Relationship between Pyrazinamide Resistance, Loss of Pyrazinamidase Activity, and Mutations in the pncA Locus in Multidrug-Resistant Clinical Isolates of Mycobacterium tuberculosis

Abstract

Sixty-two Mycobacterium tuberculosis isolates were tested for pyrazinamidase activity, and their pyrazinamide susceptibility was determined by the radiometric method. Sequencing of pncA genes in the 23 resistant strains revealed mutations in 16 pyrazinamidase-negative strains, 11 of which had not been previously described. Six isolates containing wild-type pncA might possess alternative resistance mechanisms.

Pyrazinamide (PZA), one of the first-line drugs for tuberculosis treatment, is assumed to be a prodrug, converted by the bacterial enzyme pyrazinamidase (PZase) to the toxic pyrazinoic acid (POA), the target of which remains unknown (7). Since most PZA-resistant Mycobacterium tuberculosis strains have lost PZase activity, enzyme activity assays are sometimes used to replace PZA susceptibility testing, which is problematic because of growth problems at the acid pH (±5.5) required for in vitro activity of PZA (2, 4). Single mutations in the pncA gene encoding this PZase are considered the major PZA resistance mechanism in M. tuberculosis (5, 11, 12, 15). Our objective was to gain further insight into PZA resistance mechanisms and their molecular-biological background by comparing different methods for detection of PZA resistance in M. tuberculosis.

Fifty-four multidrug-resistant M. tuberculosis strains were isolated from clinical specimens collected in Bangladesh, Azerbaijan, Siberia, Belgium, Abkhazia, Rwanda, Congo-Brazzaville, Kazakhstan, and Canada, and eight isolates were obtained from the Scottish Mycobacteria Reference Laboratory (SMRL). M. tuberculosis H37Ra (NCTC 7417) was included as a PZA-susceptible control strain. One Mycobacterium bovis isolate from Belgium and one from Egypt from the collection of the Institute of Tropical Medicine were used as PZA-resistant control organisms. M. tuberculosis strains were grown to confluence on Löwenstein-Jensen medium, and M. bovis strains were grown on Stonebrink medium.

Isolates were tested for PZA susceptibility by the BACTEC radiometric method (3, 10, 13) at 100 μg of PZA per ml according to the manufacturer’s instructions (14).

The PZase activity of all 65 isolates was assayed, following the modified version of the Wayne (17) test, as described by Jakschik (6). For each isolate, one test tube and one control tube were prepared. The test medium contained 6.5 g of Dubos broth base (Difco Laboratories, Detroit, Mich.), 1.0 g of PZA (Janssen Chimica, Geel, Belgium), 2.0 g of sodium pyruvate (Janssen Chimica), and 15.0 g of agar (Difco Laboratories) in 1 liter of water. In the control medium, PZA was omitted. Freshly prepared autoclaved media were divided into 5-ml aliquots in glass tubes and were allowed to cool in the upright position. Next, 0.2 to 0.3 ml of sterilized water was aseptically added to both test and control tubes. Tubes were inoculated with a loopful of a freshly grown culture of each isolate by stabbing the agar two to three times onto the bottom. After incubation at 37°C for 4 days, 1.0 ml of a freshly prepared 1% ferro(II) ammonia sulfate (Acros Chimica, Geel, Belgium) solution was added to each tube, and the tubes were kept at 4°C for 4 h. A PZase activity assay was considered positive when a dark red to brownish ring appeared on the agar-water interface.

Genomic DNA was extracted as described previously (16). pncA genes were amplified by PCR (9) with primers P1 and P6 (12) to obtain DNA fragments of about 720 bp. The cycling parameters were 94°C for 1 min, 56°C for 1 min, and 72°C for 1 min; 40 cycles were performed. These PCR products were used to perform cycle sequencing by using fluorescence-labelled dideoxynucleotide terminators with the PRISM Ready Reaction Dye Deoxy Terminator Cycle Sequencing kit (Applied Biosystems, Foster City, Calif.) and primers P1, P3, P4, and P6 (12) according to the manufacturer’s instructions. Sequence data were generated with an automated DNA sequencer, model 373A (Applied Biosystems).

Thirty-nine isolates were susceptible to PZA; they all displayed PZase activity. The pncA sequence of these isolates was not determined, because earlier investigations (11, 12, 15) of sequences in 75 PZA-susceptible M. tuberculosis strains did not reveal any nucleotide change. Twenty-three isolates were resistant to PZA. For these strains, sequencing of the pncA open reading frame and an 85-bp upstream, putative regulatory region was performed. The results obtained are summarized in Table 1. In 16 PZA-resistant isolates, the lack of PZase activity was correlated with mutations leading to amino acid substitutions (10 of 16), large deletions (1 of 16), frameshifts (4 of 16) by single nucleotide insertion or deletion, or promoter mutations (1 of 16). In four PZA-resistant isolates lacking PZase activity, wild-type pncA sequence was present. Three PZA-resistant isolates displayed PZase activity: the wild-type pncA sequence was present in two of them, while the other contained a nucleotide change leading to an amino acid substitution (isolate 98-32 [Ala to Val at codon 171]). For strains with discrepant results between susceptibility testing and the PZase assay (PZA resistant, but PZase positive) or between the PZase assay and pncA sequencing (PZase negative, but pncA wild type), the results obtained were confirmed in a second test round.

TABLE 1.

PZase activity, changes in pncA nucleotide sequence, and corresponding amino acid changes of the 23 PZA-resistant M. tuberculosis isolates in this study

Isolate	Origin	PZase assay result	Change(s) in:
			Nucleotide sequence	Amino acid sequence
8251	Rwanda	−	Δ C at 512	Frameshift
94-735	Congo	+	No	No
97-56	Siberia	−	No	No
97-58	Siberia	−	No	No
97-69	Siberia	−	No	No
97-553	Belgium	−	C to G at 514 and T to C at 515	L172A
97-557	Belgium	−	C to G at 514 and T to C at 515	L172A
97-578	Bangladesh	−	C to T at 425a	T142M
97-718	Azerbaijan	−	T to G at 254	L85R
97-720	Azerbaijan	−	C insertion 407–408	Frameshift
97-826	Canada	−	T to C at 307	Y103H
97-854	Bangladesh	−	G to A at 511	A171T
97-983	Bangladesh	−	No	No
97-1006	Bangladesh	−	A to C at 410a	H137P
97-1052	Azerbaijan	+	No	No
97-1069	Azerbaijan	−	Δ 68 bp 195–263	Frameshift
97-1174	Azerbaijan	−	C to A at 185	P62H
98-29	Scotland	−	G insertion 221–222	Frameshift
98-30	Scotland	−	G insertion 221–222	Frameshift
98-31	Scotland	−	C to T at 151	H51Y
98-32	Scotland	+	C to T at 512	A171V
98-34	Scotland	−	G to C at 406	D136H
98-36	Scotland	−	A to G at −11b	No

^aPreviously described mutations (12). 

^bPreviously described mutation (15). 

Detection of PZase activity, as described by Wayne (17) and modified by Jakschik (6), is an easy and very rapid technique suited for any mycobacterial laboratory. Considering BACTEC PZA susceptibility determination as a reference, PZase negativity was correlated with PZA resistance in all 20 strains in this study. However, for only 39 of the 42 PZase-positive strains, a PZA-susceptible pattern was obtained on BACTEC. This confirms the findings of Butler and Kilburn (1) and Miller et al. (8) that PZA-resistant isolates are not always PZase negative. Our conclusion is that PZase test results should only be interpreted in one direction—a negative result indicating PZA resistance—whereas care should be taken in interpreting positive test results.

Mutations were present in 17 (74%) of the 23 PZA-resistant isolates as defined by the BACTEC radiometric method. The six strains lacking pncA mutations are divided into two groups by the PZase test: one group of two strains was PZase positive, and a second group of four strains was PZase negative. Resistance to PZA in the first group may be due to mutations in the POA drug target, which has not yet been found. The second group could have undergone downregulation of the pncA gene, directed on the translation phase or on regulation of transcription, as an alternative to inactivation of the gene product by point mutations in the open reading frame. Two of these strains had a completely identical restriction fragment length polymorphism pattern (97-56 and 97-58 [data not shown]). Additional research on these strains should help to elucidate the mechanism of action of the antituberculous agent POA.