The temporal dynamics of lncRNA Firre-mediated epigenetic and transcriptional regulation

Abstract

Numerous studies have now demonstrated that lncRNAs can influence gene expression programs leading to cell and organismal phenotypes. Typically, lncRNA perturbations and concomitant changes in gene expression are measured on the timescale of many hours to days. Thus, we currently lack a temporally grounded understanding of the primary, secondary, and tertiary relationships of lncRNA-mediated transcriptional and epigenetic regulation-a prerequisite to elucidating lncRNA mechanisms. To begin to address when and where a lncRNA regulates gene expression, we genetically engineer cell lines to temporally induce the lncRNA Firre. Using this approach, we are able to monitor lncRNA transcriptional regulatory events from 15 min to four days. We observe that upon induction, Firre RNA regulates epigenetic and transcriptional states in trans within 30 min. These early regulatory events result in much larger transcriptional changes after 12 h, well before current studies monitor lncRNA regulation. Moreover, Firre-mediated gene expression changes are epigenetically remembered for days. Overall, this study suggests that lncRNAs can rapidly regulate gene expression by establishing persistent epigenetic and transcriptional states.

Fig. 1. Temporal gene regulation in Firre transgene mESC lines.

a Schematic showing the mouse blastocyst-derived mESC lines used in this study. The Firre^RESCUE and Firre^OE cell lines contain a dox-inducible Firre transgene in a Firre KO and WT background, respectively. The Firre^KO-CTL and WT cell lines do not induce Firre expression and serve as dox controls. b, c Firre expression levels in the Firre^RESCUE (b) and Firre^OE (c) line (red) as well as the Firre^KO-CTL (b) and WT (c) line (black) across time after the addition of dox as measured by RNA-seq. Numbers indicate fold expression changes over Firre WT levels at each time point. d, e Expression log₂ fold change of significantly differentially expressed genes that are affected by dox treatment in the Firre^KO-CTL (d) and WT (e) control cell line. Fold changes are relative to the zero-time point. f Linear model to derive genes that are significantly differentially regulated upon Firre transgene induction and not due to dox treatment (likelihood ratio test). g Heatmap of gene expression changes of early Firre-responsive genes over time as determined by the linear model in (f). h Gene expression changes of representative genes due to Firre in Firre-inducing (red) but not in dox control (black) cells of Firre WT and KO background. i Maximum intensity projections of smRNA FISH images for Shf exon (magenta) and Firre exon signals (white) in Firre^RESCUE and Firre^OE mESC colonies after 0, 2, 4, and 6 h of dox treatment. Nuclei are stained with Hoechst (blue), Firre exon in white, and Shf exon in magenta. Scale bar is 5 μm. (j) Quantification of Shf exon FISH signals in Firre^RESCUE and Firre^OE mESCs. The horizontal line indicates the mean and the vertical line the non-outlier range of quantified cells.

Fig. 2. Changes in chromatin accessibility upon induction of Firre.

a Heatmap showing chromatin accessibility log₂ fold changes across time. Increased chromatin accessibility relative to the zero-time point is depicted in shades of red, decreased chromatin accessibility in blue. Black tick marks indicate differentially expressed genes. b Venn diagram showing the overlap of early Firre-responsive genes and Firre-mediated chromatin accessibility sites. c TSS metaplot of the density of ATAC-seq peak centers relative to Firre-responsive gene TSSs. Red tick marks indicate the peak centers. d Genome browser tracks of chromatin accessibility changes across time within Rapgef4 in Firre^RESCUE cells. e Heatmap of chromatin accessibility and gene expression changes at each time point after induction of Firre. Time points where chromatin accessibility was not measured are indicated in gray. f Plot of chromatin accessibility (red) and gene expression (black) signal relative to the maximum measured over time. The dashed line represents the time difference in half-maximum values.

Fig. 3. Rapid nascent transcriptional changes upon Firre induction.

a Schematic of PRO-seq experimental setup. b Genome browser tracks of bidirectional transcription of genes that were (Shf, Duox1) and were not (Ep300) differentially expressed upon Firre induction. Red and gray areas indicate reads on the plus and minus strand, respectively. Arrows denote the bidirectional center. c Metaplot of PRO-seq signal at the differentially expressed sites of bidirectional transcription (padj < 0.05, n = 550, Wald test) around the bidirectional center (μ) at the 0-, 15-, and 30-min time points. d Genome browser tracks showing the regulatory events at the Apod locus upon Firre induction. The arrow indicates an internal TSS. e Volcano plot of the log₂ fold changes of nascent transcription over the gene body at 30 vs 0 min. Genes with padj <0.05 (Wald test) are indicated in red.

Fig. 4. Persistence of Firre-mediated transcriptional regulation over time.

a, b Expression levels of Firre in Firre^RESCUE (red) and Firre^KO-CTL mESCs (black) (a) and Firre^OE (red) and WT mESCs (black) (b) across time after the addition of dox as measured by RNA-seq. The dashed line indicates Firre WT expression levels. c Individual gene plots showing the log₂ expression fold change in the Firre-inducing cell lines compared to the zero-time point for all measured time points. d Heatmap of the expression fold changes of significantly differentially expressed genes over time upon Firre induction independent of dox effects. Black tick marks represent genes whose expression was rescued, i.e., that were downregulated in the knockout and upregulated upon transgene induction, or vice versa. e Heatmap of the expression of 68 rescued genes across each time point. Top and bottom heatmap represent the Firre^KO and WT expression levels, respectively. The heatmaps in between represent the temporal timing of when genes are rescued. f PCA of the rescued gene expression levels in WT, Firre^KO and Firre^RESCUE cells at each time point. The primary component PC1 that represents 75% of the variation in the data is plotted. g Model of Firre-mediated gene regulation depicting Firre forming a lncRNP by binding to an unknown protein, which either engages an activator protein to bind to chromatin or evicts a repressor protein from chromatin, ultimately leading to the upregulation of Firre target genes.