Pyrazinamide Resistance and pncA Gene Mutations in Mycobacterium tuberculosis

Abstract

Thirty-four pyrazinamide-resistant and 37 pyrazinamide-susceptible Mycobacterium tuberculosis complex strains were analyzed for pncA gene mutations. None of the sensitive strains had any mutations, apart from silent mutations, whereas all but one resistant strain showed pncA mutations. By using sequencing as a means of early resistance detection, the inconsistency of phenotypic pyrazinamide assays can be circumvented.

Pyrazinamide (PZA) is an effective first-line drug for the treatment of tuberculosis (TB). It is a prodrug that requires conversion into its active form, pyrazinoic acid, by the bacterial enzyme pyrazinamidase (PZase), which is encoded by the 561-nucleotide (nt) pncA gene (19, 22). Mutations in pncA result in lost or reduced PZase activity, and such mutations are thus considered to be the primary mechanism of PZA resistance in Mycobacterium tuberculosis (6, 22). pncA gene mutations have been detected in PZA-resistant clinical isolates of M. tuberculosis, as well as in Mycobacterium bovis, which is naturally resistant to the drug (19).

Among clinical isolates, nucleotide changes can be found throughout the pncA gene and have been shown previously to correlate closely with phenotypic PZA resistance (2, 4, 6, 7, 9-13, 16, 18, 20). Through site-directed mutagenesis, nine codons in pncA were demonstrated previously to be vital for the catalytic activity and metal ion binding capacity of PZase (21). However, PZA-resistant strains without pncA mutations may occur with significant prevalence, as shown for a majority of clinical isolates in a study from Brazil (15).

PZA is active only at low pHs, which occur in vivo when inflammatory cells produce lactic acid (22). Since a low pH is inhibitory to the in vitro growth of M. tuberculosis, conventional drug susceptibility testing (DST) of PZA on solid medium is of limited value. The radiometric BACTEC 460 system (Becton Dickinson Biosciences, Sparks, MD) is considered to be a reference method for the reliable and rapid testing of the susceptibilities of M. tuberculosis isolates to the first-line anti-TB drugs. For PZA testing, there is a modified test protocol using an acidified culture medium (pH 6.0). Due to the fact that many laboratories have already discontinued or will discontinue their use of the BACTEC 460 system, there is a pronounced need to identify substitutes for this reference technique. The newer liquid culture-based methods, such as the nonradiometric BACTEC MGIT 960 (Becton Dickinson Biosciences, Sparks, MD) and the BacT/ALERT 3D (bioMérieux Inc., Durham, NC) methods, utilize protocols adapted for PZA DST but are not yet as well documented as the BACTEC 460 method (1, 3, 8, 17).

When the role of the pncA gene was first described, it became possible to use a molecular marker for PZA resistance (19). Molecular detection of PZA resistance-related mutations could be used for the early identification of resistant strains. A DNA-based method may also increase the accuracy of PZA susceptibility testing, since difficulties (such as pH-sensitive strains and discrepancies in results from various methods) may complicate the phenotypic DST methods. Additionally, by using pncA sequencing for clinical specimens, such as smear-positive sputum specimens, the DST turnaround time could be shortened significantly.

In the present study, we used conventional sequencing to evaluate the role of pncA gene mutations as markers for the detection of PZA resistance in M. tuberculosis.

A total of 72 clinical isolates from our strain collection were selected, of which 70 were M. tuberculosis, one was M. bovis, and one was M. bovis BCG. Among these isolates, 35 were previously found to be resistant to PZA, and 21 of these PZA-resistant isolates were also multidrug-resistant TB strains. The selected PZA-susceptible isolates had varied drug susceptibility patterns and included 10 isolates that were multidrug-resistant TB strains. The PZA-susceptible strain H37Rv (ATCC 25618) and a PZA-resistant M. bovis strain (ATCC 19210) were also included in the study as reference control strains. All strains were grown on Löwenstein-Jensen egg medium for 3 to 4 weeks at 37°C prior to DST.

Susceptibility to PZA was determined by the radiometric BACTEC 460 assay according to the instructions of the manufacturer (Becton Dickinson Biosciences, Sparks, MD). Briefly, 0.1 ml of a bacterial suspension with a McFarland standard of approximately 1 was inoculated into a BACTEC 12B medium vial supplemented with reconstitution fluid, and after reaching a growth index (GI) of greater than 300, the vial contents were used as the inoculum source for the subsequent PZA test. For each strain, two BACTEC PZA test medium culture vials were inoculated, one drug-free control vial supplemented with reconstitution fluid and one vial containing 100 mg of PZA/liter. The test was interpreted when the contents of the control vial reached a GI of 200. A strain was considered to be resistant if the GI for the PZA vial was >11% of the GI for the control vial and was considered to be PZA susceptible if the GI was <9% of the GI of the control. Of the 72 clinical isolates, 34 were determined to be resistant to PZA (at 100 mg/liter) when tested by the BACTEC 460 reference method.

The 561-nt pncA gene, along with surplus regions of approximately 200 nt up- and downstream of the gene, was sequenced using the pncA_F3 (AAGGCCGCGATGACACCTCT) and pncA_R4 (GTGTCGTAGAAGCGGCCGAT) primers. These primers were used in a standard PCR to give a template for the subsequent sequencing reactions. The pncA_F3 and pncA_R4 primers, as well as the P3-F (ATCAGCGACTACCTGGCCGA) and P4-R (GATTGCCGACGTGTCCAGAC) primers, were used to subdivide the PCR fragment into two overlapping bidirectional sequencing reaction fragments. The sequencing reactions were performed using the BigDye Terminator cycle sequencing kit and a 3100 genetic analyzer (Applied Biosystems, Inc., Foster City, CA). Retrieved sequences were then analyzed with the (ClustalW) vector NTI Advance 9 software (InfoMax, Inc.) using the wild-type (wt) H37Rv strain's pncA gene (Rv2043c) as the master sequence.

A high degree of correlation between nucleotide changes in the pncA gene and resistance to PZA as determined using the BACTEC 460 system was seen. As in previous studies (2, 14, 15, 18), mutations were distributed along the whole pncA gene in our clinical isolates (Table 1).

TABLE 1.

pncA gene sequences and PZA drug susceptibilities for 71 clinical isolates belonging to the M. tuberculosis complex

No. of isolates	Susceptibility to PZAa	pncA gene description or nucleotide change(s)b	PZase description or amino acid change(s)
34	S	wt	wt
1	S	C insertion between nt −3 and −2	wt
1	S	G insertion between nt −33 and −32	wt
1	S	C(195)→T	Ser(65)→Ser
1	R	wt	wt
1	R	A(−11) →G	wt
1	R	T(17)→C	Ile(6)→Thr
1	R	C deletion at nt 99	Frameshift
1b	R	G(117)→A, C(169)→G	Ala(39)→Ala, His(57)→Asp
3	R	A(139)→G	Thr(47)→Ala
2	R	C(153)→A	His(51)→Gln
1	R	C(160)→T	Pro(54)→Ser
1	R	C(161)→T	Pro(54)→Leu
1c	R	C(169)→G	His(57)→Asp
1	R	T(172)→C	Phe(58)→Leu
1	R	C(174)→G	Phe(58)→Leu
1	R	C(195)→T, GC insertion between nt 528 and 529	Ser(65)→Ser, frameshift
1	R	T(199)→C	Ser(67)→Pro
1	R	T(202)→C	Trp(68)→Arg
1	R	T(269)→G	Ile(90)→Ser
1	R	T(307)→C	Tyr(103)→His
1	R	G(368)→C	Arg(123)→Pro
1	R	G(373)→T	Val(125)→Phe
2	R	T(374)→A	Val(125)→Asp
1	R	GG insertion between nt 391 and 392	Frameshift
1	R	G(394)→A	Gly(132)→Ser
1	R	G(395)→C	Gly(132)→Ala
1	R	G(413)→A	Cys(138)→Tyr
1	R	C(425)→A	Thr(142)→Lys
1	R	12-nt insertion between nt 444 and 445	In-frame insertion between amino acids 148 and 149
1	R	T(464)→G	Val(155)→Gly
2	R	T(515)→C	Leu(172)→Pro
1	R	T(524)→C	Met(175)→Thr

^aS, sensitive; R, resistant.

^bM. bovis.

^cM. bovis BCG.

For the 34 strains phenotypically resistant to PZA, mutations in the pncA gene were detected in 32 cases, giving the rapid molecular test a sensitivity of 94.1%. Sequencing showed wt pncA sequences in 36 of the 37 PZA-susceptible strains, giving a corresponding specificity of 97.3%. The aberrant strain contained a silent mutation, Ser65Ser. Considering that this strain had an unchanged protein sequence, the results from the phenotypic DST of all sensitive strains correlated fully with the sequencing data. In addition, of the 37 clinical isolates susceptible to PZA, 2 strains had 1-nt insertions upstream of the gene (a G insertion between nt −33 and −32 and a C insertion between nt −3 and −2 relative to the pncA start codon).

Two phenotypically PZA-resistant strains had a wt pncA nucleotide sequence. In one of these strains, a nucleotide change (A to G) in the putative regulatory area, located 11 nt upstream of the pncA gene, was detected. As opposed to the mutations upstream of the gene in the sensitive strains, the A-to-G substitution has been found previously in PZA-resistant strains that showed wt pncA sequences, thereby indicating that this mutation affects the regulation of protein synthesis (1, 4, 9, 20). For the resistant strain without any demonstrated mutation, there may be an alternative mechanism of resistance, such as an elevated efflux pump mechanism, suggested earlier to be related to low-level PZA resistance in strains with a wt pncA genotype (18, 22).

One strain with a wt pncA gene repeatedly yielded borderline results in the BACTEC testing. It thus could not be classified as phenotypically resistant or susceptible and was therefore excluded from the study. If a higher concentration of the drug had been used, the strain would most likely have been considered susceptible, which is in line with our finding of a wt sequence. This outcome may strengthen the suggestion of using an increased PZA concentration for routine testing (4, 23).

As expected, the M. tuberculosis H37Rv reference strain demonstrated a wt pncA sequence and the M. bovis PZA-resistant control strain had a His57Asp mutation, a trait typical of M. bovis (19). This substitution was also found in both the M. bovis and the M. bovis BCG clinical isolates, but in no other isolate.

A very clear majority of the resistance-related mutations identified were single-nucleotide substitutions (Table 1), while multiple mutations (in two strains), insertions (in three strains), and a deletion (in one strain) were seen more rarely. The deletion and two of the insertions were out of frame, giving missense mutations and truncated amino acid chains. A strain with a 12-nt insertion in between codons 148 and 149, despite the mutation's being in frame, was also found to be PZA resistant. A similar mutation, but differently positioned, has been reported earlier and was postulated to be caused by slipped-strand mispairing (20).

The presence of the Thr47Ala mutation, which seems to be specific for Beijing strain W isolates of M. tuberculosis, was recently reported in strains found to be PZA sensitive in the BACTEC 460 system (5). Among our isolates, this substitution was detected only in Beijing strains, but all three of our isolates with this substitution were PZA resistant, indicating a general problem in the reproducibility of the results of the phenotypic assays. Three other substitutions (Ile6Thr, Pro54Ser, and Val125Asp) were found among four other strains belonging to the Beijing family.

We consider that the analysis of the pncA gene provides rapid and useful information regarding PZA susceptibility in M. tuberculosis and may therefore contribute to early optimization of treatment. The high degree of diversity of pncA gene mutations among PZA-resistant isolates may also be epidemiologically useful in predicting linked cases of TB. Furthermore, pncA sequencing may be a reliable alternative to phenotypic PZA DST techniques, which sometimes give irreproducible results and discordance among findings from different laboratories.