Molecular epidemiology of Bartonella species isolated from ground squirrels and other rodents in northern California

Summary

Bartonella spp. are endemic in wild rodents in many parts of the world. A study conducted in two northern California counties (Sonoma and Yolo) sampling California ground squirrels (Otospermophilus beecheyi) and four other rodent species (Peromyscus maniculatus, P. boylii, P. truei and Neotoma fuscipes) led to the isolation of small Gram-negative bacilli which were identified as Bartonella spp. based on colony morphology, polymerase chain reaction–restriction fragment length polymorphism (PCR–RFLP) and partial gene sequencing. Overall, Bartonella spp. were isolated from the blood of 71% (32/45) of the ground squirrels and one third (22/66) of the other rodents. PCR–RFLP analysis of the gltA and 16S rRNA genes yielded seven unique profiles, four for the ground squirrels and three for the other rodents. Isolates from each PCR–RFLP profiles were submitted for partial sequencing. Ground squirrel isolates were most closely related to B. washoensis, whereas the other rodent isolates were closest to B. vinsonii subsp. vinsonii and B. vinsonii subsp. arupensis. Two of these three species or subspecies are known zoonotic agents.

INTRODUCTION

Bartonella are emerging pathogens of increasing human and veterinary medical importance. Members of this genus are small, curved, Gram-negative, rod-shaped bacteria that parasitize erythrocytes of their hosts [1]. Bartonella are isolated from a wide range of mammalian species, including humans. New species, many of them not yet fully described have been isolated from a wide range of rodents [2–4]. A common feature of most of these bacteria is transmission by arthropod vectors, especially fleas, as shown for cats and various rodent species [5–7]. Several rodent-borne Bartonella spp. are zoonotic causing a wide range of clinical manifestations from prolonged fever to endocarditis, such as B. elizabethae, B. grahamii, B. washoensis, B. vinsonii subsp. arupensis or Candidatus Bartonella volans [8–14].

Little is known on the distribution of Bartonella spp. in wild rodents, including ground squirrels (Otospermophilus becheeyi) in California, as only a limited number of studies were conducted in that state or neighbouring states [13, 15]. The objective of the present study was to evaluate the prevalence of Bartonella spp. in ground squirrels and other small rodents from northern central California and determine which Bartonella spp. were infecting these small mammals.

MATERIALS AND METHODS

Study area and sampling procedures

One of the study areas was located in a semi-rural residential subdivision on the eastern slope of the foothills in southern Sonoma County, California (122° 46′ W, 38° 26′ N). The specific trapping sites were two non-adjoining properties of ~2 ha each, at 420–450 m elevation, with a creek flowing through both properties year round [16]. The predominant vegetation included California live oak (Quercus agrifolia) and California bay trees (Umbellularia californica) that provided a dense leaf litter. The climate is Mediterranean, with cool, moist winters and hot, dry summers. The other study area was located in Davis, CA at two sites, one in south Davis and the other in west Davis, Yolo County (121° 44′ W, 38° 32′ N).

Rodents

From 28 October to 2 February 1998, 110 Sherman traps (H. B. Sherman Traps, USA) were placed once a month in five series of trap lines, each ~200 m long with a trap placed every ~8·5 m. Traps were set in the evening. The following morning, trapped animals were processed at the field site. Animals were anesthetized with methoxyflurane (Metofane; Schering-Plough, USA), weighed, measured and species, sex and age (adult, juvenile) were recorded. A minimum of 500 µl blood was collected via retro-orbital puncture in a Monoject EDTA samplette™ tube (Medtronic, USA). The animals were ear-tagged, then returned to the original trap location. Procedures followed an approved animal use and care protocol through the University of California, Davis [16].

Ground squirrels

Ground squirrels were trapped using Tomahawk live traps baited with oats at two sites on the Davis Campus of the University of California over a 2-week period in June 1999. Both trap sites were located in open fields along roadways, and the predominant vegetation was dry grassland, typical of the Central Valley region of California in the summer. Traps were set at night and checked daily. Captured ground squirrels were brought to a central processing location for examination and blood collection. Prior to euthanasia, animals were anaesthetized with CO₂, information on age and sex was recorded and blood was drawn by intra-cardiac puncture into 2 ml plastic EDTA tubes. This study was conducted under protocol 9275 approved by the University of California, Davis Institutional Animal Care and Use Committee (IACUC) [17].

Culture and isolation of Bartonella

Isolation of Bartonella spp. was performed for all samples. Frozen blood was thawed at room temperature and spun for 30 min at 5000  g . Supernatant was removed and 125 µl M199 isolation medium was added to the pellet, as described previously [18]. Blood was plated on fresh heart infusion agar supplemented with 5% fresh rabbit blood. For ground squirrel blood, two plates were inoculated, one with 250 µl and the other one with the rest of the volume ranging from 25 to 350 µl. Rodent samples were inoculated onto one agar plate using a 50:50 buffer/blood volume mixture) due to the small blood volume available. Agar plates were incubated at 35 °C in an aerobic atmosphere of 5% CO₂ for 3–4 weeks and checked regularly (at least twice a week) for bacterial growth. The typically observed Bartonella colonies became visible on the primary plates within 3–7 days. Colonies tentatively identified as Bartonella-like were streaked for isolation onto new agar plates. Most of the isolates appeared small, circular, raised, greyish-white or translucent, and slightly adherent to the surface of the agar. For DNA extraction, single colonies were picked and placed in sterile 1·5 ml vials containing 100 µl sterile double deionized water and frozen at −80 °C. The remaining colonies were harvested and placed in screw-cap vials with 0·5 ml brain heart infusion medium and stored at −80 °C.

DNA extraction

Crude DNA extracts suitable for use in polymerase chain reactions (PCRs) were obtained from the culture isolates. Samples frozen in water were allowed to thaw at room temperature before heating at 100 °C for 15 min. Vials were vortexed for 10 s while still hot, then spun in a micro-centrifuge for 10 min at 16 000 g at 4 °C. Five microlitres of the supernatant was placed into a sterile 500 µl vial containing 45 µl sterile double deionized water. This 1:10 dilution was used as template in all subsequent PCR reactions. The remaining supernatant was transferred to sterile vials and stored at −80 °C.

PCR–restriction fragment length polymorphism (PCR–RFLP)

Rodent and ground squirrel isolates were identified using PCR–RFLP analysis of the citrate synthase (gltA) and 16S rRNA genes, as described previously [19–21]. A 381 base-pair (bp) fragment of the gltA gene was amplified using two oligonucleotides homologous to the gltA gene of B. henselae Houston I (BhCS781.p and BhCS1137.n) [19, 21]. All PCR amplifications were performed with a PTC-200 DNA Engine (Bio-Rad, USA). PCR amplifications were performed with 5 µl sample in a mixture containing 50 mm KCl, 10 mm Tris–HCl, 1·5 mm MgCl₂, 0·001% gelatin, 0·1% Brij-35, 200 µm of each deoxynucleotide triphosphate, 0·5 µm of each primer, and 0·2 U thermostable Ampli-Taq DNA polymerase (ThermoFisher Scientific, USA). This mixture was incubated at 95 °C for 2 min and amplified for 40 cycles at 95 °C for 1 min, 50 °C for 1 min, and 72 °C for 1 min, and subsequently at 72 °C for 5 min. Amplified products were verified by gel electrophoresis on a 2% agarose gel in 0·5 × Tris-borate EDTA buffer followed by staining with ethidium bromide and visualization by a UV transilluminator.

The amplified product of the gltA gene obtained with the set of primers suggested by Norman et al. [19] was digested with TaqI (Promega, USA), HhaI, MseI and AciI (New England BioLabs, USA) restriction endonucleases. A 1400 bp segment was amplified using two eubacterial universal primers specific for the 16S rRNA gene: P8 (5’-AGAGTTTGATCCTGGCTCAG-3’) and Pc1544 (5’-AAGGAGGTGATCCAGCCGCA-3’) [22]. The amplified product of the 16S rRNA gene was digested with DdeI (Boehringer GmbH, Germany) restriction endonuclease. The digestion conditions used were those recommended by the enzymes' manufacturer. Banding patterns were compared with those of B. henselae (ATCC no. 49 882) and B. clarridgeiae (ATCC no. 51 734).

Species identification

The PCR products of the eight selected isolates used for DNA sequencing were purified with Microcon centrifugal filter devices (Millipore Corp., USA) and sequenced with a fluorescence-based automated sequencing system (Davis Sequencing, USA). The BLASTN program of GCG software (Wisconsin Sequence Analysis Package, v. 10; Genetics Computer Group, USA) was applied to determine the bacterial species and subspecies closest to the sequencing results of the Bartonella-positive samples, based on the DNA sequence similarity of 273 bp of the gltA gene. Then, the GAP program was used for sequence alignments and determination of the percentage of DNA similarity between the sequences of the gltA gene for each Bartonella-positive sample and the closest bacterial species and subspecies.

Statistical analysis

Univariate analysis by χ ² and Fisher's exact tests was applied to evaluate risk factors associated with Bartonella bacteraemia in rodents and ground squirrels. χ ² for trend was used to compare groups of rodents and ground squirrels by location, age and sex. Two-tailed Fisher's exact test was used for cell values of <5 when calculating odds ratios. The level of significance was established at P < 0·05.

RESULTS

One hundred and thirty-seven ground squirrels and other rodents, representing five different species, were captured from two main sites in northern central California (Table 1). The 76 rodents captured in Sonoma County included 35 deer mice (Peromyscus maniculatus), including 18 males and 26 adults, 32 dusky-footed woodrats (Neotoma fuscipes), including nine males and 21 adults, six brush mice (Peromyscus boylii) including three males and five adults and three pinyon mice (Peromyscus truei) including two males and one adult. Blood samples were also collected from 61 ground squirrels including 24 males and 29 adults, with 43 animals (21 males and 18 adults) captured in South Davis and 18 animals (three males and 11 adults) captured in West Davis, Yolo County.

Table 1. Rodents tested for Bartonella by geographical location, species, age and sex, northern California, 1996–1999

* Sex not recorded for one individual.

† Age not recorded for one individual.

‡ Sex not recorded for two individuals.

Blood cultures were performed from all 137 ground squirrels and other rodents. However, major bacterial or fungal contamination led to discarding of 10 rodent samples (7 N. fuscipes, 3 P. maniculatus) and 16 ground squirrel samples (14 from South Davis, 2 from West Davis). Isolates identified as possible Bartonella organisms, were recovered from 22 (33·3%) of the 66 non-contaminated rodent blood cultures and from 37 (82·2%) of the 45 non-contaminated ground squirrel cultures (Table 2). Colonies tentatively identified as Bartonella were slow growing, generally visible within 3–7 days of plating. Most appeared small (0·5–1·0 mm), raised, circular, greyish-white or translucent and were slightly adherent to the agar surface.

Table 2. Bartonella infection in wild rodents from northern central California, 1996–1999

* Three cultures contaminated were excluded.

† Seven cultures contaminated.

‡ Sixteen cultures contaminated.

The 22 rodent isolates were confirmed to be Bartonella by PCR of a fragment of the gltA gene. Bartonella isolates were obtained from four different rodent species: 15 (47%) P. maniculatus, three (12%) N. fuscipes, two (33·3%), P. boylii and two (66·6%) P. truei. The level of bacteraemia ranged from 10 to >8000 colony-forming units (c.f.u.)/ml. The median level of bacteraemia was 150 c.f.u./ml. After excluding the 10 animals for which blood cultures were contaminated, 16 (33·3%) of the 48 adults and four (22·2%) of the 18 juveniles were Bartonella bacteraemic. There was no statistically significant difference in age or sex in those rodents detected to be Bartonella bacteraemic (P > 0·05).

For the ground squirrels, 32 of the 37 culture-positive animals were confirmed to be Bartonella spp. by PCR of a fragment of the gltA gene. Overall, prevalence was 71% (32/45). Bacteraemia levels ranged from 4 to 1728 c.f.u./ml with a median of 228 c.f.u./ml. Bacteraemia prevalence did not differ significantly by trapping site, age or sex.

To assess the genotypic diversity in the 22 rodents and 32 ground squirrel isolates, a portion (381 bp fragment) of the citrate synthase (gltA) and 16S rRNA (1400 bp fragment) genes were amplified by PCR and analysed by RFLP [22]. For the gltA gene, digestion was performed using TaqI, HhaI, MseI and AciI restriction endonucleases, which allowed identification of six distinct profiles, four for ground squirrels and two for other rodents (data not shown). Similarly, analysis of the 16S rRNA gene revealed three unique profiles (one for ground squirrels and two for other rodents) following digestion with DdeI restriction endonuclease. Combining PCR–RFLP data for both genes led to seven different profiles (four for ground squirrels and three for other rodents).

To fully identify the Bartonella spp. or subspecies isolated, 1–2 isolates of each of the seven PCR–RFLP profile groupings were selected for partial sequencing of the gltA and 16S rRNA genes based on the strength and clarity of the gel banding pattern. The eight isolates selected included four obtained from ground squirrels (isolate nos. Sb20, Sb19, Sb9, Sb57), one from a deer mouse (Pm9804), one from a dusky-footed woodrat (Nf9730) and the two pinyon mouse isolates (Pt9809 and Pt9811). A 289 bp fragment of the gltA gene from these eight isolates was compared to Bartonella sequence data available in GenBank. The mice isolates were most similar to B. vinsonii spp., including B. vinsonii subsp. arupensis for the deer mouse (99·31% homology) and the dusky-footed woodrat (97·23% homology) and B. vinsonii subsp. vinsonii for the two pinyon mice (97·58% and 97·23% homology) (Table 3). For the four ground squirrel isolates, the gltA sequences were closest to B. washoensis (homology range: 96·2–100%) (Table 4). These data were also confirmed by the 16S rRNA sequences on 515 bp (99·2–100% homology with B. washoensis for the ground squirrels, but was less discriminative for the three B. vinsonii subspecies for the rodents). They were still the closest to B. vinsonii subsp. arupensis or B. vinsonii subsp. vinsonii (homology range: 99·2–99·8%) compared to B. vinsonii subsp. berkhoffii (homology range: 99–99·4%). In addition to known Bartonella spp., the rodent isolates were also closely related to isolates previously described from rodents of the same species [2]. The deer mouse isolate (Pm9804) was closest to strains D1, D2, and D7 also isolated from P. maniculatus P. leucopus and P. gossypinus in southeastern USA and the two pinyon mice isolates were closest to an isolate from a cotton rat (Sigmodon hispidus) B1 and a Peromyscus spp. D1 [2].

Table 3. DNA similarity values based on 289 bp of the citrate synthase gene (gltA) of the four selected other rodent strains compared to those of Bartonella strains in GenBank (bold >97%)

Table 4. DNA similarity values based on 289 bp of the citrate synthase gene of the four selected ground squirrel strains compared to those of Bartonella strains in GenBank (bold >96%)

All other strains described by Kosoy et al. [2], Ellis et al. [3] and Birtles & Raoult [4] presented in Table 3 had values of <95% similarity.

DISCUSSION

The present study reports the identification of B. washoensis in California ground squirrels, B. vinsonii subsp. vinsonii and B. vinsonii subsp. arupensis from four other rodent species from northern California. To our knowledge, this is the first report of the presence of a Bartonella species in pinyon mice (P. truei). The overall prevalence of Bartonella in rodents from Sonoma (33·3%) was comparable with reports from other regions of North America [2, 23, 24] and varied significantly among host species, which is similar to the pattern observed in other areas. For example, the high prevalence reported in Peromyscus spp. (33–67%) falls within the range observed in the United States [2, 24, 25]. More specifically, the strains isolated from rodents were quite similar to strains isolated from rodents in western USA [24, 25] and northern Mexico [26], and from ground squirrels and prairie dogs [13, 23, 24]. The isolates from rodents from Sonoma were also closely related to strains isolated from small rodents from the southeast of the United States including strains of group D described by Kosoy et al. [2], isolated from P. maniculatus and P. leucopus, suggestive of adaptation of these Bartonella spp. to their specific hosts. In contrast to the identification of B. vinsonii subsp. vinsonii in woodrats from New Mexico [27], our N. fuscipes isolate was closer to B. vinsonii subsp. arupensis, indicating that woodrats can be reservoirs for these two B. vinsonii subspecies. Two of the species isolated in northern California (B. vinsonii subsp. arupensis and B. washoensis) have been described as zoonotic pathogens [13, 14]. This warrants further studies to understand their potential role in causing human diseases in northern California, similar to what has been reported for urban rats in Los Angeles [15], where B. rochalimae, a known human and canine pathogen [28, 29], was isolated from 37 (18·5%) rats. Most rodent-borne Bartonella spp. are flea-borne [30], and several of the ground squirrels were flea infested at time of capture (data not shown). For instance, B. vinsonii subsp. vinsonii DNA was detected from Ctenophthalmus pseudagyrtes, a flea of insectivores and small rodents [30]. Similarly, in a human case of B. washoensis meningitis and early sepsis, Oropsylla montana fleas were implicated as the vector for disease transmission, as O. montana is the most common flea species to parasitize California ground squirrels and 8/11 flea pools were PCR positive for B. washoensis [31].

Based on our gltA results, it is likely that phylogenic group D described by Kosoy et al. [2] corresponds to B. vinsonii subsp. arupensis, with similar rodent species of the genus Peromyscus being infected. Similarly, Kosoy's phylogenic group B infecting cotton rats (Sigmodon hispidus), including strain Sh9282GH, is more likely to correspond to B. vinsonii subsp. vinsonii, as detected in our pinyon mice. It also suggests that strain SH9282 GA isolated from a rat (Rattus rattus) is closest to B. vinsonii subsp. vinsonii rather than to B. vinsonii subsp. berkhoffii [3]. However, our isolates are quite distinct from Bartonella isolates described either in rodents from Peru or in a wood mouse (Apodemus sylvaticus) from the UK [4]. In contrast to other reports [32], none of the rodents were co-infected with various Bartonella strains, as previously shown for ground squirrels, which appear to be an exclusive host of B. washoensis [13, 24] or P. maniculatus, usually infected with variants of B. vinsonii subsp. arupensis [25].