Main

The genus Bartonella consists of Gram-negative bacteria that parasitize erythrocytes and endothelial cells of mammalian hosts. Bartonellae are associated with diverse mammalian taxa (Chomel et al., 2009) and 12 Bartonella strains or species are classified as zoonotic agents (Chomel et al., 2006; Kosoy et al., 2008). Vector-borne Bartonella transmission has been reported in several systems (Chomel et al., 1996; Bown et al., 2004) and a variety of hematophagous arthropods including fleas, lice and sandflies have been implicated in Bartonella transmission among host individuals (Chomel et al., 2009).

Bartonellae are commonly detected in mammalian fleas and apparent specialization to particular mammalian hosts has been observed in Asia (Castle et al., 2004) and North America (Bai et al., 2008). In some cases, Bartonella detected in a host belongs to different species than the Bartonella detected in its fleas, and Bartonella can be detected in fleas collected from hosts that were apparently uninfected (Gabriel et al., 2009). This could result from fleas taking bloodmeals from multiple host species and Bartonella persisting and replicating in the flea gut (Finkelstein et al., 2002; Gabriel et al., 2009). However, many fleas are highly host-specific and are rarely detected on atypical hosts (Brinkerhoff, 2008). Low rates of host switching by fleas may therefore preclude acquisition of multiple Bartonella species unless fleas acquire Bartonella by mechanisms additional to the ingestion of infectious blood.

In Colorado grasslands, two dominant rodent species are the black-tailed prairie dog (Cynomys ludovicianus) and the North American deer mouse (Peromyscus maniculatus), each of which is strongly associated with one Bartonella species; B. washoensis in C. ludovicianus (Bai et al., 2008) and B. vinsonii subsp. arupensis in P. maniculatus (Bai Y, unpublished). Field sampling of over 4500 mammals revealed that the flea assemblages of these hosts are distinct with P. maniculatus parasitized primarily by Aetheca wagneri (∼93% of P. maniculatus fleas) and C. ludovicianus parasitized by Oropsylla hirsuta (∼98% of C. ludovicianus fleas) (Brinkerhoff, 2008). Carnivores in these grasslands acquire a variety of flea species and are commonly parasitized by Pulex fleas, which have wide host ranges, including rodents (Brinkerhoff, 2008). To explore relationships between Bartonella, fleas and mammalian hosts, we characterized Bartonella DNA sequences detected in fleas and blood collected from three mammalian host species.

Samples were collected from mammalian hosts (C. ludovicianus, P. maniculatus and red fox (Vulpes vulpes)) in Boulder County, Colorado, in 2005 and 2006 (methods in Brinkerhoff et al., 2008, 2009). Flea tissues were separated by dissection under stereomicroscopic observation. First, fleas were fixed to slides with commercial chemical bond and hemolymph was aspirated from between abdominal tergites. Second, the midgut and proventriculus were removed intact and were washed in sterile phosphate-buffered saline solution. Third, reproductive tissues were separated and washed in sterile phosphate-buffered saline. DNA from all tissues was extracted using the DNeasy tissue kits (Qiagen, Valencia, CA, USA). Amplification of the Bartonella citrate synthase gene (gltA) by PCR was used to detect the presence of Bartonella DNA (Norman et al., 1995), and 312 bp of the resulting amplicon was sequenced and aligned with known Bartonella species sequences (Inoue et al., 2008). For each Bartonella-positive tissue, four additional sequences were cloned and amplified from purified plasmid DNA. Bartonella in blood samples was cultured and identified using molecular methods (Bai et al., 2008).

Bartonella was detected in the blood of two P. maniculatus and in at least one tissue in 9 of 11 fleas (Table 1). Identification of Bartonella species was based on phylogenetic similarity to known Bartonella species; all DNA detected in this study matched to either the B. washoensis or the B. vinsonii arupensis clade with high bootstrap support (Figure 1). Five fleas carried Bartonella in multiple tissues, and DNA from two Bartonella species was detected in tissues of two fleas (Table 1, Figure 1). Cloning revealed that some fleas carried DNA of one sequence type, whereas others carried multiple sequence types (Table 1).

Table 1 Bartonella strains detected in multiple tissues (hemolymph, midgut and reproductive) of fleas collected from five individual mammalian hosts in Boulder County, Colorado
Full size table
Figure 1
figure 1

Neighbor-joining tree showing inferred relationships between Bartonella citrate synthase (gltA) amplicons derived from field-collected mammalian fleas and known Bartonella-type species. This phylogeny was constructed using the Kimura two-parameter model, and bootstrap values were calculated after 1000 replicates. Analysis was conducted using MEGA 4.0 software (Tamura and Dudley, 2007).

Full size image

B. washoensis is associated with ground squirrels and, in our system, is only known from prairie dogs (Bai et al., 2008). We detected B. washoensis-like DNA in fleas collected from prairie dogs, deer mice and fox (Table 1). B. v. arupensis is associated with Peromyscus mice (Jardine et al., 2005) and was detected in P. maniculatus blood samples, yet was also detected in prairie dog and deer mouse fleas (Table 1). Furthermore, Bartonella DNA was detected in five of six fleas collected from culture-negative hosts. These results suggest that fleas harbor richer Bartonella assemblages than their hosts, and that mammals may be exposed to a wider spectrum of Bartonella than those causing bacteremia. Abbot et al. (2007) reported similar patterns of coinfection in rodent fleas and suggested that host specialization may account for mammal–Bartonella relationships. Bartonellae can ‘jump’ among host species (Jardine et al., 2005; Bai et al., 2007) and fleas have been implicated in facilitating such events (Bai et al., 2007). Our results show that fleas can acquire Bartonellae that are characteristic of multiple host species; however, flea-mediated transmission of Bartonella between prairie dogs and deer mice seems unlikely as there is little flea exchange between these hosts (Brinkerhoff, 2008; Salkeld and Stapp, 2008). Note, however, that PCR detection of Bartonella in fleas does not necessarily indicate active infection in the host just as lack of Bartonella detection by culture of host blood does not rule out low-level infection.

Bartonella detection in reproductive tissue suggests that vertical transmission is a possible alternative mechanism to horizontal transmission in fleas. If vertical transmission of Bartonella within flea lineages exists, rare host-switching events could lead to the accumulation and maintenance of atypical Bartonella species in fleas. In a study of blood-sucking flies, 100% of pupae tested positive for Bartonella DNA (Halos et al., 2004), but we are unaware of any such reports in fleas. Bartonellae may also be transmitted through arthropod feces; louse feces have been shown to support B. quintana and are implicated in inoculating humans with this pathogen (Raoult and Roux, 1999). Adult flea feces can support Bartonella (Finkelstein et al., 2002) and are an important food source for larvae.

This project was funded by a fellowship to RJB from the East Asia and Pacific Summer Institutes program of the National Science Foundation (award no. OISE-0714395) and by the Japan Society for the Promotion of Science. A number of undergraduate students at the Nihon University College of Bioresource Sciences assisted with various aspects of this research. In particular, we would like to thank Yasuhito Izumi for demonstrating to us the technique for flea tissue separation.