STRC Calcium Oscillation Acoustic Therapy

Chronic above-threshold sound, delivered through a hearing aid, drives OHC apical Ca²⁺ into the AC1-CREB pathway and tonically up-regulates endogenous STRC transcription. No gene, no drug, no surgery — the therapy is sound itself. Expected STRC protein fold 1.3-2.6× at 45-60 dB LAeq. Mechanism, monotonicity, and parameter robustness (index 0.91) all computationally confirmed (2026-04-20).

The hidden assumption this breaks

Earth treats sound as input to be measured (audiogram, ABR, DPOAE). Nobody treats sound as regulatory signal at the transcriptional layer. Yet Ca²⁺ signalling is one of the best-characterised signal-transduction codes in biology (Dolmetsch 1998, De Koninck & Schulman 1998) and the OHC is uniquely positioned to receive that signal from acoustic input.

The chain (post Phase 1B pivot)

Acoustic input (any broadband sound ≥45 dB)
→ MET channel opening → apical Ca²⁺ rise (to ~500 nM at 45 dB, ~1 µM at 60 dB)
→ Ca₄·CaM binds AC1 → cAMP rises (~350 nM steady)
→ PKA holoenzyme dissociation → PKAc phosphorylates CREB at Ser133
→ CRE-driven STRC transcription (1.3–2.6× induction)
→ endogenous STRC protein increase
→ HTC reformation → Hopf amplifier recovery

For Misha, the relevant allele is the c.4976A>C (E1659A) maternal one. The protein made from this allele is folded; it just binds weakly. More of it → occupancy saturates even weak binding. Sound-driven up-regulation squeezes therapeutic utility out of the mutant allele he already has.

Mechanism notes

Initial Phase 1 model (CaMKII vs CaN frequency-decoder) gave wrong directionality — CaN saturates before CaMKII at OHC sub-µM Ca²⁺, so RBM24-P drops with sound. Pivoted to AC1-CREB in Phase 1B; that pathway is monotonic by construction (AC1 is strictly Ca²⁺-activated via Ca₄·CaM). Details and ODE results in STRC AC1-CREB Monotonic Sound Response. Parameter sensitivity confirmed robust (index 0.91) in STRC AC1-CREB Parameter Robustness.

AM modulation is not the active ingredient. CREB cycles on ~2 min dephosphorylation timescale — too slow to frequency-decode sub-Hz FM. The cochlea is a long-time-constant transcriptional integrator in this regime, not a frequency decoder. Touch Grass acoustic protocol simplifies accordingly: chronic LAeq, not carrier FM.

Acoustic protocol

ParameterTarget
Daily LAeq dose45-60 dB, 1-3 h/day minimum (6-8 h ambient target)
Carrier frequency2-4 kHz (OHC peak sensitivity; matches Misha’s moderate-loss region)
Safety ceiling<85 dB 8-hr LAeq (NIOSH) to avoid NIHL
FM / AM shapenone required (content doesn’t matter — LAeq dose is active ingredient)
Duty cyclecontinuous during waking hours OR 30-60 min dedicated sessions + ambient overlay
Expected onsetpCREB 6-24 h; STRC protein 2-4 weeks

Three delivery modes, escalating in user engagement:

  1. Ambient overlay — hearing-aid DSP or Touch Grass iOS app ensures input stays ≥45 dB whenever Misha is wearing the aid. All-day coverage, zero user burden.
  2. Dedicated session — 30-60 min/day of therapeutic soundscape (Touch Grass). Active engagement, higher SNR.
  3. Hybrid — dedicated dawn/dusk + ambient during day. Recommended for induction phase.

Touch Grass integration — same instrument, new mode

Touch Grass already produces controllable audio across multiple engines (RainStick, OceanDrum, GrainEngine, ModalOscillatorBank). The “STRC therapy mode” is a software flag — no new device. Since content doesn’t matter (Phase 1B finding), any Touch Grass engine output at LAeq ≥45 dB is therapeutic.

Delivery chain

Touch Grass iOS app → iPhone audio → BLE Audio / MFi profile → Hearing aid →
acoustic output into ear canal → tympanic membrane → ossicles →
basilar membrane → OHC stereocilia → MET channels → Ca²⁺ → AC1-CREB → STRC

Firmware integration

  1. App-level (ship next release): add “STRC Therapy” engine in Touch Grass. Ships as audio update, no hardware change.
  2. OS-level: no change — iOS ↔ BLE Audio ↔ MFi hearing aid is standard.
  3. Aid-level (optional): if manufacturer opens DSP hook (Phonak Target, ReSound Smart Fit, etc.), load an “STRC therapy” program that keeps SPL ≥45 dB.
  4. Monitoring hook: Touch Grass logs LAeq exposure duration per day. Feed weekly DPOAE / monthly ABR.

Session design for Misha

  • Induction (first 3 mo): 2 × 45-min daily active sessions + 4-6 h ambient overlay. DPOAE weekly, ABR monthly.
  • Maintenance (mo 4+): ambient overlay 4-8 h during waking hours. Dedicated sessions optional.
  • Evaluation gate at 6 mo: see kill criteria.

Why the hearing aid is the perfect dosing device

  • Worn daily — compliance automatic
  • Titrated to Misha’s loss profile — SPL at OHC is correct by design
  • Acoustically precise — known frequency response
  • Pediatric-safe — decades of pediatric hearing-aid safety data
  • Reversible — stop the therapy mode, the dose stops. No residual.
  • Titratable — LAeq envelope tunable in minutes per session

The therapy is the device Misha already owns. Zero extra clinical burden.

Kill criteria and go/no-go gates

GatePassKillAction on kill
Phase 1 (ODE directionality)Monotonic Ca → STRC protein upNon-monotonic or wrong signPivot pathway (done 2026-04-20 Phase 1B, passed)
Phase 2 (parameter robustness)Robustness index ≥0.7<0.4Hypothesis parameter-fragile (done 2026-04-20 Phase 2, index 0.91 passed)
Phase 3 (pCREB IHC)≥3× OHC nuclear pCREB signal with sound<1.5×Not AC1-CREB-mediated — reconsider
Phase 4 (STRC mRNA qPCR)≥2× STRC mRNA at 6-24 h<1.5×CRE-STRC coupling weak — pin K_TXN_MAX_FOLD
Phase 5 (chronic 4-wk exposure + ABR)≥5 dB DPOAE or ABR improvement @ 2 kHzNo improvementPivot to STRC Pharmacochaperone Virtual Screen E1659A or AAV path
Phase 6 (pilot in Misha, 6 mo)≥5 dB ABR improvementNo improvementNo harm done (sound was always safe); switch to small-molecule or gene therapy

Resources and validation partners

Compute: trivial — 7-variable ODE runs on MacBook. Already complete.

Wet validation partners:

  • Jeffrey Holt lab — OHC explant Ca²⁺ imaging, pCREB IHC, DPOAE. Warm from Derstroff 2026.
  • Shu lab — explant capacity (Tang Honghai), in-vivo mouse STRC KO.
  • Audiology (Misha) — already scheduled via his audiologist.

Budget:

Line itemCost
Computational$0 (complete)
pCREB IHC + STRC qPCR in explants$3-8 k reagents (lab covers labor)
Touch Grass firmware (internal)$0 marginal
Audiology monitoringalready covered
Total to pilot-in-Misha (6-mo evaluation)$6-13 k

By far the cheapest of the three top alien hypotheses. No drug synthesis, no gene delivery, no nanoparticle fabrication — sound through a device he already wears.

Risks and mitigations (condensed)

RiskMitigation
E1659A residual affinity too weak — 2× upregulation insufficientCombine with STRC Pharmacochaperone Virtual Screen E1659A (quantity × affinity additive)
Global AC1-CREB activation has off-target footprintTranscriptome monitoring; restrict to cochlear OHCs via localised SPL dose
Chronic exposure damages hearingStay <85 dB 8-hr LAeq (NIOSH); audiological monitoring
Slow onset (30-day protein t½) → no detectable change for 2-3 months6-month evaluation window; chronic maintenance framing
Placebo / expectation bias in pilotBlind LAeq-on/off periods via hearing-aid remote; DPOAE as objective measure
Adult OHC expression of AC1 lowWu 2011 shows AC1 is expressed in adult cochlea; confirm by scRNA-seq atlas

Why not already published

  • Cochlear researchers rarely study intracellular Ca²⁺ → transcription — that vocabulary belongs to immunology (T-cell) and neuroscience (hippocampal LTP).
  • Immunologists and neuroscientists rarely study cochlear transcription.
  • The bridging question — “can sound regulate cochlear transcription via classical Ca²⁺ signalling?” — falls through the disciplinary gap.
  • Sonogenetic researchers focus on AAV + synthetic promoter (Wu 2023, Pan 2018). They don’t ask “can endogenous transcription be driven by ambient sound alone?”

The elegance

Misha wears a hearing aid. The aid receives broadband sound. That sound’s LAeq dose drives AC1-CREB → STRC transcription in the same OHCs that are the target of therapy. The organ receiving the therapy is the organ it treats. Maximum substrate-content alignment. No device, no drug, no DNA to deliver.

Files / Models

Full phase results in child notes:

Scripts + raw outputs in ~/STRC/models/:

  • ca_oscillation_rbm24_ode.py / _results.json — Phase 1 (superseded, retained for audit)
  • ca_oscillation_ac1_creb_pivot.py / _results.json — Phase 1B AC1-CREB ODE
  • ca_oscillation_phase2_sensitivity.py / .json / .png — Phase 2 robustness
  • ca_oscillation_phase3_bifurcation.py / .json — Phase 3 eigenvalue + hysteresis + smooth-Ca control

Connections