STRC Pharmacochaperone Virtual Screen Ranked Leads
A 29-compound literature-curated library was scored against the K1141 pocket pharmacophore through two orthogonal filters (2D property match + 3D shape fit). Salicylic acid leads (final 1.79), followed by nicotinic acid (1.68), cyclopropane-phenyl-COOH (1.68), indole-3-acetic acid (1.66), and naphthalene-2-carboxylic acid (1.64). All top-5 are fragment-sized (MW 123–175), rule-of-3-compliant, commercially cheap (<$200/g), and pose with zero protein clashes in the best ETKDG-embedded conformer. Diflunisal — the known transthyretin pharmacochaperone — ranks #8, providing external proof-of-concept for the “small-acid-rescues-misfold” mode.
Two-stage scoring
Phase 3B — 2D pharmacophore match (weight recipe)
Composite = 0.30 anion + 0.15 anion-arom distance (target 5.7 Å ± 1.5 Å) + 0.15 size-fit + 0.10 rule-of-3 + 0.10 Lipinski + 0.10 H-bond donor + 0.10 QED.
Phase 3C — 3D shape fit
For each candidate, 30 ETKDG conformers × 24 rotations about the (K1141-NZ → pocket centroid) axis. For each pose, ligand’s acidic oxygen is placed 2.8 Å from K1141-NZ. Score = (heavy atoms in 5.5 Å pocket sphere + atoms in 2–4 Å productive contact with protein) / (clashes < 2 Å + 1) / total heavy atoms.
Final = 3B composite × 3C shape score.
Ranked top-10
| # | Compound | 3B | Shape | Final | Fill | Clash | d(Ar-F1646) |
|---|---|---|---|---|---|---|---|
| 1 | salicylic acid | 0.896 | 2.000 | 1.792 | 10 | 0 | 4.96 Å |
| 2 | nicotinic acid | 0.843 | 2.000 | 1.685 | 9 | 0 | 6.41 Å |
| 3 | cyclopropane-1-carboxyphenyl | 0.840 | 2.000 | 1.679 | 12 | 0 | 4.17 Å |
| 4 | indole-3-acetic acid | 0.901 | 1.846 | 1.664 | 12 | 0 | 5.16 Å |
| 5 | naphthalene-2-carboxylic acid | 0.855 | 1.923 | 1.645 | 12 | 0 | 4.69 Å |
| 6 | 4-chlorobenzoate | 0.851 | 1.900 | 1.617 | 9 | 0 | 5.21 Å |
| 7 | tolfenamic acid | 0.867 | 1.765 | 1.530 | 17 | 0 | 8.56 Å |
| 8 | diflunisal | 0.864 | 1.389 | 1.200 | 8 | 0 | 8.50 Å |
| 9 | fenbufen | 0.823 | 1.368 | 1.127 | 12 | 0 | 9.67 Å |
| 10 | probenecid | 0.812 | 1.158 | 0.940 | 8 | 0 | 8.71 Å |
Fill = heavy atoms within 5.5 Å of pocket centroid after docking. Clash = atoms within 2 Å of any protein heavy atom.
Controls validate the scoring
Non-acidic controls rank at the bottom of Phase 3B and were excluded from Phase 3C:
- caffeine (3B=0.404)
- glucose (0.379)
- atenolol (0.414)
- ivacaftor, tezacaftor, elexacaftor (0.32-0.39 — these CFTR drugs fail the K1141 anion criterion; they use sulfonamide scaffolds that bind CFTR at a different interface)
This is expected — the library is biased toward carboxylate anchors by design. The pharmacochaperone class matching our pocket is small acid + aromatic ring, not the VX-809/445 family scaffold (which requires the specific NBD1/TMD binding environment of CFTR).
Wet-lab shortlist
Recommended for STRC E1659A rescue assay (HEK stable line, surface biotinylation or GFP-STRC surface expression):
- Salicylic acid — cheap, endogenous metabolite, BBB/labyrinth-permeable, no known toxicity at therapeutic doses
- Indole-3-acetic acid — endogenous auxin-class metabolite, studied for misfolding rescue in TTR/CFTR
- Cyclopropane-1-carboxyphenyl (minimal VX-809 scaffold) — pharmacochaperone-native pharmacophore
- Diflunisal — known TTR pharmacochaperone (FDA-approved); valuable positive comparator
- 4-chlorobenzoate or naphthalene-2-carboxylic acid — fragment library staples for SAR extension
Starting concentrations 10 µM → 1 mM (typical fragment screening range); expect hits at 100 µM–1 mM. Read-out: cell-surface STRC (flow cytometry), normalised to WT-STRC reference; rescue ≥30% is a hit.
Limitations
- Library is 29 compounds, not full 40k Enamine or 2k DSi-Poised. False negatives likely among VX-809 analogues with non-canonical acid equivalents (sulfonamide, acyl sulfonamide).
- 3C shape fit is an approximation of full docking — it assumes rigid protein, no induced fit, and uses only the WT pocket shape. AutoDock Vina or GNINA would rescore more rigorously.
- No solvent model, entropy, or ΔG estimation. Rankings are geometric + pharmacophore consistency only.
- Clinical dosing, pharmacokinetics (ear-drop delivery), and off-target effects not assessed.
Next steps (Phase 4 and beyond)
- Phase 4 — Vina/GNINA rescore when docking tools become available. Box: 18 × 18 × 18 Å at (7.7, −5.4, −41.5). Top 50 from DSi-Poised (2k) + Enamine fragments (40k).
- Phase 5 — MD validation: 50 ns × 3 replicates per top-5 lead; metrics = K1141-contact persistence >60%, loop Cα-RMSD <2 Å vs WT.
- Phase 6 — FEP+ for relative ΔΔG between top leads.
- Wet lab — Jeffrey Holt collaboration: HEK293-STRC surface rescue assay with salicylate/indole-acetate/cyclopropane-phenyl-COOH/diflunisal at 10 µM–1 mM.
Replication
cd ~/STRC/models
python3 pharmacochaperone_phase3a_pharmacophore.py # target pharmacophore
python3 pharmacochaperone_phase3b_virtual_screen.py # 2D screen
python3 pharmacochaperone_phase3c_shape_fit.py # 3D shape fitDeterministic (random seed 42). Adding compounds = add SMILES to LIBRARY list in 3B script.
Files / Models
~/STRC/models/pharmacochaperone_phase3b_virtual_screen.py— 2D screen~/STRC/models/pharmacochaperone_phase3b_virtual_screen.json— 29-compound ranked table~/STRC/models/pharmacochaperone_phase3c_shape_fit.py— 3D shape fit~/STRC/models/pharmacochaperone_phase3c_shape_fit.json— top-10 poses with K1141 + F1646 distances~/STRC/models/pharmacochaperone_phase3a_pharmacophore.json— pharmacophore input (Phase 3A)
Connections
[part-of]STRC Pharmacochaperone Virtual Screen E1659A — parent hypothesis[see-also]STRC Pharmacophore Model K1141 Pocket — pharmacophore target this screen matched against[see-also]STRC Pharmacochaperone K1141 Fragment Pocket — Phase 2B pocket geometry[applies]Misha — E1659A patient[about]Jeffrey Holt — intended wet-lab collaborator for rescue assay