AI-ACCELERATED DRUG DISCOVERY

Novel inhibitors design for BRD4 and SIRT1

Successful targeting of two nuclear proteins

AI revolutionizes the drug discovery process

Background

  • Target #1: Human Silent Mating Type Information Regulation 2 Homolog 1 (SIRT1)some text
    • SIRT1 deacetylates histones and various other proteins involved in multiple cell signaling pathways.
  • Target #2: Human Bromodomain-containing protein 4 (BRD4)some text
    • A transcription factor that plays an important role in cancer development.
    • Member of a family of at least four proteins sharing the same structural motif.
Structures of SIRT1 and BRD4

Methodology

Virtual screening: The stock collection of 3.2M compounds was subject to virtual screening. 1000 top-ranked candidate compounds were selected. 100 compounds were used for experimental validation for both targets.

The two-stage virtual screening workflow implemented in the Receptor.AI drug discovery platform was used.

  1. In the first stage, the ligand-based and structure-based drug-target interaction AI models were used for initial screening. After that, the most relevant ADMET endpoints were evaluated, and the proteome-wide selectivity profiles were computed for the top compounds.

In the secondary screening stage, molecular docking with AI rescoring, which predicts the activity based on the docking poses, is performed for the molecules shortlisted during the initial screening.

Validation - SIRT1

100 best-ranked hit compounds were experimentally validated by our partners. 5 of them have shown high affinity to SIRT1. The compound with the highest affinity was selected for further activity evaluation.

  • 94 compounds have been tested in DSF against SIRT1.
  • Primary hits were subjected to counter-screening for selectivity against BRD4, CA, and Abl SH2. The acceptable values of quality control indicate the robustness of the screening process.
  • With the selection criteria '|ΔTm SIRT1| > 3*ΔTm RSD DMSO SIRT1', 4 compounds demonstrated a significant binding effect on SIRT1 Tm.

1 compound showed a selective effect on the target protein, following the selection criteria 
'|ΔTmDAbl SH2, CA, BRD4| > 3*TmD RSDDMSO Abl SH2, CA, BRD4'.

Validation - BRD4

100 best-ranked hit candidates were passed for experimental validation to our partners. 17 of them have shown affinity to BRD4. These compounds are currently subject to in-depth experimental validation.

  • 7 best compounds were picked for further activity evaluation. 3 compounds have shown notable biological activity:some text
    • RAIBD40009 470 nM,
    • RAIBD40062 730 nM,
    • RAIBD40067 4.8  uM.
  • 100 compounds have been tested in DSF against BRD4.
  • The acceptable values of quality control indicate the robustness of the performed screen.
  • Following the selection criteria ‘| dTmDBRD4-ligand | > 3*TmDRSD DMSO BRD4’, 17 compounds demonstrate significant binding effects on BRD4.
  • 8 of them belong to known classes of BRD4 inhibitors.

There are 9 entirely new BRD4 hits, belonging to 8 diverse chemical classes.

Discovered SIRT1 and BRD4 inhibitors

RAIST10058 (SIRT1)

This compound has shown outstanding nanomolar activity (IC50 = 90 nM) measured by a functional assay based on the release of the fluorophore probe from acetylated p53-AFC. This makes it de facto a lead compound.

RAIBD40009 (BRD4)

This compound, belonging to the previously unknown class of BRD4 inhibitors, has shown the best activity in experiments (470 nM).