Because of the emergence of strains of Mycobacterium tuberculosis resistant to first-line antituberculosis agents, one of the second-line drugs, p-aminosalicylate (PAS), has regained importance in the treatment of tuberculosis. The mode of action of PAS, however, remains controversial as to whether it inhibits mycobactin or folate biosynthesis. To unravel this, we have studied the effect of PAS on wild-type Mycobacterium smegmatis and its mutants (gene knockouts of the salicylate pathway –trpE2, entC and entD). The wild type had no sensitivity to PAS (MIC>400 μg mL−1), whereas the mutants were hypersensitive, with 1 μg mL−1 inhibiting growth. The sulphonamides, trimethoprim and dapsone, had little effect on the growth of either the mutants or the wild type. In addition, PAS at 0.5 μg mL−1 increased the accumulation of salicylate with the wild type and mutants. These results support our hypothesis that PAS targets the conversion of salicylate to mycobactin, thus preventing iron acquisition from the host.
Because of the emergence of strains of Mycobacterium tuberculosis that are resistant to currently used drugs, both singly and as multidrug combination therapies, new chemotherapeutic targets need to be identified and thus new drugs need to be created. Iron is essential for the growth of all pathogenic bacteria (Ratledge & Dover, 2000; Schaible & Kaufmann, 2004), and its concentration and availability are critical in determining the outcome of infection particularly with pathogenic mycobacteria (Ratledge, 2004; McDermid & Prentice, 2006; Boelaert et al., 2007). As the mechanism of iron acquisition by mycobacteria is unique to these bacteria, this provides a number of possible targets for drug action that will not be found in other microorganisms or, and most importantly, in the host. Such suggestions have already been made on the basis of mutants of pathogenic mycobacteria losing their virulence in animal models when components of iron acquisition mechanism have been deleted (De Voss et al., 2000; Luo et al., 2005; Somu et al., 2006).
The central molecule that is involved in iron acquisition in almost all mycobacteria is mycobactin. This is a lipophilic, small-molecular-weight siderophore that is located in the envelope of mycobacteria in close proximity to the cytoplasmic membrane (Ratledge, 1999). Although it has a very high affinity for iron (Ks∼1036), it does not directly sequester iron from the host as it is insufficiently water soluble for this task and cannot come into direct contact with any iron-containing molecules of the host; instead, a related siderophore, carboxymycobactin, is secreted by pathogenic mycobacteria, which is then the functional extracellular siderophore. Both mycobactin and carboxymycobactin are considered to be synthesized by a common pathway, with divergence to the two siderophores occurring at one of the last stages (Ratledge, 2004).
The pathway for mycobactin/carboxymycobactin involves the initial synthesis of salicylic acid via the shikimic acid pathway; this is then linked to various amino acids or their derivatives to yield the final siderophore (Quadri et al., 1998). Deletion of any one of the three genes (trpE2, entC or entD) that are involved in the biosynthesis of salicylate from chorismic acid in Mycobacterium smegmatis results in the impairment of growth particularly under conditions when iron is at a limiting concentration (Nagachar & Ratledge, 2010). Similar results were reported when salicylate-requiring auxotrophs of M. smegmatis were generated by random mutagenesis (Ratledge & Hall, 1972; Adilakshmi et al., 2000). It is therefore our contention that the antitubercular drug p-aminosalicylate (PAS) acts as an analogue of salicylic acid and either inhibits its synthesis or, more likely, its onward conversion to mycobactin.
PAS was one of the first antituberculosis drugs (Lehmann, 1946). As its discovery pre-dated the elucidation of the structure of mycobactin (Snow, 1965), it was suggested both then and later by numerous writers (e.g. Winder, 1964) that its mode of action was that of an antifolate drug as it seemingly could be regarded as an analogue of p-aminobenzoate, the aromatic precursor of folic acid. More recent evidence suggests that the linkage of PAS to folate metabolism could be at the level of thymidylate synthase (ThyA), whose gene, when mutated, leads to PAS resistance in M. tuberculosis (Rengarajan et al., 2004; Mathys et al., 2009). Not all clinical isolates of PAS-resistant strains, however, have mutations in thyA (Rengarajan et al., 2004) or in other genes of the folate metabolic pathway (Mathys et al., 2009). Mathys et al. (2009) have therefore suggested that PAS may be a prodrug that is activated only in the presence of a functional ThyA enzyme. However, these findings do not indicate a possible site of action of PAS, only that it may need activation before it becomes inhibitory. As we have been studying the mechanism of salicylate biosynthesis in M. smegmatis (Nagachar & Ratledge, 2010), we have extended this work to investigate the effect of PAS on the various mutants in which one of the genes involved in the biosynthesis of salicylic acid has been specifically deleted. Our results show that these mutants are hypersensitive to PAS while there is no change in their responses to antifolate compounds.
Materials and methods
Bacterial strains and growth
Mycobacterium smegmatis mc2155 and its mutants were grown in a chemically defined (glycerol/asparagine) minimal medium (Ratledge & Hall, 1971). The medium (100 mL in 250-mL conical flasks with shaking at 37 °C) was supplemented with Fe2+ at 0.01 μg mL−1 (for iron-deficient growth) or at 2 μg mL−1 (for iron-sufficient growth). Antimycobacterial agents were added to the culture medium at the time of inoculation. Growth was measured as the OD600 nm after 7 days of growth and converted to the cell dry weight based on OD600 nm 1=0.83 mg mL−1.
The supplements, mycobactin and carboxymycobactin, used were extracted and purified from M. smegmatis NCIMB 8548 (Ratledge & Ewing, 1996). PAS, salicylic acid and trimethoprim were from Sigma; stock solutions were prepared in ethanol (PAS and salicylic acid) and DMSO (trimethoprim).
Creation of gene knockout mutants of M. smegmatis
trpE2, entC and entD genes in the wild-type strain M. smegmatis were partially deleted and the respective gene knockout mutants were created by homologous recombination as described previously (Nagachar & Ratledge, 2010). entDtrpE2, a double knockout, was also created where both entD and trpE2 genes were deleted together internally.
Estimation of salicylic acid
Mycobacterium smegmatis, wild type and mutants grown in minimal medium for 7 days were harvested by centrifugation at 10 000 g for 20 min at 4 °C. The pH of the supernatant was adjusted to 1.5 using concentrated H2SO4 and then extracted twice with equal volumes of ethyl acetate. The ethyl acetate extract was dried under vacuum; the residue was dissolved in 5 mL 0.1 M KH2PO4/KOH buffer, pH 7, and salicylic acid was estimated spectrofluorimetrically by its fluorescence at 410 nm following excitation at 305 nm. The extraction efficiency of PAS was only 1% when extracted for salicylate with ethyl acetate and its response in the spectrofluorimeter was 5% of that of salicylate. Hence, the readings were not affected by the presence of PAS.
Effect of PAS on the growth of wild-type M. smegmatis and its mutants under iron deficiency and sufficiency
Mycobacterium smegmatis, being a saprophytic mycobacterium, is much less sensitive to PAS than pathogenic mycobacteria. Nevertheless, it provides a useful model to study the effects of antimicrobial agents including PAS. The conversion of chorismic acid to salicylic acid in M. smegmatis involves the participation of three genes: trpE2 codes for isochorismate synthase and entC and entD code for salicylate synthase (Nagachar & Ratledge, 2010). Knockout mutants of each of these genes, as well as a double knockout, entDtrpE2, were produced and studied (Nagachar & Ratledge, 2010). As has been observed previously (Brown & Ratledge, 1975; Adilakshmi et al., 2000), PAS is less inhibitory to mycobacteria when they are grown under iron-deficient conditions and this was confirmed in this present work (Fig. 1). This, we suggest, is due to iron-deficiently grown cells being upregulated for mycobactin biosynthesis as part of the response to iron deprivation and that this includes an increase in salicylate synthesis. Therefore, if our proposal is correct that PAS is an antisalicylate compound, then, because there will be more copies of the salicylate-metabolizing enzymes present in iron-deficient cells than in iron-sufficient ones, the efficacy of PAS will be substantially decreased by iron deficiency. However, it was very surprising that the hypersensitivity of the salicylate knockout mutants to PAS was observed under all growth conditions (Fig. 1).
Complete inhibition of growth of mutants was achieved (Fig. 1b) under iron-sufficient conditions and 90–95% inhibition under iron-deficient conditions by 1 μg PAS mL−1 (Fig. 1a), whereas the growth of the wild type was only 50% inhibited with 400 μg PAS mL−1 (Fig. 1c) under iron-deficient conditions. The results given in Fig. 1 and elsewhere were taken from cells grown for 7 days, which corresponded to the maximum growth yield; growth (as the OD600 nm) was, however, monitored daily and similar patterns of inhibition were observed on each occasion, but the maximum effect was at the end of growth, which is therefore recorded here.
Accumulation of salicylate in the presence of PAS
These results, shown in Fig. 2, once more provide strong evidence that the mechanism of action of PAS is connected with salicylate metabolism probably by inhibiting its conversion to mycobactin, which is clearly indicated by the accumulation of salicylate. If PAS were to inhibit salicylate biosynthesis, then it should decrease the synthesis of salicylate, but if it blocks salicylate conversion to mycobactin, then the accumulation of salicylate should increase.
To determine whether PAS leads to an increased or a decreased production of salicylate, and thus to establish its likely site of action, the wild type and mutants were grown iron deficiently with a subinhibitory concentration of PAS (0.5 μg mL−1) and the amounts of salicylate produced were then determined spectrofluorimetrically. The results (Fig. 2) showed a clear increase in salicylate accumulation when the wild type and mutants were treated with PAS, suggesting that the action of PAS lies after the formation of salicylate and is therefore in the subsequent conversion of salicylate to mycobactin. It has to be noted that none of the mutants created were complete auxotrophs of salicylate; hence, some small amounts of salicylate were produced (Nagachar & Ratledge, 2010).
Effect of PAS in the presence of salicylate
To determine whether salicylate can relieve the inhibition of PAS, the wild type and mutants were grown with PAS from 1 to 15 μg mL−1 and with subinhibitory concentrations of salicylate, 1 μg mL−1 (Fig. 3). (It should be noted that salicylate itself inhibits the growth of M. smegmatis above 10 μg mL−1, Nagachar & Ratledge, 2010). The toxic effect of PAS was counteracted by the addition of salicylate to the medium and the growth of the mutant entC was similar to its parent strain (Fig. 3). Similar results were obtained with the other mutants, trpE2, entD and entDtrpE2. Similarly, and like salicylate, mycobactin and carboxymycobactin also successfully relieved the toxic effect of PAS and the growth of mutants was now similar to the wild type.
Sulphonamides and PAS
Sulphonamides are structural analogues of p-aminobenzoic acid (PABA) and trimethoprim is an analogue of dihydrofolic acid. However, because of the structural similarities between PAS and PABA, PAS was originally proposed as an antifolate compound (see e.g. Winder, 1964). Despite the evidence to support PAS being a salicylate analogue (e.g. Brown & Ratledge, 1975; Adilakshmi et al., 2000), assertions are periodically made to suggest that PAS may indeed be an antifolate compound and targets the folate biosynthesis pathway (Rengarajan et al., 2004).
Effect of sulphonamides, trimethoprim and dapsone on the growth of wild-type M. smegmatis and its mutants
To determine whether the knockout mutants (all with a functional thyA gene) of our study are resistant or sensitive to the antifolate compounds, the wild type and its mutants were grown iron deficiently with different concentrations of sulphonamides, including trimethoprim, ranging from 1 to 250 μg mL−1 in the minimal medium and the growth was measured after 7 days. No significant sensitivity to trimethoprim (at <10 μg mL−1) was exhibited by either wild type or the mutants. Under iron-deficient growth conditions, 80% inhibition was achieved by 50–100 μgtrimethoprim mL−1 and complete inhibition by 250 μg mL−1 (Fig. 4a). Under the same conditions, only 15% inhibition of growth was achieved by 100 μg sulphanilamide mL−1 (Fig. 4b); with sulphanilic acid, growth was inhibited only 50% with 250 μg mL−1 (data not shown). There was therefore no change in the sensitivity of the salicylate knockout mutants to trimethoprim or the sulphonamides.
Diaminodiphenylsulphone (dapsone) is an antileprosy compound and is widely used in the treatment of Mycobacterium leprae infections. In M. smegmatis and M. leprae, dapsone resistance also leads to sulphonamide resistance (Rees, 1967; Morrison, 1971). Although work on its site of action is rather sparce, evidence has been presented that it is, in fact, an antifolate compound acting as an inhibitor of dihydropteroic acid synthetase (Kulkarni & Seydel, 1983). However, dapsone, at low concentrations (<10 μg mL−1), showed no significant inhibition of the growth of wild-type M. smegmatis or the salicylate knockout mutants. Moreover, inhibition was only achieved with 250 μg dapsone mL−1 under iron-deficient conditions (Fig. 4c). These results provide strong evidence that the mechanism of action of sulphonamides and related antifolate compounds is not connected with the salicylate metabolism as there was no change in the response of the PAS-hypersensitive mutants to these compounds.
The evidence being presented in this paper is strongly supportive of our previous contention that PAS acts as an antimycobacterial agent by targeting the conversion of salicylate to mycobactin and carboxymycobactin (Ratledge & Brown, 1972; Brown & Ratledge, 1975). This is probably by the inhibition of salicylate kinase (Adilakshmi et al., 2000), which converts salicylate via salicyloyl–AMP to salicyloyl–serine as part of the mycobactin/carboxymycobactin pathway (Ratledge, 2004). If PAS acted on another pathway, for example the PABA/folate pathway, then it would be very difficult to account for why the present knockout mutants of salicylate biosynthesis are hypersensitive to PAS. There is an increase by over two orders of magnitude of the inhibitory effect of PAS in these mutants. In our view, the reason for this hypersensitivity is that salicylate synthesis is absent (or extremely low) in the knockout mutants and thus PAS can directly inhibit salicylate kinase without competition from the natural substrate, salicylate. Furthermore, the reversal of PAS inhibition in the mutants by salicylate, mycobactin and carboxymycobactin again strongly supports this hypothesis.
Despite this and our previous advancement of this hypothesis, some arguments asserting that PAS is a metabolic analogue of PABA and interferes with the synthesis of folic acid continue to be advanced. Rengarajan et al. (2004) based their proposal for PAS being an antifolate inhibitor on evidence showing that when the thymidylate synthase (thyA) gene in Mycobacterium bovis was disrupted, this led to resistance towards PAS and also to known antifolate compounds. In addition, clinical isolates of M. tuberculosis that were resistant to PAS harboured mutations in thyA, but this was only in three out of eight isolates and therefore presumably the other five did not. A more recent study of Mathys et al. (2009) found that 63% of PAS-resistant clinical isolates of M. tuberculosis had no mutations in any of the nine genes they studied including six genes of the folate metabolic pathway. They did find, though, that specific mutations in the thyA gene were associated with increased PAS resistance and this then led them to suggest that PAS may, like other antimycobacterials (e.g. isoniazid and ethionamide), be a prodrug requiring activation by a functional ThyA enzyme, and thus when ThyA is inactive, PAS will not be converted to its active form. This view would then reconcile the views of Rengarajan et al. (2004) while still being in keeping with our own observations and conclusions regarding the action of PAS as a salicylate analogue. We would like to point out, though, that, in one respect, PAS and PABA do share some common ground: both PAS and PABA are taken up into the mycobacterial cell by the same transport system, but salicylate transport is by a separate one (Brown & Ratledge, 1975). Thus, our knockout mutants would be unchanged with respect to PAS uptake. It might just be possible that PAS is both an inhibitor of mycobactin biosynthesis as well as a folate analogue (although our personal view is that this is unlikely). This would, though, distinguish PAS from those compounds that are only antifolate compounds and are completely ineffective against mycobacteria. The specificity of PAS towards mycobacteria has to rest in it being an inhibitor of some metabolic activity that is only found in the mycobacteria, and for this reason, we continue to believe that PAS is a salicylate analogue and works by inhibiting mycobactin synthesis – which, of course, is a sequence only found in the mycobacteria.
The mode of action of PAS has never been particularly clear. Because it was established as an antimycobacterial agent well before the structure of mycobactin was elucidated (see Introduction), its mode of action was asserted to be that of an antifolate agent and it was thus, like the sulphonamide drugs, an analogue of PABA. However, it was never clear why the sulphonamides were completely ineffective against mycobacterial infections and why PAS was ineffective against other bacteria and so specific for mycobacteria. (This contrary evidence was elegantly summarized by Winder 1982). Unfortunately, once the original assertion had been made that PAS was an antifolate drug, this became widely accepted and written into many standard textbooks covering the mode of action of antimicrobial agents; this view has been very hard to reverse. However, once mycobactin had been discovered and the active synthesis and accumulation of salicylic acid by mycobacteria had been established, it appeared, at least to us, that PAS was more likely to be an inhibitor of mycobactin biosynthesis (Ratledge & Brown, 1972). Our subsequent work (Brown & Ratledge, 1975; Adilakshmi et al., 2000) has provided support for this view. Of course, definitive proof of PAS being an inhibitor of mycobactin biosynthesis must await the development of appropriate assays for the individual enzymes of the pathway, but these assays may be difficult to achieve due to the complexity of the reactions and the apparent need for carrier proteins to be attached to the various intermediates (Quadri et al., 1998; Ratledge, 2004).
Our hypothesis on the mode of action of PAS is now considerably strengthened with these present results. It does occur to us, though, that as the effectiveness of PAS is considerably enhanced by preventing salicylate biosynthesis – i.e. using the salicylate knockout mutants – then its efficacy as an antituberculosis agent should be similarly increased by administering it along with an inhibitor of salicylate synthase as has been achieved recently by Payne et al. (2009) with the design and synthesis of various analogues of 2,5-dihydrochorismic acid. If such analogues could be readily produced and were sufficiently stable for clinical use, then renewed attempts to develop further derivatives of PAS would appear worthwhile (e.g. see Patole et al., 2006). The dual administration of two inhibitors, one preventing the synthesis of salicylate and the other stopping its conversion to mycobactin, could therefore be an extremely effective way of preventing the growth of mycobacteria and could therefore be useful in the treatment of tuberculosis.
We thank Overseas Research Studentships (UK) for a research studentship to N.N. We also thank Dr Andrew Boa, Department of Chemistry, University of Hull, UK, for helpful discussions.