Role of Whole-Genome Sequencing in Characterizing the Mechanism of Action of para-Aminosalicylic Acid and Its Resistance

para-Aminosalicylic acid (PAS) remains one of the drugs of last resort for the treatment of tuberculosis, but its mechanism of action is still not completely understood. The main aim of this project was to identify new potential mechanisms of action and resistance to PAS by performing whole-genome sequencing on PAS-resistant laboratory mutants. A new variant in the folC gene was identified, as well as some other mutations that require further study.

ABSTRACT

para-Aminosalicylic acid (PAS) remains one of the drugs of last resort for the treatment of tuberculosis, but its mechanism of action is still not completely understood. The main aim of this project was to identify new potential mechanisms of action and resistance to PAS by performing whole-genome sequencing on PAS-resistant laboratory mutants. A new variant in the folC gene was identified, as well as some other mutations that require further study.

INTRODUCTION

para-Aminosalicylic acid (PAS), also known as 4-aminosalicylic acid, was one of the first chemotherapeutic agents to be used against tuberculosis (TB), and it is currently an orphan drug available to treat only extensively drug-resistant disease (1). Despite being used for decades, its mechanism of action is not completely understood. It has been proposed that, being an analogue of para-aminobenzoic acid (PABA), PAS competes with PABA for dihydropteroate synthase, interfering in the process of folate synthesis (2). A study using transposon mutagenesis identified mutations in the thyA gene that were also present in clinical isolates resistant to PAS (3). The gene thyA encodes a thymidylate synthase enzyme (essential for DNA replication and repair), and its deletion has been demonstrated to confer resistance to PAS (4). Other studies have identified various missense mutations in folC (encoding a dihydrofolate synthase) and ribD (alternative dihydrofolate reductase) that conferred resistance to PAS in laboratory and clinical isolates of Mycobacterium tuberculosis (5–7). Nevertheless, mutations in folC were detected in only 34.8% of resistant clinical isolates, whereas mutations of thyA and ribD were detected in 26.0% and 5.8%, respectively (6). Hence, other mechanisms of resistance to the drug might exist. Efflux pumps have been described conferring cross-resistance to PAS and other chemotherapeutic agents, including streptomycin (8). The main aim of this work was to investigate potential new mechanisms of action and resistance to PAS by performing whole-genome sequencing (WGS) on PAS-resistant laboratory mutants.

PAS-resistant laboratory mutants of reference strain H37Rv were spontaneously selected by growth on 7H10 medium with the addition of PAS. M. tuberculosis isolates were cultured in 20-ml aliquots of sterile 7H9 broth and incubated for 14 days, achieving log phase and a colony count of 0.5 to 1 McFarland (150 to 300 × 10⁶/ml). After the incubation period, the bacteria were concentrated by centrifugation at 10,000 × g for 10 minutes, and the entire sediment was inoculated onto preprepared 7H10 plates containing different concentrations, i.e., 2 and 4 μg/ml, of PAS (the critical concentration for PAS is 2 μg/ml) (9). After incubation at 37°C for at least 14 days, spontaneous mutants grew on the plates and were then selected for sequencing. All selected mutants and the parent reference strain H37Rv were subcultured on Lowenstein-Jensen slopes, DNA was extracted using the CTAB (cetyltrimethylammonium bromide) method, and WGS analysis was performed as previously described (10).

It was possible to grow only one PAS mutant for each critical concentration (2 and 4 μg/ml) from a culture containing 150 to 300 × 10⁶ bacteria/ml. Because the experiment was repeated on two separate occasions, we were able to select a total of four mutants. These resistant mutants were designated PAS2 (1st and 2nd positions) and PAS4 (1st and 2nd positions).

Compared to the sequenced reference strain H37Rv, a total of seven nonsynonymous single-nucleotide polymorphisms (SNPs) affecting four different genes were identified in the four PAS-resistant mutants. Both PAS4 mutants showed variants in the folC gene, a known mutation in position 2747141 and a new mutation in position 2747195 (Table 1). There was evidence of a V58I variant in the Rv3218 gene in both of the PAS2 and one of the PAS4 mutants, but reads matching both the reference and variant base were found at this site, and this call is uncertain. At low coverage, a total of 27 SNPs were also found in the rrs and rrl genes (16s and 23s RNA genes) of all four mutants. However, these SNPs remain unconfirmed because of the low coverage.

TABLE 1.

Genes involved in the PAS-resistant mutants, with the respective function and SNP position in the genome

Genea	Function	SNPb position for:
		PAS2		PAS4
		1st	2nd	1st	2nd
Rv1392 (metK)	S-adenosylmethionine synthetase			1566981
Rv2447c (folC)	Folylpolyglutamate synthase			2747195 (D135G)	2747141 (E153A)
Rv3218	Hypothetical protein	3594639c (V58I)	3594639c (V58I)	3594639 (V58I)
Rv3759c (proX)-Rv3760 intergenic	Possible osmoprotectant binding lipoprotein; conserved membrane protein				4205442

^aAt low coverage, a total of 27 SNPs were found in the rrs and rrl genes (16s and 23s RNA genes) of all four mutants. Data not shown because of low coverage.

^bOnly high-quality nonsynonymous and intergenic SNPs were considered. Hypothetical amino acid changes caused by SNP are shown in parentheses.

^cMixed base calls and therefore lower-quality evidence in these mutants for these sites.

WGS has been used to determine the mechanism of action of antituberculous agents. In the case of bedaquiline (BDQ), researchers selected and sequenced BDQ-resistant Mycobacterium smegmatis strains and identified mutations in the proton pump of ATP synthase associated with resistance (11). In the case of PAS-resistant mutants, WGS analysis indicated that it is necessary to revisit the folate metabolic pathway to fully understand our data. The folate biosynthetic pathway starts when the aromatic precursor chorismate is converted to PABA and coupled with pteridine to generate dihydropteroate. The protein encoded by folC, dihydrofolate synthetase, adds glutamate to the dihydropteroate forming dihydrofolate (12). Mutations in the folC gene have been found to be associated with PAS resistance but in <35% of the cases, but the same researchers did not find any mutation in the folP1 gene (4). Here, both PAS4-resistant mutants gained an SNP affecting the folC gene (Table 1). Mutation E153A has been reported to confer resistance to PAS (12), and it is already included in some online databases for WGS analysis. The new mutation, D135G, was not previously associated with PAS resistance. However, it has been shown to be important for the linkage of α-helices in the folC protein structure (5), and it may represent an additional mechanism of resistance to PAS. It is interesting to note that such variants in the folC gene did not develop in the PAS2 mutants at lower concentrations, raising the hypothesis of an association with high-level resistance as noted in other antituberculous drugs (13). This may have some clinical relevance in that higher levels of resistance may be managed and potentially prevented by optimizing the dosage as previously described by other authors (14). It is also worth highlighting the fascinating (albeit unconfirmed) finding of rrs and rrl gene mutations. These genes are linked to the ribosomes, including the conversion from tRNAMet to tRNAfMet (Fig. 1), and hence the synthesis of proteins in general. This would hypothetically resemble the mechanism of resistance of streptomycin, even if streptomycin resistance is more frequently associated with mutations in the rpsL gene rather than rrs/rrl (15). Because PAS and streptomycin have been companion drugs for decades, and at a time when next-generation sequencing technologies were not available, this mechanism of action/resistance may have been missed and is certainly worth further research. Our analysis is based on only four resistant mutants because it was difficult to select PAS-resistant mutants in the laboratory. This may be because of the essentiality of genes involved and the negative impact that a deletion and/or an altered gene function would have on cell survival (16). In addition, the presence of mutations in the genome needs additional confirmation that such mutations encode significant metabolic changes.

FIG 1.

Folate metabolism in M. tuberculosis. DHPP, 7,8-dihydropterin pyrophosphate; PABA, para-aminobenzoic acid; H₂Pte, dihydropteroate; H₂PteGlu, dihydrofolate; H₄PteGlu, tetrahydrofolate; C1-H₄PteGlu, various single-carbon-modified species of H₄PteGlu; ThyX, thymidylate synthase; tRNA^Met, methionyl-tRNA; tRNA^fMet, N-formylmethionyl-tRNA; Fmt, formyl-methionine transferase.

After nearly 70 years of clinical use of PAS, WGS analysis may help in elucidating its mechanism of action, but further studies are still needed.

Data availability.

The sequence data generated have been deposited in the European Nucleotide Archive database hosted by The European Bioinformatics Institute under BioProject accession no. PRJEB36463 (SRA accession no. ERP119659).