A combination screening to identify enhancers of para-aminosalicylic acid against Mycobacterium tuberculosis

Abstract

Para-aminosalicylic acid (PAS) is an antibiotic that was largely used for the multi-therapy of tuberculosis in the twentieth century. To try to overcome the inconvenience of its low efficacy and poor tolerance, we searched for novel chemical entities able to synergize with PAS using a combination screening against growing axenic Mycobacterium tuberculosis. The screening was performed at a sub-inhibitory concentration of PAS on a library of about 100,000 small molecules. Selected hit compounds were analyzed by dose-response and further probed with an intracellular macrophage assay. Scaffolds with potential additive effect with PAS are reported, opening interesting prospects for mechanism of action studies. We also report here evidence of a yet unknown bio-activation mechanism, involving activation of pyrido[1,2-a]pyrimidin-4-one (PP) derivatives through the Rv3087 protein.

Figure 1

Simplified representation of the folate synthesis pathway in M. tuberculosis and the proposed mechanism of action for para-aminosalicylic acid (PAS). Incorporation of PAS instead of para-aminobenzoic acid (PABA) in the pathway, through the action of FolP1 (dihydropteroate synthase, DHPS), lead to the formation of an hydroxylated 7,8-dihydropteroate (DHP) intermediate, itself converted into an anti-folate metabolite by FolC (dihydrofolate synthase, DHFS), resulting in downstream inhibition of FolA (dihydrofolate reductase, DHFR).

Figure 2

Screening optimization and validation. (a) Dose–response curve of PAS against H37Rv-GFP and selection of the highest, non-inhibitory concentration for the screening. The curve contained 20 doses, starting from 20 μM with twofold dilution between each dose and was repeated 4 times with 3 replicates each time. All values were pooled and average ± SD are shown on the graph (n = 12). The acceptable range of concentration for PAS is indicated in green, the 3 corresponding concentrations from the dose–response curve are indicated. The red, dashed line indicates the position of the 100 nM concentration. The best fit for the dose–response (using a sigmoidal dose–response model, 4 parameters) is indicated as a blue dashed line. (b) Confirmation of the effect of PAS on the growth of H37Rv-GFP, at three selected concentrations. Experiment was repeated 3 times and the average ± SD for a representative result are shown (n = 12). (c) Aggregated raw fluorescence values (RFU) for the positive (1 μg/mL rifampicin, RIF) and negative (1% DMSO) controls, for each of the two screening sets (PAS-treated, untreated). The black dashed line indicates equal RFU values for both sets (Y = X).

Figure 3

Distribution of the compound inhibition data for both screening sets (PAS-treated, untreated). Hits were selected at an inhibition threshold of 50% and grouped into 3 categories, as shown with dashed boxes on the graph: PAS-additive (top-left), PAS-independent (top-right) and PAS-antagonist (bottom-right). The number of hits in each box is indicated. The red-dashed line indicates equal percentage of inhibition for both sets (Y = X).

Figure 4

Structure of the main hit clusters identified. (a) Aminopyrazolo[1,5-a]pyrimidine (APP). (b) Adamantyl amide (AA). (c) 3,3-difluoro-2-azetidinone (DFA). (d) purine-like (PL) compounds.

Figure 5

Evaluation of DFA and PL additivity with PAS using a colony-forming unit (CFU) assay. H37Rv wild-type bacteria were incubated with (a) DFA and (b) PL compounds at various concentrations, in presence or absence of PAS (100 nM). Aliquots were withdrawn at regular interval and plated on solid agar for CFU enumeration. Data shown here are for 7 days of incubation (see Fig. S18 for additional data at day 2) and represent average ± SD for plating made in duplicate. The horizontal dotted line indicates the CFU counts obtained at day 0 (7.06 ± 0.3).