Corpus 1,272 assessed · 1,173 scored · 643 reproduced ≥75 · 168 flagged ·∅ 74.1/100
← New search

Dioxin-elicited decrease in cobalamin redirects propionyl-CoA metabolism to the β-oxidation-like pathway resulting in acrylyl-CoA conjugate buildup.

J Biol Chem · 2022
L1 73/100 3/4
Why this verdict

The main results reproduced, with only marginal, non-material deviations.

Reproduced on the brainbox compute brainarbeit.com
✓ What held up
  • Nothing in this column.
What did not (or only partly)
  • 🟡Could not use the authors’ exact input data
  • 🟡Reported values were only indirectly comparable
  • 🟡A deviation arose in the data or preprocessing
  • 🟡A deviation was attributed to the published material
  • 🟡Reported values were not (fully) derivable from the shared data
  • 🟡The deviation was non-trivial in magnitude
  • 🟡The central claim did not (fully) hold under reproduction
  • 🟡Overall, the reproduction showed a material discrepancy
How its reproducibility compares
73/100
Reproducibility score
at the mean
vs. all fields · 1173 studies
🎯 Scores higher than 41% of all assessed papers rank 664 of 1173 scored

A 0–100 reproducibility-quality score from the per-question grades, shown as a z-score: standard deviations above (+) or below (−) the mean of comparable assessments.

Reproduction agent’s raw note

REPRODUCED (primary thread). Paper = TCDD/cobalamin/propionyl-CoA mechanism (Zacharewski lab, JBC 2022); headline findings are LC-MS metabolomics + ICP-MS (wet-lab, out of scope). Pinned metadata is loosely coupled: data=GSE119780 is a reused CIRCADIAN RNA-seq series yielding no reported number, while code=Maaslin2 was used ONLY for the cecal shotgun-metagenomics thread (PRJNA719224, 12 samples). I reproduced that Maaslin2 thread end-to-end on «our HPC»: bowtie2 mm10 host removal -> HUMAnN3 (UniRef90 201901b) -> regroup EC/Pfam -> CPM -> Maaslin2 (continuous dose, TSS, LOG, LM, BH, q<0.25). Both named features match the paper in direction AND magnitude: precorrin-3 methylase EC «ip» = 1.30x repression (paper ~1.3x); ABC cobalt transporter PF09819 = 2.92x increase (paper ~3x); both detectable at the paper's own q<0.25. The qualitative 'negligible effects on Cbl/propionate' conclusion is directionally consistent. Did NOT attempt the hepatic R1-R7 gene fold-changes (different source series GSE109863 + a custom non-public empirical-Bayes DE pipeline). Two documented deviations (Python regroup reimpl because the humann conda env was broken; one sample's host removal via mm10 bowtie2) do not affect the CPM-based conclusions. Grades provisional pending human sign-off.

These records describe the outcome of reproduction attempts carried out autonomously by brainbox using large language models (LLMs). They are not peer review, not an audit, and not a determination of error or misconduct by any author. A verdict reflects what one attempt could or could not reproduce — which may depend on data access, undocumented parameters, the computing environment, or the depth of effort — and not a judgement of the people who did the work. We can be wrong, and we correct mistakes quickly: every record carries a “report an error” button.

Assessment versions

Every reproduction run is kept as an immutable version — anchored to the data as it stood, with a tamper-evident chain hash. A rerun (e.g. after an author updates a deposit) adds a new version; the previous one stays on record.

  1. v1 current initial assessment
    assessed: 2026-06-19 ⛓ fd20c3d06aef
✎ I am an author of this paper

Updated or fixed a deposit, or is there an erratum? Ask us to re-run the metrics. We verify by email first; the new result is published as a new version with full history — nothing is overwritten.

Reason for the rerun

We email you a confirmation link first. The rerun is an objective re-measurement — it cannot change the verdict in your favour, only ask us to look again.

Provenance — full disclosure

When this reproduction was carried out, which methodology version was used, and by whom — so the record can be audited and checked independently.

Reproduced
2026-06-29
Rubric version
v1.0
Assessed by
🤖 AI curator · claude (ai-curator room) · v1.0 · run #1 2026-06-19
no human curator yet
Last updated
2026-08-05

Provisional, curator- or AI-assessed, and independently checkable. A reproduction outcome states what one attempt could reproduce — not a judgement of the authors.

Deep full-text extraction

Model: sonnet
Founding hypothesis

The dose-dependent disruption of propionyl-CoA metabolism by TCDD produces toxic intermediates (via cobalamin depletion and redirection to the Cbl-independent β-oxidation-like pathway) that contribute to TCDD hepatotoxicity and progression of steatosis to steatohepatitis with fibrosis.

Core claims
  • TCDD dose-dependently decreases cobalamin (Cbl) levels, compromising MUT activity and redirecting propionyl-CoA metabolism from the canonical Cbl-dependent carboxylation pathway to the alternate Cbl-independent β-oxidation-like pathway. finding
  • Acrylyl-CoA, a reactive intermediate of the Cbl-independent pathway, accumulates in liver after TCDD treatment, as evidenced by increased S-(2-carboxyethyl)-L-cysteine (SCEC), its spontaneous conjugate with cysteine. finding
  • TCDD represses genes in both the canonical Cbl-dependent carboxylase pathway (Pcca, Pccb) and the alternate Cbl-independent β-oxidation-like pathway (Echs1, Adhfe1, Aldh6a1). finding
  • TCDD inhibits methylmalonyl-CoA mutase (MUT) activity, particularly at lower doses. finding
  • TCDD decreases serum Cbl and hepatic cobalt levels while having negligible effects on gene expression for Cbl absorption, transport, trafficking, or derivatization to AdoCbl. finding
  • TCDD induces Acod1 (aconitate decarboxylase 1) and dose-dependently increases hepatic itaconate levels. finding
  • Itaconate is proposed to be activated to itaconyl-CoA, a MUT suicide inactivator that adducts adenosylcobalamin, inhibiting MUT and depleting Cbl. mechanism
  • Elevated SCEC confirms both acrylyl-CoA accumulation and inhibition of short chain enoyl-CoA hydratase (ECHS1) activity. finding
Experimental setups
Assay System Perturbation Readout Platform
Targeted LC-MS/MS metabolomics liver extracts, male mice (C57BL/6) TCDD oral gavage every 4 days for 28 days (dose response) S-(2-carboxyethyl)-L-cysteine (SCEC) levels LC-MS/MS
Untargeted metabolomics hepatic extracts, mice TCDD oral gavage every 4 days for 28 days annotation of acrylyl-CoA and 3-hydroxypropionyl-CoA intermediates
ChIP-seq liver, male mice single bolus 30 μg/kg TCDD, 2 h AhR genomic binding/enrichment at propionyl-CoA pathway genes
RNA-seq (time course) liver, male C57BL/6 mice (n=3) single bolus 30 μg/kg TCDD differential gene expression over time (propionyl-CoA metabolism genes)
RNA-seq (dose response) liver, male C57BL/6 mice (n=3) TCDD oral gavage every 4 days for 28 days, multiple doses differential expression of propionyl-CoA and Cbl metabolism genes
RNA-seq intestinal segments (duodenum, jejunum, ileum, colon), mice (n=3) TCDD oral gavage every 4 days for 28 days, dose response expression of Cbl absorption/transport genes (e.g. Cubn)
ELISA serum, male C57BL/6 mice (n=4-5) TCDD oral gavage every 4 days for 28 days, dose response serum cobalamin levels ELISA assay
Inductively coupled plasma mass spectrometry (ICP-MS) liver extracts, mice (n=4-5) TCDD oral gavage every 4 days for 28 days, dose response hepatic cobalt levels ICP-MS
Key results
  • Hepatic SCEC levels increased at 30 μg/kg TCDD vs vehicle 15.62-fold ± 2.59
  • AhR enrichment associated with repression of Pcca and Pccb (propionyl-CoA carboxylase subunits) 2.7-fold and 2.9-fold
  • Echs1 and Adhfe1 repressed by TCDD (AhR enrichment detected) 1.5-fold and 2.2-fold
  • Aldh6a1 repressed by 30 μg/kg TCDD without detectable AhR enrichment 2.2-fold
  • Serum cobalamin and hepatic cobalt levels dose-dependently decreased by TCDD
  • Cubn repressed in duodenum, jejunum, proximal ileum, and colon but induced in distal ileum 4.2-, 16.7-, 4.6-, 2.0-fold repression; 1.9-fold induction (distal ileum)
  • Tcn2 repressed only at highest TCDD dose 1.5-fold at 30 μg/kg
  • Gut microbiome Cbl biosynthesis gene (precorrin-3 methylase) modestly repressed; ABC cobalt transporter genes increased 1.3-fold repression; 3-fold increase
Key statistics
  • fold_change 15.62 ± 2.59 (SCEC fold-change at 30 μg/kg TCDD vs vehicle, liver extracts)
  • pvalue p ≤ 0.05 (SCEC significance, one-way ANOVA with Dunnett's post-hoc)
  • fold_change 2.7 and 2.9-fold repression (Pcca and Pccb repression, ChIP-seq/RNA-seq 2 h after 30 μg/kg TCDD bolus)
  • fold_change 1.5-fold repression (Echs1 repression)
  • fold_change 16.7-fold repression (Cubn repression in jejunum)
  • fold_change 1.9-fold induction (Cubn induction in distal ileum)
  • pvalue p < 0.05 (Serum Cbl and hepatic cobalt reductions, Dunnett's post-hoc test)
  • fold_change 3-fold increase (ABC cobalt transporter gene abundance in cecal metagenomics)

Statistical methods review

Model: sonnet

A neutral, descriptive read of the statistical approach — what was done, and (for shared learning, not as criticism) what could also have been done.

The study used a dose-response design in mice (multiple TCDD doses vs. vehicle control, typically n=3-5 per group) with targeted LC-MS/MS, RNA-seq, ChIP-seq, and metagenomic analyses. Group comparisons for targeted metabolite (SCEC) and biochemical (serum cobalamin, hepatic cobalt) measurements were tested with one-way ANOVA and Dunnett's post-hoc test against vehicle, with significance denoted at p ≤ 0.05/p < 0.05. Differential gene expression from RNA-seq was assessed using a posterior probability threshold (P1(t) > 0.80) rather than a conventional p-value cutoff, and results throughout are largely summarized as fold-changes.

Replicationbiological Sample sizeGroup sizes stated per experiment (e.g., n=3 for RNA-seq time course/dose-response/ChIP-seq contexts, n=5 for targeted LC-MS/MS, n=4-5 for serum/hepatic cobalamin-cobalt measures); no formal power analysis or a priori sample-size justification described in the excerpt GroupsVehicle (sesame oil) vs. multiple ascending TCDD doses (0.01-30 μg/kg range) in male mice Pairingunpaired Randomization/blindingnot stated Dispersionmixed Exact p-valuesno Effect sizesyes Confidence intervalsno Multiplicity correctionDunnett's post-hoc test (following one-way ANOVA)
Statistical tests used
Test Applied to n Assumptions
one-way ANOVA with Dunnett's post-hoc test Table 1 - hepatic S-(2-carboxyethyl)-L-cysteine fold-change across TCDD doses vs. vehicle n = 5 not stated
one-way ANOVA with Dunnett's post-hoc test Figure 3 - serum cobalamin and hepatic cobalt levels across TCDD doses vs. vehicle n = 4-5 not stated
Bayesian posterior probability threshold (P1(t) > 0.80) for differential expression RNA-seq analyses of AhR target genes and propionyl-CoA/cobalamin pathway genes (Figs. 1, 2, 4) n = 3 not stated
Approaches that could also have been used
  • Dose-response comparisons in Table 1 and Figure 3 used one-way ANOVA with Dunnett's post-hoc test to compare each TCDD dose to vehicle.
    Could also: Because the doses form a natural ordered series, a trend test such as Jonckheere-Terpstra or a regression-based dose-response (e.g., a linear or benchmark-dose model) could also be used. — Trend-based approaches directly test for a monotonic dose-dependent relationship and can increase power to detect a graded response, complementing the pairwise dose-vs-control comparisons from Dunnett's test.
  • Table 1 reports variability as ± SEM while Figure 3 reports ± SD for similarly sized groups (n = 4-5).
    Could also: Reporting SD (or a 95% confidence interval) consistently across figures/tables would also convey the spread of the data. — SEM scales down with sample size and can visually understate variability in small groups, whereas SD and CIs are often preferred for directly communicating the dispersion of individual biological replicates.
  • Significance for the targeted metabolite and biochemical assays was reported as a threshold (p ≤ 0.05 or p < 0.05) with an asterisk rather than as exact values.
    Could also: Reporting exact p-values (e.g., p = 0.032) alongside the threshold could also be used. — Exact p-values convey more information about the strength of evidence against the null hypothesis and facilitate meta-analysis or re-evaluation under different alpha levels.
  • RNA-seq differential expression was flagged using a Bayesian posterior probability threshold (P1(t) > 0.80) rather than a conventional frequentist p-value/FDR approach.
    Could also: A standard count-based differential expression pipeline with multiple-testing correction (e.g., DESeq2 or edgeR with Benjamini-Hochberg FDR) could also be applied to the same RNA-seq data. — FDR-controlled frequentist methods are widely used and allow direct comparison of significance calls across studies that report q-values, complementing the Bayesian posterior-probability framework used here.
  • Group sizes (n = 3 for RNA-seq/ChIP-seq groups, n = 5 for LC-MS/MS, n = 4-5 for cobalamin/cobalt assays) were used without a stated power analysis.
    Could also: An a priori power/sample-size calculation based on expected effect size and variance could also be reported. — Stating the power analysis or expected effect size used to select sample sizes helps readers gauge the sensitivity of the study to detect the reported dose-dependent effects, particularly for small n groups.
  • The excerpt does not describe randomization or blinding procedures for dose administration or outcome measurement.
    Could also: Explicit randomized allocation of animals to dose groups and blinded outcome assessment (e.g., blinded LC-MS/MS or histological scoring) could also be described. — Documenting randomization and blinding is a standard practice that helps readers assess the risk of allocation or assessment bias in animal dosing studies.

What was reproduced

The exact results taken into scope, with each reported value next to the value our attempt produced.

Scope — pmid-35931118

Paper: Orlowska K, Fling RR, Nault R, Sink WJ, Schilmiller AL, Zacharewski T. "Dioxin-elicited decrease in cobalamin redirects propionyl-CoA metabolism to the β-oxidation-like pathway resulting in acrylyl-CoA conjugate buildup." J Biol Chem 2022;298(9):102301. PMID 35931118 · PMCID PMC9418907 · DOI 10.1016/j.jbc.2022.102301.

What kind of paper this is

A mechanistic toxicology study. TCDD (2,3,7,8-tetrachlorodibenzo-p-dioxin) is gavaged into male C57BL/6 mice; the authors use targeted/untargeted LC-MS metabolomics (the headline finding: build-up of S-(2-carboxyethyl)-L-cysteine, an acrylyl-CoA cysteine conjugate), ICP-MS (cobalt), hepatic RNA-seq (dose-response gene expression), and cecal shotgun metagenomics to argue that TCDD lowers cobalamin (B12), inhibiting methylmalonyl-CoA mutase and redirecting propionyl-CoA through a β-oxidation-like pathway.

Pinned metadata vs reality (IMPORTANT — harvested metadata is loosely coupled)

The room was seeded with data=GSE119780 and code=github.com/biobakery/Maaslin2. These belong to two different analysis threads and do not pair:

  • GSE119780 is the lab's circadian hepatic RNA-seq series ("RNA-Seq Analysis of TCDD-Elicited Changes in Circadian Hepatic Gene Expression", 48 samples, 8 timepoints). It is one of nine reused GEO RNA-seq/ChIP-seq series cited in the Data Availability statement (GSE109863, GSE87519, GSE119780, GSE97634, GSE87542, GSE90097, GSE171942, GSE89430, GSE171941). It is not the source of the headline hepatic fold-changes (those come from the dose-response liver RNA-seq, GSE109863), and the paper reports no GSE119780-specific numbers.
  • Maaslin2 was used only for the shotgun-metagenomics thread, whose data is PRJNA719224 (NCBI BioProject; 12 cecal metagenomes), NOT GSE119780.

In scope (pipeline-derived, reproducible)

PRIMARY target — Maaslin2 metagenomics (matches the pinned tool exactly). The metagenomics methods are fully specified, so this is the most faithful reproducible pipeline. Data: PRJNA719224, 12 cecal shotgun metagenomes (vehicle / 0.3 / 3 / 30 µg/kg TCDD, n=3 each). Documented pipeline:

  1. Host removal of Mus musculus (GRCm38.p6) reads via bowtie2 + samtools + bedtools.
  2. HUMAnN3 (default) → UniRef90 (UniProt UniRef90, Jan 2019).
  3. humann_regroup_table → Enzyme Commission (EC) and PFAM features.
  4. humann_renorm_table → copies per million reads (cpm).
  5. Maaslin2 default settings: TSS normalization, GLM analysis method, BH correction.

Reported metagenomic outcomes to compare against (Results + Fig S1/S2):

  • Precorrin-3 methylase (EC «ip»): ~1.3-fold repression.
  • ABC cobalt transporters (PFAM PF09819): ~3-fold increase.
  • Overall qualitative claim: "negligible effects on microbial Cbl and propionate metabolism."

Secondary / limited reproducibility (documented, not the primary attempt)

  • Hepatic dose-response gene fold-changes (Fig 2/5: Pcca −2.7×, Pccb −2.9×, Mmab −2.4×, Tcn2 −1.5×, Echs1 −1.5×, Adhfe1 −2.2× at 30 µg/kg; Acod1 induction). Source = dose-response liver RNA-seq (GSE109863, not the pinned GSE119780). The DE method is the lab's custom empirical-Bayes posterior-probability pipeline (|FC|≥1.5, P1(t)≥0.8) — not a standard public package (edgeR/DESeq2/ limma). Exact 1:1 reproduction is therefore limited; a standard-pipeline approximation could be attempted but would not be method-identical.

Out of scope (wet-lab / instrument / manual — not attempted)

  • LC-MS targeted & untargeted metabolomics (incl. the S-(2-carboxyethyl)-L-cysteine headline; raw data Metabolomics Workbench ST001379).
  • ICP-MS cobalt quantification.
  • qRT-PCR validation; enzyme-activity assays; all wet-lab measurements.

Honesty notes

  • The primary attempt directly exercises the pinned tool (Maaslin2) on the paper's own data (PRJNA719224) per the documented parameters — per the brief
Figures / tables: Fig S1Fig 2Fig 5Fig 6
M1
Reported
precorrin-3 methylase (EC «ip»): ~1.3-fold repression with TCDD (Results, Figs S1-S2)
Reproduced
1.30-fold repression at 30 ug/kg vs vehicle (CPM 7.608->5.872=0.772x); Maaslin2 dose coef -0.0138, q=0.101
within tolerance
M2
Reported
ABC cobalt transporters (PFAM PF09819): ~3-fold increase with TCDD (Results, Figs S1-S2)
Reproduced
2.92-fold increase at 30 ug/kg vs vehicle (CPM 1.777->5.193); Maaslin2 dose coef +0.0552, q=0.151
within tolerance
M3
Reported
'TCDD elicited negligible effects on microbial Cbl and propionate metabolism' (qualitative)
Reproduced
Consistent: named Cbl features modest (1.3x down / 2.9x up); broad metagenome does shift (135/1576 EC, 252/5469 Pfam q<0.25) but Cbl/propionate-specific effects modest
partial
R1-R7
Reported
hepatic dose-response fold-changes (Pcca -2.7x, Pccb -2.9x, Mmab -2.4x, Tcn2 -1.5x, Echs1 -1.5x, Adhfe1 -2.2x at 30 ug/kg; Acod1 induction)
Reproduced
NOT_ATTEMPTED
m.public.grade.not-attempted

Assessments & scoring basis

Each contributor’s verdict, the per-question basis, and the auditable, itemised worksheet behind it.

🤖 AI curator · claude (ai-curator room) · v1.0 L1 73/100

An automated assessment. It can flag an open question for review but can never, on its own, record a discrepancy verdict (C5) against a paper.

🟡1. Data identity
🟡2. Endpoint comparability
🟡3. Location of the main deviation
🟡4. Cause of the deviation
🟡5. Derivability / plausibility
🟡6. Severity of the deviation
🟡7. Core claim
🟡8. Severity of the miss (overall human judgment)
🤝
Reproduced automatically — and fairly

Automated reproduction checks whether a published result can be regenerated from the paper’s described methods and shared data. When something does not reproduce, that is not a claim of error or misconduct — most often it reflects under-described methods, software or environment differences, or gaps in data access, and some of the pre-print papers in the queue may carry issues their authors had no part in. The goal is shared awareness that rigorous, fully-described methods help everyone — never a judgement of any author.

Are you an author? We would genuinely like to hear from you — to clarify the record, add data or code, re-run the pipeline after an accession update, and publish your response right next to the assessment. Everything here is open and auditable.

🚩 Report an error in this record

Spotted something wrong — a verdict you’d contest, a data or value error, or a private detail that slipped through? Tell us, with a short justification. Authors and readers are equally welcome to write in; we review every report.

Prefer email, or the form below not working? Contact us at support@doesitreproduce.com.

Reproduction footprint

claude-opus-4-8

Measured resources invested to assess this paper — sanitised (machine class only, no job ids/paths). Compute = HPC accounting (SLURM); tokens = the AI agent's session.

102.5 k
tokens (I/O) · 4.2 M incl. cache
24 min
runtime
Per-job HPC accounting not captured for this run — the runtime shown is the reproduction’s measured wall-clock time.