Skip to content

Food-Grade HDAC Inhibitor Screen for Q141K-ABCG2 Trafficking Rescue: Computational Analysis (comp-007)

Status: Complete — 2026-05-05
Experiment folder: experiments/comp-007-food-grade-hdaci-screen/
Informs: validation-experiments.md §1.22

Summary

Stage 1 in silico screen of 7 food-grade or GRAS-classified HDAC inhibitor candidates for Q141K-ABCG2 trafficking rescue. Ranked on three axes: class I HDAC potency (HDAC½/3 IC50), HDAC6 selectivity (off-target cardiotoxicity avoidance), and gut-enriched exposure (1 − oral bioavailability).

Top 3 advancing to Stage 2:

Rank Compound Mean HDAC1-3 IC50 HDAC6 sel. Gut BA Composite Confidence
1 Butyrate 12,000 nM 167× confirmed 5% 0.3734 HIGH
2 Sulforaphane 5,000 nM* uncharacterized 70% 0.0900 LOW
3 PEITC 10,000 nM* uncharacterized 60% 0.0600 LOW

* = estimated from nuclear extract or by analogy; isoform-specific biochemical IC50 unavailable

All rankings are Mechanistic Extrapolation per Open Enzyme evidence-level convention. No wet-lab data generated at Stage 1.

Background: Q141K ABCG2 and HDAC Inhibition

ABCG2-Q141K is the most common clinically relevant variant of the urate transporter ABCG2, carried by ~12% of all alleles (~25% East Asian, ~4% European). The Q141K substitution destabilizes the NBD2 ATP-binding domain, causing ER retention via ERAD rather than apical membrane trafficking. The consequence: reduced luminal urate efflux → higher serum urate → increased gout risk and impaired Q141K-ABCG2 function in absorptive epithelial cells (Basseville et al. 2012, PMID 22472121; mechanistic extrapolation — no direct Q141K clinical HDACi data beyond Basseville's in vitro demonstration).

The HDAC-mediated rescue pathway (Basseville 2012, in vitro, human cell lines expressing Q141K-ABCG2):

Class I HDACi (HDAC1/2/3)
  → histone hyperacetylation
  → HSF1 nuclear translocation
  → Hsp90α/β transcriptional upregulation
  → chaperone-assisted refolding of misfolded Q141K NBD2
  → partial membrane-trafficking rescue (~30–50% surface expression restoration at 1 mM butyrate)

This is distinct from butyrate's ABCG2-inducing effect in wild-type cells, which is PPARγ-mediated and does not require HDAC inhibition (Xie et al. 2020 — see abcg2-modulators.md §6).

Why HDAC6 is the off-target: HDAC6 is a cytoplasmic tubulin deacetylase (class IIb). Pan-HDAC inhibitors (vorinostat, romidepsin) hit HDAC6 → tubulin hyperacetylation → actin/microtubule dysregulation → cardiac ion-channel effects → QT prolongation, cardiomyopathy. Any food-grade HDACi candidate must spare HDAC6 to be safe for chronic oral use. Selectivity ratio (HDAC6/class-I IC50) > 10× is the minimum acceptable; butyrate achieves 167× in biochemical assay.

Scoring Formula

composite_score = potency_score × selectivity_score × gut_selectivity_score
Factor Definition Notes
potency_score 1 / geomean(HDAC½/3 IC50 nM), max-normalized to [0,1] Higher = more potent at class I inhibition
selectivity_score HDAC6_IC50 / (HDAC6_IC50 + mean_classI_IC50), midpoint ratio = 10 Sigmoid-like; confirmed 222× → 0.957; unknown HDAC6 → penalty 0.30
gut_selectivity_score 1 − oral_bioavailability_fraction Higher = more gut-enriched; rewards low-bioavailability compounds

Confidence tiers: HIGH = biochemical recombinant assay (ChEMBL); LOW = nuclear extract or estimated from cellular data; DATA_UNAVAILABLE = no usable IC50. DATA_UNAVAILABLE compounds receive composite = 0.

The selectivity penalty (0.30) for unknown-HDAC6 compounds means they can rank above 0 but are substantially penalized compared to compounds with confirmed HDAC6 avoidance. If isoform-specific HDAC6 IC50 data were measured and found to be > 1 mM for sulforaphane or PEITC, their composite scores would increase ~3-fold.

Candidate-by-Candidate Analysis

1. Butyrate (rank 1, composite 0.3734, HIGH confidence)

Butyrate (sodium butyrate; n-butyric acid) is the benchmark food-grade HDACi and the only candidate with HIGH-confidence biochemical IC50 data for all four HDAC isoforms.

IC50 (biochemical, recombinant; ACS Med Chem Lett 2011, ChEMBL activity IDs 5207850–5207858): - HDAC1: 16,000 nM - HDAC2: 12,000 nM - HDAC3: 9,000 nM - HDAC6: > 2,000,000 nM (> 2 mM; no inhibition detected at 2 mM) - HDAC6 selectivity ratio: ~167× over HDAC½/3 geometric mean (12,172 nM)

Mechanism: Short-chain fatty acid; carboxylate coordinates the catalytic zinc of class I HDAC active sites. The HDAC6 active-site cavity is bulkier and accommodates acetylated tubulin peptides — the small carboxylate headgroup of butyrate is insufficient to engage it efficiently, which explains the class I selectivity structurally.

Gut vs. systemic: Colonic luminal concentration 0.5–2 mM from fermentable fiber (PMID 19197985); oral bioavailability fraction ~0.05 (colonocytes consume >90% as energy substrate before portal entry). Gut selectivity score: 0.95 — extremely favorable.

Limitation for Q141K rescue: Butyrate delivery at 1 mM is microbiome- and fiber-dependent. Individuals on low-fiber diets or with microbiome dysbiosis may have substantially lower colonic butyrate than 0.5 mM. A direct supplement or alternative food-grade HDACi with more predictable dosing would be meaningful for Q141K carriers.

2. Sulforaphane (rank 2, composite 0.0900, LOW confidence)

Sulforaphane (SFN; 4-methylsulfinylbutyl isothiocyanate) is derived from glucoraphanin hydrolysis by myrosinase in broccoli sprouts.

IC50 (estimated; no isoform-specific biochemical data available from ChEMBL or primary literature): - Effective class I HDAC estimate: 5,000 nM (from PBMC HDAC activity measurements; Myzak 2006 PMID 16537909, Su 2014 PMID 24441674) - HDAC6 IC50: uncharacterized

Mechanism: SFN is metabolized via the mercapturic acid pathway; the active metabolites SFN-Cys and SFN-N-acetylcysteine are proposed to engage the HDAC active-site zinc via thiol/sulfinyl coordination. HDAC inhibition in PBMC and cell-line studies is well-documented (Myzak 2006, Su 2014), but these are protein-level and nuclear-extract measurements, not isoform-specific biochemical assays. ChEMBL CHEMBL48802 has no HDAC bioactivity records. Critically, Choi 2018 (PMID 29482060) documents SFN-mediated downregulation of HDAC½/3 protein levels — this is epigenetic reprogramming (transcriptional), not direct zinc-chelation inhibition. The distinction matters: indirect, transcription-mediated HDAC reduction may have different kinetics, dose-response, and isoform specificity than direct inhibition.

Gut vs. systemic: ~70–80% oral bioavailability (absorbed in proximal small intestine); peak plasma 2–3 μM at 100 g broccoli sprout dose. High oral bioavailability substantially reduces its gut-selectivity score (0.30). The enterocyte exposure during absorption may be higher than plasma levels suggest, but this cannot be quantified without intracellular SFN measurements.

Key Stage 2 question: Does SFN inhibit HDAC½/3 preferentially over HDAC6 in a Caco-2 nuclear extract? If HDAC6 sparing is confirmed, SFN moves up substantially in the ranking.

3. PEITC (rank 3, composite 0.0600, LOW confidence)

Phenethyl isothiocyanate (PEITC) is derived from gluconasturtiin in watercress and garden cress.

IC50 (estimated; no ChEMBL record, no primary isoform-specific IC50): - Effective class I HDAC estimate: 10,000 nM (Gupta 2019 PMID 31228495; HDAC1 protein reduction in colorectal model — not a direct enzyme IC50) - HDAC6 IC50: uncharacterized

Mechanism: Similar to SFN — isothiocyanate metabolism via mercapturic acid pathway → active thiol metabolites → proposed zinc coordination. HDAC1 protein downregulation documented in cell lines (Gupta 2019); mechanism is extrapolated from SFN analogy. Less studied than SFN.

Gut vs. systemic: ~60% oral bioavailability; gut-selectivity score 0.40. Gut concentrations during absorption are 3–10 μM (estimated from watercress consumption PK).

4. Allyl Mercaptan (rank 4, composite 0.0225, LOW confidence)

Allyl mercaptan (AM; 2-propene-1-thiol) is produced from garlic-derived DADS and DATS by intestinal/hepatic metabolism.

IC50 (estimated from bulk nuclear extract; no isoform-specific data): - 200 μM AM → 92% total HDAC activity inhibition in Caco-2 nuclear extracts (Druesne 2004 PMID 14976134) - Effective class I HDAC estimate: 50,000 nM (Ki-equivalent from 92% inhibition at 200 μM, assuming competitive inhibition; rough estimate only) - HDAC6 IC50: uncharacterized

Mechanism: Free thiol directly coordinates catalytic zinc; competitive inhibition confirmed in nuclear extract. Most potent garlic-derived HDACi in Druesne 2004. Isoform selectivity unknown.

Gut vs. systemic: Volatile compound; primarily a gut-lumen and intestinal-cell metabolite from DADS. Oral BA ~25%; gut exposure is high relative to systemic.

5. DADS (rank 5, composite 0.0009, LOW confidence)

Diallyl disulfide (DADS) is a major garlic oil component; AM is its proximal intracellular metabolite.

IC50 (estimated from nuclear extract): - 200 μM DADS → 29% HDAC inhibition in Caco-2 nuclear extracts (Druesne 2004) — much weaker than AM - Effective class I HDAC estimate: ~1,000,000 nM (weak direct inhibitor; AM is the active species)

DADS's low direct potency reflects that it requires intracellular conversion to AM. Its score (0.0009) reflects this: DADS is better conceptualized as an AM pro-drug than a standalone HDACi.

6–7. Caffeic Acid and Ferulic Acid (composite 0.0000, DATA_UNAVAILABLE)

Both hydroxycinnamic acids show histone hyperacetylation in cell studies (PMID 27588384) consistent with HDAC inhibition, but no isoform-specific IC50 is available from ChEMBL (CHEMBL145, CHEMBL32749 — 192 and 249 activities respectively, none against HDAC targets) or primary literature. These compounds are not scored and do not advance to Stage 2 without a targeted HDAC½/3 biochemical screen first.

The proposed mechanism (catechol metal chelation for caffeic acid; similar but methoxy-reduced for ferulic acid) is plausible but unquantified. If biochemical IC50 data emerge, they could re-enter the ranking.

Stage 2 Recommendation

Advance to Stage 2 (paired Caco-2 / HepG2 HDAC activity assay):

  1. Butyrate — confirmed class I selectivity; Stage 2 is primarily a confirmation run and a reference standard for the cellular assay format
  2. Sulforaphane — the critical Stage 2 question is HDAC6 isoform selectivity; Stage 2 must include a HDAC6-selective substrate assay (acetylated α-tubulin peptide) alongside the class I substrate
  3. PEITC — same Stage 2 requirement as SFN for HDAC6 characterization; lower estimated class I potency than SFN but plausible mechanism

Do not advance to Stage 2 without biochemical data first: - Caffeic Acid, Ferulic Acid — isoform IC50 screen against recombinant HDAC½/3/6 needed before cellular work is interpretable

Effectively ruled out at Stage 1: - DADS — prodrug for AM; study AM directly in Stage 2 rather than DADS - Allyl Mercaptan (rank 4) — could be included in Stage 2 if reagent is available; lower priority than top 3

Key Limitations

Heterogeneous assay formats. Butyrate IC50 values are from biochemical recombinant assay (highest confidence). SFN values are from PBMC nuclear extract or protein-level measurements (confounded by non-HDAC effects). Direct comparison across formats is approximate.

HDAC6 selectivity unknown for most candidates. The selectivity penalty (0.30) is arbitrary; if HDAC6 IC50 were measured at > 1 mM for SFN or PEITC, their composite scores would increase ~3-fold and their rankings relative to butyrate would tighten substantially.

Gut-enrichment proxy is coarse. Oral bioavailability (1 − BA) is a surrogate for gut exposure. Sulforaphane is highly bioavailable (70%) but enterocytes encounter SFN during absorption — intracellular enterocyte SFN concentration during transit may be 10–100× higher than plasma. The proxy underestimates SFN's intestinal epithelial activity.

Sulforaphane's mechanism differs from butyrate's. Direct zinc-chelation (butyrate) vs. indirect protein-level downregulation (SFN) via transcriptional reprogramming. Kinetics, dose-response shape, and isoform specificity may differ substantially. Stage 2 cellular data will be the first direct comparison in a Q141K-relevant cell context.

No Q141K trafficking rescue data for any compound except the vorinostat class. Basseville 2012 used the pharmacological tool compound and confirmed the pathway. That butyrate's class I profile recapitulates the pathway is a mechanistic extrapolation; Stage 3 is the first direct test.

Cross-References