STRC Research

STRC Pharmacochaperone Phase 4c WT Decoy

7 min read

STRC Pharmacochaperone Phase 4c — all 5 leads prefer WT over E1659A by 0.40–0.55 kcal/mol; competition-mode sub-mechanism flagged, fold-stabilizer sub-mechanism still open; pivot to Phase 4f interface-rescue MM-GBSA as the definitive test

Paired rigid-receptor Vina docking of the 9-compound roster against WT STRC (job4-wildtype.cif) and the E1659A mutant (job3-mutant.cif). Expectation was that rescue ligands would prefer E1659A (K1141 free to accept a carboxylate) over WT (K1141 already salt-bridged to E1659). All 5 leads did the opposite. They prefer WT by a mean of 0.46 kcal/mol, while the positive control diflunisal is flat (−0.04) and the polar negatives are near-flat (−0.14 average). The leads are the only class with a systematic WT preference — so the signal is specific, not a global pocket-geometry artifact. This falsifies the aspect (a) “carboxylate replaces E1659” sub-mechanism as written. The aspect (b) “loop-capping fold-stabilizer” sub-mechanism remains untested; rigid-receptor Vina cannot see induced-fit rescue. Hypothesis stays S-tier pending Phase 4f / Phase 5.

Method

  • Roster: same 9 compounds as Phase 4b (5 Phase 3C leads + diflunisal positive + 3 polar negatives). Ligand PDBQTs reused from Phase 4b prep (identical conformer, protonation, and Meeko output).
  • Targets: WT full-length job4-wildtype.cif (chain A, offset 0) vs E1659A full-length job3-mutant.cif (chain A, offset 0).
  • Box: 18 × 18 × 18 Å at K1141 Cα + 3 Å toward loop-1642-1651 centroid, derived per-CIF in its own AF3 frame so the pocket is targeted identically despite different absolute coordinates.
    • WT box centre: (4.53, −7.94, −42.35) Å
    • E1659A box centre: (15.78, −44.66, 16.87) Å
  • Vina: exhaustiveness 32, 9 modes, 8 CPU; best-mode affinity reported.

Results

CompoundRoleΔG_WTΔG_E1659AΔ(WT − mut)
salicylic-acidlead−4.64−4.21−0.43
nicotinic-acidlead−4.37−3.97−0.40
cyclopropane-phenyl-COOHlead−5.40−4.96−0.44
indole-3-acetic-acidlead−5.56−5.10−0.46
naphthalene-2-COOHlead−6.01−5.46−0.55
diflunisalpositive−6.12−6.08−0.04
glucosenegative−4.03−4.12+0.09
ureanegative−3.29−2.96−0.33
acetamidenegative−3.09−2.93−0.16
Class meansleads −0.455, positive −0.043, negatives −0.138

Negative ΔΔG = ligand prefers WT. Positive ΔΔG = ligand prefers mutant (the hypothesized direction).

Gate: ≥3 of 5 leads with ΔΔG ≥ +1 kcal/mol. Result: 0/5. Verdict FAIL.

Interpretation

Two competing readings, neither of which is dispositive on its own:

Reading 1 — anti-hypothesis (parsimonious)

The Phase 3A pharmacophore was defined on WT STRC (Phase 2B pocket scan used job4-wildtype.cif). In WT, K1141’s salt-bridge partner E1659 is present, creating a pocket framework that includes K1141 flanked by a native carboxylate. A carboxylate ligand can bind in a geometry that exploits this framework — e.g. a bidentate clamp where the ligand’s COOH and E1659’s COOH both contact K1141 from different directions. In E1659A, E1659’s carboxylate is gone, the framework collapses, and the ligand loses the auxiliary anchor. The ligand prefers WT because it likes WT — the pocket in WT is deeper, more ordered, and offers a second H-bond partner that Alanine cannot provide.

If this reading is correct, the aspect (a) mechanism in STRC Pharmacochaperone Virtual Screen E1659A — “A ligand there (a) replaces the lost E1659 carboxylate contribution via K1141 salt-bridge / H-bond” — is falsified. The ligand is not replacing E1659; it’s enjoying E1659.

Reading 2 — rigid-receptor blind spot

Vina is rigid-receptor — it does not sample sidechain rotamers, loop conformations, or induced fit. The parent hypothesis specifies a second mechanism aspect: “(b) caps the loop against its LRR-face anchor in WT-like geometry”. This is the VX-809 / lumacaftor class — a tertiary-fold stabilizer whose therapeutic value is NOT that it out-scores on a static receptor but that it shifts the conformational ensemble toward the WT-like state. The value appears in MD (ligand keeps the loop in WT-like position; loop spends more frames near WT coordinates), NOT in static docking.

Meanwhile, Phase 4a measured the E1659A pocket as 70 ų vs WT’s 99 ų — the mutant pocket has collapsed 30% (the loop shifts 3 Å; K1141 moves 3.4 Å). On rigid Vina, a smaller pocket just scores worse for any ligand purely by reducing buried SASA. That would give a uniform WT preference across all compounds.

But the leads’ WT preference is 3–10× larger than the positive control’s and the negatives’. Uniform pocket-geometry penalty cannot explain a class-specific signal. So this reading only partially rescues the hypothesis — the 30% pocket collapse explains some of the ΔΔG, but not the lead-specificity.

The most defensible reading

Both effects are present. Aspect (a) “carboxylate replaces E1659” looks wrong as written — if it were right, leads would benefit from mutation, not penalty. Aspect (b) “loop-capping fold-stabilizer” is untested — the therapeutically dominant mechanism of VX-809-class pharmacochaperones only shows up in MD, not rigid docking. The pocket-geometry penalty accounts for some of the WT preference but not all.

Decision

QuestionVerdict
Is Phase 4c a PASS?No.
Is the hypothesis falsified?No — the fold-stabilizer mechanism is untested.
Has the competition-mode sub-mechanism been weakened?Yes, materially.
Should we run Phase 4d (K1141A decoy)?Skip. Same rigid-receptor limitation — testing a mechanism class that just flagged.
Should we run Phase 4f (interface-rescue MM-GBSA)?Yes. This measures the therapeutic claim directly: does ligand binding improve the TMEM145 interface energy toward WT levels, regardless of which pocket geometry is preferred.
Should we run Phase 5 (GROMACS MD)?Yes, after Phase 4f. Only MD can show induced-fit rescue. Even a weak ligand that holds the loop in WT position for 60% of frames is therapeutically valuable.
Does Pharmacochaperone stay S-tier?Yes, but mechanism axis flagged. Next two gates (4f, 5) decide.

Tier math

Current: mechanism 4, delivery 4, Misha-fit 4 → tier S (min 4).

After Phase 4c: the aspect (a) sub-mechanism has a real negative signal. If we down-weight the mechanism axis from 4 → 3, tier ≈ min(3,4,4) = 3 → A-tier. But we haven’t falsified the full mechanism — only one of two specified aspects. Policy call: hold at S until Phase 4f and Phase 5 are complete. If Phase 4f fails (no interface-energy recovery even with induced-fit allowance), demote. If Phase 4f passes, Phase 4c’s competition-mode failure is irrelevant (the clinical claim was always about interface rescue, not pocket competition).

Files / Models

  • ~/STRC/models/pharmacochaperone_phase4c_wt_decoy.py — driver (reuses Phase 4b receptor-prep / box / ligand-prep helpers)
  • ~/STRC/models/pharmacochaperone_phase4c_wt_decoy.json — per-compound ΔG_WT, ΔG_E1659A, paired Δ(WT−mut), class-mean summary
  • ~/STRC/models/docking_runs/4c/{wt_full,e1659a_mutant}_chainA.pdbqt — prepared receptors for each target
  • ~/STRC/models/docking_runs/4c/poses/*.pdbqt — 18 best-mode poses (9 compounds × 2 targets)

Ranking delta

  • STRC Pharmacochaperone Virtual Screen E1659A: no tier change — stays S. Mechanism axis flagged (not downgraded yet): aspect (a) “carboxylate replaces E1659” has a real negative signal on rigid-receptor Vina; aspect (b) “loop-capping fold-stabilizer” untested and is the therapeutically dominant claim. Evidence depth +1 with a negative sign. Next step column updated from “run Phase 4c WT decoy dock” → “run Phase 4f interface-rescue MM-GBSA on Ultra-Mini × TMEM145 (the therapeutic claim directly) — skip Phase 4d, it shares the rigid-receptor limitation that just flagged”.
  • All other S/A/B/C hypotheses: no change. No ripple to Mini-STRC or Strategy B LNP — those hypotheses don’t share a pharmacochaperone mechanism.

Connections

Graph View

Backlinks