Pyrazinamide Resistance Is Caused by Two Distinct Mechanisms: Prevention of Coenzyme A Depletion and Loss of Virulence Factor Synthesis

Comparing and Validating Machine Learning Models for Mycobacterium tuberculosis Drug Discovery

Tuberculosis is a global health dilemma. In 2016, the WHO reported 10.4 million incidences and 1.7 million deaths. The need to develop new treatments for those infected with Mycobacterium tuberculosis ( Mtb) has led to many large-scale phenotypic screens and many thousands of new active compounds identified in vitro. However, with limited funding, efforts to discover new active molecules against Mtb needs to be more efficient. Several computational machine learning approaches have been shown to have good enrichment and hit rates. We have curated small molecule Mtb data and developed new models with a total of 18,886 molecules with activity cutoffs of 10 μM, 1 μM, and 100 nM. These data sets were used to evaluate different machine learning methods (including deep learning) and metrics and to generate predictions for additional molecules published in 2017. One Mtb model, a combined in vitro and in vivo data Bayesian model at a 100 nM activity yielded the following metrics for 5-fold cross validation: accuracy = 0.88, precision = 0.22, recall = 0.91, specificity = 0.88, kappa = 0.31, and MCC = 0.41. We have also curated an evaluation set ( n = 153 compounds) published in 2017, and when used to test our model, it showed the comparable statistics (accuracy = 0.83, precision = 0.27, recall = 1.00, specificity = 0.81, kappa = 0.36, and MCC = 0.47). We have also compared these models with additional machine learning algorithms showing Bayesian machine learning models constructed with literature Mtb data generated by different laboratories generally were equivalent to or outperformed deep neural networks with external test sets. Finally, we have also compared our training and test sets to show they were suitably diverse and different in order to represent useful evaluation sets. Such Mtb machine learning models could help prioritize compounds for testing in vitro and in vivo.

Genetics and roadblocks of drug resistant tuberculosis

Considering the extensive evolutionary history of Mycobacterium tuberculosis, anti-Tuberculosis (TB) drug therapy exerts a recent selective pressure. However, in a microorganism devoid of horizontal gene transfer and with a strictly clonal populational structure such as M. tuberculosis the usual, but not sole, path to overcome drug susceptibility is through de novo mutations on a relatively strict set of genes. The possible allelic diversity that can be associated with drug resistance through several mechanisms such as target alteration or target overexpression, will dictate how these genes can become associated with drug resistance. The success demonstrated by this pathogenic microbe in this latter process and its ability to spread is currently one of the major obstacles to an effective TB elimination. This article reviews the action mechanism of the more important anti-TB drugs, including bedaquiline and delamanid, along with new findings on specific resistance mechanisms. With the development, validation and endorsement of new in vitro molecular tests for drug resistance, knowledge on these resistance mechanisms and microevolutionary dynamics leading to the emergence and fixation of drug resistance mutations within the host is highly important. Additionally, the fitness toll imposed by resistance development is also herein discussed together with known compensatory mechanisms. By elucidating the possible mechanisms that enable one strain to reacquire the original fitness levels, it will be theoretically possible to make more informed decisions and develop novel strategies that can force M. tuberculosis microevolutionary trajectory down through a path of decreasing fitness levels.

Impact of immunopathology on the antituberculous activity of pyrazinamide

In the 1970s, inclusion of pyrazinamide (PZA) in the drug regimen of tuberculosis (TB) patients for the first 2 mo achieved a drastic reduction of therapy duration. Until now, however, the mechanisms underlying PZA's unique contribution to efficacy have remained controversial, and animal efficacy data vary across species. To understand how PZA kills bacterial populations present in critical lung lesion compartments, we first characterized a rabbit model of active TB, showing striking similarities in lesion types and fates to nonhuman primate models deemed the most appropriate surrogates of human TB. We next employed this model with lesion-centric molecular and bacteriology readouts to demonstrate that PZA exhibits potent activity against Mycobacterium tuberculosis residing in difficult-to-sterilize necrotic lesions. Our data also indicate that PZA is slow acting, suggesting that PZA administration beyond the first 2 mo may accelerate the cure. In conclusion, we provide a pharmacodynamic explanation for PZA's treatment-shortening effect and deliver new tools to dissect the contribution of immune response versus drug at the lesion level.

Identification of Novel Mutations in LprG (rv1411c), rv0521, rv3630, rv0010c, ppsC, and cyp128 Associated with Pyrazinoic Acid/Pyrazinamide Resistance in Mycobacterium tuberculosis

Despite progress, the mechanisms of action and resistance of pyrazinamide (PZA) are not well understood. We characterized 109 mutants resistant to pyrazinoic acid (POA), the active form of PZA, and found that while most (n = 101 [93%]) mutants had panD mutations and 4 (4%) had clpC1 mutations (S91G), new mutations in lprG (rv1411c) and rv0521 (n = 4 [4%]), rv3630, rv0010c, ppsC, and cyp128 (cytochrome P450 128) were identified, shedding new light on the mechanisms of action and resistance of PZA in M. tuberculosis.

Phthiocerol dimycocerosates promote access to the cytosol and intracellular burden of Mycobacterium tuberculosis in lymphatic endothelial cells

Background

Phthiocerol dimycocerosates (PDIM), glycolipids found on the outer surface of virulent members of the Mycobacterium tuberculosis (Mtb) complex, are a major contributing factor to the pathogenesis of Mtb. Myelocytic cells, such as macrophages and dendritic cells, are the primary hosts for Mtb after infection and previous studies have shown multiple roles for PDIM in supporting Mtb in these cells. However, Mtb can infect other cell types. We previously showed that Mtb efficiently replicates in human lymphatic endothelial cells (hLECs) and that the hLEC cytosol acts as a reservoir for Mtb in humans. Here, we examined the role of PDIM in Mtb translocation to the cytosol in hLECs.

Results

Analysis of a Mtb mutant unable to produce PDIM showed less co-localisation of bacteria with the membrane damage marker Galectin-8 (Gal8), indicating that PDIM strongly contribute to phagosomal membrane damage. Lack of this Mtb lipid also leads to a reduction in the proportion of Mtb co-localising with markers of macroautophagic removal of intracellular bacteria (xenophagy) such as ubiquitin, p62 and NDP52. hLEC imaging with transmission electron microscopy shows that Mtb mutants lacking PDIM are much less frequently localised in the cytosol, leading to a lower intracellular burden.

Conclusions

PDIM is needed for the disruption of the phagosome membrane in hLEC, helping Mtb avoid the hydrolytic phagolysosomal milieu. It facilitates the translocation of Mtb into the cytosol, and the decreased intracellular burden of Mtb lacking PDIM indicates that the cytosol is the preferred replicative niche for Mtb in these cells. We hypothesise that pharmacological targeting of PDIM synthesis in Mtb would reduce the formation of a lymphatic reservoir of Mtb in humans.

Phenotypic and genotypic characterization of pyrazinamide resistance among multidrug-resistant Mycobacterium tuberculosis clinical isolates in Hangzhou, China

OBJECTIVES

Pyrazinamide (PZA) is a crucial first-line tuberculosis (TB) drug recommended for both drug-susceptible and multidrug-resistant Mycobacterium tuberculosis. This study aimed to evaluate the performance of the sequencing method of pncA, rpsA and panD mutations in detecting PZA resistance in multidrug-resistant (MDR) TB isolates.

METHODS

We sequenced the pncA, rpsA and panD genes and performed PZA susceptibility tests across 291 MDR-TB isolates to evaluate the performance of the sequencing method of these genes in detecting PZA resistance.

RESULTS

Results showed that 145 (90.0%) of 161 PZA phenotypic resistant isolates had mutations in pncA. Among the 16 isolates (10.0%) which did not have mutations in pncA, ten and three isolates had mutations in rpsA and panD, respectively. The sequencing method for detecting mutations in pncA alone had 90.1% (95% confidence interval (CI), 84.4-94.2) sensitivity and 92.3% (95% CI, 86.3-96.3) specificity. The combination of all three genes increased the sensitivity from 90.1% (95% CI, 84.4-94.2) to 98.1% (95% CI, 94.7-99.6) (p < 0.001) while the specificity remained unchanged. In 120 PZA-susceptible and 16 PZA-resistant isolates without pncA mutations, rpsA/panD mutations were correlated with PZA resistance.

CONCLUSIONS

PZA resistance was largely associated with mutations in pncA. Mutations in rpsA and panD were also associated with PZA resistance in MDR isolates expressing wild-type pncA. The detection of mutations in pncA, rpsA and panD can be useful for the determination of PZA resistance.

Pyrazinoic Acid Inhibits Mycobacterial Coenzyme A Biosynthesis by Binding to Aspartate Decarboxylase PanD

Previously, we showed that a major in vitro and in vivo mechanism of resistance to pyrazinoic acid (POA), the bioactive component of the critical tuberculosis (TB) prodrug pyrazinamide (PZA), involves missense mutations in the aspartate decarboxylase PanD, an enzyme required for coenzyme A biosynthesis. What is the mechanism of action of POA? Upon demonstrating that treatment of M. bovis BCG with POA resulted in a depletion of intracellular coenzyme A and confirming that this POA-mediated depletion is prevented by either missense mutations in PanD or exogenous supplementation of pantothenate, we hypothesized that POA binds to PanD and that this binding blocks the biosynthetic pathway. Here, we confirm both hypotheses. First, metabolomic analyses showed that POA treatment resulted in a reduction of the concentrations of all coenzyme A precursors downstream of the PanD-mediated catalytic step. Second, using isothermal titration calorimetry, we established that POA, but not its prodrug PZA, binds to PanD. Binding was abolished for mutant PanD proteins. Taken together, these findings support a mechanism of action of POA in which the bioactive component of PZA inhibits coenzyme A biosynthesis via binding to aspartate decarboxylase PanD. Together with previous works, these results establish PanD as a genetically, metabolically, and biophysically validated target of PZA.

Identification of Novel Efflux Proteins Rv0191, Rv3756c, Rv3008, and Rv1667c Involved in Pyrazinamide Resistance in Mycobacterium tuberculosis

Pyrazinamide (PZA) is a critical drug used for the treatment of tuberculosis (TB). PZA is a prodrug that requires conversion to the active component pyrazinoic acid (POA) by pyrazinamidase (PZase) encoded by the pncA gene. Although resistance to PZA is mostly caused by pncA mutations and less commonly by rpsA, panD, and clpC1 mutations, clinical strains without these mutations are known to exist. While efflux of POA was demonstrated in Mycobacterium tuberculosis previously, the efflux proteins involved have not been identified. Here we performed POA binding studies with an M. tuberculosis proteome microarray and identified four efflux proteins (Rv0191, Rv3756c, Rv3008, and Rv1667c) that bind POA. Overexpression of the four efflux pump genes in M. tuberculosis caused low-level resistance to PZA and POA but not to other drugs. Furthermore, addition of efflux pump inhibitors such as reserpine, piperine, and verapamil caused increased susceptibility to PZA in M. tuberculosis strains overexpressing the efflux proteins Rv0191, Rv3756c, Rv3008, and Rv1667c. Our studies indicate that these four efflux proteins may be responsible for PZA/POA efflux and cause PZA resistance in M. tuberculosis Future studies are needed to assess their roles in PZA resistance in clinical strains.

Anti-tubercular Activity of Pyrazinamide is Independent of trans-Translation and RpsA

Pyrazinamide (PZA) is a first line anti-tubercular drug for which the mechanism of action remains unresolved. Recently, it was proposed that the active form of PZA, pyrazinoic acid (POA), disrupts the ribosome rescue process of trans-translation in Mycobacterium tuberculosis. This model suggested that POA binds within the carboxy-terminal domain of ribosomal protein S1 (RpsA) and inhibits trans-translation leading to accumulation of stalled ribosomes. Here, we demonstrate that M. tuberculosis RpsA interacts with single stranded RNA, but not with POA. Further, we show that an rpsA polymorphism previously identified in a PZA resistant strain does not confer PZA resistance when reconstructed in a laboratory strain. Finally, by utilizing an in vitro trans-translation assay with purified M. tuberculosis ribosomes we find that an interfering oligonucleotide can inhibit trans-translation, yet POA does not inhibit trans-translation. Based on these findings, we conclude that the action of PZA is entirely independent of RpsA and trans-translation in M. tuberculosis.

In Vivo-Selected Pyrazinoic Acid-Resistant Mycobacterium tuberculosis Strains Harbor Missense Mutations in the Aspartate Decarboxylase PanD and the Unfoldase ClpC1

Through mutant selection on agar containing pyrazinoic acid (POA), the bioactive form of the prodrug pyrazinamide (PZA), we recently showed that missense mutations in the aspartate decarboxylase PanD and the unfoldase ClpC1, and loss-of-function mutation of polyketide synthases Mas and PpsA-E involved in phthiocerol dimycocerosate synthesis, cause resistance to POA and PZA in Mycobacterium tuberculosis. Here we first asked whether these in vitro-selected POA/PZA-resistant mutants are attenuated in vivo, to potentially explain the lack of evidence of these mutations among PZA-resistant clinical isolates. Infection of mice with panD, clpC1, and mas/ppsA-E mutants showed that whereas growth of clpC1 and mas/ppsA-E mutants was attenuated, the panD mutant grew as well as the wild-type. To determine whether these resistance mechanisms can emerge within the host, mice infected with wild-type M. tuberculosis were treated with POA, and POA-resistant colonies were confirmed for PZA and POA resistance. Genome sequencing revealed that 82 and 18% of the strains contained missense mutations in panD and clpC1, respectively. Consistent with their lower fitness and POA resistance level, independent mas/ppsA-E mutants were not found. In conclusion, we show that the POA/PZA resistance mechanisms due to panD and clpC1 missense mutations are recapitulated in vivo. Whereas the representative clpC1 mutant was attenuated for growth in the mouse infection model, providing a possible explanation for their absence among clinical isolates, the growth kinetics of the representative panD mutant was unaffected. Why POA/PZA resistance-conferring panD mutations are observed in POA-treated mice but not yet among clinical strains isolated from PZA-treated patients remains to be determined.

Missense Mutations in the Unfoldase ClpC1 of the Caseinolytic Protease Complex Are Associated with Pyrazinamide Resistance in Mycobacterium tuberculosis

Previously, we showed that mutations in Mycobacterium tuberculosispanD, involved in coenzyme A biosynthesis, cause resistance against pyrazinoic acid, the bioactive component of the prodrug pyrazinamide. To identify additional resistance mechanisms, we isolated mutants resistant against pyrazinoic acid and subjected panD wild-type strains to whole-genome sequencing. Eight of the nine resistant strains harbored missense mutations in the unfoldase ClpC1 associated with the caseinolytic protease complex.

Long-Chain Fatty Acyl Coenzyme A Ligase FadD2 Mediates Intrinsic Pyrazinamide Resistance in Mycobacterium tuberculosis

Pyrazinamide (PZA) is a first-line tuberculosis (TB) drug that has been in clinical use for 60 years yet still has an unresolved mechanism of action. Based upon the observation that the minimum concentration of PZA required to inhibit the growth of Mycobacterium tuberculosis is approximately 1,000-fold higher than that of other first-line drugs, we hypothesized that M. tuberculosis expresses factors that mediate intrinsic resistance to PZA. To identify genes associated with intrinsic PZA resistance, a library of transposon-mutagenized Mycobacterium bovis BCG strains was screened for strains showing hypersusceptibility to the active form of PZA, pyrazinoic acid (POA). Disruption of the long-chain fatty acyl coenzyme A (CoA) ligase FadD2 enhanced POA susceptibility by 16-fold on agar medium, and the wild-type level of susceptibility was restored upon expression of fadD2 from an integrating mycobacterial vector. Consistent with the recent observation that POA perturbs mycobacterial CoA metabolism, the fadD2 mutant strain was more vulnerable to POA-mediated CoA depletion than the wild-type strain. Ectopic expression of the M. tuberculosis pyrazinamidase PncA, necessary for conversion of PZA to POA, in the fadD2 transposon insertion mutant conferred at least a 16-fold increase in PZA susceptibility under active growth conditions in liquid culture at neutral pH. Importantly, deletion of fadD2 in M. tuberculosis strain H37Rv also resulted in enhanced susceptibility to POA. These results indicate that FadD2 is associated with intrinsic PZA and POA resistance and provide a proof of concept for the target-based potentiation of PZA activity in M. tuberculosis.

Validation of CoaBC as a Bactericidal Target in the Coenzyme A Pathway of Mycobacterium tuberculosis

Mycobacterium tuberculosis relies on its own ability to biosynthesize coenzyme A to meet the needs of the myriad enzymatic reactions that depend on this cofactor for activity. As such, the essential pantothenate and coenzyme A biosynthesis pathways have attracted attention as targets for tuberculosis drug development. To identify the optimal step for coenzyme A pathway disruption in M. tuberculosis, we constructed and characterized a panel of conditional knockdown mutants in coenzyme A pathway genes. Here, we report that silencing of coaBC was bactericidal in vitro, whereas silencing of panB, panC, or coaE was bacteriostatic over the same time course. Silencing of coaBC was likewise bactericidal in vivo, whether initiated at infection or during either the acute or chronic stages of infection, confirming that CoaBC is required for M. tuberculosis to grow and persist in mice and arguing against significant CoaBC bypass via transport and assimilation of host-derived pantetheine in this animal model. These results provide convincing genetic validation of CoaBC as a new bactericidal drug target.

Recent advancements in the development of anti-tuberculosis drugs

Modern chemotherapy has significantly improved patient outcomes against drug-sensitive tuberculosis. However, the rapid emergence of drug-resistant tuberculosis, together with the bacterium's ability to persist and remain latent present a major public health challenge. To overcome this problem, research into novel anti-tuberculosis targets and drug candidates is thus of paramount importance. This review article provides an overview of tuberculosis highlighting the recent advances and tools that are employed in the field of anti-tuberculosis drug discovery. The predominant focus is on anti-tuberculosis agents that are currently in the pipeline, i.e. clinical trials.