Molecular Detection and Phylogenetic Analyses of Diverse Bartonella Species in Bat Ectoparasites Collected from Yunnan Province, China

Bartonella species has been validated as blood-borne bacteria in mammals and has a substantial opportunity to be harbored by a variety of hematophagous arthropod vectors. Bats, along with their ectoparasites, are recognized worldwide as one of the natural reservoir hosts for these bacteria. However, there have been few investigations of Bartonella bacteria toward a broad range of obligated bat ectoparasites in China. Here, molecular detection of Bartonella species was performed to survey the infection among bat ectoparasites and follow-up phylogenetic analyses to further characterize the evolutionary relationships of the genus. A total of 434 bat ectoparasites involving four types of arthropods, namely, bat mites, bat tick, bat fleas, and bat flies (further divided into traditionally fly-like bat flies and wingless bat flies) were collected in 10 trapping sites in Yunnan Province, southwestern China. Bartonella was detected by PCR amplification and sequencing through four gene target fragments (gltA, ftsZ, rpoB, and ITS). Accordingly, diverse Bartonella species were discovered, including both the validated species and the novel genotypes, which were characterized into several geographical regions with high prevalence. Phylogenetic analyses based on gltA and multi-locus concatenated sequences both demonstrated strong phylogeny–trait associations of Bartonella species from bats and their parasitic arthropods, suggesting the occurrence of host switches and emphasizing the potential connecting vector role of these ectoparasites. Nevertheless, the maintenance and transmission of Bartonella in both bat and hemoparasite populations have not been fully understood, as well as the risk of spillage to humans, which warrants in-depth experimental studies focusing on these mammals and their ectoparasites.


Introduction
The genus Bartonella is a clade of Alphaproteobacteria that contains over 45 species of fastidious, facultative intracellular, Gram-negative bacilli that can globally infect mammalian hosts, and transmission between hosts by hematophagous arthropod vectors [1,2]. Moreover, with the respective reservoir hosts constantly increasing, there are also numerous unclassified Bartonella species isolated from animal reservoirs that have not yet been fully characterized, and it is very likely that there are more distinct species than those that have been recognized in previous studies [3]. Currently, at least 13 Bartonella species are known to be capable of infecting humans and causing a broad spectrum of diseases, including endocarditis, myocarditis, neuroretinitis, meningitis, splenomegaly, lymphadenopathy, and neurologic disorders [4].

Bat Ectoparasites Sampling, Mixing, and Species Identification
A total of 434 bat ectoparasites were collected from 10 cities/counties in Yunnan Province of China between 2012 and 2014, and initial identification was according to the morphological traits-including 295 bat flies, 113 bat mites, 21 bat fleas, and 5 bat ticks ( Figure 1, and details are shown in Table S5). All samples were mixed into 40 pools on the basis of the information regarding morphological identification, collected date, and location, and the subsequent species identification was confirmed by sequencing and analyzing the cytochrome c oxidase (COI) gene for each pool. The comparison of obtained COI sequences in the BOLD database and GenBank showed that there are 28 pools of 295 flies belonging to the superfamily Hippoboscoidea (25 pools of 284 wingless bat flies belonging to the family Nycteribiidae and 3 pools of 11 traditionally fly-like bat flies belong to the family Streblidae), 6 pools of 113 bat mites belonging to the family Spinturnicidae, 3 pools of 21 bat fleas belong to the family Ischnopsyllidae, and 3 pools of 5 bat ticks belonging to the family Ixodidae (Table 1, and phylogenies are shown in Figure S1). The result of molecular biology methods based on COI sequences was consistent with initial morphological identification and a previous study that identified through the use of the meta-transcriptomic approach [25].
belonging to the superfamily Hippoboscoidea (25 pools of 284 wingless bat flies belonging to the family Nycteribiidae and 3 pools of 11 traditionally fly-like bat flies belong to the family Streblidae), 6 pools of 113 bat mites belonging to the family Spinturnicidae, 3 pools of 21 bat fleas belong to the family Ischnopsyllidae, and 3 pools of 5 bat ticks belonging to the family Ixodidae (Table 1, and phylogenies are shown in Figure S1). The result of molecular biology methods based on COI sequences was consistent with initial morphological identification and a previous study that identified through the use of the meta-transcriptomic approach [25].

Bartonella Detection
For all mixed ectoparasite pools, four genes (gltA, ftsZ, rpoB and ITS) were used to detect Bartonella. A total of 20 pools involving four types of bat ectoparasites collected in the current study were found to harbor Bartonella by gltA screening and the following detection of targeted ftsZ genes: 2 (66.7%) of 3 bat flea pools, 1 (33.3%) pool of 3 bat tick pools, 4 (66.7%) of 6 bat mite pools, and 13 (52.0%) pools of 25 wingless bat fly pools were positive to detection, and all three fly-like bat fly pools were negative (Table 1). Several pools were negative to PCR detection in genes rpoB and ITS, and therefore, we failed to obtain a part of some novel strains' targeted gene fragment sequences (Table S1). The bat ectoparasite populations of all 10 trapping sites in Yunnan province are shown to be widely infected with Bartonella, wherein the wingless bat fly showed the highest prevalence ratio of Bartonella (eight of nine sites were positive to detection), whereas the traditional bat fly was the lowest (all sites are negative to detection) ( Table 1). Moreover, several sites with more than one type of hemoparasite harbored Bartonella, such as Wanding Town (bat flea and wingless bat fly) and Mengla County (bat mite and wingless bat fly) ( Table 1).

Identification of Bartonella
We identified the novel detected strains by nucleotide sequence similarities and phylogenies of the gene gltA, the most common target for Bartonella detection with good discriminatory power of delimitation. According to the nucleotide BLAST analysis of approximately 380 bp gltA fragment sequences in the GenBank, the similarity of those detected from bat ectoparasites in this study with the validated species/strains ranged from 88.42% in the family Ischnopsyllidae (YNWD/BC02) to 100% in the family Spinturnicidae (YNML2/BM03 and YNML2/BM04) ( Table 2). Out of all 20 novel strains, there were 12 strains of novel species, tentatively named Bartonella sp., which shared <96% sequence similarity with known species/strains discovered by previous studies, while the other 8 strains were the same as the previously described genotype with >96% identity. Likewise, the phylogeny analysis by the maximum likelihood tree based on the gltA sequences of obtained and representative Bartonella strains associated with bats and bat ectoparasites put themselves into the same clade with the closest strains to those showing the highest identity ( Figure 2).
On the other hand, the clustering traits of Bartonella sp. in the current study were neither strongly associated with host family taxonomy nor geographic origin, the same as the strains from bats and their parasites in previous studies ( Figure 2). Firstly, the strains from different host arthropods in the same trapping sites closely clustered in the same clade, for instance, the strain YNML1/BF12 from bat flies pool and YNML1/BMo2 from bat mites pool (and shown 98.9% similarity). Secondly, the strains detected in the same type of ectoparasite (bat fly) from Mojiang County, YNMJ/BF15 and YNMJ/BF16, did not cluster in the same clade (86.8% similarity), likewise with the flea-borne strains YNWD/BC02 and YNWD/BC03 from Wanding Town (85.3% similarity).

Phylogenetic Analyses
Three gene (gltA, ftsZ, and rpoB) fragment sequences were concatenated and, together with validated Bartonella species/strains that were discovered in nine types of hosts at relevant taxonomic scales of the order level, to infer a more comprehensive phylogeny of this genus (Table S3). Ultimately, an approximately 2017 bp length of concatenated sequences for three loci, 380-bp, 786-bp, and 851-bp of genes gltA, ftsZ, and rpoB partial sequences, respectively. The concatenated sequences and reference strains were used to construct the phylogenetic tree using the maximum likelihood method.
Phylogenetic analysis indicates that clades of Bartonella lineages seem broadly hostspecific within the host order, except for those associated with bats ( Figure 3). Bats and their bug-associated strains formed several clades and external branches to other mammalian orders, which were dispersal distributed across the tree. The tree put all strains of the current study into two close clades of monophyletic groups associated with bats and closely    using the MLSA approach. The host species of each strain is shown behind the bacteria names, and the different colors indicate different host order: Arthropoda (red), Artiodactyla (orange), Carnivora (golden), Chiroptera (green), Dasyuromorphia (aqua), Diprotodontia (cyan), Primates (pink), Rodentia (rose), and Peramelemorphia (purple). Within the phylogenetic tree, the Bartonella strains newly identified here are marked with gray shaded squares, and the strains associated with bat bugs are marked with black pentagrams. Hyphomicrobium album strain XQ2 was used for the outgroup and is indicated with blue.

Discussion
A total of 434 bat-parasitic arthropods collected from 10 trapping sites in Yunnan were used to perform a molecular investigation of Bartonella. Morphological identification and confirmation by analysis based on the cytochrome c oxidase (COI) gene revealed five families of bat-specific ectoparasites were contained, namely, Nycteribiidae (wingless spider-like bat flies), Streblidae (traditionally fly-like bat flies), Spinturnicidae (bat mites), Ischnopsyllidae (bat fleas), and Ixodidae (bat ticks). All bat ticks were identified as Ixodes vespertilionis or Ixodes collaris, two species of ticks that exclusively parasitize bats [39,40]. All bat mites and bat fleas were unclassified species of Spinturnicidae sp. and Thaumapsylla sp., respectively. In contrast, diverse bat flies were discovered, including both named species such as Eucampsipoda africana/sundaica, Penicillidia monoceros, and Brachytarsina kanoiand, and the potentially novel species tentatively named Nycteribiidae sp., Phthiridium sp., Figure 3. Maximum likelihood tree of the genus Bartonella based on three-loci (gltA, ftsZ, and rpoB) using the MLSA approach. The host species of each strain is shown behind the bacteria names, and the different colors indicate different host order: Arthropoda (red), Artiodactyla (orange), Carnivora (golden), Chiroptera (green), Dasyuromorphia (aqua), Diprotodontia (cyan), Primates (pink), Rodentia (rose), and Peramelemorphia (purple). Within the phylogenetic tree, the Bartonella strains newly identified here are marked with gray shaded squares, and the strains associated with bat bugs are marked with black pentagrams. Hyphomicrobium album strain XQ2 was used for the outgroup and is indicated with blue.

Discussion
A total of 434 bat-parasitic arthropods collected from 10 trapping sites in Yunnan were used to perform a molecular investigation of Bartonella. Morphological identification and confirmation by analysis based on the cytochrome c oxidase (COI) gene revealed five families of bat-specific ectoparasites were contained, namely, Nycteribiidae (wingless spider-like bat flies), Streblidae (traditionally fly-like bat flies), Spinturnicidae (bat mites), Ischnopsyllidae (bat fleas), and Ixodidae (bat ticks). All bat ticks were identified as Ixodes vespertilionis or Ixodes collaris, two species of ticks that exclusively parasitize bats [39,40]. All bat mites and bat fleas were unclassified species of Spinturnicidae sp. and Thaumapsylla sp., respectively. In contrast, diverse bat flies were discovered, including both named species such as Eucampsipoda africana/sundaica, Penicillidia monoceros, and Brachytarsina kanoiand, and the potentially novel species tentatively named Nycteribiidae sp., Phthiridium sp., Nycteribia sp., and Brachytarsina sp. (Table S1). Host species information was initially identified according to morphological traits, and all specimens were later mixed on the basis of the above information and further confirmed by sequencing and analyzing the COI gene for each pool. Therefore, the accuracy of initial information was highly dependent on experienced field biologists, which was important for the subsequent pooling and gene identification. We believe the combination of morphological characteristics and molecular analysis will better define these bat ectoparasites in the future.
Although diverse Bartonella strains/genotypes have been identified in bats and their ectoparasites worldwide [16,19,26,27], including in China [16,41,42], there have been few reports in China. In this study, molecular detection of four gene target fragments (gltA, ftsZ, rpoB, and ITS) was performed to survey the infection of Bartonella among bat ectoparasites as described above. In sum, 20 mixed pools were positive for detecting the gene gltA and ftsZ, while 17 and 18 were positive for genes rpoB and ITS, respectively. Four types of ectoparasites, namely, bat mites, bat fleas, bat ticks, and wingless bat flies, were discovered to harbor Bartonella bacteria. Moreover, the bat ectoparasite populations of all 10 trapping sites were found to be commonly infected with the Bartonella bacteria (Table 1), which somehow reflects a high prevalence ratio of these bacteria in Yunnan province. To our knowledge, it is the first time these bacteria have been detected in bat fleas and ticks in China, and there have been few discoveries around the world regarding Bartonella infections in these two types of bat-parasitic arthropods.
Due to the gltA gene having good discriminatory power of delimitation [43,44], limited genetic sequencing has permitted the tentative identification of Bartonella. We propose that the detected strains be considered a new species if a >327 bp gltA fragment shares <96.0% sequence similarity with the validated species, whereas those with >96% identity are the same genotype [43]. Accordingly, 12 strains of novel species and 8 strains of validated genotypes were detected, revealing bat ectoparasites infected with a diversity of these bacteria (Table 1), and with many novel genotypes not overlapping with those from bats (Figure 1), suggesting the role of natural reservoirs of Bartonella for these arthropods. On the other hand, the close relationship of strains from bat mites and bat flies indicated that ectoparasites serve as the potential connecting vector of Bartonella bacteria (Figure 1), which further supports the same idea of the previous study [27]. Nevertheless, there is an essential difference between proven vector competence and vector potential, requiring further experimental vector transmission studies to confirm this.
Previous studies showed that Bartonella lineages are broadly host-specific within orders [19,45,46]; however, host-switching seems to occur between "closely related" animals. The maximum likelihood tree demonstrated a strong phylogeny-trait association of Bartonella spp. from bats and their parasitic arthropods, forming several complex bat/ectoparasite-associated Bartonella lineages (Figures 2 and 3). We suspect the high host specificity and co-feeding behavior of obligated ectoparasites is one of the vital factors [47]. The long history of co-evolution between mammals, parasitic ectoparasites, and harboring bacteria may significantly cause the complex Bartonella lineages that cluster genetically similar strains from different host types. Given that diverse bacteria are harbored by bats [12,13], the same event of reservoir spillover to parasitic arthropods and divergent adaption can occur in other lineages of host-bacterial systems. On the other hand, the Bartonella associated with bat bugs and bat-borne strains form the external branches to other mammalian orders within the tree (Figure 3), which again confirms the previous view that bats have a deep influence on the radiation of mammal-associated Bartonella bacteria [19]. In addition, the previous studies demonstrated that the diversification of mammal-infecting eubartonellae started almost exactly when bats began their evolutionary radiation [19,48,49], also supporting the fact that bats play a crucial role in the genomic evolution of these bacteria. Even though these potentially confounding factors are discovered, in-depth analyses are necessary for understanding the coevolutionary patterns and frequency of host-switching events.
Bats are the natural reservoir for many pathogens [9][10][11][12][13], including Bartonella bacteria [19,38,41]. A bat-associated Bartonella species, proposed as Bartonella rousetti (Bartonella sp. strain R-191), was corroborated, potentially being capable of infecting humans by serologic evidence [5]. Although phylogenetic analysis shows strain R-191 closely clustered with the fly-associated strain YNBS/BF03 in this study (Figure 3), whether these ectoparasite-borne Bartonella strains hold the same infecting capability still relies on further experimental infection studies. We only focused on the Bartonella infection of bat ectoparasites in this study but lacked a detection of their host bats; therefore, an investigation of bat populations in these trapping sites and even wider regions should be taken up to clear the infection situation of these bacteria. Moreover, serology is critical in diagnosing Bartonella infections [3], and corresponding surveillance of the local human populations is also needed. More studies are required to elucidate the correlations between exposure routes and the pathogenicity of ectoparasite-borne Bartonella sp. in humans.
Overall, molecular detection of Bartonella in the bat ectoparasites revealed diverse bacterial species infection, and sequencing of multi-locus and phylogenetic analysis allows for a deeper understanding of host ecologies and latent potential for vectors and the evolutionary traits of these bacteria. However, there are some limitations revealed in this preliminary investigation. Firstly, as previously described, the number of arthropod samples especially bat ticks was limited, except for wingless bat flies (Figure 1) [25]. In addition, a mixed operation of specimens and sampling bias towards ectoparasites complicated the interpretation of relationships between the bacteria and hosts. Secondly, although the specimens were collected alive from the body surface of stuck bats and left in gauze bags for up to 24 h to digest the sucking blood before being transferred to liquid nitrogen, we cannot entirely eliminate the possibility that the bacteria were detected from the blood of bats but have no infection of bat ectoparasites because of the no-cultivation of the organism. Lastly, few bat ectoparasite-associated references are contained for the MLSA analysis because of a large number of sequence gaps (gltA, ftsZ, or rpoB) among the published strains, which limits a complete phylogeny of bacteria from bats and their parasitic arthropods. With increasing sampling and a wider variety of bats and their ectoparasite, we expect additional Bartonella strains/species to be identified, and more available data can be analyzed towards ecologies and phylogenies of these bacteria and their hosts.

Ethics Statement
The procedures and protocols of sample collection and processing were reviewed and approved by the Medical Ethics Committee of the Yunnan Institute of Endemic Diseases Control and Prevention (20160002). All the experiments were performed with approval by the Biosafety Committee of the Yunnan Institute of Endemic Diseases Control and Prevention.

Sample Collection
Bats were captured using sticky nets in orchards and caves of 10 counties/cities/towns in central and southwest Yunnan Province, including Xiangyun County, Shuangbai County, Baoshan City, Tengchong City, Mangshi City, Wanding Town, Yongde County, Menglian County, Mojiang County, and Mengla County (Figure 1). All visible ectoparasite specimens were collected from the body surface of stuck bats carefully using tweezers, then the bats were set free. The collected bat ectoparasites were placed into tubes with records and transported back to the local laboratory soon as possible; then, they were left in gauze bags for up to 24 h to digest the blood. Initial identification was made using morphological features such as the head, abdomen, wings, and legs under a stereo microscope, and following this, they were sealed, frozen in liquid nitrogen, and transported to our laboratory and kept at −80 • C.

Sample Mixing, Deoxyribonucleic Acid Extraction, and Host Species Identification
The samples were poured into a precooled sterile grinding mortar and washed with 2 mL minimal essential medium (MEM); then, the liquid was discarded and we added 1 mL of MEM containing 10% penicillin-streptomycin solution. All bat ectoparasites were homogenized individually at low temperatures until the disappearance of the tissue mince [50]. The grinding fluid was centrifuged at 18,000 rpm and 4 • C for 30 min. All 434 supernatant fractions were combined into 40 pools on the basis of the information regarding morphological identification, collected date, and location. Accordingly, 284 wingless bat flies were mixed into 25 pools, 11 fly-like bat flies into 3 pools, 113 bat mites into 6 pools, 21 bat fleas into 3 pools, and 5 bat ticks into 3 pools (Table S1). A total volume of 200 µL of mixing homogenate was used to extract DNA using a TIANamp Genomic DNA Kit (Tiangen Biotech, Beijing, China). Subsequent species identification was based on the cytochrome c oxidase subunit I (COI) gene. The COI gene sequences of all mixing pools were obtained by Sanger sequencing as previously described (Table S4) [50,51], and the sequences were confirmed by comparison against the BOLD database (http://www.boldsystems.org/ (accessed on 7 September 2022)) and NCBI nucleotide database.

Bartonella Detection and Identification
Bartonella bacteria were detected by conventional PCR amplification and sequencing of several a variety of genes (gltA, rpoB, ftsZ, and ITS) primer pairs as described previously (Table S4) [52][53][54][55]. The pools were first screened with the partial gltA gene because it is the most common target for Bartonella detection and identification [43]. On the basis of the initial PCR result, the gltA-positive pools were also detected with the genes ftsZ, rpoB, and ITS to further confirm and characterize the bacteria. The PCR reaction was performed in a 25 µL mixture containing 12.5 µL 2 × DreamTaq Green PCR Master Mix (Thermo Scientific, Lithuania), 1 µL of 10 µM each forward and reverse primer (Sangon Biotech Co., Ltd., Shanghai, China), 8.5 µL nuclease-free water, and 2 µL sample DNA. PCR was performed with one denaturation cycle at 95 • C for 5 min; 40 amplification cycles at 94 • C for 30 s, 50 • C for 60 s, and 72 • C for 90 s; and an additional final extension at 72 • C for 10 min. All PCR products were separated on a 1.2% agarose gel electrophoresis and visualized by E-Gel Imager (Tanon 2500B) with GoldView staining, and those observed bands of the expected size were purified and subsequently sequenced by Sangon Biotech. Sequences were determined and assembled using the SeqMan program implemented in the DNASTAR software package (Lasergene). Assembled sequences were compared to known sequences in GenBank using the Nucleotide BLAST (BLAST + 2.13.0), and identification of the obtained Bartonella sp. was according to the homology.

Phylogenetic Analysis
The single locus, especially gltA, was used to analyze the phylogenetic resolution between Bartonella species and subspecies because this gene has good discriminatory power to delimit genotypes and species of these bacteria [44]. The gltA reference sequences, which contained the representative published strains identified from bats and their ectoparasites, were used for the classification of novel strains were collected from GenBank (Table S2). However, previous Bartonella-associated studies have shown the limitations of the gltA gene in individually resolving phylogenetic relationships because of the occurrence of genetic recombination [56], and therefore, we amplified and sequenced the additional two protein-coding loci (ftsZ and rpoB) to further characterize the phylogenetic traits using the multi-locus sequence analysis (MLSA) approach [19,53]. In contrast, the gene ITS was not included due to many sequence gaps among the reference strains. The validated Bartonella species/strains that were discovered in nine types of hosts at relevant taxonomic scales of the order level were used for MSLA, aiming to characterize the evolution of the bacterium (Table S3). MAFFT was used to align the nucleotide sequences [57], and the terminal sequences were removed manually and then pruned sequences using trimAl [58]. Phylogenetic trees were constructed using PhyML by the maximum likelihood method and a bootstrap value of 1000, with the GTR + G substitution model and SPR tree topology optimization algorithm [59]. The phylogenetic trees were illustrated by using ggtree objects implemented in RStudio.

Conclusions
Continuous discoveries of Bartonella from bat ectoparasites worldwide suggest these arthropods may serve as both natural reservoirs and potential connecting vectors. However, few investigations have focused on these bacteria towards bat-parasitic ectoparasites in China, especially bat fleas and ticks. We surveyed the Bartonella infection in a broad range of obligated bat ectoparasites, including bat flies, bat mites, bat fleas, and bat ticks, finding that these arthropod populations widely harbor diverse Bartonella species with a high prevalence ratio. Bacterial phylogenies combine the host's taxonomy, indicating the occurrence of reservoir spillover to bat ectoparasites. Moreover, a strong phylogeny-trait association between Bartonella sp. in these blood-sucking arthropods and bat hosts demonstrated the vector potential of ectoparasites. Nevertheless, the molecular epidemiological characteristics of these bacteria still have not been fully understood, which is recommended for follow-up surveillance of both bats and local human populations in this region and in even wider regions of Yunnan Province in China.

Supplementary Materials:
The following supporting information can be downloaded at https://www. mdpi.com/article/10.3390/pathogens11111283/s1, Table S1: Summary of sample mixed pools and the result of Bartonella detection. Table S2: GenBank accession and information of reference sequences for gltA. Table S3: GenBank accession and information of reference sequences for MLSA. Table S4: Primers used in the current study. Table S5: Sample distribution of 10 trapping sites. Figure S1: Maximum likelihood tree for 40 mixed pools of bat ectoparasites based on the partial nucleotide sequence of the COI gene.