Functional genomic profiling of O-GlcNAc reveals its context-specific interplay with RNA polymerase II

Abstract

Background: How reversible glycosylation of DNA-bound proteins acts on transcription remains scarcely understood. O-linked β-N-acetylglucosamine (O-GlcNAc) is the only known form of glycosylation modifying nuclear proteins, including RNA polymerase II (RNA Pol II) and many transcription factors. Yet, the regulatory function of the O-GlcNAc modification in mammalian chromatin remains unclear.

Results: Here, we combine genome-wide profiling of O-GlcNAc-modified proteins with perturbations of intracellular glycosylation, RNA Pol II-degron, and super-resolution microscopy. Genomic profiling of O-GlcNAc-modified proteins shows a non-random distribution across the genome, with high densities in heterochromatin regions as well as on actively transcribed gene promoters. Large-scale intersection of the O-GlcNAc signal at promoters with public ChIP-seq datasets identifies a high overlap with RNA Pol II and specific cofactors. Knockdown of O-GlcNAc Transferase (Ogt) shows that most direct target genes are downregulated, supporting a global positive role of O-GlcNAc on the transcription of cellular genes. Rapid degradation of RNA Pol II results in the decrease of the O-GlcNAc levels at promoters encoding transcription factors and DNA modifying enzymes. RNA Pol II depletion also unexpectedly causes an increase of O-GlcNAc levels at a set of promoters encoding for the transcription machinery.

Conclusions: This study provides a deconvoluted genomic profiling of O-GlcNAc-modified proteins in murine and human cells. Perturbations of O-GlcNAc or RNA Pol II uncover a context-specific reciprocal functional interplay between the transcription machinery and the O-GlcNAc modification.

Keywords: CUT&RUN; ChIP-seq; ChIP‐Atlas; Degron; Glycosylation; O-GlcNAc; O-linked β-N-acetylglucosamine; OGT; Promoter; RNA polymerase II; RNA-seq; Transcription.

Fig. 1

O-GlcNAc-modified proteins occupy heterochromatin regions and gene promoters. A Upset plot showing the distribution of O-GlcNAc peaks (527 and 523 peaks for replicates 1 and 2) profiled by CUT&RUN in murine ESCs across the following functional genomic regions: Active promoters (79 and 84 peaks), transcription initiation (64 and 71 peaks), heterochromatin (42 and 52 peaks), bivalent promoters (29 and 30 peaks), Polycomb (PcG) domains (24 and 23 peaks), transcription elongation (23 and 12 peaks), and transcription termination (9 and 11 peaks). Two technical replicates are shown. B Representative genomic tracks of O-GlcNAc occupancy patterns in correlation with H3K4me3 (Sequence Read Archive (SRA): SRX5382140 [80]) as a marker of gene promoter and H3K9me3 (SRA: SRR925652 [81]) as a marker of heterochromatin regions. At the bottom, Refseq genes, LINE1 (L1), and LTR retrotransposons are shown. C Violin plot showing the expression levels of genes using a publicly available RNA-seq (SRA: SRR11294181 [31]) with promoters highly occupied by O-GlcNAc (CUT&RUN) modified proteins (n = 236), promoters without detectable O-GlcNAc signal (n = 236), randomly selected promoters (n = 236), and all promoters (n = 21,085). The boundaries of the overlaid box plot show the data above the 1st and within the 3.^rd quartiles, whiskers indicate minimum and maximum values, and the horizontal bar in the box plot shows the median. Differences in median expression levels were assessed with a Mann–Whitney-Wilcoxon two-sided test. *: p < 0.05; **: p < 0.01. D Percentages of signal overlap at promoters between O-GlcNAc peaks and 21,205 ChIP-Atlas datasets in mouse ESCs. Proteins previously described to be O-GlcNAcylated are indicated in green. The highest overlap is found with the pan-O-GlcNAc ChIP-seq GSE93539. E Genes-stack plots showing the O-GlcNAc CUT&RUN signal (left) and RNA Pol II (right) at all Ensembl genes (n = 55,634). F Heatmap showing O-GlcNAc CUT&RUN signal along with ChIP-seq signal of proteins involved in transcription with high genomic overlap with O-GlcNAc + / − 1 kb around O-GlcNAc peak centers, ordered by k-mean clustering on RNA Pol II signal on the union of the replicate peaks (702 peaks). All rows are centered on O-GlcNAc peaks based on the ranking of signals. The percentages of overlap with each O-GlcNAc replicate are RNA Pol II (72/76%), TBP (68/67%), TAF12 (56/57%), NELFA (45/46%), MED1 (58%), MED12 (49/48%), MED24 (57/58%), MED26 (47/49%), and DR1 (75/69%)

Fig. 2

O-GlcNAc is required to sustain the expression of a set of target genes. A Western blot detection of OGT, pan-O-GlcNAc (RL2 antibody), and histone H3 (loading control) in total protein extracts from WT ESCs transfected with control siRNA (non-targeting) or with siRNA against Ogt. The blots are representative of two independent experiments. B Volcano plot showing differential gene expression between ESCs transfected with siRNA control and siRNA anti-Ogt. Differentially expressed genes included 836 downregulated (dark blue) and 592 upregulated (dark red) genes (adj. p-value < 0.05, Wald Test, any log2FC). Among these, 44 downregulated and 12 upregulated genes have a fold-change higher than two. Thirty-one downregulated (light blue) and two upregulated (orange) genes have an overlapping O-GlcNAc peak in WT cells. C Representative genomic tracks of O-GlcNAc regulated promoters as defined by high occupancy of O-GlcNAc-modified proteins in WT cells (top CUT&RUN track) and downregulated upon global removal of O-GlcNAc (bottom RNA-seq track). D Gene set enrichment analysis of the up-, down-, and down-O-GlcNAc regulated genes shown in panel B. The number of genes enriched in gene ontology (GO) molecular function (MF) are 1461, 1324, and 31, respectively. The gene ratio reflected by the size of dots indicates the proportion of genes matching a GO set

Fig. 3

Super-resolution imaging of RNA Pol II clusters after nuclear O-GlcNAc perturbation. A Schematic representation of the Tet-ON inducible transgene of bacterial OGA BtGH84, fused to a localization peptide (NLS) and the knock-in of the epitope tags HA and SPOT to endogenous Polr2a gene encoding RNA Pol II. B Western blot (WB) detection of the kinetic depletion of O-GlcNAc (detected by WGA, top), following ectopic expression of the bacterial OGA homolog BtGH84 fused to a localization peptide, namely BtGH84-NLS. The bottom panels show the WB detection of OGT and Lamin A/C (loading control) at the indicated time points following doxycycline induction of BtGH84-NLS expression. C Quantification by normalized optical density of the WB detection of SOX2-O-GlcNAc at the indicated time points after expression of BtGH84-NLS. The blots are shown in Additional file 1: Fig. S2D. D Immunoprecipitation of endogenous RNA Pol II using the ESC line described in A whereby Polr2a was targeted with a knock-in SPOT epitope tag. The immunoprecipitation was performed with magnetic beads coated with anti-SPOT antibodies. The efficiency of RNA Pol II IP and RNA Pol II O-GlcNAc levels were probed by WB analysis at different time points after Dox-induction of BtGH84-NLS expression. E Quantification of the western blot shown in D by the normalized optical density of O-GlcNAc on immunoprecipitated RNA Pol II. The ratio IP RNA Pol II / O-GlcNAc is plotted at different time points after BtGH84-NLS expression. F Representative micrographs of RNA Pol II and DNA (DAPI) acquired by stimulated emission depletion (STED) microscopy in ESCs and NPCs before and after prolonged nuclear O-GlcNAc depletion by expression of BtGH84-NLS (48 h and 5 days, respectively). Scale bars indicate 2 μm. G Distribution of fluorescence intensity quantification of RNA Pol II clusters from STED images before and after BtGH84 induction in ESCs and NPCs (t-test. Not-significant (ns): p > 0.05; *: p < 0.05; **: p < 0.01; ***: p < 0.001; ****: p < 0.0001)

Fig. 4

RNA Pol II-dependent O-GlcNAcylation. A Western blot showing the TIR1-mediated RNA Pol II degradation in DLD-1 cells upon 14-h Doxycycline/Auxin treatment. B Percentages of signal overlap at promoters between 3450 and 2630 O-GlcNAc peaks (before and after Dox treatment, replicate 1) and 2834 ChIP-Atlas datasets in human DLD-1 cells. Proteins previously described to be O-GlcNAcylated are indicated in green. The highest overlap is found with RNA Pol II ChIP-seq GSM5237208. The overlap percentages before and after Dox treatment respectively are RNA Pol II (70/83%), NELFCD (56/39%), SUPT5H (62/55%), NCBP1 (22%), NELFE (40/70%), and INTS3 (12/76%). C Genes-stack plots of RNA Pol II ChIP-seq (GEO: GSM5237208, SRA: SRX10580013 [82]) and O-GlcNAc CUT&RUN signals before (middle panel) and after RNA Pol II degradation (right panel). Signal is represented at 21,519 genes (GRCh38 Ensembl) + / − 2 kb. D Left: K-mean clustering of 6544 O-GlcNAc occupancy peaks (union of replicates), + / − 1 kb around their centers. Five different clusters were defined based on the O-GlcNAc signal before and after RNA Pol II removal by Dox/Aux treatment. Right: Corresponding RNA Pol II ChIP-seq signal (SRA: SRX11070611 and SRX11070613, respectively [54]). E Genome browser view illustrating RNA Pol II (SRA: SRX11070611 and SRX11070613 respectively [54]) and O-GlcNAc signal upon Dox/Aux treatment. As an example of cluster 4 (gain of O-GlcNAc), ARMC5 is shown. To illustrate cluster 5 (loss of O-GlcNAc), the TAF8 gene is shown. F Gene set enrichment comparison using molecular functions for clusters 2 (n = 239), 4 (n = 347), and 5 (n = 751); clusters 1 and 3 did not highlight any significant enrichment. The gene ratio reflected by the size of dots indicates the proportion of genes matching a GO set