Previous studies suggested that pretomanid and delamanid have multiple targets: one involving cell wall synthesis in actively growing Mycobacterium tuberculosis (Mtb), and one that affects latent phase Mtb, possibly through a respiration pathway. Often, a drug target is discovered through the genomic sequencing of mutants of Mtb (or similar strains) that have gained spontaneous resistance to the drug. This method proved unsuccessful for pretomanid/delamanid though, with mutations found in enzymes that are likely to activate the drugs, rather than the actual target. Many antibiotics are taken as ‘pro-drugs’, an inactive form that is then activated by an enzyme unique to the target bacteria. This reduces the risk of toxic side-effects for the host, as the active form of the drug can be very reactive, and also enables the drug to be targeted to a specific bacterial species. The flip side of this is that the activating enzymes are often not essential, and resistance can develop if their genes are mutated.

In this research, we confirmed that pretomanid and delamanid do target cell wall synthesis in Mtb by inhibiting the reductase enzyme, DprE2. This forms a complex with DprE1 (an oxidase enzyme) to catalyse the final step in the synthesis of the lipid-linked arabinose sugar donor, which is essential to build the arabinan domains of the Mtb cell wall. DprE1 has already been extensively studied as the target of benzothiazinone drugs, which are progressing through clinical trials. And so, because DprE1 is such a good drug target and DprE2 is also essential, we had previously set out to try to find drugs that target DprE2, using a method called target over-expression. This method cleverly identifies drugs from a library that bind to a chosen protein target that is over-expressed in Mycobacterium bovis BCG. The protein acts as a molecular ‘mop’, sequestering the drug and thus increasing the concentration of drug required to ‘kill’ the bacteria (Fig. 1). Two drugs from the GSK TB library were shown to bind to DprE2, and interestingly, studies to identify spontaneous resistant mutants of these drugs showed mutations in the same activation pathway as for pretomanid and delamanid.

 Fig. 1 Target over-expression explained.

The fact that these drugs all share a common activation pathway and have similar structural features led us to speculate that the target could also be the same, and so target over-expression was used to confirm that both pretomanid and delamanid also bind to DprE2. However, while this method demonstrates a drug-protein interaction, it does not prove that these drugs inhibit the enzyme's activity. Inhibition was demonstrated using in vitro assays previously developed for both DprE1 and DprE2. These assays required the drug to be activated in order to show inhibition. There has been much speculation on the active form of these drugs, from a des-nitro form (releasing NO) to a NAD adduct of the drug (Fig. 2). This latter possibility was intriguing as DprE2 uses a reduced nicotine adenine cofactor. 

Fig. 2 Chemical structures of pretomanid and its predicted active compounds. The reactive nitro group of pretomanid is highlighted in red.

A simple experiment looking at the activation requirements of pretomanid told us a lot about the active form of the drug (Fig. 3). The presence of NAD(H) is essential during activation to produce a drug that inhibits DprE2. If NADP(H) or no cofactor was present, the resulting metabolites did not inhibit DprE2 activity. However, the presence of DprE2 during activation replaced the requisite for NAD(H). This led to two conclusions: firstly, it’s highly probable that the active form of pretomanid is an NAD adduct; secondly, the natural cofactor for DprE2 must be NADH which co-purifies with the enzyme in sufficient quantities to form the adduct with pretomanid when no NAD(H) is added. Once we were able to successfully activate pretomanid, enzyme kinetic assays then looked at the mode of inhibition. The active pretomanid competes with the NAD(P)H cofactor for the active site of DprE2, which would correlate with the active form being an NAD-pretomanid adduct.

 Fig. 3 Activation requirements for pretomanid prodrug.

It is our hope that this work will not only spark further research into  DprE2 as a drug-target, but also help our understanding of these new antibiotics, bringing with it future improvements down the drug discovery pipeline.