Genetic diversity of Bartonella species in small mammals in the Qaidam Basin, western China

Investigation of the prevalence and diversity of Bartonella infections in small mammals in the Qaidam Basin, western China, could provide a scientific basis for the control and prevention of Bartonella infections in humans. Accordingly, in this study, small mammals were captured using snap traps in Wulan County and Ge’ermu City, Qaidam Basin, China. Spleen and brain tissues were collected and cultured to isolate Bartonella strains. The suspected positive colonies were detected with polymerase chain reaction amplification and sequencing of gltA, ftsZ, RNA polymerase beta subunit (rpoB) and ribC genes. Among 101 small mammals, 39 were positive for Bartonella, with the infection rate of 38.61%. The infection rate in different tissues (spleens and brains) (χ2 = 0.112, P = 0.738) and gender (χ2 = 1.927, P = 0.165) of small mammals did not have statistical difference, but that in different habitats had statistical difference (χ2 = 10.361, P = 0.016). Through genetic evolution analysis, 40 Bartonella strains were identified (two different Bartonella species were detected in one small mammal), including B. grahamii (30), B. jaculi (3), B. krasnovii (3) and Candidatus B. gerbillinarum (4), which showed rodent-specific characteristics. B. grahamii was the dominant epidemic strain (accounted for 75.0%). Furthermore, phylogenetic analysis showed that B. grahamii in the Qaidam Basin, might be close to the strains isolated from Japan and China. Overall, we observed a high prevalence of Bartonella infection in small mammals in the Qaidam Basin. B. grahamii may cause human disease, and the pathogenicity of the others Bartonella species needs further study, the corresponding prevention and control measures should be taken into consideration.

There were 45 small mammals of three species captured in farmlands, with a Bartonella infection rate of 40.00% (18/45). Twenty small mammals of three species were captured in forests, with an infection rate of 60.00% (12/20). Additionally, 10 small mammals of one specie were captured in meadows, with no Bartonella infection, and 26 small mammals of four species were captured in semi-desert areas, with an infection rate of 34.62% (9/26). Thus, the infections rates in different habitats were significantly different (χ 2 = 10.361, P = 0.016) ( Table 2).  We used the maximum likelihood (ML), neighbor-joining (NJ), minimum-evolution (ME), and unweighted pair-group method with arithmetic mean to construct phylogenetic trees and obtained the same results; thus, the ML method was used for further analyses. Phylogenetic trees were constructed based on the DNA sequences of the concatenations of gltA, ftsZ, rpoB and ribC genes (2483 bp), and all isolates clustered into four clusters, i.e., clusters I to IV (Fig. 2). Strains belonging to cluster I were closely related to B. grahamii, strains belonging to cluster II were closely related to B. krasnovii, strains belonging to cluster III were closely related to Candidatus B. gerbillinarum and strains belonging to cluster IV were closely related to B. jaculi. In cluster III, MM82QH-GEM was separated from the other three isolates by a long distance and showed 92.06% identity with reference Candidatus B. gerbillinarum, indicating that this isolate might be a new species of Bartonella 24 .
Overall, our findings indicated that B. grahamii, B. jaculi, B. krasnovi, and Candidatus B. gerbillinarum were prevalent in the Qaidam Basin, and that B. grahamii was the dominant Bartonella species. Bartonella was not detected in the small mammals from two of six trapping sites, and the distribution of Bartonella species showed obvious geographical differences. Moreover, B. grahamii and B. jaculi were distributed in Wulan County, whereas Candidatus B. gerbillinarum and B. krasnovii were distributed in Ge' ermu City (Fig. 3).
Interestingly, our study also showed an association between Bartonella species and small mammal species. B. grahamii was specific for Cricetulus longicaudatus, B. jaculi was specific for Allactaga sibirica, and Candidatus B. gerbillinarum and B. krasnovii were specific for Meriones meridianus. In addition, two Bartonella species, i.e., Candidatus B. gerbillinarum and B. krasnovii were isolated from different tissues in one Meriones meridianu. Candidatus B. gerbillinarum was isolated from the brain, and B. krasnovii was isolated from the spleen, suggesting that one rodent could carry more than one Bartonella species (Table 3). It indicated the coinfection phenomenon existed in Bartonella species, which was consistent with previous studies 25,26 . Phylogenetic analysis. Phylogenetic analysis based on gltA sequences showed that B. grahamii in the Qaidam Basin was mainly clustered into three clusters. Some strains from Cricetulus longicaudatus clustered with B. grahamii from Apodemus speciosus in Japan, some strains from Cricetulus longicaudatus clustered with B. grahamii from Ochotona curzoniae in China, and one strain from Microtus oeconomus was clustered separately, indicating the genetic diversity of B. grahamii prevalent in the Qaidam Basin.  (Fig. 4).

Discussion
Bartonella species are highly prevalent in small mammals worldwide, which have the close contact with humans. It is of great significance to investigate the epidemiological and ecological characteristics of Bartonella infection in small mammals from different areas. In this study, we observed the prevalence and genetic diversity of Bartonella species in small mammals in the Qaidam Basin. The infection rate of Bartonella species in small mammals was 38.61%, which was higher than that of 18.99% in Ochotona curzoniae in our previous study 22 , and higher than   www.nature.com/scientificreports/ that in most areas of China 27 , but lower than that in many other countries, including Russia (60-83%), Canada (48-90%), Netherlands (72%), etc 17 . In addition, the infection rate was significantly different in different habitats (farmlands, forests, meadows, and semi-desert areas), with the highest infection rates observed in forests. The nutritional requirements of Bartonella make it difficult to culture in vitro 28 . Generally, the spleen tissue is used in Bartonella culture. In this study, brain tissue was successfully used for Bartonella isolation for the first time. Additionally, we found that the positive rates in different tissues (spleens and brains) of small mammals did not differ significantly. Moreover, we detected two Bartonella species in the same small mammal, i.e., B. krasnovii in the spleen and Candidatus B. gerbillinarum in the brain, indicating the complexity of Bartonella infection in small mammals. A previous study indicated that there might be a high Bartonella coinfection rate in rodents 29 . Therefore, in order to further explore the coinfection of Bartonella species, additional studies of multi-tissue culture, multi-clone detection, and multiple PCR detection by using well-defined species or genotype PCR primer sets are needed 25,26,29 .
Several rodent-associated Bartonella species have been implicated as the causative agents of human disease. Here, we obtained four Bartonella species in rodents, B. grahamii, B. jaculi, B. krasnovii and Candidatus B. gerbillinarum. Especially, B. grahamii was detected in Cricetulus longicaudatus specificly, B. jaculi was detected in Allactaga sibirica specificly, Candidatus B. gerbillinarum and B. krasnovii were detected in Meriones meridianus specificly, indicating rodent-specific characteristics. B. grahamii was the dominant epidemic strain circulating in the Qaidam Basin, which was associated with neuroretinitis and cat scratch disease (CSD) in immunocompromised people 11,12 , suggesting that Bartonella species detected in Cricetulus longicaudatus may have the ability to cause human disease. It was reported that Bartonella species antibodies and DNA were detected in cerebrospinal fluid of cats and dogs 30,31 , suggesting the possible relationship between Bartonella infection and central nervous system disease. Bartonella isolation from the rodent brain supported this observation. Until now, B. henselae and B. quintana were reported to cause central nervous system infection 32,33 , however, the effects of B. grahamii on the central nervous system needs further investigation.
Through phylogenetic analysis based on gltA sequences collected from the GenBank, B. grahamii from Cricetulus longicaudatus in the Qaidam Basin was mainly clustered with B. grahamii from Apodemus speciosus in Japan 34 and Ochotona curzoniae in China 22 . Some B. grahamii prevalent in the Qaidam Basin might have high homology with strains from Japan, which was consistent with the previous study 34 . B. jaculi, Candidatus B. gerbillinarum and B. krasnovii have only been reported in Eygpt 35 and Israel 29 , which pathogenicity was not clear and needs further study. However, in this study, Bartonella infection was not detected in Ochotona curzoniae, possibly because of the differences in sampling sites and the small number of samples collected, which need further investigation.

Conclusions
This study provided a better understanding of the prevalence and genetic diversity of Bartonella species in small mammals from the Qaidam Basin. Four Bartonella species were detected in rodents, among which B. grahamii was the dominant strain, and potentially pathogenic to humans. Additionally, Bartonella were isolated from rodent brains for the first time, and two Bartonella species were detected in different tissues of the same rodent, which indicated the complexity of Bartonella infection and the necessity for multi-tissue culture, multi-clone detection and multiple PCR detection. Our results raise the potential threats to public health by the Bartonella species, and surveillance of Bartonella in animals and investigation of suspected clinical cases in humans need to be strengthened in the Qaidam Basin.

Materials and methods
Ethical statement. This study was approved by the Ethics Committee of Chinese Center for Disease Con- Bartonella culture. Approximately 20 mg of each spleen and brain sample from the small mammal was homogenized by adding 200 μL sterilized trypsin soy broth (BD Biosciences, Franklin Lakes, NJ, USA), plated onto two trypsin soy agars containing 5% (vol/vol) defiber sheep blood (BD Biosciences), and incubated at 37 °C in an atmosphere containing 5% CO 2 . Pure colonies of Bartonella species were obtained according to previous methods 22

Phylogenetic analysis.
The sequences generated in this study were submitted to the GenBank (accession numbers MT815283-MT815438). The nucleotide sequence homology was blasted against reported Bartonella species sequences in the GenBank using the BLAST program at the National Center for Biotechnology Information Website (http://blast .ncbi.nlm.nih.gov/Blast .cgi). Phylogenetic tree was created using the maximum-likelihood method with MEGA version 7.0, and bootstrap values were calculated with 1000 replicates 41,42 (incomplete and poor quality sequences were excluded from phylogenetic analysis). Brucella abortus was used as the outgroup.
Statistical analysis. The positive rates of Bartonella in different habitats, genders and tissues of small mammals were analyzed using the Chi-square test. All data were analyzed using SPSS 22.0 (SPSS, Inc., Chicago, IL, USA). P < 0.05 was considered statistically significant.
Consent to publish. All the authors consent to publish the article in its present form.

Data availability
The data supporting the conclusions of this article are included within the article.