TL;DR
This study uses X-ray crystallography data to identify "multiclass ligands" (MCLs) that bind to proteins from different classes. They found that these promiscuous compounds are surprisingly flexible and hydrophilic, and importantly, they don't adopt a single "canonical" binding shape, but rather exhibit conformational versatility. This work sheds light on the molecular basis of polypharmacology, which is crucial for both understanding fundamental biology and designing more effective drugs.
What Problem Does This Paper Address?
In drug discovery and basic research, compounds that can interact with multiple protein targets (multitarget or promiscuous compounds) are of great interest. These compounds can lead to polypharmacological effects, potentially increasing therapeutic efficacy but also contributing to unwanted side effects. Understanding why and how these molecules achieve multitarget activity at a molecular level is a significant challenge. While previous studies have hinted at promiscuity, a systematic, structure-based analysis comparing how these ligands bind to diverse protein classes has been lacking. This paper aims to fill that gap by identifying and characterizing such ligands using structural data.
Methods
The researchers employed a large-scale computational approach combining several data sources:
- X-ray Crystallography Data: They extracted a vast number of protein-ligand complex structures from the Protein Data Bank (PDB).
- Ligand and Protein Classification: Proteins were classified using Gene Ontology (GO) terms to group functionally related families, allowing for the identification of ligands binding to proteins from distinct classes. Ligands were filtered for molecular weight and a minimum potency threshold (≥ 10 µM) to ensure biologically relevant interactions.
- Binding Mode Analysis: For identified multiclass ligands (MCLs), their bound conformations were analyzed. Pairwise Tanimoto shape similarity was calculated using the ROCS software to quantify how similar the ligand's shape is when bound to different proteins.
- Molecular Property Analysis: Standard molecular descriptors like hydrogen bond donors/acceptors, rotatable bonds (as a measure of flexibility), and LogP (hydrophobicity) were calculated for the identified ligands.
- Protein Structure Comparison: The structural similarity of the target proteins was assessed using RMSD of alpha-carbon traces.
Datasets
The primary dataset used was the Protein Data Bank (PDB), from which X-ray structures of protein-ligand complexes were extracted. Activity data for ligands was sourced from ChEMBL and PDBbind. The study identified 515 ligands with multitarget activity, including 70 "multiclass ligands" (MCLs) that bind to proteins from different classes.
Key Insights and Ideas
- Conformational Versatility is Key: The most striking finding is that MCLs don't rely on a single, conserved binding shape to interact with diverse targets. Instead, they exhibit significant flexibility and can adopt different conformations, allowing them to fit into various protein binding sites.
- Surprising Hydrophilicity: Contrary to a common assumption that promiscuous compounds might be more hydrophobic to interact with diverse environments, MCLs were found to be predominantly hydrophilic.
- Flexibility Matters: MCLs tend to be more flexible (more rotatable bonds) than single-class ligands, which likely contributes to their ability to adapt their shape for binding to different targets.
- No "Canonical" Promiscuous Shape: The study debunked the idea of a universal binding shape that defines promiscuity. The diversity in binding modes is a hallmark of MCLs.
- Implications for Drug Design: Understanding this conformational adaptability and physicochemical properties of MCLs provides valuable insights for designing new drugs with pre-defined multitarget activity, potentially leading to more effective therapies with fewer side effects.
Key Results
- 70 Multiclass Ligands (MCLs) Identified: The analysis identified 70 MCLs that bind to proteins from at least two different GO classes, out of 515 total multitarget ligands.
- Wide Range of Shape Similarity: Pairwise comparisons of MCL binding modes showed a broad distribution of shape similarity values, ranging from very low (~0.3) to nearly identical (~1.0).
- MCLs More Flexible and Hydrophilic: MCLs had a slightly higher number of hydrogen bond donors and acceptors and were significantly more hydrophilic (lower median LogP) than single-class ligands (SCLs).
- SCLs Bind More Consistently: In contrast to MCLs, SCLs showed a stronger tendency to bind their targets (within the same class) with higher shape similarity, as expected.
Limitations and Future Work
- Limited Structural Data: While the study leveraged a large dataset, the number of available X-ray structures for protein-ligand complexes is still limited, especially for certain protein classes.
- Potency Threshold: The use of a potency threshold (≥ 10 µM) might exclude some weakly binding but still relevant promiscuous compounds.
- Focus on Binding Shape: The analysis primarily focused on ligand shape similarity. Future work could explore other aspects of ligand-target interactions, such as specific pharmacophore features or electrostatic complementarity.
- Experimental Validation: While the study provides strong computational evidence, experimental validation of the identified MCLs and their binding modes across diverse targets would be beneficial.
- Designing for Promiscuity: The authors suggest that their findings can guide the design of new multitarget ligands by providing templates for compounds that exhibit conformational versatility. Further research could focus on developing computational tools to predict or design such adaptable molecules.
Summary updated