|Home | About | Journals | Submit | Contact Us | Français|
The most likely animal source of a human case of cardiac disease in Washoe County, Nev., was identified by comparison of DNA sequences of three genes (citrate synthase gltA, 60-kDa heat shock protein gene groEL, and 16S rRNA gene) of Bartonella washoensis cultured from the human patient in question and of Bartonella isolates obtained from the following Nevada rodents: Peromyscus maniculatus (17 isolates), Tamias minimus (11 isolates), Spermophilus lateralis (3 isolates), and Spermophilus beecheyi (7 isolates). Sequence analyses of gltA amplicons obtained from Bartonella from the rodents demonstrated considerable heterogeneity and resulted in the identification of 16 genetic variants that were clustered within three groups in phylogenetic analysis. Each of the three groups was associated with a rodent genus, Peromyscus, Tamias, or Spermophilus. The gltA, 16S rRNA gene, and groEL sequences of a Bartonella isolate obtained from a California ground squirrel (S. beecheyi) were completely identical to homologous sequences of B. washoensis, strongly suggesting that these animals were the source of infection in the human case.
Bartonella species bacteria that are recognized as human pathogens include species that are associated with domestic cats and are transmitted by cat fleas (B. henselae) as well as other species that have not been associated with animal reservoirs but are transmitted by human lice (B. quintana) or by sand flies (B. bacilliformis). Recent observations support a role for other Bartonella species as human pathogens. Among the most likely sources of such pathogens is exposure to Bartonella-infected rodents or their ectoparasites. High prevalences of various Bartonella strains among rodents have been demonstrated in North America, Asia, and Europe (3, 16, 24). Considering the range of animal reservoirs and the types of insects that have been implicated in the transmission of Bartonella species, human exposure to these bacteria may be more common than presently realized (5). This statement is supported by the isolation of Bartonella organisms from patients that were identical or closely related to Bartonella species obtained from rodents, including B. elizabethae, B. vinsonii subsp. arupensis, and B. washoensis (5, 7, 10, 23). Reports of patients with unrecognized illnesses who had antibodies to antigens derived from rodent-associated Bartonella strains also suggest that human exposures to these agents are more common than previously believed (9, 15).
The sequences of three genes (citrate synthase gltA, 60-kDa heat shock protein gene groEL, and 16S rRNA gene) of a novel Bartonella strain were submitted to GenBank in 1998 (accession numbers AF050108, AF071193, and AF070463). This strain, which was isolated by R. L. Regnery et al. in 1995 from a patient with cardiac disease from Washoe County, Nev., contained sequences that were different from the sequences of all previously described Bartonella species and isolates. Regnery et al. designated this isolate Bartonella washoensis. A rodent reservoir for this Bartonella species was implicated but never identified (5). Shortly after the occurrence of this case, the area surrounding the patient's residence was trapped for rodents by M. Murray. The gltA sequences of Bartonella isolates obtained from the three rodents captured at the case site were submitted to GenBank by Regnery et al. in 1998 and were assigned accession numbers AF071187, AF071188, and AF071189. These sequences demonstrated differing levels of homology with the human isolate, with the highest percentage of identity (96.4%) being observed between B. washoensis and an isolate from a least chipmunk (GenBank accession number AF071189).
The goal of our investigation was to reevaluate and, if possible, identify the most likely animal source of the Bartonella infection in the above-mentioned human case in Nevada. To achieve this goal, our objectives were as follows: (i) to collect blood samples from rodent species in Washoe County, Nev., and to culture Bartonella microorganisms from these animals; (ii) to characterize Bartonella isolates obtained from these rodents by DNA sequencing of PCR amplicons derived from the gltA, groEL, and 16S rRNA genes of Bartonella; and (iii) to compare the rodent isolates of Bartonella with the B. washoensis strain obtained from the human patient.
All mammals were live trapped using a combination of Sherman live traps and Tomahawk live traps baited with Farmer's Brand sweet grain. Animals were anesthetized with Metofane prior to collection of blood samples, which were obtained from these animals by using previously described procedures (18). Basically, small rodents (e.g., Peromyscus maniculatus) were bled by the retro-orbital technique using capillary tubes coated with heparin. Larger animals were bled via cardiac puncture by using 1- and 3-ml syringes fitted with 22-gauge, 1.5-in. needles. Blood samples were immediately placed on dry ice and later stored at −70°C.
Details of the procedures used to isolate Bartonella from rodent blood have been published previously (16). Briefly, rodent blood samples diluted 1:4 in brain heart infusion medium (Becton Dickinson, Cockeysville, Md.) supplemented with 5% amphotericin B were used for isolation. Aliquots of 0.1 ml of the blood were applied to heart infusion agar plates supplemented with 5% rabbit blood (Becton Dickinson). The plates were incubated at 35°C in an aerobic atmosphere of 5% CO2 and held for 10 to 24 days. The cultures were examined daily for bacterial growth, and material from colonies that were tentatively identified as Bartonella spp. were picked with an inoculating loop and streaked onto a new agar plate. Bartonella colonies were later collected from the new agar plate, placed in brain heart infusion medium supplemented with 10% glycerol, and stored at −70°C.
DNA was extracted from cultures by using a QIAamp kit (Qiagen, Chatsworth, Calif.). Bacteria cultures that were tentatively identified as Bartonella by colonial and bacterial morphology were initially confirmed as such by PCR amplification of Bartonella-specific sequences of the citrate synthase (gltA) gene. The oligonucleotide primers used in these PCR assays generated a 379-bp amplicon from gltA (19). The gltA primers were BhCS781.p (5′-GGGGACCAGCTCATGGTGG-3′) and BhCS1137.n (5′AATGCAAAAAGAACAGTAAACA-3′). All PCR amplifications were carried out in a PTC200 DNA-Engine (MJ Research, Inc., Waltham, Mass.) for 35 cycles with the following cycle parameters: 95°C for 30 s, 45°C for 30 s, and 72°C for 30 s.
The groEL gene and 16S rRNA gene were amplified from Bartonella cell suspensions in brain heart infusion broth as follows. A 50-μl aliquot of the Bartonella suspension was boiled for 10 min in a microcentrifuge tube, followed by centrifugation to pellet the cellular debris. Five microliters of the resulting supernatant was then added to the PCR mixture. The PCR mixes contained 10 mM Tris-HCl (pH 8.3), 50 mM KCl, 1.5 mM MgCl2, 0.001% gelatin, 200 μM (each) dATP, dCTP, dGTP, and dTTP, 0.4 μM concentrations of each primer, and 2.5 U of Taq DNA polymerase (AmpliTaq; Perkin-Elmer Cetus). The cycling parameters were 94°C for 30 s, 50°C for 30 s, and 72°C for 60 s for 40 cycles. The entire groEL coding sequence was amplified by using the primers derived by Haake et al. (12) (GenBank accession number U78514). The sequence of the forward primer, GroEL-1F, was 5′-ATGGCTGCTAAAGAAGT-3′, corresponding to nucleotide positions 1 to 17. The sequence of the reverse primer, GroEL-1R, was 5′-GAAGTCCATGCCGCCCAT- 3′, corresponding to nucleotide positions 1641 to 1624. The resulting amplicon was 1,641 bp in length.
The 16S rRNA primer pairs have broad specificity and can be used to amplify a portion of the 16S rRNA from all eubacteria (21). Previously, Avidor et al. (1) used these primers to amplify the 16S rRNA gene of B. henselae. The sequences for this pair were 5′-CCGCACAAGCGGTGGAGCA-3′ (p93E) and 5′-AGGCCCGGGAACGTATTCAC- 3′ (p13B). The resulting amplicon was approximately 460 bp in length. To verify the presence of amplicons of appropriate size, PCR products were first electrophoresed in 1% agarose gels according to standard methods and the ethidium bromide-stained amplicon bands were then visualized by using a UV light source.
In order to provide more conclusive confirmation of isolates as being Bartonella, PCR-positive products amplified by using the gltA primers were purified with a PCR purification kit (Qiagen, Santa Clarita, Calif.) and sequenced in both directions by using an ABI Prism DNA sequencing kit for dye terminator cycle sequencing ready reaction (PE Biosystems ABI, Foster City, Calif.) and the same primers used for the PCR assays. Thirty cycles of a thermocycle program (96°C for 15 s, 50°C for 5 s, and 60°C for 4 min) were performed with a Gene Amp PCR System 9600 thermocycler (Perkin Elmer, Norwalk, Conn.). The products of this reaction were purified in Centri-Sep spin columns (Princeton Separations, Adelphi, N.J.) and then sequenced by using an ABI model 377 sequencer. Removal of the primer sequences from the full-length amplicons resulted in 338-bp sequences that were aligned with other Bartonella strains and analyzed using the phylogenetic analysis program PAUP (for the Macintosh).
DNA sequencing of the groEL amplicon was performed with primers internal to the pair used to amplify the coding region. The groEL-2F sequencing primer (5′-ATGCTGCTGTTGATGAAGTGG-3′) corresponded to nucleotide positions 362 to 382 of the B. henseleae sequence referenced above (12). The groEL-2R sequencing primer (5′-CAACAATACCTTCTTC-3′) corresponded to nucleotide positions 1237 to 1222. These primers spanned an internal groEL fragment of 875 nucleotides. DNA sequencing of the 16S rRNA amplicon was performed by using the same primers described above for the amplification.
The four unique gltA sequences identified among Bartonella isolates from P. maniculatus have been submitted to GenBank and assigned accession numbers AY064533, AY064534, AY064535, and AY064536. The five novel gltA sequences from isolates obtained from Tamias minimus have been assigned accession numbers AF451159, AF451160, AF451161, AF451162, and AF451163. The five novel gltA sequences from isolates from Spermophilus beecheyi and S. lateralis have been assigned accession numbers AY071858, AY071859, AY071860, AY071861 and AF470616. The accession numbers for the groEL sequences from isolates Sb944nv and Sb1963nv are AF484066 and AF484067, respectively.
In November 1995, a 70-year-old male resident of Washoe County, Nev., developed fever and myocarditis. Based on positive bacterial culture, Bartonella infection was implicated as the source of his illness. The Washoe County Health Department determined that the most likely site of exposure in this case was the patient's home site at Red Rock, which is located approximately 25 km north-northwest of the city of Reno at an elevation of 1,731 m above sea level (Fig. (Fig.1).1). Dominant vegetation species in the area include Piñon-Juniper woodland, sagebrush (Artemesia tridentata), rabbit-brush (Chrysothamnus nauseosus), and an introduced grass species, Cheat grass (Bromus tectorum). The residence is built on a gently sloping lot that faces southeast.
Rodents were collected at seven sites within Washoe county and at one site within Storey County, located east of Washoe county. Three study sites were within urban settings located in the central part of Reno: Washoe County Golf Course (39.50°N, 119.81°W; elevation, 1,390 m), Idlewild Park (39.35°N, 119.86°W; elevation 1,372 m), and Governor's Bowl Park (39.53°N, 119.79°W; elevation, 1,342 m) (Fig. (Fig.11).
Three study sites were located in the southern suburb of Reno: Galena Creek Park (39.35°N, 119.88°W; elevation, 2,280 m), Mount Rose Campground (39.31°N, 119.89°W; elevation, 2,744 m), and South Meadows Sports Complex (39.40°N, 119.76°W; elevation, 1,451 m) (Fig. (Fig.1).1). Galena Creek Park is a regional park located on the east slope of the Sierra Nevada Range and is approximately 19 km south-southwest of Reno. Traps were also set at Galena Forest Estates, a residential development located immediately east of Galena Creek Park. Mount Rose Campground is located between the cities of Reno and Incline Village. South Meadows Sports Complex is located in an agricultural area approximately 11 km south of Reno.
Vya Station (41.59°N, 119.86°W; elevation, 2,677 m) is a Washoe County Road Department maintenance facility that is located in a remote, northern portion of the county (Fig. (Fig.1).1). This region is sparsely inhabited and was surveyed primarily because of the presence of large rodent populations in the resident employees' quarters. One study site, Virginia City Highlands (39.39°N, 119.63°W; elevation, 1,756 m), is situated within 10 km of Reno but is located in Storey County (Fig. (Fig.11).
We captured a total of 111 rodents belonging to six species: Ord kangaroo rat Dipodomys ordi (3 animals), bushytail woodrat Neotoma cineria (1 animal), deer mouse P. maniculatus (29 animals), California ground squirrel S. beecheyi (41 animals), golden-mantled squirrel Spermophilus lateralis (12 animals), and least chipmunk T. minimus (25 animals). The numbers of rodents, by species, captured at each collection site are presented in Table Table11.
The California ground squirrel was the only rodent species captured within the urban settings located in the central part of Reno. The dominant species in rodent communities within the sites located in the southern suburb of the city were S. beecheyi and S. lateralis, another ground squirrel species. Two species, the deer mouse and the least chipmunk, were dominant in the Vya Station site located in northern Washoe County. Kangaroo rats were captured only in the Virginia City Highlands site.
Bartonella cultures were obtained from 37 (33.3%) of the 111 mammals tested. Bartonella-positive rodents were found in each of the eight study sites, with the exception of Washoe County Golf Course. The highest prevalences of Bartonella infection (up to 50%) were found in the Idlewild Park, Governor's Bowl Park, and Vya Station sites. The host species for these Bartonella were P. maniculatus, S. beecheyi, S. lateralis, and T. minimus (Table (Table1).1). The prevalences of Bartonella-positive rodents by individual species, irrespective of site, were as follows: P. maniculatus, 53.3% (17 of 30 animals); T. minimus, 44.0% (11 of 25 animals); S. lateralis, 25.0% (3 of 12 animals); and S. beecheyi, 17.1% (7 of 41 animals).
Comparisons of gltA sequences of Bartonella isolates obtained from rodents demonstrated their heterogeneity and resulted in the identification of 16 unique genetic variants (sequences). Phylogenetic analysis demonstrated that these 16 genetic variants were clustered within three groups (Fig. (Fig.22).
Four unique sequences were identified among the Bartonella isolates obtained from eight P. maniculatus (Pm1780nv, Pm1783nv, Pm1784nv, Pm1786nv, Pm1857nv, Pm1868nv, Pm1917nv, and Pm1949nv); all of these isolates clustered in the same clade and appeared to represent a single distinct genotype (Fig. (Fig.2).2). All of the sequences except one (from isolate Pm1857nv [GenBank accession no. AY064536]) shared very close relationships, with no more than a 2-nucleotide difference between the gltA sequences of each isolate. The sequences from isolates from P. maniculatus were phylogenetically related to other isolates obtained from Peromyscus species in the eastern United States (13, 16).
Isolates from least chipmunks (T. minimus) also appeared to be very closely related. Six unique sequence patterns could be distinguished among the 11 isolates obtained from chipmunks (Tm916nv, Tm917nv, Tm918nv, Tm919nv, Tm920nv, Tm1781nv, Tm1794nv, Tm1795nv, Tm1870nv, Tm1872nv, and Tm1950nv) (Fig. (Fig.2).2). Five of these were novel sequences (see Materials and Methods), and one had been submitted previously to GenBank (accession no. AF071189).
DNA sequencing also indicated that isolates from ground squirrels were relatively closely related to one another. The three Bartonella isolates from S. lateralis (SL311nv, SL943nv, and SL979nv) were identical to each other but were distinct from isolates obtained from S. beecheyi. Two sequences (GenBank accession no. AY071858 and AY071860) from four isolates from S. beecheyi (Sb1659nv, Sb1865nv, Sb1879nv, and Sb1916nv) were closely related to the genetic variant that appeared to be specific for S. lateralis, whereas two other sequences (GenBank accession no. AY071859 and AF470616) obtained from S. beecheyi isolates were more distantly related. Regardless of the differences between sequences from the ground squirrel isolates examined in the study, the phylogenetic analysis demonstrated that these isolates were closely related to each other when compared not only with isolates found in other rodent species in this study but also with all other rodent-related Bartonella sequences that have been submitted to GenBank (3, 10, 13, 16, 24).
The gltA sequences of the Bartonella isolates (Sb944nv and Sb1963) obtained from two California ground squirrels were completely identical to homologous sequences of the strain of B. washoensis cultured from the human patient.
DNA sequencing of the 16S rRNA amplicons from isolates Sb944nv and Sb1963nv produced 395 and 415 bp of unambiguous sequence data, respectively. Complete identity to the corresponding DNA segment of the B. washoensis 16S rRNA gene was found for both Sb944nv and Sb1963nv, i.e., 395 of 395 nucleotides and 415 of 415 nucleotides, respectively, were identical.
DNA sequencing of the groEL amplicons from isolates Sb944nv and Sb1963nv produced 785 and 690 bp of unambiguous sequence data, respectively. The sequence of isolate Sb944nv demonstrated complete identity to the corresponding segment from B. washoensis (785 of 785 nucleotides). The sequence of isolate Sb1963nv had a 1-bp mismatch with the corresponding segment of B. washoensis, i.e., 689 of 690 nucleotides were identical. The substitution occurred at position 394 of the B. washoensis sequence (C for A). The sequence of isolate Sb944nv shared the same position mismatch as that of isolate Sb1963nv.
The squirrels infected with B. washoensis were found in Idlewild Park and Governor's Bowl Park, both of which are located in the central part of Reno. Idlewild Park is a municipal park located on the banks of the Truckee River. It is a typical urban park with areas of turf interspersed with trees of various species and several baseball fields and picnic areas. The infected ground squirrel was trapped along a rock wall. Governor's Bowl Park is located on land immediately adjacent to the junction of highways US 395 and Interstate 80. It is heavily used by Reno residents, primarily for baseball and soccer games. Its topography is essentially bowl-like, with sloping sides covered with decorative juniper shrubbery. The infected squirrel was collected on a landscaped perimeter planted with juniper.
This is the first report genetically linking a Bartonella species isolated from a human to a Bartonella species isolated from an indigenous North American rodent. The ability to accurately identify the strains of infectious agents that cause disease is central to epidemiological surveillance and public health decisions. Analyses of a diverse array of Bartonella strains that have been recovered from a wide range of wild mammals provide a useful tool for identification of potential animal sources of human cases of Bartonella infection. The finding that Bartonella strains obtained from ground squirrels in Nevada are identical to the strain isolated from the patient provides evidence for the transmission of Bartonella species from rodents to humans.
Comparisons of sequences from seven genes have been used for phylogenetic analyses of Bartonella species, but the present knowledge about Bartonella taxonomy is supported mostly by the sequence information of three genes, namely the 16S rRNA, citrate synthase (gltA), and 60-kDa heatshock protein (groEL) genes (4, 14, 17). DNA sequencing of these target genes, as well as of other Bartonella genus- and species-specific genes, should prove valuable in providing information on the regional diversity and classification of rodent isolates and the molecular epidemiology of future human bartonellosis cases.
The finding of Bartonella strains in California ground squirrels which are identical or very similar to B. washoensis is particularly interesting from an epidemiological perspective because these rodents are often closely associated with areas of human habitation, recreational areas, and agricultural lands, making it likely that humans will come into contact with infected animals. Both sites where squirrels infected with B. washoensis were identified lie within urban areas of the city of Reno. These sites offer excellent habitats that support large colonies of California ground squirrels, as well as being very popular recreational areas. Washoe County is a good example of an area experiencing rapid growth as residential development occurs in previously undisturbed areas. The amount of human interaction with California ground squirrels is likely to increase.
The role of California ground squirrels in the maintenance and transmission of plague (Yersinia pestis) is well documented. While much remains to be determined concerning the incidence of disease, mode of transmission, and possible vector species associated with B. washoensis, it appears likely that contact between people and California ground squirrels remains a significant link in the epidemiology of the disease.
Investigators were unable to determine how the abovementioned human patient with myocarditis became infected with B. washoensis. While the patient denied handling California ground squirrels, making this route of acquisition unlikely, his residence is located in a rural setting with habitat suitable for populations of different rodent species. Other more well-known species of Bartonella are believed to be transmitted, at least in part, by arthropod vectors. For example, B. bacilliformis, the etiologic agent of Carrion's Disease, is transmitted by sand flies, and B. quintana, which causes trench fever, can be transmitted by human body lice (5). Recently, cat fleas (Ctenocephalides felis) have been reported to be capable of transmitting B. henselae to cats, perhaps through the contamination of cat skin with flea feces (8, 11). Ticks also have been reported recently to be naturally infected with various species of Bartonella, although transmission of these agents by ticks has yet to be demonstrated (6, 22).
The implication of cat fleas as vectors of B. henselae is noteworthy because S. beecheyi are often heavily infested with Oropsylla montana, a flea species that readily feeds on humans and generally is considered to be the most important vector of human plague in the United States (2). These squirrels also often support heavy infestations of another flea species, Hoplospyllus anomalus, that is known to feed at least occasionally on humans (20). Humans are most likely to be exposed to the bites of ground squirrel fleas when the normal hosts of these insects die as a result of injury, predation, or disease. In the western United States, humans are most at risk of plague exposure when plague epizootics cause high mortality among S. beecheyi or other major hosts of O. montana (Spermophilus variegatus and S. lateralis), thereby forcing infectious fleas to seek alternative hosts, including humans. If California ground squirrel fleas transmit B. washoensis to humans, then it is possible that the risk of human exposure to this bacteria will increase significantly following the occurrence of plague epizootics. Although it is possible that other arthropods can transmit B. washoensis to humans, it should be noted that ticks are rarely encountered on California ground squirrels from the Reno area.
The identification of this new human pathogen in a common peridomestic rodent of the far western United States adds public health significance to rodent sanitation. If fleas or other arthropods are identified as vectors, then it also will be important to monitor flea densities and apply appropriate flea control when the risk of flea bite exposure is high, especially in areas such as parks and campgrounds that have high public usage. Further investigation is needed to define the role of ectoparasites in the route of transmission. It is also important to continue efforts to identify additional human cases of B. washoensis infection in order to better evaluate the extent and importance of this disease agent.
We thank David Thain and staff of the Nevada Department of Agriculture, Animal Disease Laboratory, Reno, for use of laboratory facilities for processing; Scott Monsen, Dan Ariaz, Jeff Brasel, Dennis Cerfoglio, and Judith Saum (Washoe County District Health Department) for assistance in environmental assessments and rodent trapping; Kiyotaka Tsuchiya (Division of Vector-Borne Infectious Diseases, Centers for Disease Control and Prevention, Fort Collins, Colo.) for assistance with DNA sequencing; and David Dennis and Duane Gubler (Division of Vector-Borne Infectious Diseases, Centers for Disease Control and Prevention) for their support of the research program “Role of rodent-associated Bartonellae as sources of undiagnosed illness in humans in the American Southwest.”