Introducing RpsA Point Mutations Δ438A and D123A into the Chromosome of Mycobacterium tuberculosis Confirms Their Role in Causing Resistance to Pyrazinamide

Pyrazinamide (PZA) is a unique frontline drug for shortening tuberculosis (TB) treatment, but its mechanisms of action are elusive. We previously found one PZA-resistant strain that harbors an alanine deletion at position 438 (Δ438A) in RpsA, a target of PZA associated with PZA resistance, but its role in causing PZA resistance has been inconclusive.

ABSTRACT

Pyrazinamide (PZA) is a unique frontline drug for shortening tuberculosis (TB) treatment, but its mechanisms of action are elusive. We previously found one PZA-resistant strain that harbors an alanine deletion at position 438 (Δ438A) in RpsA, a target of PZA associated with PZA resistance, but its role in causing PZA resistance has been inconclusive. Here, we introduced the RpsA Δ438A mutation along with the D123A mutation into the Mycobacterium tuberculosis chromosome and demonstrated that these RspA mutations are indeed responsible for PZA resistance.

INTRODUCTION

Pyrazinamide (PZA) is an important first-line drug that plays a unique role in shortening tuberculosis (TB) therapy from 9 to 12 months to 6 months due to its unique sterilizing activity in killing Mycobacterium tuberculosis persisters that are not killed by other TB drugs (1, 2). Despite its high activity in vivo, PZA is a peculiar drug that has virtually no activity in vitro under normal culture conditions (2, 3) but is active only at a low pH (4, 5). Recent studies of new drugs, such as bedaquiline, PA-824, and moxifloxacin, have highlighted the essentiality of PZA, as these new agents mainly work in conjunction with PZA but cannot replace it (6, 7). Thus, PZA is a key drug that will likely play a major role in new drug regimens for the treatment of TB and drug-resistant TB.

It is well established that mutations in the pncA gene encoding nicotinamidase/pyrazinamidase (PZase) involved in the conversion of PZA to its active form pyrazinoic acid (POA) confer the major mechanism of PZA resistance (8, 9), accounting for 85% of all PZA resistance (9). However, some PZA-resistant strains without pncA mutations may have mutations in potential targets of PZA, including the ribosomal protein S1 (RpsA) involved in trans-translation (10–14); aspartate decarboxylase (PanD) (15, 16) involved in the synthesis of β-alanine, a precursor for pantothenate and coenzyme A (CoA) biosynthesis; and the ATP-dependent protease ClpC1 (17, 18). We have previously identified a 3-base pair “GCC” deletion resulting in the loss of an alanine at amino acid 438 in RpsA in a low-level PZA-resistant clinical isolate, DHM444 (MIC of 200 to 300 μg/ml PZA compared with 100 μg/ml in susceptible M. tuberculosis), without the pncA mutation (10). rpsA mutations were subsequently shown to be associated with PZA resistance in clinical strains (14, 19, 20). However, certain nonsynonymous mutations in the gene encoding the RpsA protein (A364G) seem to occur in some PZA-susceptible strains (21). In addition, a recent study by Dillon et al. claims that RpsA is not a target of PZA and that the RpsA Δ438A mutation does not cause PZA resistance (22). These discordant results have raised doubts about RpsA being a target of PZA and the role of the RpsA Δ438A mutation in PZA resistance.

In this study, to more convincingly address the role of the RpsA Δ438A mutation in PZA resistance, we transferred this mutation into the genome of M. tuberculosis H37Rv by homologous recombination. It is worth noting that the same strategy has been successfully used to demonstrate inhA and certain embB point mutations as being responsible for isoniazid (INH) and ethambutol resistance, respectively (23, 24). The RpsA Δ438A mutation was created by a two-step allelic exchange method as described (25). Briefly, a 3,440-bp fragment spanning the rpsA Δ438A deletion mutation was amplified by PCR with DHM444 genomic DNA using primers containing HindIII and PacI restriction sites (underlined) in the forward primer (FrpsA 5′-CGGAAGCTTCCACACCACGTTCAACCAGAC-3′) and reverse primer (RrpsA 5′-GCTTAATTAAGCACGCGCTTGTGCCACAGAG-3′), respectively. The PCR fragment was then cloned into the p2NIL vector followed by insertion of a PacI cassette containing sacB and lacZ. The recombinant plasmid was transformed into M. tuberculosis H37Rv as described (16). The desired sucrose-resistant but kanamycin-susceptible transformants were analyzed by PCR and sequenced to confirm that the transformed M. tuberculosis had the correct RpsA Δ438A mutation (Fig. 1). We determined the PZA MICs for the RpsA Δ438A mutant and the parent strain M. tuberculosis H37Rv using the proportion method and found them to be 200 μg/ml and 100 μg/ml (pH 5.8), respectively. In addition, we determined the PZA and POA MICs for the RpsA Δ438A mutant in liquid 7H9 medium (pH 5.8) using a microdilution method. The PZA MIC for the M. tuberculosis RpsA Δ438A mutant strain was 300 μg/ml, while the parent strain H37Rv was susceptible at this concentration (Fig. 2A). The POA MIC for the M. tuberculosis RpsA Δ438A mutant strain was 50 μg/ml compared with 25 μg/ml for the parent strain H37Rv (Fig. 2B). On the other hand, both strains had the same MIC for the control drugs INH (0.03 μg/ml) and rifampin (RIF) (0.12 μg/ml) (Fig. 2C and D). These results clearly demonstrated that the transfer of the RpsA Δ438A mutation into the genome of M. tuberculosis caused resistance to PZA and POA specifically but not to INH and RIF. This finding provides conclusive evidence that the RpsA Δ438A mutation is indeed responsible for the PZA resistance in the strain DHM444.

FIG 1.

Confirmation of RpsA Δ438A and D123A point mutation construction by Sanger sequencing. Chromatogram of partial rpsA sequence showing the “GCC” (coding for alanine) (A) deletion in the constructed RpsAΔ438A mutant from PZA-resistant strain DHM444. (B) Chromatogram of partial rpsA sequence showing the presence of “GCC” (alanine) in the M. tuberculosis H37Rv parent strain. (C) Alignments of partial rpsA sequence from M. tuberculosis H37Rv wild type and the constructed RpsA Δ438A mutant at the nucleotide level (top) and amino acid level (bottom). (D and E) Chromatogram of partial rpsA sequence showing “GAC” (wild type) change to “GCC” which is contained in one PZA-resistant clinical isolate. (F) Alignments of partial rpsA sequence from M. tuberculosis H37Rv wild type and the constructed RpsA D123A mutant at the nucleotide level (top) and amino acid level (bottom).

FIG 2.

Drug susceptibility testing (DST) results of the constructed RpsA Δ438A mutant strain compared with the control strain H37Rv. The DST was performed in 7H9/ADC broth using the microdilution method with pH 5.8 for PZA and POA DST and pH 7.0 for INH and RIF DST. The constructed RpsA Δ438A mutant strain was more resistant to PZA (A) and POA (B) than the control strain H37Rv but was as susceptible to INH (C) and RIF (D) as the control strain H37Rv. (A) The PZA MIC of the M. tuberculosis H37Rv control strain was 300 μg/ml, while the M. tuberculosis RpsA Δ438A mutant strain was resistant at this concentration. (B) The POA MIC of M. tuberculosis H37Rv parent strain was 25 μg/ml, while the POA MIC for the M. tuberculosis H37Rv RpsA Δ438A mutant strain was 50 μg/ml. (C) Isoniazid (INH) MICs for the parent strain M. tuberculosis H37Rv and the M. tuberculosis H37Rv RpsA Δ438A mutant were the same (0.03 μg/ml). (D) The RIF MICs for the parent strain M. tuberculosis H37Rv and the M. tuberculosis H37Rv RpsA Δ438A mutant were the same (0.12 μg/ml). The red boxes are used to highlight the difference in PZA and POA susceptibility between the RpsA Δ438A mutant strain and the parent strain H37Rv.

It is worth discussing the possible causes for the discrepant results between that of the Baughn group (22) and the findings presented here. One possible reason is the unusual method used by that group to determine the PZA susceptibility by using optical density at 600 nm (OD₆₀₀) and MIC₉₀. Neither optical density OD₆₀₀ nor MIC₉₀ is an accepted method for PZA susceptibility testing of M. tuberculosis. It is very likely that the suboptimal PZA MIC determination using OD₆₀₀ and MIC₉₀ is not accurate or sensitive enough to detect the low-level resistance (about 2-fold) conferred by RpsA mutations in the Dillon study (22). PZA susceptibility testing is notoriously difficult because of the acid pH (below pH 6) requirement for drug activity to be manifest and many factors affect the susceptibility results (26). The inability to observe PZA or POA resistance in those rpsA-overexpression or point mutation strains could be attributed to the questionable method used in that study (22). Another argument used by Dillon et al. against the RpsA Δ438A mutation conferring PZA resistance (22) is that the PZA resistance of DHM444 could be due to defective PZase activity reported in a previous study (27). To address this question, we tested PZase activity for DHM444 as described (15) and found that contrary to the previous report (22), the DHM444 strain still had PZase activity as did the wild-type strain M. tuberculosis H37Rv, while the pncA mutation (T254C and L85P) strain clearly exhibited reduced PZase activity (Fig. 3). Thus, the low-level PZA resistance in DHM444 could not be attributed to defective PZase activity, as claimed by Dillon et al. (22), and this study provides clear evidence that the DHM444 PZA resistance is most likely is due to the RpsA Δ438A mutation, as demonstrated (Fig. 2A and B).

FIG 3.

PZase activities of PZA-resistant DHM444 strain and other control strains. PZase assay is based on the color reaction of ferrous iron (II) with POA, which is converted from PZA by PZase from M. tuberculosis H37Rv, DHM444, and PncA mutant (L85P) strain. PZA was added to 0.5-ml log phase cultures in triplicate at 15 mg/ml at 37°C for 48 h. After incubation, 30 μl of 20% ammonium iron (II) sulfate was added to the reaction mixtures. The PZA-resistant DHM444 strain and H37Rv control strain show positive PZase activity, as seen by dark brown color (left two strains), whereas the PncA mutant (L85P) and no PZA control (right two strains) show reduced PZase activity with light brown or no color.

In addition, in our previous study we found another low-level PZA-resistant clinical isolate that has two RpsA mutations, namely, T5S and D123A, with the wild-type pncA gene (10). To provide further proof that RpsA point mutations can cause PZA resistance, we constructed another RpsA mutation D123A into the genome of M. tuberculosis H37Rv. Site-directed mutagenesis of rpsA was performed with primers S1D123AF (5′-GCTCAAGGAGAAGGCCGAGGCCGTCAAGG-3′) and S1D123AR (5′-CCTTGACGGCCTCGGCCTTCTCCTTGAGC-3′) using the QuikChange site-directed mutagenesis kit, as described by the manufacturer (Agilent Technologies). Then, the point mutation RpsA D123A was transferred to the genome of the parental strain M. tuberculosis H37Rv, as described as above. We found that, indeed, the PZA MIC for the M. tuberculosis RpsA D123A mutant was increased by 2-fold (200 μg/ml PZA) compared with the parental strain M. tuberculosis H37Rv (100 μg/ml PZA) on 7H11 agar plates at pH 5.8. This evidence indicates that N-terminal nonsynonymous RpsA mutations may indirectly confer PZA resistance even though the active POA binding site is located in the RpsA C terminus (28, 29).

Dillon et al. claimed that the antituberculosis activity of PZA is independent of trans-translation and RpsA (22). However, there are reasons to believe that the trans-translation assay in that study did not work properly. First, their SDS-PAGE gel showed several protein products or bands from their trans-translation assay (figure in the supplement of 22) when it should have been a single band; by contrast, in our trans-translation assay, we always get a single protein band (10) rather than the multiple bands seen in the Dillon study (22). The presence of multiple protein bands in that trans-translation system could be due to the use of impure ribosome preparations or contamination of their ribosome preparation with other bacterial ribosomes and nucleic acids such that POA could not inhibit the trans-translation due to the contaminating system. Second, another sign that their trans-translation assay in the Dillon et al. study did not work properly is the presence of a large high molecular weight blob (>95% of the radioactive signal) at the top of the gel (which was cut out from the gel whose complete picture was put in the supplement of the Dillon paper) (22). In contrast, in our trans-translation assay we do not see a huge protein blob on top of the gel (10). Thus, for the above reasons, we do not believe they have sound evidence to arrive at the conclusion that PZA/POA does not inhibit trans-translation in M. tuberculosis. In addition, Dillon et al. claimed that POA did not bind RpsA in the isothermal titration calorimetry (ITC) experiment. However, there could be many reasons why the ITC did not work, including problems with RpsA protein purity and inappropriate protein concentrations, drug concentrations, and assay conditions. In fact, it is not even clear what concentrations of the RpsA and POA were used in their ITC (22). Evidence demonstrating that POA binds to RpsA in ITC has been reproduced by at least two different groups in three separate studies (10, 28, 29). While the ITC monitoring is just one of the measures that indicates that RpsA binds to POA, the more confirmative three-dimensional (3D) crystallography and nuclear magnetic resonance (NMR) data have convincingly demonstrated that it is the C-terminal part of RpsA that indeed binds to POA (28, 29).

The conclusion that RpsA is a target of PZA (10) is supported by the association of RpsA mutations and PZA resistance in clinical strains (11–14, 19, 20) and by biochemical and structural studies showing specific binding of the drug to RpsA (28, 29, 30). Despite the fact that rpsA overexpression caused elevated PZA resistance (10), introducing a point mutation into the chromosome, though challenging, is by far the most convincing method to provide genetic evidence supporting that a specific mutation is the cause for drug resistance. Our findings that the RpsA Δ438A mutation and the D123A mutation when introduced into the M. tuberculosis chromosome are indeed responsible for PZA resistance provide further proof that RpsA is a real target of PZA. However, some questions remain to be addressed, such as the role of certain rpsA mutations in some seemingly PZA-susceptible strains. Future studies are needed to construct more isogenic strains with rpsA mutations to determine whether some rpsA mutations are involved in PZA resistance, while others may not or may confer a borderline level of resistance that is easily mischaracterized as PZA susceptible due to insensitive PZA drug susceptibility testing.