Designing molecular glue to stabilize membrane complex

Utilizing Receptor.AI’s Induced Proximity platform to design
small molecule stabilizer for the retromer complex

7

potent hits out of
387 candidates

250%

target complex concentration
increase

400%

higher potency compared to
best existing rivals

*Virtual Screening workflow

01/ Background

  • The goal is to design “molecular glue” to stabilize the membrane complex associated with autophagy and neurodegeneration.
  • The complex consists of 2 proteins with unknown binding sites.
  • Only 2 low-quality ligands are known (mediocre binding affinity).

02/ Methodology

  • Novel binding pocket at the protein-protein interface identified by Receptor.AI's proprietary pocket detection AI model.
  • 8M stock library subjected to virtual screening.
  • 1000 top-ranked candidate compounds were selected.
    387 used for experimental validation.
  • Experimental validation was conducted on the HEK 293 cell line to assess modulation of target expression (Western blot).
  • The activity of identified hit compounds, along with the absence of cytotoxicity, was substantiated through dose-response analysis.

03/ Results

  • 7 hit compounds with over 150% increase in target concentration identified at the 50 μM compound concentration.
  • 5 hits with over 125% increase at the 25 μM concentration.
  • 4 active scaffolds were chosen for future series expansion.
*Assay results for the two rounds with different
compound concentration. Green is for rivals
*Dose-Response relationship for Compound 2.
Treatment time 24 and 48 hours.

04/ Best Compound

  • The best hit (2) shows over 250% target concentration increase.
  • 400% higher potency compared to the best existing rivals.
  • Exhibits a nearly sub-micromolar EC50.