The deubiquitinase Usp9x regulates PRC2-mediated chromatin reprogramming during mouse development

Abstract

Pluripotent cells of the mammalian embryo undergo extensive chromatin rewiring to prepare for lineage commitment after implantation. Repressive H3K27me3, deposited by Polycomb Repressive Complex 2 (PRC2), is reallocated from large blankets in pre-implantation embryos to mark promoters of developmental genes. The regulation of this global redistribution of H3K27me3 is poorly understood. Here we report a post-translational mechanism that destabilizes PRC2 to constrict H3K27me3 during lineage commitment. Using an auxin-inducible degron system, we show that the deubiquitinase Usp9x is required for mouse embryonic stem (ES) cell self-renewal. Usp9x-high ES cells have high PRC2 levels and bear a chromatin and transcriptional signature of the pre-implantation embryo, whereas Usp9x-low ES cells resemble the post-implantation, gastrulating epiblast. We show that Usp9x interacts with, deubiquitinates and stabilizes PRC2. Deletion of Usp9x in post-implantation embryos results in the derepression of genes that normally gain H3K27me3 after gastrulation, followed by the appearance of morphological abnormalities at E9.5, pointing to a recurrent link between Usp9x and PRC2 during development. Usp9x is a marker of "stemness" and is mutated in various neurological disorders and cancers. Our results unveil a Usp9x-PRC2 regulatory axis that is critical at peri-implantation and may be redeployed in other stem cell fate transitions and disease states.

Fig. 1. Usp9x promotes ES cell self-renewal and a transcriptional signature of preimplantation linked to PRC2 activity.

a Schematic of an auxin-inducible degron (AID) system for acute Usp9x depletion in mouse embryonic stem (ES) cells with representative flow cytometry plot of GFP (AID-Usp9x) expression in Usp9x-low and Usp9x-high ES cells. Right: western blot of endogenous Usp9x level in sorted cell fractions (see Supplementary Fig. 1b). b Quantification and representative images of colony formation assays. Usp9x-low ES cells display a self-renewal deficit. AP, Alkaline Phosphatase. c Principal Component (PC) Analysis of gene expression by RNA-seq. 8h: 8 h auxin. No auxin: AID-Usp9x cells with vehicle treatment. 48h: 8 h auxin followed by 48 h recovery without auxin. Flag: Flag-Usp9x cells after 8 h auxin and 48 h recovery. d The transcriptional signatures of Usp9x-high or Usp9x-low ES cells correlate with different stages of peri-implantation development by Gene Set Enrichment Analysis (GSEA). Genes differentially expressed between Usp9x-high or Usp9x-low ES cells and controls were used in each case. See Methods for references. DE, differentially expressed (relative to controls); NS, not significant (FDR > 0.05); NES, Normalized Enrichment Score. e Usp9x mRNA expression in the epiblast declines from pre- to postimplantation^,. f Flow cytometry plot measuring median fluorescence intensity of GFP (Usp9x expression) in Usp9x-high and Usp9x-low ES cells after 8 h auxin treatment and 48 h recovery (without auxin). g Fold-change in expression of all genes at 48 h relative to control cells, showing hypotranscription in Usp9x-high ES cells and hypertranscription in Usp9x-low ES cells. h Heatmaps with summary profile plot of Suz12 binding (data from wild-type ES cells) over the genes upregulated in Usp9x-low cells or a random subset (N = 1310). i Boxplots showing repression (in Usp9x-high) or induction (in Usp9x-low) of Suz12 target genes, compared to a random subset (N = 3350). Western blots represent at least two biological replicates (a). Data are mean ± s.d. of four replicates from two independent experiments (b), mean ± s.d. of 3–4 replicates (e), representative of three experiments (f, h). Boxplot hinges (g, i) show the first and third quartiles, whiskers show ±1.5*inter-quartile range (IQR) and center line shows median of three biological replicates. ****P < 2.2 × 10⁻¹⁶. P-values by one-way ANOVA with Dunnett’s multiple comparisons test to the no-auxin condition (b), two-tailed Student’s t-tests with Welch’s correction (e), two-tailed Wilcoxon rank-sum test (g), and ANOVA with pairwise two-tailed Wilcoxon tests (i).

Fig. 2. Usp9x-mutant embryos arrest at E9.5–E11.5 and display defective repression of early lineage programs marked by H3K27me3.

a Genetic cross to delete Usp9x in epiblast derivatives of postimplantation embryos. Quantification of recovered (live) male embryos at several postimplantation stages (right). b Sample images and quantification of control and mutant embryo phenotypes. Relative to controls (left), E9.5 embryos show variable developmental delay, with closed arrow indicating an open anterior neuropore. E11.5 embryos show a range of phenotypes, from hemorrhage to severe delay and death (tally includes dead embryos). Open arrow indicates pericardial edema. Scale bars = 250 µm (E9.5), 2.8 mm (E11.5), with N indicated. c MA plots of expression changes by RNA-seq in two litters of Usp9x mutants versus controls (at E8.5). 3 mutants and 3 controls were sequenced per litter (N = 12 embryos total; see Supplementary Fig. 3b-d). d Enrichr analysis of the top-enriched transcription factors (TF) that bind to the genes upregulated in Usp9x-mutant embryos in various cell types. e Expression of the 71 genes upregulated in Usp9x mutants during wild-type development. FC, Fold-change relative to E6.5 embryos. f Distribution and boxplot quantification of H3K27me3 levels over the promoters of genes upregulated in Usp9x mutants (10 kb upstream, 1 kb downstream of TSS). g Representative genome browser tracks of H3K27me3 in wild-type embryos (E6.5–E8.5) at the Nodal locus. Known enhancer elements are highlighted and show gains of H3K27me3. h Nodal mRNA expression in wild-type development. i Nodal mRNA expression in E8.5 Usp9x-mutant or control embryos. Boxplot hinges show the first and third quartiles, whiskers show ±1.5*IQR and center line shows median of 2–3 biological replicates (e, f). Data are representative of 2–3 biological replicates (g), mean ± s.e.m. of 2–3 biological replicates (h), or 6 biological replicates (i). P-values by χ² test (a, b), Fisher’s exact test (d), two-tailed Student’s t-tests with Welch’s correction (e, i), two-tailed Wilcoxon rank-sum tests (f), and ANOVA with Dunnett’s multiple comparison test to E6.5 (h). χ² = 19.78 (a), 85.19 (b, top), 147.8 (b, bottom).

Fig. 3. Usp9x mediates a pre- to postimplantation switch in global H3K27me3 levels.

a Quantification and representative western blot of H3K27me3 from histone extracts. b ChIP recovery before sequencing recapitulates the global gain of H3K27me3 in Usp9x-high ES cells. c Heatmaps of H3K27me3 ChIP-seq signal in Usp9x-high and Usp9x-low ES cells, showing H3K27me3 spreading in Usp9x-high cells. d Profile plot depicting the mean signal of coverage shown in c. e Usp9x-high ES cells carry more H3K27me3 over repetitive elements compared to untreated (no-auxin) cells. Each point represents an individual element. Dotted lines are |log₂(0.7)|. f Cumulative enrichment plots of H3K27me3 enrichment in non-overlapping genomic bins of in vivo developmental stages (left) and ES cell states (middle, right)^,,. Preimplantation stages (2C-ICM, top) or preimplantation-like ES cell states (2i, Usp9x-high) display H3K27me3 enrichment. Epi, epiblast; mor. morula; ICM, inner cell mass. Enrichment (x-axis) is log₂(H3K27me3/input +0.5). g PCA plot clustering Usp9x-high and Usp9x-low ES cells among the samples shown in f based on genome-wide H3K27me3 distributions. Each point represents a biological replicate. Data are mean of two biological replicates (a, b), sum of two biological replicates (d, e). ****P < 2.2 × 10⁻¹⁶ by two-tailed Kolmogorov–Smirnov test comparing the average of biological replicates.

Fig. 4. Usp9x is a PRC2 deubiquitinase.

a CNN western blots for Suz12 and Ezh2 proteins in whole-cell extracts. b Co-immunoprecipitation (IP) and western blot showing a reciprocal interaction between endogenous Ezh2 and Usp9x in wild-type ES cells. c Acute Usp9x depletion over a time course from 0–8 h leads to gain of ubiquitinated species of Suz12 and Ezh2 and destabilization of their protein levels. HA-Ub, HA-tagged ubiquitin; (Ub)_n, polyubiquitination. d Overexpressing a catalytic mutant (C1566S) versus wild-type (wt) form of the Usp9x catalytic domain (DUB) leads to gain of Suz12 and Ezh2 ubiquitin levels. AID-Usp9x cells were treated with auxin for 8 h to deplete endogenous Usp9x. Asterisk (*) designates the expected sizes for non-ubiquitinated species. Western blots are representative of at least two biological replicates.

Fig. 5. Model for the Usp9x-PRC2 regulatory axis in early development.

Usp9x is a PRC2 deubiquitinase that promotes diffuse H3K27me3 deposition and a preimplantation-like transcriptional state. Loss of Usp9x leads to PRC2 destabilization, restricted H3K27me3 deposition and global hypertranscription with priming of postimplantation lineages in ES cells. After gastrulation, Usp9x is required for timely silencing of developmental genes that are PRC2 targets and for normal development beyond E8.5, suggesting redeployment of the Usp9x-PRC2 connection later in development.