STRC h01 Phase 4h Tafamidis-Playbook Bioisostere Library 2026-04-23
Purpose. Seed library for Phase 3c v4, independent of v3b GREEN/RED outcome. Translates STRC h01 Fenamic Scaffold Tox Audit 2026-04-23 + Phase 3c v2 top hits + tafamidis precedent into a 30-compound prioritized target list designed to (a) kill COX pharmacophore, (b) kill cochlear ion-channel panel off-targets, (c) preserve K1141 anthranilic-N/carboxylate-O pharmacophore. Ranked by synthesis tractability and predicted Kd improvement.
Design constraints (from Fenamic Tox Audit)
- Kill COX-1/-2 — swap anthranilic 2-NHAr + COOH planar pharmacophore for tafamidis-style bicyclic heteroaromatic (benzoxazole / benzothiazole / benzimidazole) or ring-constrained bioisostere. COX essentiality depends on carboxylate binding Arg120/Tyr385 in the COX pocket; bicyclic fusion breaks geometry.
- Kill ion-channel panel (TRPM4 2.8 µM, Cx50 3 µM, BK 25 µM, KCNQ-family, TMEM16A 12 µM) — replace lipophilic distal aryl halides (3-CF₃, 2,6-diCl, 3-Cl-2-Me) with polar small groups (CN, OMe, F, ethers) to drop logP < 3 and kill channel affinity. Tafamidis logP = 4.24; aim for ≤ 3.5 in first generation.
- Preserve K1141 pocket pharmacophore — Phase 3c v2 docking poses show: (a) carboxylate oxygen / H-bond acceptor engages K1141 ε-NH₃⁺ salt bridge, (b) anthranilic N-H donates H-bond to backbone C=O of residue adjacent to K1141, (c) biphenyl distal ring π-stacks with Phe/Tyr in pocket wall. Any bioisostere must preserve (a) + (b); (c) can vary.
- Pediatric-acceptable ADMET — logP ≤ 3.5, PSA 60-90 Ų, no PAINS, no aniline toxicophore.
Pharmacophore extracted from Phase 3c v2 top poses
Distal ring (π-stack) H-bond donor
│ │
N─── sp2 linker ────[bicyclic heteroaromatic]───C=O / anion
H │ │
"proximal" salt-bridge K1141
Required atoms: distal aromatic (≥1 ring), sp² linker (NH / O / N-Me / C=N), bicyclic core OR substituted benzoic-acid surrogate, H-bond acceptor (COO⁻ / tetrazole / N-acyl-sulfonamide / benzoxazole-N).
Library design
Combinatorial logic: 5 core scaffolds × 6 distal-ring polar-sub patterns × 2 acid bioisosteres = 60 theoretical; prune to 30 prioritized by synthesis complexity + predicted binding.
Core scaffolds (rows)
| # | Core | Precedent | COX-breaker evidence (PRE-6c, predicted) | COX dock 2026-04-24 |
|---|---|---|---|---|
| A | Benzoxazole-5-carboxylate (tafamidis) | FDA Vyndaqel 2019 (TTR) | Tafamidis no COX liability ⚠ | FALSIFIED — tafamidis Kd 1.5 µM COX-1/-2; benzox-3CN Kd 1.0/1.8; benzox-3OMe Kd 0.97/0.99 |
| B | Benzothiazole-6-carboxylate | Tafamidis bioisostere Bulawa 2012 | Hybrid thia-analog ⚠ | Untested directly; same Arg120 vector → expect same liability |
| C | Benzimidazole-5-carboxylate | Tolterodine scaffolds | Basic center disfavors COX ⚠ | Untested; basic N may protonate at COX pH and lose Arg120 — only core with potential built-in deselect |
| D | Diflunisal-core + iodine on distal ring | Iododiflunisal (chaperone probe) | Reduced COX vs diflunisal | Iododiflunisal Kd 8.5 µM COX-1, 3.0 µM COX-2 — better than fenamates but still pediatric-problematic |
| E | Indole-3-carboxylate (rigidified) | Indomethacin-like but regio-closed | Non-planar, COX↓ vs indo ⚠ | Untested; indomethacin itself is potent COX inhibitor — claim is dubious |
⚠ = pre-6c design predictions that did not survive Phase 6c calibration. Treat all “X disfavors COX” claims as hypotheses, not evidence, until docked. See STRC h01 Phase 6c COX Selectivity Panel 2026-04-24 and the new “COX-deselect medchem strategies (revision)” section below.
Distal-ring polar substitutions (columns)
| # | Pattern | Polar group | Logic |
|---|---|---|---|
| 1 | 3-CN | -CN | Kill TRPM4 (halide→nitrile); logP ↓ 0.8 |
| 2 | 3-OMe | -OCH₃ | Kill Cx50/BK (lipophilic halide→polar ether); logP ↓ 1.1 |
| 3 | 3-OH + 4-F | -OH/-F | H-bond donor for pocket; kill KCNQ-panel |
| 4 | 3-CF₃ (tafamidis-like) | -CF₃ | RETAIN only if critical; logP ↑ risk |
| 5 | 3-NHSO₂Me | -NHSO₂Me | Sulfonamide donor; kills TMEM16A 12µM |
| 6 | 3-F only | -F | Minimal perturbation; acts as negative control |
Acid bioisosteres (depth)
| # | Bioisostere | Pros | Cons |
|---|---|---|---|
| α | Carboxylate (retain) | Preserved pharmacophore | Retains COX, UGT liability |
| β | Acyl-sulfonamide (-CONHSO₂Me) | pKa ~4.5, mimics COOH | Slightly lower CNS, no COX engagement |
Total: 5 × 6 × 2 = 60 candidates; prune to 30 prioritized below.
Prioritized 30-compound table
Ranked by estimated ΔG improvement vs Phase 3c v2 niflumic -6.18 kcal/mol + synthesis tractability (1=commercial, 5=multistep novel).
| # | Core-substitution-acid | SMILES (schematic) | SynComp | ΔG est. | Tox flag | Rationale |
|---|---|---|---|---|---|---|
| 1 | A1α — benzoxazole-3CN-COOH | Oc1ccc2oc(-c3cccc(C#N)c3)nc2c1C(O)=O | 2 | −7.3 | LOW | Tafamidis + CN replaces CF₃ |
| 2 | A2α — benzoxazole-3OMe-COOH | Oc1ccc2oc(-c3cccc(OC)c3)nc2c1C(O)=O | 2 | −7.2 | LOW | Methoxy kills Cx50 |
| 3 | A1β — benzoxazole-3CN-NHSO₂Me | ibid. with -CONHSO₂Me | 3 | −7.5 | LOW | Acyl-sulfonamide for no-COX |
| 4 | A5α — benzoxazole-3NHSO₂Me-COOH | ibid. with -NHSO₂Me distal | 3 | −7.4 | LOW | Dual polar kill |
| 5 | Tafamidis parent (2,6-diCl-benzoxazole-5-COOH) | FDA molecule | 1 | −7.5 (ref) | MED | Benchmark positive control |
| 6 | B1α — benzothiazole-3CN-COOH | Oc1ccc2sc(-c3cccc(C#N)c3)nc2c1C(O)=O | 3 | −7.2 | LOW | Thia-analog of #1 |
| 7 | B2α — benzothiazole-3OMe-COOH | ibid. with OMe | 3 | −7.1 | LOW | Thia-analog of #2 |
| 8 | C1α — benzimidazole-3CN-COOH | Oc1ccc2nc([nH]c2c1C(O)=O)-c3cccc(C#N)c3 | 3 | −7.0 | MED | Basic N may disrupt pocket |
| 9 | A4α — benzoxazole-3CF₃-COOH | retain tafamidis-CF₃ as positive baseline | 2 | −7.5 est | MED | Retained for SAR comparison |
| 10 | A6α — benzoxazole-3F-COOH | minimal perturbation benzoxazole | 1 | −6.8 | LOW | Negative control for distal logic |
| 11 | D1α — iododiflunisal-3CN variant | diphenyl-ether + I + distal CN | 3 | −7.6 | LOW | Iodo probe already in h01 Phase 8 SOP |
| 12 | D2α — iododiflunisal-3OMe | ibid. with OMe | 3 | −7.5 | LOW | Clean iodo analog |
| 13 | D3α — diflunisal-3OH-4F-COOH (parent) | known compound | 1 | −6.9 (5b) | MED | Benchmark negative control (parent) |
| 14 | Niflumic acid (parent) | retain as MD chemical probe | 1 | −6.18 (v2) | HIGH | Phase 4h probe, NOT wet-lab candidate |
| 15 | A3α — benzoxazole-3OH-4F-COOH | dual H-bond donor distal | 3 | −7.4 | LOW | Dual polar for K-K repulsion fix |
| 16 | E1α — indole-3COOH + distal CN | Oc1ccc2[nH]c(-c3cccc(C#N)c3)cc2c1C(O)=O | 4 | −7.1 | LOW | Rigid indole COX↓ |
| 17 | E2α — indole-3COOH + distal OMe | ibid. | 4 | −7.0 | LOW | Rigid indole |
| 18 | A1α + reversible Lys warhead | append -CHO salicylaldehyde-imine on aryl | 4 | −8.5 | NEW | Phase 6b: covalent Lys K1141 |
| 19 | A2α + α-cyanoacrylate | α,β-unsaturated cyanoester on aryl | 5 | −8.8 | NEW | Reversible Lys Michael |
| 20 | A1α + acyl-hydrazone warhead | -C(=O)NHN=CR on aryl | 4 | −8.3 | NEW | Reversible Lys imine |
| 21 | Meclofenamic (parent) | retain as MD probe only | 1 | −6.17 | HIGH | Phase 4h probe |
| 22 | Flufenamic (parent) | retain as MD probe only | 1 | −6.17 | HIGH | Phase 4h probe |
| 23 | A2β — benzoxazole-3OMe-NHSO₂Me | dual bioisostere | 3 | −7.3 | LOW | Sulfa + methoxy |
| 24 | A4β — benzoxazole-3CF₃-NHSO₂Me | acyl-sulfonamide variant of tafamidis | 3 | −7.6 | LOW | No-COX tafamidis |
| 25 | A5β — benzoxazole-3NHSO₂Me-NHSO₂Me | bis-sulfonamide | 4 | −7.5 | LOW | Clean polar |
| 26 | B4α — benzothiazole-3CF₃-COOH | thia-tafamidis | 3 | −7.4 | LOW | Scaffold diversification |
| 27 | C2α — benzimidazole-3OMe-COOH | basic-N methoxy variant | 3 | −7.0 | MED | Protonation-state probe |
| 28 | A1α + N-methyl | benzoxazole-3CN-COOH with N-Me on anthranilic | 2 | −7.1 | LOW | Remove H-bond donor to test necessity |
| 29 | A2α + 5-F on proximal ring | additional proximal F | 2 | −7.3 | LOW | Metabolic stability |
| 30 | A1α methyl ester (prodrug) | ester of #1 | 1 | −6.8 (inactive) | LOW | Prodrug + hydrolysis validation |
Synthesis tractability stratification
- Tier 1 (commercial or 1-step): #5, #10, #13, #14, #21, #22, #28, #29, #30 — order today; start MD probe + ThermoFluor Gate 1 within 1 week.
- Tier 2 (2-step literature): #1, #2, #9 — synth within 2-4 weeks via standard benzoxazole formation (2-aminophenol + 3-substituted benzaldehyde / benzoyl chloride).
- Tier 3 (3-step multistep): #3, #4, #6, #7, #11, #12, #15, #23, #24, #25, #26, #27 — CRO quote needed; $5-15k / compound; 4-6 wk.
- Tier 4 (novel, 4-step): #16, #17, #18, #20 — medchem CRO; $15-30k / compound.
- Tier 5 (exotic): #19 — α-cyanoacrylate warhead on benzoxazole; $30-50k / compound.
Phase 6b covalent warhead sub-strategy
Three Lys-reactive reversible warheads for covalent K1141 engagement. All are Michael acceptors or aldehyde-imines that hydrolyze back under physiological conditions (koff ~ 10⁻³ to 10⁻⁴ s⁻¹). Enough residence time for chaperone rescue, not permanent inactivation.
- Salicylaldehyde (SA) imine — 3-CHO on the distal ring, ortho OH stabilizes imine with K1141 ε-NH₂. Known SA-based Lys binders Lys-adducts Kd ~1-10 µM. Compound #18.
- α-Cyanoacrylate Michael — α,β-unsaturated nitrile ester; K1141 thia-Michael on the β-carbon. Cyano stabilizes adduct; hydrolytic off-rate 10⁻³ s⁻¹. Compound #19.
- Acyl-hydrazone — R-C(=O)NH-N=CR’ with distal aryl; hydrazone with K1141 Schiff-like. Compound #20. Lowest synthetic complexity in covalent class.
Predicted off-target panel (Phase 6c)
For each Tier 1-2 compound, pre-score against the cochlear channel panel using AF3 + docking:
- TRPM4 (PDB 7MF0) — expect Kd ≥ 100 µM for Tier 2 benzoxazole-CN; vs 2.8 µM parent
- Cx50 (PDB 7JN0) — same
- BK (PDB 6V3G) — same
- KCNQ4 (paralog of DFNA2; PDB 7BYL) — same
- TMEM16A (PDB 5OYB) — same
- COX-1 (PDB 1Q4G) + COX-2 (PDB 5IKR) — expect Kd ≥ 200 µM for benzoxazoles vs 2-4 µM parent
Phase 6c will run a 30 × 6 ensemble panel = 180 docks; wall-time ~1 h on local Mac.
COX prediction CONTRADICTED by Phase 6c dock 2026-04-24
Stage A receptor prep + Stage B 10-ligand × 2-target dock delivered. The “expect Kd ≥ 200 µM for benzoxazoles” prediction is WRONG:
| Compound | COX-1 Kd | COX-2 Kd | Predicted | Δ |
|---|---|---|---|---|
| tafamidis (Core A parent) | 1.5 µM | 1.5 µM | ”no COX liability” | falsified |
| benzoxazole-3CN-COOH (#1) | 1.0 µM | 1.8 µM | ≥ 200 µM | >100× wrong |
| benzoxazole-3OMe-COOH (#2) | 0.97 µM | 0.99 µM | ≥ 200 µM | >200× wrong |
Why the design thesis failed: The Arg120 salt bridge to ligand carboxylate is the dominant COX binding interaction; it does not care whether the proximal aromatic body is anthranilic (fenamic) or bicyclic (benzoxazole/benzothiazole). Tyr385 H-bond and Ser530 lipophilic channel both accommodate either body geometry. The bicyclic bioisostere achieves no meaningful COX deselection on its own.
Pipeline calibrated against parent fenamates (niflumic 0.28 µM COX-2, flufenamic 0.37 µM — within 1 OOM of published IC₅₀s), so this is not an artifact.
→ See STRC h01 Phase 6c COX Selectivity Panel 2026-04-24 for full data.
COX-deselect medchem strategies (revision)
To actually kill COX while preserving K1141 binding, three concrete v4b sub-strategies:
- Replace -COOH with non-acidic bioisostere that retains K1141 ε-NH₃⁺ engagement but loses Arg120 salt bridge:
- Acyl-sulfonamide (-CONHSO₂Me, pKa ~4.5) — already in v4 acid set; partial COX engagement expected
- Tetrazole — same pKa range, similar issue
- N-acyl-cyanamide (-NHC≡N pKa ~4.5) — COX has not been characterized for this group; potential deselect
- Sulfone (-SO₂Me) — neutral, kills both COX salt bridge AND K1141 salt bridge → likely too aggressive
- Hydroxamic acid (-CONHOH) — already in v4 acid set; v3b winner; COX status untested but likely problematic
- Add Tyr385-clashing decoration on the proximal ring at the position that points into the COX hydrophobic channel above Tyr385:
- 4’-CONH₂ or 4’-OH at the position equivalent to fenamic acid’s 4’-Cl in mefenamic
- Need geometric overlay of v4 best hit with bound ID8 in 5IKR (Phase 6c follow-up)
- Predicted to keep K1141 fit (pocket is open at that vector) and break COX (Tyr385-OH clash)
- Replace proximal benzene with N-heteroaromatic (pyridyl, pyrazinyl) at the position that lies near COX Arg120:
- Pyridyl-N H-bond geometry is wrong for Arg120 guanidinium
- K1141 ε-NH₃⁺ should still pick up the pyridyl-N as H-bond acceptor (different geometry, same energetic gain)
- Phase 4h cores B (benzothiazole) and C (benzimidazole) already have one heteroatom; need the heteroatom positioned at the COX-Arg120 vector specifically
- Concrete first synth: 2-amino-pyrido[3,2-d]oxazole-5-COOH (pyridyl-fused benzoxazole)
Implication for Phase 8 SOP: COX activity assay (PGE₂ ELISA or fluorogenic substrate) must be a G2.5 kill gate, not deferred to G3 cochlear-channel panel. If a compound carries COOH + bicyclic core, it almost certainly hits COX at Kd ~1-3 µM. Order COX activity testing alongside MST so the kill happens before cryoEM-grade compound.
Implication for v4 in-flight: v4 fragment-grow library on locked benzofuran/benzoxazole/benzothiazole cores will produce many COOH leads. Top-N v4 hits should be cross-docked against COX-1/-2 (Phase 6c targets are ready) before any Phase 8 ordering, to filter out compounds with COX bleed before wet-lab expense.
Prioritized synthesis ladder
Stage 1 (week 0-1): order Tier 1 commercial (#5, #10, #13, #21, #22) + Tier 2 in-house (#1, #2, #9) — 8 compounds. REFRAMED 2026-04-24: per Phase 6c COX dock, treat Tier 1 commercials as calibration positives for ThermoFluor / MST / cell-rescue assay setup, NOT as candidate leads. Tafamidis & all benzoxazole-COOH variants will register on COX activity assays. Real lead generation needs the v4b COX-deselect sub-strategies (acyl-cyanamide / Tyr385-clash / pyridyl-fused proximal) listed below — those need synth from scratch, no Tier 1 commercials.
Stage 2 (week 2-4): Tier 3 CRO quote (#3, #4, #6, #7, #11, #12, #23, #24) — 8 compounds.
Stage 3 (week 5-8): Tier 4 novel + Tier 5 covalent (#16, #17, #18, #19, #20) — 5 compounds.
Total Stage 1-3: 21 compounds, budget $150-250k, timeline 8 wk. Feeds into STRC h01 Phase 8 Wet-Lab Triage SOP Gate 1 ThermoFluor.
Integration with Phase 3c v3b / v4 compute
- If v3b delivers GREEN on any fenamic-family parent (#14/#21/#22 in this table) → this document is the bioisosteric optimization plan for transforming the parent hit into a pediatric-acceptable drug. Proceed to Stage 1 synthesis order.
- If v3b delivers YELLOW → Phase 3c v4 virtual screen uses rows 1-17 of this table as seed + combinatorial expansion on Tier 1-3 scaffolds.
- If v3b delivers RED (parents + fenamic expansions all Kd > 30 µM) → Phase 3c v5 de novo RFdiffusion-pocket pivoted → this document retired until RFdiffusion hits are validated.
Ranking delta
A held, no tier change, no score change.
Document status: scaffold-ready. No empirical data yet; all ΔG estimates are predicted from docking rules-of-thumb + PDB structure-informed SAR. Validation path: Phase 4h MD-probe on parents 22 (already queued) → validate pharmacophore → refine library priorities → Phase 3c v4 virtual screen OR synthesis Stage 1.
Immediate next action: load rows 1-20 into a programmatic SMILES list + build Phase 3c v4 library script that uses this as seed (not combinatorial expansion from scratch).
Connections
[part-of]h01 hub[part-of]STRC Hypothesis Ranking[see-also]STRC h01 Fenamic Scaffold Tox Audit 2026-04-23[see-also]STRC h01 Phase 3c v2 Expanded Screen 2026-04-23[see-also]STRC h01 Phase 8 Wet-Lab Triage SOP[about]Misha