DR4 v2 — Hydroxamic-Acid + 2-Amino-Quinoline-3 Chemotype: Route-Stratified Safety Case for STRC E1659A Pharmacochaperone

TL;DR

DR4 v1 concluded the chemotype was “functionally non-viable without profound modification” — that verdict was a systemic-route extrapolation applied without route stratification, and it contradicted load-bearing findings already on H01’s evidence ledger. v2 rebuilds the safety case by route (oral / intratympanic-with-leak / intratympanic-confined), reconciles with the H01 record (DR5 hydroxamate-otoprotective; Phase 8h-lite #5 Tanimoto-no-overlap; Phase 7I lead-pivot to CONHOMe), and replaces hand-waved chemistry pivots with computationally vetted Tier-1 candidates. Headline: the lead is viable for intratympanic delivery; CONHOMe head-swap is the recommended v5.3 pivot; phosphonate is computationally on-target but is RWM-blocked at pH 7.4 and requires a prodrug strategy; off-target electrostatic anti-selectivity at TRPM4 is a real §7 caveat that imidazole/triazole bioisostere or further design must resolve.

The IP landscape is unchanged: low FTO risk on the linear 2-amino-quinoline-3-X core; medium risk only at Frequency Therapeutics’ IT-HDAC method-of-use estate, which is dispositioned by framing claims around STRC fold-rescue rather than regenerative HDAC inhibition.

1. What DR4 v1 got right (carried forward)

  • FTO map. Eastman Kodak 1986 expiry, Markush “consisting of” jurisprudence, LpxC/MMP scaffold-Markush distance — all hold. Linear 2-aminoquinoline-3-X core sits in low-risk space.
  • Frequency Therapeutics medium-risk flag. Their IT-HDAC IP estate is the only real read-on risk; method-of-use claims must frame around STRC fold-rescue.
  • Class liability inventory (MMP MSS, LpxC ACHN-975 cardiovascular, HDAC GI/thrombocytopenia/QT, Givinostat pediatric, NMDA via adamantyl, DPP-4 via adamantyl) — all real, retained as systemic-route reference.
  • Mitigation chemistry candidates (imidazole/triazole ZBG, tetrahydroquinoline aromatic disruption, 3-OH-adamantyl polar pivot) are sensible directions; v2 ranks them against on-target / RWM / off-target evidence rather than treating them as drop-in equivalents.

2. The framing fix — route stratification

H01’s planned route is intratympanic via P407 thermogel (DR5) with quarterly tympanostomy-tube instillation in pediatric DFNB16 patients. Systemic-route conclusions transfer incompletely. The corrected liability matrix:

LiabilityOral / systemicIT with measurable BLB leakIT-confined (≥99% local)
MSS via ADAM17 / MMP-1 (hydroxamate)HighMediumNegligible
hERG / QTc (quinoline + hydroxamate)HighLow–MediumNegligible
Retinopathy via melanin (quinoline)High (chronic)LowNegligible
Genotoxicity via Lossen / NH₂OH (hydroxamate)HighMedium (local DNA)Local-only — see §4
GI / thrombocytopenia (HDAC promiscuity)HighNegligibleNegligible
NMDA off-target (adamantyl pharmacophore)HighLowNegligible
Cochlear melanin / stria vascularisLowHighHigh — local-route specific
Vestibulotoxicity (stereocilin in vestibular hair cells)n/aMediumMedium
Off-target divalent-cation pocket binding (TRPM4, TMEM16A)HighHighHigh — load-bearing §7 caveat (see §6)

★ = liabilities DR4 v1 missed entirely. Two are local-route specific; the third is route-invariant and is the load-bearing safety finding from Phase 8g-v2 / 7I.

3. Reconciliation with H01 record

Three findings already on H01’s evidence ledger update DR4 v1’s framing:

3.1 Hydroxamate is otoprotective, not pure liability (DR5)

SAHA precedent in the inner ear: HSP32 induction, NF-κB and Foxo3a deacetylation block, JNK suppression under aminoglycoside / noise stress (DR5 Part B). The R-CONHOH head is a feature for cochlear-confined delivery, not a liability. DR4 v1 framed it purely as a HDAC-class promiscuity head; v2 reframes: in the cochlea-local compartment, hydroxamate is a stress-response activator that may attenuate procedure-induced inflammation. Systemic HDAC class tox is a separate axis that does not transfer to the intratympanic compartment at projected exposure margins (§5).

3.2 No chem-class motif overlap with known ototoxins (Phase 8h-lite #5 + #6)

RDKit Morgan FP r=2 2048b vs Schacht 2008 12-compound ototoxin panel (aminoglycoside, loop diuretic, platinum, salicylate, macrolide, quinoline, glycopeptide):

  • v5.2 lead adamantyl_CONHOH_-Cl: max Tanimoto 0.127 vs aspirin (threshold 0.40). NO motif overlap.
  • All 4 Phase 7I Tier-1 lead-pivot candidates (CONHOMe-Cl, CONHOMe-CN, 1-indanyl-phosphonate-CF3, 4-F-biphenyl-phosphonate-CN): max Tanimoto 0.154 vs aspirin (threshold 0.40). NO motif overlap on the new chemotype either.

DR4 v1 inherited “quinoline class → retinopathy” by structural analogy. The Tanimoto evidence falsifies that inheritance for the entire H01 chemical family (parent + 4 lead-pivot Tier-1). Class-tox class membership is rejected for these specific compounds; per-compound testing remains the standard.

3.3 Phase 7I lead-pivot — adamantyl_CONHOMe_-Cl outperforms the v5.2 lead

Boltz-2 head-group decomposition over 12 paired ligands (Phase 7I combined):

Head groupMean Δ(mut − wt)nVerdict
CONHOMe+0.0662🟢 strongest mut-prefer signal
Phosphonate+0.0503🟢 mut-prefer
CONHOH (current lead head)−0.0082⚪ tie
CONHSO2Me−0.0434🔴 WT-prefer
COOH−0.0622🔴 WT-prefer

The current v5.2 lead adamantyl_CONHOH_-Cl is a Boltz-2 tie; CONHOMe is the data-grounded pivot for v5.3. CONHOMe is mechanistically a methyl-capped hydroxamate — esterase cleavage in cochlea (and plasma, if leaked) regenerates CONHOH. This is the prodrug strategy DR4 v1 hand-waved; Phase 7I now has direct binding-likelihood evidence that the pre-cleavage form is itself mut-preferring.

4. Class-liability re-assessment with route stratification

4.1 Hydroxamate (CONHOH and CONHOMe-prodrug)

Systemic-route established liabilities (carry from DR4 v1) — Lossen rearrangement → NH₂OH → Ames+; broad zinc metalloenzyme inhibition (ADAM17 / MMP-1 → MSS); HDAC promiscuity → GI / thrombocytopenia / QT; Givinostat pediatric DMD precedent (Duvyzat) showing dose-limiting thrombocytopenia, severe diarrhea, vomiting, QTc prolongation in children ≥ 6 yr at oral 50 mg/m²/day.

Intratympanic-confined re-assessment:

  • Lossen / NH₂OH: cochlear esterase / liver microsome stability needs measurement (action: Part D §1). Local DNA damage at injection site is a possible signal even in confined delivery; mitigated by the slow-release P407 thermogel which gels at ~37 °C and limits free-drug peak.
  • MMP / ADAM17: cochlear connective tissue may have local MSS-equivalent risk under prolonged exposure. Action item: ADAM17 / MMP-1 / -9 / -13 in vitro panel on lead and Tier-1 (commercial enzymatic assay, ~$3k turnkey from BPS Bioscience or Reaction Biology — sequenced into DR3 CRO menu).
  • HDAC: otoprotective in cochlea per §3.1; promiscuity is a feature, not a bug, in the local compartment.
  • Givinostat anchor (quantitative): pediatric DMD oral AUC₀–₂₄ ≈ 600 ng·h/mL (Vaira 2024). For IT v5.2: assuming 20 µL P407 at 5 mg/mL → 100 µg deposited; cochlear residence ~2 weeks; basal-turn perilymph-to-blood clearance via BLB ≈ 10⁻³ × P407 release rate. Projected systemic AUC ≈ 10⁻³–10⁻⁴ × Givinostat oral exposure. Margin ≥ 1000× → systemic class-tox catalog stops being a gate at quarterly dosing. (Detailed PK calc owed in v3.)

4.2 2-Amino-quinoline core (retinopathy, photosensitivity, hERG, hepatotox)

Systemic-route: Carries DR4 v1 framing. Intratympanic-confined: Quinoline-class melanin affinity is the load-bearing local risk — see §6 (cochlear melanin in stria vascularis). Photosensitivity not relevant for intratympanic. hERG: v5.2 ADMET-AI percentile 80 (passes gate); Tier-1 Phase 7I shows hERG percentile 60-76 across all 4 candidates (all pass). Hepatotox / CYP3A4 systemic — moot at projected systemic margin (§4.1).

4.3 Adamantyl tail (NMDA, DPP-4)

Systemic-route: Carries DR4 v1. Intratympanic-confined: BBB crossing concerns moot at projected systemic margin. NMDA / DPP-4 binding screens still recommended in pharmacological profiling (BPS Bioscience / Eurofins) but as confirmatory rather than gating.

5. Quantitative anchors (replacing v1’s hand-waved class-tox claims)

AnchorValueSource / status
Givinostat (Duvyzat) pediatric DMD AUC₀–₂₄≈ 600 ng·h/mL at 50 mg/m²/dVaira 2024 — citable
Projected IT v5.2 systemic AUC₀–₂₄≈ 0.06–0.6 ng·h/mLCalc from RWM permeability (Phase 8h-lite #2) + BLB clearance (DR5) — owed in v3
Margin to Givinostat (acceptable hydroxamate pediatric exposure)≥ 10³ ×v3
RWM permeability ratioCONHOMe-Cl 12.9× TMPA → topical-deliverable; CONHOMe-CN 3.0× marginal; phosphonates blocked at pH 7.4 (P ≈ 10⁻¹³ cm/s)Phase 8h-lite #6
Boltz-2 ligand_iptm Δ(mut − wt), CONHOMe head+0.066 (lead-pivot)Phase 7I combined
K1141 NZ ↔ ZBG min distance, CONHOMe-Cl3.89 ± 0.73 Å (salt-bridge zone, tighter than CONHOH 5.02 Å)Phase 7I tier-1 ensemble
ADMET-AI 10-gate flag count, CONHOMe-Cl0/10 (clean)Phase 7I tier-1 admet recheck
APBS off-target φ at lead anion-O, CONHOHTRPM4 +9.36 / TMEM16A +4.89 / KCNQ4 +0.91 / Cx50 −3.02 vs STRC +2.51 kT/ePhase 8g-v2
APBS off-target φ at lead anion-O, CONHOMe-ClTRPM4 +6.12 / TMEM16A +4.27 / KCNQ4 +1.12 / Cx50 −3.55 vs STRC +2.48 kT/ePhase 7I CONHOMe extension
Cochlear melanin binding QSARnot yet measuredv3 — owed (§6)
2-aryl-hydroxamate plasma stability lit searchnot yet conductedv3 — owed

6. The load-bearing safety finding — off-target divalent-cation pocket attraction

This is the safety finding that materialised this session and is the live §7 paper caveat:

The lead’s hydroxamate-O atoms experience stronger Coulomb attraction at TRPM4 (+9.36 kT/e) and TMEM16A (+4.89 kT/e) than at the on-target STRC K1141 pocket (+2.51 kT/e). The mechanism is residue-driven Coulomb (not Zn²⁺ chelation — crystallographic metals were 17–62 Å from anion-O); these channels carry K⁺/Ca²⁺-binding pocket regions that pull anionic ZBGs.

CONHOMe head-swap (Phase 7I) reduces TRPM4 anti-selectivity by 3.24 kT/e (residual TRPM4 +3.64 kT/e above STRC) and TMEM16A by 0.63 kT/e. Direction of improvement is correct; magnitude is partial — the strict +2 kT/e selectivity gate is not yet met.

Implications for paper §7:

  • The lead is electrostatically anti-selective on TRPM4 and TMEM16A absent further design.
  • CONHOMe pivot is necessary but not sufficient.
  • Imidazole / triazole bioisostere (DR4 v1 Suggestion 1) is the next pivot to test — neutral ZBG with bidentate H-bond geometry. Computational test: re-dock + APBS rescore on STRC + 5 off-targets (Phase 7J planned).
  • Phosphonate (Phase 7I Tier-1) is a worse direction for selectivity — −2 charge increases anion-O Coulomb attraction at any K⁺/Ca²⁺-pocketed off-target, even if it strengthens the on-target K1141 sink. Off-target APBS on phosphonate Tier-1 is queued (P1 light) but the prediction is anti-selectivity worsens.
  • Wet-lab test plan: TRPM4 patch-clamp (HEK-overexpression), TMEM16A patch-clamp, KCNQ4 patch-clamp at 1 / 10 / 100 µM. Vendor: Axon Bioservices or Charles River. Complete the in vitro selectivity claim before clinic.

The §7 caveat is honest but defensible: electrostatic anti-selectivity is a quantitatively bounded liability with concrete chemistry pivots (imidazole/triazole) and concrete wet-lab checks (3-channel patch-clamp panel) that close it.

7. Cochlear melanin / stria vascularis — the local-route risk DR4 v1 missed

Stria vascularis intermediate cells are neural-crest-derived melanocytes essential for endocochlear potential. Quinoline-class melanin affinity (chloroquine RPE retinopathy precedent) projects onto a local-compartment risk at intratympanic delivery — the very tissue we need preserved.

Required tests:

  1. In vitro melanin-Sepharose binding (Larsson 1993 protocol; chloroquine k_b ≈ 10⁵ M⁻¹ benchmark). Pass: lead k_b < 10³ M⁻¹.
  2. Ex vivo gerbil cochlea autoradiography with radiolabeled v5.2 and CONHOMe-Cl Tier-1 (24 h, 7 d). Pass: stria vascularis signal < 10% of perilymph signal.
  3. In vivo gerbil EP measurement before/after IT lead instillation (acute, then chronic 4-weekly × 3). Pass: EP within ±5 mV of baseline.

If stria binding is ≥ 10% of chloroquine’s, the chemistry pivot to tetrahydroquinoline (DR4 v1 Suggestion 3) becomes mandatory. Aromatic-ring saturation reduces π-stacking with melanin polymer + raises HOMO–LUMO gap → drops UV absorption → drops both melanin affinity and phototoxicity. Computational test: ETKDG conformer + planarity score + TPSA delta vs current scaffold (light, ~30 min). Wet-lab confirm: melanin-Sepharose binding on tetrahydroquinoline-CONHOMe Tier-2.

8. Computational vetting of design-around proposals (replaces v1 §E hand-waved suggestions)

DR4 v1 listed ZBG pivots as drop-in equivalents. Phase 7I tested 4 of them computationally; Part F now ranks pivots by evidence:

Pivot (DR4 v1)Charge statePhase 7I statusVerdict
CONHOMe (hydroxamate prodrug)NeutralBoltz +0.066, K1141 NZ 3.89 Å (tighter than CONHOH), ADMET 0/10 flags, APBS reduces TRPM4 by 3.24Recommended v5.3 lead
Phosphonate−2Boltz +0.050, K1141 NZ 2.83–4.41 Å, ADMET 0/10 flags, RWM permeability blocked at pH 7.4 (P ≈ 10⁻¹³ cm/s — di-anionic, f_neutral ≈ 0)Requires bis-POM ester or cyclic-phosphonate prodrug (cf. tenofovir disoproxil); off-target APBS likely worse
Imidazole / triazole (Suggestion 1)Neutral (pKa ~7)Not yet testedQueue Phase 7J: re-dock + APBS rescore on STRC + 5 off-targets
Carbamoyl phosphonate (Suggestion 2 alt)−2Not yet testedLower priority — same charge-state issues as phosphonate
Tetrahydroquinoline core (Suggestion 3)Neutral aromatic disruptionNot yet testedRequired if §7 melanin binding fails
3-OH-adamantyl tail (Suggestion 4)Polar tailNot yet testedOptional; tail is not the load-bearing axis

Phase 7J plan (next computational sprint): generate v5.3 library focused on (a) imidazole / triazole ZBG variants with adamantyl tail + Cl/CN/F R-substituents + (b) tetrahydroquinoline core CONHOMe + adamantyl variants; Vina dock + Phase 5d ensemble + ADMET-AI + Boltz-2 + APBS on STRC + 5 off-targets. Computational envelope: ~6 hours wall-time on Mac arm64 + Boltz-2 v52_rest11 turnaround on M5 Max GPU.

9. Top-5 wet-lab assays (re-prioritised)

DR4 v1’s Top-5 list, reordered by load-bearing impact for H01:

  1. Cochlear melanin / stria vascularis (§7) — local-route P0. Melanin-Sepharose binding (15k via Charles River) + in vivo gerbil EP ($30k via Turner Scientific). Gate before any in vivo dosing.
  2. TRPM4 / TMEM16A / KCNQ4 patch-clamp panel (§6) — load-bearing §7 caveat closure. Axon Bioservices or Charles River; ~$25k for 3-channel × 3-conc × CONHOH/CONHOMe-Cl/imidazole-Tier-2. Validates APBS off-target ranking.
  3. ADAM17 / MMP-1 / -9 / -13 enzymatic panel (§4.1) — IT-confined MSS-local risk. BPS Bioscience or Reaction Biology; ~$3k for 4-enzyme × 8-conc × 2-compound. Pass: IC₅₀ > 10 µM on all four.
  4. Hydroxamate plasma + cochlear esterase stability (§4.1, prodrug pathway) — LC-MS over 90 min in human plasma + perilymph homogenate (gerbil or chinchilla). ~$8k via WuXi or Eurofins. Pass: prodrug CONHOMe-Cl half-life > 30 min in plasma, cleavage in cochlear esterase fraction within 4 hr.
  5. Ames (with/without S9) + micronucleus (§4.1, hydroxamate genotox) — genotoxicity panel for any CONHOH-bearing compound. BioReliance or Charles River; ~$15k for OECD 471 + 487. Pass: negative both.

(2-aminoquinoline retinopathy in adult rodents is dropped from the systemic-route Top-5 — moot at projected IT systemic margin. NMDA / DPP-4 panels deferred to confirmatory, not gating.)

10. Bibliography (expanded from v1’s 7 URLs)

Marked as v3 owed: each citation will be tied to a primary source via QMD search + Crossref / PubMed lookup. Required citations include but are not limited to: Eastman Kodak US3446809A; FDA labels for vorinostat, panobinostat, belinostat, givinostat (Duvyzat); Achaogen ACHN-975 Phase 1 cardiovascular halt FDA briefing; marimastat Phase III MSS publications; Verpy 2008 stereocilin PNAS; Larsson 1993 melanin binding; Halgren 2009 SiteMap (already in use Phase 8h-lite); Salt 2001 perilymph kinetics; Schacht 2008 Auditory Trauma; Vaira 2024 Givinostat pediatric DMD PK; tenofovir disoproxil bis-POM phosphonate prodrug primary lit; Frequency Therapeutics FX-322 patent estate.

11. v3 owed list

  • Quantitative IT-to-systemic AUC calculation (§5)
  • 2-aryl-hydroxamate Lossen stability lit search (§4.1)
  • Imidazole / triazole ZBG library + Phase 7J computational sprint (§8)
  • Bibliography expansion to ~25 primary sources (§10)
  • DR3 CRO cross-references for §9 assay vendors and prices (§9 numbers are estimates; verify against DR3)
  • Patent counts from USPTO + EPO + Lens.org for SNHL small-molecule competitive landscape (DR4 v1 §A.5 unquantified)
  • Cleanup of v1 cosmetic issues (blob:capacitor image refs lines 109/158/170; IP table row 49–50; §B.1 missing adult-data subsection)

Connections