Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2025 May 1.
Published in final edited form as: Tuberculosis (Edinb). 2024 Feb 27;146:102500. doi: 10.1016/j.tube.2024.102500

Computational Drug Repositioning Identifies Niclosamide and Tribromsalan as Inhibitors of Mycobacterium tuberculosis and Mycobacterium abscessus

Jeremy J Yang 1,2,3,#, Aaron Goff 4,#, David J Wild 1,2, Ying Ding 1,2,5, Ayano Annis 6, Randy Kerber 2, Brian Foote 2, Anurag Passi 7, Joel L Duerksen 2, Shelley London 2, Ana C Puhl 8, Thomas R Lane 8, Miriam Braunstein 6, Simon J Waddell 4, Sean Ekins 8,*
PMCID: PMC10978224  NIHMSID: NIHMS1972153  PMID: 38432118

Abstract

Tuberculosis (TB) is still a major global health challenge, killing over 1.5 million people each year, and hence, there is a need to identify and develop novel treatments for Mycobacterium tuberculosis (M. tuberculosis). The prevalence of infections caused by nontuberculous mycobacteria (NTM) is also increasing and has overtaken TB cases in the United States and much of the developed world. Mycobacterium abscessus (M. abscessus) is one of the most frequently encountered NTM and is difficult to treat. We describe the use of drug-disease association using a semantic knowledge graph approach combined with machine learning models that has enabled the identification of several molecules for testing anti-mycobacterial activity. We established that niclosamide (M. tuberculosis IC90 2.95 μM; M. abscessus IC90 59.1 μM) and tribromsalan (M. tuberculosis IC90 76.92 μM; M. abscessus IC90 147.4 μM) inhibit M. tuberculosis and M. abscessus in vitro. To investigate the mode of action, we determined the transcriptional response of M. tuberculosis and M. abscessus to both compounds in axenic log phase, demonstrating a broad effect on gene expression that differed from known M. tuberculosis inhibitors. Both compounds elicited transcriptional responses indicative of respiratory pathway stress and the dysregulation of fatty acid metabolism.

Keywords: Drug discovery, machine learning, Mycobacterium abscessus, Mycobacterium tuberculosis, transcriptome profiling

1. Introduction

Tuberculosis (TB) is a major world health challenge in need of new drug treatments. The latest WHO annual TB report described 1.6 million fatalities due to TB and 10.6 million new TB cases in 2021 [1] Where facilities and drugs are available, drug-sensitive TB treatment has an 85% success rate; however, drug regimens for multidrug-resistant (MDR) Mycobacterium tuberculosis (M. tuberculosis) is a challenge as the treatments currently available in many parts of the world have decreased efficacy, and debilitating side effects that range from liver toxicity, hearing loss, and psychosis [2–4]. Several reviews of recent TB drug research have described the pipeline and challenges in drug discovery [5, 6]. New drugs for TB have been recently approved for use in the U.S. (delamanid, bedaquiline and pretomanid), these are now recommended as part of combination therapy for multi-drug resistant TB [7]. Delamanid and pretomanid work by blocking the synthesis of mycolic acids in M. tuberculosis, thus destabilizing its cell wall while bedaquiline inhibits the proton pump for ATP synthase [8, 9]. Further new drugs with more diverse targets are needed to populate the early-stage pipeline.

While TB is a well-known disease with evidence-based treatment strategies, in contrast nontuberculous mycobacteria (NTM) are under-researched with drug regimens based on limited empirical data [10]. NTM infections are of concern to patients with pulmonary diseases like cystic fibrosis [10], where antibiotic courses may be frequent and may interact with their other treatments. NTM are present in water and soil [11], and are able to survive in diverse environments [12]. Mycobacterium abscessus (M. abscessus) is particularly difficult to treat due to intrinsic resistance to common antibiotics [13]. A lack of correlation between in vitro and in vivo drug efficacy also hinders drug discovery efforts [12, 14]. Hence novel drugs with new mechanisms of action are required.

Drug repositioning or repurposing is one strategy to try to identify anti-mycobacterial molecules [15, 16]. Many agents have been tested in humans for other diseases using predominantly high throughput screening approaches [17–20], however there have been fewer efforts using machine learning [21]. Because repositioning builds upon previous research and development efforts, new candidate therapies are closer to clinical trials, reducing timeframes and costs, and speeding their review by medicines regulators, and, if approved, their integration into health care.

The novel drug-disease associations reported herein were identified using a semantic knowledge graph (KG) database, integrating several leading biomedical resources to accelerate discovery of new treatments for diseases, along with a first-in-kind tool that allows association finding across semantically linked data sets such as these [22]. The two key sources for building the KG were the open Phenotypic Drug Discovery Resource (PDDR) [23] and the OpenPHACTS Open Pharmacological Space (OPS) [24]. Here, we describe our modelling approach and the identification of molecules that we tested in vitro against M. tuberculosis and M. abscessus that led to the identification of active molecules. We further explored the impact of drug action by transcriptome profiling.

2. Methods

2.1. Chemicals and reagents

All reagents and solvents were purchased from commercial suppliers and used without further purification. All compounds were found and procured via eMolecules, Inc. (eMolecules.com). Niclosamide and tribromsalan came from Vitas M Labs (Causeway Bay, Hong Kong) and Alinda (Moscow, Russia), respectively. Compounds were resuspended in DMSO (MilliporeSigma). Resazurin sodium salt, tyloxapol, Triton X 100, and kanamycin sulfate salt were purchased from MilliporeSigma.

2.2. Drug-disease association using a semantic knowledge graph

Central to the computational methodology employed in this project was a knowledge graph (KG) built from OPS, an EU-funded research program which developed a large scale, semantic triple-store integrating numerous leading datasets including ChEBI [25], ChEMBL [26], ConceptWiki [27], DisGeNET [28], DrugBank [29], ENZYME, FDA Adverse Events (FAERS) [30], Gene Ontology [31], neXTProt [32], UniProt [33] and WikiPathways [24]. The PDDR [23] is a result of a unique collaboration between Lilly’s Open Innovation in Drug Discovery (OIDD) program and NCATS. NCATS provided physical samples of around 2,500 marketed drugs, from the National Pharmaceutical Collection (NPC). These drugs were tested in the OIDD phenotypic assays, and the results made publicly available. To maximize the impact of these data, we mapped the dataset into other public data sources using modern, semantic formats with support from NCATS, Lilly and Indiana CTSI. The result was the PDDR, including a KG freely available on the NCATS website. We have created a semantically rich KG, and analytic tools that can find important indirect associations across complex biomedical data sets [22]. Methods for building and querying the KG are described in a patent application [34]. Applications include predicting drug-target interactions, and a wide range of biomedical sciences challenges including screening analytics, cheminformatics-based modeling, drug repositioning, and target identification. P3 (Predictive Phenotypic Profiler) is a KG user interface developed for knowledge discovery (Figure 1). Data from the OPS and the PDDR was combined and semantically mapped. 157 existing M. tuberculosis-related compounds and drugs were identified in this combined dataset, and manual annotations created that link the drugs to a single “Tuberculosis” disease classifier. P3 was used to identify strong associative connections between drugs, targets (from the OPS set), drugs and phenotypic assays (from the PDDR) and TB. In this way, both the experimental drug-target associations from OPS and the experimental phenotypic results from the PDDR were used together to identify associations between existing drugs and M. tuberculosis. Graph analytics algorithms were used to evaluate evidence, score, and rank associations (Figure 2). Drugs that met a “strong association” threshold with M. tuberculosis were extracted for further manual analysis and refinement. Algorithmic criteria for association strength were based on simple graph-analytic measures which assess the strength of association as a combination of relevance and confidence, where relevance is based on semantic proximity, i.e. graph distance to TB, and confidence based on weight of evidence, i.e. quantity of evidence path instances. These methods are comparable with other KG-based knowledge discovery approaches such as by Himmelstein et al. [35], and Bizon et al. [36]. Subsequent to the KG search stage, the manual refinement stage involved evaluating each strongly associated compound, based on known current indications, toxicity profiles and druggability. Compounds that passed this initial evaluation were selected for computational follow-up and wet lab study. We now present the key conceptual basis of association finding and strength assessment, illustrated by Figures 1, 2a and 2b, to provide readers with an overview of our methods.

Figure 1.

Figure 1.

Open in a new tab

Metagraph visualization representing schema connecting drugs, targets, phenotypic assays, diseases, and other entities. Association strength is based on simple graph-analytic measures which assess the strength of association as a combination of relevance and confidence, where relevance is based on semantic proximity, i.e. graph distance to TB, and confidence based on weight of evidence, i.e. quantity of evidence path instances.

Figure 2.

Figure 2.

Open in a new tab

a. Representative evidence subgraph associating M. tuberculosis/M. abscessus pathways, targets, and drugs (green human proteins, red disease outcomes). b. Cytoscape network of M. tuberculosis/M. abscessus (disease) and M. tuberculosis/M. abscessus (finding) to list of potentially anti-mycobacterial compounds. Linking drug action to human immune mediators involved in response to mycobacterial disease states. Association strength is based on simple graph-analytic measures which assess the strength of association as a combination of relevance and confidence, where relevance is based on semantic proximity, i.e. graph distance to TB, and confidence based on weight of evidence, i.e. quantity of evidence path instances.

2.3. Assay Central

Molecules were scored with the previously published M. tuberculosis Bayesian Models using Assay Central software to provide an additional approach to evaluating the selected molecules from the initial data mining [37]. The machine learning model cut-offs were for actives at 100 nM, 1 μM and 10 μM. These models include a domain metric which represents the chemical space covered by a model relative to the ChEMBL 25 database where a higher value indicates the training data is likely more generally applicable (ranging from 0-1) [38].

2.4. Resazurin Microtiter Assay (REMA)

Compounds were tested on Mycobacterium abscessus ATCC 19977 (smooth) and Mycobacterium tuberculosis H37Rv using a REMA assay, as described previously [39, 40]. In brief, compounds were prepared in DMSO or kanamycin or amikacin, as positive controls, were prepared in deionized water and sterile filtered. Using non-treated polystyrene 96-well plates (Corning), drugs were serially two-fold diluted in triplicate in 7H9 broth (Difco) supplemented with albumin dextrose saline (ADS; 10 g/L bovine serum albumin fraction V, 4 g/L dextrose, 1.6 g/L NaCl), 0.5% glycerol, and 0.1% Tyloxapol (7AGT). M. abscessus ATCC 19977 and M. tuberculosis H37Rv were pre-grown in 7AGT until mid-logarithmic growth was reached. Mycobacterial cultures were added at a final density of 1 X 105 cells/well. All wells, including test compounds and kanamycin controls, contained a final concentration of 1% DMSO and 200 μL total volume for M. abscessus or 1% DMSO and 100 μL total volume for M. tuberculosis. Plates were incubated for 48hrs for M. abscessus or 96hrs for M. tuberculosis at 37°C at 80-100 rpm before adding 20 μL for M. abscessus or 10 μL for M. tuberculosis of resazurin solution (125 μg/mL in phosphate buffered saline). Following the addition of resazurin, plates were incubated in the dark for an additional 24 hrs for M. abscessus or 96 hrs for M. tuberculosis. Fluorescence was measured with an excitation at 544nm and emission at 590nm with a plate reader. These MIC data are plotted in Figure 3 and Figure 4.

Figure 3.

Figure 3.

Open in a new tab

Inhibition of M. tuberculosis H37Rv by niclosamide (IC50 1.92 μM, IC90 2.95 μM), tribromsalan (IC50 67.16 μM, IC90 76.92 μM) and the control kanamycin (Kan) (IC50 1.06 μM, IC90 4.22 μM). Error bars are from two independent experiments.

Figure 4.

Figure 4.

Open in a new tab

Inhibition of M. abscessus ATCC 19977 by niclosamide (IC50 32.23 μM, IC90 59.1 μM), tribromsalan (IC50 99.75 μM, IC90 147.4 μM) and the controls kanamycin (Kan) (IC50 12.11 μM, IC90 16.07 μM) and amikacin (Amik) (IC50 4.78 μM, IC90 7.15 μM). Error bars are from two independent experiments.

2.5. Determination of Minimum Inhibitory Concentrations (MICs) prior to mode of action transcriptomics

MICs of niclosamide and tribromsalan were measured against Mycobacterium tuberculosis H37Rv in axenic log phase culture in vitro to determine conditions to map the drug exposure signatures. Drugs were prepared as 10 mM stock solutions in DMSO and stored at −20’C in aliquots to avoid freeze-thawing. M. tuberculosis was cultured in Middlebrook 7H9 broth (with 0.05% Tween 80, 10% ADC), and 96-well microtitre plates containing two-fold dilutions of each compound were inoculated with M. tuberculosis to a final concentration of 1 x 105 – 5 x 105 CFU/mL. Plates were incubated for 7 days at 37’C and MIC values then determined by REMA following 16 h incubation with CellTiter-Blue (Promega, Wisconsin, United States). Wells containing media only were used to correct for background and fluorescence was measured on a plate reader (Glomax Discover, Promega) to determine the MIC, calculated as the concentration of drug that inhibited 90% growth. MICs were determined from three independent biological replicates, each with three technical replicates per plate. Niclosamide and tribromsalan MICs were also determined for Mycobacterium abscessus ATCC19977 as described above incubating with drug for 3 days. These MIC data are plotted in Figure S1.

2.6. Mycobacterial Gene Expression Profiling

Log phase M. tuberculosis or M. abscessus (4-day culture shaking at 150 rpm, OD 0.2-0.4) were treated with 10x MIC niclosamide or tribromsalan for 4 hours, or vehicle control only (final concentration 0.75% sterile DMSO). Following drug exposure, mycobacterial RNA was extracted using an established GTC/TRIzol method [41]. Samples were extracted from three independent biological replicates on different days. The resulting mycobacterial RNA was DNase-treated and purified using the RNeasy spin columns (Qiagen, Hilden, Germany). RNA yield and quality were assessed by NanoDrop One (Thermo Scientific, Waltham, MA, USA) and Agilent 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA, USA). RNA yields from all samples were similar and of good quality. Mycobacterial ribosomal RNA was depleted using RiboCop META kit (Lexogen, Vienna, Austria) resulting in mRNA suitable for RNA sequencing. Strand-specific RNA-seq libraries were prepared using the NEBNext Ultra II directional RNA library prep kit (New England Biolabs, Ipswitch, MA, USA) and pooled in-house.

2.7. RNA sequencing and analysis

Libraries were single-end sequenced using an Illumina NextSeq500 (San Diego, CA, United States) with Med Output Illumina kit 150 cycles PE 2x 75 cycles. The raw sequencing run was demultiplexed on the Illumina BaseSpace platform, generating fastq.gz files for subsequent analysis. Single-end raw sequencing reads from all samples passed quality control (FastQC) [42], and reads were mapped to the Mycobacterium tuberculosis H37Rv reference genome using Hisat2 [43]. Each sample yielded between 10 – 22 million reads and 97-99% of the sample reads mapped uniquely to the M. tuberculosis genome. FeatureCounts from the Subread package v1.5.2 was used to quantify gene expression. Differentially expressed genes were identified in drug-treated M. tuberculosis compared to drug-free vehicle control M. tuberculosis in DESeq2 R package v3.6.0, normalized with the RLE method, using the Wald test with Benjamini and Hochberg multiple testing correction [44]. Only primary alignments were counted, ignoring reads mapping to multiple locations. Differentially expressed genes were considered significant with a log2-fold change (L2FC) <−3 or >3 with a corrected p-value <0.05. Differentially expressed genes are detailed in Table S3.

The hypergeometric function using p-value < 0.05 was used to identify significant overlaps with previously published transcriptional signatures and functional categories. Differentially expressed genes were also mapped to metabolic pathways in M. tuberculosis using DAVID Functional Annotation tool [45] and gene set enrichment analysis [46].

3. Results

3.1. Drug disease association using a semantic knowledge graph

Data from the OPS and the PDDR was combined and semantically mapped to identify 157 existing M. tuberculosis-related compounds and drugs. Drugs that met a “strong association” threshold with M. tuberculosis were extracted for further manual analysis and refinement. Fourteen hit compounds were initially identified using our drug-disease association, a semantic linked data approach, that met the criteria of strong association (Table S1). Each compound was manually triaged for “chemical liabilities”, and at the time, we were not aware of any of these compounds being tested in vitro against M. tuberculosis. Of the 14 hit drugs, two were known topical antibacterials that are unlikely to be successful by the oral route but could be considered for other routes (hexachlorophene, tribromsalan), and three were drugs known to have severe toxicity issues (bithionol, gossypol, captan). This demonstrates that our approach can generate new repositioning hypotheses from existing knowledge via SEMAP graph analytics.

3.2. Anti-mycobacterial activity prediction using Assay Central

All molecules were also scored with M. tuberculosis machine learning models described previously (Table S2), and several had predicted promising activity at the activity threshold providing a semi-quantitative indicator of likely activity in which higher prediction scores are better. Interestingly, niclosamide and tribromsalan have borderline scores at the 10μM threshold suggesting that they may have activity in this range and this as well as molecule availability was used to select them for further testing.

3.3. Anti-mycobacterial activity in vitro

REMA was used to determine the in vitro activity against M. tuberculosis and M. abscessus. Niclosamide and tribromsalan demonstrated promising μM in vitro activity against M. tuberculosis (Figure 3) and the former is performing like the control kanamycin. For M. abscessus, both niclosamide and tribromsalan have much higher IC90 values than the controls kanamycin and amikacin (Figure 4) and they are noticeably higher than for M. tuberculosis.

3.4. Mycobacterial response to drug exposure

After confirming the MIC of niclosamide and tribromsalan in an independent laboratory (Figure S1), the M. tuberculosis and M. abscessus transcriptional signatures to drug exposure were determined to help categorize drug mode of action. The transcriptional responses to niclosamide and tribromsalan were broad, both in number of genes impacted and in magnitude of differential expression, with ~500 genes significantly differentially expressed (corrected p<0.05, L2FC <−3/>3) after either drug treatment. The M. tuberculosis drug signatures are detailed in Figure 5, and M. abscessus drug responses in Figure S2. Differentially expressed genes are listed in Table S3.

Figure 5.

Figure 5.

Open in a new tab

Niclosamide and tribromsalan M. tuberculosis transcriptional signatures. (A) M. tuberculosis summary of differential expression after exposure to niclosamide or tribromsalan compared to carrier control. (B) Venn diagram of overlapping significantly expressed genes (L2FC 3) after exposure to niclosamide or tribromsalan. (C) Heatmap of M. tuberculosis responses to niclosamide or tribromsalan relative to drug-free carrier control. Conditions as columns, genes as rows; red coloring highlighting significantly induced genes, blue repressed genes. The final two columns indicate in which comparison each gene was identified (n, niclosamide; t, tribromsalan), showing similar responses to both compounds.

A principal component analysis separated niclosamide and tribromsalan treatments from other established anti-mycobacterial drugs, suggesting a novel mode of action (Figure S3). Pathway enrichment analysis identified biotin-carboxyl carrier protein assembly as significantly impacted by both niclosamide and tribromsalan exposure (PPS scores of 60 and 33 for genes induced by niclosamide and tribromsalan, respectively). Fatty acid metabolism pathways were also highlighted, for example mycolate biosynthesis (PPS scores for niclosamide of 41 or 101 for up or down regulated genes, respectively; and for tribromsalan PPS scores of 47 or 52, respectively) and beta oxidation of fatty acids (PPS scores for niclosamide of 59 or 42 for up or down regulated genes, respectively; and for tribromsalan PPS score of 47 for upregulated genes) categories were enriched in both induced and down-regulated gene signatures, indicating a general dysregulation of these systems.

Bedaquiline was the most similar anti-tuberculosis drug signature to niclosamide and tribromsalan (hypergeometric probabilities of 4.69x10−74 and 1.74x10−66, respectively) of the anti-mycobacterial drugs profiled by RNA-seq under the same experimental conditions. Comparison to the Boshoff et al. [47] drug signature dataset revealed that both drugs clustered with ‘agents besides nitric oxide that inhibit respiration’ (for example for niclosamide, induction of gene clusters 13, 46, 140, 63 and repression of gene clusters 36, 139, 71). In agreement with this analysis, the Boshoff et al. treatment responses most like niclosamide and tribromsalan were CCCP (carbonyl cyanide m-chlorophenyl hydrazine), triclosan and valinomycin (hypergeometric probabilities all <1.5x10−15); all compounds that disrupt membrane-respiratory function. As expected in response to oxidative or nitrosative stress, the regulon of the transcription factor SigH was significantly induced by exposure to both drugs (p-value of regulon enrichment <6.0x10−8) [48].

4. Discussion

We have described how niclosamide and tribromsalan were identified by drug-disease association KG and tested in vitro to demonstrate their μM activity against M. tuberculosis and M. abscessus. Since initiating this study, the anti-mycobacterial activity of niclosamide has been described in vitro against M. tuberculosis [49] and M. abscessus [50]. The activity of tribromsalan against M. tuberculosis and M. abscessus has not previously been reported to our knowledge. Both compounds share some structural similarity to a recently reported inhibitor of M. tuberculosis MmpL3 (HC2149) [51], so this may represent a potential target for future evaluation. The synthesis of related salicylanilides has been reported that are structurally similar to niclosamide and tribromsalan [52]. To our knowledge tribromsalan represents a potentially useful compound which may provide an inhaled drug [53] for treating M. tuberculosis and M. abscessus that could overcome some of the limitations of the topical use of this drug, such as reports of photosensitizing. Future testing of tribromsalan against other mycobacteria such as the M. avium complex as well as drug resistant isolates of M. tuberculosis and M. abscessus may help to understand the utility and target of this drug in mycobacteria. The in vitro data presented here would suggest that tribromsalan is less potent than kanamycin or niclosamide against M. abscessus, which is a difficult disease to find novel drugs for [40] other than repurposing well-known FDA approved antibacterials [12, 14].

In this study, we profiled the response of M. tuberculosis H37Rv and M. abscessus ATCC19977 to x10 MIC drug for 4h in comparison to drug-free carrier controls. The first point to note is the breadth and magnitude of the transcriptional response, using standard significance cutoffs of p<0.05 with multiple testing correction and log 2-fold change (L2FC +1/−1) resulted in >2,000 significantly differentially expressed genes (DEG) for either niclosamide or tribromsalan. In comparison, isoniazid, a first line M. tuberculosis drug targeting FAS-II, profiled using the same conditions yielded ~100 DEG [54]. This suggests that whatever niclosamide and tribromsalan are targeting, the impact on M. tuberculosis (and M. abscessus) is immediate and affects multiple systems. From comparison to other M. tuberculosis drug signatures (specifically the Boshoff dataset [47]) it is clear that niclosamide and tribromsalan inhibit respiration (likely from substantial oxidative or nitrosative stress) and dysregulate fatty acid biosynthetic pathways. This is also reflected in the regulons differentially regulated, particularly SigH (Rv3223c) induced by reactive oxidative/nitrosative intermediates, and Rv0023 and Rv0081, regulators implicated in changing respiratory status due to hypoxia. However the mycobacterial responses to these compounds are greater than a respiratory shock, and reflect observations that niclosamide, and the related antiparasitic drug nitazoxanide, destroy mycobacterial proton motive force [55]. These observations suggest a number of possibilities: (1) these drugs are specific inhibitors of a membrane energetics pathway that quickly affects proton motif force, respiration and cell wall biosynthesis; (2) they inhibit a key cell wall component that results in immediate collapse of cell membrane structure, affecting respiratory processes and cell wall function; (3) they are less specific, inserting into or disrupting membranes/lipid-rich cell walls, which destroys membrane potential and respiratory processes. Moving forwards, it will be useful to have tools to discriminate these possibilities for mycobacterial drug discovery. Testing these drugs in combination with other anti-mycobacterial drugs, for example the ATP synthase inhibitor bedaquiline, may further help to reveal function, especially in M. abscessus where new drug options are urgently required [56].

It is important to note that all experimental analyses were blinded with respect to the molecule identity and for the transcriptomics no prior evidence of proposed mechanism of action was provided. The transcript analysis therefore is unsupervised and based on multigene comparisons. The combination of machine learning and mode of action techniques applied here, has identified, and characterized two repurposed drugs, niclosamide and tribromsalan, contributing to drug discovery efforts to improve M. tuberculosis and NTM multi-drug regimens.

Supplementary Material

1
2

Acknowledgments

SE kindly acknowledges NIH NIGMS funding to develop the Assay Central software from R44GM122196. SJW and AG acknowledge funding from the National Centre for the Replacement, Refinement and Reduction of Animals in Research (NC/R001669/1), and the University of Sussex Higher Education Innovation Fund (HEIF).

Footnotes

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Conflicts of interest

S.E. is owner, and A.C.P. and T.R.L. are employees of Collaborations Pharmaceuticals, Inc. J.J.Y., D.J.W., Y.D., B.F., and J.L.D., are employees, and D.J.W., B.F., R.K., and S.L., are board members, of Data2Discovery, Inc., which has applied for patent protections covering discoveries described herein. All others have no competing interests.

Research data for this article

Fully annotated M. tuberculosis RNA-seq data have been deposited in EBI BioStudies; accession number E-MTAB-13761.

References

  • [1].WHO, Global Tuberculois Report, https://www.who.int/teams/global-tuberculosis-programme/tb-reports/global-tuberculosis-report-2022
  • [2].Seung KJ, Keshavjee S, Rich ML. Multidrug-Resistant Tuberculosis and Extensively Drug-Resistant Tuberculosis. Cold Spring Harb Perspect Med 2015;5:a017863. doi: 10.1101/cshperspect.aOl7863 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [3].Arnold A, Cooke GS, Kon OM, Dedicoat M, Lipman M, Loyse A, Chis Ster I, Harrison TS. Adverse Effects and Choice between the Injectable Agents Amikacin and Capreomycin in Multidrug-Resistant Tuberculosis. Antimicrob Agents Chemother 2017;61 doi: 10.1128/AAC.02586-16 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [4].Anon, Inadequate Treatment, https://www.tballiance.org/why-new-tb-drugs/inadequate-treatment
  • [5].Zumla A, Chakaya J, Centis R, D’Ambrosio L, Mwaba P, Bates M, Kapata N, Nyirenda T, Chanda D, Mfinanga S, Hoelscher M, Maeurer M, Migliori GB. Tuberculosis treatment and management--an update on treatment regimens, trials, new drugs, and adjunct therapies. Lancet Respir Med 2015;3:220–234. doi: 10.1016/S2213-2600(15)00063-6 [DOI] [PubMed] [Google Scholar]
  • [6].Makarov V, Salina E, Reynolds RC, Kyaw Zin PP, Ekins S. Molecule Property Analyses of Active Compounds for Mycobacterium Tuberculosis. J Med Chem 2020;63:8917–8955. doi: 10.1021/acs.jmedchem.9b02075 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [7].WHO, WHO consolidated guidelines on tuberculosis. Module 4: treatment - drug-resistant tuberculosis treatment, 2022. update, https://www.who.int/publications/i/item/9789240063129 [PubMed]
  • [8].Esposito S, Bianchini S, Blasi F. Bedaquiline and delamanid in tuberculosis. Expert Opin Pharmacother 2015;16:2319–2330. doi: 10.1517/14656566.2015.1080240 [DOI] [PubMed] [Google Scholar]
  • [9].Dawson R, Diacon AH, Everitt D, van Niekerk C, Donald PR, Burger DA, Schall R, Spigelman M, Conradie A, Eisenach K, Venter A, Ive P, Page-Shipp L, Variava E, Reither K, Ntinginya NE, Pym A, von Groote-Bidlingmaier F, Mendel CM. Efficiency and safety of the combination of moxifloxacin, pretomanid (PA-824), and pyrazinamide during the first 8 weeks of antituberculosis treatment: a phase 2b, open-label, partly randomised trial in patients with drug-susceptible or drug-resistant pulmonary tuberculosis. Lancet 2015. doi: 10.1016/S0140-6736(14)62002-X [DOI] [PubMed] [Google Scholar]
  • [10].Egorova A, Jackson M, Gavrilyuk V, Makarov V. Pipeline of anti-Mycobacterium abscessus small molecules: Repurposable drugs and promising novel chemical entities. Med Res Rev 2021. doi: 10.1002/med.21798 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [11].Honda JR, Hasan NA, Davidson RM, Williams MD, Epperson LE, Reynolds PR, Smith T, Iakhiaeva E, Bankowski MJ, Wallace RJ Jr., Chan ED, Falkinham JO 3rd, Strong M Environmental Nontuberculous Mycobacteria in the Hawaiian Islands. PLoS Negl Trop Dis 2016;10:e0005068. doi: 10.1371/journal.pntd.0005068 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [12].Lopeman RC, Harrison J, Desai M, Cox JAG. Mycobacterium abscessus: Environmental Bacterium Turned Clinical Nightmare. Microorganisms 2019;7 doi: 10.3390/microorganisms7030090 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [13].Cantillon D, Goff A, Taylor S, Salehi E, Fidler K, Stoneham S, Waddell SJ. Searching for new therapeutic options for the uncommon pathogen Mycobacterium chimaera: an open drug discovery approach. Lancet Microbe 2022;3:e382–e391. doi: 10.1016/S2666-5247(21)00326-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [14].Ganapathy US, Dartois V, Dick T. Repositioning rifamycins for Mycobacterium abscessus lung disease. Expert Opin Drug Discov 2019;14:867–878. doi: 10.1080/17460441.2019.1629414 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [15].Mori G, Orena BS, Franch C, Mitchenall LA, Godbol AA, Rodrigues L, Aguilar-Pérezd C, Zemanováf J, Huszár S, Forbak M, Lane TR, Sabbahh M, Deboosere N, Frita R, Vandeputtei A, Hoffmann E, Russo R, Connell N, Veilleux C, Kumar R, Kumar P, Freundlich JS, Brodin P, Aínsa JA, Nagaraja V, Maxwell A, Mikuţová K, Pasca MR, Ekins S. The EU Approved Antimalarial Pyronaridine Shows Antitubercular Activity and Synergy with Rifampicin, Targeting RNA Polymerase. Tuberculosis 2018;112:98–109. [DOI] [PubMed] [Google Scholar]
  • [16].Ekins S, Williams AJ, Krasowski MD, Freundlich JS. In silico repositioning of approved drugs for rare and neglected diseases. Drug Discov Today 2011;16:298–310. doi: 10.1016/j.drudis.2011.02.016 [DOI] [PubMed] [Google Scholar]
  • [17].Madrid PB, Panchal RG, Warren TK, Shurtleff AC, Endsley AN, Green CE, Kolokoltsov AA, Davey RA, Manger ID, Gilfillan L, Bavari S, Tanga MJ. Evaluation of Ebola Virus Inhibitors for Drug Repurposing. ACS Inf Dis 2015;1:317–326. [DOI] [PubMed] [Google Scholar]
  • [18].Ekins S, Gerlach J, Zorn KM, Antonio BM, Lin Z, Gerlach A. Repurposing Approved Drugs as Inhibitors of K(v)7.1 and Na(v)1.8 to Treat Pitt Hopkins Syndrome. Pharm Res 2019;36:137. doi: 10.1007/s11095-019-2671-y [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [19].Blatt J, Farag S, Corey SJ, Sarrimanolis Z, Muratov E, Fourches D, Tropsha A, Janzen WP. Expanding the scope of drug repurposing in pediatrics: The Children’s Pharmacy Collaborative. Drug Discov Today 2014;19:1696–1698. doi: 10.1016/j.drudis.2014.08.003 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [20].Appleby BS, Cummings JL. Discovering new treatments for Alzheimer’s disease by repurposing approved medications. Curr Top Med Chem 2013;13:2306–2327. [DOI] [PubMed] [Google Scholar]
  • [21].Hernandez HW, Soeung M, Zorn KM, Ashoura N, Mottin M, Andrade CH, Caffrey CR, De Siqueira-Neto JL, Ekins S. High Throughput and Computational Repurposing for Neglected Diseases Pharm Res 2018;36 doi: doi: 10.1007/s11095-018-2558-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [22].Chen B, Ding Y, Wild DJ. Assessing drug target association using semantic linked data. PLoS Comput Biol 2012;8:e1002574. doi: 10.1371/journal.pcbi.1002574 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [23].Lee JA, Shinn P, Jaken S, Oliver S, Willard FS, Heidler S, Peery RB, Oler J, Chu S, Southall N, Dexheimer TS, Smallwood J, Huang R, Guha R, Jadhav A, Cox K, Austin CP, Simeonov A, Sittampalam GS, Husain S, Franklin N, Wild DJ, Yang JJ, Sutherland JJ, Thomas CJ. Novel Phenotypic Outcomes Identified for a Public Collection of Approved Drugs from a Publicly Accessible Panel of Assays. PLoS One 2015;10:e0130796. doi: 10.1371/journal.pone.0130796 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [24].Williams AJ, Harland L, Groth P, Pettifer S, Chichester C, Willighagen EL, Evelo CT, Blomberg N, Ecker G, Goble C, Mons B. Open PHACTS: semantic interoperability for drug discovery. Drug Discov Today 2012;17:1188–1198. doi: 10.1016/j.drudis.2012.05.016 [DOI] [PubMed] [Google Scholar]
  • [25].Degtyarenko K, Hastings J, de Matos P, Ennis M. ChEBI: an open bioinformatics and cheminformatics resource. Curr Protoc Bioinformatics 2009;Chapter 14:14 19 11–14 19 20. doi: 10.1002/0471250953.bi1409s26 [DOI] [PubMed] [Google Scholar]
  • [26].Gaulton A, Hersey A, Nowotka M, Bento AP, Chambers J, Mendez D, Mutowo P, Atkinson F, Bellis LJ, Cibrian-Uhalte E, Davies M, Dedman N, Karlsson A, Magarinos MP, Overington JP, Papadatos G, Smit I, Leach AR. The ChEMBL database in 2017. Nucleic Acids Res 2017;45:D945–D954. doi: 10.1093/nar/gkw1074 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [27].Mons B, Ashburner M, Chichester C, van Mulligen E, Weeber M, den Dunnen J, van Ommen GJ, Musen M, Cockerill M, Hermjakob H, Mons A, Packer A, Pacheco R, Lewis S, Berkeley A, Melton W, Barris N, Wales J, Meijssen G, Moeller E, Roes PJ, Borner K, Bairoch A. Calling on a million minds for community annotation in WikiProteins. Genome Biol 2008;9:R89. doi: 10.1186/gb-2008-9-5-r89 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [28].Pinero J, Ramirez-Anguita JM, Sauch-Pitarch J, Ronzano F, Centeno E, Sanz F, Furlong LI. The DisGeNET knowledge platform for disease genomics: 2019 update. Nucleic Acids Res 2020;48:D845–D855. doi: 10.1093/nar/gkz1021 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [29].Wishart DS, Feunang YD, Guo AC, Lo EJ, Marcu A, Grant JR, Sajed T, Johnson D, Li C, Sayeeda Z, Assempour N, Iynkkaran I, Liu Y, Maciejewski A, Gale N, Wilson A, Chin L, Cummings R, Le D, Pon A, Knox C, Wilson M. DrugBank 5.0: a major update to the DrugBank database for 2018. Nucleic Acids Res 2018;46:D1074–D1082. doi: 10.1093/nar/gkx1037 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [30].Fang H, Su Z, Wang Y, Miller A, Liu Z, Howard PC, Tong W, Lin SM. Exploring the FDA adverse event reporting system to generate hypotheses for monitoring of disease characteristics. Clin Pharmacol Ther 2014;95:496–498. doi: 10.1038/clpt.2014.17 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [31].Harris MA, Clark J, Ireland A, Lomax J, Ashburner M, Foulger R, Eilbeck K,Lewis S, Marshall B, Mungall C, Richter J, Rubin GM, Blake JA, Bult C, Dolan M, Drabkin H, Eppig JT, Hill DP, Ni L, Ringwald M, Balakrishnan R, Cherry JM, Christie KR, Costanzo MC, Dwight SS, Engel S, Fisk DG, Hirschman JE, Hong EL, Nash RS, Sethuraman A, Theesfeld CL, Botstein D, Dolinski K, Feierbach B, Berardini T, Mundodi S, Rhee SY, Apweiler R, Barrell D, Camon E, Dimmer E, Lee V, Chisholm R, Gaudet P, Kibbe W, Kishore R, Schwarz EM, Sternberg P, Gwinn M, Hannick L, Wortman J, Berriman M, Wood V, de la Cruz N, Tonellato P, Jaiswal P, Seigfried T, White R, Gene Ontology C. The Gene Ontology (GO) database and informatics resource. Nucleic Acids Res 2004;32:D258–261. doi: 10.1093/nar/gkh036 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [32].Zahn-Zabal M, Michel PA, Gateau A, Nikitin F, Schaeffer M, Audot E, Gaudet P, Duek PD, Teixeira D, Rech de Laval V, Samarasinghe K, Bairoch A, Lane L. The neXtProt knowledgebase in 2020: data, tools and usability improvements. Nucleic Acids Res 2020;48:D328–D334. doi: 10.1093/nar/gkz995 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [33].UniProt Consortium T. UniProt: the universal protein knowledgebase. Nucleic Acids Res 2018;46:2699. doi: 10.1093/nar/gky092 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [34].Kerber RG, Ding Y, Foote B, Wild JW. Object oriented system and method having semantic substructures for machine learning. 2017. [Google Scholar]
  • [35].Himmelstein DS, Baranzini SE. Heterogeneous Network Edge Prediction: A Data Integration Approach to Prioritize Disease-Associated Genes. PLoS Comput Biol 2015;11:e1004259. doi: 10.1371/journal.pcbi.1004259 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [36].Bizon C, Cox S, Balhoff J, Kebede Y, Wang P, Morton K, Fecho K, Tropsha A. ROBOKOP KG and KGB: Integrated Knowledge Graphs from Federated Sources. J Chem Inf Model 2019;59:4968–4973. doi: 10.1021/acs.jcim.9b00683 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [37].Lane T, Russo DP, Zorn KM, Clark AM, Korotcov A, Tkachenko V, Reynolds RC, Perryman AL, Freundlich JS, Ekins S. Comparing and Validating Machine Learning Models for Mycobacterium tuberculosis Drug Discovery. Mol Pharm 2018;15:4346–4360. doi: 10.1021/acs.molpharmaceut.8b00083 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [38].Zorn KM, Lane TR, Russo DP, Clark AM, Makarov V, Ekins S. Multiple Machine Learning Comparisons of HIV Cell-based and Reverse Transcriptase Data Sets. Mol Pharm 2019;16:1620–1632. doi: 10.1021/acs.molpharmaceut.8b01297 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [39].Bird KE, Xander C, Murcia S, Schmalstig AA, Wang X, Emanuele MJ, Braunstein M, Bowers AA. Thiopeptides Induce Proteasome-Independent Activation of Cellular Mitophagy. ACS Chem Biol 2020;15:2164–2174. doi: 10.1021/acschembio.0c00364 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [40].Schmalstig AA, Zorn KM, Murcia S, Robinson A, Savina S, Komarova E, Makarov V, Braunstein M, Ekins S. Mycobacterium abscessus drug discovery using machine learning. Tuberculosis (Edinb) 2022;132:102168. doi: 10.1016/j.tube.2022.102168 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [41].Waddell SJ, Stabler RA, Laing K, Kremer L, Reynolds RC, Besra GS. The use of microarray analysis to determine the gene expression profiles of Mycobacterium tuberculosis in response to anti-bacterial compounds. Tuberculosis (Edinb) 2004;84:263–274. doi: 10.1016/j.tube.2003.12.005 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [42].Andrew S, “FastQC: A quality control tool for high throughput sequence data http://www.bioinformatics.babraham.ac.uk/projects/fastqc/ [Google Scholar]
  • [43].Kim D, Langmead B, Salzberg SL. HISAT: a fast spliced aligner with low memory requirements. Nat Methods 2015;12:357–360. doi: 10.1038/nmeth.3317 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [44].Love MI, Huber W, Anders S. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol 2014;15:550. doi: 10.1186/s13059-014-0550-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [45].Huang da W, Sherman BT, Lempicki RA. Systematic and integrative analysis of large gene lists using DAVID bioinformatics resources. Nat Protoc 2009;4:44–57. doi: 10.1038/nprot.2008.211 [DOI] [PubMed] [Google Scholar]
  • [46].Caspi R, Billington R, Ferrer L, Foerster H, Fulcher CA, Keseler IM, Kothari A, Krummenacker M, Latendresse M, Mueller LA, Ong Q, Paley S, Subhraveti P, Weaver DS, Karp PD. The MetaCyc database of metabolic pathways and enzymes and the BioCyc collection of pathway/genome databases. Nucleic Acids Res 2016;44:D471–480. doi: 10.1093/nar/gkv1164 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [47].Boshoff HI, Myers TG, Copp BR, McNeil MR, Wilson MA, Barry CE 3rd., The transcriptional responses of Mycobacterium tuberculosis to inhibitors of metabolism: novel insights into drug mechanisms of action. J Biol Chem 2004;279:40174–40184. doi: 10.1074/jbc.M406796200 [DOI] [PubMed] [Google Scholar]
  • [48].Rustad TR, Minch KJ, Ma S, Winkler JK, Hobbs S, Hickey M, Brabant W, Turkarslan S, Price ND, Baliga NS, Sherman DR. Mapping and manipulating the Mycobacterium tuberculosis transcriptome using a transcription factor overexpression-derived regulatory network. Genome Biol 2014;15:502. doi: 10.1186/PREACCEPT-1701638048134699 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [49].Fan X, Xu J, Files M, Cirillo JD, Endsley JJ, Zhou J, Endsley MA. Dual activity of niclosamide to suppress replication of integrated HIV-1 and Mycobacterium tuberculosis (Beijing). Tuberculosis (Edinb) 2019;116S:S28–S33. doi: 10.1016/j.tube.2019.04.008 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [50].Berube BJ, Castro L, Russell D, Ovechkina Y, Parish T. Novel Screen to Assess Bactericidal Activity of Compounds Against Non-replicating Mycobacterium abscessus. Front Microbiol 2018;9:2417. doi: 10.3389/fmicb.2018.02417 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [51].Williams JT, Haiderer ER, Coulson GB, Conner KN, Ellsworth E, Chen C, Alvarez-Cabrera N, Li W, Jackson M, Dick T, Abramovitch RB. Identification of New MmpL3 Inhibitors by Untargeted and Targeted Mutant Screens Defines MmpL3 Domains with Differential Resistance. Antimicrob Agents Chemother 2019;63 doi: 10.1128/AAC.00547-19 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [52].Baranyai Z, Kratky M, Vinsova J, Szabo N, Senoner Z, Horvati K, Stolarikova J, David S, Bosze S. Combating highly resistant emerging pathogen Mycobacterium abscessus and Mycobacterium tuberculosis with novel salicylanilide esters and carbamates. Eur J Med Chem 2015;101:692–704. doi: 10.1016/j.ejmech.2015.07.001 [DOI] [PubMed] [Google Scholar]
  • [53].Braunstein M, Hickey AJ, Ekins S. Why Wait? The Case for Treating Tuberculosis with Inhaled Drugs. Pharm Res 2019;36:166. doi: 10.1007/s11095-019-2704-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [54].Tudo G, Laing K, Mitchison DA, Butcher PD, Waddell SJ. Examining the basis of isoniazid tolerance in nonreplicating Mycobacterium tuberculosis using transcriptional profiling. Future Med Chem 2010;2:1371–1383. doi: 10.4155/fmc.10.219 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [55].de Carvalho LP, Darby CM, Rhee KY, Nathan C. Nitazoxanide Disrupts Membrane Potential and Intrabacterial pH Homeostasis of Mycobacterium tuberculosis. ACS Med Chem Lett 2011;2:849–854. doi: 10.1021/ml200157f [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [56].Lee SFK, Laughon BE, McHugh TD, Lipman M. New drugs to treat difficult tuberculous and nontuberculous mycobacterial pulmonary disease. Curr Opin Pulm Med 2019;25:271–280. doi: 10.1097/MCP.0000000000000570 [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

1
NIHMS1972153-supplement-1.docx (1.2MB, docx)
2
NIHMS1972153-supplement-2.xlsx (179.8KB, xlsx)

RESOURCES