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 confirmed at necropsy for all 18 animals. Four additional cases of endocarditis were identified at necropsy after an apparent sudden death. Most of the damaged valves of these 22 animals had large, cauliflower-like 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 amplification 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 amplification was performed on gltA-positive valves for the following Bartonella-specific genes or genomic region: rpoB, ribC, groEL, internal transcribed spacer (ITS) of 16S–23S rRNA (9,10). Amplification products, including those for the 16S rRNA, were subsequently sequenced.

Serology by indirect fluorescent 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-specific gltA gene amplification. For both cows, the normal-appearing control valve was PCR negative for this gene. The PCR-positive cows (nos. 04–927 and 05–1406) were the 2 oldest cows (13 and 15 years old, respectively). ITS 16–23 rRNA amplification 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). Amplification 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. Amplification of all the other genes studied were PCR negative for normal-appearing control valves of both cows. The genes rpoB (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 isolates. 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)

This is the first description, to our knowledge, of endocarditis associated with Bartonella in cattle. PCR amplification of the gltA gene, used for identification 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 amplification of 1 vegetative valve gave a fragment of ≈190 bp, which is the size expected for B. bovis (10). The lack of PCR amplification of the same genes from healthy-appearing valves indicated that the PCR amplification obtained with the vegetative valves was not the result of a B. bovis bacteremia. No definitive evidence exists that B. bovis had induced the primary lesion leading to the endocarditis. However, PCR amplification 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.

Moreover, the high IFA antibody titer against B. bovis antigen and the low antibody titers of the PCR-negative endocarditis cases reinforced the likely role of B. bovis as the causative agent of these 2 bovine endocarditis cases. High antibody titers are commonly observed in human and canine cases of Bartonella endocarditis. In fact, high antibody titers are considered a major diagnostic criterion for Bartonella endocarditis in humans (3). Finally, the 2 B. bovis–infected vegetative valves were aortic and mitral valves, which are the most frequently involved valves in human and canine Bartonella endocarditis cases (5,13,14).

Two (9.1%) of the 22 endocarditis cases were Bartonella DNA positive. This percentage is within the range of 3% reported for human cases of endocarditis (14) and 19% (5) to 28% (13) reported for dogs. The 2 cases occurred in the oldest animals (Table), which suggests that B. bovis endocarditis could develop in geriatric cows, following chronic bacteremia in an overtly healthy animal. Nearly 30% of cattle >7 years of age are reportedly Bartonella bacteremic (12).

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. berkhoffii in humans, cats, and dogs, respectively (6,14,15). This study confirms that B. bovis can cause endocarditis in cows like B. henselae and B quintana in their respective feline and human reservoirs.