In 2013, we visited a dairy herd in Jura, France, on 3 occasions. The herd contained ≈30 lactating cows. During physical examination of all teats, one third of the cows had lesions. In most cases, only 1 teat/cow had a lesion, which was a single, painless nodule of variable size that was localized in the dermis (Figure 1, panel A). Early-stage lesions, which had a well-demarcated indurated area, were observed only in a few cows. A few nodules evolved to ulcers, and most ulcers showed cicatrization and fibrosis.

To establish a diagnosis, we obtained 3 biopsy specimens from 3 cows (1 specimen/cow) under local anesthesia (lidocaine) from intact nodules by using a skin biopsy punch (diameter 8 mm). Each biopsy specimen was divided into 2 parts: half was placed in formalin, and half was placed in sterile 0.9% NaCl.

During histopathologic examination, superficial and deep dermis showed nodular and interstitial dermatitis (Figure 1, panel B) and an infiltrate composed predominantly of large, pale, foamy macrophages mixed with lymphocytes and few plasma cells (Figure 1, panel C). Some macrophages contained 2 or 3 nuclei and a granular cytoplasm. Multinucleated giant cells were rarely observed. There were large foci of necrosis. Ziehl-Neelsen staining indicated acid-fast bacteria deposited in clumps and resembling globi of leprosy within macrophages and necrotic foci exclusively on samples 1 and 2 (Figure 1, panel D). These observations resulted in the diagnosis of bovine nodular thelitis. Mycobacterial culture was not attempted because of its inherent difficulty (3). A molecular approach was used to characterize bacteria from the 3 biopsy specimens.

Biopsy specimens were washed 3 times with DNA-free sterile water, and epidermis was removed from the dermis and hypodermis to avoid contamination with cutaneous flora. The dermis and hypodermis were homogenized at 7,400 rpm for 70 s in a proteinase K solution (1 mg/mL) (Sigma, St. Louis, MO, USA) by using a MagNA Lyser (Roche Molecular Diagnostics, Mannheim, Germany), followed by incubation for 18 h at 55°C in a shaking dry bath incubator. Whole DNA was extracted from the lysed tissue by using the MagnaPure Compact Kit (Roche Molecular Diagnostics).

For the phylogenetic investigations, a multigene sequencing approach was used (5). Double-stranded partial sequences of the genes for 65-kDa heat shock protein (hsp65 [groEL2]), the β-subunit of RNA polymerase gene (rpoB), superoxide dismutase (sodA), elongation factor Tuf (tuf), 16S rRNA, and transfer–messenger (tmRNA) were obtained by real-time PCR by using primers (Table) designed for this study (rpoB) or described previously (5,6). For each gene, chromatograms corresponding to the 2 strands of the 3 amplified products (only 2 amplified products for the 16S rRNA gene) were compared by using Staden software (http://staden.sourceforge.net/) and edited to remove ambiguous bases (7).

For each gene, no differences were detected among 3 sequences (2 sequences for the 16S rRNA gene), and a consensus sequence was subsequently used for each gene (392 bp for hsp65 [KJ095005], 951 bp for rpoB [KJ095009], 442 bp for sodA [KJ095006], 718 bp for tuf [KJ095008], 359 bp for the tmRNA gene [KJ095007], and 369 bp for the 16S rRNA gene [KJ095004]). To make phylogenetic comparisons, we selected a subset of Mycobacterium spp. sequences for each gene to obtain a good representation of the genomic diversity among the slow-growing mycobacteria by using data from Mignard and Flandrois (5). Sequences were obtained from GenBank/European Molecular Biology Laboratory/DNA Data Bank of Japan databases. The MAFFT program (http://mafft.cbrc.jp/alignment/software/) was used to align nucleotide sequences. Alignments were verified by using SeaView (http://www.molecularevolution.org/software/alignment/seaview). Divergent and ambiguously aligned blocks were removed by using Gblocks (http://molevol.cmina.csic.es/castresana/Gblocks_server.html) to ensure an accurate alignment before phylogenetic reconstructions. Phylogenies were inferred from sequences by calculating observed genetic distances by using PhyML (http://code.google.com/p/phyml/) with the general time reversible evolutionary model (7). Mycobacterium setuense was used as an outgroup because it is a species that is borderline to the slow-growing mycobacteria (8). Resulting tree topologies were evaluated by bootstrap analysis with 1,000 resamples.

The phylogenetic tree based on rpoB sequences (Figure 2, panel A) was consistent with that obtained for hsp65 sequences (Figure 2, panel B). On the basis of rpoB and hsp65 phylogenies, the unidentified organism was phylogenetically similar to M. leprae and M. lepromatosis but belonged to a clearly distinct branch. Similar to these species, it shares a common ancestor with M. haemophilum. The same well-sustained relationship with M. leprae and separation from M. haemophilum was inferred from sodA, tuf, and tmRNA gene phylogenies, but the lack of corresponding sequences for M. lepromatosis impaired this analysis.

On the basis of 16S rRNA gene phylogeny (Figure 2, panel C), the unknown organism was a member of the M. leprae cluster and was related to M. lepromatosis and an unknown feline Mycobacterium sp. that causes leprosy-like symptoms in cats (9). This organism was phylogenetically related to M. lepromatosis, and the 2 species constituted a separate branch from that of M. leprae (bootstrap value 88%). The absence of congruence between the phylogenetic trees inferred from the 16S rRNA gene and the hsp65 and rpoB genes could be related to a short consensus sequence (369 bp) in a position that is not optimal for mycobacteria discrimination.

Phylogenetic investigations strongly supported the conclusion that an undescribed species of the genus Mycobacterium that is related to M. leprae and M. lepromatosis, the causative agents of tuberculoid and lepromatous leprosy, and a diffuse form of lepromatous leprosy (10,11), respectively, was characterized in cows with bovine nodular thelitis. M. leprae and M. lepromatosis are not cultivable on artificial media. M. haemophilum, which is closely related to M. leprae and M. lepromatosis (11,12), is a slow-growing mycobacterium that requires hemin and a low temperature for growth (12). Attempts to cultivate mycobacteria from cows with bovine nodular thelitis were mostly unsuccessful and necessitated multiple samples (3).

M. leprae strains were found to be clonal (13,14). In accordance with reductive evolution of its genome (15), this species is an obligate parasite that infects humans and armadillos (13). In contrast, the environment could be a reservoir of M. haemophilum (12). The natural habitat of the causative agent of bovine nodular thelitis is unknown, and zoonotic transmission had never been observed in breeder cows.

The M. leprae cluster might have an animal origin. Genetic studies of multiple bovine and feline mycobacteria characterized from animals with nodular thelitis and leprosy-like syndromes, respectively, are currently in progress and should provide useful information.