According to the most recent WHO report, 10.4 million people developed tuberculosis (TB) and 1.8 million died from this disease in 2015 (WHO, 2016), making TB the leading cause of death from an infectious disease. The threat that TB presents to global health has been significantly heightened by the evolution and spread of drug-resistant TB: in 2015, a staggering 480,000 people across the world developed multi-drug resistant (MDR)-TB, defined as TB that is resistant to isoniazid (INH) and rifampicin (RIF), with or without resistance to other first-line anti-tubercular drugs. Of these, 9.5% had extensively drug-resistant (XDR)-TB, which is resistant to INH and RIF (i.e., MDR-TB) in addition to any fluoroquinolone and at least one of the injectable second-line drugs, kanamycin, amikacin, or capreomycin. Unfortunately, this alarming situation has continued to worsen with the ongoing evolution of XDR-TB to forms of the disease that are functionally untreatable with existing antibiotics (Dheda et al., 2014).

Drug-sensitive TB is treated with a standard “short-course” regimen comprising a 2-month intensive phase of treatment with four drugs—INH, RIF, pyrazinamide (PZA), and ethambutol (EMB)—followed by four additional months of treatment with INH and RIF in a continuation phase. Under optimal conditions, this regimen is highly effective at achieving durable cure of drug-sensitive TB. However, non-adherence to this protracted therapeutic regimen is common among TB patients and may result in the emergence of drug resistance through the acquisition of chromosomal mutations in the aetiological agent, Mycobacterium tuberculosis (M. tuberculosis), leading to prolonged infectiousness and poor treatment outcomes (Dheda et al., 2014). Drug-resistant TB is far more challenging to treat, requiring the administration of combinations of second- and third-line drugs that are more toxic, more expensive, and less efficacious. As a result, this form of the disease is associated with substantial morbidity and mortality, while consuming a disproportionate share of national budgets for TB control in disease-endemic countries—thus compromising TB control programmes (Dheda et al., 2014, 2017).

The need for new TB drugs for the treatment of drug-susceptible as well as drug-resistant TB is therefore clear and urgent. After decades of neglect, a renewed interest in TB drug development in the late 1990s, which coincided with major scientific advances including the completion of the first genome sequence of M. tuberculosis (Cole et al., 1998), has resulted in a pipeline populated with new as well as repurposed drugs and drug combinations at various stages of development (http://www.newtbdrugs.org). A number of criteria are being used to guide this process: for example, all new TB drugs should: (i) have novel mechanisms of action to permit their use in the treatment of drug-resistant forms of the disease; (ii) have significant treatment-shortening potential when combined with other agents; (iii) be safe and tolerable; (iv) simplify treatment by reducing the pill burden and dosing frequency; and, (v) be compatible with antiretroviral drugs to enable treatment of patients co-infected with HIV (Zumla et al., 2013, 2014). The ability to meet these criteria is dependent upon the quality of compounds that enter the pipeline at the lead optimization stage. The identification of high-quality leads has, in turn, been critically reliant on harnessing biological insight from studies on M. tuberculosis pathogenesis in various models of infection. A major theme emerging from this work is the biological complexity of TB at the level of both host and pathogen, with the genotypic and phenotypic heterogeneity of M. tuberculosis posing particularly onerous challenges for new TB drug discovery, as discussed below.

Genome-wide mutagenesis studies in M. tuberculosis (Long et al., 2015) have identified genes that are (conditionally) essential for growth and survival of the bacillus in vitro (Sassetti et al., 2003; DeJesus et al., 2017), in macrophages (Rengarajan et al., 2005), and in animal models of infection (Sassetti and Rubin, 2003). This information has underpinned target-based drug discovery efforts aimed at crippling essential cellular functions in M. tuberculosis. However, as in other areas of antimicrobial drug discovery (Payne et al., 2007), the approach has met with very limited success in the TB field, and has been confounded by a general lack of information about target vulnerability as well as the impact of compound metabolism, permeability, and efflux on efficacy. For this reason, small molecules that potently inhibit M. tuberculosis enzymes in biochemical assays have failed to translate into leads with activity against the bacillus in vitro and/or in vivo. In contrast, phenotypic screening, in which compound libraries are screened for activity against M. tuberculosis to identify molecules with whole-cell activity, has been far more successful, and has delivered the clinically approved drugs, bedaquiline (Sirturo) and delamanid (Deltyba), a number of drug candidates that are currently in development (Mdluli et al., 2015; Singh and Mizrahi, 2017)—including griselimycin (Kling et al., 2015), PA-824 (pretomanid) (Stover et al., 2000), PBTZ169 (Makarov et al., 2014), and Q203 (Pethe et al., 2013)—and other promising leads such as the Pks13 inhibitor, TAM16 (Aggarwal et al., 2017). It is worth noting, however, that this approach, too, has its challenges as mechanisms of action (MOA) of potent molecules with whole-cell activity can be difficult to elucidate, thereby complicating the progression of individual compounds or compound series through the pipeline. Importantly, though, there are signs indicating greater integration of the two approaches: on the one hand, target-based whole-cell screening, in which hit identification from phenotypic screening is biased toward prioritized targets and pathways, has begun to gain traction (Abrahams et al., 2012) while, on the other hand, screening collections of whole-cell actives identified by phenotypic approaches against high-value M. tuberculosis targets offers the prospect of discovering new drug-target pairs as starting points for hit-to-lead (H2L) programs (Esposito et al., 2017).

Genotypic and phenotypic heterogeneity of M. tuberculosis must be taken into account from the earliest stage of TB drug discovery. Genotypic heterogeneity is managed by screening promising hits for activity against representatives from the major strain lineages of M. tuberculosis (Coscolla and Gagneux, 2014) and against panels of drug-resistant strains (e.g., Aggarwal et al., 2017; Blondiaux et al., 2017). The other major mechanism underlying differential drug susceptibility in M. tuberculosis is phenotypic antibiotic tolerance (Aldridge et al., 2014; Brauner et al., 2016), which is thought to be the main reason why prolonged TB therapy is required in order to achieve relapse-free cure (Kester and Fortune, 2014; Gold and Nathan, 2017). Antibiotic efficacy can be influenced profoundly by the physiology, metabolic state, and growth rate of the organism, with most TB drugs showing significantly reduced efficacy against M. tuberculosis in slow- or non-growing states (Baer et al., 2015). Thus, drugs that target cellular processes required to support bacterial growth tend to have reduced efficacy against slow- or non-growing organisms (Gold and Nathan, 2017). Mycobacterium tuberculosis encounters complex, hostile environments during transmission, infection, and disease (Pai et al., 2016). As an exquisitely adapted human pathogen endowed with a rich and highly flexible metabolic repertoire (Baughn and Rhee, 2014; Warner, 2014), the bacillus is able to adapt its physiology and metabolism in response to the conditions encountered during each of these stages. These conditions include intracellular residence in macrophages and other phagocytic cells, exposure to nitrosative and oxidative stress, hypoxia, nutrient deprivation, alterations in carbon source availabilities, and low pH (Baer et al., 2015). In a single patient, therefore, M. tuberculosis infection can be characterized by mixed populations of intracellular and extracellular bacilli in a variety of metabolic states and with variable growth rates. This complicates treatment (Dartois and Barry, 2013) and has led to the suggestion that TB should be treated as a polybacterial infection (Evangelopoulos and McHugh, 2015). The problem is further complicated by the impact of lesion heterogeneity on drug pharmacokinetic/pharmacodynamic (PK/PD) parameters (Dartois, 2014). To address this complexity, assays designed to recapitulate at least some of the conditions encountered during infection have been incorporated into drug screening cascades with the aim of identifying “pan-active” compounds with the ability to kill M. tuberculosis in as wide a range of metabolic states as possible.

TB drugs fall into a relatively small number of mechanistic classes. A defining characteristic of the tubercle bacillus is its unusual and highly complex cell envelope, which has a number of distinguishing features including the mycolyl-arabinogalactan-peptidoglycan complex that links the peptidoglycan to the mycobacterial outer membrane. Not surprisingly, a disproportionate number of TB drugs act on biogenesis of the cell envelope; these include INH and ethambutol (EMB), the second-line agent, D-cycloserine, and those that act on the new targets, DprE1 (e.g., PBTZ169) (Makarov et al., 2014), MmpL3 (e.g., BM212 and other chemotypes) (Xu et al., 2017), and Pks13 (TAM16) (Aggarwal et al., 2017). Other drugs target transcription (RIF), protein synthesis (e.g., linezolid), and energy metabolism (bedaquiline, Q203). Furthermore, and consistent with the formidable capacity of M. tuberculosis to metabolize xenobiotics (Awasthi and Freundlich, 2017), prodrugs are common in the TB drug arsenal and, for compounds such as PZA, delamanid, and pretomanid, the respective active metabolites have pleiotropic effects on mycobacterial metabolism (Matsumoto et al., 2006; Singh et al., 2008; Anthony et al., 2016).

An important, albeit small, category of TB drugs includes those that target DNA replication. Until recently, these have been limited exclusively to the fluoroquinolones, in particular, moxifloxacin and gatifloxacin, which inhibit DNA gyrase, and are widely used for the treatment of MDR-TB. However, another component of the DNA replication machinery has emerged as an exciting new target for TB drug development through the discovery that griselimycins target the β-clamp protein, DnaN (Kling et al., 2015). In the following sections, we consider the DNA replication and repair pathways of M. tuberculosis as potential sources of new targets for TB drug development. This terrain has been extensively reviewed recently, perhaps signaling the increasing appreciation of DNA metabolism as underrepresented among common antibiotic targets. The interested reader is encouraged to consult a number of excellent articles, both specific to M. tuberculosis (Plocinska et al., 2017) and of more general interest (Robinson et al., 2012; Sanyal and Doig, 2012; van Eijk et al., 2017).

The M. tuberculosis genome comprises approximately 3950 genes (Cole et al., 1998; Wang and Chen, 2013), of which ~10% (461 genes) are absolutely required for growth and survival of the bacillus under standard aerobic growth conditions in vitro (DeJesus et al., 2017). Among the “essential” genes, 15 encode components of the DNA replication machinery; these include the DnaA replication initiator, PriA helicase loader, DnaB helicase, DnaG primase, SSB, clamp loader subunits (τ/γ, δ, δ′), DNA polymerases I and III, DnaN β-clamp, DNA ligase I, and type I and II topoisomerases (Ditse et al., 2017). It is notable that effective inhibitory agents are available for only a small number of these essential mycobacterial proteins (Table 1), with DNA gyrase representing the only clinically validated target—of the fluoroquinolones, which are used in treatment of MDR-TB. This implies considerable scope for developing new compounds targeting the other essential DNA replication components, as has been proposed recently for M. tuberculosis as well as other bacterial pathogens (Robinson et al., 2012; Sanyal and Doig, 2012; Plocinska et al., 2017; van Eijk et al., 2017). In turn, it also suggests the possible utility in investigating the potential antimycobacterial efficacies of compounds developed for use against homologous DNA replication and repair proteins in other bacteria (Table 2).

As applies to antibiotic development in general, overcoming the natural defenses of the target organism—in particular, the permeability barrier presented by the (myco)bacterial cell wall, and the capacity for xenobiotic extrusion via multiple efflux pumps—is often a key challenge, particularly in converting hits from biochemical assays into whole-cell actives. Avoiding compound metabolism (degradation or modification) by the target bacillus or its human host can present an additional obstacle. For DNA replication and repair specifically, the non-availability to date of purified forms of many of the mycobacterial proteins and/or reconstituted complexes has further restricted the number of in vitro screens against purified proteins, and has required that researchers rely on homology models developed using template structures from other bacteria. Importantly, this can also complicate any assessment of the druggability and ligandability of the target protein—both key additional factors in determining the success of the antibiotic development process, and which render gene essentiality alone insufficient for target validation (Hopkins and Groom, 2002; Edfeldt et al., 2011). It is pleasing to note, therefore, that several recent successes in expressing different components of the mycobacterial DNA metabolic machinery (Gong et al., 2004; Rock et al., 2015; Gu et al., 2016; Banos-Mateos et al., 2017) suggest this critical roadblock will be overcome shortly.

Targeting DNA and the array of proteins which ensure its replication and maintenance within the cell presents an additional challenge, namely ensuring specificity of the applied drug for its target organism. This can be onerous given that the proteins which interact with and modify this macromolecule have retained many key features and commonalities as they have evolved in different species. For TB, which requires lengthy treatment, the need to avoid toxicity in the human host presents an additional major challenge, and one which is likely to exclude drugs which target DNA directly, such as DNA intercalating agents (Zhang et al., 2017), and inducers of replication stress in mammalian cells (i.e., anticancer compounds). Instead, antitubercular chemotherapies need to be designed to exploit specific nuances of, and vulnerabilities within, the complement of mycobacterial DNA replication and repair proteins (Mizrahi and Huberts, 1996; Rock et al., 2015).

Despite all these challenges, there have been some exciting recent discoveries—for example, the novel DnaN-targeting natural product antibiotic, griselimycin (Kling et al., 2015), and the DnaE inhibitor, nargenicin (Young et al., 2016)—which support the potential for new drug discovery in this area, and also suggest that natural product sources are likely to offer the most promising new agents (Wright, 2017). In the ensuing sections, we discuss the very limited number of validated and experimental anti-TB drugs targeting DNA replication, and provide brief updates on recent progress suggesting the potential to develop additional experimental compounds to inhibit other components of the mycobacterial replication machinery.

Although the MOA of antifolate drugs such as sulfamethoxazole, trimethoprim, and para-aminosalicylic acid includes depletion of dNTP pools, preventing DNA replication, the impact of these agents on M. tuberculosis is polypharmacologic as it also involves inhibition of RNA and protein synthesis (Minato et al., 2015). Therefore, in its strictest sense, there are no anti-TB drugs in clinical use which directly target the DNA biosynthetic machinery in mycobacteria. That said, a handful of very exciting recent studies have established the utility of a number of compounds that prevent DNA synthesis by targeting novel Pol III holoenzyme components in M. tuberculosis.

Together with nine other proteins (namely, Pol IIIα, the β₂ sliding clamp, ε proofreading subunit, τ, δ, and δ′, DnaA, DNA ligase, and Pol I), the DnaB helicase, DnaG primase, and SSB constitute the basic replication module that is found across almost all sequenced bacterial genomes (McHenry, 2011; Robinson et al., 2012). DnaB and DnaG form the helicase-primase complex which, in combination with the PriA helicase loader, functions as the mycobacterial primosome: there are no identifiable mycobacterial homologs of DnaC, DnaT, PriB, or PriC. As core proteins, these represent compelling drug targets and, while much further work is required, some recent progress (summarized below) suggests that a validated clinical candidate targeting different components of the primosome is a genuine possibility.

Within any cell, ssDNA generated during DNA replication (as well as other processes, including exposure to genotoxic stress) is vulnerable to damage and prone to form secondary structures that can restrict DNA metabolic processes with potentially lethal consequences. SSB proteins have evolved to protect ssDNA, and so are essential to bacillary viability during normal replication as well as under DNA damaging conditions. Several recent high-throughput screens have been successful in identifying small-molecule inhibitors of SSB-protein interactions (Lu et al., 2010; Marceau et al., 2013; Glanzer et al., 2016). These include an attempt to identify inhibitors of SSB that might disrupt both DNA replication and SOS-mediated resistance pathways within Gram-positive and Gram-negative bacteria (Glanzer et al., 2016): following in vitro screening, six molecules were identified which successfully inhibited a broad range of bacterial SSBs, with a further four exhibiting species-specific activity—thereby establishing the potential for both broad-spectrum and species-targeted use. Notably, five of the six compounds were found to have whole-cell activity against a variety of the tested species, of which a single compound, 9-hydroxyphenylfluoron, was associated with minimal activity against the human SSB homolog. While the potential utility of this approach remains to be determined for M. tuberculosis, these results suggest the value of investigating SSB as novel anti-mycobacterial target.

An analogous approach sought to identify compounds that specifically target the eight highly-conserved residues at the C-terminus of Klebsiella pneumonia SSB with the objective of inhibiting interactions between SSB and other proteins (Voter et al., 2017). Using the interaction between SSB and the essential protein helicase, PriA, as basis for a high-throughput screen of more than 72,000 compounds, this study aimed to identify small molecules capable of inhibiting SSB interactions. Seven SSB-PriA interaction inhibitors were found to bind to SSB, with a further two binding PriA, all with IC₅₀ values below 40 μM. No data were presented on the activity (or lack thereof) of these compounds in whole-cell assays; however, this work reinforces a common theme which suggests that protein-protein interaction inhibitors may be of specific value in inhibiting the large complex of proteins which enables DNA replication. In a similar vein, two potential PriA inhibitors, kaempferol and myricetin, were shown to inhibit the ATP hydrolysis activity of S. aureus PriA in vitro (Huang et al., 2015). While these compounds were also not validated in whole-cell assays, they too represent encouraging steps in the effort to identify antibiotics that target primosome proteins and, importantly, provide useful insight into the isolation of tractable pharmacophores for optimization against the mycobacterial homologs as part of rational structure-activity relationship (SAR) efforts.

The DnaG primase synthesizes primers for lagging strand Okazaki fragments. An early study investigating plant-derived natural products discovered two phenolic monosaccharides from Polygonum cuspidatum with low micromolar IC₅₀ values against E. coli DnaG (Hegde et al., 2004). Similarly, another molecule from Penicillium verrucosum was shown to inhibit E. coli primase activity in biochemical assays (Chu et al., 2003). However, whole-cell activity was attainable only in a mutant E. coli strain deficient in the lipopolysaccharide layer of the cell wall as well as the AcrAB efflux system, reinforcing the potential obstacles associated with permeation and efflux as part of antibiotic discovery. This is echoed in another study which identified two classes of compounds with efficacy against E. coli DnaG in vitro and in efflux pump-deficient whole-cell assays (Agarwal et al., 2007). In that case, in vitro and whole-cell activity analyses of related pyrido-thieno-pyrimidines and benzo-pyrimido-furans identified numerous hits with attractive IC₅₀ and MIC values, again suggesting the potential to identify novel primase inhibitors as possible anti-mycobacterial agents. In this context, it is worth noting the “natural validation” of DnaG as a suitable target for inhibiting replication: in many bacteria including M. tuberculosis, endogenous production of guanosine tetra- and penta-phosphate, (p)ppGpp, as part of the stringent response prevents the function of the replicative primase, curtailing bacterial growth (Maciąg et al., 2010).

The interaction of primase with the ssDNA template is facilitated by the replicative DNA helicase encoded by DnaB, another essential protein in M. tuberculosis (Sassetti et al., 2003; DeJesus et al., 2017). A number of flavonols have been shown to inhibit DnaB function in other bacteria (Griep et al., 2007; Lin and Huang, 2012), though no reports exist regarding the activity of these (or other) compounds in M. tuberculosis. Unusually, DnaB is among five mycobacterial proteins that contain inteins, two others of which are also involved in DNA replication: GyrA and RecA. This observation recently prompted the interesting proposal from Dziadek and colleagues (Plocinska et al., 2017) to block the protein splicing machinery as part of a polypharmacologic approach that would prevent activation of these intein-containing proteins, potentially disrupting multiple pathways simultaneously.

Replication of the chromosomal DNA requires controlled alterations of the DNA topology to ensure processive synthesis while limiting the stresses imposed by negative supercoiling and concatenation of the double-stranded DNA molecule. The type II topoisomerase, DNA gyrase, functions to relieve torsional strain by introducing transient double-strand DNA (dsDNA) breaks which generate negative supercoils in the bacterial chromosome. Unlike those bacteria which rely on two type II topoisomerase enzymes—DNA gyrase and TopoIV—to accomplish these tasks, M. tuberculosis employs only a GyrA₂B₂ gyrase comprising gyrA-encoded supercoiling subunits and gyrB-encoded ATPase proteins. As a drug target, DNA gyrase represents one of the most successful in antibiotic history, primarily of the fluoroquinolones which have been used to treat both Gram-negative and Gram-positive bacterial pathogens. A series of chemical scaffolds has been employed in developing successive generations of fluoroquinolones, all of which function as topoisomerase II poisons, stabilizing the cleaved DNA-topoisomerase II complex and so resulting in a large number of double-stranded DNA breaks within the replicating bacillus which are thought to overwhelm the repair machinery, triggering a cascade of events that results in bacterial death (Dwyer et al., 2015). Fluoroquinolones are currently used as second-line anti-TB agents; however, the imperative to reduce the duration of therapy has seen several large clinical trials of novel combination regimens comprising a fluoroquinolone as frontline agent (Gillespie et al., 2014; Jindani et al., 2014; Merle et al., 2014). Although unsuccessful, these trials yielded valuable lessons about the types of preclinical data which might better inform the design of new therapies (Warner and Mizrahi, 2014), as well as the potentially critical role of drug distribution and lesion penetration in ensuring efficacy (Prideaux et al., 2015).

Other classes of gyrase inhibitors include the aminocoumarins, such as novobiocin, which were the first of many natural products found to act as gyrase inhibitors (Barreiro and Ullán, 2016). Since these compounds preferentially inhibit ATPase (GyrB) function (Lewis et al., 1996), they are less vulnerable to pre-existing resistance against the fluoroquinolones, which generally maps to mutations in gyrA (Chopra et al., 2012). Moreover, in contrast to the fluoroquinolones which result in dsDNA breaks and so upregulate the mycobacterial DNA damage response (Gillespie et al., 2005), the risk of aminocoumarin-induced mutagenesis is likely to be lower, especially in M. tuberculosis in which exposure to novobiocin does not trigger expression of the SOS regulon (Boshoff et al., 2004). However, the relatively poor penetration of aminocoumarins across cell membranes, their limited solubility, and the development of the synthetic fluoroquinolones have limited the clinical utility of this compound class (Barreiro and Ullán, 2016). In addition, issues with cytotoxicity remain a major hurdle, particularly for TB which requires extended therapeutic duration. Recent progress in the development of novel bacterial topoisomerase inhibitors (NBTIs) targeting DNA gyrase (Grillot et al., 2014; Blanco et al., 2015; Jeankumar et al., 2015a,b; Locher et al., 2015) nevertheless suggests that alternatives to the fluoroquinolones might become available in the future.

In contrast to the Type II enzymes, Type I topoisomerases have been very sparsely explored for antibiotic drug discovery. These enzymes, which cause single-stranded nicks in relaxing the DNA, perform an essential function in remodeling the chromosome for various processes including DNA replication and recombination, RNA transcription, and condensation and therefore represent an attractive target (Tse-Dinh, 2016). For this reason, Sridevi et al. conducted virtual screens of two chemical libraries for the capacity to dock with M. tuberculosis TopA (Sridevi et al., 2015). Subsequent in vitro verification of the putative hit compounds identified three with activity against purified TopA: amasacrine, tryptanthrin, and hydroxycamptothecin, a derivative of the anticancer topoisomerase inhibitor camptothecin (Wall et al., 1966). The latter hit compound was subsequently modified with terminal hydrophobic moieties to yield a library of fifteen 7-ethyl-10-hydroxycomptothecin derivatives which exhibited activity against both drug-susceptible and XDR M. tuberculosis, with MICs as low as 5.92 and 2.95 μM, respectively—a significant improvement over previous TopA inhibitors (Godbole et al., 2014, 2015). Moreover, the XDR isolates exhibited enhanced susceptibility to five of the hydroxycamptothecin derivatives relative to the drug-susceptible strains, suggesting that this might offer an attractive target in these otherwise highly resistant forms. Furthermore, four hydroxycamptothecin derivatives were identified to be more effective at inhibiting the resuscitation of non-replicating persisters in both nutrient starvation as well as oxidative and nitrosative stress models. These results, together with compounds identified in other studies and which still require validation in whole-cell assays (Ravishankar et al., 2015; Sandhaus et al., 2016), highlight the possibility of successfully and specifically inhibiting TopA as a novel therapeutic target for drug-susceptible and drug-resistant TB.

The notion of developing “anti-evolution” drugs to prevent the function of mutagenic repair pathways in M. tuberculosis has been discussed previously (Warner, 2010). This strategy seems likely to be especially appropriate for M. tuberculosis as adaptive evolution of this organism depends solely on chromosomal rearrangements and point mutations, and all drug resistance arises through spontaneous mutations in target or complementary genes (Galagan, 2014). These factors suggest that inducible mutagenic mechanisms—such as the imuA'-imuB/dnaE2 mycobacterial mutasome (Warner et al., 2010)—might drive the evolution of M. tuberculosis within its host. The limited distribution of ImuA′ and ImuB among sequenced bacterial genomes therefore identifies the mutasome as a compelling target for limiting drug resistance. In some ways, this strategy is analogous to targeting virulence factors (Liu et al., 2008) and assumes that the selective pressure to mutate to antibiotic resistance is not as great where the pathway is essential for pathogenesis but not survival (Clatworthy et al., 2007). Moreover, inhibiting mutagenesis should be effective in immune compromised individuals, and might facilitate clinical trials by identifying compounds that could supplement existing regimens without compromising efficacy.

To this end, several approaches appear worth pursuing: in the first, recent evidence suggests that selective inhibition of DnaE2 by anilinouracils might be possible (Jadaun et al., 2015), provided structural data are available to enable rational identification of compounds which target this alternative α subunit, and not the replicative DnaE1, the structure of which was recently elucidated (Banos-Mateos et al., 2017). For this reason, nargenicin may not be appropriate, though dual targeting of both DnaE proteins might nevertheless represent a profitable strategy. Secondly, targeting the PHP domain exonuclease of DnaE1 provides another attractive option as inactivation of this domain was found to render M. smegmatis hypersensitive to the chain-terminating adenosine analog, ara-A (Rock et al., 2015). The recent determination of the structure of the PHP domain, which lacks a human homolog, has created an opportunity for structure-guided design of inhibitors against this exonuclease (Banos-Mateos et al., 2017). In the third approach, the identification of griselimycin supports the potential for developing novel protein-protein interaction inhibitors designed to disrupt mutasome function. Further work is underway in our laboratory to elucidate the molecular interactions which are essential to DnaE2-dependent mutagenesis, with genetic evidence indicating that preventing ImuB from functioning as “hub” protein might collapse this pathway (Warner et al., 2010). In conclusion, the possibility of targeting replication and repair mechanisms implicated in the evolution of drug resistance seems a challenge worth tackling: if successful, it is proposed that these compounds might be co-administered with other agents in novel combination therapies designed to protect existing antibiotics.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.