STRC Research

STRC mRNA-LNP PKPD v2 OHC Tropism

8 min read

STRC mRNA-LNP PK/PD v2 — 1-3% OHC tropism + ER/UPR ceiling refines Delivery 2/5

Question

STRC LNP Cochlear Tropism Literature Scan (2026-04-22) confirmed the Strategy B Delivery score 2/5 was correct: 5% cochlear-tropic LNP-mRNA chassis is plausible-but-unbuilt (12-18 mo engineering), 20% OHC-targeted is aspirational, 50% is sci-fi. The first-pass v1 PKPD model (STRC mRNA-LNP Strategy B Full-Length) used 0.8 % / 5 % / 20 % tropism rows. Two refinements were flagged as next-step:

  1. Add a realistic 1-3 % tropism row between v1’s untargeted floor and the aspirational 5 %. This is the chassis-engineering window peptide-prestin-A666 LNP-mRNA hybrids could plausibly reach (Wang 2018 demonstrates ~1 % OHC drug-payload targeting in vivo).
  2. Add an ER/UPR ~5× soft ceiling on per-OHC STRC fold. Chronic mRNA-driven exogenous protein production above ~5× endogenous triggers unfolded-protein-response (UPR), capping benefit independent of dose. OHCs almost certainly have lower ER capacity than hepatocytes (small cytoplasm, heavy mechanical load).

Question: do these two refinements change the cochlea-mean fold attainable under any realistic regimen? Do they alter the Strategy B Delivery 2/5 verdict?

Method

mrna_lnp_pkpd_integration_v2_ohc_tropism.py extends the v1 3-compartment ODE (m → p → f) with:

  • New tropism levels: realistic_1pct = 0.01 and realistic_3pct = 0.03, sitting between untargeted = 0.008 and cochlear_tropic = 0.05.
  • New ER soft cap on the f_target term: f_target_capped = 1 + (ER - 1) · tanh((f_target_uncapped - 1) / (ER - 1)) with ER = 5.0. Maps:
    • 1× → 1× (identity)
    • 5× (the ceiling) → 4.05×
    • ∞ → 5× asymptote
  • Per-regimen flag er_stress_risk = (peak_per_OHC_fold ≥ 0.8 · ER) = ≥ 4.0×.

Sweep: 5 tropism levels × 2 mRNA modifications × 6 dose intervals × 5 per-dose magnitudes = 300 regimens. Runtime ~30 s on CPU.

Result — Strategy B confirmed: cannot reach therapeutic cochlea-mean fold at any realistic tropism

Coverage table (best regimen per tropism, m¹Ψ-modified mRNA):

TropismOHC fractionTheoretical ceiling cochlea-meanMax attainable cochlea-mean foldn therapeutic per-OHCn ER-stress flagged
untargeted0.8 %1.0161.015390
realistic_1pct (new)1.0 %1.0201.018390
realistic_3pct (new)3.0 %1.0601.055390
cochlear_tropic5.0 %1.1001.092390
OHC_targeted (sci-fi)20.0 %1.4001.370390

The therapeutic threshold is 2.0× cochlea-mean fold (per STRC Rescue Threshold). Even the sci-fi 20 % OHC-targeted scenario only reaches 1.37× — a 32 % shortfall against the threshold.

The 1-3 % realistic window gives 1.018-1.055× cochlea-mean fold — essentially indistinguishable from the 0.8 % untargeted baseline at the cochlea level. The per-OHC PKPD is unchanged (each transfected OHC reaches 2.4-2.8× peak), but the tropism-weighted population mean is dominated by the (1 - eff_ohc) ≈ 97-99 % of OHCs that receive zero LNP.

ER/UPR ceiling does not bind in current parameter regime: peak per-OHC fold across all 300 regimens is 2.37-2.85×, well below the 4.0× flag threshold. The 5× soft cap is correctly future-proofed for higher MAX_BOOST scenarios but does not modify v2 outcomes. 0 / 300 regimens flagged for ER stress risk.

Best dose-efficient regimen (m¹Ψ, T=42 days, D=200 mol/OHC intracellular): per-OHC tx_trough_fold = 2.10×, peak = 2.66×, annual extracellular dose = 90,000 mol/OHC. Tropism doesn’t change this — only the cochlea-mean weighting.

Why the cochlea-mean ceiling is so hard

The arithmetic is brutal: cochlea-mean fold = eff_ohc × tx_fold + (1 − eff_ohc) × 1.0 . With max attainable per-OHC tx_fold ≈ 3.0× (the v1 MAX_BOOST), the cochlea-mean ceiling = 1 + 2 × eff_ohc. To clear 2.0× cochlea-mean threshold needs eff_ohc ≥ 0.5 (50 % OHCs transfected) — far beyond any plausible LNP technology.

This is the core reason Strategy B Delivery sits at 2/5 and not higher. The model now makes that explicit. Three theoretical paths to clear the 2.0× cochlea-mean gate without 50 % tropism:

  1. Raise per-OHC MAX_BOOST beyond 3 (e.g., dual-vector RBM24 + something, or stronger 5’-UTR/Kozak). At MAX_BOOST = 6, cochlea-mean = 1 + 5 × eff_ohc clears 2× at eff_ohc = 0.2 (the OHC-targeted scenario). This would trigger ER stress under v2 model — peak per-OHC at 6× exceeds the 4× flag threshold, capping back to ~4.5× via the tanh, partially relieving the gain. Net realistic ceiling probably stays below 1.5× cochlea-mean even with MAX_BOOST = 6.
  2. Change therapeutic threshold from 2× to 1.3×. If STRC Rescue Threshold is wrong (or audiogram coupling is steeper than STRC ABR Transfer Function Model predicts), 1.3× cochlea-mean might be enough — that’s reachable at OHC_targeted 20 %. Speculative, requires re-deriving the rescue threshold from real OTOF clinical data, not assumed.
  3. Strategy B + AAV Mini-STRC stack: Strategy B contributes ~5-10 % cochlea-mean fold, AAV Mini-STRC contributes ~60 % per-OHC at 60-100 % transduction (per OTOF clinical precedent). Stacking makes Strategy B a re-dosable maintenance layer on top of AAV. This re-positions Strategy B from primary therapy to long-term maintenance/boost — which matches the original STRC mRNA-LNP Strategy B Full-Length framing as “non-curative but durable” and is consistent with the Mini-STRC delivery 5/5 upgrade from 2026-04-21.

Path 3 is the realistic positioning. Strategy B is not a standalone primary therapy; it is a maintenance layer to durable AAV-mediated rescue. This was implicit in the original framing; v2 makes it numerically explicit.

Implications for the hypothesis

Strategy B mRNA-LNP holds at S-tier (still active compute) but the next-step shifts from “model the 1-3% tropism row + ER ceiling” → “frame Strategy B as AAV Mini-STRC maintenance layer in the integrated stack model”. The numbers above are sufficient to retire the standalone-Strategy-B framing.

Specific actionable items:

  • Update Strategy B Delivery interpretation: 2/5 score reflects “alone subtherapeutic, useful as adjunct to AAV.” This was the original meaning but stated weakly in STRC mRNA Therapy Hypothesis (which was the precursor Strategy A, not B). Make the same point explicitly for B.
  • Mini-STRC + Strategy B stack model: not yet built. Required input is realistic AAV transduction rate (Anc80L65 baseline ~60 % OHC at 1×10¹³ vg, Bedrosian 2025 OTOF data) and Strategy B as additive on the un-AAV’d 30-40 % fraction. Approximate: stack reaches ~60 % AAV @ 3× per-OHC + 40 % un-AAV with Strategy B 5 % @ 2.5× → cochlea mean = 0.6 × 3 + 0.4 × (0.05 × 2.5 + 0.95 × 1) = 1.8 + 0.42 = 2.22× cochlea-mean fold — clears the 2.0× threshold by stacking.
  • Mark v1 → v2 superseded: keep v1 for historical comparison but new analyses use v2.

Limitations

  • ER ceiling parameter: 5× is a literature-informed guess, not OHC-specific data. Could be 3× (more conservative) or 10× (more optimistic) — neither end changes the fundamental cochlea-mean-fold conclusion (delivery weighting dominates).
  • Hill K: still inherited from v1 dose-response paper. Not refined.
  • No formal stack model yet: the AAV + Strategy B integrated model is sketched in the implications section but not implemented in code. Would be next-step.
  • No safety / off-target compartment: assumes all delivered mRNA reaches its intended OHC; no liver / hepatic capture model. For LNP-mRNA in vivo, ~70-90 % of dose typically lands in liver — so the “annual extracellular per-OHC” numbers vastly underestimate total-body dose burden. Real-world dosing constrained by liver tolerance, not OHC PKPD.
  • m¹Ψ vs unmod assumed equivalent for max boost: in reality m¹Ψ adds modest translation efficiency boost (~1.3-1.5×) on top of stability gain. Not modeled.

Files

  • Driver: ~/STRC/models/mrna_lnp_pkpd_integration_v2_ohc_tropism.py
  • Output JSON: ~/STRC/models/mrna_lnp_pkpd_integration_v2_ohc_tropism.json
  • v1 (preserved): mrna_lnp_pkpd_integration.py + mrna_lnp_pkpd_integration.json

Ranking delta

  • STRC mRNA-LNP Strategy B Full-Length: no tier change (stays S). Mechanism still 3/5, Delivery still 2/5, Misha-fit still 5/5. Evidence depth +1 (1-3 % tropism row added; ER ceiling future-proofed). The Delivery 2/5 score is now numerically anchored: even sci-fi 20 % tropism gives only 1.37× cochlea-mean (vs 2.0× threshold). Next step changed from “(a) extend Strategy B PK/PD model with 1-3% OHC tropism row + ER/UPR ~5× soft ceiling; (b) ask Holt for unpublished cochlear LNP groups” → “(a) build integrated AAV + Strategy B stack model — Strategy B as maintenance layer on un-AAV’d OHC fraction; (b) ask Holt for unpublished cochlear LNP groups (carry-over).” Stack model preliminary estimate: 2.22× cochlea-mean clears the threshold.
  • STRC mRNA Therapy Hypothesis (Strategy A / RBM24): no change. Same delivery-layer constraint applies; same maintenance-layer reframing logic.
  • STRC Mini-STRC Single-Vector Hypothesis: no change but reinforced as the dominant paternal-allele therapy. Strategy B’s role as adjunct rather than primary makes Mini-STRC the foundation of any realistic combination therapy.
  • STRC Rescue Threshold: no change but flagged for reconsideration — if 1.3× cochlea-mean is sufficient (vs assumed 2×), all delivery hypotheses jump tier. This is a meta-prior worth re-deriving from OTOF / DFNB9 clinical audiograms.
  • All other S/A/B/C tier hypotheses: no change.

Connections

Graph View

Backlinks