CoChat
Sign Up
Cited by 902Histone Deacetylase Inhibitors ResearchCholinesterase and Neurodegenerative DiseasesEnzyme function and inhibition

Rational Design and Simple Chemistry Yield a Superior, Neuroprotective HDAC6 Inhibitor, Tubastatin A

Open access
Butler, Kyle V., Kalin, Jay, Brochier, Camille, Vistoli, Guilio, Langley, Brett, Kozikowski, Alan P.·2010·Journal of the American Chemical Society
DOI 10.1021/ja102758vPMC PMC2916045
Open this paper in CoChat — free account

Abstract

Structure-based drug design combined with homology modeling techniques were used to develop potent inhibitors of HDAC6 that display superior selectivity for the HDAC6 isozyme compared to other inhibitors. These inhibitors can be assembled in a few synthetic steps, and thus are readily scaled up for in vivo studies. An optimized compound from this series, designated Tubastatin A, was tested in primary cortical neuron cultures in which it was found to induce elevated levels of acetylated alpha-tubulin, but not histone, consistent with its HDAC6 selectivity. Tubastatin A also conferred dose-dependent protection in primary cortical neuron cultures against glutathione depletion-induced oxidative stress. Importantly, when given alone at all concentrations tested, this hydroxamate-containing HDAC6-selective compound displayed no neuronal toxicity, thus, forecasting the potential application of this agent and its analogues to neurodegenerative conditions.

AI snapshot

Written from the paper’s full text.

TL;DR

Structure-based design and homology modeling produced carbazole hydroxamic acid HDAC6 inhibitors, with Tubastatin A (compound 6) showing high HDAC6 potency and selectivity. Tubastatin A induced α-tubulin acetylation and dose-dependently protected primary cortical neurons from oxidative stress without neuronal toxicity when tested alone.

Key findings

  • Tubastatin A (6) had an HDAC6 IC50 of 15 nM and over 1000-fold selectivity against HDAC1; compound 7 was even more potent with an HDAC6 IC50 of 1.4 nM and approximately 3700-fold selectivity against HDAC1.
  • Against all 11 HDAC isoforms, Tubastatin A maintained over 1000-fold selectivity except at HDAC8, where it had approximately 57-fold selectivity; Tubacin had an HDAC6 IC50 of 4 nM and approximately 350-fold selectivity over HDAC1.
  • In primary cortical neuron cultures, Tubastatin A preferentially induced α-tubulin hyperacetylation at 2.5 µM, with slight histone hyperacetylation at 10 µM.
  • Tubastatin A gave dose-dependent protection against homocysteic acid-induced neuronal death starting at 5 µM with near complete protection at 10 µM, and showed no toxicity when tested alone at all concentrations shown.

Methods

  • HDAC1 and HDAC6 homology models were built using I-TASSER with resolved HDAC crystal structures as templates; catalytic zinc and chelating residues were extracted from human HDAC8 complexed with trichostatin A (PDB: 3FOR), and docking used AutoDock4.0.
  • Inhibitor selectivity was monitored by comparing IC50 values at purified human HDAC6 versus HDAC1; Tubastatin A and Tubacin were tested at all 11 HDAC isoforms.
  • Neuroprotection assays used primary cortical neuron cultures from fetal Sprague-Dawley rats (embryonic day 17), with oxidative stress induced by 5 mM homocysteic acid and viability assessed after 24 h by MTT assay.

Limitations

  • Stated: crystal structures for HDAC1 and HDAC6 were lacking, so homology models were used; Tubastatin A has moderate HDAC8 activity (approximately 57-fold selectivity); multiple in vivo studies are deferred to future reports.
  • Evident from the text: neuroprotection data are limited to primary cortical neuron cultures and a single homocysteic acid oxidative-stress model, with no in vivo efficacy or toxicity data presented in this paper.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.1021/ja102758v

Cites

29 of 29 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Claims the paper makes

0 of 7 cited sources checked · 0 supported · 0 flagged

  • Findingown data

    Tubastatin A conferred dose-dependent protection in primary cortical neuron cultures against glutathione depletion-induced oxidative stress.

    “Tubastatin A also conferred dose-dependent protection in primary cortical neuron cultures against glutathione depletion-induced oxidative stress.”
    Quote in the paperSubstantial claim
    • Rests on an assumption: Under the conditions of primary cortical neuron cultures, the cells are not susceptible to glutamate-mediated excitotoxicity.
    • Rests on an assumption: The HCA-induced glutathione depletion model in primary cortical neurons is a valid model of neurodegeneration relevant to human neurodegenerative conditions.
  • Findingown data

    The developed HDAC6 inhibitors display superior selectivity for the HDAC6 isozyme compared to other inhibitors.

    “Structure-based drug design combined with homology modeling techniques were used to develop potent inhibitors of HDAC6 that display superior selectivity for the HDAC6 isozyme compared to other inhibitors.”
    Quote in the paperSubstantial claim
    • Rests on an assumption: The homology models of HDAC1 and HDAC6 accurately represent the catalytic pockets of the actual enzymes.
    • Rests on an assumption: The D-B distance measured in the energy-minimized homology models reflects the actual channel dimensions of HDAC1 and HDAC6.
  • Findingown data

    Tubastatin A displayed no neuronal toxicity when given alone at all concentrations tested.

    “Importantly, when given alone at all concentrations tested, this hydroxamate-containing HDAC6-selective compound displayed no neuronal toxicity”
    • Also stated in: § Analysis of Tubastatin A in a model of neurodegeneration · plain
      “Importantly, when tested alone at all of the concentrations shown, Tubastatin A exhibited no toxicity”
    Quote in the paperSubstantial claim
    • Rests on an assumption: Under the conditions of primary cortical neuron cultures, the cells are not susceptible to glutamate-mediated excitotoxicity.
  • Findingown data

    Rational drug design combined with homology modeling yielded potent HDAC6 inhibitors.

    “Structure-based drug design combined with homology modeling techniques were used to develop potent inhibitors of HDAC6 that display superior selectivity for the HDAC6 isozyme compared to other inhibitors.”
    Quote in the paperSubstantial claim
    • Rests on an assumption: The homology models of HDAC1 and HDAC6 accurately represent the catalytic pockets of the actual enzymes.
    • Rests on an assumption: The D-B distance measured in the energy-minimized homology models reflects the actual channel dimensions of HDAC1 and HDAC6.
  • Findingown data

    Tubastatin A displayed dose-dependent protection against HCA-induced neuronal cell death starting at 5 μM with near complete protection at 10 μM.

    “Tubastatin A displayed dose-dependent protection against HCA-induced neuronal cell death starting at 5 µM with near complete protection at 10 µM. (Figure 3)”
    Quote in the paperSubstantial claim
    • Rests on an assumption: Under the conditions of primary cortical neuron cultures, the cells are not susceptible to glutamate-mediated excitotoxicity.
    • Rests on an assumption: The HCA-induced glutathione depletion model in primary cortical neurons is a valid model of neurodegeneration relevant to human neurodegenerative conditions.
  • May combine two claims. 20
  • Classified as probably not a substantial claim. 7

  • May combine two claims.

    Findingown data

    Tubastatin A induced elevated levels of acetylated α-tubulin, but not histone, in primary cortical neuron cultures.

    “Tubastatin A, was tested in primary cortical neuron cultures in which it was found to induce elevated levels of acetylated α-tubulin, but not histone, consistent with its HDAC6 selectivity.”
    Quote in the paperSubstantial claim
    • Rests on an assumption: Under the conditions of primary cortical neuron cultures, the cells are not susceptible to glutamate-mediated excitotoxicity.
    • Rests on an assumption: The selectivity of Tubastatin A in enzyme assays translates to selective HDAC6 inhibition in cells.

  • May combine two claims.

    Findingreasoning

    These inhibitors can be assembled in a few synthetic steps and are readily scaled up for in vivo studies.

    “These inhibitors can be assembled in a few synthetic steps, as thus are readily scaled up for in vivo studies.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    Analogs of 1 (compounds 2-4) were less potent than the parent, but all preferentially inhibited HDAC6.

    “Analogs of 1 (compounds 2–4) were less potent than the parent, but all preferentially inhibited HDAC6.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingreasoning

    Rational drug design principles combined with homology modeling have made it possible to create these novel chemical entities which can be assembled in multi-gram quantities in a few days time.

    “Rational drug design principles combined with homology modeling have made it possible to create these novel chemical entities which can be assembled in multi-gram quantities in a few days time.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    In HCA-induced neurodegeneration assays, TSA was moderately neuroprotective at 0.5 μM, although protection declined at higher concentrations due to dose-dependent neurotoxicity.

    “In HCA-induced neurodegeneration assays, TSA was moderately neuroprotective at 0.5 µM, although protection declined at higher concentrations due to dose-dependant neurotoxicity.”
    Quote in the paperSubstantial claim
    • Rests on an assumption: Under the conditions of primary cortical neuron cultures, the cells are not susceptible to glutamate-mediated excitotoxicity.
    • Rests on an assumption: The HCA-induced glutathione depletion model in primary cortical neurons is a valid model of neurodegeneration relevant to human neurodegenerative conditions.

  • May combine two claims.

    Findingreasoning

    When attached to a zinc-chelating hydroxamic acid by an appropriate linker, carbazole fits the D region of HDAC6, whereas it clashes with the B and D regions in HDAC1.

    “When attached to a zinc-chelating hydroxamic acid by an appropriate linker, the carbazole nicely fits the D region of HDAC6, whereas it clashes with the B and D regions in HDAC1 (Figure 3, supporting information).”
    Quote in the paperSubstantial claim
    • Rests on an assumption: The homology models of HDAC1 and HDAC6 accurately represent the catalytic pockets of the actual enzymes.
    • Rests on an assumption: The D-B distance measured in the energy-minimized homology models reflects the actual channel dimensions of HDAC1 and HDAC6.

  • May combine two claims.

    Findingown data

    Compounds 6 and 7 have ClogP = 2.41, tPSA = 57, and AlogPs water solubility = 45.2 mg/l for 6 and 43.7 mg/l for 7.

    “These agents are also very druggable: ClogP = 2.41 (KOWWIN) and tPSA = 57 for both 6 and 7; AlogPs water solubility = 45.2 mg/l for 6 and 43.7 mg/l for 7.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingreasoning

    Modifications to the tricyclic group did not significantly enhance inhibitory activity when employed with the n-pentyl linker, but greatly enhanced activity and selectivity when used with the tolyl linker.

    “We note that modifications to the tricyclic group did not significantly enhance inhibitory activity when employed in conjunction with the n-pentyl linker, but greatly enhanced activity and selectivity when used with the tolyl linker”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingnone given

    For compound 5, the angle between the rotational axis of the tolyl linker and the C(sp3)-N bond is fixed at ~113°, forcing the carbazole to lay against the catalytic channel rim.

    “For 5, the angle between the rotational axis of the tolyl linker and the C(sp3)-N bond connecting linker and cap is fixed at ~113°, forcing the carbazole to lay against the catalytic channel rim, stressing the reported differences between the two rim regions.”
    Quote in the paperSubstantial claim
    • Rests on an assumption: The homology models of HDAC1 and HDAC6 accurately represent the catalytic pockets of the actual enzymes.
    • Rests on an assumption: The D-B distance measured in the energy-minimized homology models reflects the actual channel dimensions of HDAC1 and HDAC6.

  • May combine two claims.

    Findingown data

    The measured D-B distance in the energy minimized conformation is 12.5 Å in HDAC1 compared to 17.5 Å in HDAC6.

    “The structural manifestation of this variation is that the measured D-B distance in the energy minimized conformation is 12.5 Å in HDAC1, compared to 17.5 Å in HDAC6.”
    Quote in the paperSubstantial claim
    • Rests on an assumption: The homology models of HDAC1 and HDAC6 accurately represent the catalytic pockets of the actual enzymes.
    • Rests on an assumption: The D-B distance measured in the energy-minimized homology models reflects the actual channel dimensions of HDAC1 and HDAC6.

  • May combine two claims.

    Findingown data

    The modeled catalytic channels of HDAC1 and HDAC6 differ in shape; the HDAC6 channel appears wider and shallower.

    “While lined with conserved residues, the modeled catalytic channels of HDAC1 and HDAC6 also differ in shape. The HDAC6 channel appears wider and shallower”
    Quote in the paperSubstantial claim
    • Rests on an assumption: The homology models of HDAC1 and HDAC6 accurately represent the catalytic pockets of the actual enzymes.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Findingnone given

    The facile, three-step syntheses of compounds 6 and 7 allow easy scale-up for in vivo studies.

    “Furthermore, their facile, three-step syntheses allow easy scale-up for in vivo studies.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    Tubacin potently inhibited HDAC6, with an IC50 value of 4 nM and approximately 350-fold selectivity over HDAC1.

    “Tubacin was found to potently inhibit HDAC6, with an IC50 value of 4 nM and approximately 350-fold selectivity over HDAC1.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    The active site is highly conserved between HDAC1 and HDAC6, but the dimensions of the catalytic channel rim differ greatly.

    “Analysis of the two modeled catalytic pockets revealed that while the active site is highly conserved, the dimensions of the catalytic channel rim differ greatly between the two isozymes.”
    Quote in the paperSubstantial claim
    • Rests on an assumption: The homology models of HDAC1 and HDAC6 accurately represent the catalytic pockets of the actual enzymes.

  • May combine two claims.

    Findingown data

    The tetrahydro-β-carboline 7 was even better, with an IC50 of 1.4 nM at HDAC6 and approximately 3700-fold selectivity against HDAC1.

    “The tetrahydro-β-carboline, 7, was even better, with an IC50 of 1.4 nM at HDAC6 and approximately 3700-fold selectivity against HDAC1.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    The tetrahydro-γ-carboline analog 6 was superior to the parent in HDAC6 potency and selectivity, with an IC50 of 15 nM at HDAC6 and over 1000-fold selectivity against HDAC1.

    “The tetrahydro-γ-carboline analog, 6, was superior to the parent in HDAC6 potency and selectivity, with an IC50 of 15 nM at HDAC6 and over 1000-fold selectivity against HDAC1.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    Tubastatin A was substantially more selective than Tubacin at all isozymes except HDAC8 and maintained over 1000-fold selectivity against all isoforms excluding HDAC8, where it had approximately 57-fold selectivity.

    “Tubastatin A was substantially more selective than Tubacin at all isozymes except HDAC8 and maintained over 1000-fold selectivity against all isoforms excluding HDAC8, where it had approximately 57-fold selectivity.”
    Quote in the paperSubstantial claim
    • Rests on an assumption: The selectivity of Tubastatin A in enzyme assays translates to selective HDAC6 inhibition in cells.

  • Classified as probably not a substantial claim.

    Premisecitation

    Isozyme selective inhibitors may hold greater therapeutic promise than their nonselective counterparts.

    “isozyme selective inhibitors may hold greater therapeutic promise than their nonselective counterparts.10”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites [10] Thomas E. A., 2009, Mol. Neurobiol. source

      Not checked in the free run.

  • Classified as probably not a substantial claim.

    Interpretationreasoning

    Replacement of the traditional alkyl chain linker with bulkier and shorter aromatic moieties might further enhance HDAC6 selectivity.

    “suggesting that replacement of the traditional alkyl chain linker with bulkier and shorter aromatic moieties might further enhance HDAC6 selectivity.”
    Quote in the paperSubstantial claim
    • Rests on an assumption: The homology models of HDAC1 and HDAC6 accurately represent the catalytic pockets of the actual enzymes.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Study design and conductnone given

    A preliminary energy minimization was performed on the homology models to avoid high-energy interactions, with backbone atoms fixed to preserve the predicted folding.

    “A preliminary energy minimization was performed on the models to avoid high-energy interactions, in which the backbone atoms were fixed to preserve the predicted folding.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Premisecitation

    Tubacin's non-drug-like structure, high lipophilicity, and tedious synthesis make it more useful as a research tool than a drug.

    “The use of Tubacin in models of disease has helped to validate, in part, HDAC6 as a drug target, but its non-drug-like structure, high lipophilicity (ClogP = 6.36 (KOWWIN)), and tedious synthesis conspire to make it more useful as a research tool than a drug.21”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites [21] Haggarty S. J., 2003, Proc. Natl. Acad. Sci. U.S.A. source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Study design and conductnone given

    The complete homology models were carefully checked to avoid unphysical occurrences such as cis peptide bonds, wrong configurations or colliding side-chains.

    “The complete models were carefully checked to avoid unphysical occurrences such as cis peptide bonds, wrong configurations or colliding side-chains.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Interpretationreasoning

    The slight histone hyperacetylation at 10 μM may reflect inhibition of the limited histone deacetylase activity of HDAC6 or slight inhibition of class I HDACs at higher concentrations.

    “which may reflect inhibition of the limited histone deacetylase activity of HDAC6 or slight inhibition of class I HDACs at these higher concentrations.”
    Quote in the paperNot a substantial claim
    • Rests on an assumption: The selectivity of Tubastatin A in enzyme assays translates to selective HDAC6 inhibition in cells.

  • May combine two claims.

    Interpretationown data

    These compounds confirm the structure-based design approach, as less than 30 compounds were made to arrive at these findings.

    “These compounds confirm our structure based design approach, as less than 30 compounds were made to arrive at these findings.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Interpretationnone given

    The tertiary amine group can be used to form pharmaceutically useful salts, facilitating compound solubilization.

    “The tertiary amine group can be used to form pharmaceutically useful salts, thus facilitating compound solubilization.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Interpretationown data

    Neurotoxicity is likely a product of class I HDAC inhibition, and not a property inherent to hydroxamic acids.

    “indicating that neurotoxicity is likely a product of class I HDAC inhibition, and not a property inherent to hydroxamic acids.”
    Quote in the paperSubstantial claim
    • Rests on an assumption: The lack of toxicity of Tubastatin A when tested alone indicates that neurotoxicity is due to class I HDAC inhibition rather than a property of hydroxamic acids.
26 of 26

  • The paper relies on this without support.

    Under the conditions of primary cortical neuron cultures, the cells are not susceptible to glutamate-mediated excitotoxicity.

    “Under these conditions, the cells are not susceptible to glutamate-mediated excitotoxicity.”

  • The paper relies on this without support.

    The homology models of HDAC1 and HDAC6 accurately represent the catalytic pockets of the actual enzymes.

    “we generated reliable models for these isozymes by employing homology techniques.”

  • The paper relies on this without support.

    The D-B distance measured in the energy-minimized homology models reflects the actual channel dimensions of HDAC1 and HDAC6.

    “The structural manifestation of this variation is that the measured D-B distance in the energy minimized conformation is 12.5 Å in HDAC1, compared to 17.5 Å in HDAC6.”

  • The paper relies on this without support.

    The HCA-induced glutathione depletion model in primary cortical neurons is a valid model of neurodegeneration relevant to human neurodegenerative conditions.

    “Tubastatin A was examined in a model of oxidative stress induced by homocysteic acid (HCA). This model leads to depletion of glutathione, the cells major intracellular antioxidant.”

  • The paper relies on this without support.

    The neuroprotective effect of Tubastatin A in the HCA model is mediated by HDAC6 inhibition.

    “These results provide further evidence that the inhibition of HDAC6 may provide a means for achieving therapeutic intervention in neurodegenerative conditions.”

  • The paper relies on this without support.

    The selectivity of Tubastatin A in enzyme assays translates to selective HDAC6 inhibition in cells.

    “Selectivity of Tubastatin A was analyzed in the more complex cellular environment by comparing α-tubulin hyperacetylation, corresponding to HDAC6 inhibition, with histone hyperacetylation, which corresponds to class I HDAC inhibition.”

  • The paper relies on this without support.

    The lack of toxicity of Tubastatin A when tested alone indicates that neurotoxicity is due to class I HDAC inhibition rather than a property of hydroxamic acids.

    “Importantly, when tested alone at all of the concentrations shown, Tubastatin A exhibited no toxicity, indicating that neurotoxicity is likely a product of class I HDAC inhibition, and not a property inherent to hydroxamic acids.”
7 of 7

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

Cites

Flagged first, then by citations

Cited by

Sorted by citations

Related

Authors
Butler, K. V., Kalin, J., Brochier, C., Vistoli, G., Langley, B., & Kozikowski, A. P.
Affiliations
  • University of Milan (IT)
  • Drug Discovery Program, Department of Medicinal Chemistry and Pharmacognosy, University of Illinois at Chicago, 833 South Wood, Chicago, Illinois 60612, Burke Medical Research Institute, 785 Mamaroneck Avenue, White Plains, New York 10605, and Dipartimento di Scienze Farmaceutiche “Pietro Pratesi”, Università degli Studi di Milano, Via Mangiagalli 25, I-20133 Milan, Italy
Corresponding author
Kozikowski, Alan P.
ORCID
Venue
Journal of the American Chemical Society
Publisher
American Chemical Society (ACS)
Published
2010-07-09
Volume
132
Issue
31
Pages
10842-10846
ISSN
  • 0002-7863 (ISSN-L)
  • 1520-5126
DOI
10.1021/ja102758v
PMC
PMC2916045
OpenAlex
W2063078660
Semantic Scholar
5fc0542f71e788eef76d1e405e7d2c2c14738530
Cited by
801
References
29
Citations by year
2026: 50 · 2025: 49 · 2024: 51 · 2023: 59 · 2022: 55 · 2021: 53 · 2020: 56 · 2019: 65 · 2018: 57 · 2017: 51 · 2016: 54 · 2015: 53 · 2014: 46 · 2013: 50 · 2012: 24
FWCI
14.45
Type
journal article
Language
en
Open access
Locations
  • Journal of the American Chemical Society (publishedVersion)
  • PubMed (publishedVersion)
  • Europe PMC (PubMed Central) (acceptedVersion)
  • PubMed Central (acceptedVersion)
Funders
National Institute on Aging (R01AG022941)
Topics
  • Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › Histone Deacetylase Inhibitors Research
  • Health Sciences › Medicine › Pharmacology › Cholinesterase and Neurodegenerative Diseases
  • Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › Enzyme function and inhibition
Keywords
tubastatin A, HDAC6 inhibitors, HDAC6, structure-based drug design, homology modeling, hydroxamate, acetylated alpha-tubulin, cortical neurons, oxidative stress, glutathione depletion, neuroprotection, neurodegenerative diseases, neurotoxicity
MeSH terms
Histone Deacetylase 6, Anilides, Dose-Response Relationship, Drug, Enzyme Inhibitors, Histone Deacetylases, Homocysteine, Humans, Hydroxamic Acids, Indoles, Isoenzymes, Models, Molecular, Neurons, Stereoisomerism, Structure-Activity Relationship, Drug Design, Molecular Structure, Cell Death, Oxidative Stress, Neuroprotective Agents

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Claims read by deepseek-v4.1-flash under definition version 3 (extract-v3) on 2026-10-09.

Classified by jev-1.13-20260917 (gold-guideline-v0.2) on 2026-10-09.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review