Molecular Detection of Bartonella Species in Fleas Collected from Dogs and Cats from Costa Rica

Introduction

B artonella is a bacterial genus comprising more than 30 different species, some of which have a zoonotic potential. They are arthropod-borne bacteria parasitizing mammalian erythrocytes and endothelial cells (Chomel and Kasten 2010, Harms and Dehio 2012). Bartonella henselae is the etiological agent of cat scratch disease (CSD), and Bartonella clarridgeiae is the cause of a cat scratch–like disease in humans. Felines are the principal reservoir hosts of both Bartonella spp., and their vector is the cat flea, Ctenocephalides felis. Transmission to humans occurs by contamination of broken skin with the bacterium, usually through a cat scratch (Chomel et al. 2006). Cats are also reservoirs of Bartonella koehlerae, which has been associated with human diseases as well (Avidor et al. 2004, Chomel et al. 2006). Bartonella rochalimae and Bartonella vinsonii subsp. berkhoffii are commonly detected in dogs (causing bacteremia), and their fleas and have also reported to cause disease in humans (Yore et al. 2014). In spite of their wide geographical distribution, there are no previous reports on the presence of Bartonella species in Costa Rica or the rest of Central America. Therefore, the aim of this study was to evaluate the presence of Bartonella spp. in fleas collected from dogs and cats in Costa Rica.

Materials and Methods

Samples of fleas were collected from dogs and cats (with permission from owners) as part of a recent project on rickettsial diseases in several regions of the Caribbean slope of Costa Rica with reports or suspicions of human rickettsioses: Turrialba (9°54′N, 83°41′W; elevation, 650 m.s.l.), La Virgen (10°23′N, 84°08′W; elevation, 190 m.s.l.), Limón (9°59′N, 83°02′W; elevation, 5 m.s.l.), Cahuita (9°44′N, 82°50′W; elevation, 5 m.s.l.), Guápiles (10°13′N, 83°47′W; elevation, 260 m.s.l.), Jiménez (10°12′N, 83°44′W; elevation, 230 m.s.l.), and Guácimo (10°12′N, 83°41′W; elevation, 110 m.s.l.). This region is characterized by continuous wet (annual rainfall of 2500 to 3000 mm) and warm (mean temperatures of 23–26°C) conditions without distinct seasonality. The collection sites and methods have been described elsewhere (Troyo et al. 2012a). Briefly, permission from owners was obtained to search dogs and cats for ectoparasites, which were collected with the aid of combs and forceps. Fleas were identified and those from each household were processed independently in pools of one to 10 fleas, separated according to flea species and vertebrate host. A total 93 flea pools were analyzed for detection of Bartonella DNA: 72 C. felis pools (from 68 dogs and 4 cats), and 21 Pulex simulans pools (all from dogs). Fleas were stored at −20°C until DNA extraction.

DNA from the flea pools was extracted by the DNeasy Blood & Tissue Kit (Qiagen, Santa Clarita, CA), according to the manufacturer's instructions. Conventional PCR assays for the citrate synthase (gltA) and the β subunit RNA polymerase (rpoB) gene fragments of Bartonella spp. were performed as previously described, using primers pairs BhCS.781p and BhCS.1137n for gltA and 1400F and 2300R for rpoB that yield approximately 380-bp and 825-bp fragments, respectively (Harrus et al. 2009). DNA from B. henselae and Bartonella grahamii were used as positive controls, Brucella abortus DNA and water were used as negative controls. For DNA sequencing, PCR products were treated with alkaline phosphatase and exonuclease I (Thermo Scientific) and were sent to Macrogen Inc. (Korea). The sequences obtained were compared to those previously deposited in GenBank.

Results

Eight of the 93 flea pools (8.6%) were found positive for Bartonella DNA, as shown in Table 1. B. vinsonii subsp. berkhoffii strain Winnie and B. rochalimae sequence were detected in flea pools collected from dogs, as well as a sequences closely related to an uncultured Bartonella sp. (BR10 clone, acc. no. GU056200.1) found in Rhipicephalus ticks of dogs from Taiwan (Tsai et al. 2011). B. henselae and B. clarridgeiae DNA were identified in flea pools collected from cats. From five out of the eight positive pools, both gltA and rpoB amplicons were sequenced successfully. With the exception of pool 11-TP-2-2, which identified the B. vinsonii subsp. berkhoffii gltA gene fragment and B. rochalimae rpoB gene fragment, the other four flea pools retrieved sequences that matched the same Bartonella species in both targeted genes. Although C. felis is the most common flea species known to infest domestic dogs and cats worldwide, only three (4.2%) pools were found positive (two of which were from cats), whereas five (22.7%) of P. simulans pools were found positive. Of note, previous investigations identified Rickettsia felis–like DNA in two (2.8%) of the C. felis pools containing Bartonella DNA in this study, 9-LMG-1 and 11-TP-9-1 (Troyo et al. 2012b).

Discussion

The overall proportion of positive fleas (8.6%) found in this study is similar to recent surveys in America (Yore et al. 2014). B. vinsonii subsp. berkhoffii was the most frequent species identified in dog flea pools. It is known that infection with B. vinsonii subsp. berkhoffii may cause prolonged bacteremia in dogs (Kordick et al. 1998). Given that all animals were apparently healthy at the time of ectoparasite collection, fleas positive for Bartonella DNA likely represent feeding from either subclinical or persistent bacteremic hosts. Studies in North and South America have reported Bartonella in Pulex spp. (Parola et al. 2002, Yore et al. 2014). These fleas are common on dogs and can also feed on humans; therefore, this genus may be considered an important vector of Bartonella spp., and dogs would be a potential source of this zoonotic agent to humans (Yore et al. 2014). Two C. felis pools from cats were positive for B. henselae or B. clarridgeiae, which are frequently associated with feline hosts, but can be transmitted to dogs and humans (Chomel et al. 2006).

Only one flea pool (P. simulans) contained DNA fragments of two different Bartonella species (B. vinsonii subsp. berkhoffii and B. rochalimae). It is probable that more pools contained different fleas, or single fleas, that were infected with more than one Bartonella species. However, the detection method used in this study could overlook these different and/or co-infections, as it is biased toward the dominant species present in the sample.

To date, no clinical cases attributed to Bartonella spp. have been reported in Costa Rica. Considering the lack of clinical suspicions and appropriate laboratory tests, these infections may present as under-recognized causes of acute febrile illness with some severe manifestations such as endocarditis, angiomatosis, lymphadenitis, peliosis hepatis, meningitis, and neuroretinitis (Edouard et al. 2015). To the best of our knowledge, this is the first molecular detection of Bartonella spp. in fleas from Central America. Previous studies in Costa Rica include serologic evidence (immunoglobulin G [IgG]) of infection with Bartonella in pumas and a study in the tropical rainforest where no evidence of infection with Bartonella was detected by PCR in white-nose coatis (Chomel et al. 2004, Mehrkens et al. 2013). In Guatemala, Bartonella spp. have been detected in bats, and a study reported 20.6% of cattle infected with Bartonella bovis (Bai et al. 2011, 2013). The detection of zoonotic Bartonella spp. in this study and the previous detection of R. felis highlight the importance of flea control in this region.

This study provides baseline data useful for the surveillance, prevention, and control of rickettsioses and bartonelloses in Costa Rica. Bartonella species are increasingly being associated with animal and human illnesses; thus, the identification of their potential reservoirs and an increased understanding of disease ecology and epidemiology are of utmost public health relevance.