. monacensis in Japan. A recent study indicated that Rickettsia sp. In56, a rickettsial strain reported from ticks in Japan 21 , might be a synonymous of R. monacensis 22 . Although several isolates of Rickettsia sp. In56 have been obtained from Japanese ticks 23 , lack of their sequence information prevents a direct comparison between Rickettsia sp. In56 and R. monacensis reported elsewhere. Nevertheless, the sequence analysis of multiple genes (gltA, ompA, ompB, htrA, and 16S rRNA) conducted in the present study confirmed the presence of R. monacensis in Japan (Figs 1 and 2). R. monacensis was initially isolated from I. ricinus collected from the English Garden in Germany using ISE6 cells 24 and has been detected from the same tick species in Europe and neighbouring countries [25][26][27][28][29] . I. nipponensis and I. sinensis are considered as main vectors of R. monacensis in China and Korea, respectively 30,31 . In our study, R. monacensis was detected from four I. nipponensis samples collected in the Tohoku and Kansai regions, while none of the other tick species carried R. monacensis ( Fig. 1 and Table 2). These results may suggest the relatively wide distribution of R. monacensis and a strong association of R. monacensis with I. nipponensis in Japan. This SFG rickettsiae caused Mediterranean spotted fever-like symptoms in humans in several countries 32,33 . More recently, the agent was isolated from the blood of a patient with an acute febrile illness in Korea 22 . Thus, clinicians should be aware of R. monacensis as a possible cause of non-JSF rickettsiosis in Japan.
Although we tried to characterise SFG rickettsiae with each prospective gltA genotype in further detail by sequencing five additional rickettsial genes, ompA, ompB, htrA, sca4, and 16S rRNA, the amplification was not successful for some genes ( Table 3). The ompA and sca4 genes were amplified only from one third of the tested gltA genotypes. Considering the relatively high rickettsial abundance in the tested samples (Table 3), PCR failure is either because some of the SFG rickettsiae lack these genes as shown in TG rickettsiae that do not possess ompA gene 34 , or because there are nucleotide mismatches in the primer annealing sites. PCR failures of variable genes such as ompA, ompB, and sca4 are common issues in the genetic characterisation of SFG rickettsiae 34,35 . Thus further attempts including the development of universal primers and/or bacterial isolation followed by whole genome sequencing are required to determine the phylogenetic positions of uncharacterised Rickettsia spp.
In a previous nationwide survey of SFG rickettsiae conducted in Japan, Gaowa et al. 16 classified the detected rickettsiae (n = 181) into five groups (groups 1-5) based on the gltA sequences 16 . Groups 1 and 2 were respectively identified as R. japonica and R. tamurae, whereas groups 3, 4, and 5, showing high sequence similarity with Rickettsia sp. LON-13, R. raoultii, and Candidatus R. principis, respectively, were not classified as validated rickettsial species 16 . In agreement with their report, we detected gltA sequences corresponding to groups 3 (G6),  4 (G2), and 5 (G1, G11, G12, G14, and G15) (Fig. 2). Unfortunately, limited information is available about these uncharacterised Rickettsia spp. In our study, G6 and G2 were respectively detected in H. longicornis and H. hystricis with high infection rates (62.8% and 57.8%, respectively) ( Table 1), warranting further studies on the effect of these infections for the survival and reproductive fitness of their hosts. We detected two gltA genotypes (G7 and G9), which were allocated into distinct clusters from Rickettsia spp. previously reported from Japan (Fig. 2). G7 and G9 showed the highest gltA sequence similarity with Rickettsia spp. reported from Kenya (KT257873) and Hungary (EU853834), respectively. Rickettsia sp. reported from Kenya was detected in Rhipicephalus maculatus 36 , while the one from Hungary was detected in H. inermis and provisionally named as Candidatus R. hungarica 37 . Since the sequences of other genes were not available from those Rickettsia spp., it was difficult to evaluate the degree of genetic relatedness in more detail. Nonetheless, the presence of closely related species in two geographically remote areas may indicate the worldwide distribution of these poorly characterised SFG rickettsiae. Since the present study provided the sequences of multiple genes of those rickettsiae, the information is useful in the classification of SFG rickettsiae.
In the present study, we found a strong association between rickettsial genotypes and their host tick species; 13 out of 15 gltA genotypes were detected in only one single tick species (Fig. 1 and Table 3). Furthermore, there was minimal geographical restriction for the 11 gltA genotypes that were recovered from multiple geographical regions (Table 2). These observations may indicate that most of the SFG rickettsiae species are found in ticks but not in vertebrate hosts in the natural environment. However, further examinations are needed to confirm this hypothesis by observing transstadial and transovarial transmission of these SFG rickettsiae in ticks. The effect of these rickettsial infections on tick physiology and reproduction remains to be elucidated.
Although the sampling was conducted at several JSF-endemic areas in Mie, Kagoshima, and Kumamoto prefectures, none of the ticks were infected with R. japonica. Considering the low level of genomic plasticity within R. japonica isolates 38 , it was hardly expected that a real-time PCR assay for gltA would result in false-negatives. The positive rate of R. japonica infection in the questing ticks was as low as 0.86% (18 out of 2,099), even in endemic areas as is the case in Shimane prefecture 39 . Collectively, the failure in detection of R. japonica might be partly attributed to the sample selection procedure with which only a maximum of 10 individual ticks per species, stage/ sex, and site were tested for SFG rickettsiae infection. Therefore, it should be noted that the present study might not fully disclose the diversity of SFG rickettsiae in Japan, which warrants further investigations by employing a larger number of samples.
The present study has extended our knowledge on the diversity of SFG rickettsiae prevalent in Japan. In addition to previously recognised rickettsial species such as R. asiatica, R. helvetica, R. monacensis (formerly reported as Rickettsia sp. In56), R. tamurae, and Candidatus R. tarasevichiae, several uncharacterised Rickettsia spp. including ones showing high similarities with those designated as novel Rickettsia spp. detected in geographically remote countries such as Kenya and Hungary were discovered. A strong association between rickettsial genotypes and their host ticks suggests a long-term relationship between SFG rickettsiae and ticks. Further investigations on the potential roles of these SFG rickettsiae on ticks are required to understand the mechanisms underlying widespread existence of genetically variable rickettsiae in ticks.  (Fig. 3). All field-collected ticks were transferred to small Petri dishes and preserved in an incubator at 16 °C until use.

Materials and
Tick species identification. Tick species were identified morphologically using standard keys under a stereomicroscope 40,41 . When more than 10 ticks with the same species and stage/sex were collected from the same sampling sites, a maximum of 10 individual ticks were analysed per species, stage/sex and site. A total of 2,189 individuals (103 nymphs and 2,086 adults) in four genera were examined in this study. These included one species in the genus Amblyomma (A. testudinarium, n = 85), one species in the genus Dermacentor (D. taiwanensis, n = 12), 10  Real-time pCR. All samples were first screened for rickettsial gltA using real-time PCR to detect SFG and TG rickettsiae as described previously 43 . The primers and probes used are shown in Table 4. Reactions were performed in a 20.0 μL of reaction mixture containing 10.0 μL of THUNDERBIRD Probe qPCR Mix (Toyobo, Osaka, Japan), 300 nM of each primer, 200 nM of probe, 5.0 μL of template DNA, and distilled water. The reaction was carried out in a C1000 Thermal Cycler with a CFX96 Real-Time PCR Detection System (BioRad Laboratories, Hercules, CA) at conditions of 50 °C for 3 min, 95 °C for 1 min, and 40 cycles of 95 °C for 15 sec and 60 °C for 1 min. Each run included a blank control and serially diluted plasmid standards (10 6 , 10 4 , and 10 2 copies/reaction) as described previously 35     sequencing. The amplified PCR products were purified using a Wizard ® SV Gel and PCR Clean-Up System Kit (Promega, USA). Sanger sequencing was conducted using the BigDye Terminator version 3.1 Cycle Sequencing Kit (Applied Biosystems, Foster City, CA) and an ABI Prism3130x genetic analyser according to the manufacturers' instructions. The sequences data were assembled using ATGC software version 6.0.4 (GENETYX, Tokyo, Japan). The sequences obtained were submitted to the DNA Data Bank of Japan (DDBJ) (http://www. ddbj.nig.ac.jp) under accession numbers (gltA: LC379427-LC379443; ompA: LC379461-LC379465; ompB: LC379466-LC379476; htrA: LC379444-LC379460; sca4: LC379477-LC379482; 16S rRNA: LC379483-LC379494).
phylogenetic analysis. The nucleotide sequences obtained were aligned with representative sequences of known rickettsial species available on GenBank using ClustalW 1.6 as implemented in MEGA 7 49 . After manual modification of the alignments, phylogenetic trees were constructed using the maximum likelihood method using Kimura 2-parameter with bootstrap tests of 1,000 replicates via MEGA. R. bellii was included as an outgroup for the bases of the trees for gltA, ompB, and htrA, while R. typhi, R. akari, and Ehrlichia chaffeensis were used as outgroups for sca4, ompA, 16S rRNA, respectively. In order to generate a phylogenetic tree of tick species that was positive for Rickettsia spp., partial nucleotide sequences of mitochondrial 16S rRNA gene obtained from GenBank were used.

Data Availability
All data discussed in the manuscript is included in the paper.