Skip to content

Testosterone × Intestinal ABCG2 Suppression — Evidence Mining (comp-016)

Question

Does primary literature support the load-bearing claim in androgen-urate-axis.md §"Why this matters for the platform" that androgens directly suppress intestinal ABCG2 expression at magnitudes consistent with a structural ceiling on the gut-lumen-sink platform efficacy in male/androgen-dominant patients?

Verdict

WEAK / UNCONFIRMED (provisional — abstract-level scan only; full-text grep-verification pending).

Of 17 curated primary sources from a focused PubMed scan (2026-05-07), zero primary studies demonstrate androgen-driven intestinal ABCG2 suppression directly. No castration-restoration study exists; no testosterone-administration study shows decreased intestinal ABCG2 mRNA or protein in vivo; no published cohort reports increased intestinal ABCG2 protein after androgen ablation. The intestinal compartment IS sex-dimorphic in a urate-relevant way (Hoque 2020 Nature Communications — male Q140K mice show 88% intestinal ABCG2 protein loss + severe hyperuricemia; female Q140K mice are protected), but the mechanistic driver appears to be estradiol POSITIVE on the female side (Yu 2021 — estradiol upregulates intestinal ABCG2 via PI3K/Akt) rather than androgen NEGATIVE on the male side. One primary study directly contradicts the platform thesis at the in vitro intestinal level (Klyushova 2023 — testosterone INDUCES ABCG2 in Caco-2). One study found NO baseline sex difference in healthy rat intestinal ABCG2 across all four segments (MacLean 2008).

This is the verdict-tier the manual-literature-mining.md §"Pre-commit verification gate" rule was written for: a load-bearing claim that the primary literature does not directly support, propagated across multiple wiki pages on the strength of summary-tier reasoning.

Why this matters

The Open Enzyme platform thesis depends on the structural-ceiling argument in two places:

  1. androgen-urate-axis.md §"Why this matters for the platform" says: "androgen-driven ABCG2 suppression hits the gut-lumen-sink pathway directly... structural ceiling on platform efficacy in the primary demographic." This is the load-bearing framing the comp-016 scan is testing.
  2. abcg2-modulators.md §1 (Androgens entry) says: "AR-mediated transcriptional repression of ABCG2 in gut and kidney." This presupposes a direct AR mechanism on intestinal ABCG2 — which the primary literature does not support.

If the structural-ceiling argument is wrong (or weakly supported), the implications cascade: - The "men have a lower asymptote on platform efficacy" framing should be softened from a structural claim to an empirical-magnitude-uncertain claim. - The "rescue lever" framing changes — instead of "anti-androgen pharmacology to relieve AR repression of intestinal ABCG2" (no primary evidence supports this lever exists at intestinal AR level), the better-supported lever is "estrogenic PI3K/Akt agonism on intestinal enterocytes" or "PPARγ-mediated induction via fermentable-fiber butyrate" (which is the mechanism abcg2-modulators.md already favors as Tier 1). - The Q141K-positive male subset still has the strongest case for a genotype-stratified intervention — but the case rests on the genetic-loss-of-function vulnerability (Hoque 2020), not on additional androgen-driven suppression on top of the variant.

Method summary

Approach: WebSearch + WebFetch via PubMed, PMC, journal landing pages, aggregator sites. Six SKILL-prompt-defined sub-questions queried in parallel; results aggregated into per-study structured records (inputs/studies.json). Direction-of-effect for each study scored against the load-bearing claim. Per-study load-bearing-for-the-intestinal-claim flag set on the conjunction of: (a) intestinal tissue measured, (b) hormone or genotype manipulation as the variable, © ABCG2 readout. Aggregate verdict by tally of supporting / contradicting / neutral studies in the load-bearing subset.

Tool limitations encountered: WebFetch was 403-blocked from PMC, ScienceDirect, Springer, MDPI, Frontiers, PubMed, doi.org redirects. All primary-literature claims were extracted at abstract-level + WebSearch-result-summary-level, NOT full-text grep-verified per manual-literature-mining.md §"Pre-commit verification gate." This is acknowledged in the limitations section.

Multilingual scan: NOT executed in this run (CNKI / J-STAGE / KISS queries deferred to a Phase 2 follow-up — see Limitations).

Key results

The load-bearing-for-the-intestinal-claim subset (3 of 17 studies)

Study Tissue Intervention Direction Key magnitude
S01 Hoque 2020 Nature Communications PMID 32488095 mouse intestine + kidney Q140K knock-in (genetic LoF) Supports sex-dimorphism (males vulnerable) 88% intestinal ABCG2 protein loss in male Q140K homozygotes; 53% heterozygotes; 44% renal — intestinal more vulnerable than renal. Female Q140K protected.
S03 Yu 2021 Nutr Metab PMID 34144706 mouse intestine + Caco-2 estradiol benzoate IP Supports female-positive arm; not male-negative Estradiol upregulates intestinal ABCG2 via PI3K/Akt; LY294002 partially blocks. Direction confirmed; specific fold-change not extracted from abstract.
S04 Klyushova 2023 Biochem Moscow DOI 10.1134/S1990747823050100 Caco-2 T at 1, 10, 100 μM, 24h CONTRADICTS Testosterone INCREASES ABCG2 in Caco-2 via PXR/FXR (xenobiotic-sensor) pathway, not AR. Direction opposite to platform thesis at in vitro intestinal level.

No fourth load-bearing study exists — no primary paper has done the experiment that would directly answer the question (orchiectomy + measure intestinal ABCG2; or T administration → intestinal ABCG2 mRNA/protein in vivo).

The supportive-but-not-load-bearing studies (cohort-level androgen → urate)

Study Direction Magnitude Caveat
S02 Sakamoto 2018 PMID 30557349 Supports T → urate up −0.66 mg/dL at 6 months ADT (n=150 ADT vs 339 surgery) Mechanism not isolated to intestine; consistent with URAT1-only
S08 Yahyaoui 2008 PMID 18349066 Supports T → urate up FtM 2-year T administration significantly raised serum UA + decreased FEUA Renal-FEUA mechanism implied; intestinal ABCG2 not measured
S15 KNIGHT/ENIGI 2024-2025 Supports T → urate up, dose-dependent FtM cohort, magnitude not extracted Renal-FEUA mechanism cited; intestinal not measured
S16 Adolescent boys PMC8405811 Supports T → urate up Pubertal UA rise correlates with T + low SHBG Mechanism not measured at tissue level
S14 Hak Choi 2008 NHANES PMID 18822120 Supports estrogen → urate down HRT users −0.24 mg/dL adjusted Cohort observational

These cohort studies confirm androgens are a real and clinically meaningful urate-elevating signal in human physiology. They do NOT distinguish renal (URAT1) vs intestinal (ABCG2) mechanism. The −0.66 mg/dL Sakamoto magnitude is consistent with URAT1 being the dominant transporter affected — URAT1 changes alone can produce this size of effect (per the Hosoyamada 2010 mouse work).

The mechanism studies

Study Tissue Mechanism finding
S07 Jeong 2015 PMID 25615818 LNCaP prostate cancer Androgen withdrawal → ↑CREB-P → ↑CRTC2 nuclear translocation → ↑BCRP via −329 CRE on promoter. Indirect mechanism in cancer cells; no direct AR-ARE on the BCRP promoter.
S05 Tanaka 2005 PMID 15567169 rat kidney + mouse liver Sex-dimorphic ABCG2 driven by estradiol-suppressive on females, not testosterone-inductive on males. The female-low pattern is the load-bearing direction.

Receptor mechanism finding: there is no published direct AR → ABCG2 promoter binding. The closest mechanism is the indirect CREB/CRTC2 axis in prostate cancer cells (Jeong 2015). No data extending this to intestinal epithelium.

The contradictory studies

Study Tissue Finding
S04 Klyushova 2023 Caco-2 T induces ABCG2 (direction opposite to thesis)
S10 MacLean 2008 PMID 18378562 rat intestine (full scan) NO sex difference in baseline ABCG2 across duodenum, jejunum, ileum, colon — explicit null finding

Mechanism reframing — what the evidence actually supports

The wiki's current implicit model:

Androgens → AR → ↓ABCG2 transcription → ↓intestinal ABCG2 protein → ↓luminal urate → structural ceiling on gut-sink platform

The literature-supported model:

Females have estradiol-driven intestinal ABCG2 upregulation (Yu 2021, PI3K/Akt) → higher gut-sink baseline
Males lack this estradiol drive → lower gut-sink baseline (modest, magnitude uncertain)
Q141K-positive males (regardless of androgen state) are vulnerable to severe loss-of-function (Hoque 2020) — independent of androgen
ADT in human cohorts lowers serum UA at clinically meaningful magnitude (Sakamoto 2018, −0.66 mg/dL) but mechanism likely renal URAT1-dominant

The two models have substantially different platform-design implications:

Implication "Androgens suppress ABCG2" model "Females have estradiol-driven ABCG2 upregulation; males lack it" model
Intervention lever Anti-androgen pharmacology on intestinal AR Estrogenic PI3K/Akt induction OR PPARγ butyrate route
Magnitude expected from rescue Substantial (relieve a strong tonic suppression) Modest (induce signal that's normally absent)
Q141K stratification Genotype × androgen status Genotype × estrogen status; androgen status secondary
Female-specific therapy Less needed (no androgen-suppression) Females already at higher baseline; therapy more leveraged in males
Confidence interval on the asymptote difference Predicted to be substantial Empirically near-null in healthy baseline (MacLean 2008); only emerges under disease/genetic stress

Correction note (post-comp-017 full-text re-read, 2026-05-07)

A subsequent Tier-0 follow-up experiment (comp-017) did the full-text re-read of the four anchor papers (Hoque 2020, Yu 2021, Klyushova 2023, MacLean 2008) and surfaced numerical and methodological refinements to comp-016's abstract-level numbers. The qualitative direction of comp-016's verdict (T → intestinal ABCG2 suppression WEAK / UNCONFIRMED) is unchanged and arguably strengthened. The corrections:

  • Hoque 2020 PMID 32488095 — "88% intestinal protein loss" is the COMBINED Western + apical IHC measurement for homozygotes. The Western-jejunum-only number is 78% (~1.8× the 44% renal loss, not 2.0× as comp-016's framing implied). The 53% number is the heterozygote combined. UA flux loss in the ligation loop is 40%. Female FEUA p=0.6263 (strong null). Paper does NOT invoke AR. Both numbers (78% and 88%) are real measurements; comp-016's "88%" framing is correct as the combined-assay result but the Western-only number is more conservative for the platform-thesis ceiling argument.
  • Yu 2021 PMID 34144706 — Caco-2 active concentration is 100 µM EB, 5–6 orders above physiological serum E2 (~30–500 pmol/L). Paper notes "without a dose-dependent effect." Female mouse arm is OVARIECTOMIZED + EB replacement (high-contrast pharmacological model, NOT healthy comparison). The mechanism EXISTS at strong-pharmacological tier; magnitude at physiological E2 is unestablished. The "estradiol upregulates intestinal ABCG2 via PI3K/Akt" reframe in this page is correct as a mechanism claim but should not be cited as supporting a measurable population-level sex-dimorphism in healthy adults.
  • Klyushova 2023 — ALL three sex hormones (T, E2, P) at all three concentrations (1, 10, 100 µM) INCREASE ABCG2 via PXR/FXR. This is a xenobiotic-sensor response, not hormone-receptor-specific. Lowest active T concentration is ~30–100× above physiological free T. The "T INDUCES not suppresses ABCG2" framing in this page is correct, but the mechanism is xenobiotic-tier, not androgen-axis-specific.
  • MacLean 2008 null finding STRENGTHENED by Tubic-Grozdanis 2020 replication (P-gp dimorphic, BCRP not).
  • NEW finding (Hosoyamada 2010 PMID 20589576 full-text): T → renal Smct1 mRNA AND protein ↑, GLUT9 attenuated, BUT URAT1 mRNA-only (protein UNCHANGED in non-orchiectomized animals). The renal-arm mechanism propagated into androgen-urate-axis.md was overstated; the protein-level driver of male hyperuricemia in mouse renal tissue is Smct1 + GLUT9 attenuation, not URAT1 protein elevation. comp-017's full-text re-read surfaced this; the parent page is updated correspondingly.

Net implication for the platform thesis: comp-016's WEAK / UNCONFIRMED verdict survives unchanged in qualitative direction. The "structural ceiling on platform efficacy in androgen-dominant patients" framing softens further than comp-016 already softened it — at healthy baseline, the sex-dimorphism magnitude is empirically near-null; the asymptote difference manifests primarily under disease-state genetic stress (Q141K) or strong-pharmacological perturbation. The Q141K rescue thesis is unaffected and remains the platform's strongest pharmacogenomic differentiator.

Limitations

  1. Abstract-level + search-summary-level only. All 17 studies are anchored to abstracts and WebSearch-result-summary text, NOT full-text grep-verified per manual-literature-mining.md Rule 4. The Hoque 2020 "88% intestinal protein loss" magnitude, the Sakamoto 2018 "−0.66 mg/dL" magnitude, and the Yu 2021 PI3K/Akt mechanism are all from search-summary abstracts — should be grep-verified against the published papers in a follow-up Paperclip MCP run before they are quoted as load-bearing in any wiki page that depends on those magnitudes. Confidence tier: "verified-against-summary" not "verified-against-primary." [comp-017 partially closed this for the four anchor papers — see Correction note above.]

  2. WebFetch 403-blocked across PMC, journals, PubMed. A future run with Paperclip MCP (per the methodology in manual-literature-mining.md) could pull full-text content.lines for grep verification.

  3. Multilingual scan NOT executed. Per CLAUDE.md §"Global-multilingual research by default," CNKI / J-STAGE / KISS / WanFang queries should have run for testosterone × intestinal ABCG2. Deferred to Phase 2. Likelihood-of-finding-something-new is uncertain — the field is dominated by Western pharma-DMPK and rheumatology research for ABCG2 specifically; Chinese/Japanese gout literature tends to focus more on URAT1 and XO than ABCG2.

  4. The 17-study scan is non-exhaustive. Other relevant primary literature may exist that this scan missed. The verdict is robust to this in the qualitative direction (it would take a single primary study showing direct T → intestinal ABCG2 suppression in vivo to flip the verdict to PARTIAL; that study has not surfaced in this scan).

  5. The verdict logic is conservative. A different reviewer could reasonably argue the indirect Hoque 2020 + Sakamoto 2018 + Yu 2021 + FtM-cohort triangulation supports a PARTIAL verdict ("the sex-dimorphism is real even if the mechanism is estradiol-positive rather than androgen-negative"). This page records the WEAK_UNCONFIRMED verdict as the strict literal answer to the SPECIFIC claim "androgens directly suppress intestinal ABCG2"; the broader claim "intestinal ABCG2 is sex-dimorphic in a urate-relevant way" is well-supported and downstream pages should retain it.

  6. The directionality difference matters substantively. "Androgens suppress" vs "estradiol upregulates the female arm and males lack the signal" sound similar but predict different rescue interventions and different platform-design choices. The conservative verdict on the SPECIFIC claim is meaningful for the platform design.

Impact on experimental priorities

Wet-lab / experimental question Pre-comp-016 priority Post-comp-016 priority
validation-experiments.md §1.14 (TNFα + butyrate ABCG2 rescue in Caco-2) Confirmation experiment Stays as confirmation — TNFα suppression and butyrate induction are well-supported; orthogonal to androgen question
validation-experiments.md §1.22 (Q141K rescue HDAC inhibitor screen — comp-007) Confirmation experiment Stays as confirmation — Q141K rescue is butyrate-mediated, orthogonal to androgen direct suppression
Hypothetical animal-model experiment: castration → intestinal ABCG2 measurement Not yet specified Now PROMOTED — the cheapest experimental killshot for this question. Direct measurement that would either confirm or falsify the androgen-direct-suppression mechanism. Estimated cost: ~$30K–60K, 8 weeks; standard mouse castration + T-replacement + jejunal/ileal Western blot + qPCR for ABCG2. Could potentially extend to FtM cohort biopsy data if a collaborating clinical group exists.
Hypothetical pharmacogenomic Q141K × androgen-status stratification of fiber RCT Not yet specified Worth noting as Phase 2 — would resolve whether the platform thesis is androgen-modulated or estrogen-driven at the human level. Higher cost and longer timeline.

Cross-references

Pre-commit verification gate disclosure

Per manual-literature-mining.md §"Pre-commit verification gate":

Sources WebSearch-summary-verified, NOT full-text grep-verified (the limitation that bounds confidence on this page's quantitative claims): - All 17 primary studies were extracted at abstract / search-summary level. The page cites magnitudes (88%, 53%, 44%, −0.66 mg/dL, etc.) that come from search-result text describing the published abstracts. - This is explicitly acknowledged as a verification gap. The page should not be cited as a primary source for any of these magnitudes; consumers should grep-verify against the published papers via Paperclip or full-text PDF before using the magnitudes in load-bearing downstream analyses.

The qualitative verdict (WEAK_UNCONFIRMED) is robust to this caveat because it turns on the direction of evidence ("no primary paper directly demonstrates T → intestinal ABCG2 suppression in vivo"), not on the magnitudes. A future Paperclip-MCP run would refine the magnitudes and confirm/refute exact numbers, but is unlikely to surface a primary study that would flip the qualitative verdict.

What WOULD flip the verdict: - A primary paper showing castration → measurable increase in intestinal ABCG2 mRNA or protein (would shift verdict toward PARTIAL or CONFIRMED) - A primary paper showing T administration → measurable decrease in intestinal ABCG2 in vivo (would shift verdict toward PARTIAL or CONFIRMED) - A pharmacogenomic study with sex-stratified Q141K data showing androgen status modulates the variant's effect on serum UA at meaningful magnitude (would partially support the platform thesis) - A direct AR-ARE binding ChIP-seq result on the ABCG2 promoter (would shift the receptor-mechanism question toward CONFIRMED for the molecular arm)

If any of these surface in a future scan, the verdict should be revisited and this page updated.