To the Editor: Mycobacterium ulcerans is an emerging human pathogen responsible for Buruli ulcer, a necrotizing skin disease most commonly found in West Africa, but outbreaks have also been reported in the Americas, Australia, and Asia (1). Environmental sources of infection and mode of transmission are not completely known. M. ulcerans grows slowly at 32°C, requiring 6–8 weeks for colonies to be visible in primary culture. Differentiation from M. marinum, which also causes skin infections, is important, since M. marinum can usually be treated with antimicrobial agents, whereas M. ulcerans most often does not respond favorably to drug therapy, and treatment is usually by surgical excision (2). M. shinshuense, initially isolated from a child in Japan, is phenotypically and genetically related but biochemically distinct from M. ulcerans (3).

In the last decade, several DNA-based techniques for mycobacterial identification have been developed. Rapid molecular detection and differentiation of organisms that cause skin infections directly from tissue or exudates could be of great value for early treatment. Some techniques, especially those that include nucleic acid amplification, could be used directly on clinical samples. The accepted standard for molecular identification of mycobacteria is sequencing analysis of 2 hypervariable regions identified in 16S rRNA gene. M. marinum and M. ulcerans share identical 5´-16S rDNA and 16S-23S rRNA gene spacer sequences (4). Polymerase chain reaction (PCR)-dependent methods are based on the 16S rRNA gene (5), the hsp65 gene (6) or the insertion sequence IS2404 (7). Recently, a novel category of variable number tandem repeats that could distinguish M. marinum and M. ulcerans genotypes has been described (8).

Polymorphisms in the 3´-16S rDNA region discriminate M. ulcerans from M. marinum and M. shinshuense (5). These polymorphisms also allow the separation of M. ulcerans into 3 subgroups according to geographic origin and variable phenotypic differences. IS2404 discriminates M. ulcerans from M. marinum (7). It has been used in restriction fragment length polymorphism analysis applied to a comparable number of M. ulcerans and M. marinum strains, confirming that this sequence is present in high copy numbers in M. ulcerans but absent in M. marinum. Nevertheless, an unusual mycobacterium was recently isolated that is closely related to M. marinum by phenotypic tests, lipid pattern, and partial 16S rDNA sequencing but presents low copy numbers of this element (9).

PCR-restriction enzyme analysis (PRA) of a 441-bp fragment of the hsp65 gene is a rapid, easy, and inexpensive method for identifying mycobacteria (10). Devallois et al. (6) described the PRA-hsp65 pattern of 1 M. ulcerans strain ATCC 33728 that originated in Japan. This isolate was considered a new species that resembled M. ulcerans and was named M. shinshuense (3).

We report here the usefulness of PRA-hsp65 to differentiate M. ulcerans strains from different geographic areas. Since Buruli ulcer cases have been reported on 5 continents, we studied 33 M. ulcerans strains that originated from Africa (Benin, Zaire, Ghana, Congo, Angola, Côte d'Ivoire, Togo), Asia (China, Malaysia), Australia (Papua New Guinea, Australia), the Caribbean (Mexico, Surinam, French Guiana), 1 M. shinshuense from Japan, 1 M. marinum isolate and 1 IS2404-positive M. marinum isolate from France (9). All strains were identified at the Institute of Tropical Medicine, the World Health Organization Collaborating Centre for the Diagnosis and Surveillance of Mycobacterium ulcerans Infection by IS2404 PCR and biochemical tests (Table).

DNA extracted from cultures by 3 freeze-boiling cycles was used for amplification, according to the protocol described by Leao et al. (10). Gel images were analyzed by using GelCompar II v. 2.5 (AppliedMaths, Sint-Martens-Latem, Belgium). Two distinct M. ulcerans PRA-hsp65 patterns were identified. Of 36 strains, 34 had a PRA-hsp65 pattern indistinguishable from that of M. marinum [BstEII and HaeIII (bp) of 235/210/0 and 145/105/80] at the Swiss PRAsite (http://app.chuv.ch/prasite/index.html). Two strains, 1 each from Japan and China, showed a different pattern [BstEII and HaeIII (bp) of 235/210/0 and 190/105/80], that described by Devallois et al. (6).

We have shown that PRA-hsp65 analysis performed on several M. ulcerans strains from different geographic areas produced different patterns. In fact, the unique PRA-hsp65 profile of the M. ulcerans strain previously published (6) was the most rarely found pattern among the profiles found in this study. This work helps to clarify the PRA-hsp65 patterns of M. ulcerans found in different countries. Because the epidemiology of Buruli ulcer is poorly understood, new molecular tools are still needed to differentiate M. ulcerans from different geographic settings, mainly in Africa, where the disease is more prevalent. The PRA-hsp65 method represents a rapid, easy, and inexpensive technique to differentiate M. shinshuense from M. ulcerans and M. marinum.