Journal: Molecules

Article Title: Identification of Novel PPARγ Partial Agonists Based on Virtual Screening Strategy: In Silico and In Vitro Experimental Validation

doi: 10.3390/molecules29204881

Figure Lengend Snippet: Top four compounds from virtual screening of the Targetmol L6000 Natural Product Library via Autodock Vina.

Article Snippet: Thus, the present study’s aim is (a) virtual screening of the Targetmol L6000 Natural Product Library to search for potential hits that function as PPARγ partial agonists with the ability to block cyclin-dependent kinase 5 (CDK5)-mediated phosphorylation of PPARγ-Ser245 in the absence of classical transcription activity of PPARγ via the AF-2 helix “lock” mechanism; (b) further examination of the binding stability of the system utilizing molecular dynamics and molecular mechanics Poisson–Boltzmann surface area (MM-PBSA); (c) testing the PPARγ binding affinity and agonistic activity of the selected hit.

Techniques:

Journal: bioRxiv

Article Title: Sequence-based drug design using transformers

doi: 10.1101/2023.11.27.568880

Figure Lengend Snippet: (a , b) Scatter plot of the correlation between the experimental pIC 50 values and the binding parameters (p K d and pIC 50 values) using Ligand-Transformer for complexes in the EGFR LTC -290 dataset; all binding parameters are in molar units: (a) predicted p K d values without transfer learning (base model), and ( b ) predicted pIC 50 values post ten-fold validation with transfer learning. ( c ) Scatter plot representing the distribution of predicted distances between E762 and G857 of EGFR LTC in various ligand-binding states. The y-axis represents the predicted distance between the E762 Cβ and G857 Cα atoms, while the x-axis depicts the experimental pIC 50 values of the ligands. The data points are coloured based on their reported binding sites in the literatures: orthosteric (blue), allosteric (green), and dual (orange). The kernel density estimations (KDE) plot for the distribution of predicted distances are indicated adjacent to y coordinates. ( d) Comparative visualization of EGFR kinase domain crystal structures in the active (PDD 2ITX) and inactive (PDBID: 2GS7) states. The αC-helix is depicted as a ribbon structure in light blue and gray, representing the active and inactive states, respectively. The AMP-PNP ligand bound to PDB 2ITX is visualized as a stick-and-ball structure. Residues E762 and G857 of EGFR (corresponding to residues E738 and G833 in PDB 2GS7) are shown as sticks. Nitrogen, oxygen, and phosphorus atoms are colored blue, red, and orange, respectively. ( e) Line graph of the predicted binding modes of an orthosteric inhibitor (8r , colored in red) and an allosteric inhibitor (4i , colored in blue) with EGFR LTC . The line graph illustrates the predicted minimal distances between each residue and the ligand. The error bars represent the predicted confidence scores pMAE (see Algorithm 8 in Supplementary Iinformation). The bottom part shows sequence annotation of EGFR based on UniProt P00533. ATP binding sites are shown in red dots, active site is shown in green dot, and the region of αC-helix is shown in blue box. ( f) In vitro kinase inhibition assays for selected compounds (C1 to C11), depicted as a percentage of activity relative to a DMSO control. The graph presents the mean ± standard deviation, with each ligand tested in duplicate. The corresponding IC 50 values are presented for each compound.

Article Snippet: The library used for virtual screening for EGFR LTC inhibitors in this study contained a total of 9,090 compounds, all sourced from TargetMol.

Techniques: Binding Assay, Ligand Binding Assay, Residue, Sequencing, In Vitro, Inhibition, Activity Assay, Control, Standard Deviation