Methionine antagonizes para-aminosalicylic acid activity via affecting folate precursor biosynthesis pathway in Mycobacterium tuberculosis

para-Aminosalicylic acid (PAS) is a second-line anti-tubercular drug that is used for the treatment of drug-resistant tuberculosis (TB). PAS efficacy in the treatment of TB is limited by its lower potency against Mycobacterium tuberculosis relative to many other drugs in the TB treatment arsenal. It is known that intrinsic metabolites, such as para-aminobenzoic acid (PABA) and methionine, antagonize PAS and structurally related anti-folate drugs. While the basis for PABA-mediated antagonism of anti-folates is understood, the mechanism for methionine-based antagonism remains undefined. In the present study, we used both targeted and untargeted approaches to identify factors associated with methionine-mediated antagonism of PAS activity. We found that synthesis of folate precursors as well as a putative amino acid transporter play crucial roles in this process. We also discovered that intracellular biotin confers intrinsic PAS resistance in a methionine-independent manner. Collectively, our results demonstrate that methionine-mediated antagonism of anti-folate drugs occurs through sustained production of folate precursors.


Introduction 27
Mycobacterium tuberculosis is responsible for approximately 10.4 million new cases of active 28 tuberculosis (TB) and 1.3 million deaths annually (WHO, 2017). While TB chemotherapeutic 29 intervention is highly successful in curing drug-susceptible TB infections, therapy is challenging, in 30 part, because it requires a minimum of 6 months of treatment with drugs associated with adverse 31 reactions. In addition, the emergence of drug-resistant strains of M. tuberculosis has dramatically 32 increased the complexity and cost of TB treatment (Gandhi et al. 2006, Gehre et al. 2016). Therefore, 33 the development of more efficacious TB chemotherapy regimens is imperative to improve treatment 34 outcomes. 35 para-Aminosalicylic acid (PAS) was the second drug to be developed exclusively for TB 36 chemotherapy (Lehmann 1946). Although PAS was a cornerstone agent of early multidrug TB 37 therapies, the introduction of more potent anti-tubercular agents into TB treatment regimens greatly 38 diminished its usage (Minato et al. 2015). Emergence of M. tuberculosis strains with resistance to 39 first-line anti-tubercular agents led to the resurgence of PAS as an second-line drug to treat infections 40 that failed to respond to standard short-course therapy (Donald and Diacon 2015). However, 41 compared to many other anti-tubercular drugs, PAS is less potent and is associated with a high rate of 42 gastrointestinal distress which limits its use to the treatment of multi-drug resistant TB for which 43 there are few other treatment options (Zumla et al. 2013). Thus, it is important to develop novel 44 strategies to enhance PAS potency, limit adverse reactions and improve treatment success rates. dihydropteroate synthase (DHPS) (Youmans et al. 1947). We previously reported that intracellular 57 PABA mediates intrinsic resistance to PAS in M. tuberculosis, and disruption of this critical node in 58 folate biosynthesis can potentiate antifolate action, including that of sulfa drugs (Thiede et al. 2016). 59 Methionine is a potent antagonist of PAS in M. tuberculosis (Hedgecock 1956), yet, the basis for this 60 antagonism remains poorly understood. Because disruption of the folate pathway in M. tuberculosis 61 results in depletion of metabolites within multiple essential folate-dependent pathways (Chakraborty,  62 Gruber, Barry inspecting growth relative to growth on the no-drug control plates after grown at 37ºC for 21 days. 131 All anti-tubercular compounds employed in this study were dissolved in DMSO. The highest 132 concentration of DMSO in the growth media was 2.5%. 133

Analysis of Growth Kinetics 134
Logarithmically growing Mycobacterium strains were washed twice in an equal volume of fresh 135 medium. Cells were diluted to an OD 600 of 0.01 in 30-ml square bottles (Nalgene) and supplements 136 with or without drug were added at the described concentrations. Cultures were shaken (100 rpm) 137 and OD 600 were measured at various time points over a 14-day time-course. 138

Methionine Utilization Experiments 139
M. bovis BCG strains were grown to mid-log phase in 7H9 broth and washed twice with sulfate-free 140 Sautons medium. Resuspended cells were diluted to an OD 600 of 0.01 in sulfate-free Sautons 141 medium. Cultures were then incubated for 5 days to exhaust remaining sulfur. Exhausted cells were 142 aliquoted into 30-ml square bottles (Nalgene) and sulfur-containing metabolites were added at the 143 given concentrations. Cultures were incubated at 37ºC and shaken (100 rpm). The fold-change in 144 OD 600 (as a ratio of the final OD 600 /initial OD 600 ) was assessed following 1 week of incubation after 145 the addition of metabolites. 146

Identification of M. bovis BCG genes involved in methionine-mediated antagonism of PAS 148
A library of M. bovis BCG transposon insertion mutant strains was constructed using the phAE180 149 mycobacteriophage containing a mariner-family transposable element. To identify genes associated 150 with methionine-mediated PAS antagonism, approximately 10,000 BCG::himar1 mutants were 151 screened following the approach outlined in Figure 2. Determination of the PAS MIC on 7H10 agar 152 plates confirmed that 0.25 µg/ml was sufficient to fully inhibit growth of the M. bovis BCG parental 153 strain. Screening was then undertaken on 7H10 agar plates containing 10 µg/ml methionine and 5 154 µg/ml PAS (Met-PAS plate). Growth of M. bovis BCG on Met-PAS plates was identical to growth 155 seen on control 7H10 plates, which confirmed methionine-mediated PAS antagonism. 156 BCG::himar1 insertion mutants which exhibited observable growth inhibition on the Met-PAS plates 157 in comparison to the growth on 7H10 agar plates containing 10 µg/ml methionine (Met plate) were 158 isolated. We then identified the himar1 insertion sites within the 35 BCG::himar1 mutants that had 159 reproducible growth defects on Met-PAS plates compared to the growth on Met plates (Figure 2 and 160 Table 2). Among these mutants, one strain with a himar1 insertion located within BCG_3282c, 161 encoding a putative amino acid/polyamine/organocation (APC) superfamily transporter (Elbourne et  162 al. 2017, Jack et al. 2000), showed the most severe growth defect on Met-PAS plates suggesting 163 BCG_3282c plays a major role in methionine-mediated antagonism of PAS. We also assessed the 164 susceptibility of each mutant strain to PAS by measuring PAS MICs on 7H10 agar plates (Table 2). 165 We observed the BCG_3282c mutant possessed wild-type PAS susceptibility suggesting this 166 mutation is associated exclusively with methionine-mediated PAS antagonism. 167 Although most mutant strains that were analyzed showed a similar level of PAS tolerance as the 168 parent M. bovis BCG, four mutants (with transposon insertions in bioB, ftsH, metB, and BCG_1906c) 169 were found to be more susceptible to PAS in the absence of methionine, indicating that the disrupted 170 genes may be involved in intrinsic resistance to PAS (Table 2). 171

BCG_3282c is essential for methionine-mediated antagonism of PAS in M. bovis BCG. 172
Based upon the observation that methionine failed to antagonize PAS activity in the BCG_3282c 173 mutant, we further characterized the function of this gene. bioactivation (Figure 1). Thus, once PAS is activated to pterin-PAS, N-methylation should not affect 215 its anti-tubercular activity. We confirmed pterin-PAS was active against wild-type M. bovis BCG at a 216 comparable molar concentration to PAS (Table 3). Surprisingly, pterin-PAS was still potently 217 antagonized by methionine suggesting methionine-mediated PAS antagonism does not occur by 218 methylation of PAS to inhibit PAS bioactivation. 219 It is also known that intracellular PABA levels affect PAS susceptibility in M. tuberculosis (Thiede,220 Kordus, Turman, Buonomo, Aldrich, Minato and Baughn 2016). Since PABA biosynthesis is 221 essential for Mycobacterium survival in vitro, we hypothesized that methionine may affect PAS 222 activity. PabB, aminodeoxychorismate synthase, is one of the essential enzymes required to convert 223 chorismate to PABA in M. tuberculosis ( Figure 4A). Consequently, a M. tuberculosis H37Ra pabB 224 deletion strain is a PABA auxotroph and relies upon exogenous sources of PABA for growth ( Figure  225 4B). The folate precursor dihydropteroate is produced from PABA and DHPPP ( Figure 4A). We 226 found that pteroic acid, an oxidized form of dihydropteroate can also support the growth of the M. 227 tuberculosis H37Ra pabB deletion strain ( Figure 4B) (Table 3). Taken together, these data demonstrated that a functional PABA 235 biosynthetic pathway is essential for methionine to antagonize PAS in M. tuberculosis. 236

Biotin cofactor biosynthesis is essential for intrinsic resistance to PAS and other anti-237 tubercular drugs 238
Our screening also identified several mutations that conferred increased susceptibility to PAS even in 239 the absence of methionine. One strain, harboring a himar1 insertion within bioB, encoding biotin 240 synthase, showed increased susceptibility to PAS both in the presence and absence of methionine 241 (Table 2). BioB is a radical SAM-dependent enzyme required for the final step in the synthesis of 242 biotin. We confirmed the bioB::himar1 strain exhibited biotin auxotrophy, which could be chemically 243 complemented by a minimum of 0.05 µg/ml biotin supplementation for restoration of growth ( Figure  244 5A). We speculated that susceptibility of the bioB::himar1 strain to PAS was dependent upon 245 intracellular concentrations of biotin. Thus, we examined the PAS susceptibility of the bioB::himar1 246 strain using media containing minimal (0.05 µg/ml) or excess (5 µg/ml) concentrations of biotin 247 ( Figure 5B). We observed the bioB::himar1 strain was far more susceptible to PAS (8-fold decrease 248 in MIC 90 ) in minimal biotin medium, and that excess biotin medium was sufficient to restore 249 susceptibility back to near wild-type levels. Interestingly, the bioB::himar1 strain was also more 250 susceptible to sulfamethoxazole (SMX) and rifampicin (RIF), but maintained wild-type susceptibility 251 to isoniazid, indicating that alterations in susceptibility profiles are drug-specific ( Figure 5B). 252

Discussion 253
Methionine is the only folate-dependent metabolite known to antagonize certain anti-folate drugs in 254 M. tuberculosis and other bacterial species. Interestingly, anti-folate drugs antagonized by 255 methionine are also antagonized by PABA, a folate precursor. Although the molecular mechanism of 256 PABA-mediated anti-folate antagonism is well understood, how methionine antagonizes anti-folate 257 drugs has yet to be elucidated. Our findings revealed that methionine-mediated PAS antagonism is 258 linked to synthesis of folate precursors. 259 One strain isolated from our screen harboring a himar1 disruption within the predicted amino acid 260 permease BCG_3282c fully sensitized M. bovis BCG to PAS in the presence of normally antagonistic 261 concentrations of methionine. In addition, the himar1 disruption within BCG_3282c prevented M. 262 bovis BCG from assimilating sulfur derived from methionine. BCG_3282c belongs to the APC 263 superfamily of transporters and our data suggested that BCG_3282c is likely responsible for uptake 264 of methionine in vitro. The most well-studied methionine transport system in bacteria is the MetD 265 ABC PAS antagonism, such that supplementation with exogenous biotin was required to observe 317 antagonism by methionine (Hedgecock 1956). However, our study found that biotin supplementation 318 was non-essential for methionine to antagonize PAS in M. bovis BCG suggesting that the effect of 319 biotin on PAS susceptibility is independent of the precise mechanism of antagonism, and the initial 320 observations in M. tuberculosis by Hedgecock may be a strain specific phenotype. 321 In summary, the mechanistic basis of methionine-mediated PAS antagonism was examined. Over 30 322 novel modulators of PAS susceptibility were identified by Himar1 transposon mutagenesis. 323 However, with exception of the putative amino acid permease BCG_3282c, none of the functions 324 identified were found to be directly involved in antagonism. Upon closer examination, de novo 325 biosynthesis of PABA was determined as essential for methionine-mediated antagonism, revealing a 326 previously unappreciated relationship between methionine and folate precursor synthesis. Further 327 studies are needed to reveal the precise mechanism of this process.

Conflict of Interest Statement 339
The authors declare that the research was conducted in the absence of any commercial or financial 340 relationships that could be construed as a potential conflict of interest. 341     plates were subjected to secondary screening. These 35 mutants that passed the secondary screening 533 were collected and insertion site locations were determined. 534 strains were grown to an OD 600 of approximately 0.5, washed three times to remove residual sulfate 539 with sulfate-free Sautons medium, and resuspended in sulfate-free Sautons medium to a starting 540 OD 600 of 0.01 and cells were starved for sulfur for 5 days. Following the exhaust period, sulfur-541 sources were added, and cells were incubated for 7 days to resume growth. The fold-change in 542 growth was assessed as a ratio of the final OD 600 over the starting OD 600 following the exhaust period 543 (final OD 600 /starting OD 600 ). p-values of pairwise comparisons (denoted by lines) were calculated 544 using the Student t test. *, p < 0.05, **, p.<.0.005, ns indicates no significant difference (p > 0.05).