Skip to content

Chassis-Pending Interventions

Interventions that hit Open Enzyme chokepoints with mechanism validated, but the engineering / implementation / delivery chassis is not yet identified.

These are promising things that map to what we want to do. The mechanism is real. The target chokepoint is documented. The chassis question is open. They are not deprioritized, not off-platform, not "rejected for not fitting koji." They are tracked here so the next time someone asks "what could we be missing?" the answer is visible rather than scattered.

Why this page exists

Open Enzyme is a chokepoint-first, chassis-second platform. The platform mission is to disrupt the gout / NLRP3 / urate-disposal cascade; engineered A. oryzae koji is the highest-priority chassis because it harmoniously hits multiple chokepoints in one strain (uricase + lactoferrin + carnosine + DAF SCR1-4) and matches the food-grade / open-source / home-fermentable / democratized-access positioning. Koji is one expression of the mission, not the mission itself.

The risk this page mitigates: when a finding lands that hits a real chokepoint but doesn't fit koji, the recommendation step can quietly filter it as "off-platform." That filter is the failure mode the umbrella CLAUDE.md warns against ("don't ask 'does this fit the current chassis,' ask 'what open question might this tool answer?'"). This page makes the no-chassis-filter check structural rather than opportunistic.

The honest status of an entry here: the intervention is real, the mechanism is validated, multiple candidate chassis exist or are conceivable, we don't know which is right yet. Chassis selection is the next question, not the filter that kills the first one. See synthesis/strategic-reflections/2026-05-15-chassis-is-downstream-of-chokepoint.md for the discipline-level reflection that produced this page.

How to read this page

Each entry has a fixed shape:

  • Intervention — what the intervention does, mechanistically
  • Chokepoint(s) hit — which OE chokepoint(s) the intervention targets, with cross-reference
  • Evidence level — Clinical Trial / Animal Model / In Vitro / Mechanistic Extrapolation
  • Why not koji — the specific reason the current koji chassis isn't right (not a value judgment; a mechanism / regulatory / engineering reason)
  • Candidate chassis — the chassis options being considered. The chassis is open. Listed candidates are options, not commitments.
  • Cheapest first move — the next step that doesn't require committing to a specific chassis. Usually a lit scan, a low-cost wet-lab experiment, an n=1 self-experiment, or a partner conversation.
  • Cross-reference — the canonical wiki page(s) where this intervention is treated in mechanism depth

When an entry's chassis question resolves, the entry migrates: either to a new dedicated scope page in the wiki (if a chassis is selected and a new track begins — e.g., engineered-lbp-chassis.md if the chassis is engineered LBPs), or to a peer-track page already in the wiki (e.g., compounding-pharmacy-track.md if the answer is "compounded small molecule"). The chassis-pending page stays as the index for the open questions, not the implementation roadmap.

Current chassis-pending interventions

1. Purine-degrading bacteria (PDB) restoration / engineered PDB pathway expression

Intervention. The 2,8-dioxopurine bacterial gene cluster degrades uric acid anaerobically to butyrate + acetate, hits CP6 (urate degradation) directly, and compounds via SCFA effects on multiple downstream nodes (ABCG2 induction + Q141K trafficking rescue + NLRP3 dampening + XO inhibition). The most common gout-associated ABCG2 variant (Q141K, ~3–15% population frequency) is HDAC-rescued by butyrate — making PDB-derived butyrate a natural genotype-targeted therapy via endogenous gut bacteria producing the molecule that fixes a genetic variant.

Chokepoint(s) hit. CP6 (urate degradation), CP2 (NLRP3 dampening via butyrate), gut ABCG2 induction (PPARγ axis), Q141K trafficking rescue (HDAC inhibition). See purine-degrading-bacteria.md for the full mechanism + evidence inventory.

Evidence level. Animal Model (CBT2.0 engineered EcN, −63% plasma UA in hyperuricemic mice — Li et al. 2025 Life Metabolism PMID 41070194); Human Retrospective Cohort (Stanford n=14K clindamycin vs Bactrim, HR 1.30 for incident gout — Liu et al. 2023 Cell PMID 37541197); Human Observational (FARMM antibiotic depletion n=30 fecal urate +40–50%); Mechanistic Extrapolation (quantitative SUA effect in typical gout patient with intact renal function).

Why not koji. DOPDH (the first-step enzyme) requires SelD selenophosphate synthase — prokaryote-specific, not present in A. oryzae. The pathway is obligate anaerobic; koji fermentation is aerobic. The gene cluster is 8 enzymes; even if SelD were resolved, koji is the wrong substrate (eukaryote with peroxisomal compartmentalization optimized for different chemistry).

Candidate chassis. Multiple, all open: 1. Engineered E. coli Nissle 1917 expressing the full PDB cluster (CBT2.0 precedent in Li 2025) — facultative anaerobe, EcN safety / probiotic record, already used in PULSE uricase work, native SelD present 2. Defined-strain anaerobic probiotic (Clostridium sporogenes, Lacrimispora saccharolytica, Enterocloster bolteae) — naturally express the cluster but oxygen-sensitive manufacturing is a barrier 3. FMT from PDB-rich donors — case reports exist for gout FMT; regulatory pathway exists for some indications 4. Prebiotic enrichment — inulin/FOS/resistant starch enriches PDB-positive Lachnospiraceae and Ruminococcaceae; ~10% SUA reduction in animal/small-human trials; doesn't require an engineered organism 5. Dietary cofactor adequacy (selenium) — selenium-dependent DOPDH runs ~27× faster than the sulfur variant; selenium deficiency could phenocopy PDB depletion without changing bacterial abundance; trivially cheap if relevant

Cheapest first move. Two parallel: - Serum selenium on next blood panel (~$40–80 standard clinical) — answers whether the cofactor side of the question is gating Brian's gut PDB function. Already added to self-experiment-protocol.md §11.0. - Cranberry juice n=1 (4 weeks unsweetened, ~$20) — tests the parallel Alistipes indistinctus / hippuric acid → ABCG2 axis via direct dietary benzoate → glycine conjugation → hippuric acid, without needing bacterial colonization. Different mechanism, same downstream node (ABCG2). See abcg2-modulators.md Alistipes Tier 2.

Cross-reference. purine-degrading-bacteria.md (mechanism), abcg2-modulators.md (PPARγ/ABCG2 axis), gut-lumen-sink.md (PULSE context for EcN chassis option), engineered-lbp-chassis.md (LBP framework for anaerobic options).

2. Kidney-tropic siRNA against URAT1 mRNA

Intervention. Sequence-specific siRNA knockdown of URAT1 mRNA in renal proximal tubule cells, delivered via kidney-tropic conjugate chemistry (folate-receptor, megalin-binding, or related approaches). Eliminates the dose-dependent off-target profile of small-molecule URAT1 inhibitors (benzbromarone hepatotoxicity, lesinurad cardiovascular signal). Quarterly SC dosing precedent from GalNAc-siRNA approvals (inclisiran for PCSK9, patisiran for TTR — both liver-tropic, kidney-tropic chemistry is the active research class).

Chokepoint(s) hit. Renal URAT1 reabsorption — the single largest reabsorption step in the renal urate handling chain. GLUT9 is a parallel target. Renal urate disposal sits on a different mechanism axis from gut-lumen disposal (the koji thesis); the two are complementary, not substitutional. See sirna-urat1-modality.md.

Evidence level. Mechanistic Extrapolation for gout specifically. Clinical Trial precedent for the delivery class (inclisiran, patisiran approved for non-renal targets). No clinical program for URAT1 specifically as of 2026-05-15.

Why not koji. RNA platforms are not produced or delivered via engineered fermentation chassis. Different production stack (chemical synthesis or in vitro transcription), different formulation stack (LNP or GalNAc-like conjugates), different regulatory class (NDA pharma route).

Candidate chassis. Synthetic siRNA + kidney-tropic conjugate, delivered SC. Commercial pharma manufacturing. The OE platform's role is discovery-engine output (we surface URAT1 as a high-value target and the mRNA target-site selection); the chassis is downstream — partner with a pharma company, or spinout development.

Cheapest first move. comp-009 (URAT1 mRNA target site selection via RNAfold + accessibility scoring) — $0, ~1 week, queued in sirna-urat1-modality.md Phase 2.

Cross-reference. sirna-urat1-modality.md, modality-chokepoint-matrix.md (Renal compartment row), delivery-route-matrix.md (RNA platforms × SC cell).

3. Engineered LBP (obligate anaerobe) chassis — F. prausnitzii, Akkermansia, Bacteroides

Intervention. Live biotherapeutic products engineered from gut-native obligate anaerobes. Hits multiple chokepoints depending on the engineered payload: engineered F. prausnitzii for local butyrate at the gut crypt (ABCG2 + Q141K + NLRP3 — same compounding effect as PDB); engineered Akkermansia muciniphila for mucus-layer repair (gut barrier / TNFα leak); engineered Bacteroides for broader metabolic engineering. Solves the "transit organism vs. colonization" problem the koji chassis has — these are gut-resident, durably colonizing.

Chokepoint(s) hit. Depends on engineered payload. Gut ABCG2 induction (via SCFA), gut barrier repair (CP1 LPS / TNFα leak), gut microbiome shaping (community-level). See engineered-lbp-chassis.md.

Evidence level. Mechanistic Extrapolation + Animal Model precedent (Sonnenburg lab Bacteroides editing toolkit; Pendulum probiotic commercial Akkermansia-containing product). LBP regulatory framework (FDA 2018 guidance) defined; clinical programs exist for other indications.

Why not koji. Obligate anaerobic organisms; koji is aerobic fermentation. Manufacturing requires anaerobic bioreactor + cold-chain stabilization — categorically not home-fermentable. Regulatory: Live Biotherapeutic Product framework, not food chassis.

Candidate chassis. F. prausnitzii, Akkermansia muciniphila, Bacteroides (with established engineering toolkits). Commercial pharma manufacturing + distribution + cold chain — explicitly not the democratized-home-access track the koji chassis enables.

Cheapest first move. LBP track Phase 2 lit scans (engineering toolkit + commercial landscape + FDA LBP regulatory path) — queued in engineered-lbp-chassis.md. $0 cost, ~1–2 weeks via subagent.

Cross-reference. engineered-lbp-chassis.md is the canonical scope page.

4. Inhaled mRNA-IL-1RA pulse therapy for acute gout flare

Intervention. Lipid nanoparticle-formulated mRNA encoding IL-1 receptor antagonist (anakinra-equivalent), delivered via pulmonary inhaler. Transient expression matches the short flare window (12–72 hours). The pulmonary surface area (~70 m²) maximizes uptake; mRNA-LNP delivery for pulmonary indications is mature (CF, RSV, asthma research programs). Eliminates SC injection requirement for flare management; cost-competitive with $300K/yr canakinumab if mRNA manufacturing economics hold.

Chokepoint(s) hit. CP5a (IL-1β receptor blockade). Companion target for the existing SC anakinra / canakinumab options. See modality-chokepoint-matrix.md §"Open exploration questions" #5.

Evidence level. Mechanistic Extrapolation. No clinical program in any indication uses mRNA-IL-1RA for flare-window therapy. Adjacent precedents (mRNA vaccines IM, mRNA pulmonary research) establish the chassis feasibility.

Why not koji. mRNA / LNP is not produced or delivered via engineered fermentation. Different production stack, different formulation stack, different regulatory class.

Candidate chassis. Synthetic mRNA + LNP + inhaler device. Commercial pharma manufacturing. Clinical partner required.

Cheapest first move. Mechanism + delivery feasibility lit scan: "mRNA-IL-1RA pulse" + "pulmonary LNP for acute inflammatory indications" — $0, subagent task. Result: either confirms novel territory + bounds the chassis question, or surfaces an existing program OE didn't know about.

Computational gate — comp-033 RED single-dose + comp-036 YELLOW repeat-dose (2026-05-16): comp-033 found single-dose plasma Cmax 0.025 µg/mL = 2% of anakinra benchmark (1.5 µg/mL). comp-036 followed up with multi-dose accumulation + receptor-occupancy framing (IL-1Ra Kd vs IL-1R1: 0.1–10 nM log-uniform per Arend 1990 JCI + Schreuder 1997 Nature crystal — nM regime, not pM as initially speculated). 80%-occupancy plasma threshold: 73 ng/mL median [9–553 p05-p95]. Per-regimen verdict: QD RED (24h troughs drop below threshold; mean occupancy 0.66 but median 0% of 72h flare window above 80%); BID × 4–28 doses YELLOW (median 50–56% of flare window above 80% occupancy; best regimen but doesn't clear 95% high-confidence bar); Loading 2× + QD × 14 YELLOW (median 32% above 80%; first-day boost decays). No regimen clears the GREEN bar at current input uncertainty. Top sensitivities: Kd_nM (ρ −0.69) + translation-efficiency mass ratio (ρ +0.58) + dose (ρ +0.24). Two wet-lab measurements would tip the verdict: (1) integrated translation-efficiency mass ratio in human alveolar epithelium (ferret/NHP inhaled-LNP + BAL protein quant); (2) modern SPR Kd measurement IL-1Ra vs IL-1R1 ectodomain. Full analyses: inhaled-mrna-il1ra-pulse-computational.md (comp-033) + repeat-dose-inhaled-mrna-il1ra-pkpd-computational.md (comp-036).

Operator-relevant reframe (load-bearing): the right clinical comparison is NOT "can repeat-dose inhaled mRNA match anakinra Cmax?" (no — at the corrected Kd range, no current regimen does). The right comparison is "is partial receptor-occupancy × cleaner side-effect profile worth it vs prednisone over recurring flares × decades?" (plausibly yes — pending the two wet-lab measurements above). Prednisone's cumulative burden over 30 years of recurrent gout flares (bone loss, cataracts, adrenal suppression, glucose intolerance, mood/sleep effects) accumulates regardless of efficacy; repeat-dose inhaled mRNA at 50% flare-window occupancy may not "abort" a flare like anakinra but may meaningfully shorten/dampen with negligible cumulative steroid-class burden. The chassis-pending entry stays active under this reframe; partner conversations should lead with the prednisone-displacement argument (clean side-effect profile + cost edge), not anakinra-equivalence. Economic edge holds independently: cost/flare $2–120 vs canakinumab $300K/yr → 50–3,000× cost edge.

Why IL-1Ra over anti-IL-1β monoclonal as the mRNA payload (partner-conversation argument): the alternative mRNA cassette would encode an anti-IL-1β antibody (canakinumab-equivalent). IL-1Ra wins for four reasons that partners ask about:

  1. Mechanism breadth. IL-1Ra blocks both IL-1α AND IL-1β at the single IL-1R1 receptor — broader pathway coverage. Anti-IL-1β monoclonals only neutralize IL-1β. For gout the difference is small (IL-1β is the dominant ligand), but for COPD / ARDS / IPF the cross-indications, IL-1α also drives sterile inflammation, so the broader-mechanism payload is a feature, not a quirk.
  2. Protein size + structure. IL-1Ra is ~17 kDa, no disulfide bonds, no glycosylation required for activity — easier mRNA expression and lung-tissue translation. Antibodies are ~150 kDa with mandatory glycosylation + paired heavy/light chain assembly — substantially harder for transient pulmonary mRNA expression. Translation-efficiency mass ratio (comp-033's dominant sensitivity driver, ρ = +0.78) favors small non-glycosylated payloads by ~10×.
  3. Immunogenicity. Human IL-1Ra is endogenous (body makes its own — see nlrp3-inflammasome.md §"Chokepoint 5"); recombinant IL-1Ra is therefore essentially zero-immunogenicity. Humanized antibodies retain low but non-zero immunogenicity (anti-drug antibody response over chronic dosing).
  4. Cleanness of mechanism. IL-1Ra is purely competitive antagonism — no agonism, no ADCC, no CDC, no off-target effector function. Antibodies have Fc-mediated effector functions (ADCC / CDC / opsonization) that can produce off-target activity in some contexts.

The payload choice is structurally similar to why ankakinra and not canakinumab is the preferred reference for the inhaled-mRNA cassette: same mechanism, smaller protein, broader pathway coverage, lower immunogenicity. Partners evaluating the cassette should land on IL-1Ra not anti-IL-1β.

Cross-indication leverage — the commercial case is NOT gout alone (added 2026-05-16): gout is a low-priority indication for big pharma; nobody develops inhaled mRNA-IL-1RA for gout. But the IL-1 axis is implicated across many indications with much larger markets: COPD exacerbations (~16M US patients, ~$50B annual healthcare cost, clear IL-1β-driven neutrophilic inflammation), severe asthma — T2-low/neutrophilic phenotype (large unmet need; T2-high has biologics like dupilumab/mepolizumab, T2-low does not), ARDS / acute lung injury (~190K US cases/yr, 40% mortality), IPF, CRS from CAR-T, recurrent pericarditis. The regulatory strategy that actually works: approve for a primary indication first (most likely COPD exacerbations or ARDS by market size × mechanism fit), then off-label use spreads to gout — same playbook as anakinra (approved 2001 for RA, now widely off-label for gout / pericarditis / Schnitzler / sJIA / CAPS / COVID-CRS). Implication for partner conversations: the comp-033 Tier-A inhaled-mRNA companies (Arcturus LUNAR-CF, ReCode RCT2100, Ethris/AstraZeneca, Sanofi/Translate Bio) develop for CF/RSV/asthma, not gout. Open Enzyme's role is target validation + the multi-indication cross-leverage argument — making the case to a partner that IL-1Ra has indications beyond their initial target so an mRNA-IL-1RA cassette swap is platform-justifiable. Gout-patient access comes off-label after primary-indication approval, like anakinra. Near-term bridge for gout patients while this 5–10 year development horizon plays out: anakinra SC — see gout-action-guide.md §"This year (advanced)" for the off-label gout protocol (100 mg/day SC × 3 days, NOT intra-articular).

Cross-reference. modality-chokepoint-matrix.md (CP5a × mRNA cell), delivery-route-matrix.md (RNA platforms × inhaled cell), inhaled-mrna-il1ra-pulse-computational.md (comp-033 full analysis), disulfiram.md + gout-action-guide.md (anakinra SC bridge protocol).

5. Bacteriophage-mediated selective gut microbiome modulation

Intervention. Selective phage suppression of LPS-producing gram-negative gut species (CP1 priming relief), purine-fermenting Bacteroides species (substrate reduction upstream of urate generation), or specific dysbiosis patterns identified by 16S/metagenomics. Different from "add an organism" — phages are subtractive ecosystem-shaping, complementary to probiotic addition.

Chokepoint(s) hit. Gut microbiome shaping (community-level); CP1 (NF-κB priming via LPS reduction); upstream substrate reduction for urate generation. See modality-chokepoint-matrix.md (Bacteriophages row).

Evidence level. Animal Model + early-stage clinical for adjacent indications (AMR-associated infection, IBD). No gout programs. Approved in Eastern European jurisdictions; compassionate-use in the US.

Why not koji. Phage manufacturing is a different chassis — bacterial host strains for phage propagation, downstream purification, cold-chain stability. Not a fermentation chassis problem; a viral production chassis problem.

Candidate chassis. Established phage manufacturing companies (Adamas, Locus, Phaxiam, BiomX) for production. Clinical partner for gout-specific indication. Discovery-engine output: OE surfaces which bacterial targets are gout-relevant; production is downstream.

Cheapest first move. Bacteriophage track Phase 1 lit scan: "phage selective gut microbiome modulation × hyperuricemia / gout" — $0, subagent task.

Cross-reference. modality-chokepoint-matrix.md.

6. Intra-articular uricase ± co-formulated catalase for direct tophi dissolution

Intervention. Direct injection of uricase (with co-formulated catalase or as a uricase-catalase fusion protein) into a tophi-bearing joint. Bypasses systemic immunogenicity issue (locally bounded immune exposure), bypasses substrate-access issue at SC depot (tophi ARE concentrated urate at ~100× plasma), bypasses H2O2-in-tissue issue via co-localized catalase (Schiavon / Veronese early-2000s precedent for uricase-catalase fusion). Clinical analog: intra-articular corticosteroid for acute gout flare.

Chokepoint(s) hit. CP6 (uricase mechanism), local tophi dissolution. Sister to the existing IV pegloticase / SEL-212 system at a different delivery target (one specific joint with crystal deposition rather than systemic). See delivery-route-matrix.md §"Open exploration questions" #1.

Evidence level. Animal Model (Pickering emulsion uricase + catalase IA — J Nanobiotechnology 2025 cited in gout-kill-chain-delivery-routes.md); In Vitro precedent for the uricase-catalase fusion class (Schiavon, Veronese). No clinical program.

Why not koji. Whole-cell oral koji is the wrong format for direct intra-articular injection (live organisms in joints → septic arthritis). The chassis-as-formulation advantage of koji for H2O2 housekeeping (peroxisomal co-localization) doesn't transfer to a tissue depot. A purified uricase-catalase fusion protein needs a different production format (recombinant production in any expression system that produces correctly folded fusion protein).

Candidate chassis. Recombinant production of uricase-catalase fusion in any suitable expression host (could be koji-produced as purified protein, could be E. coli, could be Pichia pastoris). Formulation engineering layer (PLGA nanoparticles, Pickering emulsion, hydrogel depot). Clinical partner for the IA delivery trial.

Cheapest first move. comp-NNN protease-stability + folding feasibility analysis of a uricase-catalase fusion construct under shio-koji conditions (extends the comp-006 / comp-007 framework to a chimeric protein). Then a single-construct expression test if the comp-NNN returns LOW risk.

H₂O₂ biochemistry gate — resolved by comp-035 2026-05-16: GREEN across all three architectures. Reaction-diffusion analysis with Damköhler-number coupling, 20,000 Monte Carlo samples per architecture over kinetic / diffusion / geometric / joint-condition priors. Predicted steady-state [H₂O₂] at joint-tissue boundary (median, 5th–95th percentile): - Pickering emulsion (Liu 2025 PEBR geometry): 0.19 µM [0.034–1.1 µM] — GREEN - Fusion protein (Schiavon class, 1–5 nm separation): 0.034 µM [0.006–0.20 µM] — GREEN - Free co-formulated: 0.19 µM [0.005–7.2 µM, max 120 µM in worst-case URI:CAT 100:1 corner] — GREEN at reasonable stoichiometry; YELLOW at uneven URI:CAT

All three clear the 10 µM safe threshold by 5–50× margin under reference conditions. Toxicity threshold band (GREEN < 10 µM, YELLOW 10–100 µM, RED > 100 µM) was itself a comp-035 contribution — no published steady-state synovial-tissue toxicity curve existed; anchored on Schalkwijk 1986/87 (PMID 3707631) injected-GOx model + 26+ in vitro chondrocyte bolus studies + endogenous synovial baseline (~1 µM).

Substantive proximity-claim reframe (load-bearing for chassis selection): The FRET <10 nm proximity advertised in Liu 2025 is NOT the safety mechanism in the Pickering architecture. Da_shell ~5 × 10⁻³ means the 5 nm catalase shell is too thin to scavenge H₂O₂ in transit — escape fraction ~0.998. The actual safety mechanism is bulk-phase catalase scavenging from catalase distributed across all dispersed droplets in the joint volume — mathematically equivalent to free co-formulated at the same total dose. Catalase is so fast (kcat 10⁷–10⁸ s⁻¹) that bulk first-order destruction dominates regardless of proximity geometry. Pickering's actual load-bearing advantages are (a) fixed URI:CAT stoichiometry preservation in vivo, (b) catalase activity protection during storage / immune exposure, © mannose-targeted retention to tophi — not the FRET proximity claim.

Updated chassis-selection criteria (post-comp-035): choose architecture on production economics + regulatory pathway + manufacturing complexity + in vivo retention + immunogenicity — NOT on advertised proximity claims. Catalase (kcat/Km) is the dominant safety-margin driver across all three architectures (Spearman r = −0.95 to −0.97); catalase preparation quality + in vivo stability + proportional dosing are first-order chassis-selection variables.

Cheapest next wet-lab step (comp-035 handoff): Amplex Red microelectrode H₂O₂ measurement in synovial-fluid mimic with dispersed architecture + 0.5 mM urate substrate (~$2–5K per architecture). Tissue-level effects (cartilage damage, synoviocyte response) are downstream of [H₂O₂] exposure — sub-µM Amplex Red readout makes those low by construction. Chondrocyte-cytotoxicity titration only needed if Amplex Red surfaces unexpectedly high [H₂O₂].

Cross-reference. delivery-route-matrix.md, gout-kill-chain-delivery-routes.md, engineered-koji-protocol.md §"The Hydrogen Peroxide Question — and why the chassis solves it for free", intra-articular-uricase-h2o2-reaction-diffusion-computational.md (comp-035 full analysis).

7. Pharmacological chaperones for ABCG2 Q141K folding rescue

Intervention. Small molecules that bind misfolded Q141K ABCG2 and rescue trafficking from the ER aggresome to the apical brush border membrane. CFTR-corrector class precedent (ivacaftor / tezacaftor / elexacaftor for ΔF508 CFTR — multibillion-dollar therapeutic class). Same ATP-binding cassette superfamily as CFTR; same design problem. Q141K is the #1 gout-risk GWAS variant.

Chokepoint(s) hit. Gut and renal ABCG2 simultaneously (oral systemic small molecule). Pharmacologically distinct from butyrate-mediated HDAC rescue of Q141K (which is a different rescue mechanism).

Evidence level. In Vitro (academic mechanism literature, Basseville 2012 PMID 22472121 for HDAC-mediated rescue precedent); no Q141K-specific chaperone clinical programs. Mechanistic Extrapolation from CFTR-corrector class.

Why not koji. Small-molecule discovery + medicinal chemistry development is not a fermentation chassis output. The CFTR-corrector class came from pharma-style screening + structure-based design.

Candidate chassis. Small-molecule discovery campaign — AI-assisted binder design (RFdiffusion-style or related), structure-based virtual screening against ABCG2 Q141K. Compounding pharmacy track if a hit candidate has off-patent precedent in another indication (unlikely for novel chemistry, but worth checking).

Cheapest first move — COMPLETED 2026-05-16 as comp-032: GREEN. AlphaFold Q141K structure + virtual screen of 134 FDA-approved molecules against the Q141K NBD pocket. All four CFTR-corrector positive controls ranked in the top 11% — the FDA-approved drug surface is NOT empty for ABCG2 Q141K chaperone candidates. Top wet-lab-priority candidates (ranked by composite score):

  1. Lumacaftor (Tier 2, CFTR corrector) — strongest mechanistic prior; same ABC superfamily; on-patent for CF (Vertex), navigate patent landscape for off-label 503A
  2. Tafamidis (Tier 2, TTR tetramer stabilizer) — aromatic-acid stabilizer at hydrophobic interface; misfolded-state selective
  3. Ursodiol / UDCA (Tier 1, bile acid chaperone) — broad ER-stress chaperone via ATF6/Hsp70; F508del-CFTR rescue precedent; off-patent USP/NF monograph
  4. Diflunisal (Tier 1, lowest-friction first call — off-patent NSAID with USP/NF monograph + off-label ATTR-stabilization precedent) — anionic at pH 7.4, strongest electrostatic match for Q141K +1 pocket
  5. TUDCA (Tier 2, bile acid chaperone) — CNS-penetrant; F508del-CFTR + ALS-clinical-trial precedent

Next move: per-hit cell-based Q141K trafficking-rescue assay on the top 3-5 candidates in a Caco-2 Q141K-transfected line. Compounding-pharmacy partner conversation can lead with diflunisal as the Tier-1 off-patent low-friction first call. (source: abcg2-modulators.md, abcg2-q141k-chaperone-screen-computational.md)

Cross-reference. abcg2-modulators.md §"Pharmacological-chaperone route", abcg2-q141k-chaperone-screen-computational.md (comp-032 full analysis).

Multi-chassis stacks (compositions across existing entries)

Compositions where two interventions on different chassis hit complementary chokepoints and stack additively without competing for the same production / delivery resource. These are not chassis-pending entries themselves — both arms have selected chassis — but the composition is worth surfacing here because the same chassis-is-downstream-of-chokepoint discipline applies: don't filter a stack as "off-platform" just because one arm doesn't live in the koji track.

M1. Engineered PDB EcN × compounded disulfiram — urate-disposal upstream + CP6b pyroptotic-exit blockade

Composition. Engineered E. coli Nissle expressing the 2,8-dioxopurine cluster (CBT2.0 precedent, Li 2025 PMID 41070194, −63% plasma UA in mice) consumes luminal urate and produces butyrate → ABCG2 induction + NLRP3 dampening, before MSU crystals seed an inflammasome. Compounded oral disulfiram (250 mg/day; covalently modifies GSDMD Cys191) blocks the pyroptotic pore after NLRP3 fires, preventing IL-1β release and the inflammatory amplification cascade. The two arms hit the urate→inflammation axis at opposite ends — one drains the substrate, one closes the exit — and share no production stack, no formulation stack, no regulatory pathway. Disulfiram is sub-AUD dose (see comp-027 planned analysis) so co-administration with ethanol-producing live biotherapeutics is bounded by the strain's residual ethanol output, not by the pill.

Chokepoint(s) hit. CP6 urate-disposal (PDB arm, upstream of NLRP3) + CP6b GSDMD pyroptotic-exit (disulfiram arm, downstream of NLRP3). Pass 3 label correction (2026-05-15): the synthesizer's original framing "both branches of CP6" is tightened — urate-disposal is upstream of CP6 even though one wiki entry loosely labels PDB as CP6.

Chassis (both selected, neither pending). PDB arm → engineered EcN LBP (entry 1 above, chassis option A). Disulfiram arm → compounding pharmacy track (compounding-pharmacy-track.md). Multi-chassis stack, not a chassis-pending question.

Why this entry exists here. The composition was almost not surfaced because neither arm is novel individually — both are documented in their respective canonical pages. The stack is what's new, and absent an explicit "compositions" index, multi-chassis stacks risk being lost in the gap between single-modality pages. Per synthesis/strategic-reflections/2026-05-15-chassis-is-downstream-of-chokepoint.md, stacks composed across chassis deserve the same chokepoint-first treatment as single interventions.

Cheapest first move. Two parallel comp-NNNs, both completed 2026-05-16: - comp-027 — YELLOW-leaning-GREEN: narrow sub-AUD window centered on 100 mg/day (range 75–125 mg/d), where parent DSF Cmax engages GSDMD pore-formation blockade at therapeutic levels (1.3× cell-free IC50) while Me-DTC stays at or below DER hypotension threshold. Compounding-pharmacy handoff: IR capsule for 14-day titration → ER lipid-matrix 100 mg QD chronic. Gout co-administration clean (allopurinol synergistic). - comp-031 — YELLOW provisional: combined effect ~25–30% larger than PDB-alone (not 2–3× as naive sum predicts). Critical engineering handoff: route PDB and uricase to SEPARATE strains, NOT a dual-cassette EcN — substrate-competition for luminal urate means single-chassis dual-cassette gains ~nothing in additional urate consumption vs two separate strains co-administered. The butyrate-axis synergy (PDB → PPARγ/HDAC ABCG2 rescue) operates at the enterocyte nucleus from gut lumen — does NOT require co-localization at the bacterial scale. The M1 multi-chassis-stack framing here is correctly a two-strain combination probiotic, not a dual-cassette engineering target.

Cross-reference. purine-degrading-bacteria.md §"Companion intervention: compounded disulfiram"; disulfiram.md §"Companion intervention: PDB-engineered EcN"; compounding-pharmacy-track.md; computational-experiments.md comp-027 + comp-031.

Pending entries to triage into this list

These are interventions surfaced in the OE corpus that map to chokepoints but haven't yet been formalized as chassis-pending entries. Each warrants a short audit pass to confirm chassis-status:

  • Engineered exosomes carrying NLRP3 inhibitors targeted to CD163+ macrophages — see modality-chokepoint-matrix.md.
  • CRISPR / base editing in patient for Q141K → Q141 in crypt stem cells — see modality-chokepoint-matrix.md. Probably "delivery unsolved on a 5–10 year horizon" status.
  • Wearable / microneedle continuous UA monitoring — not an intervention per se, but a monitoring tool that changes intervention-titration kinetics. Different shape; possibly belongs in a separate monitoring-pending page.
  • GSDMD pore-mediated self-delivery of OE-relevant biologics (KPV / nanobodies / single SCR domains / IL-1RA — see gsdmd-pore-delivery-paradox.md §"Implication for OE biologics") — chassis-pending status: koji can produce the payloads; the chassis-question is the delivery format that gets the payload to the synovial fluid in time for the pore-opening window. Different chassis question than "what produces the molecule"; same general shape (real intervention, chassis open).
  • Engineered C1-INH (SERPING1, recombinant complement regulator) in LBP-luminal chassis — promoted from "named-but-unscoped" to "next CP0 LBP engineering gate" by comp-024 (2026-05-16). comp-024 ranked complestatin-family BGC heterologous expression as RED for LBP track (best host E. coli Nissle YELLOW 0.544; O₂-dependent NRPS tailoring chemistry incompatible with anaerobic-resident lifestyle) and surfaced C1-INH as GREEN-provisional 0.774 by comparison — single-axis problem (luminal-protease stability + glycosylation), testable via a comp-006-style protease-stability comp-NNN on SERPING1 in EcN-secreted format. Sister-to-DAF SCR1-4 (H05) — both are soluble human complement regulators expressed heterologously, both target CP0, but DAF SCR1-4 routes through koji secretion (engineered koji endgame strain) while C1-INH routes through LBP-luminal (engineered EcN). The two are mechanistically complementary: DAF accelerates convertase decay at the MSU crystal surface; C1-INH inactivates C1r/C1s + MASP-2 at the classical/lectin pathway entry point. Together they would cover convertase-formation (C1-INH) + convertase-disassembly (DAF) at two different cascade points and two different chassis. Next move: comp-NNN protease-stability + glycosylation feasibility for SERPING1 in EcN luminal-secreted format. Cross-reference: comp-024, complestatin-bgc-lbp-feasibility-computational.md, complement-c5a-gout.md, engineered-lbp-chassis.md.

When an entry leaves this page

An entry migrates off this page when its chassis is selected. Two outflows:

  1. Chassis selected → new dedicated scope page in wiki/ — example: PDB selects engineered EcN as chassis → new engineered-ecn-pdb-track.md scope page with Phase 2 follow-ups, comp-NNN, falsification card. The entry on this page becomes a one-line pointer to the new track page.
  2. Chassis selected → folds into existing peer-track page — example: a chassis-pending intervention turns out to be best-fit for the compounding pharmacy track → folds into compounding-pharmacy-track.md.

If an entry is falsified (chokepoint-fit turns out to be wrong, or mechanism doesn't survive scrutiny), it migrates to a falsification card or gets removed with a closure note. Falsification is OK; quiet filtering is not.

How decisions actually get made — there's no static rubric, by design

There is intentionally no static "promote / park / falsify" rubric on this page (or anywhere else). The decision mechanism is the wiki sweep daemon's Pass 2 (synthesizer), which re-evaluates every entry against the current corpus state on every sweep cycle. When new information lands that would shift an entry's status — a new comp-NNN output, a new wet-lab result, a new published clinical-trial readout, a new chokepoint analysis — the next sweep surfaces it as a Connection / Contradiction / Experiment / Open Question / Priority Action. The walkthrough operator then makes the actual promote / park / falsify call per-item, with each call grounded in the corpus state at that moment.

A static rubric here would be a snapshot of heuristics that drifts from the live evaluator. Items 1–5 of the 2026-05-15 sweep walkthrough are concrete evidence the mechanism works as designed: the daemon surfaced promotion-worthy recommendations (PDB×disulfiram CP6 stack, CFTR-corrector Q141K chaperone, inhaled mRNA-IL-1RA temporal complement) from chassis-pending entries; the walkthrough operator decided per-item; the actions shipped without anyone ever consulting a documented rubric. The dynamic process IS the rubric implementation.

This is why "park" and "falsify" don't appear as named statuses in the entries above — the page records interventions and their current chassis-question state, not decisions about resource allocation. Decisions live in walkthrough closure annotations (synthesis/done/) and in the canonical wiki pages that come out of promotions. The chassis-pending page is the index of the open chokepoint-fit-without-chassis state, not the decision log.

Maintenance

  • Updated on any walk-synthesis closure where a finding hits a chokepoint but lacks a chassis (see closure question in .claude/skills/walk-synthesis/SKILL.md).
  • Updated when the sweep daemon's Pass ⅔ review surfaces a chassis-expanding finding (see anti-chassis-filter language in scripts/sweep-prompt-2-synthesize.md and scripts/sweep-prompt-3-review.md).
  • The strategic reflection in synthesis/strategic-reflections/2026-05-15-chassis-is-downstream-of-chokepoint.md is the discipline anchor. The page itself is the operational manifestation of that discipline.
  • The page is not a deprioritization queue. Entries here are real interventions waiting on a real next step — not "won't-fix" or "off-platform." Treat as active research, just on a different chassis track from the koji-engineering track.