Vector transmission of Bartonella species with emphasis on the potential for tick transmission



    1. 1 Center for Comparative Medicine and Transitional Research, North Carolina State University College of Veterinary Medicine, Raleigh, North Carolina, U.S.A. and 2School of Veterinary Medicine, University of California Davis, Davis, California, U.S.A.
    Search for more papers by this author
  • 1 M. G. LEVY,

    1. 1 Center for Comparative Medicine and Transitional Research, North Carolina State University College of Veterinary Medicine, Raleigh, North Carolina, U.S.A. and 2School of Veterinary Medicine, University of California Davis, Davis, California, U.S.A.
    Search for more papers by this author
  • 1 B. B. CHOMEL,

    1. 1 Center for Comparative Medicine and Transitional Research, North Carolina State University College of Veterinary Medicine, Raleigh, North Carolina, U.S.A. and 2School of Veterinary Medicine, University of California Davis, Davis, California, U.S.A.
    Search for more papers by this author
  • and 2 E. B. BREITSCHWERDT 1

    Corresponding author
    1. 1 Center for Comparative Medicine and Transitional Research, North Carolina State University College of Veterinary Medicine, Raleigh, North Carolina, U.S.A. and 2School of Veterinary Medicine, University of California Davis, Davis, California, U.S.A.
    Search for more papers by this author

Dr Edward Breitschwerdt, North Carolina State University, College of Veterinary Medicine, Department of Clinical Sciences, 4700 Hillsborough Street, Room 454 Raleigh, North Carolina 27606, U.S.A. Tel.: + 1 919 513 8277; Fax: + 1 919 513 6336; E-mail:


AbstractBartonella species are gram-negative bacteria that infect erythrocytes, endothelial cells and macrophages, often leading to persistent blood-borne infections. Because of the ability of various Bartonella species to reside within erythrocytes of a diverse number of animal hosts, there is substantial opportunity for the potential uptake of these blood-borne bacteria by a variety of arthropod vectors that feed on animals and people. Five Bartonella species are transmitted by lice, fleas or sandflies. However, Bartonella DNA has been detected or Bartonella spp. have been cultured from numerous other arthropods. This review discusses Bartonella transmission by sandflies, lice and fleas, the potential for transmission by other vectors, and data supporting transmission by ticks. Polymerase chain reaction (PCR) or culture methods have been used to detect Bartonella in ticks, either questing or host-attached, throughout the world. Case studies and serological or molecular surveys involving humans, cats and canines provide indirect evidence supporting transmission of Bartonella species by ticks. Of potential clinical relevance, many studies have proposed co-transmission of Bartonella with other known tick-borne pathogens. Currently, critically important experimental transmission studies have not been performed for Bartonella transmission by many potential arthropod vectors, including ticks.


Bartonella spp. are gram-negative bacteria that reside within the alpha-proteobacteria class. These bacteria infect red blood cells and can invade endothelial cells, CD34+ progenitor cells and dendritic cells of their hosts, leading to persistent infections (Dehio, 2004; Boulouis et al., 2005; Mandle et al., 2005; Vermi et al., 2006). Several Bartonella species have been identified as zoonotic or potentially zoonotic agents, including: Bartonella henselae Regnery et al., Bartonella clarridgeiae Lawson & Collins, Bartonella alsatica Heller et al., Bartonella koehlerae Droz et al., Bartonella quintana Schmincke, Bartonella elizabethae Daly et al., Bartonella grahamii Birtles et al., Bartonella vinsonii subsp. arupensis Welch et al., Bartonella vinsonii subsp. berkhoffii Kordick et al., Bartonella washoensis Regnery et al. and Bartonella rochalimae Eremeeva et al. (Boulouis et al., 2005; Chomel et al., 2006; Raoult et al., 2006). As scientists, doctors and veterinarians learn more about the medical importance of the genus Bartonella, focus on known and suspected arthropod vectors increases. Because of their capability to reside within erythrocytes of a diverse number of mammalian hosts, there is substantial opportunity for various Bartonella spp. to be taken up by a variety of arthropod vectors. Recently, Bartonella DNA has been detected in blood samples obtained from loggerhead sea turtles (Caretta caretta Linnaeus), suggesting the possibility of persistent blood-borne infection in non-mammalian species (Valentine et al., 2007).

With regard to arthropods, it must be stressed, however, that there is an important difference between proven vector competence and vector potential. Documentation of vector competence is based upon experimental studies that demonstrate reliable transmission between the vector and the host. In most cases detection of Bartonella spp. in an arthropod, as determined by culture and/or polymerase chain reaction (PCR), does not provide definitive proof of vector competence and may merely represent the ingestion of Bartonella-infected blood from a bacteraemic host. Table 1 summarizes the known and potential vectors for transmission of various Bartonella spp. The vectors (Lutzomyia verrucarum [Townsend], Pediculus humanus humanus Linnaeus, Ctenocephalides felis [Bouché] and Ctenophthalmus nobilis nobilis [Rothschild]) of five Bartonella species will be reviewed and the potential for other arthropods to transmit these bacteria will be discussed. In addition, data supporting possible tick-borne transmission of Bartonella spp. will be summarized in this review. Molecular epidemiological surveys, human and canine Bartonella case reports and serological testing of dogs and people exposed to ticks provide strong evidence for tick transmission of these organisms. Experimental vector transmission studies must be performed, however, to validate the suppositions that ticks transmit Bartonella spp. to animals and human beings.

Table 1.  Known and suspected vectors for various Bartonella species.
Confirmed vectorSuspected vectorBartonella speciesReference:
Lutzomyia verrucarum (sandfly) B. bacilliformisBattistini (1929, 1931)
Lutzomyia peruensis (sandfly)B. bacilliformis and a novel Bartonella sp. resembling B. grahamiiEllis et al. (1999)
Pediculus humanus humanus (louse) B. quintanaSwift (1920)
Pediculus humanus capitis (louse)B. quintanaSasaki et al. (2006)
Rodent liceA novel rodent Bartonella sp., resembling B. henselae, B. tribocorum, B. phoceensis, and B. rattimassiliensisDurden et al. (2004); Reeves et al. (2006d)
Ctenocephalides felis (cat flea) B. henselaeKoehler et al. (1994); Chomel et al. (1996);
Ctenocephalides felisB. clarridgeiae, B. quintana and B. koehleraeRolain et al. (2003)
Ctenocephalides canis (dog flea)B. henselaeIshida et al. (2001)
Ctenophthalmus nobilis nobilis (flea) B. grahamii and B. tayloriiBown et al. (2004)
Rodent fleasResembling B. quintana, B. birtlesii, resembling B. clarridgeiae, B. elizabethae, B. koehlerae, B. doshiae, B. taylorii, B. tribocorum, B. vinsonii subsp. vinsonii, B. washoensis, and novel Bartonella spp.Parola et al. (2003); Stevenson et al. (2003); Durden et al. (2004); Reeves et al. (2005b); De Sousa et al. (2006); Loftis et al. (2006b); Marie et al. (2006); Abbot et al. (2007); Li et al. (2007); Reeves et al. (2007b); Reeves et al. (2007c)
Pulex spp. (human flea)Resembling B. vinsonii subsp. berkhoffii, a novel Bartonella sp., and B. quintanaParola et al. (2002); Rolain et al. (2005)
Sternopsylla texanus (bat flea)Novel Bartonella sp.Reeves et al. (2007c)
Various mites speciesRodent Bartonella sp. resembling B. grahamii, B. doshiae and a potentially novel Bartonella sp.Kim et al. (2005); Reeves et al. (2006a); Reeves et al. (2007a)
Lipoptena sp. (ked)Resembled B. schoenbuchensis, B. henselae, B. chomelii, and a cervid strain of BartonellaDehio et al. (2004); Reeves et al. (2006c); Halos et al. (2004)
Hippobosca equine (flies)Resembled B. schoenbuchensis, B. chomelii and a cervid strain of BartonellaHalos et al. (2004)
Melophagus ovinus (flies)Resembled B. schoenbuchensis, B. chomelii and a cervid strain of BartonellaHalos et al. (2004)
Haematobia sp. (biting flies)B. bovisChung et al. (2004)
Stomoxys sp. (biting flies)B. henselaeChung et al. (2004)

Sandfly transmission of Bartonella bacilliformis

The first Bartonella species to be described was Bartonella bacilliformis Strong et al. Arthropod transmission was proposed in the early 1900s. In 1913, C. H. T. Townsend hypothesized that Lu. verrucarum was the potential vector of B. bacilliformis, the agent of Oroya fever and verruga peruana (Townsend, 1914). Like other Bartonella species, B. bacilliformis infects erythrocytes; however, this species is somewhat unique in its ability to frequently induce a severe, life-threatening haemolytic anaemia. Initial support for the proposal that B. bacilliformis was vector-transmitted was based on the distribution and feeding habits of Lu. verrucarum relative to the distribution of cases of Oroya fever in the Peruvian Andes. Further speculation was heightened when a willing British seaman was exposed to wild sandflies. Initially, clinical symptoms appeared mild and after 4 months the individual left on a voyage. Once upon the ship, however, the seaman developed an intermittent fever and papules, presumably caused by verruga peruana. When the seaman returned to Peru, most of the symptoms had resolved and B. bacilliformis infection could not be confirmed as the cause of illness (Townsend, 1914).

After Townsend’s initial studies in 1913–14, several researchers focused their efforts on Lutzomyia spp. and other arthropods as potential vectors of B. bacilliformis. In 1928, Noguchi et al. published a manuscript detailing inoculation of monkeys (Macaca mulatta [Zimmerman]) with triturated bodies of Lutzomyia species, ticks, mites and other arthropods that were collected in known B. bacilliformis endemic areas. Bartonella bacilliformis was only cultured from the blood of sandfly-inoculated monkeys, but no lesions were apparent in these animals. In a separate experimental study, sandfly-cultured B. bacilliformis did induce nodular formations at intradermal inoculation sites, similar to the lesions reported in human bartonellosis cases (Noguchi et al., 1929).

Battistini (1929, 1931) was the first to establish direct transmission of B. bacilliformis by sandfly feeding. Twenty-three sandflies were released within an enclosure and allowed to feed on a rhesus monkey. Within 18 days, blood cultures became positive for B. bacilliformis. The species of Lutzomyia used in this study are unknown (Battistini 1929, 1931). In another experiment performed by Hertig (1942), wild-caught sandflies were permitted to feed on monkeys for several days, after which blood cultures demonstrated the presence of B. bacilliformis in these animals. The author identified the sandflies as unfed female Lu. verrucarum. Immunity experiments were also conducted by intradermal inoculation of several monkeys with cultured B. bacilliformis after initial Lu. verrucarum feedings. No nodules were produced at sites of inoculation, indicating that some immunity was conferred by prior B. bacilliformis infection induced by sandfly transmission (Hertig, 1942).

Insufficient information is available regarding the replication or survival of B. bacilliformis within the sandfly. When sandflies were fed upon infected patients, B. bacilliformis-like organisms were visible within the mid-gut, adhering to the surface of the intestines, and were also found in sandfly faeces (Hertig, 1942). Furthermore, the proboscis of many of the sandflies contained large quantities of small, rod-like organisms, similar in appearance to B. bacilliformis. Additional experiments demonstrated the ability to culture B. bacilliformis from the proboscis of two female sandflies, although the majority of the cultures were either negative or contaminated by other bacteria or fungi (Hertig, 1942). In addition, morphologically similar organisms were also apparent in male sandflies, which do not take bloodmeals, and also in unfed females. Based upon these results, it was speculated that transmission of the Bartonella-like organisms among sandflies occurred through commingling of breeding areas, contaminated water supplies and various other locations. The two cultures of B. bacilliformis obtained from the proboscis caused nodule formation at sites of inoculation in a previously uninfected monkey.

Outbreaks continue to occur in B. bacilliformis endemic and Lu. verrucarum non-endemic areas, leading to implications that other Lutzomyia sandflies or other arthropods can serve as potential vectors. Ellis et al. (1999) demonstrated that 1% of 104 wild-caught Lutzomyia peruensis (Shannon) contained B. bacilliformis by PCR analysis. Furthermore, DNA from a potentially novel Bartonella sp., resembling B. grahamii (96% similarity), was identified in another Lu. peruensis in that study (Ellis et al., 1999). However, the DNA sequence recovered from the sandfly was never deposited in GenBank and therefore no further comparison to newly characterized Bartonella bacteria has been performed. Research should continue to explore potential vectors of B. bacilliformis in non-endemic areas and define improved methods for the control of arthropod vector populations.

Louse transmission of Bartonella quintana

Infection with the agent of trench fever, Bartonella (Rochalimaea) quintana, is common in individuals displaced from their homes as a result of war, poverty and drug or alcohol abuse. Recent outbreaks have been reported in homeless individuals in Marseille (Fournier et al., 2002), the Netherlands (Fournier et al., 2002), Tokyo (Sasaki et al., 2002), rural Andean communities (Raoult et al., 1999), Moscow (Rydkina et al., 1999), various countries in Africa (Fournier et al., 2002), and in the U.S.A. in Seattle, Washington (Jackson & Spach, 1996) and the San Francisco Bay area, California (Koehler et al., 1997). Infection with B. quintana typically causes a cyclic 5-day fever accompanied by malaise and severe bone and joint pain. Endocarditis, generalized lymphadenopathy and bacillary angiomatosis in immunocompromised individuals are other frequent manifestations of B. quintana infection (Brouqui & Raoult, 2006). Historically, infection with B. quintana was thought to be limited to people with human body louse exposure. Although the mode of transmission is unknown, B. quintana was isolated from a non-human research primate (Macaca fascicularis Raffles), from dogs with endocarditis and from feral farm cats that had presumably induced B. quintana infection in a woman by bite transmission (O’Rourke et al., 2005; Kelly et al., 2006; Breitschwerdt et al., 2007b).

Pediculus humanus humanus has been the identified vector of this Bartonella species for several decades. As with sandfly transmission of B. bacilliformis, little research regarding louse transmission of B. quintana has been published in recent years. During World War I, however, a great deal of interest was focused on a relapsing fever that affected soldiers fighting in the trenches (Swift, 1920). Grätzer (1916), a doctor with the 84th Austrian Infantry Regiment, was the first to suggest a possible arthropod vector as a means of transmission. Clinical symptoms generally occurred in the winter when soldiers were often confined to close quarters, increasing the likelihood of transmission by an insect. Researchers from Germany, England and the U.S.A. demonstrated that development of trench fever-like symptoms in human patients occurred after they had been fed upon by infected lice (Swift, 1920). Swift (1920) further established that the trench fever organism, then referred to as a virus, could be transmitted to non-infected patients by escharification of the skin or injection into subcutaneous tissue with infected louse faeces. Within 5 days of feeding on an infected person, louse excreta became infectious (Bruce, 1921).

Although initially described as a virus, small cocci or bacilli, measuring 0.3−0.5 μm by 1.5 μm, were observed in the blood of infected patients and in louse faeces (Swift, 1920). Scientists Töpfer, Jungmann and Kuczynski, and da Rocha-Lima, described Rickettsia bodies within the intestinal mucosa and faeces of infected lice (Swift, 1920). It was not certain at the time, however, if the Rickettsia bodies were the cause of trench fever. Subsequent experiments also revealed that the trench fever agent was not transmitted transovarially to the offspring of infected lice (Bruce, 1921). When faeces from offspring of infected lice were escharified into the skin of non-infected patients, no clinical symptoms developed.

Scientists additionally performed immunity and challenge experiments on both naturally and experimentally infected human patients and hypothesized that only partial immunity was acquired against a second bout of trench fever. For example, Bruce (1921) describes experiments in which naturally infected patients were later inoculated with infected lice faeces. Patients were inoculated either through escharified skin or subcutaneously, at an average of 114 days after their initial bout of trench fever. Five of the eight individuals inoculated became re-infected. It was unknown, however, whether these infections constituted a relapse or a second bout of trench fever. Interestingly, individuals infected 443 days prior to louse feeding were still able to transmit the infection to non-infected lice (Bruce, 1921).

Other researchers subsequently confirmed louse transmission findings reported by scientists during or shortly after World War I. Weyer (1960) published a comprehensive review discussing the relationship between louse infestation and B. quintana transmission. In his review, Weyer (1960) states that B. quintana, referred to as Rickettsia quintana at the time, replicates extracellularly within the louse stomach and attaches to luminal epithelial cells. Based upon earlier research, Weyer (1960) also observed that louse longevity is not affected by the intraluminal presence of B. quintana in the stomach. Following intrarectal inoculation of laboratory-reared lice, Vinson et al. (1969) demonstrated viable B. quintana in the louse gut lumen using a Giemsa-stain. Furthermore, B. quintana was also visualized in faeces collected from lice feeding on an infected patient for xenodiagnostic purposes (Vinson et al., 1969). More recently, Fournier et al. (2001) demonstrated that green fluorescent protein-expressing B. quintana remained within the louse intestinal lumen and were excreted in louse faeces throughout the lifespan of an infected human body louse. Additionally, Fournier et al. (2001) reported that B. quintana is not transmitted transovarially based upon the inability to culture or PCR amplify the 16S−23S intergenic spacer region (ITS) in eggs and larvae obtained from infected lice. Seki et al. (2007) demonstrated logarithmic growth of bacteria within the louse gut and faecal matter. Within the midgut, 2 × 103 bacteria/louse were detected on day 3 and bacterial numbers increased until day 17 (maximum of 1.3 × 108 bacteria). Within the faecal matter, a maximum number of 1 × 107 bacteria were detected on day 15. Transmission of B. quintana by Pe. h. humanus occurs when adult lice become infected by way of a bloodmeal, viable organisms are maintained in the louse intestinal tract, and subsequent transmission to humans occurs by way of contamination of the louse bite site or a wound with contaminated louse faeces.

Other louse species have been identified by PCR as potential vectors of various Bartonella species. Recently, B. quintana has been detected in head lice, Pediculus humanus capitis de Geer, removed from children in Nepal (Sasaki et al., 2006). Furthermore, two sucking lice, Neohaematopinus sciuri Jancke (one nymph and one adult) and one pool of Hoplopleura sciuricola Ferris, removed from grey squirrels (Sciurus carolinensis Gmelin), harboured bacteria genetically related to Bartonella species found in other rodents (Durden et al., 2004). Interestingly, one pool containing four nymphal N. sciuri contained a Bartonella sp. closely related to B. henselae (99.6% similarity) (Durden et al., 2004). Rodents collected in Egypt, including Rattus rattus (Linnaeus) and Rattus norvegicus (Berkenhout), harboured two louse species: Polyplax spinulosa (Burmeister) and Hoplopleura pacifica Ewing that contained three Bartonella species known to infect rodents: Bartonella tribocorum Heller et al., Bartonella phoceensis Gundi and Bartonella rattimassiliensis Gundi (Reeves et al., 2006d). However, B. phoceensis was detected only in Ho. pacifica collected from a B. rattimassiliensis-infected rat. From this observation, the authors proposed that B. phoceensis might be transmitted by Ho. pacifica, but not Po. spinulosa (Reeves et al., 2006d).

Flea transmission of Bartonella species

Ctenocephalides felis has been shown experimentally to be a competent vector for transmission of B. henselae, the agent of cat scratch disease (CSD) (Chomel et al., 1996). Fleas were collected from bacteraemic cattery cats and placed on five specific-pathogen-free (SPF) kittens. Laboratory-reared kittens were negative for the presence of B. henselae DNA based on culture and immunofluorescent assay analysis prior to flea exposure. Fleas were placed on the kittens and within 2 weeks four of the five kittens were found to be bacteraemic with B. henselae. However, no negative controls were utilized within the experiment and therefore it cannot be definitively ascertained if the kittens had not been exposed to B. henselae prior to flea contact (Chomel et al., 1996). In an earlier study, Koehler et al. (1994) detected Bartonella DNA in fleas by PCR and cultured B. henselae from cat fleas collected from bacteraemic cats. In another study, Higgins et al. (1996) demonstrated that acquisition of B. henselae occurred within 3 h after Ctenoc. felis were fed on infected cat blood covered by a parafilm membrane. In this study, the bacteria remained in the flea gut for up to 9 days. Qualitative analysis demonstrated that bacterial replication also occurred in the flea gut as the numbers of Bartonella increased by day 9 post-feeding. Bartonella DNA was detected in flea faeces by PCR on day 9 and culture of flea faeces generated viable colonies on agar plates (Higgins et al., 1996). In 1998, Foil et al. successfully transmitted B. henselae via inoculation of infected flea faeces to five cats. Using a streptomycin-resistant strain of B. henselae, Finkelstein et al. (2002) observed bacterial levels of 1.80 × 103 CFU/mg in flea faeces at 2 h after collection and 3.33 × 102 CFU/mg after 72 h. These studies demonstrated that Bartonella organisms remain reproductively viable in flea faeces within the environment and that transmission to humans (CSD) most likely occurs as a result of inoculation of B. henselae-contaminated flea faeces into the skin via a scratch by a flea-infested cat (Finkelstein et al., 2002). In a study from Japan, two Ctenocephalides canis (Curtis) removed from dogs harboured B. henselae, and one C. canis collected from La Rioja, Spain also contained Bartonella DNA (Ishida et al., 2001; Blanco et al., 2006).

An experimental study involving wild-caught Ctenophthalmus n. nobilis suggested that these fleas are competent vectors for transmission of B. grahamii and Bartonella taylorii Birtles et al. to bank voles, Myodes glareolus (Schreber) (Bown et al., 2004). Fleas were removed from wild bank voles and placed on laboratory-reared bank voles for 4 weeks. Analysis by PCR demonstrated that 21 of the 28 voles became positive for Bartonella spp. using both 16S−23S ITS and citrate synthase gene (gltA) primers. DNA sequencing of PCR amplicons demonstrated that 16 voles were infected with B. taylorii and six voles were infected with B. grahamii (Bown et al., 2004). It is not known, however, if transmission occurs via inoculation of contaminated faeces, ingestion of infected fleas, or following regurgitation of Bartonella organisms while feeding, as occurs with flea transmission of Yersinia pestis Lehmann & Neumann (Chomel et al., 1996).

Ctenocephalides felis, previously mentioned in the transmission of B. henselae, also appears to be a potential vector for B. clarridgeiae, B. quintana, and B. koehlerae (Rolain et al., 2003). Bartonella henselae and/or B. clarridgeiae DNA has been detected in fleas found in the U.S.A. (Koehler et al., 1994; Chomel et al., 1996; Reeves et al., 2005b; Lappin et al., 2006), France (La Scola et al., 2002; Rolain et al., 2003), the U.K. (Shaw et al., 2004), Spain (Blanco et al., 2006), New Zealand (Kelly et al., 2004), Thailand (Parola et al., 2003), Japan (Ishida et al., 2001) and Hungary (Sréter-Lancz et al., 2006). It is possible that cat fleas (Ctenoc. felis) can maintain infection with B. quintana and transmit the organism among cats, which can subsequently transmit B. quintana to people via a bite or scratch (Breitschwerdt et al., 2007b).

Other flea species might also be important vectors for the transmission of Bartonella spp. among animals. Durden et al. (2004) examined Orchopeas howardi (Baker), a flea species commonly found on grey squirrels, S. carolinensis. One flea contained a Bartonella that was previously isolated from the same squirrel species (S. carolinensis). A pool of three Orc. howardi also harboured a Bartonella sp. related to B. quintana (Durden et al., 2004). Orchopeas howardi collected from eastern grey squirrels in a separate study harboured a Bartonella that closely resembled Bartonella birtlesii Bermond et al., a bacterium commonly found in mice (Bermond et al., 2000; Reeves et al., 2005b). During an epizootic of plague (Yersini pestis) in Colorado, 555 fleas were collected from two sites that contained numerous abandoned prairie dog (Cynomys ludovicianus [Ord]) burrows and were tested by multiplex PCR for the presence of Bartonella spp., Rickettsia spp. and Y. pestis (Stevenson et al., 2003). Fleas that were PCR positive for Bartonella species were identified as Oropsylla hirsuta (Baker) and Oropsylla tuberculatus cynomuris (Jellison). Sequencing of the PCR products demonstrated that the bacteria were most closely related to B. washoensis. Of the fleas harbouring Bartonella, 23 also contained Y. pestis DNA, the agent of bubonic plague (Stevenson et al., 2003). This observation supports the potential of co-transmission of Bartonella species and Y. pestis to animals or humans in plague-endemic regions.

A more extensive study performed by Reeves et al. (2007b) detected potentially novel Bartonella species from fleas collected from prairie dog burrows in North Dakota, Oklahoma, Texas and Wyoming (U.S.A.). Fleas that were PCR positive were identified as Oro. hirsuta, Oro. tuberculatus cynomus and Thrassis fotus (Jordan). Furthermore, one Oro. hirsuta from Oklahoma harboured both Bartonella sp. and Rickettsia sp. DNA, although the novel strains were not fully characterized (Reeves et al., 2007b). Another survey performed by Reeves et al. (2007c) examined rodent and bat fleas for the presence of Bartonella DNA. Four of 13 Xenopsylla cheopis removed from a Norway rat (R. norvegicus), two of two Ctenophthalmus pseudagyrtes Baker removed from a vole (Microtus sp.), one of 11 Sternopsylla texanus (Fox) collected from Brazilian free-tailed bats (Tadarida brasiliensis (I. Geoffroy), and one of 75 Orc. howardi removed from southern flying squirrels (Glaucomys volans [Linnaeus]) all harboured Bartonella DNA. The Bartonella species detected in the X. cheopis was identified as B. tribocorum, and B. vinsonii subsp. vinsonii Weiss & Dasch DNA was detected within Ctenoph. pseudagyrtes. All remaining PCR positives were described as potentially novel Bartonella species (Reeves et al., 2007c). Twenty-one Polygenis gwyni (Fox) were collected from eight cotton rats (Sigmodon hispidus Say & Ord) and two Eastern woodrats (Neotoma floridana [Ord]) from Georgia were also shown to harbour Bartonella DNA (Abbot et al., 2007). Twenty of 21 fleas produced the predicted PCR amplicon, whereas 12 of 16 fleas harboured mixed Bartonella DNA sequences using gltA specific primers. Sequences obtained were similar to B. vinsonii-like genogroups found in other cotton rat studies and B. clarridgeiae (Abbot et al., 2007).

Studies from outside the U.S.A. have also reported Bartonella in a variety of flea species. Bartonella quintana was detected in Pulex irritans Linnaeus removed from a pet Cercopithecus cephus (Linnaeus) monkey in Africa (Rolain et al., 2005). Furthermore, two of 30 Pulex sp., removed from humans in Peru, were found to contain a novel Bartonella species most closely related to B. rochalimae obtained from an ill patient returning from Ecuador (Parola et al., 2002; Eremeeva et al., 2007). One of the Pulex sp. also harboured a Bartonella closely related to B. vinsonii subsp. berkhoffii using ITS primers (Parola et al., 2002). Pulex irritans, removed from carcasses of red foxes (Vulpes vulpes [Linnaeus]), were found to harbour a Bartonella species closely related to Bartonella detected in X.cheopis (Rothschild) fleas (Sréter-Lancz etal., 2006). One Nosopsyllus fasciatus (Bosc) removed from a yellow rajah rat, Rattus surifer, in Thailand was also shown to harbour a unique Bartonella spp. Using phylogentic analysis, this Bartonella organism was grouped with B.grahamii (Parola etal., 2003). A rodent survey performed in Kabul, Afghanistan demonstrated infection of rodent fleas with B.quintana, B. elizabethae, B. koehlerae, Bartonella doshiae Birtles and B.taylorii (Marie etal., 2006). Two Bartonella spp. closely related to B.elizabethae were also described in four fleas, three Stenoponia tripectinata (Tiraboschi) and one Ornithophaga sp, that were removed from mice and a rat from Portugal (De Sousa etal., 2006). In another recent study, a novel Bartonella species was also identified in several X.cheopis, vector of Y.pestis, and in Leptopsylla segnis (Schönherr) removed from rodents in three cities throughout Egypt (Loftis etal., 2006b). One group of 33 X.cheopis and five of five groups of Ctenophthalmus lushuiensis Gong collected from rodents or rodent burrows in Yunnan, China were shown to harbour Bartonella DNA (Li etal., 2007). The positive X.cheopis sample was identified as B. tribocorum, whereas one of the Ctenoph. lushuiensis PCR amplicons was most closely related to B. clarridgeiae. The remaining Ctenoph. lushuiensis samples were shown to harbour Bartonella DNA closely related to other rodent Bartonella species (Li etal., 2007). Again, it must be stipulated that vector competence of the various fleas described in these recent studies has not been determined.

Other potential or suspected vectors for Bartonella transmission and miscellaneous arthropods

Keds, biting flies, mites and miscellaneous arthropods

Until recently, very little research involving mites, keds and biting flies harbouring Bartonella spp. had been published. Lipoptena spp., or deer keds, generally feed on deer but have also been found on horses, cattle and humans (Dehio etal., 2004). Bequaert (1953) was the first to describe a ‘Rickettsia melophagus’ (potentially B.melophagus) present within the midgut of all examined Hippoboscidae, suggesting a potential endosymbiotic relationship. Since that initial study, Halos etal. (2004) performed a survey involving mites, keds and biting flies. Lipoptena cervi (Linnaeus) (keds), Hippobosca equina Linnaeus (flies), and Melophagus ovinus (Linnaeus) (flies) were collected from both wild and domestic ruminants in France and Romania. All three Hippoboscidae species harboured Bartonella DNA: 94% of 48 adult Li.cervi; 100% of 12 adult Hi.equina, and 100% of 20 adult and 10 pupae M.ovinus. Both adults and pupae were recovered for all three species; however, only M.ovinus demonstrated a 100% infection of larval and adult stages, strengthening the supposition that these bacteria may serve as endosymbionts (Halos etal., 2004). Bartonella spp. detected in these fly specimens resembled Bartonella schoenbuchensis Dehio etal., Bartonella chomelii Maillard etal., and an unknown cervid strain. Interestingly, Dehio etal. (2004) demonstrated localization of B.schoenbuchensis in the midgut of Li.cervi collected from deer from Germany. These authors speculated that B.schoenbuchensis might be the cause of deer ked dermatitis in humans (Dehio etal., 2004). Using the riboflavin synthase gene, Reeves etal. (2006c) detected a Bartonella spp. closely related to B.schoenbuchensis and B.henselae in Lipoptena mazamae Rondani collected from deer in South Carolina, U.S.A.

A similar study performed by Chung etal. (2004) described Bartonella spp. in biting flies removed from cattle in northern California. Horn flies (Haematobia sp.), stable flies (Stomoxys sp.), deer flies (Chrysops spp.), and horse flies (Tabanus spp.) were examined for the presence of Bartonella DNA. Using gltA primers, these authors demonstrated that a horn fly pool contained DNA that was identified as Bartonella bovis and a stable fly contained B.henselae (Marseille type) DNA (Chung etal., 2004). Bartonella DNA was not detected in either deer flies or horse flies in this study.

The bat fly, Trichobius major Coquillett, and the Eastern bat bed bug, Cimex adjunctus Barber, are known ectoparasites of bats. One T.major, collected from Florida caverns and one of 14 C.adjunctus, collected from the Santee Caves located in South Carolina, were shown to harbour a unique Bartonella species (Reeves etal., 2005a). The Bartonella species found in T.major genetically resembled a Bartonella described in the eastern grey squirrel (Reeves etal., 2005a).

Four pools of mesostigmatid mites from Korea, removed from rodents, harboured a Bartonella sp. closely related to B. doshiae (Kim etal., 2005). Recently, Reeves etal. (2006a) demonstrated that a Steatonyssus sp., a mite removed from a bat, harboured a Bartonella sp. (96%) closely related to an unnamed Bartonella found in rodents. This DNA sequence was also 96% similar to B.grahamii. Reeves etal. (2007a) also detected a potentially novel Bartonella species from eight pools of Ornithonyssus bacoti (Hirst) removed from rats, R.rattus, in Egypt. The Bartonella sp. was most closely related to a Bartonella species (81%) described in Egyptian fleas using primers specific for the groEL gene.

The presence of Bartonella DNA has also been examined in several non-biting arthropods. Bartonella species have been described in house dust mites, Dermatophagoides farinae Hughes and Dermatophagoides pteronyssinus Trouessart (Valerio et al., 2005). These authors speculated that Bartonella infestation of house dust mites contributed to mite allergen endotoxins because of the high abundance of Bartonella lipopolysaccharides. Honey bees, Apis mellifera capensis Eschscholtz, also harbour a Bartonella sp. closely related to B. henselae (Jeyaprakash et al., 2003). These authors hypothesized that the Bartonella organisms present in honey bees were ingested or acquired through contact with the environment. Bartonella 16S DNA could not be amplified from eggs; therefore the Bartonella does not appear to be an endosymbiont in the honey bee (Jeyaprakash et al., 2003).

Molecular detection of Bartonella in ticks

Recently, there has been considerable interest in ticks as potential vectors for Bartonella species. To date, a handful of epidemiology surveys involving diverse tick populations have been performed. A number of tick species have been shown to be positive for Bartonella spp. based mainly on PCR and very rarely by culture (Table 2). In the U.S.A., two studies carried out in California have demonstrated that Ixodes pacificus Cooley & Kohls, Dermacentor occidentalis Marx and Dermacentor variabilis Say might serve as potential vectors for various Bartonella spp. (Chang et al., 2001, 2002). Twenty-nine of 151 (19.2%) individually tested questing I. pacificus ticks were PCR and sequence-positive for the gltA gene (Chang et al., 2001). By DNA sequencing of this gene, the ticks were shown to harbour Bartonella spp. that were closely related to an unnamed Bartonella of cattle and others that were related to several human pathogenic species: B. henselae, B. quintana, B. washoensis and B. vinsonii subsp. berkhoffii (Chang et al., 2001). Two ticks harboured more than one Bartonella sp.: one contained B. henselae and B. vinsonii subsp. berkhoffii and the other contained B. henselae and a cattle Bartonella strain (Chang et al., 2001). In a second study, 1119 I. pacificus, 54 D. occidentalis and 30 D. variabilis questing ticks were pooled into approximately five ticks/pool. Of the 224 I. pacificus pools, 26 were PCR positive (11.6%) using primers specific for the Bartonella gltA gene (sequencing of amplicons was not performed) (Chang et al., 2002). One of the 12 (8.3%) D. occidentalis and one of the seven (14.3%) D. variabilis pools were PCR positive (Chang et al., 2002). Bartonella henselae (Houston-1 strain) DNA was also detected in two Rhipicephalus sanguineus Latreille from California using a portion of the riboflavin synthase gene (ribC) (Wikswo et al., 2007). These are the first published reports of the presence of Bartonella DNA in I. pacificus nymphs, Dermacentor spp., and Rh. sanguineus, although presence of DNA does not confirm infection of ticks (Chang et al., 2002; Wikswo et al., 2007).

Table 2.  Ticks that harbour Bartonella spp. based on polymerase chain reaction evidence.
Tick speciesBartonella sp. present in tickLocationReference
  1. *The presence of Bartonella was ascertained by isolation.

Carios kelleyiResembling B. henselaeIowa, U.S.A.Loftis et al. (2005)
Dermacentor occidentalisUnidentified Bartonella spp.California, U.S.A.Chang et al. (2002)
Dermacentor variabilisUnidentified Bartonella spp.California, U.S.A.Chang et al. (2002)
Dermacentor reticulatesResembling B. henselae and B. quintanaRussiaRar et al. (2005)
Haemaphysalis longicornisResembling B. rattimassiliensis and B. tribocorumKoreaKim et al. (2005)
Haemaphysalis flavaUnidentified Bartonella sp.KoreaKim et al. (2005)
Ixodes nipponensisUnidentified Bartonella sp.KoreaKim et al. (2005)
Ixodes pacificusB. henselae, B. quintana, B. washoensis, B. vinsonii subsp. berkhoffii and an unknown cattle strainCalifornia, U.S.A.Chang et al. (2001); Chang et al. (2002); Holden et al. (2006)
Ixodes persulcatusUnidentified Bartonella spp.KoreaKim et al. (2005)
Ixodes persulcatusResembling B. henselae and B. quintanaRussiaRar et al. (2005)
Ixodes persulcatusB. henselaeRussiaMorozova et al. (2004)
Ixodes ricinusUnidentified Bartonella spp.AustriaSchabereiter-Gurtner et al. (2003)
Ixodes ricinusResembling B. schoenbuchensisFranceHalos et al. (2005)
Ixodes ricinusB. henselaeItalySanogo et al. (2003)
Ixodes ricinusUnidentified Bartonella spp.NetherlandsSchouls et al. (1999)
Ixodes ricinusResembling B. bacilliformis*PolandKruszewska & Tylewska-Wierzbanowska (1996)
Ixodes ricinusB. capreoliPolandSkotarczak & Adamska (2005a, b)
Ixodes ricinusB. henselaePolandPodsiadly et al. (2007)
Ixodes scapularisB. henselae and unidentified Bartonella spp.New Jersey, U.S.A.Eskow et al. (2001); Adelson et al. (2004)
Ixodes turdusResembling B. doshiaeKoreaKim et al. (2005)
Rhipicephalus sanguineusB. henselaeCalifornia, U.S.A.Wikswo et al. (2007)
Unidentified tickUnidentified Bartonella spp.PeruParola et al. (2002)

However, most research has focused on ticks outside the U.S.A. An unknown species of Bartonella reported to be most closely related to B. bacilliformis was isolated from an Ixodes ricinus Linnaeus collected from Walcz, Poland (Kruszewska & Tylewska-Wierzbanowska, 1996). The isolated bacteria were gram-negative and pleomorphic. Biochemical properties and restriction fragment length polymorphism (RFLP) provided additional support for a B. bacilliformis-like organism (Kruszewska & Tylewska-Wierzbanowska, 1996). In 1999, a survey performed in the Netherlands found 73 of 121 I. ricinus removed from 38 roe deer, Capreolus capreolus Linnaeus, were PCR positive for Bartonella DNA using 16S rRNA primers (Schouls et al., 1999). Eight of 103 (7.7%) I. ricinus collected in Poland from roe deer harboured B. capreoli Bermond et al. DNA in 2004, and another study performed by the same researchers found a 5.2% prevalence in ticks collected in this same region using primers specific for the 16S−23S ITS (Skotarczak & Adamska 2005a, 2005b). Schabereiter-Gurtner et al. (2003) found a pool of 12 I. ricinus from Austria positive for Bartonella sp. using a denaturing gradient gel electrophoresis based on a 16S rDNA PCR. A recent Polish survey of 102 I. ricinus removed from pet dogs and cats demonstrated an overall prevalence of 4.7% for the presence of Bartonella DNA (Podsiadly et al., 2007). Sequencing of positive amplicons, targeting the gltA gene, demonstrated the presence of B. henselae Houston-1 DNA in ticks. All PCR positive ticks had been removed from five dogs; however, whole blood collected from three of the five dogs did not result in positive culture of the organism. Blood was not collected at the time of tick removal (collection occurred approximately a month or more post-tick collection), which may explain the inability to grow Bartonella colonies on chocolate agar plates (Podsiadly et al., 2007).

Bartonella henselae (Houston-1 strain) was found in four of 271 I. ricinus (1.48%) removed from asymptomatic humans in Belluno Province, Italy (Sanogo et al., 2003). In this study, ticks were tested for Bartonella spp. using primers specific for three separate genes: the 60-kDa heat shock protein gene (groEL), the heme binding protein gene (pap31), and the cell division protein gene, (ftsZ). Bartonella henselae DNA was also detected in Ixodes persulcatus Schulze from the Novosibirsk region of Russia (Morozova et al., 2004). Fifty questing ticks were collected from vegetation in 2002 and 2003; upon PCR analysis with groEL primers, 22 of 50 (44%) ticks in 2002 and 19 of 50 (38%) ticks in 2003 contained Bartonella DNA (Morozova et al., 2004). In a study from western Siberia, Russia, 125 questing I. persulcatus and 84 Dermacentor reticulatus Fabricius were tested for the presence of Bartonella DNA, of which 37.6% I. persulcatus and 21.4% D. reticulatus were positive using primers targeting the Bartonella groEL gene (Rar et al., 2005). Both tick species appeared to harbour both B. henselae (99% homology) and an organism related to B. quintana (90% homology) based on sequencing analysis (Rar et al., 2005). Using ITS Bartonella primers, an unidentified tick species removed from a sheep in Peru was found to harbour a unique Bartonella species (Parola et al., 2002).

Recently, a survey from Korea tested for the presence of Bartonella in Haemaphysalis longicornis Neumann, Haemaphysalis flava Neumann, Ixodes nipponensis Kitaoka and Saito, Ixodes turdus Nakatsuji and I. persulcatus populations (Kim et al., 2005). Of the 1979 ticks, 297 were collected from a rodent host and 1682 were collected from vegetation. Polymerase chain reaction, targeting the 16 rRNA gene, detected Bartonella in 4.4% of 1173 Ha. longicornis, 2.7% of 74 Ha. flava, 5.0% of 20 I. nipponensis, 11.1% of nine I. turdus, 33.3% of three I. persulcatus, and 42.3% of 26 Ixodes spp. (Kim et al., 2005). Overall, 13 of 40 tick pools from rodents and 55 of 1265 tick pools collected from vegetation were PCR positive. Sequence analysis of an I. turdus tick pool demonstrated 99.2% homology with B. doshiae and two Ha. longicornis samples were closely related to B. rattimassiliensis and B. tribocorum (Kim et al., 2005).

Studies from central Sweden, the U.K., New Zealand, Egypt, South Carolina (U.S.A.) and Hungary did not identify any Bartonella species in the ticks examined (La Scola et al., 2004; Kelly et al., 2005; Loftis et al., 2006a; Monks et al., 2006; Reeves et al., 2006b; Sréter-Lancz et al., 2006). A total of 167 I. ricinus were collected by flagging areas of vegetation around Stockholm and Uppsala, Sweden. The majority of ticks collected were larvae and nymphs (eight adults in total), which might account for the lack of Bartonella found in this study (La Scola et al., 2004). In New Zealand, 136 adult Ha. longicornis ticks were collected from slaughtered cattle, deer and lambs, six Amblyomma sphenodonti (Dumbleton) were collected on Stephens Island and 21 Ha. longicornis were obtained from a tick collection at Massey University, Palmerston North (Kelly et al., 2005). All ticks were negative by PCR for the presence of Bartonella DNA (Kelly et al., 2005). Monks et al. (2006) attempted to identify the putative organism responsible for avian tick-related syndrome and also attempted to characterize other tick-borne pathogens of birds in the U.K. A total of 161 ticks removed from birds, predominantly Ixodes spp., one Hyalomma marginatum Koch and 14 unidentified non-Ixodes species, were examined by PCR analysis for bacterial DNA. Bartonella spp., Ehrlichia spp., Babesia spp. or Borrelia burgdorferi Johnson et al. sensu lato DNA was not found in these ticks (Monks et al., 2006). Carios capensis Neumann, argasid ticks known to parasitize brown pelicans (Pelecanus occidentalis Linnaeus), were tested for Bartonella, Borrelia, Coxiella and Rickettsia (Reeves et al., 2006b). Ticks removed from nests of these birds were not PCR positive for Bartonella spp. (Reeves et al., 2006b). Recently, Loftis et al. (2006a) examined ticks collected from domestic and wild animals from 12 sites in Egypt. A total of 1019 ticks examined by PCR were negative for Bartonella species. However, these ticks were found to harbour Anaplasma spp., Coxiella burnetii Derrick, Rickettsia spp., and Ehrlichia spp. (Loftis et al., 2006a). Bartonella DNA was also not detected in 658 ticks removed from red fox carcasses in Hungary (Sréter-Lancz et al., 2006).

Evidence supporting tick co-infection with Bartonella species

It appears that ticks might carry Bartonella spp. in conjunction with other tick-transmitted organisms. A 2004 study conducted in New Jersey demonstrated Bo. burgdorferi, Babesia microti, Anaplasma phagocytophilum Foggie and Bartonella spp. DNA in questing Ixodes scapularis Mégnin using PCR analysis (Adelson et al., 2004). A large percentage (34.5% of 107) of ticks in this study tested positive for DNA of Bartonella spp., of which 8.4% also contained Bo. burgdorferi DNA, 0.9% Ba. microti DNA, 0.9% An. phagocytophilum DNA, 0.9% contained Bo. burgdorferi and An. phagocytophilum DNA, and 0.9% also contained Ba. microti and An. phagocytophilum DNA (Adelson et al., 2004). Of 92 I. ricinus ticks collected in France in 2002, 9.8% were Bartonella PCR positive. Furthermore, 4% of these ticks were co-infected with Bartonella and Babesia, 1% with Bartonella and Bo. burgdorferi and one tick harboured all three organisms (Halos et al., 2005). Sequencing of the gltA product obtained from one adult tick yielded a B. schoenbuschensis-like sequence (96% homology). Eleven of 168 I. pacificus ticks from Santa Cruz County, CA, U.S.A. were PCR positive for B. henselae Houston-I (Holden et al., 2006). Of the Bartonella-positive ticks, 1.19% also harboured Bo. burgdorferi DNA and 2.98% harboured An. phagocytophilum DNA (Holden et al., 2006). In a 2003 and 2004 study, Carios kelleyi Cooley and Kohls, an argasid tick found on bats, was found to harbour both Bartonella and a Rickettsia spp. Upon further investigation, the Bartonella spp. in the argasid tick resembled B. henselae, based upon ITS sequencing (Loftis et al., 2005).

Clinical studies supporting tick transmission of Bartonella species in humans

Lucey et al. (1992) published the first human case reports suggestive of B. henselae infection associated with a tick bite. Two male patients complained of recurring fever, myalgia, arthralgias, headache and sensitivity to light within weeks after tick attachment. Bacteria were recovered from the blood of both patients using culture methods and PCR analysis demonstrated B. henselae infection. One of the patients did not recall any cat scratches or bites prior to the onset of clinical symptoms, and no cat contact was mentioned for the second patient (Lucey et al., 1992). Although the authors could only speculate that transmission occurred via tick bite, it is quite interesting that the B. henselae infections transpired within such a short time of tick attachment.

In a clinical survey performed in Connecticut, patients diagnosed with CSD were matched with control cases to demonstrate the likely risk factors associated with CSD (Zangwill et al., 1993). Patients were matched with individuals who were approximately the same age and each participant had to own or have exposure to cats. Questionnaire answers suggested that patients were more likely to have owned a kitten, had contact with a kitten with fleas, and to have been bitten or scratched by a kitten than control cases. However, patients were also more likely to have been bitten by a tick than controls, although additional risk factors for tick contact may have been involved. Of 56 patients, 21 had been bitten by at least one tick vs. five individuals in the control group (Zangwill et al., 1993). It is clearly evident that being bitten or scratched by a kitten increases the likelihood of contracting B. henselae; however, this study suggests that contact with ticks cannot be ruled out as a viable mode of transmission.

Subsequent cases indicate that co-infections of Bo. burgdorferi, the agent of Lyme disease, and B. henselae may have occurred in humans in both the U.S.A. and Europe. In one case series, three of four patients from New Jersey, U.S.A. had chronic Lyme disease and two had a history of tick bites (Eskow et al., 2001). No improvement in symptoms occurred despite treatment with Borrelia-specific antibiotics. Bartonella henselae, as assessed by PCR, was found in the blood of all four patients and was also amplified from several I. scapularis ticks collected at the homes of two patients (Eskow et al., 2001). Although B. henselae can be transmitted by cat scratch or bite, three of the four patients reported very limited contact with cats. In Poland, Podsiadly et al. (2003) described two patients hospitalized with clinical symptoms of neuroborreliosis suggestive of co-infection with Bo. burgdorferi and B. henselae. Patient 1 was Bo. burgdorferi seroreactive and B. henselae DNA was detected in cerebrospinal fluid. Patient 2 was seroreactive to both B. henselae and Bo. burgdorferi antigens at a titre of 1 : 32. The authors of the manuscript proposed that the exposure to B. henselae occurred via tick bite; however, contact with other arthropod vectors must be considered.

Another interesting report (published in an English abstract of a Russian text) describes a molecular survey performed using blood samples obtained from patients sustaining tick bites in the summers of 2003 and 2004 in the Novosibirsk region of Russia (Morozova et al., 2005). A nested PCR targeting the groEL gene detected Bartonella DNA closely related to B. henselae and B. quintana in patient blood samples. In another European study, Slovenian children were assessed by serology for exposure to multiple tick-borne infections within 6 weeks of a tick bite (Arnez et al., 2003). Of the 86 children tested, five were B. henselae seroreactive and four were B. quintana seroreactive. Another child, diagnosed with Lyme borreliosis, was B. henselae and B. quintana seroreactive (Arnez et al., 2003). Unfortunately, no non-tick bite control population was concurrently tested and pre-exposure samples were not available. A more recent study, performed by Breitschwerdt et al. (2007a), screened immunocompetent individuals with prior animal and arthropod contact for the presence of Bartonella species. Of 14 people tested, B. henselae and/or B. vinsonii subsp. berkhoffii DNA was detected in blood or in pre-enrichment culture or from a blood agar plate. All patients had had occupational animal contact for > 10 years and all had sustained previous arthropod exposure from fleas, ticks, biting flies, mosquitoes, lice, mites or chiggers (Breitschwerdt et al., 2007a). Studies of this type strengthen evolving evidence suggesting Bartonella species are being transmitted by currently unidentified arthropod vectors, such as ticks.

Clinical studies and serosurveys in cats, dogs and coyotes

Research involving tick transmission of Bartonella spp. in cats is very limited. A serological study performed in the U.K. found B. henselae seroreactivity in 40.6% of 69 pet cats and 41.8% of 79 feral cats (Barnes et al., 2000). There was no correlation between B. henselae antibodies and Leptospira antibodies, but there was a correlation between B. henselae and Bo. burgdorferi seropositivity (Barnes et al., 2000). These data only indicate that felines infected with B. henselae are at a higher risk for tick exposure; they do not show causality for ticks being vectors of Bartonella species.

Dogs can be infected with B. henselae, B. vinsonii subsp. berkhoffii, B. clarridgeiae, B. washoensis, B. elizabethae, B. quintana and an unnamed Bartonella species closely related to but different from B. clarridgeiae (Chomel et al., 2006). Bartonella henselae has been associated with granulomatous hepatitis, peliosis hepatitis and epistaxis in dogs and B. clarridgeiae can cause endocarditis and potentially lymphocytic hepatitis (Chomel et al., 2006). Co-infection with B. henselae and B. vinsonii subsp. berkhoffii has been reported in a dog, possibly exacerbating clinical manifestations (Diniz et al., 2007a). A serosurvey from the southeastern U.S.A. demonstrated that approximately 27.2% of sick dogs were B. henselae seroreactive, whereas only 1% and 10.1% of healthy dogs were seroreactive to B. vinsonii subsp. berkhoffii and B. henselae, respectively (Solano-Gallego et al., 2004). In California, 2.99% of 3417 sick dogs visiting the University of California Davis (UCD) School of Veterinary Medicine were seropositive for one or more Bartonella species (Henn et al., 2005). Of the seroreactors, 35.3% were only B. henselae seroreactive, 33.3% were only B. clarridgeiae seroreactive and 2% were only B. vinsonii subsp. berkhoffii seroreactive, whereas 30 dogs (29.4%) were seroreactive to a combination of Bartonella antigens (Henn et al., 2005). Bartonella washoensis and B. quintana can cause endocarditis in dogs and DNA of both organisms has been detected in Ixodes species (Chang et al., 2001; Rar et al., 2005; Chomel et al., 2006). These data again indirectly support the possibility of tick transmission of Bartonella organisms to dogs.

Bartonella infection is a common cause of endocarditis in dogs and several case reports suggest the possibility of tick transmission. Two separate case reports documented at the UCD Veterinary Medical Teaching Hospital describe dogs with aortic valve endocarditis caused by B. clarridgeiae and mitral valve endocarditis caused by B. washoensis (Chomel et al., 2001, 2003). Although fleas are the suspected vector of B. clarridgeiae, this dog also had An. phagocytophilum antibodies, indicating exposure to a known tick-borne organism (Chomel et al., 2001). The dog infected with B. washoensis had a weak IFA (immuofluorescence assay) titre to Coxiella burnetii (agent of Q fever) antigens, another bacterium known to be tick transmitted (Chomel et al., 2003). Although these observations are interesting, they do not definitively implicate ticks in the transmission of Bartonella to dogs.

Bartonella vinsonii subsp. berkhoffii, first isolated from a dog with endocarditis, has been identified as a cause of canine endocarditis and has also been associated with cardiac arrhythmias, myocarditis, granulomatous lymphadenitis and granulomatous rhinitis (Breitschwerdt et al., 1995, 1999; Pappalardo et al., 2000). In January 1996, a 4-year-old spayed greyhound was referred to the North Carolina State University (NCSU) Veterinary Teaching Hospital because of progressive submandibular swelling on the left side of the head. Seven days prior to the onset of clinical signs the owners had found an engorged tick in the left ear (Pappalardo et al., 2000). The dog was seroreactive to B. vinsonii subsp. berkhoffii antigens, silver-staining organisms were visualized in a biopsy and Bartonella DNA was amplified from the left granulomatous submandibular lymphadenitis lesion. Another dog examined at the NCSU Veterinary Teaching Hospital because of collapsing episodes (syncope) of 1 year’s duration, was B. vinsonii subsp. berkhoffii seroreactive (Tuttle et al., 2003). Despite clinical and haematological improvement following treatment with antibiotics, thrombocytopenia persisted. Based upon blood smear and PCR results, the dog was later found to be co-infected with Babesia canis vogeli Reichenow. All clinical and haematological abnormalities resolved after specific anti-Babesia treatment (Tuttle et al., 2003). Rhipicephalus sanguineus is a known vector for Ba. canis and is a suspected vector for B. vinsonii subsp. berkhoffii. Co-infection in this dog indirectly supports the potential of tick transmission of B. vinsonii subsp. berkhoffii and Ba. canis.

Although seropositivity to B. vinsonii subsp. berkhoffii antigens appears to be low among dogs from the southeastern U.S.A. (3.6% of sick dogs tested in North Carolina and Virginia were seroreactive), there was a high correlation between heavy tick burden and B. vinsonii subsp. berkhoffii seroreactivity (Pappalardo et al., 1997). Of Ehrlichia canis Donatien & Lestoquard seroreactors from this region, 36% were also seroreactive to B. vinsonii subsp. berkhoffii antigens (Pappalardo et al., 1997). Subsequently, Breitschwerdt et al. (1998) demonstrated that 42% of 12 dogs diagnosed with ehrlichiosis at the NCSU Veterinary Teaching Hospital were B. vinsonii subsp. berkhoffii seroreactive. In another study, 8.7% of 1875 U.S. military working dogs were B. vinsonii subsp. berkhoffii seroreactive (Honadel et al., 2001). In that study, 43% (13 of 30) of E. canis tested dogs were B. vinsonii subsp. berkhoffii seroreactive. Again, these studies are suggestive of possible transmission of B. vinsonii subsp. berkhoffii by Rh. sanguineus, the known vector for E. canis infection in dogs. Working or outdoor dogs, however, are exposed to other arthropods, which must also be considered as potential vectors of Bartonella species.

A kennel outbreak study performed on ill Walker hounds and their owners in North Carolina also suggested the possibility of Bartonella transmission by ticks. Of the 18 Bartonella PCR positive dogs, 17 were co-infected with at least one other tick-borne organism: E. canis, Ehrlichia chaffeensis Anderson et al., Ehrlichia ewingii Anderson et al., An. phagocytophilum, Rickettsia spp., or Ba. canis (Kordick et al., 1999). In this study, 25 of 27 (93%) dogs were B. vinsonii subsp. berkhoffii seroreactive, and all but one dog were E. canis seroreactive. On three visits to the kennel, Rh. sanguineus ticks were repeatedly removed from dogs. Of 23 people tested in association with this investigation, eight were B. henselae seroreactive, one E. chaffeensis seroreactive, and one Rickettsia rickettsii Wolbach seroreactive, although none reported illness at the time of testing (Kordick et al., 1999). Considering the presence of Rh. sanguineus ticks in the local kennel environment and the documentation of co-infection with E. canis and Ba. canis, which are vectored by Rh. sanguineus, there was strong circumstantial evidence supporting tick transmission of B. vinsonii subsp. berkhoffii. As dogs were concurrently infested with Ctenoc. felis at the time of the initial blood collections, fleas cannot be ruled out as potential vectors of B. vinsonii subsp. berkhoffii.

Similar results have been reported from studies performed outside the U.S.A. In Israel, 40 dogs with prior tick exposure and clinical signs of anorexia, lethargy and fever were examined using serology for exposure to several tick-borne organisms (Baneth et al., 1998). Only 10% of dogs were B. vinsonii subsp. berkhoffii seroreactive; however, 73% were seroreactive to at least three tick-borne organisms, including Babesia gibsoni, Ba. canis, Rickettsia conorii Brumpt Moroccan and Israeli strains, and Bo. burgdorferi (Baneth et al., 1998). Forty-nine sick dogs from Thailand were examined using both serology and PCR analysis for exposure to multiple tick-borne organisms (Suksawat et al., 2001). Of these, 38% were seroreactive to B. vinsonii subsp. berkhoffii antigens, one of the highest prevalences reported in dogs outside the U.S.A., but no dog was PCR positive (Suksawat et al., 2001). Another survey performed in Morocco established an overall B. vinsonii subsp. berkhoffii seroprevalence of 38% in 147 dogs tested (Henn et al., 2006). Furthermore, this study found a much higher seroprevalence in stray dogs (47% of 101 from Khenifra and 38% of 22 from Rabat) compared with pet dogs (4% of 24 from Rabat). Information regarding ectoparasite infestation was not collected in this study; therefore no correlations can be made regarding the higher prevalence of B. vinsonii subsp. berkhoffii in strays compared with pet dogs. By contrast with the Thai and Moroccan studies, only one dog examined from Greece was seroreactive to B. vinsonii subsp. berkhoffii antigens, although all dogs were E. canis seroreactive and 13 of 19 were E. canis PCR positive (Mylonakis et al., 2004). A survey performed on Reunion Island, located in the Indian Ocean, demonstrated that 26% of 165 dogs tested were E. canis seroreactive, whereas only 8.85% were B. vinsonii subsp. berkhoffii seroreactive (5% harboured antibodies to both organisms) (Muller et al., 2004). A more recent survey established a low serological and molecular prevalence of both B. henselae and B. vinsonii subsp. berkhoffii within a population of sick dogs from Brazil (Diniz et al., 2007b). Four of 197 (2%) dogs were seroreactive to B. henselae, whereas only three of 197 (1.5%) harboured B. vinsonii subsp. berkhoffii antibodies. Of the 197 dogs examined, B. henselae and E. canis DNA was amplified from one dog, and another dog contained B. henselae, B. vinsonii subsp. berkhoffii, and E. canis DNA based upon PCR analysis.

Coyotes (Canis latrans Say) appear to be important reservoirs for B. vinsonii subsp. berkhoffii in the U.S.A. Coyotes also have extensive environmental exposure to and infestations with ticks, more so than most domestic dogs. Of 869 blood specimens collected from California, 35% of the coyotes were seropositive by enzyme-linked immunoabsorbant assay for B. vinsonii subsp. berkhoffii antibodies (Chang et al., 1999). Another study carried out by Chang et al. (2000) determined that 28% of 109 coyotes from central coastal California were B. vinsonii subsp. berkhoffii bacteraemic and 76% of these coyotes were B. vinsonii subsp. berkhoffii seroreactive. Samples were collected in highly endemic tick areas, thereby conceivably explaining the large number of antibody- and culture-positive coyotes found in this study. As a wildlife reservoir, coyotes could serve as a potential source of B. vinsonii subsp. berkhoffii infection for pet dogs or people.

A more recent study performed by Beldomenico et al. (2005) established a strong statistical correlation between An. phagocytophilum and B. vinsonii subsp. berkhoffii seropositivity, as B. vinsonii subsp. berkhoffii seropositive coyotes were more likely to be An. phagocytophilum seroreactive than B. vinsonii subsp. berkhoffii seronegative coyotes. Bartonella vinsonii subsp. berkhoffii seroreactive coyotes were also more likely to be infected with Dirofilaria immitis (Leidy), a mosquito-transmitted pathogen. Similar results were obtained from a prospective study of sick dogs brought to the UCD School of Veterinary Medicine and diagnosed with infective endocarditis (MacDonald et al., 2004). Of the 18 endocarditis cases, 28% were culture or PCR positive for Bartonella infection; including B. vinsonii subsp. berkhoffii in three dogs, B. clarridgeiae in one dog and a B. clarridgeiae-like organism in another dog (MacDonald et al., 2004). In this study, all Bartonella seroreactive dogs were concurrently An. phagocytophilum seroreactive, an organism transmitted by I. pacificus (MacDonald et al., 2004). Furthermore, Foley et al. (2007) surveyed 97 dogs from eight villages in northern California for the presence of Bo. burgdorferi, An. phagocytophilum and B. vinsonii subsp. berkhoffii by immunofluorescence and PCR analysis. Although all dogs were negative for the presence of bacterial organisms, 17.5% were seroreactive for An. phagocytophilum, 12% for B. vinsonii subsp. berkhoffii and 4% for Bo. burgdorferi antigens. Prior use of acaricide as a preventive and dogs sleeping outdoors were not associated with pathogen exposure; however, co-exposure to both An. phagocytophilum and B. vinsonii subsp. berkhoffii was statistically significant (Foley et al., 2007) Seroepidemiological evidence derived from coyotes and dogs in California indirectly supports transmission of Bartonella spp. and An. phagocytophilum by I. pacificus.

Several Bartonella species have been implicated as cause of myocarditis, endocarditis, granulomatous lymphadenitis, granulomatous hepatitis, peliosis hepatitis, lymphocytic hepatitis and other clinical manifestations in dogs and in people, suggesting that pet dogs might serve as useful sentinels for human Bartonella infections (Chomel et al., 2006). With regard to modes of transmission, clinical studies involving both humans and dogs have suggested potential co-transmission of Bartonella spp. with other tick-borne pathogens. Co-infection with a Bartonella spp. and a known tick-borne pathogen could complicate diagnosis, treatment and patient management decisions. Increasingly, both doctors and veterinarians need to be aware of the potential of Bartonella spp. transmission by ticks.

Future research

There appears to be a growing spectrum of arthropods that might serve as potential vectors for Bartonella species. In particular, clinical and serological results derived from human and canine cases, in conjunction with molecular surveys of tick populations, strongly suggest that these arthropods should be considered as potential vectors of Bartonella. With a single exception in which a Bartonella species was isolated (Kruszewska & Tylewska-Wierzbanowska, 1996), all the reports of Bartonella spp. in ticks are based on amplification of organism-specific DNA sequences from extracted whole ticks. There is little evidence that Bartonella species are able to replicate within ticks and there is still no definitive evidence of transmission by a tick to a vertebrate host. Proof of vector competency will require experimental transmission studies with known infected ticks. Seemingly, the vector competence of numerous other arthropods should also be investigated. The location of Bartonella replication within arthropods, documentation of other potential reservoirs, and establishing whether transovarial transmission occurs in various arthropod species represent important issues that need to be resolved.