Epidemiological study of relapsing fever borreliae detected in Haemaphysalis ticks and wild animals in the western part of Japan

The genus Borrelia comprises arthropod-borne bacteria, which are infectious agents in vertebrates. They are mainly transmitted by ixodid or argasid ticks. In Hokkaido, Japan, Borrelia spp. were found in deer and Haemaphysalis ticks between 2011 and 2013; however, the study was limited to a particular area. Therefore, in the present study, we conducted large-scale surveillance of ticks and wild animals in the western part of the main island of Japan. We collected 6,407 host-seeking ticks from two regions and 1,598 larvae obtained from 32 engorged female ticks and examined them to elucidate transovarial transmission. In addition, we examined whole blood samples from 190 wild boars and 276 sika deer, as well as sera from 120 wild raccoons. We detected Borrelia spp. in Haemaphysalis flava, Haemaphysalis megaspinosa, Haemaphysalis kitaokai, Haemaphysalis longicornis, and Haemaphysalis formosensis. In addition, we isolated a strain from H. megaspinosa using Barbour-Stoenner-Kelly medium. The minimum infection rate of ticks was less than 5%. Transovarial transmission was observed in H. kitaokai. Phylogenetic analysis of the isolated strain and DNA fragments amplified from ticks identified at least four bacterial genotypes, which corresponded to the tick species detected. Bacteria were detected in 8.4%, 15%, and 0.8% of wild boars, sika deer, and raccoons, respectively. In this study, we found seasonal differences in the prevalence of bacterial genotypes in sika deer during the winter and summer. The tick activity season corresponds to the season with a high prevalence of animals. The present study suggests that a particular bacterial genotype detected in this study are defined by a particular tick species in which they are present.


Introduction
Members of the genus Borrelia in the family Spirochaetaceae are arthropod-borne infectious agents in vertebrates [1], and they are classified into three major groups based on phylogenetic PLOS

Sample collection
Ticks were collected from vegetation by flagging in Wakayama and Yamaguchi prefectures from March 2014 to August 2015 (Fig 1 and S1 Table). In these areas, no specific permission was required for collecting ticks, and this study did not involve endangered or protected species. The collected ticks were identified to the species level and stage based on their morphological features [25]. To demonstrate transovarial transmission, unfed larvae were harvested from engorged female ticks collected from wild boar (Sus scrofa) and sika deer (C. nippon) in Shimonoseki, Yamaguchi Prefecture; the wild boars and sika deer were hunter-harvested or culled for nuisance control under the Program of Prevention from the Bird and Animal Damages from November 2013 to February 2016 (Fig 1,

DNA extraction and cultivation from ticks
In total, 331 adult and 56 nymphal ticks from Shimonoseki, Yamaguchi Prefecture were longitudinally cut in half individually using a disposable knife, where one half was prepared for DNA extraction and the other half was used for borrelial cultivation (Table 1). DNA was extracted from one half of each tick using sodium hydroxide (NaOH) [26]. Briefly, ticks were lysed in 25 μl of 25 mM NaOH for 10 min at 95˚C. Subsequently, 2 μL of Tris-HCL (1 M, pH 8.0) was added for neutralization. Cultivation was performed using modified Barbour-Stoenner-Kelly medium (BSK-M) or modified Kelly-Pettenkofer medium with 10% fetal calf serum (MKP-F) and these medium were incubated at 30˚C [24,26]. The fetal bovine serum included in this medium was replaced with inactivated fetus serum collected from sika deer at the same sampling site. In total, 1,678 ticks from Shunan, Yamaguchi Prefecture and 4,342 ticks from Wakayama Prefecture were processed in pools of 1-50 ticks (mode: nymphs = 20, larvae = 50), thereby obtaining 685 tick pools (155 pools from Shunan and 530 pools from Wakayama) (Tables 2  and 3). The ticks were fractured using a multi-bead shocker (Yasui Kikai, Osaka, Japan) and were centrifuged at 2,500 rpm for 30 s; this procedure was repeated three times. Pellets were used for DNA extraction with a DNeasy Blood and Tissue Kit (Qiagen, Hilden, Germany) according to the manufacturer's instructions. Eggs were harvested from 32 engorged female ticks collected from wild boar or sika deer in Shimonoseki and reared to larvae. DNA was extracted from 1, 598 larvae, processed in 32 pools of 48 to 50 (mode = 50) larvae each, by crushing them in 50 μl of 25 mM NaOH using homogenization pestles (Funakoshi Co. Ltd, Tokyo, Japan). Cultivation of the spirochete was attempted from 5 to 10 remaining larvae from an engorged female tick that produced a positive pool. DNA extracts from 1-ml whole blood samples collected from 190 wild boars and 276 sika deer were examined using a Wizard Genomic DNA Purification Kit (Promega, Madison, WI) according to the manufacturer's instructions. DNA was extracted from the individual sera of 120 raccoons. In total, 500 μl of serum was centrifuged at 21,000 ×g and 4˚C for 5 min, before the pellet obtained was used for DNA extraction with a DNeasy Blood and Tissue Kit (Qiagen) according to the manufacturer's instructions. All extracted DNA samples were stored at −30˚C until further use. Approximately, 100 μl of whole blood from 71 wild boars and 129 sika deer that were randomly selected was used for borrelial cultivation in 1.5 ml of BSK-M at 30˚C. Cultivation was examined under dark-field microscopy (200×).

Real-time PCR of samples from ticks and wild animals
Borrelia spp. DNA in Haemaphysalis spp. were detected by real-time or quantitative PCR (qPCR) targeted at the 16S rRNA gene (16S rDNA) [15,27]. Briefly, qPCR was performed with a StepOne Real-Time PCR system (Thermo Fisher Scientific, Inc., Massachusetts, US) using a Premix Ex Taq PCR kit (Probe qPCR) (TaKaRa, Shiga, Japan). The forward and reverse primers were 16S RT-F and 16S RT-R, respectively [27]. TaqMan dye-labeled minor groove binder probes BS-16S (Thermo Fisher Scientific, Inc.) was used for detecting Borrelia spp. in Haemaphysalis spp. The sensitivity of qPCR was a minimum of 10 plasmid copies  [15]. qPCR was performed in a final volume of 12.5 μl for tick and 25 μl for wild animal samples. The amplification conditions were as follows: 95˚C for 20 s followed by 45 cycles at 95˚C for 1 s and 60˚C for 20 s. The threshold line was fixed at 0.2 to avoid detecting nonspecific fluorescence. Conventional PCR qPCR-positive tick samples were subjected to conventional PCR using KOD FX Neo (TOYOBO Co., Osaka, Japan). Part of the borrelial flagellin gene (flaB), the glycerophosphoryl diester phosphodiesterase gene (glpQ), and 16S rDNA were amplified using the primer pairs BflaPAD and BflaPDU, glpQ F and glpQ R, and rrs-F1 3-26 and rrs-R4 1542-1520, respectively [2,28]. PCR products carryover was carefully checked using distilled water as the blank control in each experiment. For the wild boar and sika deer samples, all DNA samples were subjected to nested PCR targeted at flaB using Illustra PuReTaq™ Ready-To-Go PCR beads (GE Healthcare UK Ltd, Buckinghamshire, UK) to confirm the qPCR results. The primer sets used were Bfla-PAD and BflaPDU for the first PCR and BflaPBU and BflaPCR for nested PCR, as previously described [2]. qPCR-positive raccoons samples were also confirmed by flaB nested PCR.
All PCR products were purified using a High Pure PCR Product Purification Kit (Roche Diagnostic, Basel, Switzerland) according to the manufacturer's instructions and were then directly sequenced with a BigDye Terminator v3.1 Cycle Sequencing Kit and an ABI3031 Genetic Analyzer (Thermo Fisher Scientific, Inc.). The primers used for detection and sequencing were listed in S2 Table. Samples that were positive according to qPCR and flaB sequencing were considered positive.

Multilocus Sequence Analysis (MLSA)
MLSA was performed with isolates using the primer sets listed in S2 Table based on the loci from eight genes (clpA, clpX, nifS, pepX, pyrG, recG, rplB, and uvrA), as previously described [26]. PCR was conducted at 94˚C for 30 s followed by 45 cycles at 94˚C for 30 s, 50˚C for 30 s, and 72˚C for 30 s. Sequencing was performed as described above.

Phylogenetic analysis
The sequences of flaB, glpQ, and 16S rDNA were aligned by CLUSTALW, and neighbor-joining trees were generated based on 1,000 bootstrap replicates according to the Kimura twoparameter methods in MEGA 5.2 (http://www.megasoftware.net) [29]. All positions containing alignment gaps and missing nucleotides were eliminated only in pairwise sequence comparisons (the pairwise deletion option was used). After concatenating the sequences, the MLSA phylogenetic tree was constructed based on Bayesian phylogenetic analysis, as previously described [30,31]. CLUSTALX was used to align the sequences, and phylogenetic analysis was performed with MrBayes 3.2.2 [30,31]. The general time-reversible model with sitespecific rates was used as the evolutionary model for analyses in MrBayes. The first, second, and third codon positions were defined for coding sequences. Analyses were continued for 5 × 10 7 generations or until the average standard deviation of the split frequencies was <0.01. Data were sampled every 100th generation for genes subjected to MLSA. A phylogenetic tree was constructed for MLSA sequences using FigTree v1.4.2. The genetic mean pairwise distance between bacterial genotypes was calculated using the Kimura two-parameter model in MEGA5.2. All sequence data have been deposited in DDBJ/EMBL/GenBank (accession numbers LC170019 to LC170035 and LC171370 to LC171377), and reference sequences were downloaded from DDBJ/EMBL/GenBank or the MLST database (http://www.mlst.net/ databases/).

Statistical analysis
Significant differences in prevalence were determined using Fisher's exact test. P-values of <0.05 were considered significant.
Based on the phylogenetic analysis of flaB sequences obtained from ticks, the Borrelia spp. detected formed a different branch compared with B. theileri and B. lonestari (Fig 2). Moreover, the borreliae we detected were phylogenetically distinguished according to the tick species in which they were detected. Therefore, we preliminarily designated the bacterial genotypes detected in H. flava, H. kitaokai, H. longicornis, and H. megaspinosa as Borrelia sp. HF, Borrelia sp. HK, Borrelia sp. HL, and Borrelia sp. HM, respectively (Fig 2). The bacterial genotypes and tick species detected are summarized in S3 Table. All bacterial genotypes were detected in each representative tick species, except for Borrelia sp. HF that was detected in two nymphal pools from H. formosensis and H. longicornis (S3 Table). The group mean pairwise distances for flaB among the four bacterial genotypes are shown in S4 Table. Sequencing and phylogenetic analyses of the housekeeping gene glpQ (1011 bp) were performed using two Borrelia sp. HK samples (W-31 and L-29, identical), one Borrelia sp. HF sample (W-21), and the Borrelia sp. HM tHM16w isolate (Fig 3), as well as of 16S rDNA (1490 bp) using one Borrelia sp. HM tHM16w isolate (Fig 4). The 16S rDNA sequence of Borrelia sp. HM tHM16w was identical to that of the Borrelia sp. detected in sika deer in Hokkaido (AB897890) but was slightly different from that of AB897891, which was detected in Haemaphysalis japonica, where there was one nucleotide substitution. The group mean pairwise distances for glpQ and 16S rDNA compared with B. theileri and B. lonestari are shown in S5 Table. According to the group mean pairwise distances and phylogenetic analyses of housekeeping genes, the Borrelia spp. detected were genetically more similar to B. theileri than to B. lonestari (Figs 2-4 and S5 Table). A phylogenetic tree was constructed by Bayesian phylogenetic inference using the MLSA sequences of isolate tHM16w (S1  borrelial cells were isolated from whole blood samples from wild boars and sika deer. There were no significant differences in the prevalence of Borrelia sp. HM and Borrelia sp. HF in the summer (from April to September) or winter (from October to March). However, there were seasonal differences during the winter and summer in the prevalence of Borrelia sp. HK (winter; 7.2%, summer; 0.7%, P = 0.0102) and Borrelia sp. HL (winter; 0%, summer; 8%, P = 0.0008). We also examined DNA extracted from the sera of wild raccoons in Wakayama using qPCR. Among the 120 samples, one sample was positive for Borrelia sp. HF (1/120, infection rate = 0.83%) (S6 Table). The sequences detected in wild animals collected in Yamaguchi or Wakayama Prefecture clustered in the same branch for each bacterial genotype (Fig 2).

Discussion
We detected several bacterial genotypes of Borrelia spp. from Haemaphysalis spp. collected from two regions in the western part of the main island of Japan. The prevalence was 0%-0.5% in all ticks collected and 0%-4% in each tick species (Tables 1 and 2). In a previous study, Lee et al. found that the prevalence of Borrelia spp. in Haemaphysalis adult ticks was 0.7% [15].
Several surveys of B. lonestari derived from unfed A. americanum adults have shown that the prevalence in the USA was less than 6% [32][33][34][35][36][37]. However, B. theileri has been detected in host-attached ticks. McCoy et al. reported that the prevalence of B. theileri in cattle infested with Rhipicephalus geigyi was 0.5% in Mari, while Cutler et al. detected B. theileri in 12.5% pools of animal-associated Amblyomma and Rhipicephalus spp. in Ethiopia [38,39]. Moreover, the prevalence of human pathogenic hTBRF borreliae, B. miyamotoi, in unfed ixodid ticks was less than 5% in the USA, Eurasia, and Japan [26,40,41]. Our results and those of previous investigations suggest that the prevalence of hTBRF borreliae in unfed ticks is generally less than 5%. Transovarial transmission of Borrelia sp. HK in H. kitaokai was demonstrated in the present study (Table 3), which has also been examined in B. lonestari and B. miyamotoi [35,42]. Thus, we suggest that hTBRF borreliae might be maintained in the environment via transovarial transmission.
We elucidated four bacterial genotypes among the Borrelia spp. derived from Haemaphysalis ticks (Fig 2 and S3 Table). The ticks were collected from different regions, but each bacterial genotype was detected mostly in a specific tick species. Moreover, Borrelia sp. HM detected in ticks and wild animals in Shimonoseki in the western part of Japan was not genetically distinguished from that detected in Hokkaido in the northern island of Japan in a previous study (Figs 1 and 2, Accession Number AB897887) [15]. In most cases, the genus Borrelia was transmitted by a particular tick species [4]. Thus, our results suggest that the bacterial genotypes detected in this study might be defined by the tick species from which they were detected. The group mean pairwise distances between Borrelia sp. HF, Borrelia sp. HK, and Borrelia sp. HM showed that they slightly differed in terms of flaB (mean pairwise distances over 99%). We analyzed other housekeeping genes (Figs 3 and 4), but some genes could not be amplified using the representative primer sets. Recently, MLSA was used for the intra-or inter-species characterization of Lyme disease borreliae and some hTBRF borreliae [26,30,31,43]. We used MLSA to analyze Borrelia sp. HM strain tHM16w (S1 Fig), but the amplification efficiency was low, even when we used strain-derived DNA. Moreover, nifS, pepX and uvrA were not  Relapsing fever borreliae detected in Haemaphysalis ticks in Japan amplified when we used DNA prepared from ticks (data not shown). These results may be due to mismatches in the primers, which were originally designed based on representative borrelial species. Further analysis such as genome sequencing will be required to define the genetic characteristics of the bacterial genotypes detected in this study. In a previous study, Lee et al. showed that 10.6% of sika deer (C. nippon yesoensis) were infected with Borrelia sp. in Hokkaido [15]. In the USA, a DNA fragment from B. lonestari was detected in 8.7% or 3.1% of blood samples from white-tailed deer (Odocoileus virginianus), 13% of samples from Eastern wild turkey (Meleagris gallopavo silvestris), and 7.4% of samples from migratory waterfowl [American black buck (Anas rubripes), Canada goose (Branta canadensis), mallard (Anas platyrhynchos), northern pintail (Anas acuta), ring-necked duck (Aythya collaris), and wood duck (Aix sponsa)] [44][45][46]. However, no borrelial DNA was detected in raccoons from the USA, although antibodies were detected [47]. In our survey, borrelial DNA was detected in 8.4% of wild boars, 15% of sika deer, and 0.83% of wild raccoons (Tables 4 and 5 and S6 Table). In this present study, we detected borrelial DNA using a serum sample in case of raccoons. The difference in collection may affect the low prevalence in raccoons. On the other hand, most relapsing fever borreliae were usually detected in whole blood and serum because of high bacteremia [48]. Moreover, B. miyamotoi DNA was detected in human serum [20]. From our result and previous observations, we speculated that the Borrelia spp. found in this study and B. theileri infect the order Artiodactyla. We did not detect Borrelia spp. in ticks collected from a park in Shunan, where wild boars and sika deer were absent and birds were present. On the other hand, we could not isolate the bacteria in blood samples. Lee et al. reported the average of bacteremia in sika deer blood to be 3.5 in log 10 per ml [15]. Because we observed 0.1ml blood for cultivation, the low copy number of bacteria in blood might be involved in this result.
In Shimonoseki, where we collected sika deer, it was reported that sika deer were infested with H. longicornis, Haemaphysalis yeni, H. flava, H. megaspinosa, H. kitaokai, Ixodes ovatus, and Amblyomma testudinarium. In addition, it was reported that H. flava and H. megaspinosa are active in all seasons. H. kitaokai and H. longicornis exhibit seasonal changes in their activity; they are mainly active in the winter and summer, respectively [49]. In this study, we found seasonal differences in the prevalence of bacterial genotypes in sika deer during the winter and summer, i.e., Borrelia sp. HK (detected from H. kitaokai) was detected in the winter and Borrelia sp. HL (detected from H. longicornis) was detected in the summer. The tick activity season corresponds to the season with a high prevalence of animals; therefore, we suggest that Borrelia sp. HK and Borrelia sp. HL were transiently infected and that there was no chronic bacteremia in sika deer. We only detected Borrelia sp. HM in wild boars, and H. flava, H. megaspinosa and A. testudinarium infested wild boars in our sampling site (data not shown). Moreover, H. flava, H. longicornis, and A. testudinarium infested wild boars in Shimane Prefecture, which neighbors Yamaguchi Prefecture [50]. Therefore, we suggest that wild boars are infected with a limited range of bacterial genotypes.
In this study, we detected Borrelia sp. derived from Haemaphysalis ticks in two regions in the western part of the main island of Japan. We detected four bacterial genotypes in Haemaphysalis ticks and wild animals. Our results suggest that the bacterial genotypes detected in this study are defined by the tick species in which they are present.

Originality-significance statement
The authors confirm that all of this reported work is original. This is the first report to show the prevalence of Haemaphysalis ticks associated borreliae in ticks and wild animals in the western part of the main island of Japan.
Supporting information S1