Pyrosequencing assay for rapid identification of Mycobacterium tuberculosis complex species

The Mycobacterium tuberculosis complex (MTC) includes M. tuberculosis, M. bovis and BCG-derived clones, M. africanum, M. canettii, M. microti, M. caprae and M. pinnipedii[1]. Recently, an eight member named Mycobacterium mungi was identified in banded mongooses (Mungos mungo) in Botswana [2]. While there is variable host-specificity among the different MTC members, every species, but M. mungi, has been implicated in human tuberculosis with M. tuberculosis being the most common pathogen [3–6]. The exact contribution of each species in human disease may be underestimated due to the limited capacity of laboratories to identify MTC isolates at the species level in routine practice. Indeed, phenotypic methods are time-consuming and the results are difficult to interpret. Molecular assays are hampered by the high genetic similarity reflected by the complete conservation of the 16S rRNA and rpoB genes of the MTC members [7, 8]. This situation indeed, led to propose that the various MTC species would be more accurately described as ecotypes [9]. Despite the high degree of nucleotide sequence homology, some genomic markers such as pncA, mpt40, hupB, gyrB and wbbl1 genes in addition to the regions of difference (RD) have been successfully used for the differentiation of MTC members through direct analysis such as in spoligotyping or after enzymatic digestion of PCR products [10–16]. Nevertheless, most of these techniques allow only partial discrimination between the MTC species and are too fastidious to be implemented in the routine practice of microbiology laboratories. The Genotype MTBC (Hain Lifescience, GmbH, Nehren, Germany) is a commercial DNA-strip based assay identifying MTC species through the detection of gyrB gene single nucleotide polymorphisms (SNPs) and RD1 [17]. The test is easy to use, but failed to identify M. canettii and M. pinnipeddii[17].

Pyrosequencing is a real-time non-electrophoretic DNA sequencing technique that can be easily adapted for routine use in the clinical microbiology laboratory. This technique was successfully used for the identification of non-tuberculous mycobacteria, identification of the Beijing family of M. tuberculosis and detection of rifampin-resistance among MTC members [18–20]. We here developed a pyrosequencing assay for the rapid identification of the MTC species based on polymorphisms within the glpK, pykA and gyrB genes.

The pykA (encoding for the pyruvate kinase) and glpK (encoding for the glycerol kinase) genes were previously analyzed to disclose the genetic basis of difference in the glycerol metabolism between M. tuberculosis and M. bovis in comparison with other MTC species [23, 24]. The authors reported a specific gene structure for M. tuberculosis, M. bovis and M. bovis BCG but this particularity has not been further used for identifying MTC organisms. In our study, we extended this previous observation to all MTC species and a larger collection of clinical isolates in the perspective of MTC member's identification.

In PCR-based experiments all negative controls remained negative and no PCR products were obtained when testing non-tuberculous mycobacteria and other bacteria confirming the specificity of the primer sets used. After alignment of the glpK and pykA gene sequences herein achieved and those downloaded from GenBank, we observed MTC species-specific signatures within the glpK gene. Four silent SNPs were found at nucleotide (nt) position 279 (A→G), 336 (T→C), 435 (T→A) and 1,497 (T→C) specifically within M. canettii. Insertion of one and three guanine at nt 573 was identified in M. microti and M. bovis BCG, respectively. M. bovis reference isolates also shared the same frameshift as M. microti at nucleotide 573 in addition to a G→C SNP at nt 779, but these polymorphisms were not found in the human and animal clinical M. bovis isolates (N = 10), in agreement with a previous study [23]. Thus, these SNPs further were not considered as M. bovis specific feature for subsequent analysis. M. pinnipedii specifically exhibits a G→A SNP at nt 845 and M. tuberculosis a C→T SNP at nt 1,397. Excluding M. bovis reference strains, no specific feature in the glpK gene was observed within M. bovis and M. caprae isolates. As for the pykA gene, a unique G→T SNP at nt 660 was commonly detected in M. bovis, M. caprae, M. microti and M. pinnipedii. This SNP substitutes glutamic acid with aspartic acid at codon 220, resulting in an inactive pyruvate kinase and the inability of M. bovis to use glycerol as a carbone source [23, 24]. By analyzing a larger collection of isolates, our results confirmed the presence of this SNP within M. bovis and provide the novel evidence for its occurrence also in M. caprae and M. pinnipedii which were not included in the Keating's study [23]. Accordingly, M. bovis, M. caprae and M. pinnipedii are all characterized by the inability to use glycerol and require pyruvate for growth [24]. We were not able to sequence glpK and pykA genes in M. africanum, nevertheless positive results were obtained by pyrosequencing analysis (see below). Polymorphic regions within the glpK and pykA genes were targeted for MTC species identification using a pyrosequencing approach. Since no specific signature was obtained for M. bovis and M. caprae, we combined previously described SNP of the gyrB gene to allow complete MTC discrimination [15]. In a first step, the assay was evaluated with the well characterized MTC species (N = 55). The pyrograms obtained from simplex and multiplex data were consistent with the results obtained by conventional DNA sequencing and correctly identifies MTC isolates including 22 M tuberculosis, 21 M. bovis, 2 M. bovis BCG, 3 M. caprae, 3 M. microti, 2 M. pinnipeddii, 1 M. canettii and 1 M. africanum type I. Accordingly, we propose a generalized diagram for the rapid identification of the MTC species (Figure 1). In a first step, multiplex amplification and pyrosequencing analysis of the glpK 1379 and pyk A660 nt positions in one reaction tube provide four different pyrograms: T/T and G/G specific for M. tuberculosis, T/T and T/T specific for M. africanum type I, C/C and G/G specific for M. bovis BCG and M. canettii (group 1) and C/C and T/T specific for M. bovis, M. caprae, M. microti and M. pinnipedii (group 2). If pyrogram of group 1 is obtained, sequencing of the glpK 573 may discriminate between M. bovis BCG and M. canettii. In the case where pyrogram of group 2 is obtained, analysis of the gyrB 6406 SNP may specifically identifies M. bovis (genotype C/C) from the remaining group 2 species (genotype T/T). M. pinnipedii can be identified by SNP analysis of the glpK 845 (genotype A/A), M. caprae by SNP analysis of the gyrB6307 and M. microti by sequencing the glpK 573. These steps can be done in parallel in order to minimize the delay for final identification of any MTC species. A total of 127 MTC clinical isolates were tested following this scheme and were unambiguously identified as M. tuberculosis.

Our proposed scheme allows in one step the specific identification of M. tuberculosis, the leading cause of human tuberculosis worldwide and M. africanum type I, an important cause of human tuberculosis in west Africa [6, 25]. M. bovis, an important zoonotic tuberculosis agent and M. bovis BCG a safe vaccine exceptionally responsible for disseminated disease in immuno-compromised vaccinated neonates and bladder cancer-patients, could be unambiguously identified in a second reaction after excluding M. tuberculosis and M. africanum type I [26, 27]. The remaining MTC species are less common in human disease and their identification using specific genomic marker described here could be directed by epidemiologic data.

The ability of rapid and accurate identification of MTC members has an important impact both for public health and veterinary facilities. Although a few patients infected with M. bovis have been treated by antibiotic regimens incorporating the pyrazinamide, the natural resistance of these species to pyrazinamide makes pyrazinamide generally not recommended for treating such patients [10]. Moreover, it can help epidemiologists and health care professionals to measure the relative contribution of each MTC species in human and animal disease, to identify specific MTC-specie outbreak and to rapidly intercept the impact of the zoonotic transmission particularly in case of M. bovis infections so that appropriate control measure could be undertaken.

Pyrosequencing analysis of the glpK, pykA and gyrB genes provides a robust and easy tool for the rapid identification of all MTC species that can be easily introduced in the routine laboratory activity, just requiring basics skills in PCR. The assay allows the simultaneous analyses of up to 96 isolates within 10 to 20 min after a 4-hour DNA extraction and PCR amplification rounds comprising of 3 different PCRs, and 7 different pyrosequencing steps which can be run in parallel. Moreover, pyrosequencing has several advantages compared to other molecular methods: it determines the exact sequence thereby providing the same accuracy as conventional sequencing methods, the technique dispenses with the need for labeled nucleotides, labeled primers and electrophoresis.