CoChat
Sign Up
Cited by 39Epigenetics and DNA MethylationGenomics and Chromatin DynamicsCancer-related molecular mechanisms research

Alternative promoters in CpG depleted regions are prevalently associated with epigenetic misregulation of liver cancer transcriptomes

Open access
Nepal, Chirag, Andersen, Jesper B.·2023·Nature Communications
DOI 10.1038/s41467-023-38272-4PMC PMC10175279
Open this paper in CoChat — free accountOpen PDF

Abstract

Transcriptional regulation is commonly governed by alternative promoters. However, the regulatory architecture in alternative and reference promoters, and how they differ, remains elusive. In 100 CAGE-seq libraries from hepatocellular carcinoma patients, here we annotate 4083 alternative promoters in 2926 multi-promoter genes, which are largely undetected in normal livers. These genes are enriched in oncogenic processes and predominantly show association with overall survival. Alternative promoters are narrow nucleosome depleted regions, CpG island depleted, and enriched for tissue-specific transcription factors. Globally tumors lose DNA methylation. We show hierarchical retention of intragenic DNA methylation with CG-poor regions rapidly losing methylation, while CG-rich regions retain it, a process mediated by differential SETD2 , H3K36me3, DNMT3B , and TET1 binding. This mechanism is validated in SETD2 knockdown cells and SETD2 -mutated patients. Selective DNA methylation loss in CG-poor regions makes the chromatin accessible for alternative transcription. We show alternative promoters can control tumor transcriptomes and their regulatory architecture.

AI snapshot

Written from the paper’s full text.

TL;DR

In 100 CAGE-seq libraries from hepatocellular carcinoma patients, this study annotates 4083 alternative promoters in 2926 multi-promoter genes, most undetected in normal liver and enriched in oncogenic processes and survival association. It proposes that CG-poor intragenic regions preferentially lose DNA methylation in tumors, mediated by SETD2/H3K36me3/DNMT3B/TET1, making chromatin accessible for alternative transcription.

Key findings

  • The authors annotate 4083 alternative TSSs (3052 annotated and 1031 novel) in 2926 multi-promoter genes from 50 tumors and 50 matched tumor-adjacent normal liver CAGE-seq libraries. About 50% of alternative promoters were unexpressed in normal liver at a minimum 3 TPM threshold, compared with 15% of reference promoters.
  • Alternative promoters are CpG island depleted: 48.6% (1987/4083) overlap CGIs versus 80% of reference promoters; novel alternative promoters are further CGI-depleted (P = 1.9E−61). They are narrow/sharp, nucleosome-depleted, and enriched for tissue-specific transcription factors such as HNF1A and HNF4A, with NFATC2 enriched only on nonCGI alternative promoters.
  • Multi-promoter genes are enriched in oncogenic processes and survival association: overall survival is overrepresented among MP genes (P = 3.2E−78), alternative promoter expression is associated with patient outcome (P = 5.2E−24), and survival-associated alternative promoters have shorter OS (P = 2.8E−8).
  • Tumors show hierarchical retention of intragenic DNA methylation: CG-poor regions lose methylation while CG-rich regions retain it, associated with differential SETD2, H3K36me3, DNMT3B, and TET1 binding. SETD2-mutant tumors (n = 15 vs 362 wild-type in the results) were significantly hypomethylated with demethylated CpGs enriched in gene bodies overlapping CGIs, and SETD2 mutants had high intronic reads.

Methods

  • CAGE-seq reads from 100 HCC patient libraries were mapped to hg19 with bowtie2; CTSSs were clustered into transcript clusters, and consensus TCs were retained if expressed >1 TPM in at least 15 samples and ≥3 TPM in at least one sample. Genes were classified as single-promoter or multi-promoter; the highest mean-expressed promoter expressed in normal liver was assigned as reference, and remaining promoters as alternative.
  • Novel alternative promoters were annotated by filtering intragenic CAGE TCs for miRNA/snoRNA processing, exonic post-transcriptional processing, enhancer RNAs, GG-initiation, and lack of initiator motifs, then requiring YR/YC initiation and support from RNA-seq/EST transcripts.
  • Integrated analyses included histone ChIP-seq from four HCC patients and HepG2 cells, TCGA/ENCODE ATAC-seq, TCGA HCC 450K methylation data (379 tumors and 51 tumor-adjacent tissues), RRBS/TAB-RRBS, SETD2 knockdown in HepG2, SETD2-mutant versus wild-type TCGA HCC tumors, DESeq2 differential expression, Kaplan–Meier survival analysis, and HOMER motif analysis.

Limitations

  • The paper states there is no consensus on which promoter should be assigned as reference versus alternative; a highly expressed promoter is often annotated as reference, and in some cases the annotation is interchangeable across cohorts/studies.
  • The paper states that the biological significance of alternative promoters in the full spectrum of liver cancer remains unexplored, that single-cell 5′-end methods are needed to resolve cell-type specificity, and that standard RNA-seq analyses often miss alternative promoter usage.
  • Evident from the text: the main epigenomic and functional validations rely on bulk HCC cohorts and HepG2 cells, so cell-type-specific effects and primary tumor heterogeneity are not fully resolved; the SETD2-mutant sample size is reported inconsistently between results (n = 15 vs 362) and methods (n = 16 vs 352).

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

Integrity signals

Registry

Crossref · 10.1038/s41467-023-38272-4

Cites

78 of 79 references verified0 citation errors · 1 not checkable

Statistics

Nothing to recompute.

Claims the paper makes

58 of 68 cited sources checked · 29 supported · 1 flagged · accuracy not yet measured

  • Interpretationreasoning

    Depletion of H3K4me3 on nonCGI promoters and enhancers is due to low CG density.

    “This reinforces the premise that H3K4me3 is linked to CG density47, and depletion of H3K4me3 on nonCGI promoters and enhancers is due to low CG density.”
    Quote in the paperSubstantial claim1 of 1 sources checked
    • cites [47] Thomson, J. P. et al. CpG islands influence chromatin structure via the CpG-binding protein Cfp1. Nature 464, 1082–1086 (2010). source
  • Premisecitation

    H3K4me3 marks are influenced by CG density.

    “H3K4me3 marks are influenced by CG density47.”
    Quote in the paperSubstantial claim1 of 1 sources checked
    • cites [47] Thomson, J. P. et al. CpG islands influence chromatin structure via the CpG-binding protein Cfp1. Nature 464, 1082–1086 (2010). source
  • Findingown data

    A majority of novel TSSs were supported by spliced transcripts from RNA-seq and expressed sequence tags.

    “A majority of the novel TSSs were supported by spliced transcripts (see methods) from RNA-seq25,26 and expressed sequence tag27”
    Quote in the paperSubstantial claim3 of 3 sources checked
    • cites [25] Hon, C. C. et al. An atlas of human long non-coding RNAs with accurate 5’ ends. Nature 543, 199–204 (2017). source

      The ‘Building a 5′ complete transcriptome’ section describes identifying TSSs using transcript models, including an RNA-seq assembly, and considers transcript exon number. The Methods section on FANTOM5 RNA-seq also describes assembling splice junctions. Neither passage establishes that a majority of novel TSSs were supported by spliced transcripts or identifies expressed sequence tags as support. (LLM, conf=0.76)

      Experimental — accuracy not yet measured
    • cites [26] Iyer, M. K. et al. The landscape of long noncoding RNAs in the human transcriptome. Nat. Genet. 47, 199–208 (2015). source

      The Results section, “Characterization of long RNAs” (page unavailable), reports alternative splicing and chromatin-mark enrichment near TSSs, but does not say that a majority of novel TSSs were supported by spliced transcripts. A search for expressed sequence tags also found no passage establishing that evidence for the claim. (LLM, conf=0.73)

      Experimental — accuracy not yet measured
    • cites [27] Lee, B. T. et al. The UCSC Genome Browser database: 2022 update. Nucleic Acids Res. 50, D1115–D1122 (2022). source

      The NEW GENOME BROWSER DATA section describes genome-browser tracks, but does not report whether novel TSSs were supported by spliced transcripts. Searches for “expressed sequence tag,” “EST,” and “transcription start sites” found no passage establishing that the cited browser update supplied the EST evidence or the claimed majority. (LLM, conf=0.72)

      Experimental — accuracy not yet measured
  • Findingown data

    Divergent H3K4me1 peaks were located further away from TSSs, flanking CGIs and H3K4me3 boundaries.

    “Divergent H3K4me1 peaks were located further away from TSSs (Fig. 4a, b), which flanked CGIs and H3K4me3 boundaries44–46”
    Quote in the paperSubstantial claim2 of 3 sources checked
    • cites [44] Bae, S. & Lesch, B. J. H3K4me1 distribution predicts transcription state and poising at promoters. Front. Cell Dev. Biol. 8, 289 (2020). source

      The Results (Figure 4E; page not provided) show H3K4me1 signal flanking H3K4me3 signal at active promoters, and the Results on peak distribution place many H3K4me1 peaks away from TSSs. The cited paper does not establish that these peaks flank CpG islands (CGIs), so that part of the attributed claim is unsupported. (LLM, conf=0.86)

      Experimental — accuracy not yet measured
    • cites [45] Soares, L. M. et al. Determinants of histone H3K4 methylation patterns. Mol. Cell 68, 773–785.e6 (2017). source

      The Results section (“H3K4 methylation patterns are not universal”; page unavailable) reports that H3K4me1 is generally shifted downstream relative to H3K4me2 and notes possible enrichment associated with divergent promoters. It does not report that divergent H3K4me1 peaks flank CGIs or H3K4me3 boundaries; that is the part most directly attached to this citation. (LLM, conf=0.84)

      Experimental — accuracy not yet measured
    • cites [46] Heintzman, N. D. et al. Distinct and predictive chromatin signatures of transcriptional promoters and enhancers in the human genome. Nat. Genet. 39, 311–318 (2007). source
  • Findingown data

    Hepatocytic markers, HCC signature genes, and cancer-associated gene families (oncogenes, transcription factors, and protein kinases) from GSEA were overrepresented in MP genes.

    “Notably, hepatocytic markers28, HCC signature genes29, and cancer-associated gene families (oncogenes, transcription factors, and protein kinases) from GSEA30 were overrepresented in MP genes (Fig. 1g).”
    Quote in the paperSubstantial claim3 of 3 sources checked
    • cites [28] Aran, D., Hu, Z. & Butte, A. J. xCell: digitally portraying the tissue cellular heterogeneity landscape. Genome Biol. 18, 220 (2017). source

      The Results section describes using single-sample GSEA to score cell-type signatures, and the Methods section mentions filtering genes overexpressed in cancer cell lines. Neither passage addresses whether oncogenes, transcription factors, or protein kinases were overrepresented in MP genes; searches for those families and MP genes found no supporting passage in our copy. (LLM, conf=0.76)

      Experimental — accuracy not yet measured
    • cites [29] Hoshida, Y. et al. Gene expression in fixed tissues and outcome in hepatocellular carcinoma. N. Engl. J. Med. 359, 1995–2004 (2008). source
    • cites [30] Subramanian, A. et al. Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles. Proc. Natl Acad. Sci. USA 102, 15545–15550 (2005). source
  • Classified as probably not a substantial claim. 33
  • May combine two claims. 58

  • Classified as probably not a substantial claim. May combine two claims.

    Premisecitation

    Human promoters are mostly CG-rich and have distinct promoter architecture and regulation compared to CG-poor promoters which are mostly tissue-specific.

    “Human promoters are mostly CG-rich35,62 and have distinct promoter architecture and regulation11 compared to CG-poor promoters which are mostly tissue-specific17”
    Quote in the paperSubstantial claim2 of 4 sources checked
    • cites [35] Saxonov, S., Berg, P. & Brutlag, D. L. A genome-wide analysis of CpG dinucleotides in the human genome distinguishes two distinct classes of promoters. Proc. Natl Acad. Sci. USA 103, 1412–1417 (2006). source

      The Results (Fig. 1) classify 72% of human promoters as high-CpG and describe a different CpG pattern around the transcription start site in the two classes. The Results (Fig. 3) find that expression in only a few tissues is more common in the low-CpG class, but do not establish that most low-CpG promoters are tissue-specific. The Discussion says evidence about methylation-mediated regulation of low-CpG promoters is limited, so distinct regulation is not established here. (LLM, conf=0.88)

      Experimental — accuracy not yet measured
    • cites [62] Larsen, F., Gundersen, G., Lopez, R. & Prydz, H. CpG islands as gene markers in the human genome. Genomics 13, 1095–1107 (1992). source
    • cites [11] Carninci, P. et al. Genome-wide analysis of mammalian promoter architecture and evolution. Nat. Genet. 38, 626–635 (2006). source
    • cites [17] Consortium, F. et al. A promoter-level mammalian expression atlas. Nature 507, 462–470 (2014). source

      Citation 11 is attached to the claim that CG-rich promoters have distinct architecture and regulation from CG-poor promoters. In “Features of cell-type-specific promoters” (page unavailable), the paper instead says that cell-type-specific CpG-island and non-CpG-island promoters use the same type of signals; the cited passages do not establish the claimed architectural distinction. (LLM, conf=0.82)

      Checked by a reasoning model — may be wrong.

      Experimental — accuracy not yet measured

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

    Premisecitation

    Many genes utilize alternative promoters for specific processes such as cell fate transitions in yeast and mammalian cells, vertebrate embryogenesis, and may have important roles in cancer.

    “In addition to reference promoters, many genes utilize alternative promoters for specific processes, such as cell fate transitions in yeast16 and mammalian cells17, in vertebrate embryogenesis12,18, and may have important roles in cancer19.”
    Quote in the paperNot a substantial claim4 of 4 sources checked
    • cites [16] Chia, M. et al. High-resolution analysis of cell-state transitions in yeast suggests widespread transcriptional tuning by alternative starts. Genome Biol. 22, 34 (2021). source
    • cites [17] Consortium, F. et al. A promoter-level mammalian expression atlas. Nature 507, 462–470 (2014). source

      The Abstract reports that most mammalian genes have more than one transcription start site and that promoters have cell-type-specific expression profiles. The Conclusion connects promoter-based analysis to identifying drivers of reprogramming between cell lineages. These passages support a link between promoter usage and mammalian cell states, but do not show that alternative promoters are utilized during cell-fate transitions. (LLM, conf=0.78)

      Experimental — accuracy not yet measured
    • cites [12] Nepal, C. et al. Dynamic regulation of the transcription initiation landscape at single nucleotide resolution during vertebrate embryogenesis. Genome Res. 23, 1938–1950 (2013). source
    • cites [18] Baranasic, D. et al. Multiomic atlas with functional stratification and developmental dynamics of zebrafish cis-regulatory elements. Nat. Genet. 54, 1037–1050 (2022). source

  • Classified as probably not a substantial claim. May combine two claims.

    Premisecitation

    CAGE-seq detects differently regulated transcription initiation events within the same core promoter and can accurately detect alternative promoters.

    “CAGE-seq also detects differently regulated transcription initiation events within the same core promoter14,15, and thus, it can accurately detect alternative promoters.”
    Quote in the paperSubstantial claim2 of 2 sources checked
    • cites [14] Haberle, V. et al. Two independent transcription initiation codes overlap on vertebrate core promoters. Nature 507, 381–385 (2014). source
    • cites [15] Nepal, C. et al. Dual-initiation promoters with intertwined canonical and TCT/TOP transcription start sites diversify transcript processing. Nat. Commun. 11, 168 (2020). source

      The Introduction and Results (“Differential regulation of YC and YR-initiation”; pages unavailable) report that CAGE reveals two initiation types within the same core promoter that can be independently regulated. The paper does not establish the further claim that CAGE-seq accurately detects alternative promoters as distinct from alternative initiation sites within one promoter. (LLM, conf=0.84)

      Experimental — accuracy not yet measured

  • Classified as probably not a substantial claim. May combine two claims.

    Premisecitation

    CAGE-seq determines the 5'-ends of TSSs at single nucleotide resolution and quantifies gene expression similar to RNA-seq.

    “Cap Analysis of Gene Expression Sequencing (CAGE-seq) determines the 5’-ends of TSSs at single nucleotide resolution11,12 and quantifies gene expression similar to RNA-seq13.”
    Quote in the paperSubstantial claim2 of 3 sources checked
    • cites [11] Carninci, P. et al. Genome-wide analysis of mammalian promoter architecture and evolution. Nat. Genet. 38, 626–635 (2006). source
    • cites [12] Nepal, C. et al. Dynamic regulation of the transcription initiation landscape at single nucleotide resolution during vertebrate embryogenesis. Genome Res. 23, 1938–1950 (2013). source
    • cites [13] Kawaji, H. et al. Comparison of CAGE and RNA-seq transcriptome profiling using clonally amplified and single-molecule next-generation sequencing. Genome Res. 24, 708–717 (2014). source

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

    Premisecitation

    CpGs in CG-poor regions are generally methylated, while CpGs within CGIs remain unmethylated in normal cells and aberrantly hypermethylated in cancer.

    “The CpGs in CG-poor regions are generally methylated, while CpGs within CGIs remain unmethylated in normal cells49 and aberrantly hypermethylated in cancer2,50.”
    Quote in the paperNot a substantial claim3 of 3 sources checked
    • cites [49] Hughes, A. L., Kelley, J. R. & Klose, R. J. Understanding the interplay between CpG island-associated gene promoters and H3K4 methylation. Biochim. Biophys. Acta Gene Regul. Mech. 1863, 194567 (2020). source
    • cites [2] Cancer Genome Atlas Research Network. Comprehensive and integrative genomic characterization of hepatocellular carcinoma. Cell 169, 1327–1341 e23 (2017). Electronic address, w.b.e. & Cancer Genome Atlas Research, N. source

      The Results (Methylation Profiling; page unavailable) report cancer-specific CpG hypermethylation in hepatocellular carcinoma compared with normal tissue. They do not establish that the hypermethylated CpGs are within CpG islands or that this pattern applies to cancer generally. The earlier claims about CG-poor regions and normal cells are not attached to this citation. (LLM, conf=0.86)

      Experimental — accuracy not yet measured
    • cites [50] Taniguchi, I., Iwaya, C., Ohnaka, K., Shibata, H. & Yamamoto, K. Genome-wide DNA methylation analysis reveals hypomethylation in the low-CpG promoter regions in lymphoblastoid cell lines. Hum. Genomics 11, 8 (2017). source

      The Results section ('The methylation sites located in low CpG promoters could be sensitive to demethylation in LCLs'; page unavailable) reports that nearly half of low-CpG-promoter sites were highly methylated, while almost all high-CpG-promoter sites were hypomethylated in peripheral blood cells. This supports the claimed contrast, but the paper does not establish it for CG-poor regions and CpG islands generally across normal cells. The cancer clause is attributed to other citations, not this source. (LLM, conf=0.82)

      Experimental — accuracy not yet measured

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

    Premisecitation

    Two-thirds of human genes have CGIs within their promoters.

    “Two-thirds of human genes have CGIs within their promoters35”
    Quote in the paperNot a substantial claim1 of 1 sources checked
    • cites [35] Saxonov, S., Berg, P. & Brutlag, D. L. A genome-wide analysis of CpG dinucleotides in the human genome distinguishes two distinct classes of promoters. Proc. Natl Acad. Sci. USA 103, 1412–1417 (2006). source

      The Abstract (page unavailable) reports that 72% of human promoters have high CpG content, broadly consistent with the claimed proportion. But the paper deliberately avoids defining CpG islands by a threshold, so it does not establish that two-thirds of genes have CGIs within their promoters. (LLM, conf=0.86)

      Experimental — accuracy not yet measured

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

    Premisecitation

    Transcription of alternative promoters is widespread in cancer.

    “Transcription of alternative promoters is widespread in cancer19.”
    Quote in the paperNot a substantial claim1 of 1 sources checked
    • cites [19] Demircioglu, D. et al. A Pan-cancer transcriptome analysis reveals pervasive regulation through alternative promoters. Cell 178, 1465–1477 e17 (2019). source

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

    Premisecitation

    Alternative promoters have dynamic intragenic DNA methylation across human tissues.

    “Alternative promoters have dynamic intragenic DNA methylation across human tissues20”
    Quote in the paperNot a substantial claim1 of 1 sources checked
    • cites [20] Maunakea, A. K. et al. Conserved role of intragenic DNA methylation in regulating alternative promoters. Nature 466, 253–257 (2010). source

      The SHANK3 analysis (section labeled Abstract; page unavailable) reports that intragenic alternative promoters have tissue-specific DNA methylation and that their methylation and expression patterns are similar in matched human and mouse tissues. The paper supports variation across human tissues at this locus, but its broader tissue comparisons rely substantially on mouse data; it does not establish the claim generally for alternative promoters across human tissues. (LLM, conf=0.82)

      Experimental — accuracy not yet measured

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

    Premisecitation

    Thousands of genes have CGIs in their gene bodies that can act as novel alternative promoters during development.

    “thousands of genes have CGIs in their gene bodies (intragenic CGIs) that can act as novel alternative promoters during development36.”
    Quote in the paperNot a substantial claim1 of 1 sources checked
    • cites [36] Illingworth, R. S. et al. Orphan CpG islands identify numerous conserved promoters in the mammalian genome. PLoS Genet. 6, e1001134 (2010). source

      The Discussion (‘CGI methylation is conserved even at sites distal to annotated gene promoters’; page unavailable) proposes that many intragenic orphan CGIs may serve as alternative promoters with spatially or temporally restricted use. The Concluding remarks describe thousands of orphan CGIs as functional promoters, but do not establish that thousands of distinct genes have intragenic CGIs acting as alternative promoters. (LLM, conf=0.86)

      Experimental — accuracy not yet measured

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

    Premisecitation

    Loss of DNMT3B in mouse embryonic stem cells results in spurious initiation of alternative promoters.

    “loss of DNA methyltransferase 3B (DNMT3B) in mouse embryonic stem (ES) cells was shown to result in spurious initiation of alternative promoters21.”
    Quote in the paperNot a substantial claim1 of 1 sources checked
    • cites [21] Neri, F. et al. Intragenic DNA methylation prevents spurious transcription initiation. Nature 543, 72–77 (2017). source

      The Analysis of intragenic spurious transcription section reports increased cryptic initiation within gene bodies in Dnmt3b-knockout embryonic stem cells. It also says some alternative promoters were reactivated, but found no general reactivation; Extended Data Figure 3 identifies only 4 of 2,563 genes with such reactivation. The claim conflates the broader spurious-initiation finding with the much narrower alternative-promoter finding. (LLM, conf=0.87)

      Experimental — accuracy not yet measured

  • May combine two claims.

    Interpretationcitation

    H2A.Z marks are mutually antagonistic with DNA methylation, which together with H2A.Z enrichment on CGI promoters provides an explanation for the depletion of methylation on CGI promoters.

    “H2A.Z marks are mutually antagonistic with DNA methylation51, which, together with an observed H2A.Z enrichment on CGI promoters (Fig. 4d) provides an explanation for the depletion of methylation on CGI promoters.”
    Quote in the paperNot a substantial claim1 of 1 sources checked
    • cites [51] Zilberman, D., Coleman-Derr, D., Ballinger, T. & Henikoff, S. Histone H2A.Z and DNA methylation are mutually antagonistic chromatin marks. Nature 456, 125–129 (2008). source

  • Classified as probably not a substantial claim.

    Study design and conductcitation

    We only retained those that have YR-initiation or YC-initiation initiation motif.

    “we only retained those that have YR-initiation or YC-initiation initiation motif15”
    Quote in the paperSubstantial claim1 of 1 sources checked
    • cites [15] Nepal, C. et al. Dual-initiation promoters with intertwined canonical and TCT/TOP transcription start sites diversify transcript processing. Nat. Commun. 11, 168 (2020). source

  • Classified as probably not a substantial claim.

    Study design and conductcitation

    We filtered CAGE TCs overlapping small RNAs, drosha processing site of pre-miRNAs, and 5′ ends capping of snoRNAs.

    “We filtered CAGE TCs overlapping small RNAs, drosha processing site of pre-miRNAs23, and 5′ ends capping of snoRNAs17”
    Quote in the paperSubstantial claim2 of 2 sources checked
    • cites [23] Nepal, C. et al. Transcriptional, post-transcriptional and chromatin-associated regulation of pri-miRNAs, pre-miRNAs and moRNAs. Nucleic Acids Res. 44, 3070–3081 (2016). source
    • cites [17] Consortium, F. et al. A promoter-level mammalian expression atlas. Nature 507, 462–470 (2014). source

      The ‘CAGE peak identification and thresholding’ section describes clustering CAGE tags and applying evidence thresholds to enrich for transcription start sites. It does not say that CAGE tag clusters were filtered for overlap with small RNAs, Drosha processing sites of pre-miRNAs, or capped 5′ ends of snoRNAs; searches for those terms found no supporting passage in our copy. (LLM, conf=0.72)

      Experimental — accuracy not yet measured

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

    Interpretationcitation

    Most alternative promoters might be co-expressed in the same cell.

    “which suggests that most alternative promoters might be co-expressed in the same cell”
    Quote in the paperNot a substantial claim1 of 1 sources checked
    • cites [63] Karlsson, K., Lonnerberg, P. & Linnarsson, S. Alternative TSSs are co-regulated in single cells in the mouse brain. Mol. Syst. Biol. 13, 930 (2017). source

      The Discussion (page unavailable) reports that alternative TSSs were almost always co-expressed in single cells, supporting the general idea. But it distinguishes alternative promoters—TSSs more than 1 kb apart—and says few genes expressed them; the Results (page unavailable) identify only six potential alternative-promoter pairs in the analyzed cells. The paper therefore does not establish that most alternative promoters are co-expressed. (LLM, conf=0.84)

      Experimental — accuracy not yet measured

  • Classified as probably not a substantial claim.

    Study design and conductcitation

    Differentially expressed genes and promoters were identified using DE-seq2 with a significance cut-off at an adjusted P-value of 0.05 and log2 fold-change of absolute (0.5).

    “Differentially expressed genes and promoters were identified using DE-seq232. The significance cut-off was defined at an adjusted P-value of 0.05 and log2 fold-change of absolute (0.5)”
    Quote in the paperSubstantial claim1 of 1 sources checked
    • cites [32] Love, M. I., Huber, W. & Anders, S. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol. 15, 550 (2014). source

  • Classified as probably not a substantial claim. May combine two claims.

    Premisecitation

    A highly expressed promoter is often annotated as reference promoter, which is what we used in this study.

    “a highly expressed promoter is often annotated as the reference promoter16,19, which is what we used in this study”
    Quote in the paperSubstantial claim2 of 2 sources checked
    • cites [16] Chia, M. et al. High-resolution analysis of cell-state transitions in yeast suggests widespread transcriptional tuning by alternative starts. Genome Biol. 22, 34 (2021). source
    • cites [19] Demircioglu, D. et al. A Pan-cancer transcriptome analysis reveals pervasive regulation through alternative promoters. Cell 178, 1465–1477 e17 (2019). source

      The retrieved passages have no section or page information and do not state that a highly expressed promoter is often annotated as the reference promoter. Searches for “reference promoter,” “highly expressed promoter,” and related terms returned URLs or garbled text, so our copy cannot verify the attribution. (LLM, conf=0.70)

      Experimental — accuracy not yet measured

  • Classified as probably not a substantial claim.

    Study design and conductcitation

    We filtered intragenic CAGE TCs with GG-initiation and those overlapping with coding exons and annotated enhancers.

    “We filtered intragenic CAGE TCs with GG-initiation and those overlapping with coding exons and annotated enhancers24,70”
    Quote in the paperSubstantial claim2 of 2 sources checked
    • cites [24] Andersson, R. et al. An atlas of active enhancers across human cell types and tissues. Nature 507, 455–461 (2014). source
    • cites [70] Gao, T. & Qian, J. EnhancerAtlas 2.0: an updated resource with enhancer annotation in 586 tissue/cell types across nine species. Nucleic Acids Res. 48, D58–D64 (2020). source

      The Materials and Methods sections describe filtering Rerep-Seq data and analyzing replication-timing regions, not intragenic CAGE transcription clusters. Searches for GG-initiation, CAGE clusters, coding exons, and enhancer filtering found no passage addressing the cited claim; page numbers are unavailable in our copy. (LLM, conf=0.76)

      Experimental — accuracy not yet measured

  • May combine two claims.

    Findingown data

    Intragenic CGIs have low levels of DNA demethylase (5hmC) activity, consistent with low levels of TET1 binding.

    “Intragenic CGIs have low levels of DNA demethylase (5hmC levels) activity (Fig. 5e) measured by TET-assisted reduced RRBS (TAB-RRBS)6, which is consistent with low levels of TET1 binding”
    Quote in the paperSubstantial claim1 of 1 sources checked
    • cites [6] Hlady, R. A. et al. Integrating the epigenome to identify drivers of hepatocellular carcinoma. Hepatology 69, 639–652 (2019). source
    • Rests on an assumption: DNMT3B and TET1 binding data from human embryonic stem cells are representative of the corresponding binding in HCC tumors.

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

    Premisecitation

    Aged cells (and cancer cells) generally lose DNA methylation from gene bodies.

    “It is known that aged cells (and cancer cells) generally lose DNA methylation from gene bodies59”
    Quote in the paperNot a substantial claim1 of 1 sources checked
    • cites [59] Zhou, W. et al. DNA methylation loss in late-replicating domains is linked to mitotic cell division. Nat. Genet. 50, 591–602 (2018). source

      The Abstract reports methylation loss in cancer and increasing hypomethylation with age, but locates the loss primarily in late-replicating, gene-poor partially methylated domains—not generally in gene bodies. The Results section on replication timing and H3K36me3 reports that CpGs marked by H3K36me3, a mark associated with active gene bodies, remained highly methylated. (LLM, conf=0.88)

      Experimental — accuracy not yet measured

  • May combine two claims.

    Findingcitation

    MP genes were significantly associated with OS, similar to recent observations in multiple myeloma.

    “MP genes were significantly associated with OS, similar to recent observations in multiple myeloma66”
    Quote in the paperNot a substantial claim1 of 1 sources checked
    • cites [66] Valcarcel, L. V. et al. Gene expression derived from alternative promoters improves prognostic stratification in multiple myeloma. Leukemia 35, 3012–3016 (2021). source

      The Results and discussion report that expression from alternative active promoters was significantly associated with overall survival in multiple myeloma. They do not establish that MP genes, as genes, were significantly associated with overall survival; the paper distinguishes promoter activity from overall gene expression and notes that SLAMF7 gene expression was not associated with survival. (LLM, conf=0.84)

      Experimental — accuracy not yet measured

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

    Premisecitation

    Old yeast cells have decreased H3K36me3 levels and increased intragenic transcripts, while DNA methylation decreases with age in humans.

    “Old yeast cells have decreased H3K36me3 levels and increased intragenic transcripts58, while DNA methylation decreases with age in humans59.”
    Quote in the paperNot a substantial claim2 of 2 sources checked
    • cites [58] Sen, P. et al. H3K36 methylation promotes longevity by enhancing transcriptional fidelity. Genes Dev. 29, 1362–1376 (2015). source
    • cites [59] Zhou, W. et al. DNA methylation loss in late-replicating domains is linked to mitotic cell division. Nat. Genet. 50, 591–602 (2018). source

  • May combine two claims.

    Premisecitation

    SETD2 deposits H3K36me3 after Pol2 passage, which in turn recruits DNMT3B to maintain a repressive chromatin state.

    “The methyltransferase SETD2 deposits H3K36me355 after Pol2 passage56, which in turn recruits DNMT3B21 to maintain a repressive chromatin state57.”
    Quote in the paperSubstantial claim2 of 3 sources checked
    • cites [55] Barski, A. et al. High-resolution profiling of histone methylations in the human genome. Cell 129, 823–837 (2007). source
    • cites [56] Edmunds, J. W., Mahadevan, L. C. & Clayton, A. L. Dynamic histone H3 methylation during gene induction: HYPB/Setd2 mediates all H3K36 trimethylation. EMBO J. 27, 406–420 (2008). source

      The Results section ('Setd2 is non-redundantly responsible for H3K36 trimethylation in mouse fibroblasts'; page unavailable) identifies Setd2 as necessary for virtually all H3K36 trimethylation. The Discussion ('Setd2 mediates H3K36 trimethylation at the 3′ end of active gene coding regions'; page unavailable) suggests that SETD2 may travel with Pol II and methylate K36 co-transcriptionally, but does not establish deposition specifically after Pol II passage. The DNMT3B clause has separate citations in the printed sentence and is not attributed to this source. (LLM, conf=0.85)

      Experimental — accuracy not yet measured
    • cites [21] Neri, F. et al. Intragenic DNA methylation prevents spurious transcription initiation. Nature 543, 72–77 (2017). source

  • May combine two claims.

    Interpretationreasoning

    Intragenic CGIs are less favorable to act as alternative promoters in cancer, though they can act as alternative promoters during embryonic development and cell differentiation.

    “Thus, intragenic CGIs are less favorable to act as alternative promoters in cancer, though they can act as alternative promoters during embryonic development and cell differentiation36”
    Quote in the paperSubstantial claim1 of 1 sources checked
    • cites [36] Illingworth, R. S. et al. Orphan CpG islands identify numerous conserved promoters in the mammalian genome. PLoS Genet. 6, e1001134 (2010). source

      The Discussion (‘CGI methylation is conserved even at sites distal to annotated gene promoters’; page unavailable) proposes that many intragenic orphan CGIs may be alternative promoters used in a spatially or temporally restricted fashion. But the paper does not establish their use specifically during cell differentiation, and its colorectal-cancer analysis measures CGI methylation, not whether intragenic CGIs are less likely to act as alternative promoters in cancer. (LLM, conf=0.84)

      Experimental — accuracy not yet measured

  • May combine two claims.

    Findingown data

    Regions with higher CpG density were more hypermethylated, while CpGs in nonCGI promoters were more hypomethylated.

    “Regions with higher CpG density were more hypermethylated, while CpGs in nonCGI promoters were more hypomethylated”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    Promoters for 15419 expressed genes and alternative TSSs for 3052 annotated alternative transcripts were identified.

    “Based on GENCODE transcript models, we identified promoters for 15419 expressed genes and alternative TSSs for 3052 annotated alternative transcripts.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    65% of alternative promoters are located downstream of the reference promoter and 35% are upstream.

    “Overall, 65% of alternative promoters are located downstream of the reference promoter (Fig. 1g), and 35% are upstream”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    Reference promoters were constitutively expressed in all samples, while alternative promoters were expressed either constitutively or only in a subset of patients.

    “The reference promoters (irrespective of SP or MP genes) were constitutively expressed in all samples, while alternative promoters were expressed either constitutively or only in a subset of patients”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    1489 (36.4%) alternative promoters' expression was higher than reference promoters in one or more tumor-adjacent samples, increasing to 1716 (42%) among tumors.

    “In total, 1489 (36.4%) alternative promoters’ expression was higher than reference promoters in one or more tumor-adjacent samples, which further increased to 1716 (42%) alternative promoters among tumors”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    Hundreds of differentially hypo- and hypermethylated promoters were identified and significantly overlapped with differentially expressed promoters.

    “We next analyzed CG probes (±500 bases around TSSs) and identified hundreds of differentially hypo- and hypermethylated promoters (Fig. 5a, b), which significantly overlapped with the differentially expressed promoters”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    Alternative promoters overlapping CGIs had a lower observed to expected ratio of CG dinucleotides and shorter CGIs lengths.

    “Alternative promoters overlapping CGIs had a lower observed to expected (O/E) ratio of CG dinucleotides (Supplementary Fig. 3c) and shorter CGIs lengths”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    The 15,419 expressed genes were classified into 12,493 (80.3%) single-promoter and 2926 (19.7%) multi-promoter genes.

    “Based on the number of promoters, we classified the 15,419 expressed genes into either single-promoter (SP) (12,493; 80.3%) or MP (2926; 19.7%) genes”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    Two-thirds of alternative promoters utilize a different N terminus or were located downstream of the N terminus of the protein.

    “Two-thirds of the alternative promoters were shown to utilize a different N terminus or were located downstream of the N terminus of the protein”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    Both reference and alternative promoters have a well-defined transcription initiation motif, while the TATA box was positionally constrained upstream of sharp promoters.

    “Both promoters have a well-defined transcription initiation motif, while the TATA box was positionally constrained upstream of sharp promoters”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    Pol2 was enriched at similar levels in both SP and MP genes while Pol2-Ser5 and Pol2-Ser2 were higher among MP genes.

    “RNA polymerase II (Pol2) was enriched at similar levels in both SP and MP genes while initiating Pol2 modified by phosphorylation of serine 5 (Pol2-Ser5) and elongating Pol2 modified by phosphorylation of serine 2 (Pol2-Ser2) on its carboxy-terminal domain (CTD) were higher among MP genes”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    In alternative CGI promoters, gene body histone marks presented NDR, while histones were continuous in nonCGI alternative promoters.

    “In case of alternative CGI promoters, the gene body histone (H3K79me2, H3K36me3, and H4K20me1) marks presented NDR, while histones were continuous in nonCGI alternative promoters”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    Divergent H3K27ac peaks were enriched on both CG-rich and CG-poor promoters, while they lacked H3K27me3 as expected in active genes.

    “Notably, divergent H3K27ac peaks were enriched on both CG-rich and CG-poor promoters, while they lacked H3K27me3 as expected in active genes”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    The majority of differentially downregulated promoters are MP genes, while SP genes were upregulated.

    “The majority of differentially downregulated promoters are MP genes, while SP genes were upregulated”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    Compared to nonCGI reference promoters, nonCGI alternative promoters have low levels of H3K4me3 and lower CG density.

    “Compared to nonCGI reference promoters, nonCGI alternative promoters have low levels of H3K4me3 (Fig. 4a) and lower CG density”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    Reference and alternative nonCGI promoters had low levels of H3K4me3 asymmetrically deposited downstream of TSSs, while H3K4me1 marks were non-divergent and enriched only at TSSs.

    “On the other hand, reference and alternative nonCGI promoters had low levels of H3K4me3 asymmetrically deposited downstream of TSSs, while H3K4me1 marks were non-divergent and enriched only at TSSs.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    Aberrant hypermethylation of CGI promoters was not globally observed among expressed genes, while it was prevalent among CGI promoters of unexpressed genes.

    “Aberrant hypermethylation of CGI promoters was not globally observed among expressed genes, while it was prevalent among CGI promoters of unexpressed genes”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    DNA methylation at TSSs decreased based on increasing CG density in CGI and nonCGI promoters, more evident among nonCGI promoters.

    “DNA methylation at TSSs decreased based on increasing CG density in CGI and nonCGI promoters (Supplementary Fig. 5a), which was more evident among nonCGI promoters.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    CGI promoters were enriched for H2A.Z, have a well-defined NDR, and phased +1 downstream nucleosome, whereas nonCGI promoters have undefined NDR.

    “Furthermore, CGI promoters were enriched for H2A.Z, have a well-defined NDR, and (phased +1) downstream the nucleosome, whereas nonCGI promotes have undefined NDR”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

    Findingnone given

    Alternative promoters are widespread in cancer.

    “alternative promoters are widespread in cancer”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S4 (based).

    Findingnone given

    CG-rich regions are marked by higher levels of H3K36me3 (via SETD2).

    “Since CG-rich regions are marked by higher levels of H3K36me3 (via SETD2)”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    Alternative promoters detected only in tumors were enriched for genes in signaling pathways, which globally did not downregulate reference promoters.

    “alternative promoters detected only in tumors were enriched for genes in signaling pathways (Fig. 2c), which globally did not downregulate reference promoters (Supplementary Fig. 2f)”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    Alternative promoters can downregulate reference promoters through transcriptional interference, which often coincides with important genes, including hepatocyte-specific markers.

    “alternative promoters can downregulate reference promoters through transcriptional interference, which often coincides with important genes, including hepatocyte-specific markers (Supplementary Fig. 2f)”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingnone given

    Few alternative promoters have expression levels similar to their reference promoters, and in these cases, the annotation of reference and alternative promoters is interchangeable across cohorts/studies.

    “Few alternative promoters have expression levels similar to their reference promoters, and in these cases, the annotation of reference and alternative promoters is interchangeable across cohorts/studies”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingreasoning

    These intragenic CG-rich regions have low levels of TET1 binding and, thus, have reduced active demethylation.

    “these intragenic CG-rich regions have low levels of TET1 binding and, thus, have reduced active demethylation”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    High levels of DNA methylation on intragenic CGIs remained globally unchanged in HCC tumors while flanking regions had decreased methylation.

    “High levels of DNA methylation on intragenic CGIs remained globally unchanged in HCC tumors (Fig. 5e) while flanking regions had decreased methylation.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Interpretationown data

    Alternative promoters can control tumor transcriptomes and their regulatory architecture.

    “We show alternative promoters can control tumor transcriptomes and their regulatory architecture.”
    Quote in the paperSubstantial claim

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

    Study design and conductown data

    These findings were validated using reduced representation bisulfite sequencing of HepG2 cells.

    “These findings were further validated using reduced representation bisulfite sequencing (RRBS) of HepG2 cells”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Interpretationown data

    Transcription of many alternative promoters in tumors affects key cancer-related pathways and leads to downregulation of the reference promoter.

    “Collectively, this highlights that the transcription of many alternative promoters in tumors affects key cancer-related pathways and leads to the downregulation of the reference promoter.”
    Quote in the paperSubstantial claim

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

    Study design and conductown data

    These data were validated by GRO-seq.

    “These data were validated by GRO-seq”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Interpretationown data

    Intragenic CGIs lack enrichment of H3K4me3, H3K4me1, and H3K27ac, explaining the absence of transcription.

    “we analyzed H3K4me3, H3K4me1, and H3K27ac modifications and observed no enrichment (Fig. 4c), thus explaining the absence of transcription.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Interpretationown data

    Reference and alternative promoters have distinct chromatin architecture influenced by genomic CG density, revealing that two distinct chromatin architectures coexist in MP genes.

    “In conclusion, reference and alternative promoters have distinct chromatin architecture that is influenced by genomic CG density, thus revealing that two distinct chromatin architectures coexist in MP genes.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Interpretationown data

    Alternative promoters and their usage are important in hepatocarcinogenesis and for patient outcomes.

    “Collectively, these data support that alternative promoters and their usage are important in hepatocarcinogenesis and for patient outcomes.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Interpretationown data

    Alternative promoters have distinct regulatory architecture characterized by low CG content, sharp promoter shape, and enrichment for liver tissue-specific transcription factors.

    “Thus, alternative promoters have distinct regulatory architecture characterized by low CG content, sharp promoter shape, and enriched for liver tissue-specific transcription factors.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Interpretationreasoning

    Lack of active demethylation on intragenic CGIs may explain its sustained high level of methylation to maintain the repressive state, resulting in lack of activation of alternative promoters from intragenic CGIs.

    “Thus, the lack of active demethylation on intragenic CGIs may explain its sustained high level of methylation to maintain the repressive state, resulting in lack of activation of alternative promoters from intragenic CGIs”
    Quote in the paperSubstantial claim
    • Rests on an assumption: DNMT3B and TET1 binding data from human embryonic stem cells are representative of the corresponding binding in HCC tumors.

  • May combine two claims.

    Interpretationreasoning

    This leads to intragenic CG-rich regions with more repressive chromatin marks, and that would require higher levels of DNA demethylating enzyme to make their chromatin accessible.

    “This leads to intragenic CG-rich regions with more repressive chromatin marks, and that would require higher levels of DNA demethylating enzyme to make their chromatin accessible”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Interpretationreasoning

    SETD2 mutations accelerate the epigenetic control, compromising the maintenance of DNA methylation and resulting in activation of alternative promoters.

    “as SETD2 mutations accelerate the epigenetic control, compromising the maintenance of DNA methylation and resulting in activation of alternative promoters”
    Quote in the paperSubstantial claim

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

    Study design and conductnone given

    The raw sequence reads were mapped using bowtie2 and excluded multi-mapping reads.

    “The raw sequence reads were mapped using bowtie267 and excluded multi-mapping reads”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Interpretationown data

    HCC nonCGI alternative promoters drive tissue-specific or cell type-specific transcription and are enriched for liver-specific transcription factors.

    “indicating they drive tissue-specific or cell type-specific transcription and are enriched for liver-specific transcription factors”
    Quote in the paperSubstantial claim

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

    Study design and conductown data

    We intersected ChromHMM HepG2 enhancers with other enhancers and ensured that enhancers were annotated across multiple datasets.

    “We intersected ChromHMM HepG2 enhancers with other enhancers and ensured that enhancers were annotated across multiple datasets”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Interpretationown data

    Compared to normal hepatocytes (HepaRG), HepG2 cells had lower H3K36me3 marks, suggesting a loss of H3K36me3 in tumor cells.

    “Compared to normal hepatocytes (HepaRG), HepG2 cells had lower H3K36me3 marks (Supplementary Fig. 6a, “Methods”), suggesting a loss of H3K36me3 in tumor cells.”
    Quote in the paperSubstantial claim
    • Rests on an assumption: HepG2 cells are a valid model for the epigenetic state of HCC tumor cells.

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

    Interpretationreasoning

    We propose that other tumors have similar mechanisms in activating alternative promoters, while their preferred location will be dependent upon DNA demethylated sites that are enriched for tissue-specific transcription factors of the specific tumor tissue.

    “We propose that other tumors have similar mechanisms in activating alternative promoters, while their preferred location will be dependent upon DNA demethylated sites that are enriched for tissue-specific transcription factors of the specific tumor tissue”
    Quote in the paperNot a substantial claim

  • Classified as probably not a substantial claim.

    Study design and conductnone given

    We excluded transcripts that are proximal (within 300 bases apart) to 5′ ends of annotated Ensembl transcripts.

    “We excluded transcripts that are proximal (within 300 bases apart) to 5′ ends of annotated Ensembl transcripts”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Interpretationreasoning

    Signaling pathway genes co-opt both promoters, where tumor cells utilize signaling pathways differently via alternative promoter usage (for example in CDKN2A and ERBB2).

    “This suggests that signaling pathway genes co-opt both promoters, where tumor cells utilize signaling pathways differently via alternative promoter usage (for example in CDKN2A and ERBB2 (Supplementary Fig. 1e, f))”
    Quote in the paperSubstantial claim

  • Classified as probably not a substantial claim. May combine two claims.

    Study design and conductnone given

    The P-value was adjusted for multiple corrections and P-value less than 0.05 was defined as significantly methylated CpGs.

    “The P-value was adjusted for multiple corrections and P-value less than 0.05 was defined as significantly methylated CpGs”
    Quote in the paperSubstantial claim

  • Classified as probably not a substantial claim. May combine two claims.

    Study design and conductown data

    In this study, we analyzed tumor samples and ensured that the reference promoter was expressed in the normal liver; thus, tumor-specific promoters are annotated as alternatives.

    “we analyzed tumor samples and ensured that the reference promoter was expressed in the normal liver; thus, tumor-specific promoters are annotated as alternatives”
    Quote in the paperSubstantial claim

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

    Study design and conductnone given

    The background regions were controlled for nucleotide composition and selected by default by HOMER.

    “The background regions were controlled for nucleotide composition and selected by default by HOMER”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Interpretationreasoning

    Alternative promoters with CG-poor regions in tumor tissues are sensitive to demethylation, which may facilitate an open chromatin structure and transcription initiation.

    “In contrast, alternative promoters with CG-poor regions in tumor tissues are sensitive to demethylation, which in turn may facilitate an open chromatin structure and transcription initiation.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S1 (asserted), S3 (load-bearing). May combine two claims.

    Premisecitation

    The general view on why some genes have multiple promoters is that a cell needs more expression of specific genes, and transcription from multiple promoters may add flexibility manifested through different molecular mechanisms in regulating gene expression.

    “The general view on why some genes have multiple promoters is that a cell needs more expression of specific genes, and transcription from multiple promoters may add flexibility manifested through different molecular mechanisms in regulating gene expression65”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [65] Ayoubi, T. A. & Van De Ven, W. J. Regulation of gene expression by alternative promoters. FASEB J. 10, 453–460 (1996). source

  • Classified as probably not a substantial claim. May combine two claims.

    Study design and conductreasoning

    To ensure that consensus TCs have robust expression levels, we retained TCs only if their expression was higher than 1 TPM in at least 15 samples (15% of the cohort) and had a minimum of 3 TPM at least in 1 sample.

    “To ensure that consensus TCs have robust expression levels, we retained TCs only if their expression was higher than 1 TPM in at least 15 samples (15% of the cohort) and had a minimum of 3 TPM at least in 1 sample”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S2 (checkable).

    Interpretationreasoning

    Tumor cells might utilize signaling pathways differently in a globally changed cancer background.

    “tumor cells might utilize signaling pathways differently in a globally changed cancer background”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Interpretationown data

    The chromatin structure of intragenic CG-rich and CG-poor regions have different distributions of 5mC, 5hmC, H3K36me3, and DNMT3B, leading to the pervasive initiation of alternative promoters from CG-poor regions.

    “The chromatin structure of intragenic CG-rich and CG-poor regions have different distributions of 5mC, 5hmC, H3K36me3, and DNMT3B, leading to the pervasive initiation of alternative promoters from CG-poor regions.”
    Quote in the paperSubstantial claim

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

    Interpretationreasoning

    Cell type-specific transcription might contribute to tissue-specific alternative promoters.

    “cell type-specific transcription might contribute to tissue-specific alternative promoters”
    Quote in the paperNot a substantial claim
76 of 76

  • The paper relies on this without support.

    The tumor-adjacent liver tissues used as matched controls are representative of normal liver biology.

    “we analyzed CAGE-seq data from 50 tumors and 50 matched tumor-adjacent normal liver tissues8”

  • The paper relies on this without support.

    HepG2 cells are a valid model for the epigenetic state of HCC tumor cells.

    “Compared to normal hepatocytes (HepaRG), HepG2 cells had lower H3K36me3 marks (Supplementary Fig. 6a, “Methods”), suggesting a loss of H3K36me3 in tumor cells.”

  • The paper relies on this without support.

    DNMT3B and TET1 binding data from human embryonic stem cells are representative of the corresponding binding in HCC tumors.

    “Coverage of DNMT3B binding on promoter CGIs and intragenic CGIs across human ES cells.”

  • The paper relies on this without support.

    There is no consensus on which promoter should be assigned as reference or alternative.

    “Since there is no consensus on which promoter should be assigned as a reference or alternative”

  • The paper relies on this without support.

    Reference promoters are generally ubiquitous and represent the predominant degree of transcription.

    “While reference promoters are generally ubiquitous and represent the predominant degree of transcription”
5 of 5

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

Each contradiction between two passages listed here was confirmed by an independent check.

  • SETD2 comparison uses conflicting patient-group sizes

    Major

    The results and figure describe 15 SETD2-mutant and 362 wild-type tumors, whereas Methods specifies 16 mutants and 352 wild-type tumors for the DNA methylation comparison.

  • Novel nonCGI promoter count exceeds the total novel promoter count

    Major

    The figure caption lists 1,387 novel nonCGI promoters, although the paper reports only 1,031 novel alternative promoters in total.

  • Lower expression threshold is said to increase nondetection

    Minor

    With the same samples and promoters, lowering the minimum expression needed for detection cannot increase the fraction classified as undetected.

  • Fig. 6c is identified as two different measurements

    Minor

    The results identify Fig. 6c as DNMT3B coverage, but its caption says it shows H3K36me3 coverage.

  • Conflicting P values for the Fig. 1h survival association

    Minor

    The Fig. 1h caption and the results text give different Fisher's exact-test P values for the association of multi-promoter genes with survival.

5 of 5

Cites

Flagged first, then by citations

Cited by

Still being checked.

Related

Authors
Nepal, C., & Andersen, J. B.
Affiliations
University of Copenhagen (DK)
Corresponding author
  • Nepal, Chirag
  • Andersen, Jesper B.
ORCID
Venue
Nature Communications
Publisher
Springer Science and Business Media LLC
Published
2023-05-11
Volume
14
Issue
1
Pages
2712-2712
ISSN
2041-1723 (ISSN-L)
DOI
10.1038/s41467-023-38272-4
PMC
PMC10175279
OpenAlex
W4376225499
Semantic Scholar
f85676c0e2922cb0d6f8d61e14d148ba984c38ba
Cited by
39
References
79
Citations by year
2026: 11 · 2025: 10 · 2024: 13 · 2023: 5
FWCI
2.90
Type
journal article
Language
en
Open access
Licence
Locations
  • Nature Communications (publishedVersion, cc-by)
  • PubMed (publishedVersion)
  • Research at the University of Copenhagen (University of Copenhagen) (publishedVersion, cc-by)
  • PubMed Central (submittedVersion, cc-by)
  • RePEc: Research Papers in Economics (submittedVersion)
  • DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
  • Figshare (submittedVersion, cc-by)
  • Figshare (cc-by)
Funders
Sundhed og Sygdom, Det Frie Forskningsråd (6100-00557A)
Topics
  • Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › Epigenetics and DNA Methylation
  • Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › Genomics and Chromatin Dynamics
  • Life Sciences › Biochemistry, Genetics and Molecular Biology › Cancer Research › Cancer-related molecular mechanisms research
Keywords
alternative promoters, hepatocellular carcinoma, DNA methylation loss, SETD2, H3K36me3, DNMT3B, TET1, nucleosome depletion, transcriptional regulation, chromatin accessibility, tumor transcriptome, overall survival
MeSH terms
Chromatin, Humans, Mixed Function Oxygenases, Liver Neoplasms, Nucleosomes, Proto-Oncogene Proteins, CpG Islands, DNA Methylation, Epigenesis, Genetic, Transcriptome

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 2026-10-27.

Classified by jev-1.13-20260917 (gold-guideline-v0.2) on 2026-10-27.

Cited sources are checked by gpt-6-sol.

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