Rapid Differentiation of “Mycobacterium canettii” from Other Mycobacterium tuberculosis Complex Organisms by PCR-Restriction Analysis of the hsp65 Gene

Abstract

A total of 102 isolates of the Mycobacterium tuberculosis complex, including available “M. canettii” isolates, were studied by PCR-restriction analysis of a 441-bp fragment of the hsp65 gene. PRA upon HhaI enzyme digestion (GCGC) allowed easy differentiatiation of “M. canettii” from other members of the M. tuberculosis complex (three bands of 260, 105, and 60 bp for “M. canetti,” compared to four bands of 185, 105, 75, and 60 bp for other members of the M. tuberculosis complex). Sequencing of the 441-bp hsp65 fragment of “M. canettii” isolates showed the disappearance of an HhaI site at position 235 due to a C-to-T transition that corresponded to position 631 of the homologous hsp65 gene of M. tuberculosis H37Rv. Considering that “M. canettii” may also exist as a stable rough morphotype, we suggest that the true number of “M. canettii” isolates may be underestimated in clinical microbiology laboratories.

The Mycobacterium tuberculosis complex includes five recognized members; M. tuberculosis and M. africanum are human pathogens, M. bovis is known to infect a wide range of domestic and wild animals as well as humans, M. bovis BCG is a vaccine strain with attenuated virulence, and M. microti is a pathogen of small rodents. Another member named “M. canettii” was added to this list of M. tuberculosis complex organisms in 1997 (the name is given in quotation marks since it does not yet appear on the approved list of bacterial names) (16). In 1969, Georges Canetti, of the Institut Pasteur, Paris, France, was the first to isolate “M. canettii” in France. It was isolated as a smooth-colony variant of M. tuberculosis from a 20-year-old farmer suffering from pulmonary tuberculosis. The original isolate, isolate HB3177, was reported to be virulent for guinea pigs and was temporarily designated “M. tuberculosis var. canettii (negrei).” In 1969 and 1970, two other “M. tuberculosis var. canettii” strains (strains HB3178 and HB3253) were isolated from a 54-year-old dairy-farm worker who had lived in Madagascar and who was suffering from pulmonary tuberculosis and from the pus of a patient of unknown age in Papeete, French Polynesia, who was suffering from adenitis of the armpit, respectively. These three isolates were found to constitute a homogeneous group of isolates within the M. tuberculosis complex and were deposited in the Culture Collection of the Pasteur Institute-Tuberculosis (CIPT) under the numbers CIPT140010059, CIPT140010060, and CIPT140010061, respectively. Studies of the cell wall glycolipids showed that a single phenolic glycolipid (PGL-Tb1) was produced at high levels by these isolates and may explain the smooth phenotype observed in “M. canettii” strains (8, 10).

Recently, two other “M. canettii” strains were isolated (11, 16) and were shown to be similar to the original M. canettii strain; they had in common a single copy of IS1081, two copies of IS6110, identical recA gene sequences, and identical spoligotypes (11, 15, 16). In the present investigation, we have investigated M. tuberculosis complex organisms by PCR-restriction analysis (PRA) of a 441-bp fragment of the hsp65 gene and show that the “M. canettii” taxon may easily be differentiated from other members of the M. tuberculosis complex on the basis of a specific PRA signature upon HhaI digestion.

Bacterial isolates.

A total of 102 isolates of the M. tuberculosis complex were studied, including 50 isolates of M. tuberculosis (48 clinical isolates and the type strains M. tuberculosis H37Rv and H37Ra), 20 isolates of M. bovis (19 clinical isolates and the type strain M. bovis ATCC 19210), 10 isolates of M. bovis BCG (strains 001-Pasteur, 003-USA, 004-Denmark, 005-Glaxo, 006-Athens, 007-Japan, 008-Montreal, 009-Brazil, 011-Russia, and 012-Sweden), 15 isolates of M. africanum (14 clinical isolates and the type strain M. africanum ATCC 25420), and the following 7 isolates of “M. canettii”: the original “M. tuberculosis var. canettii” strain CIPT140010059 from the Institut Pasteur as well as the same strain under the designation 9600046 maintained at the National Institute of Public Health and Environment, Bilthoven, The Netherlands; strain 910563, isolated from a patient in 1991 at the Institut Pasteur, Paris; strain 217/94, isolated from a 56-year-old Swiss patient (smooth colony type) (11), and the same strain under the designation 9701549 maintained at the National Institute of Public Health and Environment, which produced both smooth and rough colony types upon culture in our laboratory; strain So93 (smooth colony type), isolated from a 2-year-old Somali patient (16); and strain So93R (rough colony type), obtained upon culture of a single colony of So93. All isolates were grown on fresh Löwenstein-Jensen slants at 37°C, and biochemical tests and drug susceptibility testing were performed by classical procedures (4). “M. canettii” isolate 217/94 was kindly provided by Gaby Pfyffer, Swiss National Center for Mycobacteria, Zurich, Switzerland, “M. canettii” isolates 9600046, 9701549, 17727 (S093R), and 17728 (So93) were kindly provided by Dick van Soolingen, Mycobacteria Laboratory, National Institute of Public Health and Environment, whereas all other isolates including “M. canettii” isolates CIPT140010059 and 910563 were from our own culture collection.

PRA and sequencing of hsp65 gene.

DNA was prepared by a glass bead method (5), and a 5-μl aliquot of the supernatant containing the crude DNA extract was used for PCR. Amplification was performed with primers Tb11 (5′-ACCAACGATGGTGTGTCCAT) and Tb12 (5′-CTTGTCGAACCGCATACCCT), which amplified a 441-bp fragment of the hsp65 gene (positions 396 to 836 of the published sequence of M. tuberculosis H37Rv), followed by digestion of the PCR product with BstEII (Promega, Madison, Wis.) and with HaeIII, AciI, and HhaI (BioLabs Inc., Beverly, Mass). After digestion, 12 μl of the restriction digest was loaded onto a 3% (wt/vol) Metaphor agarose gel (FMC Bioproducts, Rockland, Maine). An external molecular weight marker (100 bp-ladder; AP Biotech, Uppsala, Sweden) was added to every six lanes of migration to reduce migration-related errors. The fragments were visualized using ethidium bromide; and the images were captured on video, digitized, and analyzed with Gel-Analyst software (Bioprobe Systems, Montreuil, France). The fragment lengths were calculated with Taxotron software (Taxolab, Institut Pasteur, Paris, France), as reported previously (5).

Sequencing of the hsp65 gene was performed as described previously (9). Briefly, the 21M13 forward primer TB11 (5′-T GTAAAACGACGGCCAGTACCAACGATGGTGTGTCC AT-3′) and the M13 reverse primer TB12 (5′-CAGGAAACAGCTATGACCCTTGTCGAACCGCATACCCT-3′) (underscores indicate forward and reverse primers [21M13 and M13, respectively]) were used to amplify a 441-bp portion of the hsp65 gene (9, 13). The amplification product (10 μl) was first checked by agarose gel electrophoresis, and 40 μl was purified with the QIAquick PCR purification kit (Qiagen, Courtaboeuf, France). The PCR product was sequenced with the Thermo-Sequenase fluorescence-labeled-primer cycle sequencing kit (AP-Biotech, Piscataway, N.J.) with fluorescent primers Cy5.5-Tb11 and Cy.5.0-Tb12 (Visible Genetics Inc., Toronto, Ontario, Canada) on an Opengene Long Read Tower sequencing system (Visible Genetics Inc.), according to the manufacturer's instruction. All the sequencing data are from two independent experiments, and any discrepancy was systematically checked by sequencing a third time. Sequencing data were compared by using the sequences in the GenBank database and the BlastN algorithm (9).

The biochemical and cultural characteristics and drug susceptibility patterns of the seven “M. canettii” strains studied are summarized in Table 1. The available “M. canettii” isolates were easily differentiated from M. tuberculosis on the basis of their morphologies (eugonic, smooth colonies), a negative response for niacin production and β-glucosidase enzyme activity, and resistance to pyrazinamide and streptomycin. “M. canettii” was also easily discriminated from M. bovis and M. bovis BCG on the basis of its colony morphology, its ability to reduce nitrates and to hydrolyze Tween 80 (10 days), and resistance to thiophene-2-carboxylic acid hydrazide. A positive response of “M. canettii” for urease enzyme activity further differentiated it from M. bovis, whereas the susceptibility of “M. canettii” to d-cycloserine and the resistance of “M. canettii” to clarithromycin were able to discriminate it from M. bovis BCG (Table 1).

TABLE 1.

Biochemical and cultural characteristics and drug susceptibility testing results for various subspecies of M. tuberculosis complex^a

Subspecies (no. of strains)	Colony morphology	Biochemical test results								Growth in the presence of:			Drug susceptibility				CLR MIC (μg/ml)
		Niacin production	Presence of:					Tween 80 hydrolysis	Presence of β-glucosidase	TCH	NAP	TB10	SM	d-CS	Other drugs	PZA
			Nitrate reductase	Catalase		Urease	Aryl sulfatase
				22°C	68°C
M. tuberculosis (50)	Eugonic (rough)	+	+	+	−	+	−	+	+	R	S	S	S	S	S	S	20
M. bovis BCG (10)	Eugonic (rough)	−	−	+	−	+	−	−	−	S	S	S	S	R	S	R	0.5
M. bovis (20)	Dysgonic (rough)	−	−	+	−	−	−	−	−	S	S	S	S	S	S	R	10
M. africanum (15)	Dysgonic (rough)	X	X	+	−	X	−	X	X	S	S	S	S	S	S	S	10
M. canettii CIPT140010059	Eugonic (smooth)	−	+	+	−	+	−	+	−	R	S	S	R	S	S	R	>32
M. canettii 9600046	Eugonic (smooth)	−	+	+	−	+	−	+	−	R	S	S	R	S	S	R	>32
M. canettii 910563	Eugonic (smooth)	−	+	+	−	+	−	+	−	R	S	S	R	S	S	R	>32
M. canettii 9701549	Eugonic (smooth and rough)	−	+	+	−	+	−	+	−	R	S	S	R	S	S	R	>32
M. canettii 217/94	Eugonic (smooth)	−	+	+	−	+	−	+	−	R	S	S	ND	S	S	R	>32
M. canetti 17727 (So93R)	Eugonic (rough)	−	+	+	−	+	−	+	−	R	S	S	R	S	S	R	ND
M. canetti 17728 (So93)	Eugonic (smooth)	−	+	+	−	+	−	+	−	R	S	S	R	S	S	R	>32

^aSymbols and abbreviations: +, positive result; −, negative result; X, variable results; R, drug resistant; S, drug suseptible; TCH, thiophene-2-carboxylic acid hydrazide (5 μg/ml); NAP, p-nitro-α-acetylamino-β-hydroxypropiophenone (5 μg/ml); TB10, thiacetazone (10 μg/ml); PZA, pyrazinamide (50 μg/ml); SM, streptomycin (2 μg/ml); d-CS, d-cycloserine (30 μg/ml); CLR, clarithromycin; ND, not done. Other drugs included isoniazid (0.2 μg/ml), rifampin (1 μg/ml), ofloxacin (1.5 μg/ml), kanamycin (6 μg/ml), amikacin (4 μg/ml), clofazimine (1 μg/ml), rifabutin (1 μg/ml), ciprofloxacin (1.5 μg/ml), and ethambutol (2 μg/ml). The Arylsulfatase result was read at day 3, whereas the Tween 80 hydrolysis result was read at day 10; Drug susceptibility and growth experiments were performed with 7H11 agar medium except for tests for growth in the presence of TB10 and susceptibility to d-cycloserine and ethambutol, which were performed on Löwenstein-Jensen medium. 

The smooth colony type of “M. canettii” may switch to give a stable rough colony type which is not reversible even after passage in guinea pigs (16). Interestingly, the switch from the smooth to the rough colony types did not alter the biochemical, cultural, and the drug resistance patterns of the “M. canettii” isolates (Table 1), a fact already observed at the level of the genetic markers studied previously (16). It is possible that the switch from the smooth colony type to the rough colony type is linked to changes in the lipid composition of the cell wall, as shown previously for M. avium (12), in which it is associated with large genomic deletions implicating either the ser2 gene cluster that encodes a glycopeptidolipid (GPL) haptenic oligosaccharide in serovar 2 isolates (6), or may further encompass the genes encoding lipopeptide biosynthesis, resulting in rough morphotypes that are devoid of any vestige of the GPL antigens (1). Although the molecular basis of the transition from the smooth colony type to the rough colony type is not yet known for “M. canettii,” Daffé et al. (3) have shown that, in addition to its ability to produce large amounts of triglycosyl phenol phthiocerol glycolipid (PGL-Tb1), smooth “M. canettii” strains also differ from the classical rough M. tuberculosis strains by the presence of large amounts of a characteristic lipooligosaccharides (LOSs), which is distinguished by the presence, among other sugars, of 2-O-Me-Fuc, 2-O-Me-Rha, and 4-O-Me-Rha (3). Interestingly, rough isolate So93R was devoid of this characteristic LOSs, which was present in smooth isolate So93 (16).

The results of PRA after digestions with BstEII, HaeIII, AciI, and HhaI are summarized in Fig. 1. Although the results were similar for all 102 isolates studied after the digestions with BstEII (three fragments of 235, 120, and 80 bp; Fig. 1A), HaeIII (four fragments of 155, 130, 70, and 40 bp; Fig. 1B), and AciI (three fragments of 185, 130, 90 bp; Fig. 1C), the digestion of the 441-bp PCR product with HhaI clearly differentiated “M. canettii” from all other isolates at the subspecies level (Fig. 1D); it resulted in three fragments of 260, 105, and 60 bp for “M. canettii,” whereas it resulted in four fragments of 185, 105, 75, and 60 bp for M. tuberculosis, M. bovis, and M. bovis BCG. Thus, the 185- and 75-bp fragments that characterize the M. tuberculosis complex by PRA upon HhaI digestion were apparently replaced by a 260-bp fragment in “M. canettii” strains, probably due to the disappearance of an HhaI restriction site in these strains. Sequencing of the 441-bp hsp65 fragment (Fig. 1E) showed a single mutation at position 235 (C-to-T transition), which corresponds to position 631 of the homologous hsp65 gene of M. tuberculosis H37Rv (2). The disappearance of an HhaI site yielded four fragments with theoretical sizes of 257, 103, 64, and 17 bp for “M. canettii,” whereas it yielded five fragments with theoretical sizes of 185, 103, 72, 64, and 17 bp for M. tuberculosis. However, as the fragment of 17 bp is not visible on the gels, it did correspond to a three-band profile (bands of 260, 105, and 60 bp) for “M. canettii” and a four-band profile (bands of 185, 105, 75, and 60 bp) for other members of the M. tuberculosis complex.

FIG. 1.

PRA patterns of the M. tuberculosis complex organisms upon digestion of the 441-bp PCR product with BstEII (A), HaeIII (B), AciI (C), or HhaI (D) and the sequence of the hsp65 fragment from “M. canettii” (E). Lanes 1, M. tuberculosis; lanes 2, M. bovis; lanes 3, M. bovis BCG; lanes 4, M. africanum; lanes 5, “M. canettii”; lanes C, negative control; lanes M, 100-bp ladder. The underlined portion in panel E shows the single mutation at position 235 (C-to-T transition) that is linked to the disappearance of a HhaI site and that corresponds to position 631 of the homologous hsp65 gene of M. tuberculosis H37Rv.

Although the results of PRA after BstEII, HaeIII, and AciI enzyme digestions did not discriminate among the various members of the M. tuberculosis complex, we found an excellent correlation between the theoretical sizes of the fragments based on hsp65 sequencing data and PRA results. Knowing that bands ≤40 bp are not easily visible on Metaphor gels, the values obtained were four theoretical fragments of 231, 116, 79, and 15 bp versus three observed bands of 235, 120, and 80 bp for BstEII; seven theoretical fragments of 152, 127, 69, 42, 22, 17, and 12 bp versus four bands of 155, 130, 70, and 40 bp for HaeIII; and three theoretical fragments of 190, 148, and 103 bp versus three bands of 185, 130, and 90 bp for AciI. This unambiguously corroborated the reproducibility of the PRA methodology under the experimental conditions used and justifies its routine use for the identification of mycobacteria, as established previously (5, 13, 14).

The “M. canettii” taxon may be further characterized by a specific spoligotyping signature, as recently demonstrated by the publication of 26 new “M. canettii”-specific spacers (15). Various members of the M. tuberculosis complex (M. bovis, M. microti, and M. africanum) have recently been reported to bear specific spoligotyping signatures (7, 15, 17, 18); however, PRA is easier to use and shows a greater sensitivity than spoligotyping for the rapid detection of M. tuberculosis complex isolates from smear-positive clinical samples (our unpublished results) and might be useful for determination of the true number of “M. canettii” isolates implicated in tuberculosis (human and/or animal) in the routine microbiology laboratory. Indeed, the prevalence of “M. canettii” taxon isolates may be underestimated due to their close similarity to M. tuberculosis, particularly as the smooth colony type easily reverts to a give a stable rough colony type in “M. canettii,” as shown previously (16) and during the present investigation; e.g., the original Swiss isolate 217/94 gave both smooth and rough colony types upon successive culture of derived isolate 9701549. However, irrespective of the smooth or rough colony types, all “M. canettii” isolates available so far are uniform in regard to their cultural and biochemical properties, spoligotyping signatures, and other genetic markers such as the loss of a HhaI restriction site, the presence of two to three copies of IS6110 and a single copy of IS1081 as determined by restriction fragment length polymorphism analysis, and the sequence of the recA gene. Further studies are now needed to discover the natural reservoir, host range, and mode of transmission of “M. canettii” as well as the basis for the transition from the smooth to the rough colony type, which has not been found in other members of the M. tuberculosis complex.