Structure-guided, target-based drug discovery – exploiting genome information from HIV to mycobacterial infections

Sony Malhotra, S.e. Thomas, Bernardo Ochoa Montaño, Tom L. BlundellPostępy 2016 · DOI

TL;DR

This paper reviews how structure-guided drug discovery, particularly fragment-based drug design (FBDD), has evolved from early successes with HIV protease inhibitors to its current application in developing antimicrobials against challenging pathogens like Mycobacterium tuberculosis and Mycobacterium abscessus. It highlights the power of combining structural biology with genomic information to identify and optimize drug candidates for complex bacterial targets.

What Problem Does This Paper Address?

The paper addresses the persistent challenge of discovering new antimicrobial drugs, especially against drug-resistant bacteria like mycobacteria. Traditional drug discovery methods often rely on large screening libraries, which can be inefficient and costly. The authors highlight the need for more targeted and efficient approaches, leveraging our growing understanding of pathogen genomes and protein structures. They trace the lineage of structure-guided drug design, starting with the groundbreaking work on HIV protease inhibitors, and then pivot to its application in tackling mycobacterial infections, which are particularly difficult due to the unique cell wall structure and large genomes of these bacteria. The paper aims to demonstrate how integrating genomic data with structural insights can accelerate the identification and optimization of novel drug leads.

Methods

The paper discusses several key methodologies:

  • Structure-Guided Drug Discovery: This overarching approach uses the 3D structure of a target protein to identify potential binding sites for drug molecules. This can involve:
  • Protein Crystallography: The primary experimental technique for determining high-resolution protein structures.
  • Homology/Comparative Modeling: Building structural models of proteins based on the known structures of related proteins, especially useful when experimental structures are unavailable.
  • Virtual Screening: Computational methods to screen large libraries of small molecules in silico for potential binding to a target protein. Techniques like GRID, LUDI, and SUPERSTAR are mentioned for identifying "binding hotspots."
  • Fragment-Based Drug Discovery (FBDD): This approach screens libraries of small molecules (fragments, typically <300 Da) that bind with low affinity but make well-defined interactions. These fragments are then "grown" or "linked" to create higher-affinity lead compounds. This method is presented as more efficient than traditional high-throughput screening (HTS) due to smaller library sizes and improved ligand efficiency.
  • Genomic Analysis and Annotation: The paper emphasizes the importance of whole-genome sequencing and comparative genomics to identify potential drug targets within pathogens like Mycobacterium tuberculosis (MTb) and Mycobacterium abscessus (MAb). Tools and databases like UniProt, ProteinOrtho, and Tuberculist are mentioned for analyzing gene sequences, identifying orthologs, and annotating protein functions.
  • Computational Tools for Mutation Analysis: The paper mentions the use of tools like SDM and mCSM to predict the effects of mutations on protein stability, which is relevant for understanding drug resistance mechanisms.

Datasets

The paper does not explicitly list datasets used for training or evaluation in the context of a specific computational model or benchmark. However, it heavily relies on publicly available biological databases and resources:

  • Protein Data Bank (PDB): Used for obtaining experimental protein structures.
  • UniProt: A comprehensive source of protein sequence and functional information.
  • NCBI Genomes: Used for accessing whole-genome sequences of mycobacteria.
  • Tuberculist: A resource for MTb genome annotation.
  • CHOPIN database: An in-house database developed by the authors, storing homology models for MTb proteins.
  • TB Drug Resistance Mutation Database (TBDReaMDB) and MUBII-TB-DB: Databases for drug resistance mutations.
  • TDR Targets database: A resource for neglected disease pathogens.

The paper discusses the analysis of genomic data from MTb and MAb, comparing orthologous gene clusters and identifying "MTb-only" and "MAb-only" proteins.

Key Insights and Ideas

  • Evolution of Drug Discovery: The paper beautifully illustrates the progression of drug discovery from early, structure-driven approaches (like HIV protease inhibitors) to more sophisticated, genome-informed FBDD strategies.
  • FBDD's Efficiency: FBDD is presented as a more rational and efficient way to explore chemical space compared to traditional HTS, especially for challenging targets where large, diverse libraries are less effective.
  • Genomics as a Foundation: The increasing availability of pathogen genomes is revolutionizing target identification. By comparing genomes and annotating gene products, researchers can pinpoint essential proteins and pathways for intervention.
  • Mycobacterial Challenges: The unique features of mycobacteria, such as their thick cell walls and large genomes, necessitate specialized approaches to drug discovery, often requiring molecules with different properties than those optimized for human targets.
  • Structure-Guided Annotation: Combining structural modeling with genomic data is crucial for annotating the function of unknown proteins, which is a significant challenge for pathogens like mycobacteria.
  • Fragment Merging/Linking: The paper highlights successful examples of using fragment merging and linking strategies to develop potent inhibitors, demonstrating the power of FBDD in creating complex molecules from simple starting points.

Key Results

The paper is primarily a review and perspective piece, rather than a report of specific experimental results with quantitative metrics. However, it highlights:

  • Successful application of FBDD: The paper points to the successful application of FBDD in developing antimicrobials, citing examples like inhibitors for Mycobacterium tuberculosis EthR (boosting ethionamide efficacy) and phosphopantetheine adenylyltransferase (PPAT) in Mycobacterium abscessus.
  • Identification of orthologous clusters: The analysis of MTb and MAb genomes revealed a significant number of orthologous gene clusters, providing insights into conserved functions and potential drug targets.
  • Development of computational tools: The authors' group has developed and utilized tools like CHOPIN for homology modeling and predicting mutation effects, contributing to the structural annotation of mycobacterial proteomes.
  • Improved ligand efficiency: FBDD, by design, leads to improved ligand efficiency, meaning smaller molecules can achieve higher binding affinity and potency.

Limitations and Future Work

  • Annotation Gaps: Despite advances, a significant portion of mycobacterial proteomes remains unannotated, highlighting the need for continued efforts in structural and functional characterization.
  • Experimental Validation: While computational methods and structural modeling are powerful, experimental validation (e.g., crystallography, biophysical assays) remains critical for confirming hits and optimizing leads.
  • Drug Resistance Complexity: The paper touches upon drug resistance but acknowledges the complexity of mechanisms, suggesting further work is needed to address it comprehensively.
  • Expanding Resources: The authors express a commitment to expanding their CHOPIN database to include more MTb strains, protein-protein interactions, and oligomeric models, as well as adopting more sensitive modeling pipelines.
  • MAb Proteome Study: The paper indicates ongoing work to apply the developed MTb annotation pipelines to the less-studied MAb proteome, aiming to improve understanding of its gene products and functions.
  • Target Identification and Validation: The paper notes that while genetic manipulation experiments for target validation can be challenging, homology modeling and database mining offer valuable alternatives.

Summary updated

Abstract

The use of protein crystallography in structure-guided drug discovery allows identification of potential inhibitor-binding sites and optimisation of interactions of hits and lead compounds with a target protein. An early example of this approach was the use of the structure of HIV protease in designing AIDS antivirals. More recently, use of structure-guided design with fragment-based drug discovery, which reduces the size of screening libraries by decreasing complexity, has improved ligand efficiency in drug design. Here, we discuss the use of structure-guided target identification and lead optimisation using fragment-based approaches in the development of new antimicrobials for mycobacterial infections.
BibTeX
@inproceedings{malhotra2016structure,
  author = {Sony Malhotra and S.E. Thomas and Bernardo Ochoa Montaño and Tom L. Blundell},
  title = {{Structure-guided, target-based drug discovery – exploiting genome information from HIV to mycobacterial infections}},
  booktitle = {Postępy},
  year = {2016},
  doi = {10.18388/pb.2016\_25},
}
Figure