Importance of permeability prediction for structure-based in silico drug design of novel anti-tuberculosis leads

Mycobacterium tuberculosis (M. tb), the causative agent of tuberculosis (TB), exhibits a concerning propensity to develop drug resistance. This necessitates the discovery of novel anti-tubercular drugs. However, effective drug candidates must possess the ability to permeate the unique cell wall of M. tb. This challenge arises due to the presence of mycolic acids, a hallmark component that impedes passage. Traditional wet-lab experimentation for permeability assessment of existing drug-like molecules is a laborious and resource-intensive process. Artificial intelligence (AI) and machine learning (ML) offer a powerful solution by enabling biologically informed prediction of ligands with cell wall permeability. This approach can significantly expedite the drug discovery pipeline by filtering millions of candidate molecules down to a select few hundred for further investigation including wet lab validation.

At Aganitha, we’re leveraging our Generative AI-powered Fragment Based Drug Discovery and Structure Based Drug Discovery platform to work with them on identifying novel anti-TB therapeutics as our collaboration with CSIR-CCMB. Starting with target elucidation to progress promising leads from initial hits to optimized candidates, it’s a true team effort where cutting-edge technology (deep tech) empowers scientific discovery (deep science).

Deep science and deep tech: Utilizing this virtuous cycle of innovation to accelerate drug discovery for TB

Deep science and deep tech, working together, are accelerating drug development, speeding up the processes of finding novel treatment options, screening of molecules and refining hits to leads. Out of many applications of AI in infectious diseases, some of these include: disease detection, disease prediction, modelling, diagnosis, treatment, discovery, repair, data visualisation, surgery, training, robotics, precision, treatment, risk stratification, and clinical management, etc.

Drug-resistant tuberculosis is a major public health threat to effectively control the disease. Genomic variants in M. tb drug targets or pro-drug activators, including single nucleotide polymorphisms (SNPs) and small insertions and deletions (indels), are responsible for drug resistance (DR). Recent advancements in NGS (e.g., Illumina, Oxford Nanopore (ONT)) to use these directly from sputum or DNA from limited M. tb culture (MGIT), in near real time, and at decreasing costs can be used to inform on “precision medicine” treatment decisions.

With the growing availability of large databases, increased accuracy of machine learning models, lower barriers to entry for researchers, and widespread access to public domain codes, computational methods are poised to significantly advance drug discovery for neglected tropical diseases. The use of artificial intelligence, especially machine learning, is transforming the modeling and prediction of biological activity and the development of novel medications for these diseases. Recent advancements in AI and ML for drug screening and design hold promise for guiding the development of novel drug candidates with targeted polypharmacological properties.

Predicting antibiotic resistance from next-generation sequence data:

Genotypic resistance prediction from M. tb DNA sequences offers an alternative to laboratory-based drug-susceptibility testing. User-friendly and accurate resistance prediction tools are needed to enable public health and clinical practitioners to rapidly diagnose resistance and inform treatment regimens. A lot of studies employ Machine learning, in particular to develop predictive models for identifying additional resistance, which can help guide antibiotic selection while awaiting susceptibility testing results.

For example: Genomics platform for TB (GenTB), a free and open web-based application to predict antibiotic resistance from next-generation sequence data. It offers three main features: a means for sharing, citing and crediting TB data and metadata, the prediction of resistance on genotype using a machine learning algorithm, and geographic resistance and mutation data mapping.

Enhancing Diagnosis:

Of the estimated 10 million new cases of TB worldwide each year, 3 million go undiagnosed and/or unreported. AI can analyze chest X-rays, a primary screening tool for TB, with high accuracy. This can help identify potential cases, especially in regions with limited access to trained radiologists.  Deep learning has been utilized for reliable TB Detection using chest X-Ray.

Accelerating drug discovery:

One of the major challenges in TB research is identifying new drug targets. Traditional methods often rely on prior knowledge or specific hypotheses, which can limit the scope of discovery. The development of new anti-tuberculosis drugs faces hurdles in efficiently transitioning from initial target identification to validation, particularly in target-based discovery. Similarly, scaling up studies to understand a drug’s mechanism of action (mode-of-action) needs improvement for seamless integration with high-throughput screening methods.

Omics technologies offer a powerful solution to these challenges in TB research. These approaches, like genomics, transcriptomics, and the emerging field of glycomics, allow researchers to explore the entire biological system of M. tb without relying on pre-defined assumptions. This agnostic approach unlocks a wider range of potential drug targets, as exemplified by the glycomics study which identified novel targets related to M. tb membrane synthesis accelerating drug discovery.

Artificial intelligence and machine learning (AI/ML) are playing a crucial role in developing novel therapies like small molecule drugs (SMOLs) and mRNA-based vaccines. This is coupled with the emergence of chatbots and SMS-based systems that empower patients by providing them with accessible tools to manage their TB journey. These interfaces answer basic questions about the disease, allow patients to input symptoms, connect them with healthcare workers, and provide information on referral networks and treatment protocols. This combination of cutting-edge therapeutics and patient-centric tools signifies a significant step forward in the fight against TB.

Background:

TB: An urgent unmet need

TB has been a persistent global killer, accounting for about 1.5 million fatalities each year. Further, the emergence of drug-resistant tuberculosis strains threatens to reverse decades of progress, potentially resulting in a catastrophic public health crisis. The need to act cannot be emphasized; millions of lives are at risk.

In 1882, meticulous efforts of Dr. Robert Koch led to the discovery of the casual Mycobacterium tuberculosis (M. tb), a tiny rod-shaped bacteria, responsible for causing TB a contagious and airborne disease. M. tb cunningly sustains infection within its human host by hiding and replicating inside macrophages. To thrive in the nutrient-limited environment of these immune cells, M. tb cleverly exploits specific metabolic pathways, siphoning host-derived nutrients.This evolutionary ability of the pathogen makes it challenging to target.

In 2022, TB was the second leading infectious disease killer worldwide, after COVID-19 (WHO 2023). It was also the leading killer of people with HIV and a major cause of deaths related to antimicrobial resistance. The resurgence of drug-resistant strains has further complicated the battle against this ancient disease, emphasizing the urgent need for additional and novel therapeutic strategies.

Modern medicine approaches have managed to combat TB to a large extent. Drug-sensitive tuberculosis is treated with a 6-month regimen of oral antibiotics pills (Isoniazid, Rifampicin, Pyrazinamide, and Ethambutol). Drug-resistant tuberculosis, including MDR and XDR-TB, requires extended therapy (18-24+ months) with a combination of injectables (e.g., amikacin) and second-line oral medicines (e.g., fluoroquinolones, bedaquiline).​ (NTCP) Developed in the early 20th century, the BCG (Bacillus Calmette-Guérin) vaccine is currently the only approved vaccine for TB, and it is efficacious only in children for  preventing the  disease.

The Growing threat of drug resistance

Drug-resistant TB (DR-TB) is a major impediment in the fight against the disease. Multidrug-resistant TB (MDR-TB) does not respond to at least isoniazid and rifampicin, the two most potent TB drugs. Extensively drug-resistant TB (XDR-TB) is resistant to these and additional second-line drugs, making treatment options limited, more toxic, and less effective. Estimated mortality rates for MDR-TB is around 40% and India has the second highest burden of MDR-TB in the whole world, approximately 26%. There are more than 400,000 cases of drug resistant TB each year (WHO 2023). The cure rate in MDR-TB is low (50-60%), therefore an early and accurate detection is essential for better treatment. The emergence of these resistant strains is primarily due to improper use of antibiotics, incomplete treatment regimens, and transmission from person to person. 

Patient Experience

MDR-TB and XDR-TB require longer treatment durations, often up to two years, with drugs that can have severe side effects. The treatment success rate for these resistant forms is significantly lower, highlighting the critical need for new and more effective anti-TB drugs.
Often patients upon feeling  better prematurely, usually discontinue the regimen, while stigma and isolation further discourage adherence. Efforts to identify more potent treatment modules which address these challenges are long due.

Funding, rather lack of it:

Despite the significant global burden of TB, the alarming lack of funding (falling short of US$ 1.0 billion in 2021 ) for TB research and treatment exacerbates the challenge, leaving millions at risk without adequate diagnostics, treatments, or preventive measures. Urgent action and investment are imperative to address this persistent and deadly disease.

Accelerating the process of drug design at Aganitha:

At Aganitha, we are addressing the challenges of TB using in silico tools to accelerate drug discovery. From structure-based in silico drug design for target elucidation to developing AI/ML based model for binary classification of ligands that are permeable through the cell wall of M. tb, we bring together deep science and deep tech. Stay tuned for our publication extensively detailing our research on using Generative AI and ML to speed up the drug discovery process for TB.