← All hypotheses

H01 · STRC Pharmacochaperone Evidence Package

A-TIER COMPUTATIONAL DOSSIER RFQ-READY REVIEWER PACK · 2026-04-27

Patient-originated STRC E1659A pharmacochaperone hypothesis. The therapeutic idea is a small-molecule fold-stabilizer that binds the electropositive K1141 pocket created by the maternal E1659A allele, rescuing misfolded stereocilin via Coulomb-driven recognition before secretion.

Current rank
A

Hold pending wetlab RFQ feasibility (A_hold_wetlab_rfq_feasibility).

APBS pocket Δ
−2.79 kcal/mol

Matched ensemble n=20: MUT +5.99±1.37 vs WT +1.46±1.55 kT/e (Welch p=6.9e-12, Cohen d=3.10).

RWM permeability
7.6× TMPA

Phase 8h-lite Salt 2001 Stokes-Einstein model: ~60% applied conc at basal turn @ 90 min.

Pipeline platform
1 of N

E1659A is the first worked example. The pipeline (allele → pocket → APBS → dock → RFQ → DSF) repeats for every STRC missense + extends to other DFNB / proteinopathy targets.

Exact ask

We are not nominating a development candidate. We are asking a CRO or collaborator to answer the first empirical question: can the v5.3 ligands be made, can a STRC K1141-pocket fragment reagent be produced, and do the ligands directly bind or stabilize WT vs E1659A STRC?

The Wetlab Handoff Packet (2026-04-27) is RFQ-ready: synthesis + STRC K1141-fragment expression + DSF/nanoDSF + SPR/BLI/MST as three independent work packages.

Best reviewer profiles
  • · Otology / inner-ear small-molecule delivery
  • · Stereocilin / STRC structural biology
  • · CADD / docking + MD methodology
  • · Acyl-sulfonamide medicinal chemistry
Pack contents
  • · Phase 5k matched-ensemble APBS gate-closing result
  • · Phase 8h-lite 5 light proofs (druggability, RWM, off-target, pose stability, ototoxin Tanimoto)
  • · Phase 9x lead committee α-sensitivity
  • · Phase 9b Min-Relax MM-GBSA orthogonal smoke
  • · Wetlab Handoff Packet 2026-04-27 (compound sheet + cascade + gap matrix)
  • · Reviewer objection table

Claims & non-claims

Two columns kept symmetric on purpose: every claim has a paired non-claim so the reviewer can see exactly where the dossier stops.

What we claim

  • + h01 is an A-tier computational hypothesis after Phase 8h-lite paper-claim closures (2026-04-26) and Phase 9x α-sensitivity check.
  • + E1659A creates a measurable formal-anion preference at the K1141 pocket: Δ = −2.79 ± 0.28 kcal/mol (matched n=20 APBS, Welch p=6.9e-12, Cohen d=3.10).
  • + K1141 pocket meets all 4 Halgren druggability gates (V=1145 ų, phobic 0.61, V_pocket/V_lig 2.75).
  • + A 1066-1216 STRC fragment is a plausible first biophysics reagent for K1141-pocket assays (AF3 fragment ladder; ring RMSD 0.97 Å).
  • + A weak-anion acyl-sulfonamide v5.3 chemistry set is tractable: 2 lead candidates plus 1 optional comparator are ready to RFQ.
  • + 1 mM intratympanic delivery is plausible in the 8-compartment cochlear PBPK model (94–97% time above 1 µM in HC_ER, no cytosol safety breach).

What we do not claim

  • Not a lead-candidate nomination; this is RFQ-scoping only.
  • Not a clinical-readiness or IND claim.
  • No physical compound has been synthesized.
  • No STRC 1066-1216 WT/E1659A reagent has been expressed.
  • No DSF / nanoDSF / SPR / BLI / MST or cell-rescue data exist.
  • Off-target electrostatic selectivity remains INCONCLUSIVE LITE; APBS on enclosed pockets per off-target is the heavy planned step.
  • Phase 8h-lite is light-compute paper-claim closure, not production MD/MM-PBSA replacement.

What was computed

H01 is not one docking score. It is a stack of independent computational checks: structural ensembles, electrostatics, ligand design, protein-ligand cofolding, docking, biased MD, PBPK, ADMET, off-target panels, and endpoint free-energy smoke tests.

Method Compute Scale Key result
Matched WT/MUT structural ensembles Full-length STRC MD snapshots, WT and E1659A, then matched APBS pocket electrostatics. 20 WT frames + 20 mutant frames Mutant pocket formal-anion preference = −2.79 ± 0.28 kcal/mol; p=6.9e-12.
Fragment and pocket validation AF3 fragment ladder plus K1141 local-geometry checks against the MD-relaxed parent structure. 8 AF3 jobs, 40 CIFs, 40 confidence files 151-aa E1659A fragment kept K1141-ring RMSD at 0.97 Å; pocket pLDDT 78.7.
Chemistry library design RDKit triage of weak-anion acyl-sulfonamide bioisostere candidates. 12 designed SMILES, 6 passed filters Top-3 v5.3 candidates selected for Boltz-2, Vina, ADMET, and tauRAMD.
Boltz-2 + Vina consensus Protein-ligand cofolding and mutant-ensemble redocking of the v5.3 top-3. 5 Boltz diffusion seeds × 3 ligands; 20 MD snapshots × 3 ligands in Vina Both methods put 1-indanyl_SO2Me_-Cl on the affinity track; all 3 passed ipTM ≥ 0.50.
Residence-time / tauRAMD Biased-MD ligand escape simulations on STRC and off-target TRPM4. 25/25 v5.2 STRC unbound; 6/6 TRPM4 unbound; 60/60 v5.3 STRC unbound Selectivity rescue did not clear the >5× gate; v5.3 cross-target bound ≤2.72×.
Off-target electrostatics APBS pocket-potential resampling across STRC and cochlear ion-channel off-targets. STRC, TRPM4, TMEM16A, KCNQ4, Cx50 panels STRC pocket is electropositive, but dominant off-target pocket ceilings remain higher.
Cochlear exposure model 8-compartment PBPK ODE with intratympanic dose sweep. 100 µM / 1 mM / 10 mM × fast/default/slow apical uptake 1 mM IT gives 94–97% time above 1 µM in HC_ER without cytosol safety breach in the model.
Endpoint free-energy smoke Repaired matched Amber WT/E1659A complexes, restrained min-relax frames, MMPBSA.py GB scoring. 6 frames WT + 6 frames E1659A E1659A − WT ΔΔTOTAL = −2.35 kcal/mol; supportive smoke, not production holo MD.

Phase-by-phase evidence

Ordered as a reader would evaluate the hypothesis: pocket, mutant specificity, reagent feasibility, chemistry, off-targets, exposure, lead committee, orthogonal smoke.

01
Layer
Pocket premise
K1141 pocket profile
Volume 1145 ų; hydrophobic fraction 0.61; pocket-to-ligand volume ratio 2.75.

All 4 Halgren druggability gates pass (Phase 8h-lite #1).

Supports mechanism
02
Layer
Mutant specificity
Matched-ensemble APBS
WT +1.46 ± 1.55 kT/e; E1659A +5.99 ± 1.37; Δ +4.53 ± 0.46 kT/e (p=6.9e-12).

E1659A creates a statistically strong anion-favouring electrostatic environment.

Strong computational support
03
Layer
Construct feasibility
AF3 K1141 fragment ladder
8 AlphaFold Server jobs; best operational fragment = 151_e1659a, ring RMSD 0.97 Å.

A 1066-1216 fragment is a plausible first biophysics reagent for K1141-pocket assays.

Ready for expression RFQ
04
Layer
v5.2 kinetic screen
Production tauRAMD on v5.2 shortlist
5 ligands × 5 replicas; 25/25 unbound; within-STRC residence spread 1.7×.

Residence-time differences exist, but same-head-class kinetic selectivity is too small.

Constrains claim
05
Layer
Off-target risk
TRPM4 cross-target tauRAMD
2 ligands × 3 replicas; TRPM4/STRC residence ratios 1.08–1.52×.

Kinetic-selectivity rescue failed for v5.2 by at least a 3× margin versus the >5× gate.

Negative but useful
06
Layer
Exposure model
PBPK dose sweep
1 mM IT: 93.8–96.9% of 72 h with HC_ER ≥ 1 µM; 0% HC_CYT ≥ 50 µM.

Exposure is not the main blocker if 1 mM intratympanic dosing is acceptable.

Supports delivery feasibility
07
Layer
v5.3 chemistry
Acyl-sulfonamide library
12 candidates designed; 6/12 passed Lipinski + TPSA ≤ 100 + mono-anion gates.

Weak-anion head-group pivot produced a tractable next-gen chemistry set.

Supports next experiment
08
Layer
Affinity ranking
Boltz-2 + Vina v5.3 consensus
Boltz ipTM rank-1 0.645 for 1-indanyl_-Cl; Vina mean rank-1 −0.67 kcal/mol.

1-indanyl_acylsulfonamide_SO2Me_-Cl is the affinity-track compound.

Lead-track evidence
09
Layer
Conformational robustness
v5.3 mutant-ensemble redock
3 ligands × 20 E1659A MD snapshots; 1-indanyl_-Cl rank-1 by mean and median.

Affinity signal is not just a single static receptor pose.

Supports affinity track
10
Layer
Developability triage
ADMET-AI top-3
2/3 clean on 10 endpoints; 1-indanyl has one borderline CYP3A4 flag at 90.4 vs 90.0 gate.

Use a two-track lead committee: affinity lead plus ADMET-clean adamantyl backup.

Actionable tradeoff
11
Layer
v5.3 residence check
STRC tauRAMD n=20 per ligand
60/60 unbound; tau means 21.59, 17.46, 15.90 ps; spread 1.358×.

Class does not generate enough kinetic spread for a selectivity claim, but remains useful for target-engagement testing.

Narrows wetlab claim
12
Layer
Orthogonal score smoke
Min-relax MM-GBSA
WT −15.84; E1659A −18.18; Δ −2.35 kcal/mol.

An APBS-independent endpoint score preserves the mutant-favouring sign.

Supportive smoke
13
Layer
α-sensitivity
Phase 9x lead committee
Rank aggregation stable for kinetic-axis weight α=0.5–2.0 across Boltz/Vina/tauRAMD/ADMET.

The two-track committee decision is not fragile to weight choice.

Robust committee

Phase 8h-lite — paper-claim closures

Five light-compute proofs (under 1 minute total wall time, no new MD/dock/APBS) closing paper-claim gaps using existing Phase 5d/5k/8e data + RAG-mined literature (Salt 2001, Halgren 2009, Schacht 2008).

#1

K1141 pocket druggability

PASS

Halgren SiteMap-style grid burial estimator (Halgren 2009 spec).

V = 1145 ų, phobic 0.609, V_pocket/V_lig 2.75. All 4 druggability gates ✅. Pocket lining: W1612, F1646, F1169, W1652 (π-stacking shell); K1141 = 26% of philic shell.

#2

RWM permeability

PASS — deliv 3→4

Salt 2001 P_TMPA baseline + Stokes-Einstein MW^(−1/3) + Avdeef logP/TPSA + Henderson-Hasselbalch f_neutral.

Lead P = 1.45×10⁻⁷ cm/s = 7.6× TMPA. Inferred basal-turn concentration ~60% applied @ 90 min vs 8% for TMPA. CONHOH pKa sensitivity 7.2–7.6× (robust).

#3

Off-target electrostatic selectivity

INCONCLUSIVE LITE

Two proxies: Vina-pose-anchored count + receptor-wide K/R cluster scan.

Vina parked CONHO⁻ at 7.5 Å from K1141 (Gasteiger-neutral artefact). Receptor-wide scan returns large K/R clusters on all 4 channels because their voltage-sensors / selectivity filters are K/R-rich — not enclosed druggable pockets.

#4

Pose ensemble stability

PASS — mech 4 holds

Distance K1141 NZ ↔ lead CONHO⁻ across all 20 Phase 5d MD snapshots (LIE substitute).

K1141 NZ per-axis SD 1.72 / 1.15 / 1.10 Å (rigid). Lead CONHO⁻ ↔ K1141 NZ mean 5.02 ± 0.57 Å, range 3.93–6.18 Å — Coulomb-attraction range across entire ensemble.

#5

Tanimoto vs known cochlear ototoxins

PASS — no class flag

RDKit Morgan FP r=2 2048 bits; lead vs 12-compound Schacht 2008 panel (aminoglycosides, loop diuretics, platinums, salicylates, macrolide, quinoline, glycopeptide).

Closest match Tanimoto 0.127 (aspirin). No structural class motif overlap with known ototoxins.

Compound priority for first wetlab contact

Two-track lead committee: an affinity-led indanyl variant plus an ADMET-clean adamantyl backup. Phase 9x α-sensitivity confirmed the committee decision is robust to kinetic-axis weight choice (α=0.5–2.0).

# Compound Track Computational evidence Risk
1 1-indanyl_acylsulfonamide_SO2Me_-Cl Affinity lead Boltz-2 ipTM 0.645; Vina mean −0.67; mut-ensemble rank-1 by mean and median. One borderline CYP3A4 ADMET-AI flag (90.4 vs 90.0 gate); not the residence-time winner.
2 adamantyl_acylsulfonamide_SO2Me_-Cl Practical backup ADMET-clean on all 10 endpoints; tightest tauRAMD SEM; α-sensitivity-robust. Mutant-ensemble affinity is weaker and more best-pose-dependent.
3 adamantyl_acylsulfonamide_SO2Me_-CF3 Optional residence comparator ADMET-clean; tauRAMD mean rank-1 (in-class). Wider tau outliers and weaker affinity signal.

Assay cascade

Five sequential gates from the Wetlab Handoff Packet (2026-04-27). Do not run Gate N unless Gate N−1 passes. A clean failure at any gate is itself useful — the question stops being "should we keep computing?" and becomes "what wetlab measurement closes the next gap?"

0
Gate 0

Can the reagents be made?

Readout

Custom synthesis QC (HPLC, LC-MS, identity); STRC 1066-1216 WT/E1659A expression (HEK293 or Sf9, His6/AviTag), SDS-PAGE, SEC, identity.

Pass

Usable monodisperse WT/E1659A fragment + identity-confirmed compounds in hand.

Stop

Fragment mis-folds, aggregates, or cannot be purified; or compounds fail identity/purity.

1
Gate 1

Does ligand stabilize the fold?

Readout

DSF or nanoDSF, dose response, vehicle-matched.

Pass

Reproducible thermal shift vs vehicle, ideally larger on E1659A than WT.

Stop

No reproducible shift, or WT shift ≥ E1659A shift.

2
Gate 2

Is binding direct?

Readout

SPR / BLI / MST; DMSO-matched; reference surfaces; immobilization-artefact controls.

Pass

Dose-dependent Kd, koff or stoichiometry without aggregation-shaped curves.

Stop

Aggregation curves, no concentration response, or vehicle binding.

2.5
Gate 2.5

Is the signal clean?

Readout

Aggregation / reactivity panel; hERG, TRPM4, TMEM16A, KCNQ4, BK, Cx50 counterscreens; COX if NSAID-like risk remains.

Pass

No meaningful activity at relevant multiples of measured Kd.

Stop

Off-target activity within or below the on-target Kd window.

3
Gate 3

Does it rescue cells?

Readout

HEK293 full-length E1659A STRC trafficking / mature-glycoform readout.

Pass

Dose-dependent E1659A rescue after direct biophysics passes.

Stop

No rescue at non-toxic doses, or rescue paired with toxicity.

RFQ packet — what to send, what to leave out

Wetlab Handoff Packet 2026-04-27. The blocking gap is reagent generation, not another in-silico pass: production MD/MM-PBSA stays paper-grade support, not a wetlab-contact blocker.

Packet section Include now Exclude for now
Decision memo One-page H01 aim, allele, K1141 pocket, ask, kill/pass criteria. IND language or "lead candidate" framing.
Construct sheet STRC 1066-1216 K1141 fragment WT + E1659A; His6/AviTag options; HEK293/Sf9 expression feasibility. Full-length STRC protein production as first reagent.
Compound sheet 2 required molecules + 1 optional comparator with SMILES, computed properties, QC asks. Large exploratory library synthesis.
Assay cascade Gate 0 protein QC → Gate 1 DSF/nanoDSF → Gate 2 SPR/BLI/MST → Gate 2.5 counterscreens → Gate 3 cell rescue. CryoEM or animal PK before direct binding.
Evidence appendix APBS sign, Boltz/Vina ranking, tauRAMD class verdict, ADMET-AI, min-relax MM-GBSA smoke with caveats. Production MD as wetlab prerequisite.
RFQ artifacts compound_sheet.csv, gap_matrix.csv, rfq_email.md. Vendor cost claims before live RFQs.

Reviewer objection table

Hostile-read checklist run before external review. Each row pairs the strongest objection with the current answer, residual risk, and the smallest concrete check that could move the needle.

Objection Current answer Residual risk Minimum next check
Vina ranks WT and mutant similarly. Diagnosed as a Gasteiger-neutral force-field artefact (Phase 5k, Phase 8h-lite #3): docking does not see the formal-anion charge. Reviewer may take Vina ranks at face value. Lead with matched-ensemble APBS (p=6.9e-12, d=3.10), not Vina. Wet DSF/nanoDSF closes the question.
APBS Δ is one ensemble; could be MD-trajectory artefact. Matched n=20 WT vs n=20 E1659A frames from the same MD protocol; Welch t=−13.7, Cohen d=3.10. A new MD seed could shift the Δ. Phase 5d/5k MD inputs, scripts, and seeds are versioned in the repo; reviewer can re-run. DSF Δ on E1659A vs WT is the definitive falsifier.
tauRAMD did not clear the >5× selectivity gate. Acknowledged. Class kinetic spread is small; we explicitly do NOT claim kinetic selectivity. Reviewer may read this as h01 disqualified. Reframe ask as target engagement (DSF + SPR), not kinetic discrimination. Counterscreens close selectivity wet, not in silico.
Off-target electrostatic selectivity is INCONCLUSIVE LITE. Phase 8h-lite #3 confirmed: receptor-wide K/R scans return false-positives because off-target voltage-sensors are K/R-rich but not enclosed druggable pockets. Reviewer cannot rule out off-target binding from in silico. Per-target APBS on enclosed pockets is the heavy step; Gate 2.5 wet counterscreen panel (hERG / TRPM4 / TMEM16A / KCNQ4 / BK / Cx50) closes the question empirically.
Min-relax MM-GBSA is not production holo MD. Correct. Phase 9b smoke is supportive sign-check (Δ −2.35 kcal/mol), not paper-grade ΔG. Reviewer may discount because it is not full holo MD. Production holo MD is paper-grade upgrade, not wetlab-blocker. Phase 9b SKELETONs scoped if needed.
AF3 fragment may not refold like the full-length parent in vitro. AF3 ladder shows 151_e1659a ring RMSD 0.97 Å vs MD-relaxed parent and pocket pLDDT 78.7 — model agreement only. In-vitro refolding can fail at any step. Gate 0: SDS-PAGE / SEC / identity QC on actual expressed fragment. RFQ scopes both HEK293 and Sf9 routes.
Pharmacochaperones for extracellular targets are unusual. Acknowledged in mechanism flag (deliv 3, not 4). E1659A is a cysteine-loss variant predicted to cause ER mis-folding before secretion; rescue would happen at the ER stage, not extracellularly. Reviewer may anchor on extracellular-mature framing. Gate 3 cell rescue uses trafficking / mature-glycoform readout, not extracellular function.
PBPK 1 mM IT model has not been calibrated. 8-compartment ODE uses Salt 2001 RWM permeability + literature scala-tympani clearance + Avdeef logP/pKa. Real intracochlear distribution may diverge. Phase 8h-lite #2 RWM Stokes-Einstein scaling + lead 7.6× TMPA permeability is an independent supporting check.
CYP3A4 ADMET flag on the affinity lead. Borderline: 90.4 vs 90.0 gate. The two-track committee includes the ADMET-clean adamantyl backup specifically to derisk this. Real CYP3A4 inhibition would create DDI exposure window. In-vitro CYP panel after first synthesis; Gate 2.5.
No physical compound and no STRC reagent yet. Acknowledged; the entire ask is RFQ-scoping for synthesis + fragment expression + DSF/SPR. Empirical work is the only thing that can move h01. Wetlab Handoff Packet 2026-04-27 is RFQ-ready.
Pharmacochaperone-via-intratympanic for ER misfolding rescue is a novel combination. Class works (Tafamidis, Migalastat, Trikafta — all FDA). Route works (IT steroids/gentamicin daily clinical practice; AK-OTOF AAV via RWM restored hearing in deaf DFNB9 children at CHOP, Oct 2023). The combination — small-molecule pharmacochaperone delivered intratympanically to rescue ER-misfolded cochlear protein — has not been clinically demonstrated. First-in-class regulatory + manufacturing learning curve; PK in OHC cytoplasm is modeled, not measured. Migalastat-style amenability assay (DSF/nanoDSF Gate 1) closes target-engagement question wet, not in silico. The route question is already closed by Akouos AK-OTOF; the class question is closed by Galafold/Vyndaqel/Trikafta.

Class precedents — has anyone made this kind of drug?

Yes. Four FDA-approved pharmacochaperone / corrector programs follow exactly the architecture h01 is set up to repeat: matched WT/MUT structural reasoning → small-molecule fold stabilizer → wet amenability assay → allele-by-allele label expansion. The combined precedent class now serves the majority of patients in three orphan diseases (TTR amyloidosis, Fabry, cystic fibrosis). Delivery-route precedents are covered separately in the next section.

Tafamidis (Vyndaqel / Vyndamax)

Pfizer
FDA 2019 (cardiomyopathy); EMA 2011 (polyneuropathy)
Target

Transthyretin (TTR) tetramer

Mechanism

Small-molecule stabilizer of the TTR dimer-dimer interface; blocks tetramer dissociation that drives misfolding and amyloid deposition.

Allele coverage

Wild-type ATTR + 100+ pathogenic TTR variants (V122I, V30M, T60A, …) — single molecule covers the entire protein-class disease.

Citation

Coelho 2012 NEJM (Fx-005); Maurer 2018 NEJM ATTR-ACT.

Why it matters for h01

Direct architectural precedent: matched WT/MUT ensemble + Coulomb-aware ligand design + first-in-class FDA approval for a misfolding pharmacochaperone. Used as the design reference in Phase 4h (Tafamidis Playbook Library).

Migalastat (Galafold)

Amicus Therapeutics
FDA 2018; EMA 2016
Target

α-galactosidase A (GLA) — Fabry disease

Mechanism

Allele-specific iminosugar pharmacochaperone that binds the GLA active site, stabilizes folding in the ER, and rescues lysosomal trafficking of misfolded variants.

Allele coverage

FDA-approved companion diagnostic (HEK assay) defines an "amenable mutations list" — currently ~35–50% of Fabry patients across 300+ GLA missense variants. The list expands as new variants are tested in vitro.

Citation

Germain 2016 NEJM 375:545; Wu 2011 Hum Mutat (HEK amenability assay).

Why it matters for h01

Cleanest one-to-one model for h01: one molecule, allele-by-allele clinical expansion via a wet amenability assay (= our Gate 1 DSF/nanoDSF). E1659A is the first STRC entry; the same DSF assay reads other STRC missense alleles into the list.

Ivacaftor (Kalydeco)

Vertex Pharmaceuticals
FDA 2012 (G551D); label expanded to ~38 gating mutations
Target

CFTR — cystic fibrosis

Mechanism

Potentiator of the CFTR ion channel that increases open-channel probability; rescues function of partially-folded CFTR variants at the cell membrane.

Allele coverage

Started at ~5% of CF patients (G551D only); grew to ~10% via in-vitro–driven label additions over a decade. Same molecule, expanding allele list.

Citation

Ramsey 2011 NEJM 365:1663 (STRIVE); De Boeck 2014 (KONNECTION expansion).

Why it matters for h01

Precedent for in-vitro–driven label expansion: a single molecule progressively earns reimbursable indications by passing a defined biochemical assay on each new allele.

Elexacaftor / Tezacaftor / Ivacaftor (Trikafta / Kaftrio)

Vertex Pharmaceuticals
FDA 2019 (≥12 yr); FDA 2021 (≥6 yr); FDA 2023 (≥2 yr)
Target

CFTR — cystic fibrosis

Mechanism

Triple combination of two correctors (folding) + one potentiator. Together they rescue F508del — the dominant misfolding allele — plus 177+ other CFTR mutations.

Allele coverage

~90% of all CF patients globally; transformed CF from a high-mortality disease into a manageable chronic condition within a decade.

Citation

Middleton 2019 NEJM 381:1809; Heijerman 2019 Lancet 394:1940.

Why it matters for h01

The end-state of the platform: one program, one company, one disease — eventually serving the overwhelming majority of patients. Multiple molecules, one mechanism class. Direct counter-argument to "one allele = one patient".

Delivery precedents — does the drop actually reach the cochlea?

The cochlea is buried in temporal bone, but the middle-ear cavity is accessible to any ENT clinic and the round-window membrane between the two is thin (~70 µm in humans) and permeable. The trans-tympanic → RWM → perilymph route is FDA-permitted, in daily clinical use, and has just been validated in human gene-therapy trials that restored hearing in deaf children.

Headline proof point

Akouos AK-OTOF (Eli Lilly) and Decibel DB-OTO (Regeneron) dosed their first DFNB9 patients in late 2023 via the same round-window route. Both reported functional hearing restoration in genetically deaf children within months. If a ~25 nm AAV capsid delivered through the RWM reaches outer + inner hair cells well enough to restore hearing in humans, a ~500 Da small molecule reaches those same cells with substantial margin.

Jeffrey Holt (Boston Children's / Harvard) — scientific co-founder of Akouos — is the field's anchor PI for this delivery route. Targeting his group with the h01 dossier is consistent with the existing OTOF programme architecture.

Intratympanic methylprednisolone / dexamethasone

FDA-permitted off-label; daily clinical practice for >20 years
Indication

Sudden sensorineural hearing loss (SSNHL); standard-of-care across ENT clinics globally

Route

Trans-tympanic injection → middle-ear cavity → diffusion across round-window membrane (RWM) into scala-tympani perilymph

Citation

Rauch SD et al. JAMA 2011;305(20):2071-2079 — RCT showing IT methylprednisolone non-inferior to oral steroids for SSNHL.

Why it matters for h01

Demonstrates that the trans-tympanic → RWM → perilymph route is anatomically reachable in awake adult patients with a needle in an ENT clinic, not a research-only procedure.

Intratympanic gentamicin

Standard-of-care since 1990s; thousands of patient-years
Indication

Ménière disease — chemical vestibular ablation

Route

Trans-tympanic injection of low-dose aminoglycoside; selectively ablates vestibular hair cells via RWM permeation, sparing cochlear hair cells at controlled doses.

Citation

Pfaltz CR & Thomsen J. Acta Otolaryngol Suppl reviews; long-term outcome studies through Cochrane CD008234.

Why it matters for h01

Proves that small molecules delivered intratympanically reach hair cells in pharmacologically-active concentrations — a small molecule does cross the RWM and engage cells inside the cochlea.

OTO-104 / Otividex (sustained-release dexamethasone hydrogel)

Two Phase 3 trials completed (AVERTS-1, AVERTS-2)
Indication

Ménière disease (Otonomy Phase 3; programme later wound down at company level for commercial reasons, not safety)

Route

Single intratympanic injection of poloxamer-407 thermo-gel; sustains middle-ear dexamethasone concentration for weeks; same RWM diffusion route as plain IT injection.

Citation

Otonomy clinical-trial register NCT02612337, NCT02265393.

Why it matters for h01

Prior art for the formulation upgrade we will need eventually: a single in-clinic injection that holds therapeutic middle-ear concentration for days–weeks instead of hourly drops.

AK-OTOF (Akouos / Eli Lilly) and DB-OTO (Decibel / Regeneron)

FIRST PATIENTS DOSED October 2023 (CHOP, Akouos AK-OTOF) and 2023 (DB-OTO); functional hearing restoration reported in deaf children within months. Published 2024.
Indication

Hereditary deafness DFNB9 (OTOF gene)

Route

AAV1-delivered OTOF cDNA via trans-mastoid round-window injection in the operating room; the AAV particle is far larger than a small molecule yet still reaches outer + inner hair cells.

Citation

Lustig L, Holt JR et al, AK-OTOF / CHOP first-patient case (Akouos pipeline; results presented ASGCT 2024 and published in major journal 2024). DB-OTO: Decibel/Regeneron case series 2024.

Why it matters for h01

STRONGEST proof-point in the entire dossier. If an ~25 nm AAV capsid passes the RWM and infects hair cells well enough to restore hearing in human DFNB9 children — a ~500 Da small molecule will reach those same cells with substantial margin. This is the precedent that turns "the cochlea is too hidden" from concern into solved problem.

Honest residual delivery risks

The route is real, but three specific delivery questions are not closed by precedent alone and remain in scope for the wetlab feasibility package.

Open risk

Apex-vs-base concentration gradient

Detail

Compounds entering through the RWM concentrate at the basal turn of the cochlea (high frequencies). Salt and colleagues have measured 100–1000× basal-vs-apical gradients for some markers. If STRC-dependent mechanics are needed apically, simple IT dosing will under-treat the apex.

Mitigation / next check

Sustained-release / posterior-semicircular-canal injection / canalostomy can flatten the gradient. Clinical-stage formulations exist (OTO-104 class).

Open risk

Between-patient variability (10–100×)

Detail

RWM thickness, mucosal swelling, middle-ear effusion, and patient anatomy all change effective permeability. Single-dose drops show wide PK spread in published data.

Mitigation / next check

Dose-finding in feasibility cohort; sustained-release reduces peak/trough variability; middle-ear pre-treatment protocols (mucosa-thinning, posture) used clinically.

Open risk

Perilymph ≠ cytoplasm

Detail

Reaching scala-tympani perilymph is necessary but not sufficient: the molecule still must cross the apical hair-cell membrane and reach the ER, where the misfolded STRC is. For our ~500 Da, logP 1.94, weak-anion lead, passive diffusion is the expected route, but this has not been measured for our specific compound.

Mitigation / next check

Gate 0–2 of the assay cascade addresses target engagement directly (DSF, SPR/BLI/MST on STRC fragment). Cell uptake then directly tested at Gate 3 (HEK293 full-length E1659A trafficking rescue) before any explant or animal work.

From one allele to platform

The molecule on this page is allele-specific. The pipeline behind it is not. STRC E1659A is the first worked example. Each new allele either joins the amenable list (passes Gate 1 DSF Δ) or fails cleanly — exactly the Migalastat (Galafold) model that grew from one Fabry allele to ~35–50% of all Fabry patients across 300+ GLA missense variants.

Plain-language framing

H01 is not a "treatment for one child." H01 is the first allele in a pharmacochaperone program that, if the assay cascade works, expands one allele at a time the same way Migalastat, Ivacaftor, and Trikafta did.

STRC-related hearing loss (DFNB16) is the second-most-common genetic hearing loss after GJB2. The addressable population grows in three nested layers: (1) E1659A homozygous + compound-heterozygous carriers; (2) any STRC missense allele that passes the DSF amenability assay; (3) DFNB-wide pharmacochaperone targets that fit the same pipeline.

The reusable pipeline

Five steps. Every step is gene-agnostic and rerunnable on a new allele in under a wall-week of M5 Max compute. The molecule changes per pocket; the steps don't.

1
Step
Allele triage
Input

ClinVar / gnomAD missense list for the target gene; family genotype.

Output

Candidate alleles ranked by predicted ER-misfolding signature.

Reusable as

Same triage script for every STRC missense and any other DFNB gene.

2
Step
Pocket + electrostatic proof
Input

AF3 fragment ladder + matched WT/MUT MD ensemble + APBS pocket-potential resampling.

Output

Δ formal-anion preference (kcal/mol) with effect size and p-value.

Reusable as

Phase 5d/5k pipeline reruns end-to-end on a new allele in <1 wall-week on M5 Max.

3
Step
Coulomb-aware chemistry
Input

Allele-specific pocket charge profile + Halgren druggability gates.

Output

Triaged ligand library (Lipinski + TPSA + mono-anion gates) ready for Boltz/Vina/tauRAMD.

Reusable as

Phase 8 v5 library design framework; tafamidis-playbook scaffold rotation per pocket.

4
Step
Lead committee
Input

Boltz-2 affinity + mutant-ensemble Vina + tauRAMD residence + ADMET-AI.

Output

Two-track committee (affinity lead + ADMET-clean backup) with α-sensitivity check.

Reusable as

Phase 9x α-sensitivity script generalizes to any per-allele ranking.

5
Step
Wetlab handoff (RFQ)
Input

Lead committee + protein construct sheet + assay cascade.

Output

compound_sheet.csv + gap_matrix.csv + rfq_email.md (Wetlab Handoff Packet 2026-04-27).

Reusable as

Same RFQ cascade (Gate 0 → 3) reads every new allele into the amenable list (Migalastat model).

Three expansion tracks

Three nested horizons for how the same pipeline grows the addressable population beyond E1659A.

STRC missense expansion

Year 1

Run the same Phase 5k/5e/8c/9x pipeline on the next STRC missense alleles in ClinVar with predicted K1141-zone or analogous pocket destabilization. Each allele either joins the amenable list (DSF Δ ≥ threshold) or fails cleanly.

Examples / scope

STRC ClinVar missense set (literature pull required); compound-heterozygous combos with truncating null alleles broaden the addressable population without requiring homozygous missense.

Outcome

Per-allele decision: amenable / not amenable, exactly the Migalastat model.

DFNB-wide pharmacochaperone

Year 2–3

Apply the pipeline to other DFNB hearing-loss genes where a missense allele creates a dockable pocket and ER misfolding is the disease driver. Targets to triage first: GJB2 (Cx26 missense subset), OTOA, OTOG, OTOF, MYO7A.

Examples / scope

GJB2 alone is the most common hereditary HL gene globally; even a 10–20% missense subset would dwarf STRC patient numbers.

Outcome

A pharmacochaperone platform branded around hereditary HL, not a single allele.

Pocket-driven proteinopathy class

Year 3+

The matched-ensemble APBS + Coulomb-aware ligand design framework is gene-agnostic: any monogenic missense disease where the mutation creates a charged pocket destabilization is in scope.

Examples / scope

Same mechanism class as Tafamidis (TTR), Migalastat (GLA), Trikafta (CFTR). The pipeline is gene-class agnostic.

Outcome

Cross-disease platform play; STRC E1659A becomes the published worked example.

Allele-screening criteria

A new allele enters the program only if all five criteria are met. This filter prevents wasted compute on null alleles or non-druggable pockets.

Missense allele, not null or large deletion.

Residual protein is made and plausibly foldable.

A local pocket or allosteric stabilizer site can be modeled.

Target engagement can be measured before cell rescue.

The stabilizer does not permanently block the protein function it is meant to restore.

Current claim boundary

H01 is ready to present as a computationally promoted, RFQ-packaged hypothesis for first synthesis + STRC fragment expression + DSF/SPR. It is not yet a validated drug, lead candidate, or clinical-readiness claim.

Read the page as the strongest current case for starting wetlab work, with matched-ensemble APBS as the gate-closing electrostatic evidence and Phase 8h-lite as the closing paper-claim package. Production holo MD/MM-PBSA remains an optional paper-grade upgrade — it does not block first wetlab contact. Reviewer-pack prose cannot move ranking; only new computational or experimental evidence can change h01 tier, mech, deliv, or misha_fit.

← All hypotheses