Stable pausing by RNA polymerase II provides an opportunity to target and integrate regulatory signals

Abstract

Metazoan gene expression is often regulated after the recruitment of RNA polymerase II (Pol II) to promoters, through the controlled release of promoter-proximally paused Pol II into productive RNA synthesis. Despite the prevalence of paused Pol II, very little is known about the dynamics of these early elongation complexes or the fate of the short transcription start site-associated (tss) RNAs they produce. Here, we demonstrate that paused elongation complexes can be remarkably stable, with half-lives exceeding 15 min at genes with inefficient pause release. Promoter-proximal termination by Pol II is infrequent, and released tssRNAs are targeted for rapid degradation. Further, we provide evidence that the predominant tssRNA species observed are nascent RNAs held within early elongation complexes. We propose that stable pausing of polymerase provides a temporal window of opportunity for recruitment of factors to modulate gene expression and that the nascent tssRNA represents an appealing target for these interactions.

Figure 1. scRNAs co-localize with Pol II and remain stably associated with chromatin

(A) Heat map representations of Pol II ChIP-seq reads, mRNA expression levels determined by microarray and 3’-end sequencing reads of scRNAs isolated from different cellular fractions. Shown are Drosophila genes with significant Pol II occupancy (as determined by Pol II ChIP-seq using an α-Rpb3 antibody) and >550 nt in length (N=6,152). Genes are rank ordered by decreasing number of Pol II ChIP-seq reads observed around the transcription start sites (+/− 150 bp from TSS, depicted by arrow). Arrow heads at right of heat maps indicate the position above which 50% of the scRNA-seq reads are located. Color bars at bottom indicate range for each depth-normalized data set. (B and C) Composite metagene analysis of scRNA reads 3’-ends in the chromatin-associated pellet (B) or detergent-soluble fraction (C). Shown is the average number and position of scRNA 3’-end reads per TSS at the top decile of genes when rank ordered by mRNA expression, promoter Pol II ChIP-seq signal or Pausing Index, as indicated. (D) Heat maps are shown as in (A) but with genes rank ordered by decreasing mRNA expression levels. Of the Pol II-occupied TSSs shown in (A) 4,954 are represented by unique transcripts on the microarray and are included in this panel. See also Figure S1.

Figure 2. Disruption of exosome activity reveals a low-level of termination coupled with targeted degradation of released RNAs

(A) Heat maps showing 3'-end distributions of scRNAs isolated from Mock-treated or Exosome(Rrp40)-depleted cells at genes occupied by Pol II. Genes are rank-ordered and shown as in Figure 1A. (B) Northern analysis of short RNAs generated by promoter Pol II at the Hsp70 gene in Mock-treated and Rrp40-depleted cells. RNAs derived from whole cells are shown along with species present in the chromatin or soluble fractions. To quantify the abundance of these species, a 33 nt synthetic DNA corresponding to the Hsp70 5’-end was loaded over a range of molecular equivalents per gene copy in the S2 cell genome. (C) Enrichment in 3’-oligo-adenylated species observed in Rrp40-depleted samples. Shown is the difference in number of scRNA-seq reads with 3'-A residues of the specified length in Rrp40-depleted vs. Mock-treated samples. (D) Oligo-adenylated species identified in exosome-depleted cells are derived from near TSSs. Heat maps are depicted as in (A) for the species that mapped uniquely after trimming 3’-A residues. (E) Comparison of 3’-end positions of directly mappable and 3’-oligo-adenylated scRNA species isolated from Rrp40-depleted samples shows that they both localize tightly within the region of promoter-proximal pausing, suggesting that they share a common origin. (F and G) The percentage of reads bearing non-templated A-tails is consistently low across gene quartiles ranked by promoter Pol II signal (F) and gene activity (G). See also Figure S2.

Figure 3. Quantitative PCR indicates abundant tssRNAs at paused genes, with low levels of oligo-adenylated species

(A) 3’-end locations are shown for scRNAs derived from an example gene (CG9884) in Mock-treated and Rrp40-depleted cells. The number of reads at each location, in single nucleotide intervals, is shown for each sample. Note the >10-fold difference in scale between reads that were mapped directly and those with A-tails at their 3’-ends (oligo-A), as well as the lack of oligo-A reads in cells with intact exosome activity. The gene model is shown below, with TSS depicted as an arrow. (B) Schematic of the experimental protocol for amplification of tssRNA species. Following isolation of short RNAs, an in vitro adenylation reaction (left) allows for reverse transcription of all short RNA species using an anchored oligo-dT primer. Omitting the in vitro adenylation (right) allows for detection of only those species that are oligo-adenylated in vivo. The oligo-dT primer contains a sequence complementary to the universal Reverse primer (univ. R), allowing amplification of individual tssRNAs when coupled with a gene-specific Forward primer (G-S F). (C and D) Example quantification of the scRNA read counts near promoters (C, from the TSS to 100 nt downstream) and qPCR results (D) for tssRNA species generated by the CG9884 gene (top), and CG2207 (below). Values represent the averages of ≥3 biological replicates ± SEM. The small differences in directly mappable scRNAs or total short RNAs observed at individual genes were not statistically significant. See also Figure S3.

Figure 4. Measurement of the turnover of tssRNA and promoter Pol II indicates a long-lived elongation complex whose residence time is dictated by the rate of pause release

(A and B) Quantification of the short tssRNA levels (A, green) and promoter Pol II ChIP signal (B, blue) during a time course of treatment with Trp. Plots show the decay of signal from genes with varying levels of gene activity as measured by gene body GRO-seq signal: CG4427 (top) is highly active, CG9884 (middle) has moderate activity and CG33174 (bottom) exhibits very low activity. Signals at each time point were quantified by qPCR relative to DMSO-treated cells at t = 0. For tssRNA, each point represents the mean of three biological replicates ± SEM. ChIP samples show average values and range of two replicates. (C) The relationship between gene activity and the half-lives measured for tssRNAs is shown at thirteen genes, revealing that the lifetime of early elongation complexes is inversely correlated with the rate of release into productive elongation. See also Figure S4.

Figure 5. Flavopiridol causes broad accumulation of RNAs and Pol II near promoters

(A) Heat maps showing 3'-end distributions of scRNAs isolated from DMSO-treated or FP-treated cells at genes occupied by Pol II. Genes are rank-ordered and shown as in Figure 1A. Arrow heads at right of heat maps indicate the position above which 50% of the scRNA-seq reads are located, revealing a shift in FP-treated cells. (B) Composite metagene analysis of scRNA reads 3’-ends detected in the DMSO-treated or FP-treated cells. Shown is the average number and position of scRNA 3’-end reads per TSS. (C) Heatmap showing the Fold change in scRNA reads upon FP-treatment for genes divided into groups based on their Pausing index (y-axis) and Gene activity level (Gene body GRO-seq signal, x-axis). Color bars at bottom indicate range. (D) Quantification of the change in promoter Pol II ChIP signal at genes following FP treatment. Promoter Pol II signal was quantified relative to DMSO-treated cells (set as 100%) using primer pairs centered near the TSSs for each gene. The relative pausing index and gene activity level for each gene is shown. Data represents the mean of two biological replicates ± range. See also Figure S5.

Figure 6. Generation of stable promoter-associated Pol II complexes involves the pause-inducing factor NELF

(A) Promoter Pol II ChIP signal was measured following Trp-treatment in cells that were Mock-treated or NELF-depleted. Genes shown display High (e.g. CG4427), intermediate (e.g. CG8896 and CG9884) and Low (CG33174 and CG17697) levels of gene activity as determined by GRO-seq. The percent input of promoter Pol II at each time point was compared to the signal present at t=0 (before adding Trp), and is reported as the percentage of ChIP signal remaining at each time. Values represent averages of ≥3 biological replicates ± SEM. (B) Difference in Pol II half-life between Mock-treated and NELF-depleted cells. The twelve genes analyzed as in (A) were separated into three groups based on their gene activity levels in untreated S2 cells. Whereas highly active genes with short-lived promoter Pol II occupancy were not significantly affected by NELF-depletion (High P=0.155, N=3), genes with Intermediate and Low activity displayed more long-lived promoter occupancy in untreated cells that was significantly affected by NELF RNAi (Intermediate P=0.025, N=5; Low P=0.029, N=4). Values are mean half-lives for each group of genes ± SEM. See also Figure S6.

Figure 7. Stable pausing by Pol II provides a framework for regulation

(A) Model: Pol II (red) initiates transcription and is induced to pause through the activity of NELF and other factors (not shown). Paused Pol II and nascent tssRNA (blue) remain stably associated with the promoter while awaiting the recruitment of P-TEFb (green) which triggers productive elongation. Most promoter-associated Pol II can transition to productive elongation (thick arrow); however, we detect a low level of promoter-proximal termination (small arrow), with the exosome (purple) rapidly degrading the released tssRNAs. (B) Upon depletion of NELF, elongation complexes no longer stably pause and exhibit decreased half-lives near promoters. Many genes can support rapid re-initiation of Pol II to maintain similar Pol II occupancy, but genes with inefficient re-initiation display decreased expression. (C) We propose that the tssRNAs produced by, and stably associated with, paused Pol II could play an important role in transcription activation in cis through interactions with factors that affect productive elongation and promoter chromatin.