RNA polymerase II degradation triggered by DNA repair occurs in trans and independently of how the lesion is recognized

Abstract

In response to DNA damage, RPB1, the catalytic subunit of RNA Polymerase II (RNAPII), is degraded by the ubiquitin–proteasome system. Degradation models only consider transcriptionally engaged molecules, where a stalled RNAPII complex functions as a lesion-recognition factor, and its RPB1 subunit is proposed to be subsequently degraded to facilitate access of lesion-processing nucleotide excision repair (NER) factors. This transcription-coupled repair is complemented by the global genome repair (GG-NER) system, where lesions are recognized by the XPC and DDB2 factors. Here, we show that RPB1 degradation is controlled in trans by a pathway that depends on lesion processing by NER, irrespectively of whether the lesion is recognized by RNAPII itself or by XPC–DDB2. Incomplete repair due to absence of lesion-processing factors (XPA, XPB, XPD, XPF, or XPG) enhances RPB1 degradation, indicating that the signal controlling RPB1 abundance is started by lesion recognition and continues until DNA repair is completed. Consistent with an in trans mechanism, damage-induced RPB1 degradation is not restricted to active nor phosphorylated RPB1 molecules and depends on Cullin-RING ubiquitin ligases. These findings uncover a repair-dependent mechanism controlling RPB1 levels and provide a rationale for the control of gene expression under stress, where more damage implies more repair and less RPB1 levels, hence restricting RNAPII activity.

Graphical Abstract

Graphical Abstract.

Introduction

Gene expression and lesion repair, two of the major DNA-based molecular mechanisms in a cell, necessarily interact with each other. This idea is well illustrated by the transcription-coupled nucleotide excision repair system (TC-NER), where lesion-stalled RNA polymerases initiate DNA repair [1–3]. At the time of acute DNA damage induction, as when exposing cells to ultraviolet (UV) light, every active RNAPII complex not only transcribes a nascent RNA but also serves as a DNA lesion scanner. As a direct consequence, repair of template strands in transcriptionally active genes is faster than in the rest of the genome [4–6]. Conversely, lesions in coding strands and transcriptionally silent regions are recognized by the xeroderma pigmentosum (XP) factors XPC and DDB2, also known as XPE, which act as lesion-recognition factors in the other branch of NER, the global genome NER (GG-NER) [7]. After lesion recognition by either RNAPII stalling or XPC–DDB2, both pathways converge to favour the recruitment of NER factors involved in lesion processing. These factors include XPB and XPD helicases, part of the general transcription factor TFIIH, that together with XPA scan the DNA to verify the presence of a lesion. In such a case, recruitment of the endonucleases ERCC1-XPF and XPG to excise the damaged strand is favoured, and the single-stranded DNA gap is filled by DNA synthesis and ligation [8, 9].

RPB1, the major and catalytic subunit of RNAPII, is not only unique because of its repetitive carboxy-terminal domain (CTD), which is subject to multiple regulatory post-translational modifications, but also because it is specifically degraded in response to different DNA-damaging agents. As an example, exposure to UV light induces the ubiquitin-dependent proteasomal degradation of RPB1 [10], as well as deep changes in gene expression [11–13]. RPB1 ubiquitylation in response to DNA damage has been addressed intensively, particularly in the context of RNAPII persistent staling and TC-NER, where different E3 ubiquitin ligase complexes, particularly from the Cullin-RING ubiquitin ligase (CRL) family, have been shown to modify RPB1 [6, 14–17]. Nonetheless, the widely accepted view is that upon persistent RNAPII staling, the Cockayne Syndrome B (CSB) factor recruits the E3 complex CRL4^CSA that ubiquitylates RPB1 [6, 14, 15, 18], affecting RPB1 eviction [19, 20], degradation [14, 18, 19, 21], and DNA repair [6]. However, while ubiquitylation is known to precede degradation, mechanistic details concerning RPB1 ubiquitylation and degradation, such as specific lysine residues, types of ubiquitin chains, and factors involved at different time points after damage, awaits for determination. Therefore, it is not surprising that ELOF1, a TC-NER-specific factor, greatly affects ubiquitylation but modestly degradation, while STK19, another recently identified TC-NER factor, does not prominently modulate RPB1 ubiquitylation but affects degradation similarly to ELOF1 [22]. In line with this, in UVSSA-deficient cells, RPB1 ubiquitylation is reduced with no clear impact on degradation [6, 21], although recent evidence suggests the opposite [19]. For CRL4^CSA the picture is neater, since CSA KO cells displayed both reduced RPB1 ubiquitylation and degradation, although residual ubiquitylation was observed, and the activity of CRL4^CSA is likely to modulate RPB1 within the first 1–2 h after damage [6, 18]. On RPB1’s side, its K1268 residue was shown to be ubiquitylated by CSA, affecting DNA repair and gene expression upon UV treatment [6, 18]. Concerning RPB1 levels, only one of these works assessed it, showing an impairment in RPB1 degradation after UV exposure [18]. Additionally, the Hu lab recently described important but not indispensable roles for K1268 ubiquitylation in RNAPII eviction from chromatin, suggesting that other lysine residues in RPB1 are also relevant in the control of RNAPII fate under stress [20].

Although RNAPII stalling serves as a mechanism for lesion detection and its RPB1 subunit is specifically degraded, the connection between these two, damage detection and RPB1 degradation, is not obvious. The last-resort model suggests that the stalled RNAPII molecule that acts as a lesion recognition factor is the one that is actually degraded [23]. Degradation of lesion-stalled RPB1 would be needed to clear the lesion while simultaneously favouring the recruitment of lesion-processing NER factors (XPA, XPB, XPD, ERCC1-XPF, and XPG) to induce repair. Nevertheless, it is possible that the actual RBP1 molecule stalled in front of a lesion is evicted and re-used, and therefore not necessarily degraded by the proteasome. In this regard, different eviction mechanisms and their connection to RPB1 degradation are starting to be explored [19, 20]. Beyond the last-resort model and the idea of RPB1 degradation as a way to facilitate repair, promoter-proximal paused RPB1 molecules were proposed to be degraded in vicinity of the lesion, but not at the lesion [24, 25], and even without exogenous damage [26–28]. Although with different proximity to damaged DNA, these models propose degradation of phosphorylated and transcriptionally engaged RNAPII molecules, implying a direct link between an active RNAPII complex and RPB1 degradation. Nonetheless, since transcription is inhibited after damage [6, 18, 19, 29], it is unlikely that a transcription-dependent mechanism solely governs the prominent reduction in RPB1 levels.

Apart from RPB1 degradation and TC-NER-mediated repair, DNA lesions also induce GG-NER. Immediately after UV exposure, GG-NER quickly deals with low abundance 6–4 pyrimidine–pyrimidone photoproducts (6–4PPs), but slowly repairs the more abundant cyclobutane pyrimidine dimers (CPDs). In fact, a possible connection between GG-NER and RPB1 degradation has not been addressed, being currently unknown if these two events are coupled or concurrent, i.e. just happening at the same time.

Using human keratinocytes exposed to UV light and other DNA-damaging agents, here we show that lesion processing by the NER system controls RBP1 degradation. Considering that TC-NER is at its peak immediately after damage, while GG-NER is active for longer periods of time [5, 6], RPB1 levels are controlled in a sequential manner, first by TC-NER and then by GG-NER. Furthermore, incomplete repair due to the absence of any lesion processing factor (XPA, XPB, XPD, XPF, and XPG) enhanced RPB1 degradation, suggesting that signalling for in trans degradation is initiated by lesion recognition, either by TC-NER or GG-NER, and continues until DNA repair is completed. In agreement with an in trans mechanism, we present evidence suggesting that during TC-NER, a lesion-stalled RPB1 molecule can be degraded in a CSA-dependent manner, but also, since RNAPII stalling favours lesion processing, another RPB1 molecule could be targeted in trans, and independently of CSA, for degradation. Finally, we show that unphosphorylated, transcriptionally inactive RPB1 molecules can be degraded in a Cullin-RING E3 ligase-dependent manner, indicating that a pathway dependent on lesion processing by NER controls, in trans, RPB1 abundance.

Materials and methods

Cell culture

HaCaT human keratinocytes were cultured in 5% CO₂ incubators at 37°C in Dulbecco’s modified Eagle’s medium (DMEM) (Thermo Fisher Scientific #10566016) supplemented with 10% FBS (Fetal Bovine Serum, Thermo Fisher Scientific #10270098), and 1% Primocin (Invivogen #ant-pm-1). HEK293T were cultured in 5% CO₂ incubators at 37°C in DMEM supplemented with 10% FBS (Thermo Fisher Scientific #10270098) and 1% Pen-Strep (GIBCO #15140122). HaCaT cells were routinely sub-cultured two to three times a week using a 1:4 dilution ratio, while HEK293T were sub-cultured using a 1:10 dilution ratio. All cell lines were proved to be mycoplasma-free.

Drug treatments

When indicated, cells were incubated with the following drugs, which were kept in culture medium until harvesting: IlludinS (MedChemExpress #HY-125098); MG132 (Selleckchem #S2619); Spironolactone (Selleckchem #S4054); α-amanitin (Sigma #A2263); Aphidicolin (MedChemExpress #HY-N6733); ATR inhibitor VE822 (Selleckchem # S7102); ATM inhibitor KU60019 (Selleckchem #S1570); DNA-PK inhibitor NU7441 (Selleckchem #S2638); GSK3 inhibitor CHIR-99021 (Sigma); 5,6-dichlorobenzimidazole 1-β-D-ribofuranoside (DRB) (Sigma #D1916); CDK7 inhibitor THZ1 (Selleckchem #S7449); CDK9 inhibitor BAY1251152 (Selleckchem #S8730); and MLN4924 (Selleckchem #S7109).

Plasmid construction

The oligonucleotide duplex (AGCTTGCCGCCACCATGACTAGTTATCCTTACGATGTGCCAGACTACGCTAGCG; GATCCGCTAGCGTAGTCTGGCACATCGTAAGGATAACTAGTCATGGTGGCGGCA), containing the Kozak sequence and the HA epitope, was inserted into HindIII/BamHI-treated pcDNA5/FRT/TO vector (Thermo Fisher Scientific). To obtain HA-tagged RPB1-A2A5 expression vector, the 1579 bp PvuI/NheI fragment spanning a section of the ampicillin selectable marker, CMV promoter, and HA epitope coding sequence of this resulting plasmid was transferred to pEVRF1-RPB1-A2A5 [11], employing the same restriction enzymes. An equivalent vector for the expression of wild-type (WT) HA-tagged RPB1 was derived replacing the A2A5 coding region of the former with the WT CTD equivalent contained in the XhoI/XbaI 2509 bp section of pEVRF1-RPB1 [11].

Generation of stable cell lines

CRISPR-Cas9 single DDB2 KO clones of HaCaT cells were obtained as described in [12]. CRISPR-Cas9 single XPC KO clones were obtained following the protocol described in [30]. The guide RNA (gRNA) for Cas9 targets within XPC first exon were cloned into the pSpCas9(BB)-2A-GFP (PX458), and stable transfectants were selected by fluorescence-activated cell sorting (FACS). The double KO XPC–DDB2 pools were obtained by doing the XPC KO in the DDB2 KO genetic background, and stable transfectants were selected by FACS followed by puromycin selection. Single clones were obtained by clonal-density dilution. CRISPR-Cas9 KO clones for XPA were obtained in the same manner as XPC KO clones. Sequences for each specific gRNA are presented in Supplementary Table S1.

G1-arrest

HaCaT human keratinocytes were seeded at 90% confluence in DMEM supplemented with 10% FBS and 1% Primocin. Twenty-four hours later, cells were washed twice with 1× phosphate buffered saline (PBS), and the culture medium was changed to DMEM supplemented with 0.5% FBS and 1% Primocin. Cells were further incubated for 48 h to allow for G1-arrest.

UV-irradiation

HaCaT and HEK293T cells were washed once with 1× PBS before irradiation with UVC light (254 nm). UV irradiation was performed with a CL-1000 Shortwave Crosslinker (UVP) or a custom-made UV-irradiation apparatus. UV light doses were quantified with an external UVC radiometer.

siRNA knockdown experiments

HaCaT human keratinocytes at 70% confluence (FBS 10%, Primocin 1%) were transfected using RNAiMAX (Thermo Fisher Scientific #13778075) according to manufacturer’s instructions. Pre-designed human siGenome small interfering RNAs (siRNAs) pool were obtained from Dharmacon. Twenty-four hours after transfection, when cells were at 100% confluence, culture medium was changed to DMEM supplemented with 0.5% FBS and 1% Primocin, after which cells were further incubated for 48 h. After this time, cell population was G1-arrested and siRNA-transfected 72 h before damage induction.

Western blot

Protein samples were obtained by harvesting cells with Laemmli buffer 2× [Tris–HCl, pH 6.8, 125 mM, β-Me 10%, sodium dodecyl sulphate (SDS) 4%, glycerol 20%]. Samples were subjected to SDS–polyacrylamide gel electrophoresis and transferred to a nitrocellulose membrane (GE Healthcare Life Sciences #10600002). Membranes were blocked at room temperature for 1 h with 5% milk in T-TBS (1% Tween 20–Tris buffered saline buffer) and incubated overnight at 4°C with the indicated primary antibodies diluted in 5% milk in T-TBS. Membranes were washed three times with T-TBS and incubated for 1 h at room temperature with secondary antibodies. Membranes were washed again three times with T-TBS and visualized by using an Odyssey Imaging System for fluorescent secondary antibodies or incubated with a chemiluminescent substrate ECL reagent (Thermo Scientific Super Signal #34094) for HRP-conjugated secondary antibodies.

For RPB1 quantitation, we used the FIJI software to digitally quantify IIa, IIo and the area in between, in which partially modified RPB1 molecules migrate. For a list of antibodies used throughout this work please see Supplementary Table S1.

Southern blot

WT and XPC–DDB2 dKO HaCaT cells were irradiated and harvested at the indicated time points for genomic DNA extraction with a QIAGEN QIAmp DNA Mini Kit (Qiagen #51304), according to manufacturer’s instructions. DNA samples were loaded in a 0.8% agarose gel and run for 1 h at 100 V. The agarose gel was then transferred by Southern blot to a Hybond+ membrane (Amersham Hybond-XL GE Healthcare) with SSC 10× buffer (1.5 M NaCl, 0.15 M Tris-Na Citrate, pH 5), and DNA was crosslinked by baking the membrane for 2 h at 80°C. The membrane was then blocked for 1 h at room temperature with 5% milk in T-PBS (1% Tween 20–PBS 1X), washed once with T-PBS, and incubated with primary antibodies at 4°C overnight. For a list of antibodies used throughout this work please see Supplementary Table S1. Primary antibodies were prepared in bovine serum albumin 5% T-PBS. Membranes were then washed three times with TENT buffer (20 mM Tris–HCl, pH 8, 20 mM, 137 mM NaCl, Tween 0.1%) and incubated for 1 h at room temperature with secondary antibodies. Membranes were washed again three times with TENT buffer and incubated with a chemiluminescent substrate ECL reagent for HRP-conjugated detection of secondary antibodies.

FACS analysis

WT and XPC–DDB2 dKO HaCaT cells were harvested by gently scrapping the plates in PBS 1× and centrifuged at 300 × g for 5 min. PBS 1× was aspirated, and cells were resuspended in ethanol 70% by up-and-down pipetting to avoid cell aggregation. Cells were then incubated for 2 h at 4°C for fixation, followed by a centrifugation step of 5 min at 300 × g. The remaining ethanol was washed once with PBS 1×, and cell pellets were resuspended in 1 ml of propidium iodide solution (propidium iodide 120 mM, 0.1% Triton X-100, 100 ng/µl RNaseA, PBS 1×) before incubation for 30 min at room temperature. Samples were then vortexed, and cell fluorescence was measured with a FACS Aria II flow cytometer. Data were analyzed using FlowJo software.

Statistical analysis

Western blot quantification data were analysed using GraphPad Prism 10 (v.10.2.1). Two-way analysis of variance (ANOVA) was performed, followed by Bonferroni-corrected multiple comparisons test using the single pooled variance from the ANOVA model. The family of tests comprised the contrasts displayed in each graph, and multiplicity-adjusted P values are reported. All tests were two-sided with α = 0.05. Sample sizes (n) are stated in each figure legend, and error bars represent mean ± standard error of the mean (SEM) from n independent biological replicates. Significance is represented by: *P < .05; **P < .01; ***P < .001; ****P < .0001; ns, non-significant.

Results

Given the central role of RNAPII in gene expression, it is expected that RPB1 abundance is of utmost importance for proper gene expression and cell function [31]. In this sense, datasets regarding variations in gene copy number and human disease prevalence have been published [32]. To analyse the impact of RPB1 levels in human disease, we used these datasets to study the effect of increased or decreased dosage of POLR2A, the gene encoding RPB1. As expected, results in Supplementary Fig. S1A showed that alterations in the amount of POLR2A are highly associated with the onset of pathologies. Nevertheless, and despite its relevance, mechanisms controlling RPB1 abundance are still being uncovered.

TC-NER controls RPB1 levels in human keratinocytes

Prominent variations in RPB1 levels have been historically observed under an acute response to DNA damage, where RPB1 abundance decreases due to specific proteasome-dependent degradation. To better understand the mechanisms controlling RPB1 levels, and having in mind that current degradation models rely on phosphorylated and transcriptionally engaged RNAPII molecules, we decided to force lesion recognition by active RNAPII complexes by inactivating GG-NER. To this end, we generated an XPC and DDB2 double-knockout (dKO) human keratinocyte cell line (Supplementary Fig. S1B), deficient in damage recognition and repair by GG-NER (Supplementary Fig. S1C). Keratinocytes, the most common cell type in human skin, are naturally exposed to UV light and can be easily arrested in G1 by serum withdrawal (Supplementary Fig. S1D). To avoid excessive cell death and activation of pathways due to conflicts between DNA and RNA polymerases in S-phase, we based our experimental set up in G1-arrested human keratinocytes (HaCaT cells) [33] exposed to UV light and other DNA-damaging agents. As expected, UV exposure of G1-arrested WT HaCaT cells induced the proteasome-dependent degradation of RPB1 (Supplementary Fig. S1E). Of note, RPB1 degradation in UV-treated keratinocytes was barely affected by inhibiting classical DNA damage response kinases, such as ATM, ATR, DNA-PK, and GSK3 (Supplementary Fig. S1F and G).

To confirm the widespread idea that TC-NER can induce RPB1 degradation, and also that GG-NER-deficient cells are TC-NER proficient [34], we analysed RPB1 levels in whole-cell extracts from both WT and XPC–DDB2 dKO G1-arrested HaCaT cells. As expected, results in Fig. 1A showed a comparable reduction in total RPB1 levels (unphosphorylated IIa and phosphorylated IIo isoforms) shortly after UV irradiation (3 h) of WT and XPC–DDB2 dKO cells. To confirm these results, we treated WT and XPC–DDB2 dKO cells with illudinS, a DNA-alkylating agent that, unlike UV-induced CPDs and 6–4PPs, can be recognized exclusively by TC-NER [35, 36]. Results in Fig. 1B and Supplementary Fig. S1H show that exposure of WT and XPC–DDB2 dKO cells to illudinS induced a comparable decrease in RPB1 levels in both cell types, further suggesting that TC-NER can induce RPB1 degradation and also that GG-NER deficient cells are TC-NER proficient. In line with these ideas, and within the time frame of 3 h after UV irradiation, RPB1 degradation in HaCaT cells was reduced by transfection of siRNAs directed against CSA and CSB, two major TC-NER factors (Fig. 1C and Supplementary Fig. S1I). Reasonably, CSA knockdown by siRNA transfection showed no effect in RPB1 degradation when cells were pre-treated with 5,6-dichloro-1-β-D-ribofuranosylbenzimidazole (DRB), a transcriptional inhibitor (Supplementary Fig. S1J).

Figure 1.

TC-NER controls RPB1 levels in human keratinocytes. (A) G1-arrested WT and XPC–DDB2 double KO (dKO XPC–DDB2) HaCaT human keratinocytes were irradiated with the indicated doses of UV light and harvested 3 h after. Relative RPB1 abundance was assessed by western blot using antibodies against the N-terminal domain (NTD) of RPB1 and Vinculin as a loading control. IIo and IIa indicate the phosphorylated and unphosphorylated forms of RPB1. Images of a representative experiment and mean ± SEM of RPB1 abundance relative to 0 J/m² condition are shown (n = 5 independent biological replicates). Two-way ANOVA. ****P < .0001; ns, non-significant. (B) G1-arrested WT and XPC–DDB2 double KO (dKO XPC–DDB2) HaCaT human keratinocytes were treated with the indicated doses of illudinS for 3 h. Relative RPB1 abundance was assessed as before. IIo and IIa indicate the phosphorylated and unphosphorylated forms of RPB1. Images of a representative experiment and mean ± SEM of RPB1 abundance relative to 0 nM illudinS condition are shown (n = 3 independent biological replicates). Two-way ANOVA. **P < .01; ns, non-significant. (C) WT HaCaT human keratinocytes were transfected with scramble siRNAs (siCtrl) or siRNAs directed against CSA (siCSA) or CSB (siCSB). 24 h later, cells were arrested in G1 by serum withdrawal for 48 h and then irradiated with the indicated doses of UV light. Cells were finally harvested 3 h after irradiation. IIo and IIa indicate the phosphorylated and unphosphorylated forms of RPB1. Images of a representative experiment and mean ± SEM of RPB1 abundance relative to 0 J/m² condition are shown (n = 3 independent biological replicates). Two-way ANOVA. Significance is denoted by:*P < .05; **P < .01; ***P < .001; ****P < .0001; ns, non-significant.

GG-NER controls RPB1 levels in human keratinocytes

Given that TC-NER induces not only RPB1 degradation but also lesion processing, we wondered whether total NER activity, or just TC-NER, could modulate RPB1 levels. To analyse a possible role for GG-NER in the control of RPB1 levels, we recognized that any TC-NER contribution should occur shortly after UV irradiation, when TC-NER activity accounts for most of NER-mediated DNA repair. Moreover, at longer time points after damage induction, when TC-NER activity is barely detectable [5, 6], GG-NER-deficient cells will lack any NER-mediated repair, providing an opportunity to assess a possible role for global DNA repair in the control of RPB1 abundance. To this end, we measured total RPB1 levels in GG-NER proficient (WT) and deficient (XPC–DDB2 dKO) cells 12 h after UV irradiation, where TC-NER is not mainly active. Results in Fig. 2A show higher levels of RPB1 in GG-NER-deficient than in WT cells, pointing towards a role for NER-mediated lesion processing in the control of RPB1 levels. Comparable results were obtained analysing other related timepoints after UV (Supplementary Fig. S2A), using single XPC or DDB2 knockout cells (Supplementary Figs S1B and S2B), or XPC siRNA-mediated knockdown in WT cells (Supplementary Figs S2C and D). Moreover, in contrast to what was observed at 3 h, RPB1 degradation at 12 h after UV was CSA and CSB independent (Fig. 2B).

Figure 2.

GG-NER controls RPB1 levels in human keratinocytes. (A) G1-arrested WT and XPC–DDB2 double KO (dKO XPC–DDB2) HaCaT human keratinocytes were irradiated with the indicated doses of UV light and harvested 12 h after. Relative RPB1 abundance was assessed by western blot using antibodies against the NTD of RPB1 and Vinculin as a loading control. IIo and IIa indicate the phosphorylated and unphosphorylated forms of RPB1. Images of a representative experiment and mean ± SEM of RPB1 abundance relative to 0 J/m² condition are shown (n = 5 independent biological replicates). Two-way ANOVA. ***P < .001; ****P < .0001; ns, non-significant. (B) WT HaCaT human keratinocytes were transfected with scramble siRNAs (siCtrl) or siRNAs directed against CSA (siCSA) or CSB (siCSB). Twenty-four hours later, cells were arrested in G1 by serum withdrawal for 48 h and then irradiated with the indicated doses of UV light. Cells were finally harvested 12 h after irradiation. Relative RPB1 abundance was assessed as before. IIo and IIa indicate the phosphorylated and unphosphorylated forms of RPB1. Images of a representative experiment and mean ± SEM of RPB1 abundance relative to 0 J/m² condition are shown (n = 3 independent biological replicates). Two-way ANOVA.**P < .01; ***P < .001; ****P < .0001; ns, non-significant.

These results suggest that NER-dependent lesion processing controls RPB1 abundance in G1-arrested human keratinocytes. Moreover, since TC-NER activity is maximum immediately after UV exposure, while GG-NER deals with UV-induced DNA damage for much longer, these results favour a model of sequential control of RPB1 levels, first by TC-NER and then GG-NER.

Lesion processing by NER controls RPB1 degradation in trans

The fact that there are no RPB1 molecules directly involved in GG-NER suggests the presence of a repair-dependent pathway controlling, in trans, the levels of RPB1. Results in Figs 1 and 2 suggest that NER-dependent lesion processing controls, by an unknown mechanism, RPB1 abundance. To deepen into a possible role for DNA repair in the control of RPB1 levels, we next generated an XPA knockout HaCaT cell line (XPA KO) (Supplementary Fig. S3A), in which damaged DNA is recognized but the repair process cannot be completed. In comparison to WT keratinocytes, XPA KO HaCaT cells showed lower levels of RPB1 in response to UV exposure (Fig. 3A). As this reduction proved to be proteasome-dependent (Supplementary Fig. S3B), these results suggest that lesion processing by the NER system activates a pathway that, in turn, favours RPB1 ubiquitylation and its subsequent proteasomal degradation. To rule out specific functions of XPA, we next treated WT HaCaT cells with spironolactone (SP), a drug that induces degradation of XPB [37], one of the DNA helicases operating in the NER system. In agreement with a role for lesion processing in the control of RPB1 abundance, results in Fig. 3B and Supplementary Fig. S3C and D showed that treatment with SP enhanced the UV-induced degradation of RPB1.

Figure 3.

NER-mediated DNA repair controls RPB1 degradation in trans. (A) G1-arrested WT and XPA KO HaCaT human keratinocytes were irradiated with the indicated doses of UV light and harvested 3 or 12 h after. Relative RPB1 abundance was assessed by western blot using antibodies against the NTD of RPB1 and Vinculin as a loading control. IIo and IIa indicate the phosphorylated and unphosphorylated forms of RPB1. Images of a representative experiment and mean ± SEM of RPB1 abundance relative to 0 J/m² condition are shown (n = 4 independent biological replicates). Two-way ANOVA.*P < .05; **P < .01; ****P < .0001. (B) G1-arrested WT HaCaT human keratinocytes were incubated for 1 h with spironolactone (SP, 2.5 µM) to induce XPB knockdown. Cells were then irradiated with the indicated doses of UV light and harvested 3 or 12 h after. Relative RPB1 abundance was assessed as before. IIo and IIa indicate the phosphorylated and unphosphorylated forms of RPB1. Images of a representative experiment and mean ± SEM of RPB1 abundance relative to 0 J/m² condition are shown (n = 6 independent biological replicates). Two-way ANOVA. *P < .05; **P < .01; ****P < .0001. (C) HEK293T human cells were transfected with a plasmid encoding an HA-tagged and α-amanitin-resistant version of RPB1 (left) or with a control GFP plasmid (right). Twenty-four hours after transfection, cells were treated with the indicated doses of α-amanitin and harvested 6, 12, or 24 h later. Relative levels of the HA-tagged and α-amanitin-resistant version of RPB1 were assessed by western blot using antibodies against the HA tag (left), the endogenous RPB1 (NTD antibody, right), or Vinculin as a loading control. IIo and IIa indicate the phosphorylated and unphosphorylated forms of RPB1, whether exogenous (HA-tagged) or endogenous. Images of a representative experiment and mean ± SEM of RPB1 abundance relative to 0 ng/µl of α-amanitin conditions are shown (Exogenous HA-RPB1, left panel n = 4 independent biological replicates; Endogenous RPB1, right panel n = 3 independent biological replicates). Two-way ANOVA. *P < .05; **P < .01; ****P < .0001; ns, non-significant.

Next, we inhibited the gap-filling step of the NER reaction by treating G1-arrested WT HaCaT cells with the DNA polymerase inhibitor aphidicolin. In line with results in XPA- and XPB-deficient HaCaT cells, results in Supplementary Fig. S3E showed that 3 and 12 h after adding aphidicolin, RPB1 levels were further reduced in cells where repair is started but cannot be completed, in this case by a compromised gap-filling step. RPB1 degradation was also enhanced in WT HaCaT cells transfected with anti XPA, XPD, XPF, or XPG siRNAs at 3 and 12 h after UV (Supplementary Fig. S3F and G). Altogether, these results suggest that DNA repair by the NER system controls RPB1 levels in human keratinocytes, where reduced lesion recognition implies less RPB1 degradation (Fig. 2) and an incomplete processing of the lesion results in enhanced degradation levels (Fig. 3). Moreover, impairment in lesion processing by interfering with XPA, XPB, XPD, XPF, XPG, or the gap-filling reaction consistently induced an enhancement in RPB1 degradation at both time points analysed (3 and 12 h). Remarkably, the enhancement observed at short time points (3 h) suggests that when lesion recognition is performed by TC-NER, and besides from CRL4^CSA ubiquitylation/degradation in cis, subsequent lesion processing by NER factors may induce, in trans, degradation of a distinct RPB1 molecule. To strengthen this idea, we performed two different experiments. First, lesion processing-defective cells were exposed to illudinS, a TC-NER-specific drug. In comparison to WT HaCaT cells, we observed that RPB1 degradation was enhanced in XPA KO cells (Supplementary Fig. S3H), as well as upon siRNA-mediated knockdown of XPA, XPD, and XPF (Supplementary Fig. S3I). Since a lesion-stalled RPB1 molecule can be degraded only once, this enhancement may involve other molecules whose degradation shall occur in trans to the lesion. In the second experiment, we aimed at distinguishing RPB1 molecules involved in either lesion recognition by TC-NER or targeted for degradation. To this end, we considered α-amanitin, a drug that impairs ribonucleotide incorporation by sensitive RPB1 molecules, inducing RNAPII stalling, RPB1 degradation, and changes in gene expression patterns, similarly to UV irradiation [38–40]. α-Amanitin resistance can be achieved by a point mutation in RPB1’s coding sequence (N792D) and, therefore, this drug has been widely used to replace the endogenous RPB1 pool with an exogenous and resistant version [41]. Cells transfected with α-amanitin-resistant RPB1 expression plasmids have, initially, two populations of RPB1 molecules: α-amanitin sensitive, encoded in the endogenous loci, and α-amanitin resistant exogenous molecules. After adding α-amanitin to the culture medium, gene expression is typically assessed 24–48 h later, when the endogenous RPB1 pool is barely detectable and the exogenous version is evident. In the context of an in trans degradation pathway, we hypothesized that shortly after adding α-amanitin, a sensitive and stalled RPB1 molecule might induce, through a TC-NER reaction, in trans degradation of either sensitive or resistant molecules. To test this idea, we transfected HEK293T cells with an α-amanitin-resistant RPB1 expression plasmid and then analysed the levels of both α-amanitin sensitive and resistant pools in response to α-amanitin treatment. In line with a NER-dependent in trans degradation mechanism, results in Fig. 3C and S3J show that the resistant pool is in fact partially and initially sensitive to α-amanitin but later, and in agreement with the many studies that took advantage of α-amanitin to exchange different versions of RPB1, the resistant pool is stabilized while the sensitive pool is further degraded. In this matter, we hypothesize that after adding the drug, the levels of the sensitive pool, able to target sensitive or resistant RPB1 molecules for degradation, will decrease with time. Conversely, as sensitive molecules progressively decrease, the α-amanitin resistant pool stabilizes, reaching a maximum that will ultimately depend on de novo expression of sensitive RPB1 molecules encoded in the endogenous loci.

Transcriptionally inactive RPB1 molecules can be targeted for degradation

Our evidence suggests that lesion processing, initiated by both TC- and GG-NER controls, in trans, RPB1 levels. We next wondered about the nature of the targeted RPB1 molecules, part of either transcriptionally active or inactive RNAPII complexes. To this end, we first used DRB, a CTD’s kinase inhibitor that favours the unphosphorylated and transcriptionally inactive isoform (RNAPIIa) over the phosphorylated isoform (RNAPIIo), therefore inhibiting transcription [42–44]. Results in Fig. 4A show that even when RNAPIIa greatly outnumbers RNAPIIo, a clear reduction in RPB1 levels is still observed upon UV irradiation, suggesting that transcriptionally inactive RPB1 molecules can be targeted for proteasomal degradation. Next, we used specific inhibitors for two of the main CTD kinases involved in transcriptional regulation, cyclin-dependent kinases 7 and 9 (CDK7 and CDK9). CDK7, part of TFIIH, phosphorylates RPB1’s CTD residues at positions serine 5 and favours transcriptional initiation, while CDK9, part of the pTEFb complex, phosphorylates serines at position 2, a hallmark of productive elongation [45]. UV irradiation of WT HaCaT cells treated with CDK7 inhibitor THZ1 or CDK9 inhibitor BAY1251152 showed similar results to those obtained with DRB: a clear reduction in RPB1 levels in response to UV exposure (Fig. 4B). To further confirm that RPB1’s phosphorylation state and transcriptional activity are not a prerequisite for degradation, we took advantage of a mutant version of RPB1 that cannot be phosphorylated at serines 2 and 5 because of alanine replacement (A2A5, alanine 2 and 5 mutant polymerase) [11]. Results in Fig. 4C and Supplementary Fig. S4A showed that A2A5 mutant RPB1 molecules are also degraded upon UV exposure, favouring the notion that transcriptionally inactive RPB1 molecules can be targeted for degradation. In this sense, the decrease in the unphosphorylated RNAPIIa isoform band in UV-treated cells was thus far interpreted as a conversion into the phosphorylated RNAPIIo isoform, which would later be degraded in cis to the lesion. In view of the results shown here, the observed disappearance of the RNAPIIa band may also reflect its degradation. Nevertheless, we must be careful in associating RNAPIIa or RNAPIIo bands disappearance solely to degradation, since IIa–IIo interconversion is affected by a plethora of factors.

Figure 4.

Transcriptionally inactive RPB1 molecules can be targeted for degradation. (A) G1-arrested WT human keratinocytes were incubated for 3 h with the kinase inhibitor DRB (100 µM) and then irradiated with the indicated doses of UV light. Cells were harvested 10 h after irradiation. Relative RPB1 abundance was assessed by western blot using antibodies against the NTD of RPB1 and Vinculin as a loading control. IIo and IIa indicate the phosphorylated and unphosphorylated forms of RPB1. Images of a representative experiment and mean ± SEM of RPB1 abundance relative to 0 J/m² condition are shown (n = 3 independent biological replicates). Two-way ANOVA.**P < .01; ***P < .001; ****P < .0001. (B) G1-arrested WT human keratinocytes were irradiated with 40 J/m² of UV light. Immediately after irradiation, the CDK7 inhibitor THZ1 (100 nM) or the CDK9 inhibitor BAY 1251152 (400 nM) was added, and cells were harvested 12 h after. Relative RPB1 abundance was assessed as before. IIo and IIa indicate the phosphorylated and unphosphorylated forms of RPB1. Images of a representative experiment and mean ± SEM of RPB1 abundance relative to 0 J/m² condition are shown (n = 3 independent biological replicates). Two-way ANOVA. ***P < .001; ****P < .0001. (C) HEK293T human cells were transfected with a plasmid encoding an HA-tagged version of RPB1 that cannot be phosphorylated at serines 2 and 5 because of alanine replacement (A2A5-RPB1). Forty-eight hours after transfection, cells were irradiated with the indicated doses of UV light and harvested after 6 or 12 h. Relative levels of HA-tagged A2A5-RPB1 were assessed by western blot using antibodies against the HA tag and Vinculin as a loading control. Images of a representative experiment and mean ± SEM of HA-tagged A2A5-RPB1 abundance relative to 0 J/m² condition are shown (n = 4 independent biological replicates). Two-way ANOVA.**P < .01; ***P < .001.

NER-dependent degradation of RPB1 depends on Cullin-RING ligases

Different E3 ubiquitin ligases were shown to be involved in RPB1 proteasomal degradation. Relevant for DNA repair, the family of Cullin4 RING ubiquitin ligases (CRL4) was shown to be extensively remodelled in response to UV irradiation [46]. Results in Figs 1C and 2B showed that the E3 complex CRL4^CSA, associated with TC-NER, was relevant for RPB1 degradation shortly after UV and in a transcription-dependent manner (Supplementary Fig. S1J), but not at 12 h after UV. In relation to GG-NER, our results suggest that CRL4^DDB2 is also involved in the control of RPB1 degradation (Fig. 2A and Supplementary Fig. S2A and B), but it is not clear if CRL4^DDB2 ubiquitylates RPB1 or if, as XPC, favours lesion processing and, therefore, RPB1 degradation. The fact that DDB2 is also degraded after UV (Supplementary Fig. S5A) argues against a direct role for CRL4^DDB2 in RPB1 degradation. To gather additional evidence supporting a role for CRLs in RPB1 degradation, we used the neddylation inhibitor MLN4924, which impedes NEDD8 conjugation and CRLs activation. In agreement with CRL4^CSA activity being NEDD8 dependent, results in Fig. 5A show that in HaCaT cells exposed to the TC-NER specific drug illudinS, RPB1 degradation was inhibited. Also, degradation at 3 and 12 h after UV was sensitive to MLN4924 (Fig. 5B), as well as degradation of transcriptionally inactive molecules upon DRB and UV treatments (Fig. 5C). Altogether, these results suggest that in cis and in trans RPB1 degradation pathways rely on different CRL E3 complexes, with TC-NER’s CRL4^CSA involved in in cis RPB1 ubiquitylation and a different CRL complex/es involved in the in trans mechanism, a pathway dependent on lesion processing by NER.

Figure 5.

NER-related CRL E3 ubiquitin ligases target RPB1 degradation. (A) G1-arrested WT human keratinocytes were incubated for 1 h with the neddylation inhibitor MLN4924 (10 µM) before adding the indicated doses of illudinS. Cells were harvested after 3 or 12 h. Relative RPB1 abundance was assessed by western blot using antibodies against the NTD of RPB1 and Vinculin as a loading control. IIo and IIa indicate the phosphorylated and unphosphorylated forms of RPB1. Images of a representative experiment and mean ± SEM of RPB1 abundance relative to 0 nM illudinS condition are shown (n = 3 independent biological replicates). Two-way ANOVA. *P < .05; **P < .01; ***P < .001; ns, non-significant. (B) G1-arrested WT human keratinocytes were incubated for 1 h with the neddylation inhibitor MLN4924 (10 µM) before irradiation with the indicated doses of UV light. Cells were harvested 3 or 12 h after irradiation, and the relative RPB1 abundance was assessed as before. IIo and IIa indicate the phosphorylated and unphosphorylated forms of RPB1. Images of a representative experiment and mean ± SEM of RPB1 abundance relative to 0 J/m² condition are shown (n = 3 independent biological replicates). Two-way ANOVA. **P < .01; ***P < .001; ****P < .0001; ns, non-significant. (C) G1-arrested WT human keratinocytes were incubated with the neddylation inhibitor MLN4924 for 1 h (10 µM), with the kinase inhibitor DRB (100 µM) for 3 h, or a combination of both, before irradiation with the indicated doses of UV light. Cells were harvested 3 h (left) or 12 h (right) after irradiation. Relative RPB1 abundance was assessed as before. IIo and IIa indicate the phosphorylated and unphosphorylated forms of RPB1. Images of a representative experiment and mean ± SEM of RPB1 abundance relative to 0 J/m² condition are shown (n = 3 independent biological replicates). Two-way ANOVA. *P < .05; **P < .01; ***P < .001; ****P < .0001; ns, non-significant.

Proposed model for RPB1 degradation in response to DNA damage repair by NER

The proposed model for the control of RNAPII activity under genotoxic stress highlights the role of lesion processing factors not only in DNA repair, but also in the control of RPB1 degradation (Fig. 6). Irrespective of how lesion recognition is conveyed, either by RNAPII staling (TC-NER, left) or XPC–DDB2 (GG-NER, right), lesion processing factors not only induce DNA repair but also a CRL-dependent RPB1 ubiquitylation and degradation pathway, acting in trans to the lesion. The activity of this pathway modulates the abundance of RPB1 and RNAPII activity, but not the DNA repair process. In the case of lesion recognition by TC-NER, an in cis, CSA/CSB-dependent ubiquitylation pathway also takes place, affecting DNA repair and RPB1 degradation. The model proposes that during GG-NER, only the in trans mechanism modulates degradation of RPB1, while in the case of TC-NER, both the in cis and in trans pathways modulate RPB1 levels. In this case, a phosphorylated RPB1, acting a as a damage recognition factor in TC-NER, facilities lesion processing and, hence, in trans degradation of a different RPB1 molecule.

Figure 6.

Model for the control of RPB1 levels by the NER system. The proposed model for the control of RNAPII activity under genotoxic stress highlights the role of lesion processing factors not only in DNA repair, but also in the control of RPB1 degradation. Regardless of how lesion recognition is conveyed, either by RNAPII stalling (TC-NER, left) or XPC–DDB2 (GG-NER, right), lesion processing factors not only induce DNA repair but also a CRL-dependent RPB1 ubiquitylation and degradation pathway, acting in trans to the lesion. The activity of this pathway modulates the abundance of RPB1 and RNAPII activity, but not the DNA repair process. In the case of lesion recognition by TC-NER, an in cis, CSA/CSB-dependent ubiquitylation pathway also takes place, affecting DNA repair and RPB1 degradation.

Discussion

NER connects repair and gene expression

In this work we have identified a pathway that depends on lesion processing and controls RPB1 levels, connecting two major DNA-based events: repair and transcription. As can be seen throughout this work and elsewhere, more damage implies less RPB1 levels. Consequently, these findings provide a rationale for the control of gene expression under stress: the number of lesions determines the amount of repair that, through the control of RPB1 abundance, modulates the gene expression response.

RNAPII fate in response to damage is controlled sequentially, first by TC-NER and then by GG-NER, the latter being greatly responsible for the control of RPB1 levels in human keratinocytes exposed to UV light. In addition to the results shown here, the sequential control of RNAPII levels is supported by the fact that transcription is inhibited shortly after damage and, therefore, lesion recognition by TC-NER will decrease. Conversely, lesion recognition by GG-NER is active for longer periods of time, favouring the proposed sequential control of RPB1 abundance.

A retrospective analysis of relevant published data suggests that connection between lesion processing and the control of RPB1 levels herein proposed might have been overlooked. The seminal work by Ratner and co-workers [10] presented different clues suggesting that NER-deficient cells have altered RPB1 levels in response to UV light: (i) Results in Fig. 1 show that XPD mutant fibroblasts have reduced levels of RPB1 in comparison to WT cells, (ii) Results in Fig. 6 show the same, but this time using XPA and XPG mutant fibroblasts and, (iii) Fig. 6 also shows that XPC mutant fibroblasts present higher levels of RPB1 than its WT counterpart (please see Ratner et al.,1998) . Additionally, the Sancar laboratory showed an enhancement in template-strand repair in XPC mutant fibroblasts, a result that suits the idea of the present work, where GG-NER deficiency could lead to more RPB1, hence favouring transcription and template-strand repair [47]. Therefore, we propose there might be an enhancement of TC-NER activity in GG-NER deficient cells, not only because there is no repair by GG-NER, but also because of higher levels of RPB1 in response to any damage dealt by NER. Conversely, mutant cells in lesion processing NER factors (XPA, XPB, XPD, XPF, and XPG) will deal with reduced levels of RPB1 and lower TC-NER activity. In line with this, different reports showed altered gene expression patterns in NER-deficient cells in response to damaging agents [12, 48].

In cis versus in trans degradation of RPB1

There is a substantial difference between in cis degradation or an in trans signalling cascade controlling RPB1 levels: in cis degradation would facilitate lesion clearance and repair, while in a signalling cascade, RPB1 degradation is not related to lesion clearance nor repair, but to the overall control of the gene expression response. With regard to a chromatin-loaded and phosphorylated RPB1, it is currently unknown how and when RNAPII eviction precedes degradation or re-usage. Recently, two different RNAPII eviction systems were proposed: a classic TC-NER reaction, where eviction and repair are linked [19, 20, 49], and a VCP/p97 repair-independent pathway, which may act as a backup for TC-NER to ensure RNAPII eviction [19, 20]. Although both eviction pathways were proposed to induce some level of RPB1 degradation, it is unclear how they compare to each other in the control of RPB1 fate after eviction (degradation or re-usage).

Whenever eviction is not possible, in situ degradation of lesion-stalled RPB1 molecules may act as a last resort, implying the direct interaction between the proteasome and the chromatin-engaged transcriptional complex. In this sense, interaction between the 26S proteasome and DNA has been reported in yeast [50, 51] and in humans [52, 53], supporting a model in which clearance of the lesion can be achieved directly by the proteasome. UV light, one of the most common cytotoxic agents used in research, induces different types of lesions that might well influence how RPB1 is cleared from the DNA lesion. Apart from pyrimidine dimers and many others, UV light exposure favours the crosslink between aromatic amino acids and nucleic acids [54]. In this sense, cryo-electron microscopy data of an elongating RNAPII complex showed that two phenylalanine and six tyrosine residues are sufficiently close, less that 10 angstroms, to a DNA or RNA nucleotide (Protein Data Bank accession number 8B3D). We hypothesize that a crosslink between RPB1’s residues and DNA or RNA may represent a scenario in which lesion clearance by the proteasome, in situ degradation, is favoured over different pathways for RPB1 eviction.

Phosphorylation, ubiquitylation, and degradation of RPB1

Regarding the nature of the RPB1 molecule that is degraded by the proteasome, i.e. phosphorylated or not, the conceptual framework thus far pointed towards RPB1 molecules that were part of active RNAPII complexes. In the “classic” in cis pathway, upon persistent stalling of an elongating RNAPII complex, its RPB1 molecule, likely phosphorylated at positions Ser 2 and Ser 5, is ubiquitylated by CRL4^CSA and/or others E3 complexes [6, 18, 19], then evicted from chromatin by the natural progress of the repair reaction or the VCP/p97 backup system [19, 20, 49], and finally degraded or re-used.

Recently, different mechanisms involving eviction and degradation of promoter-proximal paused RPB1 molecules (Ser5 phosphorylated) were proposed. Opposite to what we have described throughout this work, Steuerer and co-workers proposed a GSK3-dependent and DNA repair-independent mechanism for degradation of promoter-proximal paused molecules [25]. Also, degradation of promoter-proximal paused molecules in the absence of external damage was recently observed, suggesting that a quality control system operates in promoter-proximal paused molecules to avoid transcription by faulty RNAPII complexes [26–28]. In this work, we show that Ser2 and Ser5 unphosphorylated and transcriptionally inactive RPB1 molecules can also be degraded. Unphosphorylated molecules can be found in the nucleoplasm, where it was suggested that RPB1 forms transcriptionally inactive dimers, incapable of interacting with the DNA [55]. Additional research would be needed to confirm their nucleoplasmic localization and to analyse if these inactive dimers can be targeted for degradation.

Unphosphorylated RPB1 molecules can also be found in the promoter area, before pre-initiation complex (PIC) assembly and CTD phosphorylation, which is a rate-limiting step in transcription initiation. While the emerging mechanisms of RPB1 degradation involving promoter-proximal Ser5-phosphorylated molecules may act in a quality control pathway after PIC formation, the in trans mechanism proposed here is Ser5 and Ser2 independent, and would not act as a safeguard to ensure proper transcription and a faithful interaction with the DNA, but as a way to control RPB1 levels and RNAPII transcription under stress. Moreover, it was recently published that the decrease in the levels of unphosphorylated RPB1 (IIa) actively induces apoptosis, in a pathway coined Pol II degradation-dependent apoptotic response [56]. Therefore, the repair-dependent in trans degradation mechanism herein proposed may well regulate cell survival through the control of RPB1 IIa degradation.

Given the complexity of human cells and the relevance of the object of study (RPB1 levels / RNAPII activity), it is expected that different mechanisms are involved in the control of RPB1 levels. As commented before, the POLR2A gene ranked at the top of the genes whose copy number is of paramount importance for proper cell function, and it has been recently showed that variations in RPB1 levels deeply affect gene expression [31]. Therefore, it is not surprising that different mechanisms, active in different cell types or states, help to determine the abundance of RPB1. For the case herein described, the fact that NER modulates RBP1 levels suggests that RPB1 abundance might be controlled not only under an acute response to damage, but also in normal conditions. NER deals with different types of lesions, including some derived from oxidative metabolism like 5′,8-cyclopurines [57], and therefore, it is possible that high-energy demand tissues of NER-deficient patients present abnormal levels of RPB1 abundancy and, in turn, misregulation of gene expression.

Transient activation of NER-related Cullin-RING ligases

In relation to the E3 ubiquitin ligases involved in RPB1 ubiquitylation and degradation, our results are in agreement with the common notion in which CRL4^CSA is relevant for the in cis ubiquitylation reaction, as initial (3 h) degradation was sensitive to CSA and CSB levels. On the contrary, 12 h after UV, when most of repair is transcription independent, lesion processing triggered by GG-NER will activate an E3 CRL complex/es (Fig. 5) but likely not CRL4^CSA. Since proper ubiquitylation by CSA during TC-NER showed to be necessary for lesion repair, then CRL4^CSA activity may in turn activate a different CRL E3 complex/es by facilitating lesion processing. The fact that defects in the completion of the repair reaction are associated with an enhancement in RPB1 degradation (Fig. 3) suggests that CRL E3 ligases are transiently activated by lesion processing during the NER reaction, and that such a signalling will continue until DNA structure is restored, only after DNA synthesis and ligation. Supporting a model of a transient activation of different E3 complexes after damage, it has been published that CRL4 complexes are dynamically reorganized upon UV exposure [46]. In the present work, we have not identified the lysine residues involved in RPB1 degradation, which may involve not only K1268 [18] but, as suggested in [20], also others. In this regard, it has been shown that triptolide-induced RPB1 degradation persists in K1268R mutant cells [28], in agreement with the notion that multiple lysine residues contribute to RPB1 eviction and/or degradation.

RPB1 as a limiting factor under stress

The idea of an intimate connection between DNA repair and gene expression was established many years ago, when it became apparent that certain factors play roles in both processes. Thus, the notion of a dual and limiting factor operating in repair but not in transcription was conceived in an effort to comprehend the global inhibition of gene expression following DNA damage. Based on the findings presented in this study, we propose another layer in the coupling between gene expression and DNA repair, in which NER-dependent lesion processing turns RPB1 into a limiting factor, restricting RNAPII activity and hence shaping the gene expression response to DNA damage.

Supplementary data

Supplementary data is available at NAR online.

Conflict of interest

None declared.

Funding

Associazione Italiana per la Ricerca sul Cancro, AIRC, Italy (Grant/Award Number: AIRC-IG-21416; AIRC-5X1000-22759); Expresión Génica y Procesamiento del ARN en Patologías Humanas. Redes Federales de Alto Impacto. Ministerio de Ciencia, Tecnología e Innovación, 2023 (Grant/Award Number: CONVE-2023-100766162-APN-MCT); Agencia Nacional de Promoción Científica y Tecnológica (Grant/Award Number: '2020-1025', '2020-3194'); Ministero dell’Istruzione, dell’Università e della Ricerca (MIUR-PRIN- 2015SJLMB9; MIUR-PRIN202223MFOIA_01)

Data availability

The data underlying this article will be shared on reasonable request to the corresponding author.