Regulation of the RNAPII Pool Is Integral to the DNA Damage Response

Abstract

In response to transcription-blocking DNA damage, cells orchestrate a multi-pronged reaction, involving transcription-coupled DNA repair, degradation of RNA polymerase II (RNAPII), and genome-wide transcription shutdown. Here, we provide insight into how these responses are connected by the finding that ubiquitylation of RNAPII itself, at a single lysine (RPB1 K₁₂₆₈), is the focal point for DNA-damage-response coordination. K₁₂₆₈ ubiquitylation affects DNA repair and signals RNAPII degradation, essential for surviving genotoxic insult. RNAPII degradation results in a shutdown of transcriptional initiation, in the absence of which cells display dramatic transcriptome alterations. Additionally, regulation of RNAPII stability is central to transcription recovery-persistent RNAPII depletion underlies the failure of this process in Cockayne syndrome B cells. These data expose regulation of global RNAPII levels as integral to the cellular DNA-damage response and open the intriguing possibility that RNAPII pool size generally affects cell-specific transcription programs in genome instability disorders and even normal cells.

Keywords: DNA damage; RNA polymerase II; UV irradiation; transcription; ubiquitin; ubiquitylation.

Graphical abstract

Figure 1

RPB1 K₁₂₆₈ Is Important for UV-Induced Poly-ubiquitylation and Degradation (A) UV-induced RPB1 ubiquitylation sites (red) on the mammalian RNAPII structure (Bernecky et al., 2016). (B) Schematic of the RPB1 switchover system. (C) Dsk2 pulldown-western blot analysis of cells expressing RPB1 with different K → R mutations, before and after UV irradiation (20 J/m²). K₁₃₅₀R is a CRISPR KI, matched with its own control. (D) As in (C) but in K₁₂₆₈R CRISPR KI cells. (E) As in (C) and (D), but in yeast, before and after 4-NQO treatment (10 μg/mL). (F) Western blot analysis of UV-induced RPB1 degradation after 20 J/m² UV irradiation. Switchover cells were used as outlined in Figure S1A. Total RPB1 is detected with the anti-His tag antibody. Vinculin is the loading control. (G) Western blot analysis of yeast TAP-Rpb1 degradation after treatment with 10 μg/mL of 4-NQO. Tubulin is the loading control. See also Figure S1 and Table S1.

Figure S1

K₁₂₆₈ Is a Major or Sole Signal for UV-Induced RPB1 Poly-ubiquitylation and Degradation, Related to Figure 1 (A) Experimental setup: siRNA and doxycycline treatments in K → R switchover model system cell lines. (B) Western blot showing the efficiency of the switchover model system (in this example WT switchover control – K → K), two days after transfection (day 4, see A), in whole cell extracts. Total (D8L4Y) and transgenic (His-tagged) RPB1 were detected. Vinculin is used as a loading control. (C) Colony formation assay showing the efficiency of the switchover system in supporting cell survival (in this example WT switchover control – K → K is shown). (D) Sanger sequencing traces of the genomic DNA region encoding RPB1 K₁₂₆₈ (AAG) and the corresponding K → R mutation (AGG). Parental cells (WT) and a CRISPR knock-in clone E2 are shown. (E) Western blot showing levels of RPB1 (D8L4Y antibody) on chromatin in WT cells, before and after proteasome inhibition (MG-132) and UV treatments. Cells were pre-treated with 5 μM MG-132 for 3 h, then treated with 20 J/m² UV. Extracts were prepared 3 hours after UV. (F) Sequence alignment of the RPB1 unstructured loop region across representative eukaryote species. The presence of lysine (K) corresponding to human K₁₂₆₈ is marked with arrows.

Figure 2

K₁₂₆₈ Ubiquitylation Is Required for Cell Survival upon DNA Damage and Affects Repair Kinetics (A) Representative images of colony formation assays before and after UV irradiation (5 J/m²) (switchover system). (B) Quantification of colony formation assays (n = 3) as in (A), but using K₁₂₆₈R CRISPR KI cells and CSB KO control cells. Data are presented on a log₁₀ scale, as average surviving fractions ± SD. Asterisks indicate significance of differences (comparison versus WT cells) (p < 0.05, Tukey two-way ANOVA). ns, not significant (comparison versus CSB KO cells). (C) Growth assays before (left panel) and after UV irradiation (20 J/m²) (right panel). Cell confluency was monitored every 3 h using Incucyte and the data were normalized to t = 0 for each well. Data are represented at each 3 h time point as average relative confluency of 3 biological replicates ± SD. (D) Immunoprecipitation (IP) of RPB1 from chromatin, followed by western blot for RPB1, CSB and CPSF73 (control). Cells were UV-irradiated (20 J/m², or not) and collected 45 min later for IP. (E) 4SU-slotblot showing global nascent RNA production before and after UV irradiation (10 J/m²). Cells were pulse-labeled with 4SU 15 min prior to collection. Methylene blue staining is the loading control. (F) RT-qPCR measuring nascent transcription before or after UV-irradiated (20 J/m²) at LMNB1 (60 kb), EXT1 (317 kb), and PUM1 (134 kb), using primers at their 3′ ends. Data are represented as mean ± SD, normalized to the mature GAPDH transcript, and to untreated conditions. Statistically significant differences (three biological replicates) are indicated with asterisks (p < 0.05, multiple t tests, Holm-Sidak correction). (G) Schematic illustrating the relationship between DNA damage burden (purple stars) and nascent transcription on a long gene. Restart is only detected when all lesions have been removed; 50% restart indicates that all lesions have been removed from 50% of genes in the cell population. (H) Experimental approach (top) and western blot analysis (bottom), with DRB added immediately after UV irradiation, and samples collected at the indicated time points. The abundance of S2-phosphorylated (S2_P, 3E10) RPB1, as well as CSB and histone H3 (control) in chromatin is shown. Piggybac, product of transposon insertion into the CSB locus, was used as loading/specificity control. See also Figure S2.

Figure S2

K₁₂₆₈ Ubiquitylation Is Required for Cell Survival upon DNA Damage but Not for TC-NER, Related to Figure 2 (A) Growth assays before and after UV irradiation (10 J/m² and 20 J/m²), in WT and K₁₂₆₈R switchover model systems and CSB knock-out cells. Cell growth (confluency) was monitored every 3 h using Incucyte and the data were normalized to t = 0 for each well. Data are represented at each 3h time point as average relative confluency of 3 biological replicates ± SD. Please note that normalization to t = 0 results in technical variability between samples, such as the impression that K₁₂₆₈R cells grow better than WT cells in untreated condition, however this is not significant or reproducible. (B and C) Growth assays before and after the treatment with 4-NQO (0.5 μM for 1 h) (B) or cisplatin (90 μM for 1 h) (C), in WT and K₁₂₆₈R CRISPR knock-in and CSB knock-out cells. Cell growth (confluency) was monitored every 4 h (B) or every 3 h (C) using Incucyte and the data were normalized to t = 0 for each well. Data are represented at each 3h/4h time point as average relative confluency of 3 biological replicates ± SD (D) Sensitivity of yeast cells with the genotype shown on the left, to the levels of UV irradiation shown above. (E) Recruitment of GFP-tagged CSB in either parental HEK293T cells (WT, Blue) or in cells carrying the K₁₂₆₈R RPB1 mutation (K1268R, Orange). Micro-irradiation was initiated at time t = 0, and cells were imaged every second, with intensity values binned over 5 s intervals. Graphs show mean ± SEM, n = 18 cells (6 cells from each of 3 independent experiments). (F) Representative images of either WT or K₁₂₆₈R cells before and after being subjected to micro-irradiation; white triangles indicate regions of micro-irradiation, scale bars, 8 μm. (G) Gene browser snapshots showing the location of primers (red arrows) used for measuring transcription restart on two long genes, EXT1 and PUM1. (H) A sketch depicting the time frame within which all RNAPII will stall at DNA damages, upon UV irradiation of 20 J/m². Addition of DRB in the DRB run-off experiment, blocking the new release of RNAPII into elongation, is indicated in red. (I) Immunoprecipitation (IP) of RPB1 from chromatin fractions followed by western blot for RPB1, CSB and CPSF73. WT and K₁₂₆₈R CRISPR knock-in (clone E2) cells were either untreated or UV-irradiated with 20 J/m2 and collected 45 min, 24 h and 48 h later. IP was carried out with 4H8 RPB1 antibody.

Figure 3

RPB1 Degradation Is a Major Determinant of UV-Induced Shutdown of Transcription Initiation (A) Relative abundance of K₁₂₆₈ ubiquitylation, before and at different times after UV irradiation (20 J/m²), quantified by TMT Gly-Gly IP mass spectrometry. Data are normalized to untreated controls. Different stages of the transcriptional UV-response are indicated by red and blue boxes. (B) Diagram of experimental design for TT_chem-seq analysis. (C) Browser tracks from TT_chem-seq experiment, at ZNF644, NRIP1, and TIPARP. The data are normalized to yeast spike-in. RT-qPCR primers used for validation are indicated below gene panels. (D) Metagene TT_chem-seq profiles of genes ≧100 kb, in untreated cells and after UV irradiation (20 J/m²). Data are normalized to yeast spike-in. TSS, transcription start site. (E) Graphical representation of variables used for in silico simulation of RNAPII activity. (F) Simulated RNAPII activity on a 100-kb long gene, before and after DNA damage. RNAPII degradation upon stalling was either allowed (blue panels) or not (orange panels). For all parameter values used, refer to Table S3. See also Figure S3.

Figure S3

RPB1 Degradation Is a Major Determinant of UV-Induced Shutdown of Transcription Initiation, Related to Figure 3 (A) Abundance of all detected ubiquitylation sites in the proteome of WT cells, at different times after UV irradiation (20 J/m²). Each ubiquitylation site is represented as one gray line connecting different time points. K₁₂₆₈ ubiquitylation is marked as a red line. Also see Table S2. (B) RPB1 poly-ubiquitylation is inhibited by the NEDDylation inhibitor MLN4924, showing that it requires a cullin E3 ligase (C) K₁₂₆₈ ubiquitylation in WT and CSA knock-out cells, before, and at different times after UV irradiation (20 J/m²), quantified by TMT Gly-Gly IP mass-spectrometry, normalized to untreated condition. WT only is also shown in Figure 3A. Note that CSA KO cells have normal K₁₂₆₈R ubiquitylation at the earliest time-point, strongly indicating that CUL4^CSA plays no direct role in it. However, CSA KO cells have defective transcription after UV, potentially explaining why the CSA KO cells have decreased K₁₂₆₈ ubiquitylation at this time-point: only transcribing RNAPII is ubiquitylated (Anindya et al., 2007). (D) Browser tracks from the TT_chem-seq experiment, KITLG and FOXO1 genes. The data are normalized to yeast spike-in. (E) RT-qPCR measuring nascent RNA production at TSS-proximal regions of ZNF644 and NRIP1 genes. WT and K₁₂₆₈R CRISPR knock-in cells were either untreated, or collected 3 h post UV irradiation (20 J/m²). Primer locations are indicated in Figure 3C. Data are normalized to the mature GAPDH expression, and to untreated condition for each cell line. Representative experiments of three biological replicates are shown, with data represented as mean ± SD. Statistically significant differences (p < 0.05, multiple t tests, Holm-Sidak correction) in all three biological replicates are indicated with asterisks. (F) Western blot analysis of chromatin fractions assessing UV-induced RPB1 degradation after 20 J/m² UV irradiation. RNAPII half-life was estimated to be ∼1.5 h in WT cells. (G) Outline of two different scenarios for RNAPII fate at DNA damage when it cannot be degraded (K₁₂₆₈R mutant cells), with the predicted transcription activity profiles on the right – in the case where RNAPII dissociation at DNA damage does not take place (upper) or where it does (lower).

Figure 4

K₁₂₆₈ Ubiquitylation Ensures that Short Genes Also Cease Expression upon UV Irradiation (A) Simulated RNAPII activity on a long, medium, and short gene, with or without DNA damage. Three genes are competing for the same pool of RNAPII molecules. The initiation probability was weighted by the relative representation of long (>100 kb), medium (30–100 kb), and short (<30 kb) genes in the genome (0.1: 0.2: 0.7, respectively). RNAPII degradation upon stalling was either allowed (blue) or not (orange). (B) Browser tracks of TT_chem-seq data, from a long (EXT1) and two short genes (TMSB10 and FOS). The data are normalized to yeast spike-in. RT-qPCR primers used for validation are indicated below gene panels. (C) Metagene TT_chem-seq profiles of all genes in the genome, stratified by gene length (indicated in bold on the right). x axis: relative scale (TSS and TTS are indicated); y axis: reads per million mapped reads (rpm). Transcription levels in untreated cells (gray lines), and 45 min (light-colored lines) and 3 h (dark-colored lines) after UV irradiation (20 J/m²) are shown. The data are normalized to yeast spike-in. (D) Nascent RNA production after UV irradiation (20 J/m²) at TSS-proximal regions of EXT1, TMSB10, and FOS genes. RT-qPCR primer positions are indicated in (B). Data are represented as mean ± SD and normalized to the mature GAPDH transcript and to untreated conditions. (E) Simulation-predicted number of mRNA transcripts in a long, medium, and short gene, in untreated cells and 4 h post-damage, in scenarios where RNAPII degradation is allowed (WT equivalent) or not (K₁₂₆₈R equivalent). Parameter values as in (A). See also Figure S4 and Table S3.

Figure S4

K1268 Ubiquitylation Prevents Short Genes from Escaping the UV-Induced Transcription Shutdown, Related to Figure 4 (A) Browser tracks of the TT_chem-seq experiment, showing a long (PTEN), a medium (TIMP3) and two short genes (RGS16 and FOSB), before and 45 min or 3 h after UV irradiation. The data are normalized to yeast spike-in. (B) Scatter-density plots showing the genes that are differentially expressed between K₁₂₆₈R and WT cells (TT_chem-seq data) stratified by gene length, at different times after UV irradiation (20 J/m²). Each gene is represented by one dot. Plots are colored by binned spot density from low (blue) to high (red).

Figure 5

K₁₂₆₈ Ubiquitylation Is Required to Prevent Long-Term Transcriptional Defects upon Acute Exposure to UV (A) Experimental design. (B) Heatmap, showing expression over time, of the top 50 down- and upregulated genes in K₁₂₆₈R cells 8 h after UV irradiation (20 J/m²). Each column within a treatment group represents a biological replicate (r1, r2, r3), and each row represents one gene. Gene lengths are shown on the right, in shades of gray. Asterisks indicate genes mis-annotated as short, but confirmed by manual inspection to be long. (C) Browser tracks from mRNA-seq experiment at four short genes. (D) Bar plots of genes differentially expressed between K₁₂₆₈R and WT cells (logFC >1, false discovery rate [FDR] <0.01), for short, medium, and long genes, at different times after UV irradiation (20 J/m²). Positive side of y axis: upregulated genes; negative side: downregulated genes. (E) Western blot showing proteins encoded by short, immediate-early genes, at different time points after UV irradiation (20 J/m²). (F) Model depicting how DNA damage and RNAPII (red sphere) levels influence transcription and mRNA level (green lines) after UV irradiation, in the first stage (45 min equivalent, middle panel) and second stage (3 h + equivalent, bottom panel) of the transcription shutdown. See also Figure S5.

Figure S5

K1268 Ubiquitylation Is Required to Prevent Long-Term Transcriptional Defects upon Acute Exposure to UV, Related to Figure 5 (A) Number of differentially expressed genes (mRNA-seq) between K₁₂₆₈R and WT cells, at different time points after UV irradiation (20 J/m²). Black bars: upregulated genes; gray bars: downregulated genes. (B) Gene set enrichment analysis showing the enrichment of short (top two panels) and medium to long genes (bottom panels) in the differentially expressed gene datasets (24 h) between K₁₂₆₈R versus WT. (C) Browser tracks of the RNA-seq experiment, showing the expression of two short genes (EGR1 and ATF3). (D) RT-qPCR, measuring the abundance of mature, poly-adenylated transcripts of four short genes (FOS, FOSB, EGR1 and ATF3), in WT and K₁₂₆₈R cells, at different times after UV irradiation (20 J/m²). Data are normalized to GAPDH and untreated condition. A representative experiment of three biological replicates is shown; data are represented as mean ± SD. Asterisks indicate statistically significant differences in all three biological replicates (p < 0.01, multiple t tests, Holm-Sidak correction). (E) Scatter-density plots showing the UV-regulated genes in the mRNA-seq data (differentially expressed genes between each UV-treated condition and untreated condition, logFC > 1, FDR < 0.01), for K₁₂₆₈R and WT cells separately. The total number of differentially expressed genes (n) in each condition is indicated on top of the plots. Genes were stratified by gene length (short: < 30 kb; medium: 30-100 kb; long: > 100 kb), and each gene is represented by one dot. Plots are colored by binned spot density from low (blue) to high (red).

Figure 6

RPB1 Stability Determines Transcription Recovery upon UV Irradiation (A) Western blot showing the total levels of RPB1 before and after UV irradiation (10 J/m²). (B) Diagram of cell lines. (C) As in (A), but using the cell lines from (B) and testing later time points. (D) RT-qPCR measuring nascent RNA production at the end of the long EXT1 and PUM1 genes at different times after different UV doses. Data are normalized to the expression of mature GAPDH transcript, and to untreated conditions, and represented as mean ± SD. Statistically significant differences (p < 0.05, multiple t tests, Holm-Sidak correction) in all three biological replicates are indicated with asterisks. (E) Browser tracks from TT_chem-seq experiments. The data are normalized to yeast spike-in. (F) Metagene TT_chem-seq profiles. Data are normalized to yeast spike-in. See also Figure S6.

Figure S6

Other Effects of K₁₂₆₈R Mutation, Related to Figure 6 and Discussion (A) Growth assays before and after UV irradiation (10 J/m²), in switchover model cell lines represented in Figure 6B. Cell growth (confluency) was monitored every 3 h after UV irradiation using Incucyte and the data were normalized to t = 0 for each well. Data are represented at each 3h time point as average relative confluency of 3 biological replicates ± SD (B) As in Figure 6C, but with 5 J/m² UV irradiation. (C) Alternative splicing differences between K₁₂₆₈R and WT cells, at different time points after UV irradiation (20 J/m²), detected in the mRNA-seq data. Pie-chart categories show the proportions of different classes of alternative splicing events. The size of the pie-charts is proportional to the total number of differences (n, indicated on the right). (D) Enrichment of differential splicing events (K₁₂₆₈R versus WT) at different time points after UV irradiation. Enrichment was calculated by comparing the proportion of each class of events in the given UV-treated condition, to the proportion of the same class in untreated condition. (E and F) Browser tracks of the RNA-seq experiment, showing the examples of three genes (ARL5A, CHMP2B and DHPS) with alternative splicing events induced by UV irradiation preferentially in K₁₂₆₈R cells. (G) RT-qPCR measuring the abundance of alternatively spliced poly-adenylated transcripts in WT and K₁₂₆₈R cells, in untreated condition and 24 h after UV irradiation (20 J/m²). The data were normalized to the expression of the mature GAPDH transcript and untreated condition. A representative experiment of three biological replicates is shown, data are represented as mean ± SD. Asterisks indicate statistically significant differences in all three biological replicates (p < 0.01, multiple t tests, Holm-Sidak correction). (H) Analysis of transcription readthrough beyond the TTSs. Ratios of read-counts of the 4kb region downstream of the TTS and the terminal exon of all protein coding and RNA genes, derived from TT_chem-seq experiment, are plotted for WT and K₁₂₆₈R cells, in untreated conditions. (I) Immunofluorescence detection of CPDs in WT HEK293 cells, 3 h and 24 h after exposure to 15 J/m² of UV irradiation.

Figure 7

A Model for Transcription and Its Global Control by the Free RNAPII Pool Simplified model, illustrating how RNAPII levels in WT, K₁₂₆₈R, and Cockayne syndrome cells regulate the global transcriptional response to UV irradiation. mRNA, green lines; DNA lesions, purple stars. Top panels denote transcriptional shutdown, bottom panels restart. Repair of transcription-blocking DNA lesions is slower in K₁₂₆₈R cells, and little TC-NER occurs in CSB cells, partly because CSB protein is required for it, but also because TC-NER requires RNAPII, which is depleted in these cells. Combined with other mechanisms (e.g., Epanchintsev et al., 2017), such depletion means that transcription shutdown is more rapid in CSB cells (shown as thin green lines).