Available online at www.sciencedirect.com ScienceDirect Reactive dirty fragments: implications for tuberculosis drug discovery Pooja Gopal and Thomas Dick Reactive multi-target fragments, old synthetic Additional drugs include ethionamide (ETH) and para- antimycobacterials that are activated inside Mycobacterium aminosalicylic (PAS) [4]. New drugs with new mechanism tuberculosis bacilli and are smaller than the usual drug-like, of action are urgently needed to shorten the lengthy single-target molecules, represent critical components of treatment regimen of drug susceptible TB and to improve current tuberculosis chemotherapies. Recent studies showed the poor cure rates of drug resistant disease [5]. that para-aminosalicylic acid is recognized as a substrate by Current drug discovery strategies are largely based on dihydropteroate synthase and poisons the downstream folate Paul Ehrlich’s magic bullet ‘one drug–one target’ con- pathway. Pyrazinamide, a key relapse-reducing drug, is metabolized by an amidase and the reaction product interferes cept. This facilitates lead optimization as it allows simple with trans-translation, membrane potential and other targets. structure activity relationships and the use of target- However, the mechanism of action of pyrazinamide remains ill- compound co-structure guided design [6]. A few years defined and needs to be understood to rationally approach ago, after a decade of biochemical high throughput treatment shortening. The success of small dirty drugs and screening against genetically validated targets, the anti- prodrugs suggests that fragment-based whole cell screens mycobacterial discovery field largely moved back to phe- should be re-introduced in our current antimycobacterial drug notypic whole cell screens to identify compounds with discovery efforts. antimicrobial activity first, then deconvolute the target for lead optimization [7]. This strategic shift is due to the Addresses large scale failure of target-based approaches: translating Antibacterial Drug Discovery Laboratory, Department of Microbiology, biochemical enzyme inhibitors into whole cell active Yong Loo Lin School of Medicine, National University Health System, National University of Singapore, 5 Science Drive 2, Block MD4A, antimicrobials turned out to be far more difficult than Singapore 117597, Republic of Singapore anticipated [8]. A key issue is the mycobacterial cell envelope, representing a formidable permeability barrier Corresponding author: Dick, Thomas ([email protected]) [9]. Target based approaches also fail to capture prodrugs, which are critical components of anti-TB regimens. It is to note that both target-first and compound-first avenues Current Opinion in Microbiology 2014, 21:7–12 use the same concept: identify single-target, high affinity This review comes from a themed issue on Antimicrobials (nM) binders, which are non-reactive to avoid side Edited by James J Collins and Roy Kishony effects. Other recent approaches do not start with screen- ing of compound libraries, but make use of existing antibacterials that either do not work against M. tubercu- losis, or are not used for treatment of TB. These include http://dx.doi.org/10.1016/j.mib.2014.06.015 elegant remodelling of antibiotics, such as spectinomycin 1369-5274/# 2014 The Authors. Published by Elsevier Ltd. This is an [10], to make them stay inside the bacillus (prevent open access article under the CC BY-NC-ND license (http://creative- commons.org/licenses/by-nc-nd/3.0/). efflux), and re-purposing of old drugs, such as clofazimine [11]. High attrition rates, that is, limited success in all approaches, call for a multipronged ‘leaving no stone unturned’ strategy [12]. Can we find additional hit-finding Current TB chemotherapies and drug avenues at the bottom of the barrel, approaches that we discovery approaches have overlooked or not fully utilized so far? Mycobacterium tuberculosis remains the most deadly bac- terial pathogen globally [1]. Tuberculosis (TB) is treated Some of the key TB drugs, discovered in the middle of with a combination therapy consisting of isoniazid (INH), the past century by whole cell or animal model screening, rifampicin (RIF), pyrazinamide (PZA), and ethambutol are dirty fragments: they hit multiple targets and their for 2 months, followed by 4 months of treatment with molecular weights are in the range of 100–300 g/mol INH and RIF. This lengthy regimen has of course (Figure 1). The fragments are metabolized inside the implementation and compliance issues, and this in turn tubercle bacillus and only then, after being ‘activated’, fuels development of drug resistant disease [2,3]. Multi- exert their antimicrobial activity [13,14]. This type of drug resistant TB, defined as being resistant to INH and mechanism of action (MoA), polypharmacology, and phy- RIF, is treated with less potent and more toxic second line sicochemical properties, ‘extra’ small and reactive, is at drugs, fluoroquinolones and injectables (aminoglycosides odds with main stream antibacterial drug discovery: and capreomycin), and requires at least 2 years of therapy. attractive leads for medicinal chemistry should inhibit www.sciencedirect.com Current Opinion in Microbiology 2014, 21:7–12 8 Antimicrobials Figure 1 NH2 O NH S NH2 O OH HO N H2N N N N CH3 NH2 O INH ETH PAS PZA CH3 CH3 O O HO O F OH H3C O OH O H C OH OH CH H 3 3 N N H3C H3C O NH O H3C H3C H NH N O N O OH N CH O 3 CH3 MXF RIF Current Opinion in Microbiology Structures of TB fragment drugs. Isoniazid [INH, MW 137], ethionamide [ETH, MW 166], para-aminosalicylic acid [PAS, MW 153] and pyrazinamide [PZA, MW 123] in comparison with a more standard-sized synthetic drug, moxifloxacin [MXF, MW 402], and a huge natural product, rifampicin [RIF, MW 823]. MW: molecular weights in g/mol. a single target (to facilitate lead optimization), have a early 1900s, for some apparently forgotten papers. We decent size (to bind a target with high affinity), and should propose that including, or re-introducing, fragment-based not be reactive (to avoid side effects) [15]. whole cell screens against replicating and dormant bac- teria in our current approaches to TB drug discovery will Can we learn something from these old TB fragment deliver the next generation of TB drugs. drugs? Should we start screening again for these types of compounds to identify leads for the discovery of new TB MoA of reactive TB fragment drugs and why it drugs? Here we give an update on some recent develop- is useful to understand them ments in our understanding of the MoA of some of these Reactive multi-target fragment drugs, unusually small anti- unusual antibiotics, revealing new concepts and targets. mycobacterials that are activated inside the bacilli, Then we zoom into a critical sterilizing (relapse-prevent- represent critical components of TB chemotherapies ing) TB drug: pyrazinamide (PZA). We argue that the (Figure 1). INH is the key bactericidal (i.e. sputum MoA of this metabolized fragment drug remains an count-reducing) fragment drug in the first line regimen enigma, despite some recent progress, and that it is [16]. The compound is oxidized by the bacterial catalase- important to identify its targets if we are to eliminate peroxidase KatG [17], and its reactive metabolite forms persisters and substantially shorten TB therapy. adducts with NAD(P). The enoyl acyl carrier protein We review some key findings from the past few years. reductase InhA, required for synthesis of outer-membrane However, in the case of PZA, we need to go back to the mycolic acids, appears to be its major target [18]. Additional Current Opinion in Microbiology 2014, 21:7–12 www.sciencedirect.com Reactive fragments for TB drug discovery Gopal and Dick 9 targets include dihydrofolate reductase DHFR [19], and bacteriological models of persistence, but so far we do ketoacyl acyl carrier protein synthase KasA [20]. At least 16 not have clear, clinical answers. other INH adduct-binding mycobacterial proteins as well as several additional known and unknown INH resistant What we know is that inclusion of PZA in the current first mutations extend the target/MoA list [21,22 ]. ETH, less line regimen many decades ago resulted in shortening of well studied, is activated by the mono-oxygenase EthA [23] treatment from nine to six months. PZA is also considered and again curiously appears to inhibit InhA as its major a key sterilizing and relapse-preventing drug [32], indi- target [24]. More recently, the (first?) MoA of PAS was cating that it must target persisting mycobacterial popu- determined. The molecule is a structural analogue of the lations that are not killed by other drugs. Thus, learning substrate for dihydropteroate synthase (DHPS) in the from PZA’s mechanism of action, unique treatment short- folate pathway of mycobacteria. Therefore it was assumed ening properties, and lesion penetration characteristics, that PAS acts as a competitive inhibitor of this enzyme. seems to be the logical path to rationally discover drugs However, this is inconsistent with enzymology studies, in that shorten TB therapy. It may also help characterize key which PAS showed relatively weak inhibitory activity functions of persisting mycobacteria and refine the envi- against DHPS. Elegant metabolomics and genetic analyses ronment or assay conditions adopted in screening cam- revealed that, rather than being (only) an inhibitor of paigns. While we know that M. tuberculosis undergoes DHPS, PAS is recognized as its substrate, resulting in metabolic remodelling and physiological adaptation in poisoning of downstream enzymes of the folate pathway, response to microenvironments present in TB lesions including dihydrofolate reductase DHFR [25 ,26 ]. The [9], adequate recapitulation of these conditions for identification of the MoA of INH, ETH and PAS opens the high-throughput screening against persisters is mostly way for new drug discovery approaches employing those matter for debate. Defining conditions under which new chemically validated targets. In the case of INH, for PZA kills in vitro as it does in vivo would provide much instance InhA can be directly targeted, circumventing needed fresh approaches to TB drug discovery.