Bacteria have evolved numerous strategies to evade killing by antibiotics. Reduced antibiotic susceptibility may result from genetic or phenotypic alterations and can be classified as resistance, persistence, or tolerance. Although each is associated with reduced antibiotic activity, it is important to distinguish among these three separate phenomena. A clear understanding of the diverse molecular pathways through which bacteria can become unresponsive to antibiotics is critical for the development of novel and more effective therapeutic approaches.

Antibiotic resistance is the most well-understood mechanism of bacterial insensitivity. The term applies to an entire bacterial population which exhibits an increase in the minimum inhibitory concentration of a particular antibiotic under optimal bacterial growth conditions. Antibiotic resistance may arise from mutations in genes encoding antibiotic targets or antibiotic-activating enzymes, mutations leading to increased enzymatic degradation of or reduced cell permeability to the antibiotic, or from increased activity of antibiotic efflux pumps (Hoffman, 2001; Karakousis, 2009; Westblade et al., 2020).

Antibiotic tolerance and persistence are both used to describe reversible states of broad insusceptibility to antibiotics. Although their usage in the literature remains ambiguous at times, antibiotic tolerance generally is used to describe an entire population of bacteria, while persistence refers to a subpopulation of phenotypically distinct bacteria within a clonal population. Both phenomena are frequently associated with a slowly growing or dormant state (Balaban et al., 2013; Maisonneuve and Gerdes, 2014).

Persistence is a non-heritable state in which a small proportion of bacteria in a clonal population can survive transient antibiotic exposure. This is characterized by a biphasic killing curve in which most of the bacterial population is killed rapidly, leaving behind a small number of persister bacteria which are insensitive to antibiotics and are killed at a much slower rate (Maisonneuve and Gerdes, 2014; Brauner et al., 2016). Persister cells may arise in the absence of any specific stimuli, perhaps due to stochastic differences in expression of specific genes, or may result from exposure to stress conditions. Persister cells are genetically identical to the rest of the bacterial population, and the antibiotic susceptibility patterns of daughter cells mirror those of the original parental population. Persistence has been attributed to reduced growth rate, metabolism, and protein synthesis as well as to the activity of toxin–antitoxin systems (Dahl et al., 2003; Sala et al., 2014; Harms et al., 2016; Talwar et al., 2020).

A wide array of stresses can induce antibiotic tolerance, including hypoxia, nitric oxide, and nutrient starvation (Gibson et al., 2018). Antibiotic tolerance is mediated by a variety of mechanisms, including the stringent response, discussed in more detail below, and drug efflux pumps (Goossens et al., 2021). Antibiotic tolerance may have a phenotypic or genetic basis and is often associated with an altered transcriptomic and metabolic state (Dahl et al., 2003; Dutta et al., 2019; Goossens et al., 2021). Additionally, actively dividing M. tuberculosis is able to develop antibiotic tolerance within macrophages through the activity of its efflux pumps (Adams et al., 2011, 2014; Szumowski et al., 2013; Black et al., 2014; Jang et al., 2017). These pumps can export antibiotics, reducing their intracellular concentrations and antibacterial effects, and pharmacological inhibition of the efflux pumps prevents the development of this macrophage-induced tolerance (Adams et al., 2011, 2014; Jang et al., 2017).

Although antibiotic resistance, tolerance, and persistence may reflect distinct phenomena, they may also be interrelated. Previous work has shown that bacterial mutations accumulate more rapidly under persistence- and tolerance-inducing stress conditions (Balaban et al., 2013). Such mutations can confer resistance to antibiotics, thus promoting the survival of tolerant bacteria (Levin-Reisman et al., 2017).

The stringent response is a conserved bacterial adaptation to nutrient starvation and other stress conditions. It is mediated by tetra- or penta-phosphorylated guanosine [(p)ppGpp] and inorganic polyphosphate [poly(P)]. Accumulation of these two regulatory molecules in E. coli leads to downregulation of growth-related pathways through altered transcription and metabolism (Traxler et al., 2008). In M. tuberculosis, the stringent response enables bacteria to survive host defenses by inducing metabolic quiescence, thereby contributing to long-term mycobacterial survival and virulence (Primm et al., 2000). By reducing mycobacterial growth and metabolism, the stringent response also contributes to antibiotic tolerance by suppressing the activity of numerous antibiotic targets.

Disruption of the stringent response by perturbing poly(P) homeostasis has been proposed as a novel TB treatment strategy. To this end, inhibitors of PPK1 and PPK2 have been developed. A small in silico screen of 18 predicted inhibitors was performed for PPK1. The three lead compounds from this screen showed robust activity with inhibition constants ranging from 255 to 866nM (Shahbaaz et al., 2019). A high-throughput screen of 2,300 compounds for PPK2 inhibitors identified two lead compounds with >80% inhibition of enzyme activity at 100μM (M. Singh et al., 2016). A DNA aptamer with an IC₅₀ of 40nM in a biochemical assay has also been designed for PPK2 (Shum et al., 2011). The finding that a ppk2 transposon mutant displayed a 4-fold increase in the MIC of isoniazid relative to the wild-type and complemented strains is concerning, however, as this effect could counteract any potential benefits of incorporating a PPK2 inhibitor into current TB treatment regimens (Chuang et al., 2013; Singh et al., 2016).

Since regulation of (p)ppGpp levels is critical for the activation of the stringent response and for bacterial adaption to stress conditions, inhibition of the bifunctional Rel_Mtb protein has also been proposed as a treatment strategy (Wexselblatt et al., 2012; Dutta et al., 2019).

Multiple groups have sought to identify RelA inhibitors. The earliest of these inhibitors, Relacin, was identified using modeling based on the crystal structure of Rel/Spo from Streptococcus equisimilis (Wexselblatt et al., 2012). This small molecule was able to decrease (p)ppGpp levels and disrupt biofilm formation in Bacillus subtilis. It also inhibited sporulation in a dose-dependent manner (Wexselblatt et al., 2012). Deoxyguanosine-based analogs were subsequently designed to optimize the compound, and one of these had increased potency against RelA from E. coli (Wexselblatt et al., 2013). Two additional analogs (a benzoylated and an acetylated benzoylated version) were also synthesized and tested for their ability to decrease levels of (p)ppGpp in M. smegmatis (Syal et al., 2017b). The acetylated benzoylated analog had an IC₅₀ of ~40μM and caused elongation of the cells (Syal et al., 2017b), which is in line with previous morphological characterization of mycobacteria lacking rel (Gupta et al., 2016). This compound also decreased long-term survival of M. smegmatis and inhibited biofilm formation in both M. smegmatis and M. tuberculosis (Syal et al., 2017b). The activity of Relacin and its analogs against multiple bacterial species suggest that these Rel inhibitors could have broad applications as antibiotics.

A high-throughput strategy using a truncated recombinant version of Rel_Mtb and a fluorescence polarization assay to screen a library of over 2million compounds for enzymatic inhibition of (p)ppGpp synthesis activity yielded 178 hits with favorable physicochemical properties (Dutta et al., 2019). In whole-cell assays, one of these hits (X9) dosed at 2μM induced significant killing of nutrient-starved, wild-type M. tuberculosis, phenocopying the survival defect of the untreated Δrel_Mtb mutant during nutrient starvation, and hypersensitized the nutrient-starved wild type to cell wall-active agents, decreasing the MBC of isoniazid 16-fold. These findings suggest that pharmacological inhibition of the M. tuberculosis stringent response results in direct mycobacterial killing under growth-limiting conditions, as well as reversal of tolerance to cell wall-active agents.

Several natural products have also demonstrated anti-Rel activity. Vitamin C inhibits the (p)ppGpp synthesis activity of Rel and decreases cell viability and biofilm formation in M. smegmatis (Syal et al., 2017a). The compound, 4-(4,7-dimethyl-1,2,3,4-tetrahydronaphthalene-1-yl)pentanoic acid (DMNP), a derivative of erogorgiaene, was identified as an inhibitor of both Rel_Msm and RelZ in M. smegmatis (Tkachenko et al., 2021). DMNP showed antibacterial activity against M. smegmatis in a whole cell assay, and overexpression of either Rel_Msm or RelZ reduced the compound’s antibiotic effect and increased the number of persisters. DMNP further inhibited the (p)ppGpp synthase activity of purified Rel_Msm and showed similar predicted binding sites for Rel_Msm or RelZ in molecular docking simulations (Tkachenko et al., 2021).

One FDA-approved antitubercular drug appears to target the M. tuberculosis stringent response, among other putative mechanisms of action. Pyrazinamide is a critical component of current treatment regimens for drug-susceptible TB and is formulated as a prodrug which is converted by M. tuberculosis pyrazinamidase to pyrazinoic acid. Pyrazinoic acid, but not the prodrug, was found to inhibit Rv2783c, which is involved in the RNA degradosome, a multi-enzyme complex important in RNA processing and mRNA degradation, with a K_d1 of 1.05mM and a K_d2 of 3.17mM (Njire et al., 2017). This binding was completely abolished in an D67N mutant and in enzyme isolated from naturally pyrazinamide-resistant M. smegmatis. In addition to its role in mRNA processing and degradation, Rv2783c has multifunctional activities, including polymerization and phosphorolysis of single-stranded DNA and (p)ppGpp hydrolysis in vitro. Each of these activities can be inhibited by pyrazinoic acid. The point mutation which eliminated pyrazinoic acid binding to Rv2783c resulted in a 5-fold increase in the MIC of pyrazinamide against M. tuberculosis, further underscoring the importance of Rv2783c as a target for this antitubercular drug (Njire et al., 2017).

In addition to pharmacological modulation of the stringent response, recent work has focused on potentiating host immunity to M. tuberculosis stringent response factors as a treatment-shortening strategy since this pathway is important for long-term mycobacterial survival in host tissues, and is induced during chronic TB infection in the lungs of mice (Talaat et al., 2004).

Intramuscular immunization of mice with a DNA vaccine targeting four different M. tuberculosis stringent response genes (rel_Mtb, sigE, ppk2, and ppx) induced significant antigen-specific IgG responses, antigen-specific IFN-γ-producing CD4+ T cells, and rel_Mtb-specific TNF-α-producing CD4+ T cells (Chuang et al., 2016). Although DNA vaccine containing all four stringent response genes was not protective against aerosol challenge with virulent M. tuberculosis, it enhanced the activity of isoniazid in a mouse model of chronic infection, significantly reducing the M. tuberculosis burden in the lungs relative to mice receiving the control DNA vaccine (Chuang et al., 2016).

Of the four stringent response genes included in the above vaccine, isoniazid exposure of M. tuberculosis-infected macrophages increased expression of rel_Mtb most robustly (Chuang et al., 2020), prompting generation of a DNA vaccine targeting this gene alone. The rel_Mtb DNA vaccine conferred increased susceptibility of M. tuberculosis to isoniazid relative to the control DNA vaccine in a mouse model of chronic TB infection and reduced regrowth of bacteria after cessation of antibiotic treatment (Chuang et al., 2020). Thus, Rel_Mtb appears to be an important persistence antigen, which can be targeted by cellular immune responses in conjunction with antitubercular drugs to assist with clearance of M. tuberculosis infection from host tissues.