Prevalence and Genetic Diversity of Bartonella Spp. in Northern Bats (Eptesicus nilssonii) and Their Blood-Sucking Ectoparasites in Hokkaido, Japan

We investigated the prevalence of Bartonella in 123 northern bats (Eptesicus nilssonii) and their ectoparasites from Hokkaido, Japan. A total of 174 bat fleas (Ischnopsyllus needhami) and two bat bugs (Cimex japonicus) were collected from the bats. Bartonella bacteria were isolated from 32 (26.0%) of 123 bats. Though Bartonella DNA was detected in 79 (45.4%) of the bat fleas, the bacterium was isolated from only one bat flea (0.6%). The gltA sequences of the isolates were categorized into genotypes I, II, and III, which were found in both bats and their fleas. The gltA sequences of genotypes I and II showed 97.6% similarity with Bartonella strains from a Finnish E. nilssonii and a bat flea from a E. serotinus in the Netherlands. The rpoB sequences of the genotypes showed 98.9% similarity with Bartonella strain 44722 from E. serotinus in Republic of Georgia. The gltA and rpoB sequences of genotype III showed 95.9% and 96.7% similarity with Bartonella strains detected in shrews in Kenya and France, respectively. Phylogenetic analysis revealed that Bartonella isolates of genotypes I and II clustered with Bartonella strains from Eptesicus bats in Republic of Georgia and Finland, Myotis bats in Romania and the UK, and a bat flea from an Eptesicus bat in Finland. In contrast, genotype III formed a clade with B. florencae, B. acomydis, and B. birtlesii. These data suggest that northern bats in Japan harbor two Bartonella species and the bat flea serves as a potential vector of Bartonella transmission among the bats.

Bartonella are small, fastidious Gram-negative bacteria that parasitize erythrocytes and endothelial cells of various mammals. Bartonella infections have been found in various vespertilionid bat species including the genera Myotis and Eptesicus in Europe, Asia, and the Americas [5]. Candidatus Bartonella mayotimonensis was first detected from an aortic valve of a patient with endocarditis in the USA [6], and then a B. mayotimonensis-like organism was also identified in Eptesicus bats in Finland and Myotis bats in Finland, France, and Spain [7,8]. Additionally, it has been reported that the gltA sequences of Bartonella bacteria isolated from Eptesicus and Myotis bats in Republic of Georgia are closely related to those from forest workers in Poland [9,10]. These data suggest that Bartonella bacteria in Eptesicus and Myotis bat species have the potential to spill over to humans and some species may be pathogenic in humans.
Ectoparasites such as cat fleas [11], body lice [12], sand flies [13], and deer keds [14] are known to be competent vectors of B. henselae, B. quintana, B. bacilliformis, and deer associated strains through experimental infection studies or immunofluorescence analysis, respectively. Previous studies have shown that some ectoparasites, such as bat flies, may be involved in Bartonella transmission among bats. Bartonella DNA has been frequently detected in nycteribiid and streblid bat flies collected in Japan [15] and 19 other countries [5,[16][17][18][19][20][21][22]. In particular, Bartonella bacteria were isolated from a Cyclopodia bat fly on Eidolon helvum in Ghana [21] and from three Nycteribia species that infested Miniopterus fuliginosus in Japan [15]. Although there have been no controlled experiments that have proven that bat flies are vectors for Bartonella, these findings suggest that bat flies may serve as the major vector for the transmission of Bartonella among bats in multiple systems.
We previously reported that Bartonella bacteria are present in 24% (12/50) of Miniopterus fuliginosus in Japan [23]. Furthermore, detection of Bartonella has also been reported in various Asian bat species at high prevalence, such as Taphozous melanopogon (100%; 1/1), Hipposideros sp. ( [27]. Although Myotis and Eptesicus species [7] are suspected to carry zoonotic Bartonella species, no epidemiological studies of the bacterial presence in these bats and their ectoparasites have been conducted in Japan so far. The aims of the present study were to investigate the prevalence and genetic characteristics of Bartonella bacteria in northern bats (Eptesicus nilssonii) and to identify a potential ectoparasite vector transmitting the bacteria among northern bats in Japan.

Collection of Blood and Ectoparasites from Bats
Before performing the present study, we obtained permission to capture bats from the Oshima general sub-prefectural bureau, Hokkaido government (license #: Oshima 27,28,29,122,123,and 124). The bats were living within the exterior walls of an abandoned building in a dense colony as shown in the supplementary video. A total of 123 northern bats (Eptesicus nilssonii) were captured in Yakumo Town (42° 29′ N, 140° 18′ E), Hokkaido Prefecture, located in the northern part of Japan. Bat samples were collected during three sampling sessions in August 2017 (N = 49), June 2018 (N = 38), and August 2018 (N = 36) using hand nets with personal protective equipment.
Blood samples were aseptically collected from the bats via heart puncture after bats were euthanized following the guidelines of the Japan Veterinary Medical Association as previously reported [23]. The blood samples and the carcasses were immediately sent to the Laboratory of Veterinary Public Health at the Department of Veterinary Medicine, College of Bioresource Sciences, Nihon University under frozen conditions with dry ice. All of the samples were stored at − 70 °C until they were examined.
A total of 174 bat fleas and two bat bugs were obtained from the surface of the carcasses in the laboratory. The ectoparasites were morphologically identified at the species level under a stereo microscope SZX16 (Olympus, Tokyo, Japan) with taxonomic keys [28,29].

Isolation of Bartonella Bacteria from Bats
Isolation of Bartonella bacteria from the bat blood samples exactly followed a previously described method [23]. Bartonella bacteria were tentatively identified by colony morphology (small, white, round shape) and three homogenous colonies per sample were sub-cultured on a fresh blood agar plate using the same conditions as the primary culture.

Detection of Bartonella DNA from Ectoparasites
To avoid bacterial and fungal contamination on the surface of the ectoparasites, each sample was sterilized by immersing in 500 µl of 70% ethanol containing 0.35% povidoneiodine (Shionogi & Co., Ltd, Osaka, Japan) for 10 min. Then, the samples were once washed with 1 ml of 0.01 M PBS with 0.5% FBS and transferred into shatter-resistant 2.0 ml tubes (SSIbio, Lodi CA, USA). After adding 400 µl of sucrose phosphate glutamate (SPG; 10 mM sodium phosphate, 220 mM sucrose, and 0.5 mM L-glutamic acid), each sample was homogenized using a bead crusher µT-12 (TAITEC corp., Saitama, Japan) at 3,000 rpm for 1 min. Genomic DNA was extracted from the homogenate aliquot (100 µl) by InstaGene Matrix (Bio-Rad Laboratories, Inc. Hercules, CA, USA). The remaining homogenates were stored − 70 °C until further examination including Bartonella isolation.

299
Real-time PCR targeting the transfer-mRNA (ssrA) gene of the genus Bartonella was used for screening Bartonella DNA [30]. Real-time PCR assays were performed in 20 µl reaction mixtures containing 10 µl of TB Green Premix Ex Taq II (Takara Bio Inc., Shiga, Japan), 2 μL of DNA sample (approximately 10 to 20 ng/µl), 1 µl of each primer (10 nM), and 6 µl of nuclease-free water. The PCR conditions were as follows: 95 °C for 30 s, followed by 40 cycles at 95 °C for 5 s and 60 °C for 30 s. The genomic DNA from the northern bat isolate (EN2-1) was used as a positive control and nuclease-free water as a negative control in this study. The targeted DNA was amplified with a Thermal Cycler Dice Real Time System II (Takara Bio Inc.). Melting curve analysis was applied to the results with a Ct value lower than 35 cycles. Since strain EN2-1 used as positive control showed the melting temperature at 80 °C, samples with 80 ± 1 °C melting temperature were tentatively defined as Bartonella DNA positive by the assay. Each positive sample was additionally analyzed by conventional PCR targeting the citrate synthase gene (gltA) [31]. The conventional PCR products were separated on 3% agarose gels by electrophoresis and were visualized by staining with ethidium bromide under UV light. When samples tested positive for both PCRs, the samples were defined as being infected with Bartonella bacteria. A band showing the expected size for gltA (approximately 380 bp) was purified by using the Wizard SV Gel and PCR Clean-Up System (Promega, Madison, WI, USA) following the manufacturer's instructions.
The purified PCR products of the gltA gene were directly sequenced using a BigDye Terminator Cycle Sequencing Ready Reaction Kit and a Genetic Analyzer model 3130 (Applied Biosystems). When double peaks were detected in the DNA sequencing chromatograms, the PCR products were cloned using the plasmid pGEM-T Easy vector system (Promega) and resequenced. The plasmids were purified from the transformed cell lysates with a plasmid purification kit (PureYield Plasmid Miniprep System; Promega) and then 10 clones per samples were sequenced using the primers SP6 (5′-CAA GCT ATT TAG GTG ACA CTA TAG -3′) and T7 (5′-TAA TAC GAC TCA CTA TAG GG-3′).

Isolation of Bartonella Bacteria from Ectoparasites
The homogenates of Bartonella DNA-positive ectoparasites were subsequently used for the isolation of Bartonella bacteria following previously described methods [15].

PCR Amplification of the gltA and rpoB Genes from Bartonella Isolates
Genomic DNA was extracted from each sub-cultured colony by using InstaGene Matrix (Bio-Rad Laboratories) and subjected to genus-specific PCR targeting the gltA [31] and RNA polymerase beta-subunit-encoding (rpoB) genes [32]. The PCR products were separated on 2% agarose gels by electrophoresis and visualized by staining with ethidium bromide under UV light. Any samples showing the expected band sizes for gltA and rpoB (approximately 850 bp) were considered members of the genus Bartonella. The PCR products were then purified using the Wizard SV Gel and PCR Clean-Up System (Promega) and directly sequenced as previously described [23].

Genotyping
Genotypes of the gltA gene were determined when unique sequence variants with ≥ 1 nucleotide difference were found in each strain by comparing the gltA sequences using Genetyx software ver. 12 (Genetyx Corporation, Tokyo, Japan). Representative strains from each genotype were further analyzed by sequencing their rpoB genes.

Sequence Homology Analysis
The gltA and rpoB sequences of representative strains were compared with genomic sequences of prokaryotes registered in the GenBank/EMBL/DDBJ database using the BLAST program.

Phylogenetic Analysis
Based on evolutionary model selection using jModelTest 2 [33] with Akaike's information criterion corrected for finite sample sizes (AICc) [34], the generalized time-reversible substitution model with four gamma-distributed categories and a proportion of invariant sites (GTR + G + I) model was the best available model for the phylogenetic analyses based on the gltA sequences.
A phylogenetic tree of the gltA sequences was constructed using the maximum-likelihood method based on the GTR + G + I model in MEGA 7 [35]. Known Bartonella species (N = 40), bat-associated Bartonella strains (N = 420) derived from 36 countries, and Brucella melitensis 16 M as an outgroup were included in this analysis. Strain names, host types, host scientific names, countries where the bats and ectoparasites were collected are summarized together with references and the gltA accession numbers in supplementary Table 1.

Statistical Analysis
Chi-square test was applied to evaluate any statistical difference in parasite and Bartonella prevalence in bats. P value < 0.01 was considered to be statistically significant.

Ectoparasite Fauna in the Northern Bats
A total of 176 ectoparasites including 174 bat fleas and two bat bugs were collected and submitted to species identification. All bat fleas (N = 174) were morphologically identified as Ischnopsyllus needhami and the two bat bugs as Cimex japonicus [28,29]. Parasite prevalence of I. needhami and C. japonicus in bats was 57.7% (71/123) and 1.6% (2/123), respectively and the prevalence of I. needhami was significantly higher than that of C. japonicus (P < 0.01). Parasite prevalence of bat fleas varied by sampling period: 63.3% (31/49) in August 2017, 71.1% (27/38) in June 2018, and 36.1% (13/36) in August 2018, and the prevalence in June 2018 was significantly higher than that in August 2018 (P < 0.01) ( Table 1). The bat isolates were conveniently classified into three genotypes as genotypes I (33 isolates), II (33 isolates), and III (30 isolates) based on the gltA sequences. None of the bats was co-infected with multiple genotypes, though only three colonies from each sample were examined for the genotypes. Three isolates from one I. needhami were classified into the genotype I. The gltA and rpoB sequences of representative bat strains (EN2-1, EN19-2, and EN36-1) were registered in the GenBank/EMBL/DDBJ database ( Table 2).

Detection of Bartonella DNA from Bat Ectoparasites and gltA Genotypes of Bartonella DNA
Bartonella DNA was detected from 45.4% I. needhami (79/174), but not from C. japonicus. The gltA sequences from the bat fleas were also classified within the genotypes I, II, and III, as were the bat isolates. The numbers of I. needhami harboring only one genotype were 43 for genotype I, 21 for genotype II, and 10 for genotype III,    (Table 3). Since seven samples produced double chromatogram peaks for at least one locus, these were cloned and resequenced. Fleas harboring two genotypes were found in five individuals: genotypes I and II in two fleas, genotypes II and III in one flea, and genotypes I and III in two fleas.

Sequence Similarities of the gltA and rpoB Genes Between the Three Genotypes and Other Bartonella Strains
The sequence similarities between the three genotypes and the closest Bartonella strains based on BLAST searches are shown in Table 4.

Discussion
In the present study, 26% (32/123) of Japanese northern bats (E. nilssonii) were found to be infected with Bartonella bacteria. According to previous studies, Bartonella bacteria were also isolated from Eptesicus bats: E. nilssonii (33.3%; 1/3) in Finland [7] and E. serotinus (20%; 4/20) in Republic of Georgia [9]. The prevalence of Bartonella bacteria in Japanese northern bats is similar to those of previous two studies. In contrast, no Bartonella was detected in E. serotinus in China [27] or France [8]. Since these results suggest that Bartonella prevalence in Eptesicus bats may vary depending on the locations of the bat populations and/or bat species, more investigation will be necessary to clarify the causes of the differences in Bartonella prevalence in different regions and bat species.
Our study showed that Japanese northern bats were infested with two species of ectoparasites: bat fleas (Ischnopsyllus needhami) and bat bugs (Cimex japonicus). The prevalence of I. needhami was significantly higher than that of C. japonicus (P < 0.01), indicating that a dominant ectoparasite in the sampled Japanese northern bats is I. needhami. Parasite prevalence of bat fleas varied by sampling period, and that in June 2018 was significantly higher than that in August 2018 (P < 0.01). Previous studies have shown that various ectoparasites such as Ischnopsyllus spp. and Myodopsylla borealis (bat fleas), Cimex spp. (bat bug), Carios kelleyi (soft tick), spinturnicid and macronyssid mites, Basilia spp., Nycteribia spp., and Penicillidia dufourii (bat fly) parasitize Eptesicus bats in Russia and the USA [43][44][45]. However, we did not find any ectoparasites other than I. needhami and C. japonicus in this study. Hence, the prevalence and diversity of ectoparasites on Eptesicus bats may depend on the season, habitat, and/or bat species.
Bartonella DNA was frequently detected from 45.4% bat fleas (79/174) but Bartonella bacteria were isolated from only one bat flea. In the present study, the isolation rate is much lower than the detection rate of Bartonella DNA by PCR. The isolation of Bartonella from fleas was performed after identification of all ectoparasites and detection of Bartonella DNA was completed, which took about 1 to 2 months. In the previous studies, the isolation rate of Bartonella from fresh deer ked samples was reported to be considerably higher at 51.5% [14] and 73.3% [46]. Therefore, it may be important to use fresh-collected samples to measure the actual prevalence of viable bacteria in ectoparasites. Since the northern bats that we sampled were living in dense conditions within the exterior walls of a building (Supplementary video), fleas can easily move between bat bodies in the colony. Though Bartonella prevalence (28.2%; 20/71) in the bats infested with I. needhami was slightly higher than that (23.1%; 12/52) in the bats uninfested with the fleas, this difference was not statistically significant. Becker et al. [47] also reported that no significant difference was observed in the Bartonella prevalence between the bat fly-infested and -uninfested vampire bats, and suggest that the vector presence in the colony is more important for the transmission of Bartonella than abundance. Since fleas move freely around the bat bodies in the dense colonies, bats uninfested with fleas may mean only that they were not parasitized at the time of capture. When comparing the gltA genotypes of Bartonella between the bats and the fleas, multiple genotypes were detected in the fleas, but the bats were infected with only one genotype. The evidence supports the possibility that the bat fleas actively move and suck blood of northern bats in the colony, and the bat fleas are likely to transmit Bartonella bacteria to individual bats.
In the BLAST searches, the gltA and rpoB sequences of strains EN2-1 (genotype I) and EN19-2 (genotype II) showed the highest similarities (97.9% in gltA and 98.8% in rpoB) with strains 1157/3 from a northern bat in Finland, 1F40 from Ischnopsyllus variabilis in the Netherlands, and 44722 from E. serotinus in Republic of Georgia. On the other hand, EN36-1 (genotypes III) showed the highest similarities (95.9% in gltA and 96.7% in rpoB) with a shrew (Crocidura spp.) strain B28303 in Kenya and B. florencae R4 T in France. In the phylogenetic analysis based on the gltA sequences, strains EN2-1 (genotype I) and EN19-2 (genotype II) formed clade A with strains from bat fleas collected from an Eptesicus bat in the Netherlands, Eptesicus bats in Republic of Georgia and Finland, and Myotis bats in the UK and Romania. Strain EN36-1 (genotypes III) formed clade B with B. florencae, B. acomydis, and B. birtlesii from shrews, wood mouse, and golden spiny mouse, respectively. These results suggest that genotypes I and II are vesper bat-specific Bartonella species and genotype III is closely related to shrew and rodent-derived Bartonella species. Interestingly, strains in clade B were distinct from other known bat-associated Bartonella strains. A previous study [48] suggested that host switching of Bartonella is a common event as has occurred at least five times between bats and other small mammals. From the results of BLAST search and phylogenetic analysis based on the gltA sequences, such an event might have occurred in genotype III between Japanese northern bats and shrews. However, the identity of a "bridge" vector capable of transmitting Bartonella bacteria between bats and other small mammal species is currently unknown. Further investigation will be necessary to clarify the bridge vectors of northern bats.
Our study showed that northern bats in Japan harbor two species of Bartonella: vesper bat-specific Bartonella species and a B. florencae-like bacterium. Additionally, bat fleas (Ischnopsyllus needhami) were found to be the dominant ectoparasite species in these bats and they serve as a potential vector of Bartonella transmission among northern bats in Japan.