Exotic Small Mammals as Potential Reservoirs of Zoonotic Bartonella spp.

To evaluate the risk for emerging human infections caused by zoonotic Bartonella spp. from exotic small mammals, we investigated the prevalence of Bartonella spp. in 546 small mammals (28 species) that had been imported into Japan as pets from Asia, North America, Europe, and the Middle and Near East. We obtained 407 Bartonella isolates and characterized them by molecular phylogenetic analysis of the citrate synthase gene, gltA. The animals examined carried 4 zoonotic Bartonella spp. that cause human endocarditis and neuroretinitis and 6 novel Bartonella spp. at a high prevalence (26.0%, 142/546). We conclude that exotic small mammals potentially serve as reservoirs of several zoonotic Bartonella spp.

T he genus Bartonella includes a variety of gram-negative, fastidious, hemotrophic bacteria that are transmitted by blood-sucking arthropod vectors. The genus consists of 20 species and 3 subspecies; at least 11 of these species are known or suspected to be pathogenic for humans as causative agents of emerging zoonoses (1).
Many exotic animals are traded as pets around the world and have been imported into Japan without quarantine. However, no data exist on the prevalence of infection with Bartonella spp., especially in exotic pet animals. Our study objectives were to 1) examine the prevalence of Bartonella spp. infection in exotic small mammals imported into Japan from various countries, 2) compare the diversity of these Bartonella strains by analyzing the partial sequence of the citrate synthase gene (gltA), and 3) evaluate the possibility that these mammals may serve as potential reservoirs of zoonotic Bartonella spp.

Animals and Samples
For this study, 546 exotic small mammals were purchased from trading companies. The animals represented 3 orders and included 6 families, 23 genera, and 28 species (Table 1). They had been imported into Japan as pets from June 2004 through October 2007 from 8 countries in 4 geographic regions: Asia (China, Thailand, and Indonesia), Europe (the Netherlands and Czech Republic), North America (United States), and the Middle and Near East (Egypt and Pakistan). Of the 546 animals, 367 had been captured in their natural environment and 179 had been bred in the exporting countries. Heparinized blood samples were aseptically collected from each animal (anesthetized with chloroform) and centrifuged at 3,000 rpm for 15 min. Plasma was removed and the blood sample pellets were kept at -80ºC until examination.

Isolation of Bacteria
The blood sample pellets were thawed at room temperature, 100-μL supplemented Medium 199 (24) was added to each pellet, and each sample was mixed well. A 100-μL sample of each mixture was plated on 2 heart infusion agar (DIFCO, Sparks Glencoe, MI, USA) plates containing 5% defi brinated rabbit blood. The plates were incubated at 35ºC under 5% CO 2 . After 2 weeks of incubation, 2 or 3 colonies with genus Bartonella morphologic characteristics (small, gray or cream-yellow, round colonies) were picked from each plate, confi rmed to be gram-negative pleomorphic bacteria, and subcultured using the same conditions used for the original cultures.

DNA Extraction and PCR
The genomic DNA of each isolate was extracted by using InstaGene Matrix (Bio-Rad, Hercules, CA, USA). The extracted DNA was used for PCR analysis of a 312-bp part of the gltA gene to confi rm that the bacteria were from the genus Bartonella. PCR was performed by using an iCycler (Bio-Rad) with a 20-μL mixture containing 20 ng extracted DNA, 200 μM of each deoxynucleoside triphosphate, 1.5 mmol/L MgCl 2 , 0.5 U Taq DNA polymerase (Promega, Madison, WI, USA), and 1 pmol of each primer. The specifi c primer pair and PCR conditions for gltA were as previously reported (25).

DNA Sequencing and Phylogenetic Analysis
The PCR products were purifi ed by using a commercial kit (Spin Column PCR Products Purifi cation Kit; Bio Basic, Markham, Ontario, Canada). Direct DNA sequencing of the purifi ed PCR products was carried out by using dye terminator chemistry with specifi c primers (25) and a Model 3130 Genetic Analyzer (Applied Biosystems, Foster City, CA, USA). The 312-bp gltA sequences from the isolates and type strains of established Bartonella spp. in GenBank/EMBL/DDBJ were aligned with the Clustal X program (26), and a phylogenetic tree was drawn, based on the sequence data and using the neighbor-joining method (27) with the Kimura 2-parameter distance method (28) in MEGA 4 (29). Bootstrap analysis was carried out with 1,000 replications (30).

Statistical Analysis
The results were analyzed in 2×2 tables. Chi-square tests were used to examine the statistical signifi cance; p<0.05 was considered signifi cant.

Prevalence of Bartonellae
The prevalence of bartonellae in the exotic small mammals examined was 26.0% (142/546). A total of 407 isolates were obtained from the 142 bacteremic animals ( Table 1). The prevalence by animal origin was 37.3% (137/367) in captive animals and 2.8% (5/179) in animals from breeder facilities. A signifi cantly higher prevalence of bartonellae was observed in captive animals than in animals from breeder facilities (p<0.001). In the captive animals, the prevalence by region varied up to 47.2% in Asia, which is higher than the 39.7% prevalence in North America. The prevalence of bartonellae by corresponding taxonomic family of host animal ranged from 38.6% (49/127) in the family Muridae to 43.9% (69/157) in the family Sciuridae. No bartonellae were detected in animals in the families Octodontidae and Erinaceidae. Among animals from breeders, only 5 chipmunks (Tamias sibiricus) from China were found to be infected with bartonellae; no bartonellae were isolated from animals in the families Petauridae, Muridae, Octodontidae, or Dipodidae.

DNA Sequences and Phylogeny of Isolates
The 407 isolates in this study were classifi ed into 53 genotypes on the basis of DNA sequence analysis of a 312bp fragment of their gltA genes. The sequence of a genotype from a Cairo spiny mouse was identical to that of the B. elizabethae type strain (GenBank accession no. Z70009) isolated from a human patient with endocarditis (10). The other 52 genotypes were found to be novel genotypes after comparison with known Bartonella spp. The phylogenetic tree of the gltA sequences shows that the 52 novel genotypes are clearly clustered in 10 genogroups, designated A to J (Figure).
Of the 52 novel genotypes, genogroup A, which consisted of 21 genotypes (AB444954 to AB444974) isolated from 7 squirrel species, was related to B. washoensis strain Sb944nv (AF470616), which was isolated from a California ground squirrel (Spermophilus beecheyi) and was genetically identical to an isolate from a human patient with myocarditis (9). The sequence similarities of these genotypes and B. washoensis strain Sb944nv ranged from 94.2% to 97.4%. Genogroup A contained B. washoensis-like genotypes; the genotypes from each squirrel species formed a separate clade, except for the genotypes from Richardson's ground squirrels (Sp. richardsonii) and Columbian ground squirrels (Sp. columbianus), which formed a mixed clade (Figure).
In this study, 18 genotypes formed the 6 unique genogroups B to G. The DNA sequences of the genotypes in each genogroup showed relatively low similarity (82.4%-94.6%) to the type strains of known Bartonella spp., and sequence similarities between genogroups B to G were also low (87.5%-93.6%). The novel Bartonella genogroups B, C, D, and E were isolated from greater Egyptian jerboas, tricolored squirrels (Callosciurus notatus), fat-tailed gerbils (Pachyuromys duprasi), and golden spiny mice (A. russatus), respectively. The genotypes in group F were isolated from 6 animal species: large Egyptian gerbils, fat-tailed gerbils, fat sand rats (Psammomys obesus), lesser Egyptian jerboas, greater Egyptian jerboas, and bushy-tailed jirds; those in genogroup G were isolated from a bushy-tailed jird and a large Egyptian gerbil (Figure). In genogroup F, 3 of the 9 isolates from fat-tailed gerbils and the 14 isolates from fat sand rats had identical gltA DNA sequences. Fur-thermore, 5 of the 7 isolates from greater Egyptian jerboas and 2 of the 3 isolates from a lesser Egyptian jerboa also had identical sequences.
The 10 novel genotypes in genogroup I were isolated from 9 animal species, and the sequence similarities between the genotypes (AB444995 to AB445005) and B. elizabethae type strain (F9251) ranged from 95.5% to 98.7%. The DNA sequences of gltA of the 3 isolates from a Cairo spiny mouse (AB445000) were identical to that of B. elizabethae (F9251). The sequences of the 13 isolates from lesser Egyptian jerboas were identical to those of the 16 isolates from greater Egyptian jerboas.

Multiple Infections with Different Bartonella Genogroups and Genotypes
Of the 142 Bartonella-positive animals, 25 (17.6%) were found to be infected with different Bartonella genogroups or genotypes (Table 2). A lesser Egyptian jerboa carried 3 different genotypes in 2 genogroups; the other 24 animals carried 2 different genogroups or genotypes. Of these 24 animals, an American red squirrel carried a B. washoensis-like strain in genogroup A and B. clarridgeiaelike strains in genogroup J; 11 animals were infected with B. elizabethae-like strains in genogroup I and strains in genogroups B, C, D, E, F, or G, and the remaining 12 carried different genotypes in the same genogroup (Table 2).

Discussion
We report prevalence of bartonellae in exotic small mammals imported into Japan as pets. We found that 26.0% (142/546) of the animals examined had bartonellae in their blood. Prevalence among wild captive animal species was high (37.3%), signifi cantly higher (p<0.001) than that among animals from breeder facilities. Of the 179 animals (representing 9 species) from breeder facilities, only 5 Siberian chipmunks imported from a Chinese breeder were found to carry bartonellae, and these were of the same genotype as bartonellae from wild captive animals. These results suggest that animals in breeder facilities may be maintained under hygienic conditions from birth to export, so they rarely have contact with wild animals or bloodsucking arthropod vectors.  Most isolates from animals in the family Sciuridae (58.7%; 122/208) were in genogroup A and showed high sequence similarity to B. washoensis. Kosoy et al. (9) have reported that B. washoensis is widely distributed in ground squirrels in the western part of the United States and that it was isolated from a human with myocarditis in Nevada, USA. Thus, captive squirrels carrying B. washoensis-like organisms could serve as a source of infection for humans.
Animals in the family Sciuridae were also found to be carrying several genotypes of bartonellae in genogroups C, B. grahamii-like strains in genogroup H, B. elizabethaelike strains in genogroup I, and B. clarridgeiae-like strains in genogroup J. The sequence similarities between the genotypes and the related Bartonella spp. type strains ranged from 98.4% to 98.7% for B. grahamii, from 95.5% to 95.8% for B. elizabethae, and were 96.2% for B. clarridgeiae. In humans, B. grahamii, B. elizabethae, and B. clarridgeiae have been reported to cause neuroretinitis (11), endocarditis (10), and cat-scratch disease (31), respectively. These fi ndings suggest that exotic squirrels also might be a potential source of Bartonella infections in humans. Although B. clarridgeiae has mainly been isolated from cats (1), B.
clarridgeiae-like strains were isolated from an American red squirrel in this study. B. clarridgeiae-like organisms have also been isolated from yellow-necked mice (Apodemus fl avicollis) in Sweden (14) and Greece (15).
The sequence similarity of the gltA sequence (312 bp) of the B. clarridgeiae-like genotype isolated in our study to that of the strain isolated from the yellow-necked mouse (AF391788) was relatively high (97.7%). Recently, B. rochalimae, a B. clarridgeiae-like organism, was isolated from a human patient with bacteremia, fever, and splenomegaly (32). The B. clarridgeiae-like strain from the American red squirrel in this study also showed high gltA sequence similarity (96.8%) with that of B. rochalimae strain BMGH. Studies will be required to clarify the pathogenicity of B. clarridgeiae-like organisms for humans. Such studies would include 1) evaluation of the organisms' ability to invade human erythrocytes and/or endothelial cells, 2) demonstration of the presence and expression of the genes of type 4 secretion systems (VirB/VirD4 or Vbh) and Trw, and 3) comparisons of the entire genome sequences of the organisms and with those of other human pathogenic Bartonella spp.  In this study, Bartonella genogroups D, E, and G were isolated from animals in the family Muridae, and Bartonella genogroup B was isolated from animals in the family Dipodidae. These fi ndings suggest strict host specifi city between the strains in these genogroups and the host animal family. However, fi ndings also showed wide host species diversity; strains in genogroup F were isolated from 6 animal species, and strains from genogroup I (B. elizabethae-like) were isolated from 9 animal species. Bartonella strains in genogroup F were isolated from animals in the families Muridae and Dipodidae. Genogroup I (B. elizabethae-like) strains were also isolated from animals in the family Sciuridae. B. elizabethae has been isolated from different animal species, e.g., a human patient and genus Rattus rats (5,10), and B. elizabethae DNA has been isolated from a dog (33). In our study, 3 Bartonella isolates from a Cairo spiny mouse imported from Egypt had an identical gltA sequence to that of the B. elizabethae type strain. Thus, some Bartonella spp., such as B. elizabethae and B. washoensis, infect host animals in diverse families and may have zoonotic potential.
In the present study, 17.6% (25/142) of exotic animals were infected with different Bartonella genotypes or genogroups. In particular, 3 isolates from a greater Egyptian jerboa were classifi ed in 3 different genotypes. Of the 25 Bartonella-positive animals, 13 showed co-infection with different Bartonella genogroups. Of these 13 animals, 12 carried B. elizabethae-like strains in genogroup I. In contrast, strains with identical gltA sequences were isolated from 2 different animal species, such as greater Egyptian jerboas and lesser Egyptian jerboas, Siberian chipmunks and Hokkaido squirrels, and fat-tailed gerbils and fat sand rats. These fi ndings suggest that some Bartonella species have a wide host range and may be transmitted horizontally by some blood-sucking arthropod vectors with low host specifi city.
In summary, we examined the possibility that exotic small mammals may be reservoirs of zoonotic Bartonella spp. around the world. The animals in this study carried, at high prevalence, several Bartonella spp. that are human pathogens. Novel species were suggested by the fact that some of the genotypes in 6 genogroups (B to G) showed relatively low similarity (<94.6%) to known Bartonella spp. and formed independent clusters according to phylogenetic analysis based on partial gltA sequences. More taxonomic studies should sequence other housekeeping genes, such as rpoB, 16S rRNA, ftsZ, groEL, and ribC, to confi rm whether these isolates are novel Bartonella spp. (34). To prevent human infections by Bartonella spp. carried by exotic small mammals, a quarantine system for these animals should be established as early as possible. Further studies will be necessary to clarify the route of transmission among exotic small mammals and to evaluate the pathogenicity for humans and animals of the isolates belonging to novel Bartonella genotypes found in this study.