Correspondence: Tatjana Avšič-Županc, Institute of Microbiology and Immunology, Medical Faculty, Zaloška 4, 1000 Ljubljana, Slovenia. Tel.: +386 1 543 7450; fax: +386 1 543 7401; e-mail: firstname.lastname@example.org
Rodents, collected in three zoogeographical regions across Slovenia, were tested for the presence of bartonellae using direct PCR-based amplification of 16S/23S rRNA gene intergenic spacer region (ITS) fragments from splenic DNA extracts. Bartonella DNA was detected in four species of rodents, Apodemus flavicollis, Apodemus sylvaticus, Apodemus agrarius and Clethrionomys glareolus, in all three zoogeographic regions at an overall prevalence of 40.4%. The prevalence of infection varied significantly between rodent species and zoogeographical regions. Comparison of ITS sequences obtained from bartonellae revealed six sequence variants. Four of these matched the ITS sequences of the previously recognized species, Bartonella taylorii, Bartonella grahamii, Bartonella doshiae and Bartonella birtlesii, but one was new. The identity of the bartonellae from which the novel ITS sequences was obtained were further assessed by sequence analysis of cell division protein-encoding gene (ftsZ) fragments. This analysis demonstrated that the strain is most likely a representative of possible new species within the genus.
Until 1995, parasites from the family Bartonellaceae of non human mammalian hosts, who were thought to be of no medical relevance, formed a species called Grahamella. The other species in the family was Bartonella, and these were considered human pathogens.
The application of polyphasic methods to characterize and compare members of each taxon indicated that they would be better accommodated in a single genus, and therefore their unification was proposed (Birtles et al., 1995). The availability of improved molecular techniques for the characterization of bartonellae led not only changes in taxonomy but also promoted the discovery of several new species. Today, an ever increasing number of bartonellae, infecting a wide range of mammalian hosts, continue to be encountered (Birtles et al., 1999; Breitschwerdt & Kordick, 2000). In Europe, surveys of small woodland mammals in the UK, Sweden, Denmark and Greece (Birtles et al., 2001; Holmberg et al., 2003; Engbæk & Lawson, 2004; Tea et al., 2004) have shown that Bartonella infections are common, and that at least four different species, Bartonella birtlesii, Bartonella doshiae, Bartonella grahamii and Bartonella taylorii, circulate in rodent communities. Several other Bartonella strains have also been encountered. They probably represent additional rodent-associated species but, to date, they have only been partially characterized, usually on the basis of sequence analysis of citrate synthase-encoding gene (gltA) fragments (Bermond et al., 2000; Birtles et al., 2001; Holmberg et al., 2003; Engbæk & Lawson, 2004; Tea et al., 2004).
Because of their fastidious nature, bartonellae are difficult to differentiate on the basis of phenotypic properties; for example, all are more or less inert in the standard biochemical tests used in bacterial identification. However, a number of PCR-based methods for the identification and delineation of Bartonella species have now been described, targeting a range of genetic loci including fragments of 16S rRNA gene, the 16S/23S rRNA gene intergenic spacer region (ITS), the citrate synthase-encoding gene (gltA), the 60-kDa heat-shock protein-encoding gene (groEL), a cell division protein-encoding gene (ftsZ), the RNA polymerase β subunit-encoding gene (rpoB) and others (Birtles et al., 2000; Haupikian & Raoult, 2001a, 2001b). The ITS region, which lies between the 16S and 23S rRNA genes within the rRNA-encoding operon, has been characterized in several Bartonella species and has been found to be somewhat larger than observed among other bacteria, ranging between 906 and 1529 base pairs (bp). As most of the ITS is noncoding, it is prone to hypervariability and thus interspecies and interstrain ITS sequence variation is markedly higher than that observed at other genetic loci (Roux & Raoult, 1995). Birtles and colleagues described the use of PCR-based amplification of ITS fragments to detect and identify bartonellae in the blood of rodents. Direct detection was of particular use in the longitudinal survey of bartonella bacteraemia that involved the field collection of very small amounts of blood from live wild-living rodents (Birtles et al., 2000). However, although comparison of ITS sequences is useful for the allocation of detected organisms into one of the recognized Bartonella species, detection of a novel ITS sequence can be problematic. The hypervariability of this region makes meaningful alignment of sequences from different Bartonella species difficult, compromising its use as a basis for phylogenetic inference. Other more conserved loci such as gltA or ftsZ can be used to complement ITS analysis for strain identification and can also serve as a basis for phylogenetic inference (Zeaiter et al., 2002).
In this study, an attempt was made to extend the survey of the prevalence and diversity of Bartonella species in European rodents by determining the diversity of organisms infecting animals from three ecologically-distinct regions of Slovenia.
Materials and methods
Trapping of rodents
Rodents were captured between 1996 and 2000 at six sites in three different zoogeographic regions of Slovenia: the Subpannonian, Prealpine and Mediterranean regions. Ecological conditions in Slovenia are heterogenous due to different abiotic (climatic, tectonic, edaphic, orographic and lithologic) conditions across the country, and on the basis of these parameters, the country has been divided into five zoogeographic regions that possess characteristic flora and fauna (Mršič, 1997). Differences between the fauna in each region, particularly the diversity and population dynamics of rodents and their ectoparasites, may influence the ecology of the Bartonella infections they maintain. The three regions included in the current study are predominantly covered with broad-leaved forest, with the addition of coniferous forest of boreal type in the Prealpine, oak forest in the Mediterranian region and elements of mountains in the Mediterranean and Subpannonian regions (Mršič, 1997). Animals were caught in overnight strategically-placed Elliot special and Sherman type traps and were euthanized with ether. Each animal was weighed, sized, sexed and identified as to species. In the laboratory, each animal was dissected and internal organs (heart, lung, liver, kidney, spleen, bladder) were collected and stored at −70°C until analyzed. From the captured animals we selected approximately the same number of every rodent species from each zoogeographic region.
DNA extraction and PCR amplification of Bartonella ITS and ftsZ fragments
We amplified Bartonella DNA directly from spleen samples, without prior bacterial cultivation. DNA extracts were prepared from a small piece (1 mm3) of spleen from individual animals using QIAmp® DNA Mini Kit (Qiagen, Gmbh, Hilden, Germany). Each sample of undiluted extracted DNA was incorporated into a Bartonella-specific PCR targeting an ITS fragment (Minnick & Barbian, 1997). The thermal programme employed was 35 cycles of 94°C for 1 min, 55°C for 1 min and 72°C for 1 min. The success of each amplification reaction was assessed by electrophoresis of an aliquot of the reaction mix after the thermal programme in a 2% agarose gel containing ethidium bromide and examination under UV illumination. Some DNA extracts were also incorporated into a Bartonella-specific PCR targeting a ftsZ fragment, as previously described (Zeaiter et al., 2002). The success of PCR was gauged as described above.
DNA sequencing and phylogenetic analysis
Sequencing of PCR products was carried out using the BigDye Terminator Cycle Sequencing Ready Reaction Kit (ABI PRISM, PE Applied Biosystems, CA), as per the manufacturer's instructions. The products of each reaction were resolved and translated into crude sequence data on an automatic DNA sequencer (ABI 310 Genetic Analyser, PE Applied Biosystems). These data were combined and analyzed using seqman and editseq software within the lasergene (2001) package (Dnastar, Madison, WI). The new sequences obtained were compared with sequences deposited in GenBank using the nucleotide blast program (National Center for Biotechnology Information). Alignment and phylogenetic analyses of ITS and ftsZ amplicons were performed using paup software, employing parsimony and distance methods (Version 4.0b10, Sinauer Associates, Inc. Publishers, Sunderland, MA). The stability of inferred phylogenies was assessed by bootstrap analysis of 1000 randomly generated sample trees.
Data were analyzed using spss for Windows, version 13.0 (SPSS Inc., Chicago, IL). The association between prevalence of infection and small mammal species, and between prevalence of infection and zoogeographic region, was evaluated using the χ2 test, Fisher's exact test, as appropriate. P-values were two-sided and considered significant at a level of <0.05.
A total of 146 rodents belonging to four species, A. agarius (striped field mouse), A. flavicollis (yellow-necked field mouse), A. sylvaticus (long-tailed field mouse) and C. glareolus (bank vole) were surveyed for bartonellae. All rodents were captured in each of the three regions surveyed, with the exception of A. agarius, which is not encountered at the Prealpine site. When possible, 15 representatives of each species in each region and approximately equal numbers of each gender were selected. Bartonella DNA was detected in 59 rodents. The number of infected animals varied according to the species of the rodent (the differences are statistically significant, χ2 with exact P=0.004) (Table 1). Apodemus flavicollis had the highest overall prevalence of infection (62.7%) and the highest prevalence in all three regions. The lowest overall and regional prevalence was found in A. agrarius (26.6%). Regionally, the highest prevalence was found in the Subpanonnian region (51.7%) (P=0.021, Fisher's exact test) (Table 1).
Table 1. Prevalence of infection with Bartonella in small mammals captured in three zoogeoraphic regions in Slovenia
The number of infected individuals and the number tested by species and locality are provided; prevalence of infection and percent are provided in parentheses.
Overall prevelence in a zoogeographic region and in rodent species are presented in bold.
Not present in this region.
Fourteen ITS fragments were selected for sequencing on the basis of size differences observed on 2% agarose electrophoresis and the provenance of the DNA extract from which the product was obtained. Comparison of the sequence data obtained demonstrated six sequence variants among them (Table 2). Four of these sequence variants were closely related to bartonella ITS sequences held on GenBank (Table 2). However, two (AF-s2, AA-m), although almost identical to one another (99.7% identity), were distinct from the ITS sequences of either the recognized Bartonella species or other partially characterized strains. The novel ITS region sequences could be aligned with that of Bartonella clarridgeiae, although they were somewhat longer, sharing only about 68% similarity.
Table 2. Description of selected samples for sequencing, according to small mammal species, zoogeographic region and length of amplicons
Region of trapping
Species of Bartonella
GenBank accession no. ITS/ftsZ
Accession numbers are assigned by GenBank.
AA, A. agrarius; AF, A. flavicollis; AS, A. sylvaticus; CG, C. glareolus. – m, mediterranean; -p, Prealpine; -s, subpannonian. 1, first isolate; 2, second isolate. For example, CG-s2 is the second isolate from the Subpannonian region in the C. glareolus species of small mammals.
Partial (781 bp) ftsZ sequences were obtained from five samples that yielded different ITS sequences. These sequences were aligned with one another and with other bartonella ftsZ sequences available in GenBank. One Bartonella variant was identical to B. grahamii and the other two were 99.4% and 96.4% similar to B. doshiae and B. taylorii, respectively. However, two novel genotype sequences were identical to each other and 95.6% similar to B. clarridgeiae. A phylogenetic tree, constructed using parsimony and distance methods, derived from ftsZ sequence alignment showed a topology consistent with a phylogenetic tree inferred from ITS sequence alignment (Figs 1 and 2).
The prevalence of Bartonella infections in European rodents has been found to range from about 17% in Sweden to over 60% in the UK (Birtles et al., 2001; Holmberg et al., 2003; Engbæk & Lawson, 2004; Tea et al., 2004). The overall prevalence of Bartonella infection in Slovenian rodents was 40.4%, although it varied between the three zoogeographic regions examined and the rodent species surveyed, from as high as 66.7% in A. flavicollis in two regions to as low as 13.3% in A. agrarius in the Mediterranian region. The difference in the prevalence of infection might be due to different habitat preference and abundance of the rodent species, which is related to population size and density. Apodemus agrarius does not share the same habitat as A. flavicollis and C. glareolus. Whereas the latter can be found mainly in the forest, A. agrarius prefers humid territories and forest borders and it can only be found in three isolated locations nationally (Kryštufek, 1991). Apodemus flavicollis is the most common rodent in Slovenia and was found to have the highest prevalence of infection, an observation in concordance with the hypothesis of Kosoy et al. that the dominant species in a community has the highest prevalence of infection (Kosoy et al., 2000). Longitudinal studies have suggested that seasonal dynamics could influence the prevalence in bartonellae infection in rodents (Bajer et al., 2001, Kosoy et al., 2004a). However, the animals tested in this study were all collected during late summer or early fall. It is most likely that several factors including the host–parasite interactions contribute to the observed differences in the prevalence of infection among species.
The prevalence of Bartonella infection in rodents in this study, as in others conducted in Europe, is remarkably high (Birtles et al., 1995, 1999, 2001; Breitschwerdt & Kordick, 2000; Tea et al., 2004). One hypothesis suggests that the cause of such a high prevalence might be long-term bacteraemia, as seen in rodent models and in cats infected with Bartonella henselae (Bouloius et al., 2001; Schulein et al., 2001; Kosoy et al., 1997). To some extent, the high prevalence of infection observed in our study could also be explained with the detection method used. As PCR does not only detect viable bacteria, it is possible that the prevalence of infection is overestimated, due to possible detection of remaining DNA from destroyed bacteria. However, genus-specific PCR performed directly on DNA extracted from samples and further sequencing of the amplified products has been very successful for the purpose of this study. We were able to bypass a labour-intensive and time-consuming cultivation step. The sample needed was very small. This is particularly important because it indicates that a small sample (blood) could be taken without the need to sacrifice or maim the animal (Birtles et al., 2000). However, a drawback in PCR performed directly on DNA extracted from samples is that not all primers are able to detect coinfections or even mixed infections, as the PCR products from different species can be of the same length. It was shown that the course of bartonellae infections could result from a single, persistent and potentially multi-genogroup/variant infection, during which variants differentially dominate the detectable bacteraemia (Kosoy et al., 2004b).
We have been able to confirm the presence of four recognized Old World rodent-associated bartonellae in Slovenia: B. taylorii, B. grahamii, B. doshiae and B. birtlesii. Furthermore, on the basis of comparative analysis of ITS and ftsZ fragments, one additional genotype that may represent a new Bartonella species was encountered. The genotype was most similar to B. clarridgeiae. However, the degree of dissimilarity between the new genotype and the published sequences of recognized species was well above that recently proposed as a cut-off value for Bartonella species definition (La Scola et al., 2003).
A large number of genotypes from only 15 sequenced samples confirms the great natural diversity of the genus. New species of Bartonella are demonstrated regularly from animals such as rabbits and cattle to a well studied reservoir like the cat (Droz et al., 1999; Heller et al., 1999; Maillard et al., 2004). This confirms previous suspicions about the lack of knowledge concerning the diversity of bartonellae (Birtles et al., 2000).
Many authors have implied that the host specificity between bartonellae and their rodent hosts is imperfect (Kosoy et al., 2000; Holmberg et al., 2003; Engbæk & Lawson, 2004). As we did not sequence all the amplified products it is difficult to make predictions of our own. However, the difference in length of the amplified products on the electrophoresis gel allows us to make some assumptions. The shorter fragments of about 600 bp which were assigned to B. doshiae were amplified only from C. glareolus, whereas the somewhat longer, about 700 bp, fragments, which turned out to be derived from a novel genotype, were amplified from A. flavicollis and A. agrarius. The other three Bartonella species produced fragments of approximately the same length (about 650 bp) differing by 1–30 bps, therefore it is impossible to predict their distribution in rodents. However, as such a high percentage of the sequenced amplicons turned out to be B. taylorii, it is likely that at the time of the survey this was the most common Bartonella species in rodents in Slovenia.
Until recently, only three Bartonella species were associated with human disease syndromes: Bartonella quintana, which causes trench fever, Bartonella bacilliformis, which causes Carrion's disease, and B. henselae, which causes cat scratch disease (Karem et al., 2000; Chomel et al., 2004). However, in the last 15 years several other Bartonella species have also been implicated as human pathogens, albeit to a far lesser degree than the three species mentioned above (Birtles et al., 1999). Among these is the European rodent-associated species B. grahamii, which has been associated with human neurotenitis and retinal artery occlusion (Kerkhoff et al., 1999). We confirmed the presence of B. grahamii in Slovenia and extended its host range to include A. agrarius.
Comparison of phylogenetic information resulting from ITS – and ftsZ-based genotypic studies resulted in phylogenetic trees comparable to phylogenetic results yielded in other studies (Haupikian & Raoult, 2001a, b). Some of the bootstrap values in the ITS phylogenetic tree are lower, due to a higher sequential diversity. However, FtsZ protein plays an important role in bacterial cell division, therefore its nucleotide sequence is much more conserved and better suited for phylogenetic analysis. It is important to note the close phylogenetic relationship of both sequences of the novel genotype with the recognized human pathogen B. clarridgeiae. To date, only one genotype, Af82Up, isolated from rodents in Sweden clusters with B. clarridgeiae based on gltA gene sequence analysis (Kordick et al., 1997; Holmberg et al., 2003). Cultivation of the novel strain described here would permit a more comprehensive phenotypic and genotypic characterization and would also permit the development of assays that could be used in investigating human exposure to the strain.