Antibiotic tolerance & resilience in M. tuberculosis: Potential to predict & pre-empt antibiotic resistance

The sharp rise in drug-resistant tuberculosis (TB) over recent decades has posed a serious threat to global TB control efforts. Compounding this challenge is the limited progress in discovering new antitubercular drugs. With only a handful of promising treatments currently in development, humanity finds itself once again facing an old and formidable foe—TB that resists cure. In the absence of more effective antimicrobials, it is imperative to pursue innovative approaches to diagnose, anticipate, and prevent antibiotic resistance in Mycobacterium tuberculosis (MTB).

Standard procedures of laboratory diagnosis of drug resistance in Mycobacterium tuberculosis are based on critical concentrations of antitubercular drugs. By convention, critical concentrations of anti-tubercular drugs are defined as the minimum concentration required to inhibit 99 per cent of the growth of ‘wild-type’ strains of MTB¹. This concept laid the foundation for determining the minimum inhibitory concentration (MIC) of an antibiotic, which represents the lowest concentration needed to block the growth of a specific MTB strain. Consequently, this led to a simplified binary definition of antibiotic resistance, forming the technical framework for the proportion method used in drug sensitivity testing for MTB. Nowadays, genetic proxies of the proportion method are widely employed in drug sensitivity testing. The design of drug regimens for treatment is based on the results derived from these binary measures of antibiotic resistance.

Currently used molecular tests identify specific mutations in genes associated with drug targets or activators. These include single nucleotide variations in genes such as rpoB, inhA, katG, embB, pncA, rpsL, rrs, gyrA, and gyrB, which confer resistance to first- and second-line anti-tubercular drugs. Near point-of-care tests like GeneXpert MTB/RIF, Xpert MTB/XDR, TruNAT, as well as MTBDRplus and MTBDRsl, detect these mutations to identify drug-resistant MTB strains. Recent research reveals that mycobacterial populations may adopt alternative survival mechanisms beyond traditional resistance pathways. Consequently, standard drug sensitivity testing methods may fail to reliably predict treatment outcomes². Emerging evidence suggests that many MTB strains, classified as drug-sensitive by conventional tests, evade antibiotic action by forming privileged subpopulations of drug-tolerant bacteria that demonstrate resilience to antibiotics³. Unexpected treatment failures suggest the presence of an alternative state of drug sensitivity that falls short of being classified as resistance. This has led to the identification of numerous ‘stepping stone mutations,’ which contribute to functional diversity, fostering drug tolerance and antibiotic resilience.

Examples of such mutations include those found in the essential transcriptional regulator gene Rv1830, which enable MTB strains to recover more quickly after antibiotic treatment. This phenomenon is termed antibiotic resilience, distinct from traditional drug resistance or tolerance³. Other mutations include non-synonymous mutations in the dnaA gene, which encodes the highly conserved regulator of DNA replication in MTB, enhance bacterial survival during isoniazid treatment by downregulating katG, the gene responsible for activating isoniazid⁴. Similarly, MTB strains with prpR mutations exhibit conditional drug tolerance by modifying propionyl-CoA metabolism⁵, and frameshift mutations in a sequence of seven cytosines (7C) in a homopolymeric tract (HT) within the glpK gene can temporarily cause GlpK phase variation, which might play a role in drug tolerance, treatment failure, and relapse⁶. Another example of such genetic variation is related to the Rv2752c/rnj gene, which is frequently mutated in drug-resistant MTB isolates. The absence or deletion of the rnj gene has been linked to increased tolerance against high concentrations of various antibiotics⁷.

These mutations are not detected by conventional susceptibility testing but are associated with treatment failures. Existing diagnostic methods overlook these temporal phenotypes, which could explain many treatment failures in infections caused by MTB strains deemed drug-sensitive by standard tests. Identifying ‘pre-resistance’ genomic loci and polymorphisms with a higher likelihood of developing drug resistance could provide insights to foresee and prevent the emergence of resistance⁸.

In some instances where lineage or sublineage-defining variants occur in genes such as whib7 or mmpl5, certain MTB strains are known to have developed altered drug sensitivity, even to antibiotics they have not previously encountered^5,6. This phenomenon is frequently linked to genetic drift. Given the geographical distribution of MTB lineage subgroups, these findings point to the relevance of drug susceptibility diagnostics designed to identify functional variants relevant to specific regional contexts. Nevertheless, altered sensitivity might not always indicate negative outcomes. For instance, a loss-of-function mutation in the whib7 gene within a subset of L1 clade MTB strains has been linked to heightened sensitivity to macrolides⁹. Similarly, certain strains of MTB from the L1 and L4 clades possess loss-of-function mutations in the mmpl5 gene, which could increase their susceptibility to bedaquiline and clofazimine^10,11.

While drug tolerance and resilience are essentially bacterial functions, it is also suggested that specific host environments may favour bacterial adaptations that support MTB survival under antibiotic pressure^5,6. Variations in carbon sources within the host have been linked to shifts in MTB antibiotic sensitivity. For instance, glycerol deprivation in MTB induces a stress-resistance mechanism that heightens antibiotic tolerance⁶. This mechanism arises from transient frameshift mutations impairing the function of the glycerol kinase-encoding gene glpK^6,12. Similarly, prpR gene mutations facilitate tolerance to isoniazid, rifampicin, and ofloxacin during macrophage infections or in liquid media containing propionate⁵. Both glycerol and propionate phenotypes are significant in the nonreplicating state induced by hypoxia, which is a characteristic stress in the lung environment^13,14. Also, in mouse models, mutations in the cydC gene and the rv0096–rv0101 gene set are thought to provoke physiological changes in MTB that interact with isoniazid and mouse tissue, allowing MTB persistence despite isoniazid exposure¹⁵. It is therefore postulated that incorporating host environmental factors into molecular diagnostic algorithms could improve the clinical understanding of resistance phenotypes.

Detecting mutations in bacterial strains that predict the progression towards antibiotic resistance and potentially forecast treatment outcomes can pave the way for personalized drug regimens¹⁶. While individual mutations may only indicate resistance in a subset of strains, a more impactful approach would be to focus on identifying clusters of mutations that are strongly associated with future resistance or clinical outcomes. Combining MTB strain characteristics with patient-specific variables could inspire innovative strategies for optimizing treatment plans and dosage. Recognizing sublineage-defining mutations connected to drug resistance, along with their regional prevalence, could enhance precision in antibiotic selection.

The first step towards detecting mutations that predict progression towards antibiotic resistance involves identification of strains that are tolerant/resilient towards antibiotics. Metrics involving the MIC and the minimal duration required to kill the bacteria have shown encouraging results towards identifying such strains¹⁷. Another study tracked the populations of several beta-lactam-tolerant strains of bacteria over time when exposed to beta-lactam antibiotics. A method to quantify resilience has been proposed. A mathematical model has also been developed to predict antibiotic resilience and identify biological parameters responsible for resilience levels¹⁸.

While the mechanisms related to antibiotic resilience are of growing concern, strategies to counteract these mechanisms with specific molecules are in progress. Artemisinin and structural analogues have been proposed as molecules that block progress to resistance. Molecules specifically targeting persister bacterial strains are under investigation^19-21. Challenges exist in translating laboratory findings on tolerance and resilience into clinically actionable tools. Upregulation and downregulation of mutations are a dynamic process, and the potential to catch the transient window of opportunity to introduce an intervention is challenging. Individual, environmental, and situational variations would make it difficult to predict the trajectory of response to antibiotics. These challenges need to be addressed.

MTB has continuously adapted to host environments. These adaptations also drive its functional genetic diversity. The recognition of functional diversity in MTB and its influence on drug resistance has opened up promising avenues for advancing precision medicine in TB care. There is immense potential in leveraging these insights to predict and prevent treatment failures. Unravelling this diversity can lead to the development of innovative molecular approaches for more effective diagnostics and treatment regimens. It is time to pit our survival skills against those of MTB, and cover the last mile towards the goal of TB elimination.