STRC Research

DR4_review_and_enhancements

8 min read

DR4 — Review & Enhancements (what to add to make it load-bearing for H01)

TL;DR

DR4 is a useful systemic-route IP/tox survey, but as written it is mis-framed for H01 — the planned route is intratympanic, not oral, and DR4’s “functionally non-viable without profound modification” verdict is a systemic-dose conclusion applied without route stratification. It also ignores two findings already on the H01 record (DR5 hydroxamate-otoprotective SAHA precedent; Phase 8h-lite #5 Tanimoto vs 12 ototoxin classes max 0.127). With the additions below, DR4 becomes the safety-case spine for paper §7. Without them, citing DR4 verbatim would falsely block the next computational step.

What DR4 gets right (keep)

  • FTO map is structurally sound. The Eastman Kodak 1986 expiry, the Markush “consisting of” jurisprudence, and the LpxC/MMP scaffold-Markush distance argument hold. Low FTO risk for the linear 2-aminoquinoline-3-hydroxamate is defensible.
  • Frequency Therapeutics medium-risk flag is correct. Their IT-HDAC IP estate is the only real read-on risk; DR4 correctly identifies it and prescribes method-of-use claim framing around STRC fold-rescue.
  • Class liability inventory is comprehensive. MMP MSS, LpxC ACHN-975 cardiovascular, HDAC GI/thrombocytopenia/QT, Givinostat pediatric data, NMDA via adamantyl, DPP-4 via adamantyl — all real and should remain in the tox annex.
  • Mitigation menu is sensible at the chemistry level. Imidazole/triazole ZBG, tetrahydroquinoline aromatic disruption, 3-OH-adamantyl polar pivot — these are the right candidates to consider (but require computational vetting; see §3 below).

1. Critical framing error — route stratification missing

DR4 collapses oral and intratympanic exposure into one verdict. Every tox row needs splitting into three columns:

LiabilityOral / systemicIT with measurable BLB leakIT-confined (≥99% local)
MSS via ADAM17/MMP-1HighMediumNegligible
hERG / QTcHighLow–MediumNegligible
Retinopathy via melaninHigh (chronic)LowNegligible
Genotoxicity (Lossen)HighMedium (local DNA)Medium-Local — see §2
GI / thrombocytopeniaHighNegligibleNegligible
NMDA off-targetHighLowNegligible
Cochlear melanin (stria vascularis) ★LowHighHigh
Vestibulotoxicity (stereocilin in vestibular hair cells) ★n/aMediumMedium

★ = liabilities DR4 missed entirely. Local-route specific.

Action: rebuild Part E table with these three columns. Without it, DR4’s headline “non-viable without profound modification” is just wrong for the actual route.

2. Direct contradictions with H01 record DR4 must reconcile

Two findings already on H01’s evidence ledger fight DR4’s framing:

  • Hydroxamate is otoprotective, not ototoxic (DR5, SAHA precedent — HSP32 induction, NF-κB / Foxo3a deacetylation block, JNK suppression under aminoglycoside / noise stress). DR4 treats –CONHOH purely as a liability head; the literature in DR5 says it is a feature in inner ear stress. DR4 must either cite-and-rebut or absorb this — currently it does neither.
  • No chem-class overlap with known ototoxins (Phase 8h-lite #5: RDKit Morgan FP r=2 2048b vs 12-compound panel — aminoglycoside / loop diuretic / platinum / salicylate / macrolide / quinoline / glycopeptide. Max 0.127 (aspirin), threshold 0.40, NO motif overlap.) DR4 inherits “quinoline class → retinopathy” by structural analogy alone. The Tanimoto evidence falsifies that inheritance for this lead. DR4 should cite Phase 8h-lite #5 in Part C and downgrade the “established pharmacophore” framing to “watch-list, structurally distant from class”.

Action: add a “Reconciliation with H01 record” subsection in Part E that explicitly addresses these two findings. The chemotype-specific Phase 7H Boltz-2 result (anthranilate-NH-aryl mut-prefer; salicylate-COOH WT-prefer) further sharpens this — different chem-classes within quinolines behave differently against the same protein, so blanket class-tox inheritance is the wrong inferential frame.

3. K1141 NZ Coulomb sink lock-in — design-around suggestions need computational vetting

DR4 lists ZBG replacements as drop-in. They aren’t, given Phase 8h-lite #4:

Lead CONHO⁻ ↔ K1141 NZ 5.02 ± 0.57 Å across 20 Phase 5d snapshots. Coulomb attraction range, edge of direct salt-bridge zone. K1141 NZ position SD 1.11 Å (rigid pocket).

Each DR4-suggested ZBG must be re-docked / Boltz-2’d against the Phase 5d ensemble before being treated as equivalent:

DR4 suggestionCharge state at pH 7.4K1141 NZ engagement predictionRequired test
2-(1S-OH-methyl)-imidazoleNeutral (pKa ~7)Loses Coulomb sink — H-bond onlyVina re-dock + Phase 5d SD analysis
1,2,4-triazoleNeutral / weakly acidicWeak sink; possible H-bondVina re-dock
Carbamoyl phosphonate−2 chargeStronger sink, but membrane-impermeableOK if STRC binding face is extracellular (it is — confirm w/ topology); MD water-shell check
3-OH-adamantyl tailPolar atom on cageShould preserve C2-tail vectorPhase 5d snapshot re-dock + lipophilicity correction
Tetrahydroquinoline coreLoss of aromaticityMay break C3-head vectorRe-dock + pose-similarity score vs lead

Action: add a new Part F “Computational vetting of design-around proposals” — list the docking/Boltz-2 jobs needed before any chemistry pivot is greenlit. This is a 1-week computational sprint, not a safety claim.

4. Missing quantitative anchors

DR4 makes qualitative class-tox claims without exposure margins. The most useful additions:

  • Givinostat anchor. Pediatric DMD oral dose ~50 mg/m²/day, AUC₀₋₂₄ ≈ 600 ng·h/mL (Vaira 2024). Project IT v5.2 systemic AUC: RWM permeability × cochlear → BLB clearance × distribution volume → expect ~10⁻³–10⁻⁴ × Givinostat oral exposure assuming quarterly P407 thermogel instillation. If margin is ≥1000×, the Givinostat AE catalog stops being a gate. Run this calc; cite it.
  • Cochlear-melanin binding assay scoping. Use Larsson 1993 in-vitro melanin-Sepharose binding (k_b for chloroquine ≈ 10⁵ M⁻¹) as benchmark. Predict v5.2 binding from c log P 1.94 + Hansch melanin QSAR. If k_b < 10³ M⁻¹, the stria vascularis risk is acceptable for quarterly dosing. Add this as a Part C subsection.
  • Lossen rearrangement specifically for 2-aryl-hydroxamates. General hydroxamate Ames+ via metabolic activation, but the 2-amino-NH-aryl substitution geometry may stabilize the carbonyl through intramolecular H-bond. Need PubChem / ACS lit search for prior 2-aryl-hydroxamate plasma stability work. Cite or commission the experiment.
  • CYP3A4 age-stratification for any systemic leak. Pediatric DFNB16 patients are diagnosed 0–3 yr. CYP3A4 activity at age <2 yr is ~30% of adolescent. Even sub-1% systemic leak hits a different metabolic landscape than DR4’s adult-extrapolated estimates. Add age-cohort PK model.

5. Cochlear melanin — the local-route risk DR4 entirely missed

The stria vascularis intermediate cells (intermediate cells = neural-crest-derived melanocytes) carry melanin granules. They sit in the cochlear lateral wall and are essential for endocochlear potential (EP) maintenance. Aminoquinoline melanin affinity is well-established (chloroquine RPE precedent in DR4 Part C).

Local intratympanic 2-aminoquinoline could accumulate in the very tissue we need preserved. This is a local-route-specific risk that increases with intratympanic confinement (the opposite of systemic risks).

Action: add Part C.6 “Cochlear melanin and stria vascularis risk”. Tests required:

  • In vitro: melanin-Sepharose binding assay (k_b benchmark vs chloroquine).
  • Ex vivo: gerbil cochlea slice + radiolabeled v5.2 → autoradiography of stria vascularis lateral wall (24h, 7d).
  • In vivo: gerbil EP measurement before/after IT v5.2 (acute, then chronic 4-weekly × 3).

If stria binding is ≥10% of chloroquine’s, structural pivot to tetrahydroquinoline is required for IT route, not just oral.

6. Other gaps

  • No prodrug strategy section. Esterase-cleaved hydroxamate-O-acyl prodrug masks the ZBG until cochlear esterases unmask it. Perilymph esterase profile vs serum esterase profile would predict cochlear-confined activation. Worth a paragraph; ties directly to “minimize systemic exposure” goal.
  • No vestibulotoxicity discussion. Stereocilin is in vestibular hair cells (Verpy 2008). Balance, gait, vertigo screens are pediatric-relevant. Add to DR3 CRO menu cross-ref.
  • No quantification of “white space”. DR4 says SNHL small-molecule space is mostly gene therapy without numbers. Run a USPTO + EPO + Lens.org search “DFNB16 OR stereocilin OR sensorineural hearing loss” filtered by claim language for small-molecule mechanism. Quantify N patents, % small-molecule, % active vs expired.
  • DR1 / DR2 / DR3 cross-references missing. DR4 should cite DR1’s tafamidis/migalastat chronic-safety profile (the comparable for “acceptable pharmacochaperone tox”), DR2’s competitive landscape, and DR3’s CRO assignments for each Top-5 screen. Currently DR4 reads as an island.
  • Part B section 1 has a structural gap. “Approved Hydroxamic-Acid Drugs and FDA Label Signals” header → directly into “Pediatric Exposure Data” subsection without an adult-data subsection between them. Either remove the §1 placeholder header or fill in the FDA label table that the schema implies.
  • IP table row formatting bug. Line 49–50 of source: “Pharmacochaperone Mechanism” row has only 2 of 4 cells populated. Reformat.
  • Bibliography is thin. Only 7 unique URLs serve all [1][2][3][4][5] citations; same issue as DR5. Specifically missing: actual Frequency FX-322 patent numbers, ACHN-975 cardiovascular trial halt FDA briefing doc, Givinostat (Duvyzat) FDA label, marimastat Phase III MSS publications, Larsson 1993 melanin binding paper, Verpy 2008 stereocilin paper. Expand to ~25 sources.
  • Stale blob:capacitor://localhost/... image references at lines 109, 158, 170. Same clipboard-paste artefact as DR5. Decode and replace inline:
    • Line 109: I_Kr (rapid delayed rectifier potassium current)
    • Line 158: [hydroxylamine] (NH₂OH)
    • Line 170: IC₅₀

7. Concrete deliverables for DR4 v2

Ordered by load-bearing impact on H01:

  1. Route-stratified tox table (oral / IT-leak / IT-confined). Replaces current Part E table. P0.
  2. Reconciliation with H01 record subsection: hydroxamate-otoprotective (DR5) + Tanimoto-no-overlap (Phase 8h-lite #5). P0.
  3. Cochlear melanin / stria vascularis Part C.6 with binding-assay scope. P0 (new local-route risk).
  4. K1141 NZ design-around computational vetting as new Part F. P1 (1-week sprint).
  5. Givinostat AUC anchor quantifying systemic margin under IT route. P1.
  6. Bibliography expansion to ~25 specific primary sources. P1.
  7. Prodrug strategy paragraph (esterase-cleaved). P2.
  8. DR1/DR2/DR3 cross-references woven through. P2.
  9. Cleanup: blob image refs, IP table row, §B.1 header gap. P2.

Ranking delta

No change. DR4 review is a survey-quality artifact, not a mechanism/delivery proof. tier A held | mech 4 held | deliv 4 held | misha_fit 4 held. next_step unchanged: APBS on off-target enclosed pockets (P0 heavy), Boltz-2 v52_rest11 analysis, v5.3 design, wet-lab cyto panel.

DR4 in current form would, if cited verbatim into paper §7, contradict DR5’s hydroxamate-otoprotective finding and Phase 8h-lite #5 Tanimoto. Treat as draft-quality survey requiring v2 before any tox claim lands in paper prose. P1 follow-up — sequence behind APBS and Boltz-2 v52_rest11 analysis but ahead of v5.3 design freeze.

Connections

Graph View

Backlinks