“A Re-Evaluation of M. prototuberculosis”: Continuing the Debate

I n his Opinion [1], Noel Smith questions that the Mycobacterium tuberculosis complex (MTBC) is a successful clone that emerged from a much broader and ancient mycobacterial species, whose extant representatives are Mycobacterium canettii and other smooth tubercle bacilli [2]. He suggests that our conclusion is weak because he sees it as solely based on comparison of single nucleotide polymorphisms (SNPs) in six housekeeping genes and the 16 sRNA gene, and because some of these genes also showed evidence of horizontal transfer. Smith’s criticism mostly comes down to two important assertions: first, MTBC members did not emerge from the smooth strains; second, the smooth strains and MTBC members should not be seen as a single species. In response, we would stress that using comparison of SNP in housekeeping genes is now a widely accepted practice that has shown the clonal expansion of several other important bacterial pathogens from genetically broader progenitor species. Examples of such analyses are Yersinia pestis descended from Yersinia pseudotuberculosis [3], serotype Typhi from Salmonella enterica [4], Shigella clones from Escherichia coli [5], or Bordetella pertussis from Bordetella bronchiseptica [6]. For M. tuberculosis, Musser and colleagues [7] used SNP analysis of six housekeeping genes to detect the extremely reduced genetic polymorphisms that exist within the classical members of the MTBC. Most importantly, their results were later confirmed by whole genome comparisons [8–10]. It was therefore particularly exciting to find a much higher rate of sequence polymorphisms in the same six housekeeping genes that were used by Musser and colleagues when we studied smooth tubercle bacilli isolated in East Africa from patients with clinical tuberculosis, and these data represented the backbone of our published analysis [2]. However, we reemphasize here that the greater genetic polymorphism observed among the smooth tubercle bacilli is not restricted to the six tested housekeeping genes, but also applies to other regions of the genomes, including conventionally used markers for differentiation of tubercle bacilli. These results, shown in supporting Figure S1 of our article [2], demonstrate that the smooth tubercle bacilli exhibited several outstanding features that underline their differences among each other and with the MTBC members. The number of IS1081 copies constitutes one example. This insertion sequence has been so far described as being specific for tubercle bacilli [11]. Their number is highly conserved among all members of the MTBC, which carry five or six copies [11]. However, the smooth tubercle bacilli showed zero to three copies [2], indicating that this specific IS element was present in most strains of the smooth tubercle bacilli, but with a distinct copy number. In previous studies, the unusually low copy number of IS1081 was also used to differentiate the smooth tubercle bacillus M. canettii from the classical MTBC members [12,13]. Other examples illustrating the wider diversity of smooth tubercle bacilli, mentioned in Protocol S2 and Figure S1, include the variable presence of another insertion sequence specific to smooth bacilli (ISMycA1), different MIRU–VNTR variant alleles also uniquely detected in smooth bacilli, and the absence of the direct repeat region in some smooth groups. In a different study, it was also shown thatM. canettii harbored a completely different set of spacer sequences in the direct repeat region that was not present in the classical MTBC members [14]. Finally, Figure 1 shows our unpublished sequences of the 59 end of the 16S rRNA gene in different lineages of the smooth tubercle bacilli. This region, which is 100% identical within the classical members of the MTBC, shows SNP variations that were only found within the smooth tubercle bacilli. This comparison also nicely shows that some lineages of smooth tubercle bacilli differ from others in an otherwise highly conserved region. Thus, altogether, our data clearly show that smooth tubercle bacilli harbor more diversity than MTBC members alone, although they share many characteristics in common (see below). More insight in this matter will certainly be provided by genome sequencing projects on M. canettii (http://www.sanger.ac.uk/sequencing/ Mycobacterium/canetti) and other lineages of smooth tubercle bacilli. In the case of the smooth tubercle bacilli strains, we demonstrated gene mosaicism, which we interpret as the result of homologous DNA recombination among smooth bacilli strains. Taking a different view, Smith speculates that the sequence diversity of the smooth strains, and their divergence from the MTBC members, mostly results from horizontal genetic transfer (HGT) into the smooth strains of DNA segments from donor strains of another species that remains to be discovered. Under this hypothesis, MTBC members would have been left untouched by HGT, whereas