To the Editor: Fleas are involved in the natural cycle of different Bartonella spp. Among the 20 currently recognized Bartonella spp., 13 species or subspecies have been implicated in human disease. Recently, B. rochalimae was identified in a patient who had received numerous insect bites and subsequently had bacteremia, fever, and splenomegaly after visiting Peru (1). A recent study in Taiwan suggested that rodents could be a reservoir for B. rochalimae (2), but the vector or other mechanism of infection remains unknown. We amplified B. rochalimae, B. clarridgeiae, and B. henselae from fleas (Pulex irritans and Ctenocephalides felis) collected in Chile and discuss the role of these fleas as possible vectors of infection.

From 2005 through 2008, we collected 82 fleas from cats and dogs in pounds in Chile: 34 P. irritans, 37 C. felis, and 11 C. canis. Fleas were kept in 70% ethanol and sent to the Special Pathogens Laboratory of the Área de Enfermedades Infecciosas of the Hospital San Pedro, La Rioja, Spain, to be examined for Bartonella spp. Fleas were then rinsed in distilled water and dried on sterile filter paper under a laminar-flow hood. Each flea was crushed with a sterile pestle, and DNA was extracted by lysis with 0.7 M ammonium hydroxide. PCR was used to detect Bartonella DNA (according to the defining criteria for Bartonella spp.); primers targeted the RNA polymerase β-subunit–encoding gene (rpoB) and the citrate synthase gene (gltA) (3–5). PCR primers for a fragment of the 16/23S rRNA intergenic region and the heat-shock protein-encoding gene (groEL) were also used (6,7). Positive controls (B. henselae strain Marseille, kindly supplied by Unité des Rickettsies, Faculté de Médecine, Université de la Méditerranée, Marseille, France) and negative controls (sterile water instead of template DNA) were used. PCR products were purified, and both strands of each amplicon were subjected to sequence analysis. Nucleotide sequence homologies were searched by using BLAST (www.ncbi.nlm.nih.gov/blast/Blast.cgi).

When rpoB primers were used, Bartonella spp. were found in 4 C. felis (4.8%) fleas from cats and in 4 P. irritans (4.8%) fleas from dogs. The same 8 samples were positive when primers for gltA gene and 16/23S rRNA intergenic region were used. Unfortunately, none of the 82 specimens were positive when PCR primers targeting the groEL gene were used. In all experiments, negative controls remained negative.

Sequencing of the 825-bp rpoB fragments from the 4 C. felis fleas indicated that they were most closely aligned with the gene sequences of B. clarridgeiae (n = 2, >99% similarity) and B. henselae (n = 2, 100% similarity). Using gltA (380 bp), we found 100% similarity with B. clarridgeiae and B. henselae. Accordingly, the 16/23S rRNA amplicons from these specimens exhibited 100% similarity with the corresponding sequences of B. clarridgeiae (154 bp) and B. henselae (172 bp).

Amplicons for the rpoB fragment gene obtained from 4 P. irritans fleas showed highest similarity (97.2%–99.5%) with rpoB of B. rochalimae. Three were identical, and we deposited the consensus sequence in GenBank in 2006 under the name “uncultured Bartonella sp.” and accession no. DQ858956. The sequence differed from those described for all known Bartonella spp. and phylogenetically was most closely related to B. clarridgeiae (8). The sequence of the protein encoded by rpoB in these 3 specimens (protein_id ABH09235) had 3 aa changes (121I→V, 233K→I, and 274N→E) with respect to the deduced sequence of the RpoB protein for B. rochalimae. The importance of these changes remains unknown. The remaining nucleotide sequence was recently submitted to GenBank under accession no. FJ147196, designated B. rochalimae because isolation of this new Bartonella spp. was reported in 2007 (1). These 4 specimens also yielded positive PCR products for gltA (380 bp) and 16/23S rRNA (≈175 bp). Subsequent nucleotide sequence analysis showed 100% homology with the corresponding partial nucleotide sequences from B. rochalimae.

In 2002, Parola et al. (9) amplified Bartonella DNA by using PCR with Pulex spp. fleas collected from persons in Peru and suggested the existence of a new Bartonella sp. The nucleotide sequence of the 16S-23S ribosomal RNA intergenic spacer obtained from 1 genotype (clone F17688) was nearly identical to the corresponding sequence of B. rochalimae. This finding suggests that Pulex spp. fleas could be vectors.

Cat scratch disease has been reported in Chile, and B. henselae has been found in cats in Chile (10). Thus, our finding of B. henselae and B. clarridgeiae in C. felis fleas from Chile confirms the risk for exposure of humans in contact with cat fleas. Furthermore, our finding of B. rochalimae in P. irritans fleas from dogs in Chile supports the possibility that P. irritans fleas could be vectors for B. rochalimae. These findings are of public health importance because they identify possible vectors of these human pathogens.