|Home | About | Journals | Submit | Contact Us | Français|
Fleas collected from rodents in the Negev Desert in southern Israel were molecularly screened for Bartonella species. A total of 1,148 fleas, collected from 122 rodents belonging to six species, were pooled in 245 pools based on flea species, sex, and rodent host species. Two Bartonella gene fragments, corresponding to RNA polymerase B (rpoB) and citrate synthase (gltA), were targeted, and 94 and 74 flea pools were found positive by PCR, respectively. The Bartonella 16S-23S internal transcribed spacer (ITS) region was also targeted, and 66 flea pools were found to be positive by PCR. Sixteen different Bartonella gltA genotypes were detected in 94 positive flea pools collected from 5 different rodent species, indicating that fleas collected from each rodent species can harbor several Bartonella genotypes. Based on gltA analysis, identified Bartonella genotypes were highly similar or identical to strains previously detected in rodent species from different parts of the world. A gltA fragment 100% similar to Bartonella henselae was detected in one flea pool. Another 2 flea pools contained gltA fragments that were closely related to B. henselae (98% similarity). The high sequence similarities to the zoonotic pathogen B. henselae warrant further investigation.
Bartonellae are small Gram-negative bacilli belonging to the alpha-2 subdivision of the Proteobacteria. Different Bartonella species were detected in a wide range of vertebrate animals. There are currently 30 known species or subspecies, among which 14 have been associated with human diseases (7). Bartonella organisms are parasites of mammalian erythrocytes and endothelial cells and are transmitted by fleas and lice and potentially by other blood-feeding arthropods such as ticks and flies (2). Infection in the natural host commonly causes a chronic bacteremia, which is asymptomatic in most cases.
Rodents are being extensively studied and were found to have a high prevalence of Bartonella infection, with a high diversity of Bartonella spp. and strains (3). The close contacts between human and rodent populations around the world create excellent conditions for transmission of Bartonella spp. from animals to humans (28). The transmission routes of Bartonella bacteria by arthropod vectors among rodents and between rodents and other mammalian hosts have public health implications. In order to understand the extent to which rodents serve as source of human infections, investigations of rodent-borne Bartonella are essential (28). A few cases of human infections with Bartonella bacteria of rodent origin have been reported: B. elizabethae was associated with endocarditis, B. washoensis was associated with cases of myocarditis and meningitis, B. vinsonii subsp. arupensis was reported to cause fever and neurologic symptoms, and B. grahamii was isolated from the intraocular fluids of a patient with neuroretinitis (5, 11, 12, 25, 29).
An earlier survey carried out in the Tel Aviv region, Israel, demonstrated the occurrence of Bartonella strains closely related to B. elizabethae and B. tribocorum in commensal rats (Rattus rattus) (8). Another study has surveyed wild rodents and their fleas for Bartonella spp. in 19 geographical locations in Israel from the Upper Galilee in the north to Beer Sheba in the south. Bartonella DNA was detected in spleen samples of 19 out of 79 (24%) black rats (R. rattus), in 1 of 4 (25%) Cairo spiny mice (Acomys cahirinus), and in 15 of 34 (44%) flea pools collected from black rats (R. rattus) (21). The objectives of the current study were to screen fleas collected from rodents inhabiting the Negev Desert south to Beer Sheba for Bartonella infection and to compare Bartonella prevalences between male and female fleas.
Rodents were captured in the northern and central portions of the Negev Desert in October to December 2007. The climate of the region is characterized by hot and dry summers (mean daily temperature in July is 25 to 34°C) and relatively cool winters (mean daily temperature in January is 11 to 14°C). Annual rainfall ranges from 108 mm per annum in the northern part of the region to 25.8 mm per annum in its southern part. The dominant vegetation consists of Retama raetam (white weeping broom), Zygophyllum dumosum (desert shrub), Artemisia monosperma (bushy leafy shrub), Anabasis articulata (jointed anabis), Atriplex halimus (Mediterranean saltbush), Echiochilon fruticosum (North African shrub), Haloxylon persicum (white saxaul), and Calligonum comosum (abal) (18). Sherman live traps baited with millet seeds were arranged across each sampling plot in a 100- × 100-m grid using 25 or 50 traps per grid. Fleas were collected from a captured individual rodent by brushing the animal's fur thoroughly, using a toothbrush over a white plastic pan. Collected fleas were placed in Eppendorf vials with 70% alcohol and delivered to the laboratory for identification. Each rodent was identified by species, sexed, weighed (with a Pesola spring scale), marked by subcutaneously implanting a microtransponder (Trovan ID-100; United Kingdom), and released at the capture site. After counting and identification, fleas were grouped in pools ranging from 1 to 10 fleas based on their species, sex, and rodent host species. The study was approved by the Nature and National Parks Protection Authority (2007/27779).
Fleas were minced on a slide with a scalpel into minor pieces. DNA was extracted using the DNA extraction kit (Illustra Tissue Mini Spin kit; GE Healthcare, Buckinghamshire, United Kingdom) according to the manufacturer's instructions.
Molecular diagnosis of Bartonella DNA was carried out by conventional PCR targeting the 313-bp fragment of the citrate synthase gene (gltA) using primers Bhcs.781p and Bhcs.1137n (22, 26) and targeting the 16S-23S internal transcribed spacer (ITS) region (496 to 549 bp) using primers 321s and 983 (20) and by the high-resolution melt (HRM) real-time PCR analysis targeting a 194-bp fragment of the RNA polymerase B (rpoB) gene using primers 600f and 800r (21). Positive amplicons were purified using the PCR purification kit (ExoSAP-IT; USB, Ohio). DNA sequencing was carried out utilizing the BigDye Terminator cycle sequencing chemistry from Applied Biosystems (ABI), the 3700 DNA analyzer, and the ABI data collection and sequence analysis software. Further analysis was done using Sequencher software, version 4.8 for Mac (Gene Codes Corporation, Michigan). The obtained sequences were analyzed initially by BLAST through the NCBI's Mega-BLAST algorithm and further aligned with other Bartonella sequences.
Multiple alignments of the DNA fragments were generated using ClustalW 1.83 (27). Phylogenetic trees obtained from the alignments were compared by several programs from Phylip package v.3.68 (23). The Seqboot program was initially run with 100 data sets. Then, the DNAML (maximum likelihood) program was run on the Seqboot data sets using the default parameters of the DNAML program and 3-time jumbles. The Concense program was run to build the consensus tree. The NJplot program was used to draw the phylogenetic trees. All representative sequences of gltA from positive flea pools were translated to amino acid sequences and compared with each other and with closely related Bartonella species.
Newly identified sequences were deposited in GenBank under accession numbers GU354260, GU354263, GU354264,GU354268, GU354269, GU354270, GU354272 to GU354281, GU354282 to GU354306, GU937230, GU937231, GU937232, and GU937233.
A total of 245 flea pools were collected from 122 rodents: Leptopsylla algira fleas were collected from 11 Mus musculus (house mouse) mice; Synosternus cleopatrae fleas were collected from 63 Gerbillus andersoni allenbyi (Anderson's gerbil) gerbils, 7 Gerbillus pyramidum (greater Egyptian gerbil) gerbils, and 5 Meriones sacramenti (Buxton's jird) jirds; and Xenopsylla ramesis fleas were collected from 33 Meriones tristrami (Tristram's jird) jirds and 3 Dipodillus dasyurus (Wagner's gerbil) gerbils.
Ninety-four of 245 (38.3%) flea pools collected from 62/122 (50.8%) rodents contained Bartonella gltA fragments. Sixteen different Bartonella genotypes were identified in fleas collected from 5 different rodent species. Figure Figure11 describes the phylogenetic analysis of the identified Bartonella genotypes.
In 4 of 12 pools of L. algira fleas collected from 2 M. musculus mice (one male and one female flea pool from each rodent), identified Bartonella genotypes (deposited as GenBank accession number GU354260) were 96% similar to Bartonella sp. OE 1-1 (AB445001) (Fig. (Fig.11).
In 4 of 8 pools of S. cleopatrae fleas collected from 2 M. sacramenti jirds, Bartonella genotypes (deposited as GenBank accession number GU354270) were 96% similar to Bartonella sp. TL-sv4 (AJ583130). Both male and female flea pools collected from two M. sacramenti were found positive (Fig. (Fig.11).
In 55 of 116 pools of S. cleopatrae fleas collected from 35 G. andersoni allenbyi gerbils, nine different Bartonella genotypes were detected (Fig. (Fig.1).1). The most common Bartonella strain detected was 100% similar to Bartonella sp. OE 1-1, detected in 26/55 positive flea pools. Male and female flea pools collected from the same individual of G. andersoni allenbyi contained the gltA gene fragment (deposited as GenBank accession number GU354264) 98% similar to Bartonella henselae (Fig. (Fig.11).
In 6 of 14 pools of S. cleopatrae fleas collected from 3 G. pyramidum gerbils, two different Bartonella genotypes were detected. In two G. pyramidum individuals, male and female flea pools contained a gltA fragment 100% similar to Bartonella sp. OE 1-1. In another G. pyramidum individual, the male flea pool contained a gltA fragment (deposited as GenBank accession number GU354263) that was 95% similar to Bartonella sp. OE 5-1 (AB444992) (Fig. (Fig.11).
In 31 of 76 pools of X. ramesis fleas collected from 20 M. tristrami jirds, six different Bartonella strains were identified. In one female flea pool from M. tristrami, B. henselae DNA was detected (313 bp; 100% identity to GenBank accession number GU056191) (Fig. (Fig.11).
Sixty-six out of 245 (26.9%) flea pools collected from 59/122 (48.3%) rodents were found positive for the Bartonella 16S-23S ITS. All the ITS-positive samples were positive for gltA as well. The ITS sequences were grouped into two distinct phylogenetic clusters (Fig. (Fig.2):2): one cluster of Bartonella genotypes was identified from 28 flea pools collected from 26 gerbils (Gerbillus) only, and the second cluster of genotypes was identified from 37 flea pools collected from 33 jirds (Meriones) together with a single sample that was collected from a gerbil. The ITS sequences were deposited in GenBank with accession numbers GU354272 to GU354281, GU937232, GU937233, GU937231, and GU937230.
Seventy-four out of 245 (30.2%) flea pools collected from 54/122 (44.2%) rodents were found positive for the Bartonella rpoB gene fragment. All these pools were positive for the gltA fragment. Eleven different genotypes were identified in G. andersoni allenbyi, and 14 different genotypes were identified in M. tristrami (deposited as GenBank accession numbers GU354282 to GU354306).
The DNA sample detected in fleas collected from M. tristrami which contained a Bartonella gltA fragment that was identical to Bartonella henselae (GenBank accession number GU056191) contained an rpoB fragment (194 bp) that was 95% similar to B. henselae (new GenBank accession number GU354283). rpoB could not be amplified in the other DNA samples that contained gltA fragments that were 98% similar to B. henselae.
The gltA-based amino acid analysis indicated that most of the genomic substitutions between the sequences detected in this study were proteomically synonymous with a few nonsynonymous substitutions. In two genotypes (deposited as GenBank accession numbers GU354263 and GU354268), there were two amino acid substitutions compared to their closely related reference strains (Bartonella sp. OE 5-1, GenBank accession number AB444992, and Bartonella sp. OE 3-1, GenBank accession number AB444982, respectively). In the former, arginine substituted for lysine in position 108 and serine substituted for arginine in position 85, while in the latter asparagine substituted for serine in position 44 and asparagine substituted for aspartate in position 27. In two genotypes (new GenBank accession numbers GU354260 and GU354269), there was one amino acid substitution compared to their reference strains (Bartonella sp. OE 1-1, GenBank accession number AB445001, and Bartonella grahamii, GenBank accession number AY584857). In the former isoleucine substituted for asparagine in position 97 and lysine substituted for isoleucine in position 10.
Sixty-two out of the 122 (50.8%) rodents carried Bartonella DNA-positive fleas. In 62 out of the 94 (65%) flea pools which were found positive for Bartonella gltA, similar Bartonella DNAs were identified in both female and male fleas collected from the same rodent. In three cases, Bartonella DNA identified in male fleas was different from Bartonella DNA identified in female fleas collected from the same rodent host. Fourteen positive male flea pools were identified when the female pools collected from the same rodent were negative, and 12 positive female flea pools were identified when the male pools collected from the same rodent were negative.
In the current study, Bartonella DNA was detected in a high percentage of the flea pools (collected from individual rodents), suggesting a high prevalence of Bartonella spp. in rodent fleas in the Negev Desert in Israel. Moreover, about half of the rodents that were captured were infested with Bartonella- positive fleas, suggesting the high prevalence of Bartonella spp. in rodents in this region. Sixteen different Bartonella genotypes were identified in fleas collected from 5 different rodent species, suggesting that some rodent species possibly can harbor more than one Bartonella genotype. The clinical significance of these findings has yet to be elucidated.
The 313-bp gltA fragment used in this study was found to be a good target for screening fleas for Bartonella infection and for identification to species level. All flea pools that were found positive by rpoB or ITS screening were also positive by gltA. The Bartonella gltA sequence database in GenBank is the largest and most frequently updated among the different collections of Bartonella deposited sequences and therefore allows a more accurate differentiation between Bartonella species and strains. The gltA-based proteomic analysis indicated that most amino acid substitutions were synonymous, highlighting the important and critical function of the citrate synthase (gltA) enzyme in bacteria. Whether the substitutions detected in this study have any significance for the function of this enzyme warrants further investigation.
Interesting information was gained by analyzing the ITS locus among detected Bartonella genotypes. Except for one Bartonella genotype (which was found in one pool collected from a gerbil), the Bartonella ITS genotypes collected from M. tristrami grouped in one cluster, while Bartonella genotypes from flea pools collected from G. andersoni allenbyi grouped in a different cluster, suggesting that the ITS locus can assist in defining a group of Bartonella genotypes according to their rodent host. There may be at least three, not mutually exclusive reasons for the occurrence of two phylogenetic clusters. This can be related to differences in (i) rodent host species (Gerbillus and Dipodillus versus Meriones), (ii) flea vector species (S. cleopatrae versus X. ramesis), and (iii) habitat type (sand dunes versus loess plains). Indeed, in the study area, G. andersoni allenbyi is a sand dweller whereas M. tristrami inhabits loess plains. The characteristic flea of the former is S. cleopatrae, whereas the characteristic flea of the latter is X. ramesis (16).
In this study, a gltA fragment, identical to Bartonella henselae, was detected in one flea pool. The rpoB fragment of this sample was 95% similar to B. henselae strain Houston-1 (BX897699; AF171070) and to B. henselae that was previously detected in the aortic valve of a boxer dog (FJ832095). Another gltA fragment, highly similar to B. henselae, was detected in two other flea pools. Based on the 16S rRNA analysis, B. henselae was once previously reported in rodents (wood mice, Apodemus sylvaticus), in Denmark (6). The detection of Bartonella strains closely related to B. henselae, the agent of cat scratch disease, in rodent fleas suggests the zoonotic potential of these pathogens and warrants further investigation of their pathogenic capability.
Male and female fleas were found to be highly similar in their capacities to carry Bartonella organisms. In several flea pools, collected from the same rodent, male flea pools were positive while the female flea pools were negative and vice versa. As the frequencies of the latter findings were almost the same for male and female pools and were found in a relatively small number of pools compared to those for the rodents that carried positive male and female flea pools, these findings are probably incidental. However, as no characterization of Bartonella in the hosting rodent was carried out and no information on whether the fleas were hosting more than one host was available due to the characteristics of this survey, this enigma will remain unsolved. Previous studies have shown that the absolute amount of a blood meal taken by a male flea is smaller than that of a female flea meal (15, 17). Our results suggest that the difference in the amount of a blood meal between male and female fleas is not a critical factor for acquiring Bartonella bacteria. In three cases, different Bartonella strains were found in male and female flea pools collected from the same rodent. These observations could be explained by coinfection of multiple Bartonella strains in the same rodent host rather than by different associations between Bartonella genotypes and male or female fleas.
Bartonella genotypes that were detected in this study were similar or identical to Bartonella strains that were previously detected in different rodent species from a worldwide geographic distribution: Bartonella sp. OE 1-1 (AB445001), Bartonella sp. OE 5-1 (AB444992), Bartonella sp. OE 3-1 (AB444982), and Bartonella sp. OE 1-2 (AB444981) were previously detected in G. pyramidum in Egypt (9); Bartonella sp. FN 12-1 (AB445003) and Bartonella sp. FN 6-1 (AB444983) were previously detected in Pachyuromys duprasi rodents that were imported to Japan from a breeding farm in the Netherlands (9); Bartonella sp. 1-1C (EU551154) was previously detected in Rattus norvegicus in Taiwan (19); and Bartonella sp. TL-sv4 (AJ583130) was previously detected in Tatera leucogaster from South Africa (24). The explanation for the worldwide existence of similar genotypes is unknown and yet to be defined.
The most common Bartonella strain identified in this study was 100% similar to Bartonella sp. OE 1-1, which was previously detected in G. pyramidum from Egypt (9). In this study it was detected in fleas collected from 3 different rodent species: G. andersoni allenbyi, M. tristrami, and G. pyramidum. The close geographic proximity between Egypt and the Negev Desert in Israel suggests that these rodents share similar habitats and therefore similar fleas and Bartonella strains.
A study investigating the diversity of Bartonella spp. in Spermophilus richardsonii ground squirrels at multiple sites in Saskatchewan, Canada, suggested that in sites where dispersal is limited and rodent hosts are separated from their neighbors by barriers, a lower diversity of Bartonella genotypes might be expected, whereas in locations where a few rodent hosts share the same area, spillover and establishment of Bartonella genotypes from one rodent host to another are possible (10). The rodents included in this study live in two different habitats. G. andersoni allenbyi, G. pyramidum, and M. sacramenti are sand dwellers while M. tristrami, M. musculus, and D. dasyurus inhabit loess plains. The dispersal within each habitat is high and might be the reason for the great diversity of Bartonella spp. found in the current study.
Field and laboratory studies have suggested that some host specificity occurs among Bartonella spp. (13, 14). There is also evidence that a single species of Bartonella might infect more than one host species at a given site (2). In a recent study performed by our group, a single strain of Bartonella species closely related to both B. tribocorum and B. elizabethae was identified in rat spleens and in fleas collected from black rats (Rattus rattus) captured in different geographical locations in Israel (21). Our current study indicates a great diversity of Bartonella strains circulating among wild rodent fleas in the Negev Desert compared to Bartonella strains detected in fleas collected from commensal black rats in urban areas in Israel (10, 11). Recent studies have shown that fleas might carry a broader range of Bartonella spp. than do their hosts (1, 4). As no attempts were made in this study to isolate and characterize Bartonella spp. in the hosting rodents, we cannot determine their Bartonella hosting capacities in comparison to those of their fleas. The relation between the proximity to urban centers and the possible effect on Bartonella diversity needs further investigation, as wild rodents and their parasitizing fleas might be found to play a role in public health issues.
In conclusion, this study adds to the knowledge of the diversity of Bartonella spp. and strains in wild rodents in the Negev Desert in Israel. More research is required to better understand the diversity of Bartonella species in rodent communities and its relation to different geographical and climate conditions, as well as proximity to urban centers and populated areas. The finding of Bartonella spp. closely related to B. henselae, the agent of cat scratch disease, should warrant further investigation of these species.
This study was sponsored by the Niedersachsen-Israel cooperation fund (VWZN2176) and partly supported by ISF grants 297/07 (to G.I.S.) and 27/08 (to B.R.K. and I.S.K.).
We thank the Center for Absorption in Science, Ministry of Immigrant Absorption, for their financial support.
This is publication no. 687 of the Mitrani Department of Desert Ecology.
Published ahead of print on 27 August 2010.