Leveraging AI to explore structural contexts of post-translational modifications in drug binding

Abstract

Post-translational modifications (PTMs) play a crucial role in allowing cells to expand the functionality of their proteins and adaptively regulate their signaling pathways. Defects in PTMs have been linked to numerous developmental disorders and human diseases, including cancer, diabetes, heart, neurodegenerative and metabolic diseases. PTMs are important targets in drug discovery, as they can significantly influence various aspects of drug interactions including binding affinity. The structural consequences of PTMs, such as phosphorylation-induced conformational changes or their effects on ligand binding affinity, have historically been challenging to study on a large scale, primarily due to reliance on experimental methods. Recent advancements in computational power and artificial intelligence, particularly in deep learning algorithms and protein structure prediction tools like AlphaFold3, have opened new possibilities for exploring the structural context of interactions between PTMs and drugs. These AI-driven methods enable accurate modeling of protein structures including prediction of PTM-modified regions and simulation of ligand-binding dynamics on a large scale. In this work, we identified small molecule binding-associated PTMs that can influence drug binding across all human proteins listed as small molecule targets in the DrugDomain database, which we developed recently. 6,131 identified PTMs were mapped to structural domains from Evolutionary Classification of Protein Domains (ECOD) database.Scientific contribution: Using recent AI-based approaches for protein structure prediction (AlphaFold3, RoseTTAFold All-Atom, Chai-1), we generated 14,178 models of PTM-modified human proteins with docked ligands. Our results demonstrate that these methods can predict PTM effects on small molecule binding, but precise evaluation of their accuracy requires a much larger benchmarking set. We also found that phosphorylation of NADPH-Cytochrome P450 Reductase, observed in cervical and lung cancer, causes significant structural disruption in the binding pocket, potentially impairing protein function. All data and generated models are available from DrugDomain database v1.1 ( http://prodata.swmed.edu/DrugDomain/ ) and GitHub ( https://github.com/kirmedvedev/DrugDomain ). This resource is the first to our knowledge in offering structural context for small molecule binding-associated PTMs on a large scale.

Keywords: Domain; Drug discovery; Drugs; Post-translational modification; Protein structure; Protein-drug interaction; Small molecule.

Fig. 1

Distribution of small molecule binding-associated PTMs types in ECOD architecture groups. A Statistics for experimental PDB structures. B Statistics for AlphaFill models. The length of each vertical line represents the number of PTMs per ECOD A-group

Fig. 2

Average ligand RMSD for the PTM-modified and unmodified states in models generated by different approaches. A RoseTTAFold All-Atom (RFAA). B Chai-1. C KarmaDock. D AlphaFold3

Fig. 3

Structure of Tyrosine-protein phosphatase non-receptor type 11 (SHP-2) (PDB: 3O5X, shown in grey) in complex with inhibitor (II-B08, in magenta) and the modelled positions of this drug. A Unmodified state. B Zoomed-in view of modelled PTMs and experimental drug position. C PTM-modified state. Drug positions modelled by RFAA shown in green, Chai-1 in orange, KarmaDock in cyan, AlphaFold3 in slate. Experimental position of the drug is shown in magenta and thick sticks. The phosphorylated residue is shown in colors corresponding to the methods by which it was modeled

Fig. 4

Structure of Mineralocorticoid receptor (PDB: 3 VHV, shown in grey) in complex with inhibitor (PDB id: LD1, in magenta) and the modelled positions of this drug. A Unmodified state. B PTM-modified state. Drug positions modelled by RFAA shown in green, Chai-1 in orange, KarmaDock in cyan, AlphaFold3 in slate. Experimental position of the drug and Ser843 are shown in magenta and thick sticks. The phosphorylated residue is shown in colors corresponding to the methods by which it was modeled

Fig. 5

Examples of Class 2 phosphorylation site effects. A Unmodified state of inactive insulin-like growth factor 1 receptor (PDB: 3 NW7, shown in grey). B PTM-modified state of inactive insulin-like growth factor 1 receptor. Phosphorylated insulin receptor (PDB: 1IR3) shown in dark grey. Experimental position of the drug (PDB id: LGV) and Tyr1161 are shown in magenta and thick sticks. Activation loop of inactive receptor is shown in salmon, active and modeled in green. C Unmodified state of mitogen-activated protein kinase 1 (MAP2 K1) (PDB: 4LMN, shown in grey). D PTM-modified state of MAP2 K1. Experimental position of the drug (PDB id: EUI) and Ser222 are shown in magenta and thick sticks. Drug positions modelled by RFAA shown in green, Chai-1 in orange, KarmaDock in cyan, AlphaFold3 in slate. Chai-1 model structures shown in light yellow, AlphaFold3 models in light blue. The phosphorylated residue is shown in colors corresponding to the methods by which it was modeled

Fig. 6

Distribution of ligand RMSD values. A AlphaFold3 models vs experimental PDB structures. B AlphaFold3 models vs AlphaFill models. C RoseTTAFold All-Atom models vs experimental PDB structures. D RoseTTAFold All-Atom models vs AlphaFill models

Fig. 7

Phosphorylation of Tyr604 affects binding of NADP by NADPH-Cytochrome P450 Reductase. A Unmodified state of NADPH-Cytochrome P450 Reductase (PDB: 3QFR, shown in grey). B PTM-modified state of NADPH-Cytochrome P450 Reductase. Experimental position of the NADP and Tyr604 are shown in magenta and thick sticks. Drug positions modelled by RFAA shown in green, AlphaFold3 in purple, Chai-1 in orange, KarmaDock in cyan. The phosphorylated residue is shown in colors corresponding to the methods by which it was modeled. C Binding pocket of experimental structure of NADPH-Cytochrome P450 Reductase (PDB: 3QFR). D Binding pocket of unmodified state of NADPH-Cytochrome P450 Reductase modelled by AlphaFold3. E Binding pocket of PTM-modified state of NADPH-Cytochrome P450 Reductase modelled by AlphaFold3

Fig. 8

Example of the DrugDomain data webpage showing the list of small molecule binding-associated PTMs for Elongation factor 1-alpha 1 (P68104). A Table of small molecule binding-associated PTMs with links to generated models of modified structures. B AlphaFold3 model of modified structure of Elongation factor 1-alpha 1 with phosphorylated TYR29. C Chai-1 model of modified structure of Elongation factor 1-alpha 1 with phosphorylated TYR29. ECOD domains are shown in different colors