henhaobio
Sign in

bioRxivNeurosciencePreprint

H3K27 trimethylation maintains baseline network excitability in epilepsy

Jose Ezekiel Clemente EspinaiD, Olivia R. HoffmaniD, Jennifer L. KoehleriD, Barry A. Schoenike, Avtar RoopraiD

PDFDOI

In-depth summary

Claude Opus 5.5 · xhigh effort · 8 Nov 2026 · from the full text of version 2 (30 Jul 2026) · CC BY 4.0

In mice 4 days after kainate-induced status epilepticus, H3K27me3 gains coincide with repression of excitability genes; EZH2 inhibition with UNC1999 lowers seizure threshold acutely and increases chronic seizure frequency.

The authors' lab previously found that Enhancer of Zeste Homolog 2 (EZH2), the H3K27 methylase of Polycomb Repressive Complex 2, is induced after status epilepticus (SE), and that deleting it in neurons or inhibiting it systemically worsens later epilepsy; the mechanism was unknown. This work links SE-induced histone H3 lysine 27 trimethylation (H3K27me3) to repression of a neuronal excitability gene module and to the preservation of seizure threshold after injury.

Key results

  1. ShownStatus epilepticus mostly increased hippocampal H3K27me3. The gains were split almost evenly between sites already marked in naive tissue and newly methylated (de novo) sites, and the de novo sites tended to gain more signal.

    DiffBind (FDR < 0.05) found 3,442 differential H3K27me3 peaks at 4 days post-SE vs naive hippocampus: 2659 increased, 783 depleted. Of the increased peaks, 1335 overlapped naive consensus peaks (pre-marked) and 1324 did not (de novo). · Status epilepticus induces H3K27me3 enrichment…; CUT&RUN-seq analysis; Figure legends

  2. ShownAbout half of the genes that gained H3K27me3 did not change expression. The repressed subset was associated with de novo methylation and enriched for neuronal signalling and synaptic pathways, while undetected (likely silent) genes were associated with pre-marked loci.

    RNA-seq: 6225 differentially expressed genes (3482 up, 2743 down; padj < 0.05, 1.2-fold). Of 2022 genes with H3K27me3 gains, about 50% were unchanged, 287 (14%) undetected, 519 (25%) repressed and 168 (8%) induced. Distributions differed between pre-marked and de novo genes (chi-squared p=2.11e-36). Repressed targets yielded 8 neuronal ontology clusters. · H3K27me3 induction is associated with repression…; RNA-seq analysis; Ontology analysis; Figure legends

  3. SuggestedThe 241-gene excitability module is expressed mainly in mature neurons. Genes repressed after SE in dentate granule and pyramidal cells overlap it most, which points to excitatory neurons as the main site.

    Reanalysed hippocampal snRNA-seq (pilocarpine SE, naive vs 4 days; Hoffman et al.) with scanpy score_genes: module scores were higher in neurons than in non-neuronal cells and neural progenitor cells. In per-class pseudobulk DESeq2 (padj < 0.1), dentate granule and pyramidal cells showed the most significant overlap; the overlap statistics are not given in the text. · H3K27me3 induction is associated with repression…; snRNA-seq meta-analysis; Figure legends

  4. ShownSystemic EZH2 inhibition with UNC1999 after SE modestly reduced the H3K27me3 gain at the module's promoters and blunted the repression of these genes.

    UNC1999 (20 mg/kg) at 6, 24 and 48 h post-SE. H3K27me3 at module transcription start sites (±200 bp): SE + UNC1999 below SE + vehicle (p = 0.0253, 1.2-fold reduction; the figure legend calls it marginal). Module RNA z-scores: less repression with UNC1999 than with vehicle after SE (p = 2.14e-77). Kruskal-Wallis, then Benjamini-Hochberg-corrected Wilcoxon tests on gene-level distributions. · EZH2 inhibition post-SE increases network…; UNC1999 treatment; CUT&RUN-seq analysis; Figure legends

  5. ShownMice given UNC1999 after SE had a lower seizure threshold at 4 days and more spontaneous seizures at weeks 5-6, whereas vehicle mice kept their baseline threshold. The size of the acute threshold drop tracked chronic seizure frequency.

    Flurothyl time to generalized tonic-clonic seizure fell significantly from baseline to 4 days post-SE only with UNC1999 (repeated-measures 2-way ANOVA, Tukey); sham mice given UNC1999 were unchanged. Video-scored seizure frequency was higher with UNC1999 (Wilcoxon). Threshold change correlated negatively with frequency (Pearson). Group sizes, effect sizes and p-values are not given in the text. · EZH2 inhibition post-SE increases network…; Seizure threshold testing; Behavioral seizure monitoring; Figure legends

Methods, data and reproducibility

Hippocampi of naive mice and mice 4 days after kainic acid status epilepticus (SE) were profiled by H3K27me3 CUT&RUN-seq and bulk RNA-seq; methylated, repressed genes were checked in prior pilocarpine snRNA-seq. UNC1999 given 6, 24 and 48 h post-SE was tested with the same assays, flurothyl seizure threshold and video-scored chronic seizures.

  • Hippocampal H3K27me3 CUT&RUN-seq with IgG controls, naive vs 4 days after repeated low-dose kainic acid SE · mouse
  • Hippocampal bulk RNA-seq collected in tandem, naive vs 4 days post-SE · mouse
  • H3K27me3 CUT&RUN and RNA-seq from naive and post-SE mice given UNC1999 or vehicle (4 conditions) · mouse
  • Previously generated hippocampal snRNA-seq, pilocarpine SE model, naive vs 4 days post-SE (Hoffman et al.) · mouse
  • Flurothyl seizure threshold (baseline and 4 days post-SE) and video-scored behavioural seizures at weeks 5-6, UNC1999 vs vehicle · mouse
  • Sham-SE mice treated with UNC1999, seizure threshold before and after treatment · mouse
  • Western blot of hippocampal H2AK119ub with actin loading control · mouse · n = naive n=5; 4d post-SE n=8
  • EZH2, SUZ12 and CTCF CUT&RUN in naive and post-SE hippocampi (signal lost post-SE) · mouse
Methods & tools
CUT&RUN-seq, Bowtie2 2.5.4, MACS3, MSPC, DiffBind 3.16.0, ChIPseeker, salmon 1.10.3, DESeq2 1.46.0, Ontomancer (KEGG, Reactome, Biocarta, Hallmark), STRING, scanpy 1.12, flurothyl seizure threshold test
Reagents & environment
Bowtie2 2.5.4, samtools 1.2, deepTools 3.5.6, MACS3, MSPC, DiffBind 3.16.0, IRanges 2.44.0, ChIPseeker v3.22, Rsubread 2.20.0, salmon 1.10.3 (GRCm39, GENCODE M36), DESeq2 1.46.0, scanpy 1.12, R 4.4.3; reads aligned to mm39. No container or requirements file stated.
Assay conditions & compute
Run on the University of Wisconsin Center for High Throughput Computing (custom workflow for alignment, filtering, coverage and peak calling); no hardware, runtime or memory figures given.
Reproducibility gaps
CUT&RUN and RNA-seq data are to be deposited in GEO (no accession given); R and Python scripts are promised on 'github/dryad' with no link; other data are in supplements or on request. Per-group mouse and replicate numbers are not stated, except for the Western blot.
Limitations (6)
  • Bulk CUT&RUN and RNA-seq on whole hippocampus average over all cell types. Changes in small but disease-relevant populations, such as interneurons (roughly 10-15% of hippocampal neurons), can be masked; the authors add a snRNA-seq reanalysis to offset this. · stated by the authors
  • Occupancy of EZH2 and other PRC2 subunits was not mapped. EZH2, SUZ12 and CTCF CUT&RUN worked in naive hippocampi but gave IgG-like signal after SE, so H3K27me3 serves as a proxy for EZH2 activity. · stated by the authors
  • Only one time point (4 days post-SE) was profiled; whether the H3K27me3 and expression changes persist into later acute or chronic phases is unknown. · stated by the authors
  • It is unresolved whether H3K27me3 gains at genes with unchanged or undetected expression (over 77% of targets) have any function. Most are barely expressed in naive hippocampus, which makes RNA-seq quantification hard, and UNC1999 left about 90% of them unchanged. · stated by the authors
  • The link between the gene module and the seizure phenotypes is correlative. UNC1999 was given systemically and shifted module H3K27me3 by only 1.2-fold, with no gene- or cell-type-specific manipulation; the cell-type mapping uses a pilocarpine rather than a kainate SE dataset. · noted in this summary
  • Group sizes, effect sizes and p-values for the seizure-threshold and chronic-seizure results are not given in the text, and chronic seizures were counted from only four 5-6 h behavioural video sessions. · noted in this summary

Read this if you study epigenetic mechanisms of epileptogenesis or Polycomb/EZH2 function in post-mitotic neurons and want paired H3K27me3 CUT&RUN, RNA-seq and seizure-threshold data from a mouse kainate model.

Citations
0

Abstract

The processes by which epileptic insults precipitate the molecular, cellular, and network alterations in the brain that lead to epilepsy are poorly understood. We previously discovered that after status epilepticus (SE - an epilepsy inducing severe bout of seizures) the H3K27 methylase Enhancer of Zeste Homolog 2 (EZH2) is robustly induced and drives repression of genes. Both systemic pharmacological inhibition of EZH2 and deletion of EZH2 in neurons exacerbates epilepsy progression, suggesting that acute EZH2 induction may exert a net protective effect against disease progression chronically. However, the mechanisms underlying EZH2-mediated control of the putative protective and pathological pathways in disease progression are still unknown. To interrogate the mechanisms of EZH2 function post-SE, we used bulk CUT&RUN- sequencing against H3K27me3 in tandem with bulk RNA-sequencing in hippocampi of naive and 4d. post-SE mice to profile epigenomic and transcriptomic changes. Differential peak analysis showed that H3K27me3 was enriched both at loci pre-marked by H3K27me3 in the naive hippocampus as well as in loci that were de novo methylated after SE, consistent with the SE-dependent induction of EZH2 protein levels. Multi-omic integration of CUT&RUN and RNA-seq data revealed a module of genes that were coordinately H3K27me3 enriched, transcriptionally repressed, and annotated to ontological terms involved in neuronal signaling and network excitability. This result suggests that EZH2 induction may function in part to control network excitability after injury. To test this, we treated mice acutely post-SE with the EZH2 inhibitor UNC1999 and found that EZH2 inhibition attenuated H3K27me3 induction and dampened repression of target network excitability genes in response to SE. Functionally, UNC1999 treatment significantly increased seizure probability acutely and exacerbated disease severity in the chronic period. Taken together, these results suggest that EZH2 induction after injury may function to maintain network excitability to lower seizure probability acutely to protect against disease progression.

Reproducibility ledger

Nobody has reported trying to reproduce this paper yet. Did you replicate an experiment or run the code? Your report saves the next lab a week.

Sign in to file a reproducibility report.

Related papers

Similarity Map
SelectedSame fieldCross-field
THISKhan '19Lisgaras '21Subramanian '26Liu '24Mardones '22Martín‐Suár '22
75% semantic matchNeuroscience
Nadia Khan, Barry A. Schoenike, Trina Basu+8 more·PLoS ONE·Dec 31, 2019· 22 citations