Oxidative stress rapidly stabilizes promoter-proximal paused Pol II across the human genome

Abstract

Oxidative stress has pervasive effects on cells but how they respond transcriptionally upon the initial insult is incompletely understood. We developed a nuclear walk-on assay that semi-globally quantifies nascent transcripts in promoter-proximal paused RNA polymerase II (Pol II). Using this assay in conjunction with ChIP-Seq, in vitro transcription, and a chromatin retention assay, we show that within a minute, hydrogen peroxide causes accumulation of Pol II near promoters and enhancers that can best be explained by a rapid decrease in termination. Some of the accumulated polymerases slowly move or 'creep' downstream. This second effect is correlated with and probably results from loss of NELF association and function. Notably, both effects were independent of DNA damage and ADP-ribosylation. Our results demonstrate the unexpected speed at which a global transcriptional response can occur. The findings provide strong support for the residence time of paused Pol II elongation complexes being much shorter than estimated from previous studies.

Figure 1.

The nuclear walk-on assay quantitatively and precisely measures engaged Pol II. (A) Steps in the nuclear walk-on assay are illustrated. (B) Adherent HeLa cells were transfected 48 h before isolation of nuclei with 1.5 nM NELF-A siRNA (NELF KD) or lipid reagent only (Mock). Cells were also treated 1 h before isolation with 0.1% DMSO or 1 μM flavopiridol (Flavo). Nuclear walk-ons were performed using a 6 min α-³²P-CTP pulse in the absence or presence of 2 μg/ml α-amanitin (α-aman). After 6% Urea-PAGE, ethidium bromide was used to visualize nuclear RNAs. A representative gel from nuclear walk-ons performed in triplicate is shown here. (C) Phosphorimage to visualize radiolabeled nascent transcripts from B. (D) Average amanitin-sensitive Pol II nascent transcript profiles from triplicate nuclear walk-ons as represented in B and C. For each replicate, signals from each lane in the phosphorimage were normalized using quantifications of cold nuclear RNAs in the ethidium bromide stain. Then, α-amanitin-insensitive signals were subtracted from total signals. Triplicate profiles (see Supplementary Figure S1) were then averaged as described in the Materials and Methods. Inset: western blot of NELF-A; a nonspecific band (n.s.) indicates even loading. (E) PRO-Seq paired-end reads over the DDIT4 gene and upstream enhancer regions. Adherent HeLa cells were treated 1 h with either 0.1% DMSO or 1 μM flavopiridol. (F) Plot of sense or divergent PRO-Seq fragments lengths near Pol II promoters.

Figure 2.

H₂O₂ causes rapid accumulation, then creeping of promoter-proximal Pol II. (A) Capped nascent Pol II transcript profiles from a nuclear walk-on using nuclei from adherent HeLa cells treated with 0.3 mM H₂O₂ for indicated times. No H₂O₂ control profiles are duplicated in all panels. Capped transcripts were selected by treating Trizol-isolated RNAs with recombinant RNMT and SAM, and then using an immobilized anti-2,2,7-trimethylguanosine antibody. All plots in A were from the same gel (see Gel Appendix) and have the same vertical scale (relative signal). (B) The experiment in A, except all cells were treated 1 h with 1 μM flavopiridol prior to H₂O₂. All plots in B were from the same gel and have the same vertical scale (relative signal), which is reduced to match the paused Pol II peak height in A. (C) Amanitin-sensitive Pol II nascent transcript profiles from a nuclear walk-on using nuclei from adherent HeLa cells treated 1 h with 0.1% DMSO. 0.3 mM H₂O₂ was added during the last 20 s, 1 min or 3 min of treatment as indicated (n = 2). (D) The experiment in C, except all cells were treated 1 h with 1 μM flavopiridol instead of DMSO alone. DMSO was 0.1% in all conditions. Figure 2C and D were from the same gel and have the same vertical scale (relative signal). (E) Amanitin-sensitive Pol II nascent transcript profiles from a nuclear walk-on using nuclei from adherent HeLa cells treated 1 h with 1 μM flavopiridol. 0.3, 3 or 30 mM H₂O₂ were added during the last 10 min of treatment as indicated. After a 5 min α-³²P-CTP pulse, reactions were chased as indicated for 10 min with 500 μM cold ATP, UTP, GTP and CTP. All plots in Figure 2E were from the same gel and have the same vertical scale (relative signal).

Figure 3.

Effects of H₂O₂ occur genome-wide at promoters and enhancers. (A) ChIP-Seq reads over the DDIT4 gene and upstream enhancer regions. Pol II: adherent HeLa cells were left untreated (Control) or treated 1 h with 1 μM flavopiridol, and then with 0.3 mM H₂O₂ for indicated times. Pileup densities were depth-adjusted using the average number of mapped sequence reads in all 10 samples. NELF: suspension HeLa cells were treated 1 h with 1 μM flavopiridol. H3K4me3 and H3K4me1: untreated suspension HeLa cells. (B) Average Pol II ChIP-Seq occupancies ±1.5 kb around TSS within the EPDnew eukaryotic promoter database not within 1 kb of a neighboring TSS (n = 17592). The vertical axes represent depth-adjusted signal and all plots have the same vertical scale. (C) Average Pol II ChIP-Seq occupancies ±1.5 kb around a subset of enhancers, defined here as the center of non-TSS NELF peaks preferentially enriched for H3K4me1 over H3K4me3 (n = 1000). The vertical axes represent depth-adjusted signal and all plots have the same vertical scale.

Figure 4.

NELF is selectively lost from creeping elongation complexes genome-wide. (A) Pol II and NELF ChIP-Seq occupancies ±1 kb around the top 8000 EPDnew TSS with the highest median Pol II signal. Suspension HeLa cells were treated 1 h with 1 μM flavopiridol, and then 35 min with 0.3 mM H₂O₂ as indicated. Signals were depth-adjusted using the average number of mapped sequence reads in all four samples. The heatmaps were sorted by TSS with decreasing Pol II occupancy in the no H₂O₂ control. (B) Average normalized Pol II and NELF ChIP-Seq occupancies ±1.5 kb around 17592 EPDnew TSS. The vertical axes represent depth-adjusted signals that were background-subtracted and normalized using a window ±10 kb around genes. Both plots have the same vertical scale. (C) The heatmaps from A were sorted by the change in median NELF occupancy after 35 min 0.3 mM H₂O₂ treatment. (D) Average normalized Pol II ChIP-Seq occupancies ±1.5 kb around the 2000 EPDnew TSS with the largest decrease (red) or increase (blue) in median NELF occupancy after 35 min 0.3 mM H₂O₂ treatment. The vertical axes represent depth-adjusted signals that were background-subtracted and normalized using a window ±10 kb around genes. Both plots have the same vertical scale.

Figure 5.

H₂O₂ directly inhibits pausing by DSIF and NELF. (A) Elongation complexes were generated by first preincubating HeLa nuclear extract and immobilized template DNA for 30 min, and then pulsing with limiting α-³²P-CTP for 30 s. Complexes were isolated by high salt wash and incubated 10 min with 1 mM DTT or indicated concentrations of H₂O₂. Addback mixtures containing buffer alone (Mock, lanes 1–4) or also containing 0.3 pmol DSIF and 0.6 pmol NELF (lanes 5–8) were also incubated 10 min with 1 mM DTT or indicated concentrations of H₂O₂ prior to incubation with complexes for 5 min. Elongation rates were measured by chasing for 5 min with 500 μM cold ATP, UTP, GTP and CTP. 6% Urea–PAGE. (B) Amanitin-sensitive Pol II nascent transcript profiles from a nuclear walk-on using nuclei from adherent HeLa cells treated 50 min with 1 μM flavopiridol, then 10 min with 0.1% DMSO (flavopiridol only), 10 min with 50 μM PJ34, or 20 min with 50 μM PJ34. Cells were additionally treated 10 min with 0.3 mM H₂O₂ as indicated. DMSO was 0.2% in all conditions. The vertical axis represents relative signal from Pol II and all curves were from the same gel. (C) Average normalized Pol II ChIP-Seq occupancies ±1.5 kb around 17592 EPDnew TSS. Suspension HeLa cells were treated 40 min with 1 μM flavopiridol, then 20 min with either 0.05% DMSO (flavopiridol only) or 20 μM PJ34. Cells were additionally treated 10 min with 0.3 mM H₂O₂ as indicated. The vertical axis represents depth-adjusted signals that were background-subtracted and normalized using a window ±10 kb around genes.

Figure 6.

H₂O₂ inhibits Pol II termination in vitro and increases Pol II engagement in cells. (A) Preinitiation complexes were formed on immobilized template DNA in the presence of indicated concentrations of H₂O₂. Elongation complexes were pulsed 30 s with limiting α-³²P-CTP and chased as indicated for 3 min with 500 μM cold ATP, UTP, GTP and CTP. Lengths of limiting-CTP transcripts are indicated; brackets indicate sizes of capped and uncapped transcripts. 6% Urea-PAGE. (B) Amanitin-sensitive Pol II nascent transcript profiles from a nuclear walk-on using nuclei from adherent HeLa cells treated 100 min with 1 μM flavopiridol, and then 3 min with 0.3 mM H₂O₂ as indicated. 500 nM triptolide was added either during the last 30 min of flavopiridol treatment, or during all 100 min of flavopiridol treatment. The vertical axis represents relative signal from Pol II and all curves were from the same gel. (C) 0.1, 1 or 10 μM triptolide was added to preinitiation complexes 30, 10, 3 or 1 min before a 30 s limiting α-³²P-CTP pulse. Labeled non-tRNA transcripts were quantified, adjusted for loading using cold nuclear RNAs, and normalized to reactions lacking triptolide. Error bars represent S.E.M. from three replicates. (D) Preinitiation complexes were formed on immobilized template DNA in the presence of 1 μM flavopiridol alone (C; control) or in combination with 0.3 mM H₂O₂ (H). Elongation complexes were pulsed 30 s with limiting α-³²P-CTP and chased 10 min with 500 μM cold ATP, UTP, GTP and CTP. Labeled transcripts from bead-bound (B) and supernatant (S) fractions were quantified to measure termination. Runoff transcripts (508 nt) were considered bound and tRNAs were excluded. Error bars represent S.E.M. and p-values are from 3 replicates (see Gel Appendix). 12% Urea–PAGE. (E) Top panel: western blot of RPB1 (sc-55492) and Ponceau S staining. Total (T) or cytosolic (C) and nuclear (N) fractions were obtained from adherent HeLa cells treated 1 h with 0.1% DMSO or 1 μM flavopiridol, and then 10 min with 0.3 mM H₂O₂ as indicated. Shown are two representative blots for RBP1. The bottom plot pairs with the Ponceau S staining. 4–20% SDS-PAGE. Bottom panel: plot of the percentage of RPB1 in the cytosolic (C/C+N) or nuclear (N/C+N) fractions from five replicates.

Figure 7.

Mechanistic model of the rapid and global transcriptional response to H₂O₂. Illustrated are the key events Pol II transcription normally (black) or after exposure to oxidative stress (red). We propose that H₂O₂ (1) inhibits turnover of paused elongation complexes and (2) induces loss of NELF association and pausing activity.