Background

Bartonella is a group of Gram-negative, fastidious, facultative, intracellular parasitic aerobic bacilli belonging to the class Proteobacteria, order Rhizobacteria, family Bartonellaceae and genus Bartonella that parasitize the erythrocytes and vascular endothelial cells of hosts and infect humans or other mammalian hosts through blood-sucking arthropods [1]. At least 40 species of Bartonella and its subspecies are currently recognized, of which at least 15 are human pathogens [2]. The clinical manifestations of Bartonella infection in humans range from mild to life-threatening and can be acute or chronic. Known symptoms of Bartonella in humans include endocarditis, myocarditis, fever and neurological disorders, intraocular retinitis, meningitis, splenomegaly and lymph node enlargement [3,4,5,6,7]. This constellation of nonspecific and variable symptoms make Bartonella infection difficult to diagnose clinically (Additional file 1: Table S1).

Rodents are natural hosts for approximately 20 species of Bartonella [8], and Bartonella has been detected in almost 100 rodent species worldwide. Importantly, a number of human pathogenic Bartonella species, such as B. elizabethae, B. grahamii, B. vinsonii subsp. arupensis and B. washoensis, are carried by rodents [9].

Bartonella is mainly transmitted horizontally [10], with arthropods acquiring Bartonella when blood feeding on an infected host with the subsequent transfer Bartonella to another host. Sand flies, body lice and cat fleas are involved in the transmission of B. bacilliformis, B. quintana and B. henselae, respectively [1]. Fleas are considered to be the primary vectors of Bartonella transmission among rodents, and a variety of fleas have been shown to be infected by zoonotic Bartonella species such as B. henselae, B. clarridgeiae, B. quintana, B. grahamii and B. elizabethae [11,12,13,14,15]. Fleas have been shown to play an important role in the transmission and acquisition of Bartonella species in rodents, and Bartonella DNA has been detected in fleas on rodents [16], providing evidence that fleas are vectors for the transmission of Bartonella among rodents.

Currently, 28 species of rodents belonging to seven families and 14 genera and 27 species of fleas belonging to six families and 18 genera have been identified in Fujian Province (China) [17, 18]. Previous systematic investigations conducted on 10 species of Bartonella host rodents harboring Bartonella in Southeast China identified Bartonella species in rodents, including B. tribocorum, B. grahamii, B. rattimassiliensis, B. queenslandensis, B. elizabethae, B. phoceensis, B. coopersplainsensis, B. japonica and B. rochalimae [19]. To date, however, no investigations have been conducted on Bartonella vectors. In the present study, we analyzed the epidemiological and molecular characteristics of Bartonella in fleas in Southeast China by investigating Bartonella infection in several areas of this region. Our aim was to assess the public health risk of the host-vector relationship between rodents and fleas on the transmission of Bartonella in the natural habitats of Southeast China.

Methods

Ethical aspects

This study was approved by the Ethics Committee of Fujian Center for Disease Control and Prevention (No: FJCDCNT1811-2015). All rodents were treated in accordance with the Guidelines of Regulations for the Administration of Laboratory Animals of the People's Republic of China.

Sample collection and identification

Rodents were captured in seven inland and coastal cities in Southeast China, namely Zhangzhou City, Quanzhou City, Sanming City, Longyan City, Ningde City, Fuzhou City and Putian City, and one to three fleas were collected from the body surface of each captured animal. Rodents were captured in live-capture traps baited with corn. Live traps were placed every night at each surveillance point for three consecutive nights at locations where rodent activities were detected, and retrieved the following morning.

Following capture, rodents were anesthetized with ether, and fleas were collected from the body surface of the rodents and from the cloth bags in which the rodents were held. Chinese monographs were used to identify the species of trapped rodents according to body shape, tail, coat color and other morphological characteristics [20]. The fleas were identified to species under the stereomicroscope by observing the distribution of setae and spines and the morphology of important structures such as eyes and genitalia by stereomicroscope, as well as by literature references [21]. We then individually recorded flea species, region, sex, host species and habitat. The fleas were morphologically classified and counted for registration and were stored in 75% alcohol at -20 °C until examination. After fleas had been collected, all rodents were used for surveillance in other programs.

Molecular analyses

Following published guidelines [22], before DNA extraction, each individual flea was immersed in 75% ethanol for 5–10 min, followed by two to three immersions in phosphate-buffered saline (PBS). The flea samples were then immersed in the lysate for 2 h and ground to a powder. DNA was extracted using a bacterial genomic DNA extraction kit (Tianlong Science & Technology, Xi'an, China) according to the manufacturer's instructions and stored at − 20 °C. DNA was extracted in order to identify Bartonella species using a real-time PCR (qPCR) assay targeting a transfer-mRNA gene (ssrA) [23]. The primers ssrA-F (5′-GCTATGGTAATAAATGGACAATGAAATAA-3′) and ssrA-R (5′-GCTTCTGTTGCTAGGTG-3′) and the FAM-labeled probe (FAM-ACCCCGCTTAAA CCTGCG-BHQ1) were used to amplify a 301-bp fragment of the ssrA gene. qPCR amplification was performed in a 20-μl reaction mixture containing 10 μl of Premix Ex Taq (Probe qPCR; Takara, Shiga, Japan), 0.4 μl each of 10 μM forward and reverse primers, 0.2 μl of 10 μM probe, 3 μl of DNA template and double-distilled water. The qPCR conditions were: 95 °C for 5 min; then 50 cycles of 95 °C for 15 s and 60 °C for 45 s. Samples with Ct (cycle threshold) values ≤ 35 were considered to be positive for Bartonella DNA. Positive samples were then subjected to conventional PCR to amplify the 379-bp gltA gene fragment [24] using the primers BhCS781.p (5′-GGGGACCAGCTCATGGT GG-3′) and BhCS1137.n (5′-AATGCAAAAAGAACAATAAACA-3′)[24]. The conventional PCR analysis was carried out in a total reaction volume of 25 μl containing 3 μl of template DNA, 1 μl each of 10 μM forward and reverse primers, 12.5 µl Premix Taq™ (Premix Taq Version 2.0 plus dye; Takara) and 7.5 µl double-distilled water. The amplification procedure was: 95 °C for 5 min; followed by 35 cycles of 95 °C for 30 s, 58 °C for 30 s and 72 °C for 30 s; with a final cycle at 72 °C for 5 min. The PCR products were separated by electrophoresis in a 1.5% agarose gel. During all PCR amplifications, distilled water was used as the negative control and positive DNA samples obtained from previous rodent surveys [19] were used as positive controls.

Bartonella ssrA sequences were sent to Sangon Biotech Company (Sangon Biotech, Shanghai, China) for gene synthesis to construct plasmid DNA. In addition, the Bartonella DNA load was calculated for each positive flea sample using a standard curve generated from a tenfold dilution (2log10-6log10 copies/μl) of plasmid DNA encoding a 300-bp B. henselae ssrA gene fragment.

DNA sequencing and phylogenetic analysis

Positive amplification products were subsequently sent to Sangon Biotech Company (Sangon Biotech) for sequencing.

The gltA sequences were compared with the sequences of the type strains of the validated Bartonella species in the GenBank database using NCBI BLAST (https://blast.ncbi.nlm.nih.gov/Blast.cgi). After alignment of the gltA sequences by ClustalW, phylogenetic trees were created using the neighbor-joining method in MEGA 11.0 software. The best-fit nucleotide substitution model for the phylogenetic analysis was estimated based on the Bayesian information criterion (BIC) calculated using MEGA 11 software [25].

Statistical analysis

The Chi-square test (χ2) and Fisher's exact test were used to evaluate the correlations between flea species, region, gender, host species, habitat and Bartonella infection. P < 0.05 was considered to indicate statistical significance. Analysis of variance (ANOVA) and the t-test were used to compare Bartonella loads.

All statistical analyses were performed using SPSS version 23.0 statistical software (SPSS IBM Corp, Armonk, NY, USA).

Results

Flea collection and morphological identification

A total of 1119 fleas were collected in seven cities during this survey, and seven species of fleas were identified (Table 1): Xenopsylla cheopis (n = 1047), Monopsyllus anisus (n = 38), Leptopsylla segnis (n = 24 ), Ctenocephalides felis (n = 6 ), Nosopsyllus nicanus (n = 1 ), Neopsylla dispar fukienensis (n = 1) and Stivalius klossi bispiniformis (n = 2). Among these, X. cheopis was the dominant flea species collected from the rats captured Southeast China, accounting for 93.6% (1047/1119) of the total fleas. A total of 308 fleas were from Zhangzhou city, 191 fleas were from Quanzhou city, 127 fleas were from Sanming city, 60 fleas were from Longyan city, 127 fleas were from Ningde city, 34 fleas were from Fuzhou city and 272 fleas were from Putian city.

Table 1 Flea collection from seven cities in Southeast China
Full size table

Detection and quantification of Bartonella spp. DNA

Bartonella-ssrA DNA was detected in 26.2% (293/1119, 95% confidence interval [CI] 23.6–28.8%) of the tested fleas from Southeast China (Table 2). Among the fleas found, 27.1% (284/1047) of the X. cheopis, 13.2% (5/38) of the M. anisus, 8.3% (2/24) of the L. segnis and 20.0% (2/10) of the ‘other’ fleas (N. nicanus, C. felis, S. klossi bispiniformis and N. fukienensis) were positive for Bartonella, with X. cheopis having the highest prevalence of infection and L. segnis the lowest. There was a significant difference in the prevalence of Bartonella among the different flea species (χ2 = 9.48, df = 3, P = 0.024). The infection rate of female fleas (28.9%, 217/750) was greater than that of male fleas (20.6%, 76/369), and there was a significant difference in the prevalence of infection between the sex (χ2 = 8.89, df = 1, P = 0.003).

Table 2 Molecular detection of Bartonella species in fleas from rats collected in Southeast China
Full size table

In this study, seven rodent species, namely Rattus norvegicus, Rattus flavipectus, Rattus losea, Niviventer coninga, Bandicota indica, Mus musculus, and Niviventer fulvescens, and one mammal species, Suncus murinus, were captured. When fleas from N. coninga, B. indica and N. fulvescens were not taken into account, the prevalence of fleas ranged from 16.7% to 29.8% (note: host species was not recorded in four fleas; Table 2). There was a significant difference in the prevalence among different hosts (χ2 = 18.948, df = 7, P = 0.008). Two or more fleas were captured from 245 hosts, with 14.5% (33/245) of these infected with Bartonella; 30.6% (75/245) were infected by only one flea and 55.9% (137/245) were not infected.

There was a significant difference in the prevalence of Bartonella in the different regions (χ2 = 75.23, df = 6, P < 0.001), with the highest incidence (33.3%, 20/60) occurring in Longyan City and the lowest prevalence (8.8%, 3/34) occurring in Fuzhou City (Table 2). In terms of geographical location of the seven cities investigated, Ningde City, Fuzhou City, Putian City, Zhangzhou City and Quanzhou City are located in the coastal area, and Sanming City and Longyan City are located in the inland area. The prevalence of Bartonella in the coastal cities was 26.4% (246/932) and that in the inland cities was 25.1% (47/187); the difference in prevalence among these two different geographic locations was not statistically significant (χ2 = 0.13, df = 1, P > 0.05). With the exception of four fleas from unrecorded habitats, 22.1% (202/915) of the fleas collected in wildernesses/farmlands were infected with Bartonella, and 45% of fleas (90/200) collected in residential areas were infected (Table 2). Fleas from residential areas had a significantly greater prevalence of Bartonella infection than did those collected in fields/farmland (χ2 = 44.62, df = 1, P < 0.001).

The prevalence of Bartonella in flea samples showed seasonal variation (Fig. 1), increasing from 16.4% in April to 26.7% in June, then decreasing to 16.8% in July, followed by an increase to a peak infection of 39.4% from August to October. The difference in Bartonella prevalence was significantly different between the different months (χ2 = 32.08, df = 6, P < 0.001).

Fig. 1
figure 1

Monthly prevalence of Bartonella in fleas in southeast China. Filled circles represent the prevalence of Bartonella

Full size image

A standard curve was established using plasmid DNA from the B. henselae ssrA gene fragment with r2 = 0.996, a slope of − 3.62, and a y-intercept of 40.42 (Fig. 2). The Bartonella loads of the positive fleas ranged from 1.35 to 8.29 log10 copies/μl (mean ± standard deviation [SD] 2.78 ± 1.14). Flea Bartonella loads were statistically significantly different among the different regions (F = 2.178, P = 0.045), with the highest flea bacterial loads occurring in Longyan city (mean ± SD, 3.19 ± 1.13) and the lowest occurring in Putian city (mean ± SD, 2.34 ± 1.14) (Fig. 3c). The fecal bacterial load in fleas caught in wildernesses (mean ± SD, 2.99 ± 1.20) was significantly higher than that in fleas caught in residential areas (mean ± SD, 2.70 ± 1.11) (t = − 2.010, P = 0.045) (Fig. 3e). Changes in flea bacterial loads over time showed a trend similar to that of prevalence and were significantly different (F = 3.148, P = 0.005) (Fig. 3d). Differences in flea Bartonella loads among flea species (F = 1.108, P = 0.346) (Fig. 3a), sex (t = 0.553, P = 0.581) (Fig. 3b) and host species (F = 1.977, P = 0.098) (Fig. 3f) were not statistically significant.

Fig. 2
figure 2

Standard curve based on Bartonella henselae ssrA gene fragment plasmid DNA. Tenfold serial dilutions of the plasmid vector DNA were performed (2log10-6log10 copies/μl), and real-time PCR analyses were repeated three times for each dilution concentration. The slope and intercept of the regression curve are shown. Ct Cycle threshold; lg, log

Full size image
Fig. 3
figure 3

Boxplot of Bartonella loads in positive samples from fleas of different species (a), sex (b), locations (c), time points (d), habitats (e) and hosts (f). Boxes represent IQRs, and vertical lines represent the distribution of maximum and minimum values. The values on the y-axis are expressed as log DNA copies/μl

Full size image

Sequence comparison and phylogenetic analysis

In total, 114 gltA sequences were analyzed via BLAST, and the phylogenetic analysis included sequences of 18 Bartonella genotypes, six Bartonella strains previously isolated from rodents in Southeast China and 26 representative flea samples from the present study. Brucella was also included as an outgroup (Fig. 4). The phylogenetic tree showed that the Bartonella-positive samples could be divided into five different branches. A total of 35.1% (40/114) of the gltA sequences belonged to B. tribocorum, which is the dominant genotype in Southeast China and is in the same branch as KT324580 in Thailand and MW771088 in Fujian, with 100% similarity. Seven sequences were detected as B. queenslandensis, with 95.5%-100% similarity to KT324558 from Thailand and MW771064 from Fujian. Twenty-three sequences of B. elizabethae were 99.1–100% homologous to JX158352 and GU056192 from Thailand and Taiwan, as well as to MW771077 and MW771078 from Fujian. Twenty-seven B. rochalimae sequences showed 100% similarity with those of MG027988 from the USA and MW771100 from Fujian. Nine B. coopersplainsensis sequences showed 94.3–98.5% similarity with HQ444160 from Australia and MW771106 from Fujian. Although the previously investigated rodents were not the hosts of the present flea samples, their Bartonella spp. were analyzed against the present samples, and the similarity reached 96.1–100%. Interestingly, of two or more fleas from the same host, four pairs were infected with the same Bartonella species: B. tribocorum, B. rochalimae and B. elizabethae.

Fig. 4
figure 4

Phylogenetic tree based on the gltA gene of Bartonella. The phylogenetic tree was constructed using the neighbor-joining method based on the maximum composite likelihood model, and bootstrap values were calculated with 1000 replicates

Full size image

The differences in flea Bartonella loads among the different regions were significantly different (P < 0.001). Bartonella coopersplainsensis-infected fleas were found to have higher bacterial loads (mean ± SD, 3.92 ± 0.57 log10 copies/μl) than other species. Among all species, B. elizabethae had the lowest load (mean ± SD, 2.13 ± 0.566 log10 copies/μl).

Discussion

Fleas are recognized as key players in the transmission of Bartonella, as they are able to carry a high diversity of Bartonella species and transmit them efficiently among rodents [26]. This efficient transmission of Bartonella is regarded as an important factor in maintaining its high prevalence in the natural environment. In China, there are relatively few investigations on ectoparasite infections caused by Bartonella. Li DM [27, 28], who detected Bartonella from the bacteria Chlamydophila felis and Leptopsylla segnis, isolated Bartonella strains from fleas and ticks. Bartonella infection in fleas has also been found in Qinghai Province, the Qinghai-Tibet Plateau and the China-Kazakhstan Border [29,30,31]. The present study emphasized the prevalent distribution of Bartonella in fleas and the related genotypes in Southeast China, with the data showing that there was a high prevalence of Bartonella in fleas in Southeast China and that multiple Bartonella genotypes could be identified.

The reported prevalence of flea Bartonella DNA detection in various countries varies, ranging from 2.2% to 40% in Egypt, the USA, France, Chile and Japan [9, 14, 32,33,34]. The overall Bartonella infection rate in fleas in the present study was 26.2% (293/1119) according to the qPCR analyses, which is higher than that reported our previous study of Bartonella infection in rodents in Southeast China (14.6–14.9%) [19, 35]. One factor for fleas possessing such a high infection rate may be their frequent feeding and ability to move from one host to another [36]. Our results also showed that B. tribocorum was the predominant genotype of Bartonella fleas in Southeast China; this species can cause causing acute fever and bacteremia in humans. Therefore, it is necessary to evaluate the epidemiological characteristics of Bartonella in fleas.

The transmission and acquisition of Bartonella are mediated by the host specificity of fleas, flea exchange between rodents and flea abundance [37]. According to our survey, X. cheopis was the dominant flea species in Southeast China and also the most prevalent flea. Our observations are similar to the results of a survey of rodents in the USA [38], which showed that the highest prevalence usually occurred among the most common species in rodent communities. These results indicate that the increase in the prevalence of Bartonella in fleas may also be related to the dominance of flea species in the population. Moreover, the density of the hosts may also increase flea transmission and infection among the hosts, as we mainly conducted surveys in villages and surrounding farmland where captured rodents were dominated by domestic rats, such as R. norvebicus, R. flavipectus and S. murinus, which were also accompanied by a high prevalence of parasitic flea infestations on their body surfaces (20.8–29.8%). Additionally, we found that residential areas have higher infection rates than wildernesses/farmlands, which undoubtedly increases the likelihood of flea contact with humans and disease transmission. In addition, animal sex has not been identified as a risk factor for Bartonella infection in rodents from Taiwan and France [39, 40]. However, in our study, we found that the incidence of Bartonella infection was significantly greater in females than in males, and this difference may be related to the parasitism and blood-sucking habits of fleas, with females sucking a greater amount of blood more frequently and for a longer period than males.

Several previous studies have shown that the prevalence of Bartonella in rodents and their ectoparasitic fleas is influenced by seasonality, peaking from the summer to fall [41,42,43]. In the present study, we found that the prevalence of Bartonella in fleas was markedly seasonal, with a clear upward trend in the prevalence of this genus from July to October. Late summer and early fall are not only periods of prevalence of Bartonella transmission but also periods of peak vector activity [44], making this period a risky time for Bartonella transmission to other species, including humans.

Phylogenetic analysis of the Bartonella gltA gene revealed five Bartonella genotypes, namely B. tribocorum, B. queenslandensis, B. elizabethae, B. rochalimae and B. coopersplainsensis, indicating the high diversity of Bartonella in the fleas of Southeast China. Bartonella tribocorum, B. elizabethae and B. rochalimae were the major genotypes identified in this survey, and all of them are pathogenic to humans, causing endocarditis, myocarditis, fever and neurological diseases. The high diversity of Bartonella genotypes may be a result of frequent host changes in fleas and their high efficiency in transmitting Bartonella. We compared the sequences of Bartonella species previously isolated from rodents in Southeast China with those isolated in the present study; the homology was 96.2%-100%, indicating the high adaptation of Bartonella species to rodents and fleas. In addition, Bowen et al. [11] reported that 75% (21/28) of bank voles housed with wild-caught fleas for 4 weeks developed Bartonella infections, and the present study also revealed multiple groups of fleas from the same host infected with the same Bartonella genotype at the same time, suggesting that fleas may play a potential role as vectors for the transmission of Bartonella among rodents. However, it is worth noting that the PCR detection of Bartonella spp. in fleas does not necessarily mean that they actively infest the host. Consequently, the mechanism of Bartonella spp. transmission between fleas and rodents still needs to be investigated more thoroughly.

Conclusions

The present study describes the prevalence and genetic characteristics of Bartonella species in fleas in southeast China. The results showed that there was a high prevalence and diversity of Bartonella in fleas. We identified five Bartonella genotypes in fleas, of which the zoonotic B. tribocorum, B. elizabethae, and B. rochalimae will pose a threat to human health in southeast China. However, the vector capacity of fleas was not determined in this study. In future studies, the host-vector relationship of Bartonella can be further investigated via animal experiments.