Endocarditis in Cattle Caused by Bartonella bovis

This study aimed to determine the role of Bartonella as an endocarditis agent in cattle. Bartonella bovis was identified by PCR, gene sequences analysis, and specific internal transcribed spacer amplicon product size in 2 bovine endocarditis cases with high antibody titers, which demonstrates that B. bovis is a pathogen for cattle.

B acteria-induced vegetative valvular endocarditis is one of the main cardiac disorders in adult cattle (1). The prevalence of endocarditis may reach 5.2 cases per 10,000 cows (2), but the disease is often misdiagnosed and only discovered during the slaughtering process or at necropsy. Bacterial endocarditis is often linked to a primary source of infection and the presence of other infectious lesions, such as mastitis, metritis, arthritis, or liver abscesses. The most frequent pathogens isolated from cardiac valves or the bloodstream of cows with endocarditis are Arcanobacterium pyogenes (up to 90% of the strains), Streptococcus sp., and numerous Enterobacteriaceae (2).
The development of molecular techniques (PCR) led to the identifi cation of many noncultivable or poorly cultivable bacteria as agents of human endocarditis, such as Coxiella burnetii, different Bartonella species, or Tropheryma whipplei (3,4). In dogs, Bartonella species cause 8% of all bacterial endocarditis and up to 19% of noncultivable bacterial endocarditis (5). In cats, bacterial endocarditis is infrequent, but a fatal case caused by B. henselae was recently reported (6). Bartonellae can therefore be considered as potential agents of endocarditis even in their own reservoir species. Cattle host B. bovis, which has also been isolated from cats (7) and was recently suggested as the etiologic agent in a human case of bartonellosis (8). However, the role of Bartonella species in bovine endocarditis has never been explored. Therefore, our objective was to determine the putative role of Bartonella sp. as an agent of endocarditis in cattle.

The Study
Twenty-two cases of bovine endocarditis were diagnosed in adult cows (ages 5-15 years, mean 7.4 years) at the School of Veterinary Medicine, Toulouse, France, from September 2004 through June 2006. Eighteen cows were hospitalized for poor condition, anorexia, weight loss, wasting syndrome, and abnormal cardiac auscultation. Endocarditis was diagnosed at physical examination. Lesions of the cardiac valves were confi rmed at necropsy for all 18 animals. Four additional cases of endocarditis were identifi ed at necropsy after an apparent sudden death. Most of the damaged valves of these 22 animals had large, caulifl owerlike lesions.
For each cow, fragments of the vegetative valve and of 1 normal-appearing valve were collected. DNA extraction from each valve sample was performed by using Nucleospin Tissue extraction kit (Macherey-Nagel, Hoerdt, France) according to the supplier's recommendation.
PCR amplifi cation was performed on all normal-appearing and vegetative valves for the hypervariable V3 zone of the eubacterial 16S rRNA detection, and the 3′ end of citrate synthase gene (gltA) Bartonella sp. DNA detection. Additional PCR amplifi cation was performed on gltApositive valves for the following Bartonella-specifi c genes or genomic region: rpoB, ribC, groEL, internal transcribed spacer (ITS) of 16S-23S rRNA (9,10). Amplifi cation products, including those for the 16S rRNA, were subsequently sequenced.
Serology by indirect fl uorescent antibody assay (IFA) was performed as reported elsewhere (11) on the supernatant extracted from a cardiac blood clot from each cow. Vero cells infected with the type strain of B. bovis (CIP 106692 T ), B. chomelii (CIP 107869 T ), and B. schoenbuchensis (NCTC 13165 T ), respectively, were used as antigens.
The 22 vegetative valves included 8 pulmonary valves, 7 tricuspid valves, 6 aortic valves, and 1 mitral valve. The only vegetative mitral valve and 1 of the 6 aortic vegetative valves showed positive results for Bartonella-specifi c gltA gene amplifi cation. For both cows, the normal-appearing control valve was PCR negative for this gene. The PCRpositive cows (nos. 04-927 and 05-1406) were the 2 oldest cows (13 and 15 years old, respectively). ITS 16-23 rRNA amplifi cation was obtained only for the damaged valve of cow 04-927; the size of the amplicon product was ≈190 bp, which was identical to the size of the product obtained with the B. bovis reference strain (Figure). Amplifi cation of the 16S rRNA and all the other genes studied were PCR positive for the damaged valves. A 16S rRNA PCR amplicon has been obtained from the normal-appearing valve of cow 05-1406 but could not be sequenced. Amplifi cation of all the other genes studied were PCR negative for normal-appearing control valves of both cows. The genes rpoB *Ecole Nationale Vétérinaire d'Alfort, Maisons-Alfort, France; †Institut National de la Recherche Agronomique, Maisons-Alfort, France; ‡University of California, Davis, California, USA; and §Ecole Nationale Vétérinaire de Toulouse, Toulouse, France (GenBank accession nos. EF432062, EF432061), ribC (accession nos. EF432060, EF432059), gltA (accession nos. EF432055, EF432056), groEL (accession nos. EF432058, EF432057) were partially sequenced. Sequence identities were respectively 100% with B. bovis (gltA) and 99% with B. bovis (groEL, ribC, rpoB). The sequence obtained with 16S rRNA (accession no. EF432054) had a 99% identity with B. bovis.
None of the cultures on rabbit blood agar (12) of fragments from the 22 vegetative valves yielded Bartonella iso-lates. The 2 PCR-positive cows had high IFA titers (5,120 and 640 for B. bovis antigen), whereas the 20 PCR-negative cows had low or negative titers (Table) Conclusions This is the fi rst description, to our knowledge, of endocarditis associated with Bartonella in cattle. PCR amplifi cation of the gltA gene, used for identifi cation of Bartonella infection, gave an identity of 100% with the previously reported B. bovis gene sequence. The sequences of 4 additional genes (groEL, ribC, rpoB, and 16S rRNA) shared 99% identity with B. bovis genes. ITS amplifi cation of 1 vegetative valve gave a fragment of ≈190 bp, which is the size expected for B. bovis (10). The lack of PCR amplifi cation of the same genes from healthy-appearing valves indicated that the PCR amplifi cation obtained with the vegetative valves was not the result of a B. bovis bacteremia. No defi nitive evidence exists that B. bovis had induced the primary lesion leading to the endocarditis. However, PCR amplifi cation with universal primers for bacterial 16S rRNA allowed us to identify only Bartonella sequence in the damaged valves without apparent contamination with DNA from other bacteria.
Cattle are the main reservoir for B. bovis, and diseases attributed to infection with this Bartonella species in cows are scarce (12). Nevertheless, these 2 cases demonstrated that B. bovis is a potential bovine pathogen and that B. bovis can induce endocarditis in the animal reservoir host, as previously shown for B. quintana, B. henselae, and B. vinsonii subsp. berkhoffi i in humans, cats, and dogs, respectively (6,14,15). This study confi rms that B. bovis can cause endocarditis in cows like B. henselae and B quintana in their respective feline and human reservoirs.