Oxidative stress triggers RNAPII arrest through PARylation and DNA damage

Transcription, nascent transcription, RNA polymerase II (RNAPII), RNAPII 22 elongation, DNA damage, oxidative stress, transcription-coupled DNA repair 23 24 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 6, 2025. ; https://doi.org/10.1101/2025.11.06.686939doi: bioRxiv preprint 2 Summary 25 26 UV or gamma irradiation, as well as certain chemicals, generate DNA damage that disrupts 27 transcription through a variety of well -characterised mechanisms. In contrast, the 28 transcriptional response to oxidative stress remains poorly understood. Here, we describe a 29 rapid and widespread shutdown of transcription following oxidative DNA base damage. By 30 monitoring RNAPII occupancy and elongation dynamics, we demonstrate that oxidative stress 31 temporarily halts RNAPII pause release and arrests the progression of elongation complexes 32 within the gene body. We present evidence that this occurs in a unique and transient manner, 33 characterised by abrupt arrest of elongating RNAPII dead in its tracks , followed by rapid 34 transcriptional recovery as DNA lesions are repaired. We find that the restriction of initiation 35 and early elongation complexes is regulated by PARylation, whereas recovery of RNAPII 36 arrested within the gene body requires DNA repair mediated by the base excision repair (BER) 37 and single-strand break repair (SSBR) pathways. 38 39 40 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 6, 2025. ; https://doi.org/10.1101/2025.11.06.686939doi: bioRxiv preprint 3

41 42 Our genome is continuously exposed to genotoxic insults causing DNA lesions and genom ic 43 instability 1. Lesions originating from oxidative damage are among the most frequently 44 observed in the germline 2 and accumulation of mutations from oxidative damage has been 45 associated with aging, neurodegenerative diseases , and cancer 1,3. Actively transcribed genes 46 are particularly vulnerable to genomic instability, and rapid repair within genes is therefore of 47 utmost importance to prevent deleterious long-term effects 4,5. DNA repair and transcription are 48 inevitably coupled as they work on the same template. To maintain fidelity of cellular 49 processes, cells frequently respond to genotoxic stress by silencing transcription, either locally 50 or globally 4,6-9. In fact, certain types of DNA lesions can serve as a physical roadblock to 51 transcription, causing a complete halt to RNAPII elongation 7. For instance, UV-induced bulky 52 DNA lesions directly impede RNAPII progression and necessitate repair through the 53 transcription-coupled nucleotide excision repair pathway (TC -NER) to facilitate restart of 54 transcription 9-13. 55 In contrast, the direct impact on transcription is less clear in the case of non-bulky DNA 56 lesions and single-stranded breaks (SSBs) frequently induced by oxidative damage. While it is 57 known that exposure to oxidative stress results in a global repression of transcription 8,14,15, it 58 remains unknown if and how oxidative DNA lesions impact the transcription machinery in vivo. 59 Lesions originating from oxidative stress include non-bulky single-base modification such as 60 8-oxoguanine (8OG), abasic sites, and SSBs. In fact, SSBs is the most frequent type of DNA 61 lesions, with an estimated occurrence of tens-of-thousands per cell per day 1,16. Data generated 62 in vitro suggest that RNAPII is able to bypass 8OG DNA lesions 17-19, while products 63 originating from further oxidation impede RNAPII progression 20. In addition, SSBs generated 64 either as a direct result of oxidative damage, or as BER repair intermediates have been shown 65 to impair RNAPII elongation in vitro 21. However, transcription assays performed with nuclear 66 extracts from hydrogen peroxide -treated cells remained transcriptionally inactive even on an 67 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 6, 2025. ; https://doi.org/10.1101/2025.11.06.686939doi: bioRxiv preprint 4 undamaged DNA template, suggesting that DNA lesions by themselves may not be sufficient 68 to explain the global transcriptional repression observed following oxidative stress 8. 69 Besides physical impairment caused by the DNA lesion itself, the DNA damage 70 response induces a coordinated recruitment of factors involved in DNA repair to the sites of 71 damage, which can elicit either local or global repression of transcription 22,23. The recruitment 72 of DNA repair factors can be facilitated either directly by the DNA lesion itself, or through 73 recognition of lesion -stalled RNAPII, such as in TC-NER 7,9. Interestingly, rather than being 74 caused by the lesion itself, local transcriptional repression around sites of double-strand breaks 75 (DSBs) is dependent on DNA damage signalling involving ATM, DNAPK, and poly(ADP-76 ribosyl)ation (PARylation)-mediated regulation of RNAPII close to the site of damage as well 77 as changes in the chromatin environment 24-28. PARylation also serves to recruit NELF -E to 78 RNAPII stalled at DSBs 28. In addition to DSB s, PARylation is also implicated in the 79 recruitment of DNA repair factors to SSB lesions. Here , PARP1-mediated PARylation 80 stimulates recruitment of X-ray repair cross -complementing group 1 (XRCC1) as well as 81 additional factors involved in SSB repair (SSBR) 23,29,30. Interestingly, the lack of XRCC1 leads 82 to a failure of transcription recovery following oxidative stress, suggested to involve PARP1 83 ‘trapping’ on DNA , which would create a physical block consisting of an unresolved repair 84 intermediate rather than the lesion itself 15. 85 In addition to transcriptional repression induced by DNA damage, it is also becoming 86 evident that cellular stress such as heat shock result s in widespread repression of transcription 87 31-35. Multiple mechanisms have been reported to be involved in transcriptional attenuation 88 following heat shock, including promoter -proximal pausing of RNAPII 33, NELF condensate 89 formation 36, and premature transcript termination 37. 90 Using a variety of genome -wide approaches to track the transcriptional response in a 91 temporally and spatially resolved manner , we find that transcriptional activity in response to 92 oxidative stress is dynamically regulated through early RNAPII pause release, coordinated by 93 PARylation in combination with rapid and reversible arrest of RNAPII in the gene body, 94 controlled at the level of DNA damage repair. 95 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 6, 2025. ; https://doi.org/10.1101/2025.11.06.686939doi: bioRxiv preprint 5 96 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 6, 2025. ; https://doi.org/10.1101/2025.11.06.686939doi: bioRxiv preprint 6

97 98 Oxidative stress induces a rapid transcriptional shutdown 99 To investigate the immediate transcriptional response to oxidation-induced stress and DNA 100 damage, we treated cells with a brief, sublethal dose of either H2O2 or menadione, the latter of 101 which leads to intracellular ROS formation (Fig. 1a, Supplementary Fig. 1a-b). Using metabolic 102 labelling of nascent RNA with 4-thiouridine (4sU) or 5-ethynyluridine (EU) in human cells, we 103 tracked the global levels of RNA synthesis during and upon treatment recovery (Fig. 1b-d and 104 Supplementary Fig. 1c-e). Both H2O2 and menadione led to a pronounced decrease in overall 105 RNA synthesis across several cell lines (Fig. 1b-d, Supplementary Fig. 1c). In the case of brief 106 H2O2 treatment, this effect was almost immediate, with a rapid recovery of transcription activity 107 (Fig. 1b-c). In contrast , as expected, menadione induced a somewhat milder and temporally 108 delayed response in terms of repression of nascent transcription levels (Fig. 1b-d). Similarly, 109 lower concentrations of H2O2 (250 M) also resulted in a delayed reduction of nascent 110 transcriptional activity (Supplementary Fig. 1c). In contrast, a high H2O2 dose (10 mM) or 111 menadione (2 mM) resulted in a failure to recover normal transcription levels, impaired cell 112 growth, and apoptosis (Supplementary Fig. 1d-f). 113 To investigate the effects on transcription genome-wide, we performed TT chem-seq 38 114 using yeast spike -ins as an internal normali sation control to accurately quantify changes in 115 nascent transcription in a spatially resolved manner. In agreement with the global quantification 116 from dot blots and EU assays, a dramatic reduction in overall transcription activity throughout 117 the entire gene body of protein-encoding genes was observed (Fig. 1f-h). This was the case both 118 for menadione and for H2O2 (Fig. 1f-h). In the case of H2O2, 99.5 % of differentially expressed 119 protein-encoding genes were downregulated (Supplementary Fig. 1g). At lower H2O2 120 concentrations, transcriptional repression was strongest at the 30 min recovery timepoint, while 121 a lower menadione concentration led to the strongest repression at 1 -hour recovery 122 (Supplementary Fig. 2a-c), which again aligns with the quantifications from dot blots and EU 123 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 6, 2025. ; https://doi.org/10.1101/2025.11.06.686939doi: bioRxiv preprint 7 assays (Fig. 1b-c). Classical stress-response genes such as the immediate early genes FOS and 124 EGR1 were also repressed upon H2O2 and menadione treatment and with kinetics mimicking 125 the global repression (Supplementary Fig. 2d-e), highlighting that even genes later activated as 126 part of various stress-response pathways do not initially escape the transcriptional repression 127 following oxidative stress. 128 To investigate the possibility that direct inhibition of RNAPII activity might occur 129 because of its oxidation, we assembled in vitro transcription elongation complexes with purified 130 RNAPII, pretreated them with increasing concentrations of H 2O2 and performed in vitro 131 transcription assays. Whereas the transcriptional activity of RNAPII elongation complexes was 132 reduced at extremely high concentrations of H2O2 (>50 mM ), we failed to observe any 133 impairment of RNAPII activity with lo wer ( <5-10 mM) H 2O2 concentrations, making it 134 unlikely that the lack of transcriptional activity in cells is due to direct RNAPII inactivation 135 (Supplementary Fig. 3a-b). 136 The results above indicate that transcriptional activity is dramatically reduced 137 throughout the gene body almost immediately after exposure to oxidative stress. The dramatic 138 decrease in nascent transcription within gene bodies can either be due to (1) dissociation of 139 RNAPII elongation complexes from the DNA template, resulting in fewer polymerases in the 140 gene body, or (2) stalling or arrest of RNAPII resulting in disruption of RNA synthesis by those 141 RNAPII elongation complexes. Interestingly, t he reduction of transcription did not at first 142 glance appear to be caused by changes in the level of DNA -associated RNAPII as both 143 hyperphosphorylated and unphosphorylated RNAPII levels in chromatin remained unchanged 144 (Fig. 1i). 145 To more precisely track RNAPII occupancy, we used ELCAP -seq 39 which maps 146 RNAPII binding at nucleotide resolution (Fig. 1j). To ensure that all RNAPII populations were 147 effectively captured regardless of their CTD -phosphorylation status, we used cells expressing 148 FLAG-tagged RPB3 (Fig. 1j and Supplementary Fig. 3c). Short RNA fragments protected from 149 nuclease digestion were captured by FLAG-immunoprecipitation of total RNAPII followed by 150 sequencing (Supplementary Fig. 3c-e). Despite the fact that transcription activity was markedly 151 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 6, 2025. ; https://doi.org/10.1101/2025.11.06.686939doi: bioRxiv preprint 8 down during the initial response, the levels of RNAPII in the gene body remain ed largely 152 unchanged ( Fig. 1k). The only differences being a slight accumulation of early elongation 153 complexes close to the TSS ( Fig. 1k). Intriguingly, this suggests that in response to oxidative 154 stress, elongating RNAPII complex es within the gene body are not dissociated but instead 155 rapidly arrested, in a global manner leading to an immediate transcription impairment. 156 157 Oxidative stress represses RNAPII elongation within the gene body 158 To confirm that the transcriptional repression observed after oxidative stress is indeed due to 159 reduced transcription in the gene body, we assessed the effect of oxidative stress on gene body-160 associated RNAPII in the presence or absence of the CDK9 inhibitor 5,6-Dichloro-1-β-D-161 ribofuranosylbenzimidazole (DRB) (Fig. 2a). In these experiments, RNAPII was first cleared 162 from the gene body by a treatment with DRB for 3.5 hours, which prevents promoter proximal 163 pause release, while allowing all RNAPII to run until the 3’ end of genes and terminate. DRB 164 inhibition is reversible, enabling release of RNAPII complexes into the gene body upon DRB 165 washout. After the 3.5 hours DRB incubation, RNAPII was allowed to enter the gene body by 166 15 min of unperturbed transcription (RNAPII release phase), after which cells were treated with 167 H2O2 alone, or in combination with fresh DRB, and incubated for another 15 min (Fig. 2a). 168 This results in the release of RNAPII complexes into the gene body, which are in all cases 169 allowed to run for a total of 30 min (15 min unperturbed + 15 min with indicated treatments) . 170 To investigate the effect of oxidation on RNAPII complexes already in the gene body , we 171 focused our analysis on long genes (> 90 kb). In agreement with previous data showing that 172 RNAPII travels approximately 2 kb/min and with a maximum speed around 4 kb/min 38, we 173 observed that the RNAPII wavefront progressed to around 60-90 kb into the gene body in cells 174 not treated with H2O2 (Fig. 2b-e, blue tracks). By contrast, progression of the wave front into 175 the gene body was clearly reduced in response to a brief oxidative stress treatment (Fig. 2b-e, 176 red tracks ). Importantly, the wavefront reached a similar point in the gene whether new 177 transcription was inhibited by DRB or not at the time of oxidative stress treatment ( Fig. 2b-e; 178 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 6, 2025. ; https://doi.org/10.1101/2025.11.06.686939doi: bioRxiv preprint 9 most easily seen by comparing red tracks in d and e). This strongly indicates that H2O2 treatment 179 also arrested RNAPII molecules that were already in the gene body. 180 181 Arrested RNAPII complexes are not subject to ubiquitination and subsequent 182 degradation 183 A hallmark of RNAPII stalled at bulky UV lesions is polyubiquitination of the largest RNAPII 184 subunit RPB1 and subsequent RNAPII degradation 10,11. Unlike the rapid transcriptional 185 repression that occurs following oxidative damage, UV -induced transcriptional repression 186 peaks around 45 min to 3 hrs after the initial exposure (Supplementary Fig. 4a). To examine if 187 RNAPII becomes ubiquitinated following oxidative damage, we tested RPB1 188 polyubiquitination in a time -course-dependent manner using the DSK2 affinity enrichment 189 strategy for polyubiquitinated proteins 40. Although we did observe low levels of RPB1 190 polyubiquitination, these are much lower than RPB1 polyubiquitination after UV irradiation 191 (Supplementary Fig. 4b). Notably, unlike the case with UV 10, we did not observe any decrease 192 in overall levels of hyperphosphorylated transcriptionally engaged RNAPII during either the 193 oxidative stress-induced RNAPII arrest or recovery (Supplementary Fig. 4c). 194 Upon UV-irradiation, RNAPII stalls at bulky DNA lesions which serve as a roadblock 195 for polymerase elongation 7. Accordingly, removal of bulky DNA lesions by the transcription-196 coupled nucleotide excision repair (TC -NER) pathway is a prerequisite for transcriptional 197 restart following UV 10,13,41. Interestingly, TC-NER has also been implicated in lesion repair in 198 response to oxidative stress 42-45. To investigate the effects of NER following oxidative stress, 199 we tracked nascent transcription recovery in XPC and CSB knockout cells, which lack GG -200 NER and TC -NER, respectively 7. Knockout of XPC or CSB had little or no effect on the 201 transcriptional response (Supplementary Fig. 4d). This is in stark contrast to the transcriptional 202 recovery after UV, which is completely CSB -dependent 10,12,41. We also failed to detect 203 recruitment by IP–mass spectrometry of CSB, XPC and other DNA repair factors involved in 204 NER to RNAPII complexes following oxidative stress (Supplementary Table 1). Together with 205 the lack of significant RPB1 ubiquitination observed following oxidative stress (Supplementary 206 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 6, 2025. ; https://doi.org/10.1101/2025.11.06.686939doi: bioRxiv preprint 10 Fig. 4b), this indicates that RNAPII arrest following oxidative damage is mechanistically 207 distinct from RNAPII arrest observed as a consequence of UV-irradiation. 208 209 Gene body stalled RNAPII resumes transcription upon recovery from oxidative stress 210 The data above indicated that in response to oxidative stress, elongating RNAPII complex es 211 are rapidly arrested, dead in their tracks, leading to a global, temporary shutdown of nascent 212 transcription. To further explore the transcription dynamics during the recovery of 213 transcription, we tracked nascent transcript ion activity over time immediately after oxidation 214 (Fig. 3a). In agreement with the overall transcription level s indicated by dot blots and EU 215 assays, we observe d increasing restart of nascent RNAPI I transcription 30 min and 1 hour 216 following H2O2 removal (Fig. 3a-b, Supplementary Fig. 5a). Interestingly, transcription activity 217 was restored through the gene body at both time -points, with higher transcription activity 218 restoration towards the 3’end of long genes at the 1-hour timepoint (Fig. 3a-b, Supplementary 219 Fig. 5b). This again indicates that even RNAPII elongation complexes arrested by oxidation 220 inside the gene body can restart transcription. This point is clearly illustrated by tracking 221 nascent transcription recovery for individual long genes, such as FARS2 (>570 kb) (Fig. 3b). 222 As stated previously, RNAPII elongation rate measurements indicate that RNAPII travels at 223 approximately 2 kb/min and with a maximum speed around 4 kb/min 38. So, if new nascent 224 transcription during the recovery period originated only from newly initiated RNAPII or 225 RNAPII at the promoter-proximal pause site, all transcriptional activity would be restricted to 226 the first 60 -120 kb after 30 min recovery and 120 -240 kb after 1 hour recovery (Fig. 3b, 227 indicated by arrows). However, recovery of transcription activity was observed throughout the 228 gene already at 30 min and 1 hour recovery (Fig. 3b, indicated by dashed boxes ), suggesting 229 that RNAPII complexes indeed resume transcription from within the gene body following 230 oxidation-induced transcriptional arrest. 231 To further rule out the possibility that newly initiated transcription is responsible for 232 the recovery of transcriptional activity in the gene body , we finally tracked recovery in the 233 absence of new release of RNAPII into the gene body by treating cells with DRB during the 234 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 6, 2025. ; https://doi.org/10.1101/2025.11.06.686939doi: bioRxiv preprint 11 recovery period (Fig. 3c). We thus found that overall transcription activity was restored even 235 when DRB was present during recovery from H2O2 treatment, clearly seen at the 30 min and 1 236 hour time points (Fig. 3d) before already transcribing polymerases have run off the gene which 237 takes place during longer DRB incubations (Fig. 3d, lower panel s). Using TT chem-seq, we 238 further confirmed that RNAPII complexes indeed were able to restart transcription from within 239 the gene body in the presence of DRB. Indeed, as expected, transcriptional recovery for one 240 hour after oxidative stress was shifted towards the 3’end and absent in the 5’ end of genes in 241 the presence of DRB ( Fig. 3e, dashed purple line; Fig. 3f, lowest panel) , and obvious from 242 quantification of nascent transcription levels in the first 25 kb compared to nascent transcription 243 levels in the last 25 kb of genes only (Supplementary Fig. 5b). At the same time, the most 244 prominent change in RNAPII occupancy was a slightly increased pausing index (indicating 245 accumulation of early RNAPII complexes) measured with ELCAP -seq during the recovery 246 phase ( Supplementary Fig. 5c). This indicates that the response to oxidative damage is 247 fundamentally different from transcriptional repression induced by bulky DNA lesions which 248

in RPB1 ubiquitination, subsequent RNAPII degradation and transcription restart as a 249 5’ to 3’ wave front originating from the TSS by transcription from newly initiated RNAPII 250 complexes 7,10,46-48. 251 To investigate if RNAPII elongation rates change during the recovery phase across 252 genes, we computed the elongation index as the ratio of active elongation (TTchem-seq signal) 253 and RNAPII occupancy (ELCAP-seq), which was previously used as a proxy for RNAPII speed 254 49,50. As expected, this indicated a strong decrease in the RNAPII elongation index during the 255 H2O2 treatment ( Supplementary Fig. 5d). As transcription resumes, the RNAPII elongation 256 index in the gene body increase d, while the elongation index in the beginning of the gene 257 decreased slightly at the 30 min time point ( Supplementary Fig. 5d). At 1 hour recovery this 258 was even more pronounced, with the elongation index almost back to normal levels in the 3’end 259 of the gene body, while the elongation index for early gene body RNAPII complexes decreased 260 even further (Supplementary Fig. 5d). The inverse behaviour of RNAPII elongation indices for 261 early RNAPII complexes and gene body RNAPII indicate that early stalled RNAPII complexes 262 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 6, 2025. ; https://doi.org/10.1101/2025.11.06.686939doi: bioRxiv preprint 12 and gene body -arrested RNAPII complexes resume transcription differently during the 263 recovery period (Supplementary Fig. 5c-d). This could either be due to distinct mechanisms 264 controlling release of early RNAPII elongation complexes and gene body-arrested RNAPII, or 265 due to differently con figured RNAPII complexes (bound by different RNAPII associated 266 factors) impacting elongation kinetics of RNAPII complexes differently depending on their 267 position at the time of transcription block. 268 269 NELF mediated pausing of RNAPII is altered during the oxidative stress response 270 To investigate whether RNAPII complexes are remodelled during the global transcriptional 271 repression caused by oxidative stress, we investigated the RNAPII interactome via proteomic 272 analysis of RNAPII immunoprecipitates. We used two approaches (1) capturing total RNAPII 273 complexes using a FLAG-tagged RPB3 cell line that allows us to capture all RNAPII 274 complexes regardless of the phosphorylation status of the CTD, and (2) specifically capturing 275 transcriptionally engaged RNAPII using the monoclonal 4H8 antibody, which recognizes CTD 276 hyperphosphorylated RNAPII (Supplementary Fig. 6a). Notably, u sing both approaches we 277 observed an increased enrichment of NELF bound RNAPII complexes in H 2O2 treated cells 278 (Fig. 4a-b and Supplementary Fig. 6b, Supplementary Table 1 ). In addition, we found an 279 increase in 5’ capping components such as CMTR1 and RNGTT associated RNAPII complexes 280 upon treatment with H2O2 (Fig. 4a-b). This is in line with data suggesting that recruitment of 281 capping factors is dependent on NELF 51 and suggests increased early RNAPII pausing. We 282 confirmed the increased interaction between RNAPII and the NELF complex by RNAPII IP-283 western analysi s (Fig. 4c). Interestingly, while NELF recruitment to RNAPII was initially 284 increased, it decreased during the recovery period and after 1 hour and 2 hours of recovery, 285 NELF was less associated with RNAPII than even the untreated conditions (Fig. 4c). This trend 286 was also observed in RNAPII IP proteomics both at 1 hour and 2 h ours recovery 287 (Supplementary Fig. 6b). 288 To address the role of NELF -mediated RNAPII regulation during the oxidative stress 289 response, we carried out a detailed analysis of RNAPII occupancy . ELCAP -seq provides 290 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 6, 2025. ; https://doi.org/10.1101/2025.11.06.686939doi: bioRxiv preprint 13 nucleotide resolution of RNAPII binding and is therefore ideally suited to investigate RNAPII 291 promoter-proximal pausing. This uncovered an increase in promoter-proximal pausing at 30 292 min after oxidative stress (Fig. 4d and Supplementary Fig. 6c). Interestingly, pausing after 30 293 min recovery was primarily increased at the canonical pause site (20-50 bp from the TSS), also 294 known as the 1st pause site of NELF-mediated pausing 51,52. However, at later time points ( 1 295 hour and 2 hours recovery), RNAPII occupancy at the canonical pause site decreased, with the 296 RNAPII occupancy shifting downstream to the ‘2nd pause site’ coinciding with the +1 297 nucleosome 51 (Fig. 4d-f). Tellingly, a similar shift from 1st to 2nd pause site has previously been 298 described to occur after loss of NELF 51,52. To test if the increased association between NELF 299 and RNAPII correlates with the changes in promoter-proximal pausing, we plotted the relative 300 NELF binding (ascertained by NELFE ChIP-seq) after normalisation to RNAPII binding levels 301 (spike-in normalised RNAPII ChIP-seq) (Fig. 4f). During the initial repression after oxidative 302 stress, more NELF accumulated at the promoter-proximal pause site relative to RNAPII , 303 indicating active NELF recruitment, or impaired NELF release, at paused RNAPII complexes 304 (Fig. 4f). However, during recovery, less NELF was bound to promoter-proximal RNAPII, in 305 agreement with our IP results and the observed shift in RNAPII pausing from the 1st pause site 306 to the 2nd pause site (Fig. 4c-f). 307 We wondered if the increased NELF binding to RNAPII during the initial phase of the 308 oxidative stress response serves to restrict paused RNAPII elongation complexes from entering 309 productive elongation. To test this, we generated NELFC degron cell lines enabling us to track 310 nascent transcription in response to oxidative stress in cells lacking the NELF complex 311 (Supplementary Fig. 6d). Treatment with dTAG led to depletion of NELFC within 1 hour 312 (Supplementary Fig. 6e). As previously reported 51, NELFC depletion also resulted in 313 degradation of additional NELF subunits, such as NELFA (Supplementary Fig. 6e). 314 Intriguingly, however, the overall levels of nascent transcription either during the immediate 315 response to H 2O2 or in the recovery period were unaffected by loss of NELF (Supplementary 316 Fig. 6f). As NELF condensation formation has been suggested to drive transcriptional 317 repression during the heat shock response 36, we checked nuclear localisation of NELF during 318 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 6, 2025. ; https://doi.org/10.1101/2025.11.06.686939doi: bioRxiv preprint 14 the oxidative stress response. As expected from previous data by others, changes in the nuclear 319 location of NELF condensate formation was observed following heat shock, but no indications 320 of condensate formation following oxidative stress were observed (Supplementary Fig. 6g). 321 Thus, the NELF depletion experiments indicate that NELF -mediated regulation is not 322 responsible, nor required, for the widespread transcriptional repression observed upon oxidative 323 stress. Instead, we observe that the position of early RNAPII pausing is affected during the 324 oxidative stress response in a NELF-dependent manner; first accumulating at the 1st pause site 325 (increased NELF binding) followed by a shift towards the more downstream +1 nucleosome 326 pause site (decreased NELF binding). 327 328 PARylation governs early transcriptional recovery 329 Another mean s of promoting RNAPII stalling in response to DNA damage is PARylation 330 catalysed by poly(ADP-ribose) polymerase 1 (PARP1) 14. PARP1 activation leads to 331 PARylation of PARP1 itself (auto-PARylation) and other proteins which serve as scaffolds for 332 recruitment of DNA repair factors at the site of DNA lesions 23,53. To test if PARylation might 333 be relevant for the transcriptional response to oxidative stress, we first measured the level of 334 ADP-ribose on proteins in a time-dependent manner. In agreement with previous data 14, we 335 observed a strong and rapid increase in PARylation during the initial oxidative stress response 336 (Fig. 5a, left). The highest levels were observed during the H 2O2 treatment itself, coinciding 337 with the time transcriptional arrest (Fig. 5a, left). During the recovery phase PARylation levels 338 rapidly decreased and almost reached baseline levels at the 2 hours recovery timepoint (Fig. 5a, 339 left). PARylation can be modulated by treatment with either PARP inhibitor (PARPi) or 340 poly(ADP-ribose) glycohydrolase inhibitor (PARGi) 54. As expected, treatment of cells with 341 PARPi prior to oxidative stress abolished PARylation (Fig. 5a, right) while pre-treatment with 342 a PARG inhibitor (PARGi) resulted in sustained PARylation ( Fig. 5b). To test if PARylation 343 might play a functional role during the oxidative stress response we first tested the overall levels 344 of nascent transcription in cells treated with either PARPi or PARGi prior to the induction of 345 oxidative stress. Strikingly, PARPi led to faster recovery of overall transcription levels, while 346 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 6, 2025. ; https://doi.org/10.1101/2025.11.06.686939doi: bioRxiv preprint 15 sustained PARylation following treatment with PARGi resulted in reduced transcriptional 347 recovery (Fig. 5c and Supplementary Fig. 7a-b). To address how PARylation affects nascent 348 RNAPII transcription in a genome-wide manner, we performed nascent transcriptome profiling 349 by TTchem-seq. Confirming the dot blot results, we first observed higher levels of overall 4sU-350 labelled RNA at the 15 min timepoint in PARPi treated cells (Fig. 5d). Somewhat surprisingly, 351 the TTchem-seq data revealed that PARPi led to increased release of early RNAPII elongation 352 complexes into the gene body at the 15 min recovery timepoint (Fig. 5e-f). The same trend of 353 increased nascent transcription levels primarily within the first 40 kb upon PARPi pre-treatment 354 was also clearly observed at a single gene level ( Fig. 5g) and across genes regardless of gene 355 length (Fig. 5h). With an RNAPII elongation rate of 2-4 kb/min, RNAPII complexes originating 356 from newly initiated RNAPII complexes or released from the promoter proximal site would be 357 expected to travel 30 -60 kb into the gene body at the 15 min timepoint, which corresponds 358 nicely to the region in which the primary increase in RNAPII transcription al activity upon 359 PARPi treatment was observed (Fig. 5e-h). This suggests that PARylation after oxidative stress 360 inhibits either RNAPII initiation or RNAPII promoter-proximal pausing, or both. Strikingly, 361 the spatial effect of PARPi treatment on nascent transcription across the gene was sustained 362 even at the 1-hour recovery timepoint , d espite seeing similar overall levels of nascent 363 transcription between PARPi and control-treated cells at this timepoint ( compare Fig. 5c and 364 5h), again indicating that the effect of PARylation is primarily restricted to early elongation. 365 This differential effect indicates that the transient arrest of RNAPII further into the 366 gene body rel ies on other mechanisms or additional factors be yond PARylation signalling. 367 Intriguingly, NuMA, a nuclear protein associated with the mitotic spindle, binds RNAPII in a 368 PARP1-dependent manner with repair proteins TDP1 and XRCC1 and has been suggested to 369 help promote oxidative break repair 55. In apparent agreement with the notion that PARylation 370 promotes NuMA binding 55, we indeed observed increased association of NuMA with RNAPII 371 immediately after oxidative stress (Supplementary Table 1). Importantly, however, NuMA was 372 not detected in RNAPII elongation complexes captured by IP of hyperphosphorylated RNAPII, 373 but only in IPs capturing total RNAPII compl exes (Supplementary Table 1) . This is i n 374 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 6, 2025. ; https://doi.org/10.1101/2025.11.06.686939doi: bioRxiv preprint 16 agreement with the idea that NuMA associates primar ily with initiating RNAPII, and in line 375 with the findings above that PARylation primarily affects this RNAPII population. 376 377 Lack of SSBR prevents transcriptional recovery 378 A key function of PARP activ ation in response to oxidative ly induced DNA damage is to 379 facilitate recruitment of the molecular scaffold protein XRCC1 to the single-stranded DNA 380 breaks (SSB) generated either as a direct consequence of oxidative damage or by incision by 381 DNA glycosylases such as OGG1 , promoting efficient base excision repair (BER) 29,30. To 382 investigate the role of XRCC1 in the transcriptional response to oxidative stress, we generated 383 XRCC1 KOs (Supplementary Fig. 7c). In agreement with previous reports 15, XRCC1 KOs 384 showed a clear defect in recovery of nascent transcription following oxidative stress (Fig. 6a-385 b). This effect was observed even 2 and 4 hours after treatment, suggesting that XRCC1 is 386 critical for transcriptional recovery following oxidative stress (Fig. 6a-b). Using nascent 387 transcriptomics, we found that lack of XRCC1 led to an almost complete failure of RNAPII 388 transcription restart throughout the gene body ( Fig. 6c). Compared to WT cells, XRCC1 KOs 389 first and foremost displayed much less transcription recovery towards the end of long genes at 390 the 1-hour recovery time point ( Fig. 6c), indicative of failure to restart gene body -arrested 391 RNAPII complexes. Previous reports have indicated that XRCC1 is dependent on PARylation 392 for its rapid recruitment onto oxidatively induced SSB 30. Noticeably, pre-treatment of XRCC1 393 KOs with PARPi res cued the delay in overall nascent transcription levels observed in the 394 XRCC1 KOs ( Fig. 6d). This suggests that PARylation serve s a role in the transcriptional 395 response beyond simply facilitating XRCC1 recruitment to sites of DNA damage. 396 To address the transcriptional interplay between XRCC1 and active PARylation 397 genome-wide, we performed nascent transcriptome profiling by TT chem-seq of XRCC1 KOs, 398 with and without PARPi pre-treatment (Fig. 6e-h). Interestingly, when focusing on elongation 399 close to the TSS, PARPi completely reverse d the effect of XRCC1 KOs at 1 hr recovery (Fig. 400 6e). Thus, PARPi treatment both in WT and XRCC1 KOs led to enhanced transcription activity 401 early in the gene body, in accordance with the previous results showing PARylation-dependent 402 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 6, 2025. ; https://doi.org/10.1101/2025.11.06.686939doi: bioRxiv preprint 17 release of early RNAPII complexes. However, looking at metagene plots for long genes, it was 403 obvious that PARPi treatment could not fully reverse the transcription recovery towards the 404 3’end of long genes even at the 1-hour recovery timepointc. 6f). Tellingly, PARPi treatment in 405 XRCC1 KO cells did not fully rescue the lack of nascent transcription recovery towards the end 406 of genes to the same levels as PARPi treated cells, indicating a PARylation-independent effect 407 of XRCC1 on gene body recovery (Fig. 6f). This is also evident from single gene examples, 408 where transcription recovery in the beginning of CERS6 is similar for PARPi treatment in WT 409 and XRCC1 KOs, while 3’end recovery is not fully rescued by PARPi treatment in XRCC1 KOs 410 (Fig. 6g). Looking specifically at the PARPi effect in XRCC1 KOs across gene s with ranked 411 according to their lengths, we again see an effect of PARPi restricted primarily to the initial 40 412 kb region ( Fig. 6h). This suggest s that while PARPi promoted release of early elongation 413 complexes can partially ‘ override’ the lack of transcription recovery in XRCC1 KOs through 414 release of early elongation complexes, PARPi treatment cannot fully overcome transcriptional 415 defects further into the gene body (Fig. 6h). To investigate if the effect of XRCC1 might be due 416 to PARP ‘trapping’ on DNA as previously reported 56, we checked the levels of chromatin-417 associated PARP1 protein in a time-dependent manner. However, we found that PARPi did not 418 led to increased chromatin levels of PARP1, either in WT cells or in XRCC1 KOs 419 (Supplementary Fig. 7d), arguing that PARP trapping is unlikely to be the cause of the 420 transcriptional phenotypes observed upon PARPi treatment in our experiment. 421 422 Lack of OGG1 activity slows transcriptional restart 423 One of the most frequent o xidative damage-induced DNA lesions is the non-bulky 8-424 oxoguanine (8OG), which is primarily recognised by the base excision repair (BER) pathway. 425 Although RNAPII has been shown to bypass 8OG lesions present in naked DNA templates 17-426 19, it is unknown how 8OG might impact the transcription machinery in vivo. To investigate the 427 role of BER and whether BER is required for transcription recovery we chemically inhibited 428 the 8OG DNA glycosylase-1 (OGG1). Interestingly, OGG1 inhibition (OGG1i) also led to a 429 clear delay in the recovery of transcription levels (Fig. 7a-b), indicating that recovery of nascent 430 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 6, 2025. ; https://doi.org/10.1101/2025.11.06.686939doi: bioRxiv preprint 18 transcription is dependent on the BER pathway. This was again further confirmed by nascent 431 transcriptome profiling by TTchem-seq, which showed that transcriptional restart within the gene 432 body at 1 hour recovery was almost absent ( Fig. 7c-f). Intriguingly, although treatment with 433 OGG1i delayed restart of RNAPII, the level of transcriptional activity at later time points 434 recovered close to levels of untreated cells (Fig. 7a-b), suggesting that rather than preventing 435 transcriptional restart completely, BER defects such as those inflicted by lack of OGG1 activity 436 delays transcriptional restart. This contrasts with the situation in XRCC1 KOs where 437 transcription activity remained impaired even at 4 hours recovery (Fig. 6a-b). To investigate 438 how BER defects influence the PARylation -driven release of early RNAPII complexes, we 439 combined OGG1i with PARPi pre-treatment. Firstly, we wanted to see if PARPi treatment 440 could override the dramatic sustained OGG1i nascent transcription repression. Indeed, similarly 441 to the situation for XRCC1, PARPi pre -treatment result ed in recovery of overall nascent 442 transcription levels starting already at the 15 min recovery time point ( Fig. 7d) which was 443 maintained throughout the recovery period (Fig. 7d and Supplementary Fig. 7e). To determine 444 if the PARPi effect was restricted to release of early RNAPII elongation complexes, we 445 performed TT chem-seq and looked specifically at the nascent transcription recovery for long 446 genes at the 1-hour timepoint. Like the situation for XRCC1, the added transcriptional activity 447 brought about by PARPi pre-treatment in OGG1i treated cells was primarily restricted to early 448 elongation complexes ( Fig. 7e). However, when focussing also on the 3’end of long genes, 449 PARPi pre-treatment was able to overcome the OGG1i elongation block to a greater extent than 450 in XRCC1 KOs (Fig. 7f). Nevertheless, gene body transcription levels in the 3’end of long genes 451 were still greatly reduced compared to WT conditions ( Fig. 7e). Something that is also very 452 evident from an individual gene example of the long gene PAM (Fig. 7g). The spatially 453 restricted effect of PARPi ‘force d’ early RNAPII release was further confirmed by heatmaps 454 looking across both long and short genes (Fig. 7h). Thus, again PARPi treatment leads to release 455 of early RNAPII elongation complexes , an effect that can partially override the elongation 456 block observed in cells defective for OGG1-mediated repair (Fig. 7e-h, Supplementary Fig. 7e). 457 This again suggests that while PARylation restricts early elongation complexes, DNA repair is 458 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 6, 2025. ; https://doi.org/10.1101/2025.11.06.686939doi: bioRxiv preprint 19 crucial for resumption of RNAPII transcription within the gene body. An effect which becomes 459 particularly evident when looking at the 3’end of long genes ( Fig. 7f-h). Together, these data 460 thus indicate that, after oxidative damage, PARylation is important to restrict release of RNAPII 461 from promoter-proximal areas, while DNA repair is required to remove DNA lesions in the 462 path of RNAPII. 463 464 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 6, 2025. ; https://doi.org/10.1101/2025.11.06.686939doi: bioRxiv preprint 20

465 466 Genotoxic-induced transcriptional attenuation occurs in response to UV and gamma 467 irradiation, as well as oxidative damage 6-8,10. However, the transcriptional response varies 468 significantly depending on the nature of the damage. In contrast to previously described 469 mechanisms of transcriptional attenuation following UV -induced damage, we find that 470 oxidative damage induces a rapid and distinct transcriptional response, characterised by abrupt 471 arrest of RNAPII within the gene body. Even more unexpectedly, we find that these RNAPII 472 complexes restart transcription from their site of arrest. This is in contrast with the UV response, 473 where transcriptional recovery relies on new transcription initiation and proceeds in 5’ to 3’ 474 wavefront from the TSS 7,10,13,47,48. The difference most likely stems from the type of damage 475 induced. While UV induced bulky DNA lesions act as roadblocks for RNAPII, and lesion 476 stalled RNAPII itself plays a central role in recruiting of repair factors such as CSB 5,9,13, this 477 is not the case following oxidative damage. Unlike the UV response, RNAPII is neither 478 ubiquitinated and subsequently degraded, nor are NER factors such as XPC or CSB required 479 for transcriptional restart. Indeed, high doses of hydrogen peroxide for which transcription fails 480 to recover within a few hours , have been associated with RNAPII ubiquitination 8,57. We 481 therefore speculate that RNAPII ubiquitination following oxidative stress serves only as a 482 secondary response , facilitating removal of persistently stalled RNAPII from the DNA 483 template. At lower doses of oxidative damage, RNAPII restart s from within the gene body , 484 enabling a less disruptive and faster transcriptional recovery. Compared to the UV response, 485 the kinetics of RNAPII arrest and restart following oxidative damage are extremely rapid. 486 Whereas the UV -induced transcriptional repression is strongest a few hours after the initial 487 exposure and recovery occurs 6 -24 hours post-UV 13, oxidative damage-induced repression is 488 immediate and recovery of overall transcription levels are restored within 1-2 hours of recovery. 489 We suspect that th is rapid response results from arrested RNAPII complexes being able to 490 restart from within the gene body, thereby bypassing the need to physically remove RNAPII 491 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 6, 2025. ; https://doi.org/10.1101/2025.11.06.686939doi: bioRxiv preprint 21 complexes from the DNA template. This aligns with our observation that RNAPII complexes 492 arrested due to oxidative damage are differently configured in terms of their associated factors. 493 Similarly to the heat shock response , we observed an accumulation of NELF -paused 494 RNAPII during the transcriptional arrest and early recovery phase 36. However, unlike the heat 495 shock response, oxidative stress did not result in NELF condensate formation, nor was the 496 presence of NELF required for the transcriptional repression 36. In addition to its global role in 497 heat shock-mediated transcriptional downregulation, NELF is involved in local transcriptional 498 silencing through its recruitment to sites of DSBs in a PARylation-dependent manner 28. These 499 observations led us to question whether NELF might be recruited to RNAPII within the gene 500 body in response to oxidative stress. However, NELF ChIP -seq data excluded this possibility. 501 Instead we find that NELF-mediated pausing is not per se required for the transcriptional 502 response although it is impacting the pause site of RNAPII, leading to a transition of RNAPII 503 pausing from the 1 st pause site to the 2 nd pause site (+1 nucleosome) 51. These observations 504 again emphasise how global transcriptional repression can be achived through different 505 mechanisms depending on the type of cellular stress and the nature of DNA damage induced. 506 Bulky DNA lesions induced by UV are known to block RNAPII progression both in 507 vitro and in vivo, while non-bulky DNA les ions such as those typically induced by oxidative 508 damage have generally been considered not to induce RNAPII stalling 17-19. SSBs however have 509 been reported t o stall RNAPII elongation in vitro 21. This is most likely not due to the DNA 510 break itself, but rather to the aberrant 3’ termini of the gap impeding RNAPII progression 58. 511 Our data strongly suggest that RNAPII gene body arrest is dependent on DNA damage, which 512 is rapidly resolved through BER and SSBR, involving the DNA glycosylase OGG1 and 513 XRCC1. XRCC1 serves as a scaffold protein to facilitate recruitment of DNA repair enzymes 514 to SSB either generated directly enabling SSBR or to SSB generated as BER intermediates 59. 515 Consequently, lack of XRCC1 leads to a delay in SSBR and is associated with neuronal 516 dysfunction and neurological disease s, highlighting its important role in the maintenance of 517 genome stability 59,60. However, the molecular interplay between SSBR and transcription has 518 until now not been well -studied. Here we describe how the lack of XRCC1 leads to strong 519 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 6, 2025. ; https://doi.org/10.1101/2025.11.06.686939doi: bioRxiv preprint 22 defects in RNAPII gene body recovery consistent with persistent occurrence of SSB within 520 genes. Accordingly, cells lacking XRCC1 completely fail to recover transcription even at 4 521 hours after hydrogen peroxide treatment, as a consequence of sustained transcriptional 522 repression within genes. Similar to XRCC1, OGG1i treatment results in very dramatic and 523 prolonged gene body RNAPII arrest which is sustained at the 1 -hour recovery timepoint. In 524 contrast to lack of XRCC1, OGG1i -treated cells do eventually recover nascent transcription 525 levels at the 2–4-hour recovery timepoint. We suspect this difference is due to additional DNA 526 glycosylases such as NEIL1, NEIL2 and MUTYH 61,62. Indeed, NEIL1 and to a lesser extent 527 NEIL2 have been shown to function as backup BER glycosylases working both on 8OG and 528 further oxidation products such as hydantoin lesions upon OGG1 TH5487 inhibitor treatment 529 or OGG1 depletion 61. This also fits well with mice knockout phenotypes, where individual 530 BER glycosylase knockouts display only a moderate increase in mutation frequencies, while 531 double or triple knockouts, such as OGG1/MUTYH knockouts display strong phenotypes 532 characterised by high susceptibility to cancer and shortened lifespan 63. In addition, alternative 533 repair pathways that might compensate for the lack of OGG1 -mediated repair, such as NER , 534 could facilitate a secondary response in case of persistently stalled RNAPII complexes 44. In 535 contrast, XRCC1 is required both for SSBR and downstream in the BER pathway to resolve 536 SSB intermediates 59. We speculate that this explains why lack of XRCC1 leads to sustained 537 transcription repression, as lack of XRCC1 cannot easily compensated for by alternative repair 538 pathways. 539 Oxidative damage -induced transcriptional repression coincides with high levels of 540 PARylation which quickly decrease as transcription recovers. Sustained PARylation achieved 541 through PARG inhibition maintains transcriptional repression, highlighting the dynamic 542 regulation of the transcriptional response through PARylation. Traditionally , PARylation has 543 been considered as a recruitment signal for DNA repair factors and plays a crucial role in SSBR 544 through recruitment of XRCC1 23. Lack of XRCC1 has also been suggested to promote PARP 545 ‘trapping’ on DNA leading to a transcription block through maintained PARP1 binding to sites 546 of DNA lesions 56. However, while we find that cells lacking XRCC1 fail to recover gene body 547 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 6, 2025. ; https://doi.org/10.1101/2025.11.06.686939doi: bioRxiv preprint 23 transcription following oxidative stress, PARPi treatment exerts a milder effect on gene body 548 transcription, suggesting that the presence of the oxidative DNA lesions themselves rather than 549 PARylation-induced effects are the primary cause of gene body RNAPII arrest. Instead, we find 550 that PARylation controls the release of early RNAPII complexes from the TSS already at the 551 15 min recovery timepoint. Notably, if we had relied solely on quantification of overall levels 552 of nascent transcription following PARPi treatment , we would have concluded that PARPi 553 treatment completely reverses the transcription inhibition as early as 15 min recovery. 554 However, nascent transcriptomics clearly show that the effect of PARPi is more complex and 555 spatially restricted to early RNAPII elongation complexes. Intriguingly, the effect of PARP 556 inhibition of early RNAPII elongation complexes is independent of BER and SSBR DNA repair 557 factors such as OGG1 or XRCC1, as we show that PARPi treatment ‘overrides’ continued 558 transcriptional repression and promotes release of early elongation complexes even in cells 559 lacking XRCC1 or following treatment with OGG1i. Based on our results, we suggest a spatial 560 role of PARP1 which has hitherto been mistakenly overlooked. 561 The question remains h ow PARylation restricts early RNAPII elongation complexes . 562 Both NELF and the pTEFb component CycT1 have been reported as direct PARylation targets 563 14,64. PARylation of NELF has been suggested to promote elongation 64, whereas CycT1 564 PARylation induced by MNNG (N-methyl-N'-nitro-N-nitrosoguanidine) disrupts CycT1 phase 565 separation, correlating with reduced RNAPII hyperphosphorylation in vitro 14. In the case of 566 CycT1, this would suggest an inhibitory effect on RNAPII pause release. However, it remains 567 unknown if MNNG induce s a spatially separated transcriptional response like hydrogen 568 peroxide. Another possibility is that PARylation of proteins associated with early RNAPII 569 complexes signals recruitment of additional factors such as NuMA which has been suggested 570 to impact oxidative damage repair through a yet undefined mechanism 55. Thus, elucidation of 571 the precise molecular mechanism of PARylation dependent restriction of early RNAPII 572 elongation remains an exciting topic for future research and will undoubtedly provide new 573 insight into how genotoxic DNA such as that induced by oxidative stress elicit s a highly 574 regulated transcriptional response. 575 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 6, 2025. ; https://doi.org/10.1101/2025.11.06.686939doi: bioRxiv preprint 24 As PARP inhibition is not able to prevent the initial repression of transcription, we 576 suggest that the immediate and strong transcriptional attenuation reflects arrest of RNAPII by 577 oxidative DNA lesions and RNAPII -blocking repair intermediates, occurring in a PARP -578 independent manner. This fits well with our results showing that the immediate transcriptional 579 repression dramatically impacts RNAPII gene body activity. This is analogous to the situation 580 for UV-induced bulky lesions, where the DNA damage itself directly impacts the transcription 581 machinery. However, due to the type of DNA lesion and consequently how repair and RNAPII 582 stalling are dealt with, the two types of genotoxic insults led to completely distinct outcomes in 583 terms of transcriptional response. These differences are manifested in both the kinetics of the 584 transcriptional response and factor dependencies , underscoring the unique nature of 585 transcriptional regulation in response to oxidative damage. 586 587 While the coupling between transcription and nucleotide excision repair (NER) is well 588 established, known as TC-NER, and relies on DNA damage recognition via RNAPII stalling 589 within actively transcribed genes, the interplay between BER and transcription remains largely 590 unexplored. A defining feature of TC -NER is its preferential repair of bulky DNA lesions on 591 the transcribed strand of active genes 7,9,13. Our findings suggest that transcriptional recovery 592 following oxidative damage is closely linked to DNA repair mediated by BER and SSBR, 593 pointing to the possible existence of transcription -coupled BER (TC-BER) and transcription-594 coupled SSBR (TC-SSBR). It has been hypothesised that TC-BER may operate analogously to 595 TC-NER, favoring repair within transcription units. While some studies support preferential 596 BER of oxidative lesions in the transcribed strand in reconstituted transcription assays 65, others 597 report no strand bias in the repair of 8 -oxoguanine (8OG) lesions 66, leaving the mechanistic 598 basis of such coupling unresolved. Unlike TC-NER, we do not observe direct recruitment of 599 DNA repair factors to RNAPII. This absence may reflect a more transient interaction between 600 the transcription machinery and repair complexes, or the inherently rapid kinetics of BER and 601 SSBR compared to TC -NER 13,59. This underscores the need for further investigation into the 602 nature and dynamics of transcription-coupled repair beyond NER. 603 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 6, 2025. ; https://doi.org/10.1101/2025.11.06.686939doi: bioRxiv preprint 25 604 In summary, our findings demonstrate that oxidative damage triggers a dramatic and 605 immediate global repression of nascent transcription. This transcriptional attenuation is 606 characterised by increased stalling of early RNAPII complexes near the TSS, as well as RNAPII 607 arrest within the gene body. A distinctive feature of this response is that gene body –arrested 608 RNAPII can resume transcription from within the gene body itself, enabling rapid recovery. 609 Although NELF -mediated regulation influences promoter -proximal RNAPII pausing and 610 positioning, NELF is not essential for the transcriptional repression. Strikingly, we find that 611 separate mechanisms govern the transcriptional response of early RNAPII elongation 612 complexes and gene body–arrested RNAPII. Early RNAPII stalling is regulated by PARylation, 613 whereas recovery within the gene body depends on DNA repair mediated by BER and SSBR, 614 suggesting that oxidative DNA lesions in vivo impede RNAPII progression. Thus, our 615 temporally and spatially resolved characteri sation of the nascent transcriptional response to 616 oxidative damage reveals previously unrecognized layers of genotoxic-induced transcriptional 617 regulation. These mechanisms serve to transiently repress global transcription through RNAPII 618 arrest, orchestrated by PARylation and DNA damage repair pathways. 619 620 621 622 623 624 625 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 6, 2025. ; https://doi.org/10.1101/2025.11.06.686939doi: bioRxiv preprint 26 Author contributions 626 L.H.G. and Q.A.T conceived the project. Q.A.T., L.H.G., L.W., E. L., H. K.M. I., S.K., D.L.M., 627 and N.N.M. performed experiments. Q.A.T performed main experiments. L.H.G performed 628 proof-of-concept experiments, assay establishment and initial TT chem-seq experiments. L.W. 629 assisted in generating NELFC degron cell line, performed NELF microscopy experiments as 630 well as UV and heat shock experiments. E.L. performed NELF and RNAPII ChIP -seq 631 experiments. H.K.M.I. performed RNAPII Dsk2 pull -outs. S.K. performed initial EU assays. 632 D.L.M performed in vitro RNAPII transcription assays. N.N.M. generated XPC knockout cell 633 line. Q.A.T performed bioinformatic analysis , with input from H.L. L.H.G. wrote the 634 manuscript with input from Q.A.T . All authors read and approved the final version of the 635 manuscript. 636 637 Acknowledgments 638 This work was supported by grants to L.H.G. from the European Research Council (ERC 639 Agreement, TranscriptStress, 101076758), a Hallas -Møller Emerging Investigator Grant from 640 the Novo Nordisk Foundation (NNF20OC0059959), and a Sapere Aude Research Leader 641 Programme grant from the Independent Research Fund Denmark (0165 -00092B). Research at 642 the Center for Gene Expression (CGEN) is funded by the Danish National Research Foundation 643 (DNRF166). Q.A.T. was supported by the Lundbeck Postdoctoral Fellowship Programme (LF 644 R380-2021-1284). N.N.M. was supported by the EMBO Postdoctoral Fellowship (ALTF 911-645 2020). Mass spectrometry -based proteomics analyses were performed by the Proteomics 646 Research Infrastructure (PRI) at the University of Copenhagen (UCPH), supported by the Novo 647 Nordisk Foundation (grant agreement number NNF19SA0059305). We thank members of the 648 CGEN/CPR/reNEW Sequencing Facility, University of Copenhagen, for expert technical 649 assistance, and Jesper Q. Svejstrup for feedback on the manuscript. 650 651 652 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 6, 2025. ; https://doi.org/10.1101/2025.11.06.686939doi: bioRxiv preprint 27 Declaration of interest 653 654 The authors declare that they have no competing financial interests. 655 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 6, 2025. ; https://doi.org/10.1101/2025.11.06.686939doi: bioRxiv preprint 28 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681

Resource availability Plasmids used in this study have been deposited in Addgene. Any additional information required to reanalyse the data reported in this paper is available from the lead contact upon request. Experimental Model Flp-InTM T-REx HEK293 cells (HEK293, R78007, ThermoFisher Scientific, human embryonic kidney epithelial, female origin) were cultured at 37 oC with 5% CO 2 in high glucose DMEM (Biowest, L0104) supplemented with 10% v/v FBS (Gibco, 10270106), 100 U/mL penicillin, 100 μg/mL streptomycin (Gibco, 10378016) 2 mM L-glutamine. Cells were routinely passaged 2-3 times a week. HEK293 CSB KO cells have previously been described 67. All cell lines were confirmed to be mycoplasma-free. Oxidative stress induction treatments were carried out for 15 min (or 30 min for EU-labelling) with direct addition of freshly prepare 1000x concentrate to cell culture media, fresh H2O2 (maximum 3 month old kept at 4oC) to final concentration of 1 mM H 2O2 (Sigma, H1009) or freshly resuspended Menadione sodium bisulfite (MND, Sigma, M5750) in water to final concentration of 0.25 mM. After treatment the media was discarded, and cells were washed once with PBS pH = 7.4 (Gibco, 10010056) and replenished with fresh 37oC cell culture media. Inhibitors treatments were carried out by direct addition to the cell culture media with the following concentration 100 μM DRB (Sigma, D1916) or/and 10 μM Olaparib (PARPi, MedchemExpress, Cat#HY-10162) or/and 10 μM TH5487 (OGG1i, MedchemExpress, Cat#HY-125276) and 10 μM PDD 00017273 (PARGi, MedchemExpress, Cat#HY-108360). DRB inhibition was carried out 3.5 hours prior to other treatments. Inhibition 682 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 6, 2025. ; https://doi.org/10.1101/2025.11.06.686939doi: bioRxiv preprint 29 of PARPi, PARGi and OGG1i were preformed 1 hour prior to other treatments unless specified 683 otherwise. Cells treated with inhibitors were subjected to oxidative stress by direct addition of 684 oxidative reagents to cell culture media. PARPi, OGG1i and PARGi inhibitors were then 685 replenished together with cell culture media during stress recovery. DRB was either replenished 686 or not depending on experiments. UV irradiation was performed with an exact dose of 20 J/m2 687 as in 10. Heat shock (HS) was performed by incubating cells at 43°C with pre-warmed media 688 for 30 min or 1 hour. 689 690 Plasmid construction and generation of stable cell lines 691 Generation of stable cell lines 692 HEK293 Flp-In™ T-REx™ RPB3-3xFLAG cells (RPB3-FLAG) and U2OS Flp-In™ T-REx™ 693 GFP-NELFE were generated by stable integration of the pDEST/TO/FRT/RPB3 -3xFLAG or 694 pFRT/TO/GFP-NELFE plasmids respectively. The human RPB3 (POLR2C) cDNA was 695 amplified from pIRES -puro-RPB3-HA plasmid (gifted from Jesper Svejstrup) and NELFE 696 cDNA were PCR amplified with Q5® High -Fidelity DNA Polymerase (NEB) from HEK293 697 WT cells genomic DNA extracted with Quick -DNA™ Miniprep Plus Kit (Zymo Research). 698 Respective cDNA fragments were inserted through digestion and ligation into pENTR4 699 containing attB flanking sequences. RPB3 cDNA (in pENTR4 -RPB3) and NELFE cDNA (in 700 pENTR4-NELFE) were tagged through insertion into respectively pDEST /TO/3xFLAG and 701 pDEST/GFP/NELFE destination vectors using Gateway ™ LR Clonase II (Invitrogen, 702 ThermoFisher Scientific), resulting with in -frame C -terminal fusion of RPB3 to 3xFLAG 703 (3xDYKDDDDK) tag and N -terminal fusion of NELFE with EGFP. HEK293 Flp -In cells in 704 DMEM without antibiotic complementation were transfected for 48 hours with 900 ng insertion 705 plasmid pOG44 (ThermoFisher Scientific) and 100 ng donor plasmid pDEST/TO/RPB3 -706 3xFLAG or pDEST/GFP/NELFE using Lipofectamine ™ 3000 (Invitrogen, Therm oFisher 707 Scientific; Cat#L3000015), following the manufacturer’s protocol. After 24 hours, the 708 transfection cell culture media was replaced with fresh culture media for 24 hours, followed by 709 one passage prior to selection with hygromycin (100 μg/mL), zeocin (100 μg/mL), and 710 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 6, 2025. ; https://doi.org/10.1101/2025.11.06.686939doi: bioRxiv preprint 30 blasticidin (15 μg/mL) for 10 days. Single colonies were manually picked and passaged 2–3 711 times prior to testing. Tagged cell lines were treated with doxycycline (100 ng/mL) for 24 hours 712 to induce RPB3-3xFLAG or GFP-NELFE confirmed by western blotting with respectively anti-713 FLAG (Sigma, Cat#F1804) and anti-GFP (Abcam, Cat#ab290) antibodies. 714 715 Generation of knock-out cell lines 716 HEK293 Flp -In™ T-REx™ XRCC1 knockout cells (XRCC1 KO) were generated by 717 CRISPR/Cas9-mediated genome editing. The pDG458 -XRCC1/sg1&2 plasmid was 718 constructed by cloning two guide RNAs (gRNAs: 5′-TGCAGGACACGACATGGCGG-3′ and 719 5′-AGCCCACGACGTTGACATGC-3′) into BbsI-digested pDG458 containing Cas9 fused to 720 EGFP. This plasmid was transfected into HEK293 cells using Lipofectamine ™ 3000 for 48 721 hours according to manufacturer’s protocol. GFP-positive cells were then sorted using a FACS 722 system (FACSMelody ™, BD Biosciences) into two 96 -well plates. After 2 –3 passages, 723 individual clones were screened by western blotting to confirm XRCC1 knockout. Two 724 validated knockout clones (KO#3 and KO#8) were used in subsequent experiments. HEK293 725 Flp-In™ T-REx™ XPC knockout cell s (XPC KO) line was generated by CRISPR/Cas9 -726 mediated genome editing. The gRNA sequence 5’ -CAACATGGCTCGGAAACGCG-3’ was 727 ligated into the vector pSpCas9(BB) -2A-Puro (pX459) V2.0 (Addgene, #48138) to generate 728 the pX459_XPC_gRNA1 plasmid. HEK293 were transfected with the plasmid 729 pX459_XPC_gRNA1 using Lipofectamine TM 2000 for 48 hours following manufacturer 730 instructions. Transfected cells were passaged and treated with puromycin (2 μg/mL) for 72 731 hours. Cells were then washed twice with PBS, detached with trypsin and quantified to prepare 732 single cell dilutions in a 96 well plate. Two weeks after, colonies originated from a single cell 733 were evaluated for XPC knock out by western blot using the anti-XPC antibody (D1M5Y, Cell 734 Signalling, Cat#14768). Flp-In™ T-REx™ CSB knockout cell s (CSB KO) were previously 735 described.10 736 737 Generation of NELFC degron cell line 738 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 6, 2025. ; https://doi.org/10.1101/2025.11.06.686939doi: bioRxiv preprint 31 HEK293 Flp-In™ T-REx FKBP12F36V-NELFCD cells (NELFCD-degron) were generated via 739 CRISPR double break following homology recombination knock -in of a C -terminal FKBP12 740 degron tag, along with a cleavable puromycin resistance gene, into the endogenous HEK293 741 NELFCD locus with the donor plasmid pDONR -HA1/NELFCD/FKBP/PURO/HA2. To 742 generate the donor plasmid, NELFCD cDNA sequence and homology arms (HA1 and HA2; 743 >500 bp each) were amplified by PCR with Q5® High -Fidelity DNA Polymerase from 744 HEK293 genomic DNA. The degron and puromycin resistance cassette were amplified from 745 the pCRIS -PITChv2-Puro-dTAG (BRD4) plasmid (Addgene #91793). All fragments were 746 assembled into a donor plasmid (pBluescript II SK(+)) using Gibson assembly with the 747 NEBuilder® HiFi DNA Assembly Kit (NEB). To created double strand breaks and allow 748 homology recombination targeted CRISPR components were delivered to cells, the pDG458 -749 NELFCD-sg1&2 plasmid was constructed by cloning two gRNAs (5′ -750 ATTTGCAGTGAGCTTTAACG-3′ and 5′ -TGAAAGGGTTTTTCCACAAC-3′) into BbsI -751 digested pDG458 (Addgene, #100900). HEK293 cells were then co-transfected with pDONR-752 HA1/NELFCD/FKBP/PURO/HA2 and pDG458 -NELFCD-sg1&2 using Lipofectamine TM 753 3000 with manufacturer protocols in antibiotic -free cell culture media. After 24 hours, cells 754 were selected with puromycin (2 μg/mL) for 48 hours. For selection of positive clones, colonies 755 were picked after selection passaged 2 -3 times and genomic DNA extracted with 756 QuickExtract™ DNA Extraction Solution (Lucigen , Cat# LGCQE09050). A PCR fragment 757 overlapping tagged region was amplified clones were selected based on fragment size shift and 758 further confirmed by western blotting with NELFCD -specific antibodies. To further validate 759 the degron system, homozygous knock -in clones were treated with 250 nM dTAG -v1 (Tocris 760 Bioscience, Cat#6914) for 1 hour to induce in vivo NELFCD degradation. Depletion of 761 NELFCD protein and reduction of chromatin-bound NELFA protein levels were confirmed by 762 western blotting. 763 764 EU-labelling assay 765 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 6, 2025. ; https://doi.org/10.1101/2025.11.06.686939doi: bioRxiv preprint 32 Approximately 7,000 HEK293 cells were seeded in 96-well Greiner CELLSTAR® 96 well plate 766 (ThermoFischer Scientific, Cat#M0562) coated for 15 min at room temperature with poly-D-767 Lysine (Gibco, Cat#A3890401) and washed twice with PBS. For treated samples, cell culture 768 media was replaced by media complemented with indicated concentration of H2O2 or MND for 769 30 min and washed with PBS and replenished with new 37oC cell culture media. Nascent RNA 770 was pulse-labelled for 30 min with addition of cell culture media complemented with 5 -771 Ethynyluridine ( EU, Jena Bioscience, Cat#CLK -N002, final concentration 750 μM). For 772 treated samples (no recovery) both oxidative reagents and EU were added simultaneously. 773 Labelled cells were immediately fixed with 3.7 % formaldehyde diluted in PBS for 45 min in 774 the dark. The cells were then washed three times in PBS and permeabilized with 0.5 % Triton 775 X-100 in PBS for 30 min at room temperature in the dark. Nascent RNA was then visualized 776 by click-it chemistry, the EU-incorporated RNA was labelled in the dark for 2 hours with a mix 777 of 5 μM Alexa Fluor TM 488 Azid (ThermoFischer Scientific , Cat# A10266), 4 mM copper 778 sulfate (Sigma-Aldrich) and 100 mM ascorbic acid (Sigma -Aldrich). Cells were then washed 779 twice with PBS and incubated for 30 min at room temperature in the dark with 2 μg/mL DAPI 780 (Sigma-Aldrich, Cat#D9742) diluted in PBS. Cells were washed twice with PBS and kept in 781 the dark at 4oC in PBS for up to 48 hours. A total of 25 images per well were acquired with the 782 ScanR acquisition system (Olympus) controlling a motorized Olympus IX -81 wide -field 783 microscope and analysed and quantified with the ScanR Analysis software. 784 785 Total RNA extraction 786 Total RNA was extracted by adding TRIzol ™ Reagent ( ThermoFisher Scientific , 787 Cat#15596026) directly to the cell monolayer after cell culture media removal. Chloroform was 788 added to the cell/TRIzol mixture at a 1:5 volume ratio, followed by thorough mixing for 30 789 seconds and centrifugation at 12,000 g for 15 min at 4 °C. The upper aqueous phase was 790 collected and further purified by adding 1 volume of chloroform/isoamyl alcohol (24:1; Sigma-791 Aldrich, Cat#C0549), followed by vortexing and centrifugation at 12,000 × g for 5 min at 4 °C. 792 The resulting upper phase, containing total RNA, was precipitated by adding one volume of 793 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 6, 2025. ; https://doi.org/10.1101/2025.11.06.686939doi: bioRxiv preprint 33 isopropanol (Sigma -Aldrich, Cat#I9516) and 30 ng/mL GlycoBlue ™ (Invitrogen, 794 ThermoFisher Scientific, Cat#AM9515) and incubated at room temperature for 20 min. The 795 RNA was pelleted by centrifugation at 20,000 g for 30 min and washed once with ice-cold 85% 796 ethanol. The final RNA pellet was resuspended in RNase-free water. 797 798 Nascent RNA dot blot 799 Dot blots were used to detect global levels of nascent transcription based on 4-thiorudine (4sU, 800 Glentham Life Sciences, Cat#GN6085) RNA incorporation. Dot blot were performed as 801 described in 38 with minor changes. Approximately 500,000 cells were seeded in 6-well plates. 802 The next day, 4sU (final concentration 1mM) was added directly to the tissue culture for 15 803 min coordinated with H2O2 treatment and recovery time points, labelling was then arrested with 804 addition of 0.5 mL TRIzol. Total RNA was extracted and 3 μg of 4sU labelled total RNA was 805 biotinylated for 2 hours in the dark with 50 μL of 0.1 mg/ml EZ -linkTM HPDP-Biotin 806 (ThermoFischer Scientific, Cat#21341 in DMF) with 3 μL of biotin buffer (833 mM Tris -HCl 807 pH 7.4, and 83.3 mM EDTA) for a final reaction volume of 300 μL. Unreacted biotin-linker 808 was removed from RNA samples by addition of 1.1 volumes of ROTI ®Aqua-P/C/I (Roth, 809 Cat#985.1) vortexed, precipitated with 1.1 isopropanol followed by pellet centrifugation, 810 ethanol washes and resuspension in RNAse-free H2O. Purified RNA sample were resuspended 811 in 10 μL and applied to a Hybond -N+ membrane (Cytiva, Cat#RPN203B ) in a dot blot 812 apparatus (BioRad). The immobilized RNA was UV-crosslinked to the membrane with UV-C 813 0.24 J/cm2 and the membrane incubated in blocking solution (PBS pH = 7.4; 10% SDS; 1 mM 814 EDTA) for 20 min at room temperature followed by at 15 min incubation with a 1:50,000 815 dilution of 1 mg/mL streptavidin -horseradish peroxidase (HRP) ( ThermoFisher Scientific 816 Cat#N100) in blocking solution. The membrane was then washed twice in blocking solution 817 buffer for 10 min, followed by two washes in dot blot wash buffer I (PBS pH = 7.4; 1% SDS) 818 for 10 min each and two washes in dot blot wash buffer II (PBS pH = 7.4 ; 0.1% SDS) for 10 819 min each. The biotin -bound HRP was visualized with 2 min incubation of 1:4 diluted ECL 820 detection reagent (BioRad) on an ChemiDoc Imaging System (BioRad). As a loading control, 821 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 6, 2025. ; https://doi.org/10.1101/2025.11.06.686939doi: bioRxiv preprint 34 the membrane was stained with methylene blue solution (0.5 M sodium acetate; 0.5 % 822 methylene blue) for 10 min, then de-stained by several washes in water. 823 824 TTchem-seq 825 TTchem-seq was performed as described previously with minor modifications 38. In brief, 10 cm 826 dishes with 80% confluent cells RNA were labelled in vivo for 15 min with 4sU by adding it 827 directly to the tissue culture media to a final concentration of 1 mM. Labelling was terminated 828 by addition of TRIzol and total RNA was extracted by TRIzol/chloroform extraction followed 829 by isopropanol precipitation. Per samples 80 -100 μg of total RNA resuspended in 100 μL 830 RNAse-free H 2O was complemented with 1 % ( relative to total RNA concentration) of S. 831 cerevisiae 4tU-labelled RNA spike-in. Total RNA was fragmented by adding 20 μL of freshly 832 prepared 1 M NaOH per sample, thoroughly mixed and incubated for 40 min on ice to < 400 nt 833 fragments. RNA fragmentation was terminated by addition of 80 μL 1 M Tris -HCl pH = 6.8 834 and samples were immediately purified with 1 volume of ROTI ®Aqua-P/C/I (ROTH, Cat# 835 X985.1) extraction and precipitated with isopropanol. Resuspended and purified 4sU labelled 836 RNA in RNAse -free water was biotinylated in 10 mM Tris -HCl pH = 7.4; 1 mM EDTA 837 complemented with 0.5 mg/ μL MTSEA biotin-XX linker (Biotum, Cat# BT90066) for 30 min 838 in the dark at room temperature. Biotinylated RNA was immediately purified by ROTI®Aqua-839 P/C/I extraction and isopropanol precipitation and resuspended in 50 μL RNAse-free water. To 840 enrich for 4sU labelled RNA, the purified RNA was denatured at 65°C for 10 min, with 5 min 841 on ice was used, and isolated with μMACS Streptavidin Kit (Miltenyi, Cat#130 -074-101). A 842 total of 100uL μMACS Streptavidin MicroBeads were added to denatured 4sU total RNA and 843 incubated for 15 min at room temperature with agitation. Beads bound to 4sU biotinylated RNA 844 were applied to magnetic μColumns and transferred to the magnetic field of the μMACS 845 magnetic separator (Miltenyi). The columns were washed t wice with 1 mL 55oC pre-warmed 846 pull-down wash buffer (100 mM Tris -HCl pH 7.4, 10 mM EDTA, 1 M NaCl and 0.1 % (v/v) 847 Tween 20) and eluted with two consecutive washes with freshly prepared 100 mM DTT. 848 Biotinylated RNA was further cleaned up from DTT elution by ROTI®Aqua-P/C/I extraction 849 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 6, 2025. ; https://doi.org/10.1101/2025.11.06.686939doi: bioRxiv preprint 35 and isopropanol precipitation and resuspended in RNAse -free H 2O. Concentration of 4sU 850 enriched RNA was measured by Qubit (ThermoFischer Scientific , Cat# Q32852) using the 851 RNA HS assay kit and size of fragmented RNA assessed by TapeStation using RNA 852 ScreenTape Assay for TapeStation Systems (Agilent, Cat# 5067-5576). 30 ng of purified 4sU 853 biotinylated RNA was used to prepare libraries with NEBNext® Ultra ™ II RNA library prep 854 kit (NEB, Cat#E7760) according to manufacturer’s protocol (protocol 5 for fragmented RNA) 855 with ligation of two barcodes of 8 nt and 11 nt UMIs (NEB, Cat#E7416). Libraries were 856 amplified using 8 cycles of PCR and sequenced as single -end 68 bp reads on a NextSeq 2000 857 platform (Illumina). 858 859 Cell growth assays 860 Approximately 5,000 HEK293 cells per well were seeded in triplicates in 96-well plates, pre-861 coated with poly-D-Lysine (Gibco) for 15 min and washed twice with PBS. The following day 862 cell treatments were performed in 96 -wells by addition of cell culture media complemented 863 with different concentration of H 2O2 or MND for 15 min. The complemented media was 864 removed, and cells were washed once with PBS and fresh 37 oC cell culture media was 865 replenishment. Growth was monitored using an IncuCyte S3 Live -Cell Analysis System 866 (Sartorius). Images of live cells were captured every 2 hours for up to 6 days. Images were 867 analysed in the IncuCuyte S3 image analysis software, with an output of cell confluency (%). 868 869 Cell fractionation and RNAPII immunoprecipitation 870 Cell pellets from one 15 cm plate were resuspended in 1 mL volume HE buffer (10 mM HEPES-871 NaOH pH=7.5; 10 mM KCl, protease and phosphatase inhibitors: Vanadate, Sodium fluoride, 872 and ß-glycerolphosphate) and incubated 15 min on ice. Nuclei were pelleted 15 min; 1,000 g 873 and the supernatant taken as the cytoplamic fraction. The nuclear pellet was resuspended in 500 874 ml NE buffer (20 mM HEPES -NaOH pH = 7.9; 150 mM NaCl; 0.05% NP40; 10% glycerol; 875 protease and phosphatase inhibitors fresh). Chromatin were pelleted 15 min; 20,000 g and the 876 supernatant taken as the nucleoplasmic fraction. Cytoplasmic and nucleoplasmic fractions were 877 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 6, 2025. ; https://doi.org/10.1101/2025.11.06.686939doi: bioRxiv preprint 36 pooled as soluble fraction. Chromatin pellets were digested in 500 m L of CD Buffer (20 mM 878 HEPES-NaOH pH=7.9; 150 mM NaCl; 1.5 mM MgCl 2; 0.05% NP40; 10% glycerol; protease 879 and phosphatase inhibitors fresh; 250 U/ml DENARASE® (c-Lecta, Cat#20804) and incubated 880 on a rotating wheel for 1 hour at 4oC to digest nucleic acids and solubilize chromatin and release 881 chromatin-bound proteins. Digested chromatins were pelleted 15 min at 20,000 g and the 882 supernatant was taken as the enriched chromatin fraction. For immunoprecipitation, 883 DynabeadsTM Protein G Magnetic Beads (Invitrogen, ThermoFischer Scientific, Cat#10004D) 884 were washed and resuspended in 2 beads volumes of PBS + 0.02% Tween 20. The resuspended 885 beads were conjugated for 1 hour at room temperature with 1 mg 4H8 or 3E10 as indicated (for 886 capture of phosphorylated RNAPII complexes) or with anti-FLAG (Sigma, #1806, for capture 887 of total RNAPII complexes in a RPB3 -FLAG expressing cell line ). Non-antibody conjugated 888 beads were used for pull -out as negative control for mass spectrometry. Following antibody 889 incubation, unconjugated antibodies were removed by from beads by two washes with PBS + 890 0.02 % (v/v) Tween20. Protein concentrations of chromatin enriched fractions were measured 891 with Protein Assay Dye Reagent (BioRad, Cat#5000001) through absorbance measured at 495 892 nm and concentration was adjusted to the sample with the lowest concentration by addition of 893 CD buffer (constituting the ‘input’ sample for western blotting). Concentration -adjusted 894 chromatin fractions were added to antibody-conjugated beads (or unconjugated beads for 895 negative controls) and incubated for 3 hours at 4 oC with agitation. Samples were placed on a 896 magnetic rack and the supernatant was removed and kept (‘unbound’ lysate for western 897 blotting). Proteins bound to magnetic beads were then washed three times with IP wash buffer 898 (10 mM Tris -HCl pH=7.5; 150 mM NaCl; 0.05% NP40; 0.5 mM EDTA; protease and 899 phosphatase inhibitors fresh). Beads bound proteins were either extracted by boiling with 900 2xSDS sample buffer ( 10 mM Tris -HCl; pH = 6.8, 4% SDS, 0.2% bromophenol blue, 20% 901 Glycerol, 200 mM DTT) for western blotting (‘IP’ sample) or for mass spectrometry analysis, 902 the beads were washed two additional times with TBS and stored dry at -20oC until further 903 processing for proteomics. 904 905 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 6, 2025. ; https://doi.org/10.1101/2025.11.06.686939doi: bioRxiv preprint 37 Mass spectrometry 906 All samples for proteomic analysis were carried out in triplicate and samples were prepared in 907 parallel from the point of cell seeding. TBS washed frozen beads were thawed and incubated 908 for 30 min with elution buffer 1 (50 mM Tris -HCl pH = 7.5; 2 M Urea; 2 mM DTT; 20 μg/ml 909 trypsin) followed by a second elution for 5 min with elution buffer 2 (50 mM Tris -HCl pH = 910 7.5; 2 M Urea; 10 mM Chloroacetamide). Both eluates were combined and further incubated 911 at room temperature over-night. Tryptic peptide mixtures were acidified to 1% TFA and loaded 912 on Evotips (Evosep). Peptides were separated on 15 cm, 150 μM ID columns packed with C18 913 beads (1.9 μM) (Pepsep) on an Evosep ONE HPLC applying the ‘30 samples per day’ method 914 and injected via a CaptiveSpray source and 10 μm emitter into a timsTOF pro mass 915 spectrometer (Bruker) operated in PASEF mode. 916 917 Western blotting 918 For whole cell extracts, cell pellets were collected, resuspended in whole cell extraction buffer 919 (20 mM Tris -HCl pH = 7.5; 300 mM NaCl; 2.5 mM MgCl2; 1% NP40; 10% glycerol) with 920 freshly added DENARASE® (c-Lecta, Cat#20804), fresh protease and phosphatase inhibitors, 921 and then incubated rotating at 4°C for 1 hour followed by centrifugation at 10,000 g for 2 min 922 at 4°C. Supernatant was transferred to a new tube and protein concentration measured using 923 Bio-Rad protein assay reagent. For whole cell extracts , chromatin enriched fractions and IP 924 fractions, protein concentrations were adjusted to the sample with the lowest concentration with 925 the respective appropriate extraction buffer. 10 μL protein samples was complemented with 10 926 μL 2x SDS sample buffer and boiled at 98°C for 5 min. 20 μL protein sample were separated 927 by SDS-PAGE gel electrophoresis (6%, 8%, 10% or 15% acrylamide were used) in running 928 buffer (25 mM Tris pH = 8.3, 192 mM glycine, 0.1% SDS,) and transferred to nitrocellulose 929 membranes by wet transfer in transfer buffer (0.025 M Tris, 0.192 M glycine, 10% (v/v) 930 ethanol). Membranes were stained with Ponceau S solution for 10 min and washed with dH2O, 931 and a picture was acquired. Membranes were then blocked in blocking buffer (5% (w/v) milk 932 in TBS-T) and incubated with primary antibody (in 5% (w/v) milk in TBS-T) overnight at 4°C. 933 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 6, 2025. ; https://doi.org/10.1101/2025.11.06.686939doi: bioRxiv preprint 38 Membranes were then washed with TBS -T, and incubated with HRP -conjugated secondary 934 antibodies in blocking buffer for 1 hour at room temperature and then washed several times in 935 TBS-T. HRP signal detection was obtained using 1:1 ratio of each Clarity ECL Western 936 Blotting Substrates (BioRad, Cat#1705061) applied for 2min onto the membrane and 937 visualization by ChemiDoc Imaging System (BioRad). 938 939 ELCAP-seq 940 ELCAP-seq was carried out similarly to previously des cribed 39. RPB3 -FLAG cells were 941 seeded in 4x15 cm dished per samples; RPB3-FLAG was induced by adding a final 942 concentration of 100 ng/mL doxycycline for 16 hours directly to cell culture media. Induced 943 cells were treated with 1 mM H2O2 for 15 min or let recover after a PBS wash and replenishment 944 of warm cell culture media for variable recovery times. Immediately cells were collected, 945 centrifuged for 2 min at 1,000 g and the pellets frozen in liquid nitrogen. Frozen cells were then 946 immediately used for cell fractionation as described above. A volume of 25 μL DynabeadsTM 947 Protein G Magnetic Beads per sample were washed and conjugated to 1 mg anti-FLAG (Sigma, 948 #1806) as previously described in immunoprecipitation section. Protein concentrations from 949 chromatin enriched fraction were measured with Protein Assay Dye Reagent (BioRad) and 950 adjusted to the sample with the lowest concentration with CD buffer. Normalised chromatin 951 enriched fractions were then added to FLAG-conjugated magnetic beads for 3 hours at 4oC with 952 agitation. The beads were washed 4 times with ELCAP wash buffer (10 mM Tris -HCl pH = 953 7.5; 250 mM NaCl; 0.5% (v/v) NP -40; 0.5 mM EDTA; fresh protease inhibitor; protease 954 Inhibitors cocktail and fresh phosphatase inhibitors). Short RNA fragments protected by 955 RNAPII were extracted with the Quick -RNA MicroPrep kit (Zymo, Cat#R1050) according to 956 the manufacture’s instruction for short RNA extraction (17 -200 nt). Briefly, RNPAII bound 957 beads were incubated for 5 min at RT in extraction buffer (per 100 μL sample; 100 μL of 958 supplied Zymo RNA lysis Buffer, 100 μL of IP wash buffer,100 μL ethanol). RNA was isolated 959 using kit supplied Zymo-Spin™ IC columns following the manufacturer’s instructions. The 960 size distribution of extracted RNA was assessed with the Bioanalyzer Small RNA Analysis 961 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 6, 2025. ; https://doi.org/10.1101/2025.11.06.686939doi: bioRxiv preprint 39 system (Agilent). Library were prepared using 30 ng of short RNAs with NEBNext® Small 962 RNA library preparation kit (NEB, Cat# E7300) according to the manufacturer’s protocol with 963 the only modification being inclusion of a UMI containing adapter ; the i nitial ligation was 964 carried with the custom ligation adapter ‘5' -App-965 NNNNNNAGATCGGAAGAGCACACGTCT-3' (IDT) with the same ligation conditions as 966 manufacturer’s protocol. Libraries were amplified with 8 cycles of PCR and the whole library 967 was separated by size with Novex™ TBE Gels, 6% (Invitrogen, ThermoFischer Scientific, Cat# 968 EC6265BOX), stained with SYBRTMgold (Invitrogen, ThermoFischer Scientific, Cat# S11494) 969 and fragment size ranging from 147 nt to 190 nt were excised from the gel and purified 970 according to NEBNext® Small RNA library preparation kit protocol. Size selected libraries 971 were then sequenced as single-end run (68 bp reads) on a NextSeq2000 platform (Illumina). 972 973 Ubiquitinated proteins pull-down 974 DSK2 beads were prepared as previously described 40 with minor changes. In brief E. coli 975 expressing DSK2 were diluted in ice cold dilution buffer (1x PBS, 0.5% Triton X-100, protease 976 inhibitors) and sonicated (15 s ON, 30 s OFF, total 10 min ON time). HEK293 cells were either 977 treated with UV (20 J/m2) and let recover for 45 min after cell culture media replenishment or 978 with 1 mM H 2O2 for 15 min and recovery time as indicated. Cell pellets were resuspended in 979 TENT buffer (50 mM Tris-HCl pH = 7.4; 2 mM EDTA; 150 NaCl; 1 % Triton X-100; 2,5 µM 980 MgCl; protease and phosphatase inhibitors added fresh and 250 U/ml DENARASE® ) and 981 incubated 1 hour at 4 oC. Debris and non-solubilized proteins were removed by centrifugation 982 at 14,000 g for 10 min. Protein concentration of the resulting supernatant was determined with 983 Protein Assay Dye Reagent (Biorad). Lysates protein concentration was equalized before 984 adding 25 µL DSK2 beads and incubated overnight at 4oC. DSK2 beads enriched for ubiquitin 985 bound proteins were spun down at 700 g for 5 min and washed once in TENT buffer and once 986 with PBS (pH = 7.4). To release ubiquitinated proteins from the beads, samples were heated at 987 95oC for 5 min in 2x SDS sample buffer . The samples were separated from the beads by 988 centrifugation and 1% input and 20 % sample were used for western blotting. 989 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 6, 2025. ; https://doi.org/10.1101/2025.11.06.686939doi: bioRxiv preprint 40 990 Chromatin immunoprecipitation and sequencing (ChIP-seq) 991 For each sample, 80 % confluent HEK293 cells in a 15 cm plate were crosslinked with 1 % 992 formaldehyde for 15 min and stopped with 125 mM glycine for 5 min. Fixed cells were then 993 homogenized in cellular lysis buffer (5 mM Pipes pH = 8; 85 mM KCl; 0.5 % NP‐40). Nuclei 994 were then resuspended in nuclear lysis buffer (50 mM Tris pH = 8.1; 10 mM EDTA; 1 % SDS) 995 and sonicated to obtain DNA fragments between 50 bp and 500 bp (5 cycles; 30sec ON, 30sec 996 OFF, with Bioruptor pico). Samples were diluted 10 times in dilution buffer (0.01 % SDS; 1.1 997 % Triton X‐100; 1.2 mM EDTA; 16.7 mM Tris pH = 8; 167 mM NaCl). Samples were pre -998 cleared with 25 µL proteins A/G beads (ThermoFisher Scientific, Cat# 20422 ) and blocked 999 with 500 µg BSA for each 200 µg of chromatin for 2 hours at 4°C. A volume of 100 µL pre -1000 cleared chromatin was keep at -20°C for inputs and another volume of chromatin was then 1001 incubated overnight with either NELFE or total RPB1 (D8L4Y) antibodies at 4°C. 1002 Immunoprecipitation was performed using 35 µL proteins A/G beads (blocked with 500 µg 1003 BSA) for 2 hours at 4°C. Beads were then washed once with dialysis buffer (2 mM EDTA; 50 1004 mM Tris-HCl pH=8.1; 0.2% Sarkosyl), once wash buffer (100 mM Tris pH = 8.8; 500 mM 1005 LiCl; 1% NP‐40; 1% NaDoc) and once with TE (10 mM Tris pH = 8; 0.5 mM EDTA). 1006 Immunoprecipitated complexes and inputs were resuspended in TE and incubated with 1 µL of 1007 RNase A (10 mg/mL , ThermoFisher Scientific, Cat# EN0531) for 30 min at 37°C. Reverse 1008 crosslinking was performed by incubating samples overnight at 70°C with 4 µL SDS (10%) 1009 and incubation with ProteinaseK (10 mg/mL, ThermoFisher Scientifc Cat#AM2548) for 1 hour 1010 30 min at 45°C. DNA was purified by phenol/chlorophorm extraction followed by ethanol 1011 precipitation. Libraries were prepared with NEBNext® Ultra™ II DNA library prep kit (NEB, 1012 Cat#E7604) with 12 nt UMIs adaptors (NEB, Cat#E7395) according to manufacturer’s protocol 1013 and sequenced on a NextSeq2000 platform (Illumina). 1014 1015 In vitro transcription with pig thymus RNAPII 1016 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 6, 2025. ; https://doi.org/10.1101/2025.11.06.686939doi: bioRxiv preprint 41 RNAPII was purified from pig thymus as previously described 68. The elution of the 8WG16 1017 conjugated beads was dialyzed in dialyze buffer (20 mM HEPES-NaOH pH = 8; 150 mM NaCl; 1018 10 % Glycerol; 2 mM 𝛽-mercaptoethanol) and concentrated with Amicon® Ultra 4 centrifugal 1019 filter (50 kDa) (Millipore, Cat#UF8010). The elongation complex was assembled as previously 1020 described with minor modification 69. Briefly, 5 pmol of pre -annealed RNA:DNA (template 1021 strand) hybrid was mixed with an equimolar amount of pure RNAPII and incubated 20 min at 1022 30oC, followed by the addition of 10 pmol 5’ biotin -labelled non-template strand DNA for an 1023 additional 10 min. The assembled elongation complex was diluted with transcription buffer (20 1024 mM Tris -HCl pH= 7.5; 100 mM NaCl; 8 mM MgCl2; 10 μM ZnCl2; 10% glycerol). The 1025 transcription reaction was started by addition of 1 mM rNTPs for 2 min at 25 oC and stopped 1026 with 20 mM EDTA, 0.5 μL Proteinase K and 45% formamide. For treatment, H2O2 was added 1027 before the rNTPs at the indicated concentrations. Samples were incubated for 30 min at 50 oC 1028 followed by 10 min at 70 oC. RNA products were resolved in 8 M Urea, 16% polyacrylamide, 1029 1x TBE gel. Gels were scanned in a Typhoon scanner (Cytiva) to detect Cy2 fluorescence. 1030 1031 Immunofluorescence microscopy 1032 GFP-tagged NELFE U2OS cells were seeded (with 100 ng/mL doxycycline) at 60 -70% 1033 confluency onto coverslips coated with Poly-D-Lysine (Gibco) in 12-well plates. The next day 1034 cells were treated for 15 min or 30 min with 1 mM H 2O2 and for 30 min or 1 hour heat shock 1035 at 43oC. Cells were then fixed with 4% (v/v) formaldehyde for 15 min at room temperature, 1036 washed once with PBS, and mounted onto slides using VECTASHIELD® Antifade Mounting 1037 Medium with DAPI (VWR, Cat#VECTH-1200). GFP and DAPI fluorescence were visualised 1038 with an inverted confocal microscope (Zeiss LSM 900) 1039 1040 Image analysis 1041 For EU-labelling experiments, scanR software was used to quantify each isolated nuclei green 1042 signal relative to DAPI signal. For picture plotting, raw tiff images were upload to ImageJ’s 1043 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 6, 2025. ; https://doi.org/10.1101/2025.11.06.686939doi: bioRxiv preprint 42 Fiji, untreated samples (HEK293 or XRCC1 KO cells) were corrected for brightness and 1044 contrast; the same correction was then applied to all other images of the same cell line. 1045 1046 TTchem-seq analysis 1047 Raw fastq files for Read 1 were labelled with appropriate UMI reads using UMItools 70(v1.1.4) 1048 extract function followed by adapter trimming with cutadapt (v4.5) within trim_galore 1049 (v0.6.10) and read length filtering > 19nt. For the purposes of visualization, biological 1050 replicates UMI and adapter trimmed reads fastq files were merged and processed further 1051 similarly to individual replicates. In the case of XRCC1 KO, two individual clones (KO#3 and 1052 KO#8) were considered as biological duplicates. Merged and individual samples were mapped 1053 with STAR 71 (v2.7.11b) against a merged human -yeast genome assemblies ( Homo sapiens 1054 GRCh38 and Saccharomyces cerevisiae sacCer, Ensembl release 108) with options as default. 1055 Genome alignments BAM files were filtered for multimapped reads with samtools (v1.21) view 1056 function (-q 10). UMI duplicates were then removed with UMItools dedup function. A yeast 1057 gene-level count matrix was generated with Subread (v2.0.6) featureCounts 72 and used to 1058 calculate yeast size -factors with DEseq2 73 (v1.42.0) “estimateSizeFactors” function. Strand 1059 specific and spike -in normalised BigWig files were generated from strand -specific bam files 1060 using Deeptools74 (v3.5.5) bamCoverage function (--scaleFactor = 1/ yeast size-factor) with bin 1061 size of 25 nt ( -bs 25). TTchem-seq scaled plots (metagene profiles or heatmaps) were created 1062 from spike-in normalised BigWig files with DeepTools computeMatrix function (scale-regions, 1063 between –2 kb TSS to TES + 2 kb, bin size 50bp) using gene definitions from Ensembl (release 1064 108). All gene features were selected for protein coding genes non-overlapping (extended to 5 1065 kb upstream the TSS and 5 kb downstream of the TES) other coding genes (n = 11,593). ‘Long 1066 Genes’ derived from the later and were filtered with length > 90 kb (n = 2,217). Unscaled 1067 (reference-point) mode metagene profiles were centred on TSS with various distance around 1068 the TSS. TT chem-seq signal relative to u ntreated control ratio BigWig files were created using 1069 the DeepTools bigwigCompare function; the genome was partitioned to 100 bp bins for scaled 1070 metagene profiles and 50 bp bins for unscaled metagene profiles. Heatmaps were sorted by 1071 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 6, 2025. ; https://doi.org/10.1101/2025.11.06.686939doi: bioRxiv preprint 43 gene length (decreasing or increasing). Quantitative differential expression analysis of TTchem-1072 seq data for Supplementary Fig. 2d was done through DESeq2 with replacement of human 1073 estimateSizeFactor by yeast spike-in size-factors. For Supplementary Fig. 5b BigWig files were 1074 converted to bedGraph, forward and reverse strand were then merged into one bedGraph file. 1075 In R (v4.3.2), with a custom script, coverage values were extracted from each feature with 1076 regards to strand and normalised to region size. Genomic spike-in normalised track screenshots 1077 were exported from IGB as vector images. 1078 1079 ELCAP-seq analysis 1080 Raw fastq files were trimmed with cutadapt (v4.5) within trim_galore (v0.6.10) and minimum 1081 read length filtering > 19nt and maximum read length < 47 nt. For purposes of visualization, 1082 biological replicates adapter trimmed and size filtered reads fastq files were merged and 1083 processed further similarly to individual replicates. Resulting reads were mapped with STAR 1084 with ( –-alignEndsType Extend5pOfRead1) against the human genome assembly (GRCh38, 1085 Ensembl release 108). Genome alignment BAM files were filtered for multimapped reads and 1086 separated by strand with samtools view function (-q 10, -F 16 or -f 16). Single nucleotide strand 1087 specific and RPKM (read per million kilobase) normalised BigWig files were generated using 1088 Deeptools’ bamCoverage function for each stranded BAM files in single nucleotide resolution 1089 mode with the following options “ -bs 1 --Offset -1 --normalizeUsing RPKM”. ‘Pause sites’ 1090 were obtained by calculating the highest peak position within pausing region ( -20 nt , TSS, 1091 +120 nt) in the untreated sample with CAGEfightR75 . 1092 ELCAP-seq centred metagene plots (profiles or heatmaps) were created using deepTools 1093 computeMatrix in reference -point mode ( centred on TSS or the ‘pause sites’) over the same 1094 non-overlapping protein coding gene set as for TT chem-seq plots. ELCAP-seq signal relative to 1095 untreated control ratio BigWig files were created using the deepTools’ bigwigCompare 1096 function in single nucleotide mode ( -bs 1). Heatmaps were sorted by gene length (decreasing 1097 or increasing). For Supplementary Fig. 5c ratio between TT chem-seq and ELCAP -seq BigWig 1098 files for each time points were created using deepTools bigwigCompare function and scaled to 1099 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 6, 2025. ; https://doi.org/10.1101/2025.11.06.686939doi: bioRxiv preprint 44 a normali sing ratio calculated for library size dispersion between the two datasets. For 1100 Supplematery Fig. 5d ELCAP-seq BigWig files were converted to bedGraph, forward and 1101 reverse strand were then merged into one bedGraph file. In R, with a custom script, the pausing 1102 index (PI) was calculated as ratio of coverage values from TSS (TSS + 200 bp) over coverage 1103 for gene body region (TSS + 200 bp to TES – 200 bp) relative to size. The recurrence of pausing 1104 index was then plotted as distribution. 1105 1106 Mass spectrometry analysis 1107 Raw mass spectrometry data were analysed with MaxQuant (v1.6.15.0). Peak lists were 1108 searched against the human Uniprot FASTA database combined with 262 common 1109 contaminants by the integrated Andromeda search engine. False discovery rate was set to 1 % 1110 for both peptides (minimum length of 7 amino acids) and proteins. “Match between runs” 1111 (MBR) was enabled with a Match time window of 0.7, and a Match ion mobility window of 1112 0.05 min. Relative protein amounts were determined by the MaxLFQ algorithm with a 1113 minimum ratio count of two. Statistical analysis of LFQ derived protein expression data was 1114 performed using the automated analysis pipeline of the Clinical Knowledge Graph. 93 Protein 1115 entries referring to potential contaminants, proteins identified by matches to the decoy reverse 1116 database, and proteins identified only by modified sites, were removed. LFQ intensity values 1117 were normalised by log2 transformation and proteins with less than 70 % of valid values in at 1118 least one group were filtered out. The remaining missing values were imputed using the 1119 MinProb approach (random draws from a Gaussian distribution; width = 0.2 and downshift = 1120 1.8) or with Mixed imputation using KNN. 1121 1122 ChIP-seq analysis 1123 NELFE, total RPB1 and input raw fastq files were trimmed with with trim_galore (paired-end 1124 mode) with minimum read length filtering > 19 nt. Filtered and trimmed paired end reads were 1125 mapped with STAR against human assembly GRCh38 (Ensembl release 108). Genome 1126 alignment BAM files were filtered for multi-mapped reads with samtools view function (-q 10). 1127 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 6, 2025. ; https://doi.org/10.1101/2025.11.06.686939doi: bioRxiv preprint 45 Bam files for NELFE or total RPB1 samples were normalised to input with MACS76 (v2.2.9.1). 1128 Resulting bedGraph files were used to compute heatmaps. In Fig. 4e input normalised NELFE 1129 ChIP-seq signal levels were used to sort (in decreasing order) both bottom panels of ELCAP -1130 seq signals. In Fig. 4f ChIP-seq ratio (sample relative to untreated control) BigWig files were 1131 created using the deepTools bigwigCompare function, the genome was partitioned to 50 bp 1132 bins. NELFE ChIP-seq ratio BigWig files were then normalised to total RPB1 ChiP-seq levels 1133 with bigwigCompare function and plotted relative to NELFE ChIP-seq decreasing signal. 1134 1135 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 6, 2025. ; https://doi.org/10.1101/2025.11.06.686939doi: bioRxiv preprint 46 Supplemental item titles 1136 1137 Table S1: Mass spectrometry of RNAPII immunoprecipitation experiments 1138 Table S2: List of oligonucleotides 1139 1140 1141 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 6, 2025. ; https://doi.org/10.1101/2025.11.06.686939doi: bioRxiv preprint 47

1142 1. Tubbs, A., and Nussenzweig, A. (2017). Endogenous DNA Damage as a Source of Genomic 1143 Instability in Cancer. Cell 168, 644-656. 10.1016/j.cell.2017.01.002. 1144 2. Crow, J.F. (2000). The origins, patterns and implications of human spontaneous mutation. Nat 1145 Rev Genet 1, 40-47. 10.1038/35049558. 1146 3. Lodato, M.A., Rodin, R.E., Bohrson, C.L., Coulter, M.E., Barton, A.R., Kwon, M., Sherman, 1147 M.A., Vitzthum, C.M., Luquette, L.J., Yandava, C.N., et al. (2018). Aging and 1148 neurodegeneration are associated with increased mutations in single human neurons. Science 1149 359, 555-559. 10.1126/science.aao4426. 1150 4. Caron, P., van der Linden, J., and van Attikum, H. (2019). Bon voyage: A transcriptional 1151 journey around DNA breaks. DNA Repair (Amst) 82, 102686. 10.1016/j.dnarep.2019.102686. 1152 5. Lans, H., Hoeijmakers, J.H.J., Vermeulen, W., and Marteijn, J.A. (2019). The DNA damage 1153 response to transcription stress. Nat Rev Mol Cell Biol 20, 766-784. 10.1038/s41580-019-1154 0169-4. 1155 6. Wang, J., Muste Sadurni, M., and Saponaro, M. (2023). RNAPII response to transcription-1156 blocking DNA lesions in mammalian cells. FEBS J 290, 4382-4394. 10.1111/febs.16561. 1157 7. Gregersen, L.H., and Svejstrup, J.Q. (2018). The Cellular Response to Transcription-Blocking 1158 DNA Damage. Trends Biochem Sci 43, 327-341. 10.1016/j.tibs.2018.02.010. 1159 8. Heine, G.F., Horwitz, A.A., and Parvin, J.D. (2008). Multiple mechanisms contribute to 1160 inhibit transcription in response to DNA damage. J Biol Chem 283, 9555-9561. 1161 10.1074/jbc.M707700200. 1162 9. van der Meer, P.J., and Luijsterburg, M.S. (2025). The molecular basis of human 1163 transcription-coupled DNA repair. Nat Cell Biol 27, 1230-1239. 10.1038/s41556-025-01715-1164 9. 1165 10. Tufegdzic Vidakovic, A., Mitter, R., Kelly, G.P., Neumann, M., Harreman, M., Rodriguez-1166 Martinez, M., Herlihy, A., Weems, J.C., Boeing, S., Encheva, V., et al. (2020). Regulation of 1167 the RNAPII Pool Is Integral to the DNA Damage Response. Cell 180, 1245-1261 e1221. 1168 10.1016/j.cell.2020.02.009. 1169 11. Nakazawa, Y., Hara, Y., Oka, Y., Komine, O., van den Heuvel, D., Guo, C., Daigaku, Y., 1170 Isono, M., He, Y., Shimada, M., et al. (2020). Ubiquitination of DNA Damage-Stalled 1171 RNAPII Promotes Transcription-Coupled Repair. Cell 180, 1228-1244 e1224. 1172 10.1016/j.cell.2020.02.010. 1173 12. Mayne, L.V., and Lehmann, A.R. (1982). Failure of RNA synthesis to recover after UV 1174 irradiation: an early defect in cells from individuals with Cockayne's syndrome and xeroderma 1175 pigmentosum. Cancer Res 42, 1473-1478. 1176 13. Nieto Moreno, N., Olthof, A.M., and Svejstrup, J.Q. (2023). Transcription-Coupled 1177 Nucleotide Excision Repair and the Transcriptional Response to UV-Induced DNA Damage. 1178 Annu Rev Biochem 92, 81-113. 10.1146/annurev-biochem-052621-091205. 1179 14. Fu, H., Liu, R., Jia, Z., Li, R., Zhu, F., Zhu, W., Shao, Y., Jin, Y., Xue, Y., Huang, J., et al. 1180 (2022). Poly(ADP-ribosylation) of P-TEFb by PARP1 disrupts phase separation to inhibit 1181 global transcription after DNA damage. Nat Cell Biol 24, 513-525. 10.1038/s41556-022-1182 00872-5. 1183 15. Adamowicz, M., Hailstone, R., Demin, A.A., Komulainen, E., Hanzlikova, H., Brazina, J., 1184 Gautam, A., Wells, S.E., and Caldecott, K.W. (2021). XRCC1 protects transcription from 1185 toxic PARP1 activity during DNA base excision repair. Nat Cell Biol 23, 1287-1298. 1186 10.1038/s41556-021-00792-w. 1187 16. Lindahl, T. (1993). Instability and decay of the primary structure of DNA. Nature 362, 709-1188 715. 10.1038/362709a0. 1189 17. Charlet-Berguerand, N., Feuerhahn, S., Kong, S.E., Ziserman, H., Conaway, J.W., Conaway, 1190 R., and Egly, J.M. (2006). RNA polymerase II bypass of oxidative DNA damage is regulated 1191 by transcription elongation factors. EMBO J 25, 5481-5491. 10.1038/sj.emboj.7601403. 1192 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 6, 2025. ; https://doi.org/10.1101/2025.11.06.686939doi: bioRxiv preprint 48 18. Damsma, G.E., and Cramer, P. (2009). Molecular basis of transcriptional mutagenesis at 8-1193 oxoguanine. J Biol Chem 284, 31658-31663. 10.1074/jbc.M109.022764. 1194 19. Tornaletti, S., Maeda, L.S., Kolodner, R.D., and Hanawalt, P.C. (2004). Effect of 8-1195 oxoguanine on transcription elongation by T7 RNA polymerase and mammalian RNA 1196 polymerase II. DNA Repair (Amst) 3, 483-494. 10.1016/j.dnarep.2004.01.003. 1197 20. Oh, J., Fleming, A.M., Xu, J., Chong, J., Burrows, C.J., and Wang, D. (2020). RNA 1198 polymerase II stalls on oxidative DNA damage via a torsion-latch mechanism involving lone 1199 pair-pi and CH-pi interactions. Proc Natl Acad Sci U S A 117, 9338-9348. 1200 10.1073/pnas.1919904117. 1201 21. Kathe, S.D., Shen, G.P., and Wallace, S.S. (2004). Single-stranded breaks in DNA but not 1202 oxidative DNA base damages block transcriptional elongation by RNA polymerase II in HeLa 1203 cell nuclear extracts. J Biol Chem 279, 18511-18520. 10.1074/jbc.M313598200. 1204 22. Jackson, S.P., and Bartek, J. (2009). The DNA-damage response in human biology and 1205 disease. Nature 461, 1071-1078. 10.1038/nature08467. 1206 23. Ray Chaudhuri, A., and Nussenzweig, A. (2017). The multifaceted roles of PARP1 in DNA 1207 repair and chromatin remodelling. Nat Rev Mol Cell Biol 18, 610-621. 10.1038/nrm.2017.53. 1208 24. Shanbhag, N.M., Rafalska-Metcalf, I.U., Balane-Bolivar, C., Janicki, S.M., and Greenberg, 1209 R.A. (2010). ATM-dependent chromatin changes silence transcription in cis to DNA double-1210 strand breaks. Cell 141, 970-981. 10.1016/j.cell.2010.04.038. 1211 25. Kakarougkas, A., Ismail, A., Chambers, A.L., Riballo, E., Herbert, A.D., Kunzel, J., Lobrich, 1212 M., Jeggo, P.A., and Downs, J.A. (2014). Requirement for PBAF in transcriptional repression 1213 and repair at DNA breaks in actively transcribed regions of chromatin. Mol Cell 55, 723-732. 1214 10.1016/j.molcel.2014.06.028. 1215 26. Dong, C., West, K.L., Tan, X.Y., Li, J., Ishibashi, T., Yu, C.H., Sy, S.M.H., Leung, J.W.C., 1216 and Huen, M.S.Y. (2020). Screen identifies DYRK1B network as mediator of transcription 1217 repression on damaged chromatin. Proc Natl Acad Sci U S A 117, 17019-17030. 1218 10.1073/pnas.2002193117. 1219 27. Chou, D.M., Adamson, B., Dephoure, N.E., Tan, X., Nottke, A.C., Hurov, K.E., Gygi, S.P., 1220 Colaiacovo, M.P., and Elledge, S.J. (2010). A chromatin localization screen reveals poly 1221 (ADP ribose)-regulated recruitment of the repressive polycomb and NuRD complexes to sites 1222 of DNA damage. Proc Natl Acad Sci U S A 107, 18475-18480. 10.1073/pnas.1012946107. 1223 28. Awwad, S.W., Abu-Zhayia, E.R., Guttmann-Raviv, N., and Ayoub, N. (2017). NELF-E is 1224 recruited to DNA double-strand break sites to promote transcriptional repression and repair. 1225 EMBO Rep 18, 745-764. 10.15252/embr.201643191. 1226 29. El-Khamisy, S.F., Masutani, M., Suzuki, H., and Caldecott, K.W. (2003). A requirement for 1227 PARP-1 for the assembly or stability of XRCC1 nuclear foci at sites of oxidative DNA 1228 damage. Nucleic Acids Res 31, 5526-5533. 10.1093/nar/gkg761. 1229 30. Hanzlikova, H., Gittens, W., Krejcikova, K., Zeng, Z., and Caldecott, K.W. (2017). 1230 Overlapping roles for PARP1 and PARP2 in the recruitment of endogenous XRCC1 and 1231 PNKP into oxidized chromatin. Nucleic Acids Res 45, 2546-2557. 10.1093/nar/gkw1246. 1232 31. Aprile-Garcia, F., Tomar, P., Hummel, B., Khavaran, A., and Sawarkar, R. (2019). Nascent-1233 protein ubiquitination is required for heat shock-induced gene downregulation in human cells. 1234 Nat Struct Mol Biol 26, 137-146. 10.1038/s41594-018-0182-x. 1235 32. Cavadas, M.A.S., Cheong, A., and Taylor, C.T. (2017). The regulation of transcriptional 1236 repression in hypoxia. Exp Cell Res 356, 173-181. 10.1016/j.yexcr.2017.02.024. 1237 33. Mahat, D.B., Salamanca, H.H., Duarte, F.M., Danko, C.G., and Lis, J.T. (2016). Mammalian 1238 Heat Shock Response and Mechanisms Underlying Its Genome-wide Transcriptional 1239 Regulation. Mol Cell 62, 63-78. 10.1016/j.molcel.2016.02.025. 1240 34. Vihervaara, A., Duarte, F.M., and Lis, J.T. (2018). Molecular mechanisms driving 1241 transcriptional stress responses. Nat Rev Genet 19, 385-397. 10.1038/s41576-018-0001-6. 1242 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 6, 2025. ; https://doi.org/10.1101/2025.11.06.686939doi: bioRxiv preprint 49 35. Gressel, S., Schwalb, B., and Cramer, P. (2019). The pause-initiation limit restricts 1243 transcription activation in human cells. Nat Commun 10, 3603. 10.1038/s41467-019-11536-8. 1244 36. Rawat, P., Boehning, M., Hummel, B., Aprile-Garcia, F., Pandit, A.S., Eisenhardt, N., 1245 Khavaran, A., Niskanen, E., Vos, S.M., Palvimo, J.J., et al. (2021). Stress-induced nuclear 1246 condensation of NELF drives transcriptional downregulation. Mol Cell. 1247 10.1016/j.molcel.2021.01.016. 1248 37. Cugusi, S., Mitter, R., Kelly, G.P., Walker, J., Han, Z., Pisano, P., Wierer, M., Stewart, A., 1249 and Svejstrup, J.Q. (2022). Heat shock induces premature transcript termination and 1250 reconfigured the human transcriptome. Mol Cell 82, 1573-1588 e1510. 1251 10.1016/j.molcel.2022.01.007. 1252 38. Gregersen, L.H., Mitter, R., and Svejstrup, J.Q. (2020). Using TTchem-seq for profiling 1253 nascent transcription and measuring transcript elongation. Nat Protoc 15, 604-627. 1254 10.1038/s41596-019-0262-3. 1255 39. Gregersen, L.H., Mitter, R., and Svejstrup, J.Q. (2022). Elongation factor-specific capture of 1256 RNA polymerase II complexes. Cell Rep Methods 2, 100368. 10.1016/j.crmeth.2022.100368. 1257 40. Tufegdzic Vidakovic, A., Harreman, M., Dirac-Svejstrup, A.B., Boeing, S., Roy, A., Encheva, 1258 V., Neumann, M., Wilson, M., Snijders, A.P., and Svejstrup, J.Q. (2019). Analysis of RNA 1259 polymerase II ubiquitylation and proteasomal degradation. Methods 159-160, 146-156. 1260 10.1016/j.ymeth.2019.02.005. 1261 41. Gonzalo-Hansen, C., Steurer, B., Janssens, R.C., Zhou, D., van Sluis, M., Lans, H., and 1262 Marteijn, J.A. (2024). Differential processing of RNA polymerase II at DNA damage 1263 correlates with transcription-coupled repair syndrome severity. Nucleic Acids Res 52, 9596-1264 9612. 10.1093/nar/gkae618. 1265 42. Guo, J., Hanawalt, P.C., and Spivak, G. (2013). Comet-FISH with strand-specific probes 1266 reveals transcription-coupled repair of 8-oxoGuanine in human cells. Nucleic Acids Res 41, 1267 7700-7712. 10.1093/nar/gkt524. 1268 43. Kyng, K.J., May, A., Brosh, R.M., Jr., Cheng, W.H., Chen, C., Becker, K.G., and Bohr, V.A. 1269 (2003). The transcriptional response after oxidative stress is defective in Cockayne syndrome 1270 group B cells. Oncogene 22, 1135-1149. 10.1038/sj.onc.1206187. 1271 44. Ranes, M., Boeing, S., Wang, Y., Wienholz, F., Menoni, H., Walker, J., Encheva, V., 1272 Chakravarty, P., Mari, P.O., Stewart, A., et al. (2016). A ubiquitylation site in Cockayne 1273 syndrome B required for repair of oxidative DNA damage, but not for transcription-coupled 1274 nucleotide excision repair. Nucleic Acids Res 44, 5246-5255. 10.1093/nar/gkw216. 1275 45. Tuo, J., Jaruga, P., Rodriguez, H., Dizdaroglu, M., and Bohr, V.A. (2002). The cockayne 1276 syndrome group B gene product is involved in cellular repair of 8-hydroxyadenine in DNA. J 1277 Biol Chem 277, 30832-30837. 10.1074/jbc.M204814200. 1278 46. Andrade-Lima, L.C., Veloso, A., Paulsen, M.T., Menck, C.F., and Ljungman, M. (2015). 1279 DNA repair and recovery of RNA synthesis following exposure to ultraviolet light are delayed 1280 in long genes. Nucleic Acids Res 43, 2744-2756. 10.1093/nar/gkv148. 1281 47. Williamson, L., Saponaro, M., Boeing, S., East, P., Mitter, R., Kantidakis, T., Kelly, G.P., 1282 Lobley, A., Walker, J., Spencer-Dene, B., et al. (2017). UV Irradiation Induces a Non-coding 1283 RNA that Functionally Opposes the Protein Encoded by the Same Gene. Cell 168, 843-855 1284 e813. 10.1016/j.cell.2017.01.019. 1285 48. Lavigne, M.D., Konstantopoulos, D., Ntakou-Zamplara, K.Z., Liakos, A., and Fousteri, M. 1286 (2017). Global unleashing of transcription elongation waves in response to genotoxic stress 1287 restricts somatic mutation rate. Nat Commun 8, 2076. 10.1038/s41467-017-02145-4. 1288 49. Zumer, K., Maier, K.C., Farnung, L., Jaeger, M.G., Rus, P., Winter, G., and Cramer, P. 1289 (2021). Two distinct mechanisms of RNA polymerase II elongation stimulation in vivo. Mol 1290 Cell 81, 3096-3109 e3098. 10.1016/j.molcel.2021.05.028. 1291 50. Stein, C.B., Field, A.R., Mimoso, C.A., Zhao, C., Huang, K.L., Wagner, E.J., and Adelman, 1292 K. (2022). Integrator endonuclease drives promoter-proximal termination at all RNA 1293 polymerase II-transcribed loci. Mol Cell 82, 4232-4245 e4211. 10.1016/j.molcel.2022.10.004. 1294 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 6, 2025. ; https://doi.org/10.1101/2025.11.06.686939doi: bioRxiv preprint 50 51. Aoi, Y., Smith, E.R., Shah, A.P., Rendleman, E.J., Marshall, S.A., Woodfin, A.R., Chen, F.X., 1295 Shiekhattar, R., and Shilatifard, A. (2020). NELF Regulates a Promoter-Proximal Step 1296 Distinct from RNA Pol II Pause-Release. Mol Cell 78, 261-274 e265. 1297 10.1016/j.molcel.2020.02.014. 1298 52. DeBerardine, M., Booth, G.T., Versluis, P.P., and Lis, J.T. (2023). The NELF pausing 1299 checkpoint mediates the functional divergence of Cdk9. Nat Commun 14, 2762. 1300 10.1038/s41467-023-38359-y. 1301 53. Bai, P. (2015). Biology of Poly(ADP-Ribose) Polymerases: The Factotums of Cell 1302 Maintenance. Mol Cell 58, 947-958. 10.1016/j.molcel.2015.01.034. 1303 54. Slade, D., Dunstan, M.S., Barkauskaite, E., Weston, R., Lafite, P., Dixon, N., Ahel, M., Leys, 1304 D., and Ahel, I. (2011). The structure and catalytic mechanism of a poly(ADP-ribose) 1305 glycohydrolase. Nature 477, 616-620. 10.1038/nature10404. 1306 55. Ray, S., Abugable, A.A., Parker, J., Liversidge, K., Palminha, N.M., Liao, C., Acosta-Martin, 1307 A.E., Souza, C.D.S., Jurga, M., Sudbery, I., and El-Khamisy, S.F. (2022). A mechanism for 1308 oxidative damage repair at gene regulatory elements. Nature 609, 1038-1047. 1309 10.1038/s41586-022-05217-8. 1310 56. Demin, A.A., Hirota, K., Tsuda, M., Adamowicz, M., Hailstone, R., Brazina, J., Gittens, W., 1311 Kalasova, I., Shao, Z., Zha, S., et al. (2021). XRCC1 prevents toxic PARP1 trapping during 1312 DNA base excision repair. Mol Cell 81, 3018-3030 e3015. 10.1016/j.molcel.2021.05.009. 1313 57. Inukai, N., Yamaguchi, Y., Kuraoka, I., Yamada, T., Kamijo, S., Kato, J., Tanaka, K., and 1314 Handa, H. (2004). A novel hydrogen peroxide-induced phosphorylation and ubiquitination 1315 pathway leading to RNA polymerase II proteolysis. J Biol Chem 279, 8190-8195. 1316 10.1074/jbc.M311412200. 1317 58. Neil, A.J., Belotserkovskii, B.P., and Hanawalt, P.C. (2012). Transcription blockage by bulky 1318 end termini at single-strand breaks in the DNA template: differential effects of 5' and 3' 1319 adducts. Biochemistry 51, 8964-8970. 10.1021/bi301240y. 1320 59. Caldecott, K.W. (2024). Causes and consequences of DNA single-strand breaks. Trends 1321 Biochem Sci 49, 68-78. 10.1016/j.tibs.2023.11.001. 1322 60. Hoch, N.C., Hanzlikova, H., Rulten, S.L., Tetreault, M., Komulainen, E., Ju, L., Hornyak, P., 1323 Zeng, Z., Gittens, W., Rey, S.A., et al. (2017). XRCC1 mutation is associated with PARP1 1324 hyperactivation and cerebellar ataxia. Nature 541, 87-91. 10.1038/nature20790. 1325 61. Hanna, B.M.F., Michel, M., Helleday, T., and Mortusewicz, O. (2021). NEIL1 and NEIL2 1326 Are Recruited as Potential Backup for OGG1 upon OGG1 Depletion or Inhibition by TH5487. 1327 Int J Mol Sci 22. 10.3390/ijms22094542. 1328 62. David, S.S., O'Shea, V.L., and Kundu, S. (2007). Base-excision repair of oxidative DNA 1329 damage. Nature 447, 941-950. 10.1038/nature05978. 1330 63. Xie, Y., Yang, H., Cunanan, C., Okamoto, K., Shibata, D., Pan, J., Barnes, D.E., Lindahl, T., 1331 McIlhatton, M., Fishel, R., and Miller, J.H. (2004). Deficiencies in mouse Myh and Ogg1 1332

in tumor predisposition and G to T mutations in codon 12 of the K-ras oncogene in lung 1333 tumors. Cancer Res 64, 3096-3102. 10.1158/0008-5472.can-03-3834. 1334 64. Gibson, B.A., Zhang, Y., Jiang, H., Hussey, K.M., Shrimp, J.H., Lin, H., Schwede, F., Yu, Y., 1335 and Kraus, W.L. (2016). Chemical genetic discovery of PARP targets reveals a role for 1336 PARP-1 in transcription elongation. Science 353, 45-50. 10.1126/science.aaf7865. 1337 65. Banerjee, D., Mandal, S.M., Das, A., Hegde, M.L., Das, S., Bhakat, K.K., Boldogh, I., Sarkar, 1338 P.S., Mitra, S., and Hazra, T.K. (2011). Preferential repair of oxidized base damage in the 1339 transcribed genes of mammalian cells. J Biol Chem 286, 6006-6016. 1340 10.1074/jbc.M110.198796. 1341 66. Thorslund, T., Sunesen, M., Bohr, V.A., and Stevnsner, T. (2002). Repair of 8-oxoG is slower 1342 in endogenous nuclear genes than in mitochondrial DNA and is without strand bias. DNA 1343 Repair (Amst) 1, 261-273. 10.1016/s1568-7864(02)00003-4. 1344 67. Mulderrig, L., Garaycoechea, J.I., Tuong, Z.K., Millington, C.L., Dingler, F.A., Ferdinand, 1345 J.R., Gaul, L., Tadross, J.A., Arends, M.J., O'Rahilly, S., et al. (2021). Aldehyde-driven 1346 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 6, 2025. ; https://doi.org/10.1101/2025.11.06.686939doi: bioRxiv preprint 51 transcriptional stress triggers an anorexic DNA damage response. Nature 600, 158-163. 1347 10.1038/s41586-021-04133-7. 1348 68. Hu, X., Malik, S., Negroiu, C.C., Hubbard, K., Velalar, C.N., Hampton, B., Grosu, D., 1349 Catalano, J., Roeder, R.G., and Gnatt, A. (2006). A Mediator-responsive form of metazoan 1350 RNA polymerase II. Proc Natl Acad Sci U S A 103, 9506-9511. 10.1073/pnas.0603702103. 1351 69. Han, Z., Moore, G.A., Mitter, R., Lopez Martinez, D., Wan, L., Dirac Svejstrup, A.B., Rueda, 1352 D.S., and Svejstrup, J.Q. (2023). DNA-directed termination of RNA polymerase II 1353 transcription. Molecular Cell. 10.1016/j.molcel.2023.08.007. 1354 70. Smith, T., Heger, A., and Sudbery, I. (2017). UMI-tools: modeling sequencing errors in 1355 Unique Molecular Identifiers to improve quantification accuracy. Genome Res 27, 491-499. 1356 10.1101/gr.209601.116. 1357 71. Dobin, A., Davis, C.A., Schlesinger, F., Drenkow, J., Zaleski, C., Jha, S., Batut, P., Chaisson, 1358 M., and Gingeras, T.R. (2013). STAR: ultrafast universal RNA-seq aligner. Bioinformatics 1359 29, 15-21. 10.1093/bioinformatics/bts635. 1360 72. Liao, Y., Smyth, G.K., and Shi, W. (2014). featureCounts: an efficient general purpose 1361 program for assigning sequence reads to genomic features. Bioinformatics 30, 923-930. 1362 10.1093/bioinformatics/btt656. 1363 73. Love, M.I., Huber, W., and Anders, S. (2014). Moderated estimation of fold change and 1364 dispersion for RNA-seq data with DESeq2. Genome Biol 15, 550. 10.1186/s13059-014-0550-1365 8. 1366 74. Ramirez, F., Dundar, F., Diehl, S., Gruning, B.A., and Manke, T. (2014). deepTools: a 1367 flexible platform for exploring deep-sequencing data. Nucleic Acids Res 42, W187-191. 1368 10.1093/nar/gku365. 1369 75. Thodberg, M., Thieffry, A., Vitting-Seerup, K., Andersson, R., and Sandelin, A. (2019). 1370 CAGEfightR: analysis of 5'-end data using R/Bioconductor. BMC Bioinformatics 20, 487. 1371 10.1186/s12859-019-3029-5. 1372 76. Zhang, Y., Liu, T., Meyer, C.A., Eeckhoute, J., Johnson, D.S., Bernstein, B.E., Nusbaum, C., 1373 Myers, R.M., Brown, M., Li, W., and Liu, X.S. (2008). Model-based analysis of ChIP-Seq 1374 (MACS). Genome Biol 9, R137. 10.1186/gb-2008-9-9-r137. 1375 1376 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 6, 2025. ; https://doi.org/10.1101/2025.11.06.686939doi: bioRxiv preprint .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 6, 2025. ; https://doi.org/10.1101/2025.11.06.686939doi: bioRxiv preprint Fig. 1. Oxidative stress represses global nascent transcription of RNAPII without release from chromatin. a Schematic of oxidative stress inductions using either H2O2 or Menadione (MND). b Nascent transcription levels measured by nascent RNA metabolic pulse labelling (15 min) with 4 - thiouridine (4sU). Oxidative stress was induced for 15 min in HEK293 cells with 1 mM H 2O2 or 0.25 mM MND, and cells were allowed to recovery for indicated times. Staining with methylene blue as a loading control. c-d Nascent transcription levels quantified by 5 -Ethynyluridine (EU) labelling and single cell quantification by immunofluorescence microscopy (EU-assay). Per nuclei EU signal is normalised to DAPI and log10 scaled. Cells treated with H2O2 (c) or MND (d) and EU-labelled for 30 min. e Schematic of TTchem-seq profiling nascent transcriptome by 4sU incorporation into nascent RNA, fragmentation and sequencing of purified labelled RNAs. f Single-gene view of TT chem-seq data between untreated controls and 15 min H2O2 treated cells and MND treated cells. g-h Metagene analysis of TTchem-seq data between untreated controls and 15 min H2O2 treated cells (g) and MND treated cell (h). Lines represent the mean of spike-in normalised read counts between two biological replicates for a set of protein-encoding genes (non-overlapping ± 2 kb, n = 11,593). i Western blot of total RNAPII levels from chromatin-enriched cellular fractions. Total RNAPII was detected with an antibody raised against the N-terminal part of RPB1 (D8L4Y). Histone H3 serves as loading control. IIo and IIa denotes hyperphosphorylated and unphosphorylated RPB1 CTD respectively. j Schematic of ELCAP -seq experimental design . RNAPII is captured by immunoprecipitation of RPB3 -FLAG, short RNA fragments protected by RNAPII are extracted and sequencing to obtain nucleotide resolution maps of RNAPII positions (3’end corresponding to last incorporated nucleotide) . k Heatmaps of ELCAP-seq signal (left: RPKM) alongside TTchem-seq signal (right; read counts normalised to spike-in). The signal is shown as the log2 fold -change between H2O2 -treated HEK293 cells and untreated cells. Heatmaps centred on transcription start site and sorted by increasing length of gene regions (−5 kb upstream of TSS to 50 kb into the gene body) with dashed lines indicating transcript end sites (TES). .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 6, 2025. ; https://doi.org/10.1101/2025.11.06.686939doi: bioRxiv preprint .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 6, 2025. ; https://doi.org/10.1101/2025.11.06.686939doi: bioRxiv preprint Fig. 2. Oxidative stress arrest RNAPII within the gene body . a Schematic for RNAPII promoter - proximal release time points with DRB inhibition and release with and without H 2O2 treatment in HEK293 cells. RNAPII complexes are synchronised at promoter-proximal pause site by DRB treatment for 3.5 hours, then released by media washout following either re-inhibition or not and either treated or not with H2O2. b Metagene analysis of TTchem-seq data following 3.5 hours of DRB inhibition between no DRB release, 30 mi n released control and 30 min release with 15 min H2O2 treatment of HEK293 cells. Data is represented as the average spike-in normalised signal of two biological replicates. The arrow represents the change upon H2O2 treatment. c Metagene analysis of TTchem-seq data following 3.5 hours of DRB inhibition with DRB release for 15 min and re-inhibition for 15 min with and without 15min H2O2 treatment. The arrow represents the change upon H2O2 treatment. Data is represented as the mean of spike-in normalised TTchem-seq read counts over each segmented bins for the selected genes set of protein coding genes and average of two biological replicates. d-e Single-gene view of TTchem- seq data for the PPP2R2A gene for experimental conditions of (b-c) respectively. .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 6, 2025. ; https://doi.org/10.1101/2025.11.06.686939doi: bioRxiv preprint .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 6, 2025. ; https://doi.org/10.1101/2025.11.06.686939doi: bioRxiv preprint Fig. 3. Arrested RNAPII elongation complexes resume transcription from within the gene body during recovery from oxidative stress . a Metagene analysis of TTchem-seq data for long genes (>90 kb, n=2217) following H 2O2 treatment and indicated recovery time points in HEK293 cells. The line represents the mean of spike-in normalised read counts from three biological replicates. b Single-gene view of TTchem-seq data for FARS2 gene (length = 521 kb) for experimental conditions of ( a). Arrow represent distance RNAPII would travel from the TSS with an elongation speed of 2 kb/min or 4 kb/min, within the 30 min or 1 hour recovery period, respectively. c Schematic of the experimental setup used to study RNAPII restart following oxidative stress with or without DRB inhibition. d Nascent transcription levels (4sU dot blot) after 15 min treatment of HEK293 cells with 1 mM H2O2 with and without DRB. Staining with methylene blue serves as a loading control. e Metagene analysis of TTchem- seq data for long genes (>90 kb, n=2217) of untreated and 1 hour recovery timepoint following H2O2 treatment of HEK293 cells, with and without DRB inhibition. Lines is the mean of spike-in normalised read counts between two biological replicates. f Single-gene view of TTchem-seq data for FOXP4 gene for experimental conditions of (c). .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 6, 2025. ; https://doi.org/10.1101/2025.11.06.686939doi: bioRxiv preprint .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 6, 2025. ; https://doi.org/10.1101/2025.11.06.686939doi: bioRxiv preprint Fig. 4. NELF mediated RNAPII pausing is altered during the oxidative stress response . a Total RNAPII (FLAG IP in HEK293 RPB3-FLAG cells) interactome after 15 min of 1 mM H2O2 treatment. A log2 fold-change comparison of H 2O2-treated with untreated control cells is shown relative to significance as -log2 of p-value (triplicate injections). Are highlighted: RNAPII, NELF and the capping complex subunits and INTS4 (IP-WB loading control in (c)). b Phosphorylated RPB1 (4H8 IP, HEK293 cells) interactome after 15 min of 1 mM H2O2 treatment. A log2 fold-change comparison between H2O2 treated to untreated control cells is shown relative to significance as -log2 of p-value. Are highlighted: RNAPII subunits, NELF subunits, members of the capping complex and CPSF2 (IP-WB loading control) in (c) are coloured and labelled. c Western blot analysis of FLAG IP (left) and 4H8 IP (right) before and after H2O2 treatment. Total RNAPII probed with D8L4Y antibody. Phosphorylated RNAPII CTD with 4H8. Serine 5 phosphorylated RNAPII CTD with 3E8. d Metagene analysis of ELCAP-seq, RPKM signal centred on the TSS for non -overlapping coding genes (n=11 ,593) in untreated control, H2O2 treated, and recovery timepoints (30 min, 1 hour and 2 hours). The lines are the average signal of two biological replicates. Doted vertical lines are the highest RNAPII occupancy distance from TSS upon 30 min recovery and 2 hours recovery . e Heatmap analysis of RPKM ELCAP-seq in RPB3- 3xFLAG cells treated with H2O2 as in ( d) and NELFE untreated ChIP-seq signal in HEK293. The ELCAP-seq signal is shown as a log2 fold -change relative to untreated. NELFE ChIP-seq signal is normalised to IgG input . Top panel shows gene regions (pause −5 kb to pause +50 kb) sorted by increasing gene length (pause site genes n= 7 ,595), with dashed lines indicating TES. Middle panel shows promoter regions (pause site ± 5 kb) sorted by decreasing NELFE ChIP-seq signal. Bottom panel shows promoter regions (pause site ± 0.5 kb) sorted by decreasing NELFE ChIP-seq signal. f Heatmap of NELFE ChIP-seq log2 fold-change between H2O2 treated and untreated control and between 1 hour recovery and untreated control, both normalised by ratio to the log2 fold-change of respective treatment changes in total RPB1 (D8L4Y) ChIP-seq in HEK293 cells. Gene regions (± 5 kb pause site genes n = 7,595) are sorted by decreasing NELFE ChIP-seq signal in untreated condition. .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 6, 2025. ; https://doi.org/10.1101/2025.11.06.686939doi: bioRxiv preprint .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 6, 2025. ; https://doi.org/10.1101/2025.11.06.686939doi: bioRxiv preprint Fig. 5. PARylation restricts release of early RNAPII elongation complexes during transcription recovery. a-b Western blot analysis of poly -ADP-ribose levels in HEK293 cells with and without 1 hour pre-treatment with PARP inhibitor ( Olaparib) (a) or cells pre -treated for 1 hour with PARPG inhibitor (PDD 00017273) (b) and then subsequently treated with 1 mM H2O2 for 15 min and indicated times following media replenishment. Ponceau S staining serves as total protein loading contro l. c Nascent transcription levels (4sU dot blot) treated as in ( a) with and without PARPi pre-treatment or PARGi. Staining with methylene blue serves as a loading control. d 4sU labelled RNA yields measured after streptavidin pull-down preformed in duplicates of HEK293 cells treated as in (a), with or without 1 hour PARPi pre-treatment. e-f Metagene analysis of TTchem-seq for long genes (non-overlapping >90 kb, n = 2,217) of DMSO treated HEK293 cells (e) or PARPi treated HEK293 cells (f) treated with 1 mM H2O2 for 15 min with and without 15 min recovery. Lines is the mean of spike-in normalised read counts between two biological replicates. g Single-gene view of TTchem-seq data for the FOXP4 gene for experimental conditions of (E -F). h Heatmaps of TTchem-seq signal ( normalised to spike -in). The signal is shown as the log2 fold-change between PARPi treated HEK293 cells and DMSO treated cells as control. Heatmaps are centred on transcription start site and are sorted by decreasing length of genes regions (TSS to +200 kb) with dashed lines indicating TES. .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 6, 2025. ; https://doi.org/10.1101/2025.11.06.686939doi: bioRxiv preprint .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 6, 2025. ; https://doi.org/10.1101/2025.11.06.686939doi: bioRxiv preprint Fig. 6. Lack of XRCC1 blocks gene body recovery of RNAPII elongation complexes during transcription recovery. a Nascent transcription levels quantified by 5-Ethynyluridine (EU) labelling and single cell quantification by immunofluorescence microscopy (EU-assay). Per cell EU signal is normalised to DAPI and log 10 scaled. HEK293 WT and XRCC1 knockouts (KO) cells were t reated with 1 mM H2O2 and EU-labelled for 30 min. b Representative images of global nascent transcription (EU immunofluorescence) from experiments in ( a) of WT HEK293 and XRCC1 KO cells following oxidative stress conditions. All image intensities are displayed with same exposure as the u ntreated control. c Heatmaps of TTchem-seq signal (normalised to spike-in) in HEK293 WT cells and XRCC1 KO cells. The signal is shown as the log2 fold-change between 1 hour recovery relative to untreated samples for the respective cell lines. Heatmaps are centred on TSS and are sorted by decreasing length of gene regions (TSS to +200 kb) with dashed lines indicating TES. d 4sU labelled RNA yield measured after streptavidin pull-down in HEK293 cells (WT), XRCC1 KO (average for two independently generated KO cell lines), HEK293 treated with PARPi (WT + PARPi) and XRCC1-KO treated with PARPi (XRCC1 KO + PARPi). All samples were performed in duplicates. e-f Metagene analysis of TTchem-seq data for 1 hour recovery for long protein coding genes (>90 kb, n=2,217). Metagene in (e) is centred on TSS (-2 kb to TSS to +80 kb) while metagene in (f) is scaled on transcription start site and transcript end site (-2 kb to TSS to TES +2 kb). Lines is the mean of spike-in normalised read counts between two biological replicates. g Single-gene view of TTchem-seq data for the CERS6 gene for experimental conditions of (e-f). h Heatmaps of TTchem-seq signal (normalised to spike-in). The signal is shown as the log2 fold-change between WT-DMSO control and PARPi treated XRCC1-KO cells. Heatmaps are centred at the transcription start site (TSS + 200 kb) and sorted based on gene lengths. .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 6, 2025. ; https://doi.org/10.1101/2025.11.06.686939doi: bioRxiv preprint .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 6, 2025. ; https://doi.org/10.1101/2025.11.06.686939doi: bioRxiv preprint Fig. 7. Lack of OGG1 activity blocks gene body recovery of RNAPII elongation complexes during transcription recovery. a Nascent transcription levels quantified by 5-Ethynyluridine (EU) labelling and single cell quantification by immunofluorescence microscopy (EU-assay). Per cell EU signal is normalised to DAPI and log 10 scaled. HEK293 WT and OGG1i treated HEK293 cells were t reated with 1 mM H2O2 and EU-labelled for 30 min. b Representative images of global nascent transcription (EU immunofluorescence) from experiments in (a) of WT HEK293 and OGG1i treated HEK293 cells following oxidative stress conditions. All image intensities are displayed with same exposure as the untreated control. c Heatmaps of TTchem-seq signal (normalised to spike-in) in HEK293 WT cells and OGG1i treated HEK293 cells. The signal is shown as the log2 fold -change between 1 hour recovery relative to untreated samples for the respective cell lines. Heatmaps are centred on TSS and are sorted by decreasing length of gene regions (TSS to +200 kb) with dashed lines indicating TES. d 4sU labelled RNA yield measured after streptavidin pull-down in HEK293 cells (WT), HEK293 treated with OGG1i, HEK293 treated with PARPi (WT + PARPi) and OGG1i treated HEK293 with PARPi (OGG1i + PARPi). All samples were performed in duplicates. e-f Metagene analysis of TT chem-seq signal for 1 hour recovery for long protein coding genes (>90 kb, n=2,217). Metagene in (e) is centred on TSS (-2 kb to TSS to +80 kb) while metagene in (f) is scaled on transcription start site and transcript end site (- 2 kb to TSS to TES +2 kb). Lines is the mean of spike-in normalised read counts between two biological replicates. g Single-gene view of TTchem-seq data for the PAM gene for experimental conditions of (e- f). h Heatmaps of TTchem-seq signal ( normalised to spike -in). The signal is shown as the log2 fold- change between WT-DMSO control and PARPi treated OGG1i treated HEK293 cells. Heatmaps are centred at the transcription start site (TSS + 200 kb) and sorted based on gene lengths. i Model of the transcriptional response to oxidative damage. In WT cells (left) oxidative damage induced by hydrogen peroxide or menadione results in rapid RNAPII arrest, NELF-mediated promoter proximal pausing and PARylation. As transcription recovers RNAPII elongation resumes from within the gene body. NELF is disassociated from early RNAPII elongation complexes and released into the gene body due to de- PARylation. In XRCC1 KOs or upon OGGi RNAPII elongation block is maintained (right top panel), PARPi treatment results in forced release of early RNAPII elongation complexes (right middle panel), .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 6, 2025. ; https://doi.org/10.1101/2025.11.06.686939doi: bioRxiv preprint whereas PARPi in XRCC1 KOs or OGG1i leads to forced release of early RNAPII elongation complexes with a sustained elongation block. .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted November 6, 2025. ; https://doi.org/10.1101/2025.11.06.686939doi: bioRxiv preprint