Excessive reactive oxygen species induce transcription-dependent replication stress

Abstract

Elevated levels of reactive oxygen species (ROS) reduce replication fork velocity by causing dissociation of the TIMELESS-TIPIN complex from the replisome. Here, we show that ROS generated by exposure of human cells to the ribonucleotide reductase inhibitor hydroxyurea (HU) promote replication fork reversal in a manner dependent on active transcription and formation of co-transcriptional RNA:DNA hybrids (R-loops). The frequency of R-loop-dependent fork stalling events is also increased after TIMELESS depletion or a partial inhibition of replicative DNA polymerases by aphidicolin, suggesting that this phenomenon is due to a global replication slowdown. In contrast, replication arrest caused by HU-induced depletion of deoxynucleotides does not induce fork reversal but, if allowed to persist, leads to extensive R-loop-independent DNA breakage during S-phase. Our work reveals a link between oxidative stress and transcription-replication interference that causes genomic alterations recurrently found in human cancer.

Fig. 1. ROS-induced replication slowdown causes replication fork stalling.

a–e Elimination of ROS or PRDX2 inhibition prevents nascent DNA degradation in BRCA2-depleted U2OS cells upon exposure to hydroxyurea (HU). a Experimental workflow of DNA fiber assays. ConA, Conoidin A (PRDX2 inhibitor; 0.5 μM); NAC, N-acetyl cysteine, (ROS scavenger; 5 mM); NUC, exogenous nucleosides (adenosine, guanosine, thymidine and cytosine; 20 μM each), Mirin, (MRE11 inhibitor; 50 μM). A thymidine (400 μM) chase was included for all conditions to stop IdU incorporation. b Representative images of DNA replication tracts. Scale bar, 10 μm. c Western blot analysis of the extracts of U2OS cells transfected with control siRNA (siLuc) or siRNA to BRCA2 (siBRCA2). d Scatter plot of the values of IdU/CldU tract length ratio obtained from three independent experiments for indicated conditions (n ≥ 371). e Scatter plot of the values of sister IdU tract length ratio (sister fork ratio; shorter tract/longer tract) obtained from three independent experiments for indicated conditions (n ≥ 100). f–i ROS induce replication fork stalling in a PRDX2 dependent manner in U2OS cells. f Experimental workflow. ConA and NAC were present at concentrations as in (a). g Representative DNA fiber images. NT, non-treated. Scale bar, 10 μm. h Scatter plot of the values of IdU/CldU tract length ratio obtained from three independent experiments for indicated conditions (n ≥ 371). i Scatter plot of the values of sister fork ratio obtained from three independent experiments for indicated conditions (n ≥ 90). j–m ZRANB3 depletion rescues HU-induced fork stalling in U2OS cells. j Experimental workflow. k Western blot analysis of the extracts of U2OS cells transfected with indicated siRNAs. l Scatter plot of the values of IdU/CldU tract length ratio obtained from three independent experiments for indicated conditions (n ≥ 402). m Scatter plot of the values of sister fork ratio obtained from three independent experiments for indicated conditions (n ≥ 96). d, e, h, i, l, m Red horizontal lines indicate the median; p values were calculated by Kruskal-Wallis test followed by Dunn’s multiple comparisons test. Source data are provided as a Source Data file.

Fig. 2. ROS-induced fork stalling is caused by co-transcriptional R-loops.

a–d Overexpression of RNase H1 prevents replication fork stalling in HU- and H₂O₂-treated cells. a Experimental workflow. Doxycycline (Dox; 1 ng/ml) was added to induce RNase H1 (RNH1) expression. b Western blot analysis of the extracts of U2OS T-REx [RNH1(WT)-GFP] cells treated with Dox for 24 h. c Scatter plot of the values of IdU/CldU tract length ratio obtained from three independent experiments for indicated conditions (n ≥ 319). d Scatter plot of the values of sister fork ratio obtained from three independent experiments for indicated conditions (n ≥ 105). e–g RNase H1 overexpression prevents replication fork stalling in TIMELESS-depleted cells (siTIM; 10 nM); (e) Experimental workflow. f Scatter plot of the values of replication tract length (IdU + CldU) obtained from three independent experiments for indicated conditions (n ≥ 413). g Scatter plot of the values of sister fork ratio obtained from four independent experiments for indicated conditions (n ≥ 106). h–j RNase H1 overexpression prevents HU-induced nascent DNA degradation in BRCA2-depleted U2OS cells. h Experimental workflow. i Scatter plot of the values of IdU/CldU tract length ratio obtained from three independent experiments for indicated conditions (n ≥ 471). j Scatter plot of the values of sister fork ratio obtained from three independent experiments for indicated conditions (n ≥ 67). c, d, f, g, i, j Red horizontal lines indicate the median; p values were calculated by Kruskal-Wallis test with Dunn’s multiple comparisons test. k HU induces fork reversal in a ROS- and R-loop-dependent manner. Left: Representative images of ongoing (top) and reversed (bottom) forks obtained by EM. Scale bar, 200 nm. P parental arm, D daughter arm, R regressed arm. Right: Quantification of the frequency of reversed forks for indicated conditions. U2OS T-REx [RNH1(WT)-GFP] cells were treated with 500 μM HU for 1 h. Dox was added 24 h before the treatment to induce RNH1 expression. DRB (100 μM) was added 2 h before HU treatment. NAC (5 mM) was added simultaneously with HU. Data are presented as mean ± SD, n = 3–4; p values were calculated by one-way ANOVA followed by Tukey’s test. In brackets, the total number of analyzed molecules is given. Source data are provided as a Source Data file.

Fig. 3. ROS induce transcription-replication conflicts and R-loop formation in a manner dependent on PRDX2.

a, b Co-localization of PCNA and elongating form of RNA polymerase II (RNAPII pS2) in U2OS cells after 1-h exposure to HU or H₂O₂ as determined by proximity ligation assays (PLA). a Top: Experimental workflow. EdU, 5-ethynyl-2′-deoxyuridine (10 μM). Bottom: Representative galleries of PLA signal in cell nuclei exported from the ScanR analysis program. Scale bar, 10 μm. b Top: Scatter plot of the number of PLA foci in individual cells for indicated conditions (n ≥ 1508). Bottom: Scatter plot of total DAPI (x-axis) and mean EdU (y-axis) intensities in individual cells. Colors indicate the number of PLA foci and correspond to the upper plot. Representative plots from three independent experiments yielding similar results are shown. NT non-treated, A.U. arbitraty units. c HU-induced co-localization between PCNA and elongating form of RNAPII depends on ROS. The plots shown are as in (b) (n ≥ 2607). U2OS cells were treated as in (a). N-acetyl cysteine (NAC; 5 mM) was present during the HU treatment. d–g ROS-dependent formation RNH1(D210N)-GFP foci upon exposure of U2OS T-REx [RNH1(D210N)-GFP] cells to 4 mM HU for 1 h. d Top: Experimental workflow. Bottom: Representative images of DAPI, RNH1(D210N)-GFP, and PCNA channels for indicated conditions. Scale bar, 10 μm. e Scatter plot of the number of RNH1(D210N)-GFP foci in PCNA-positive nuclei for indicated conditions (n ≥ 896). A representative plot from four independent experiments yielding similar results is shown. f Plot of the median values of the data sets represented in (e). Data are presented as mean ± SEM, n = 4; p values were calculated by one-way ANOVA followed by Tukey’s test. g Scatter plot of DAPI total (x-axis) and PCNA mean (y-axis) intensities in individual cells. Colors indicate the number of RNH1(D210N)-GFP foci, as shown in the legend on the right. h HU-induced accumulation of RNH1(D210N)-GFP foci in S-phase cells depends on PRDX2. Top: Experimental workflow. Bottom: Scatter plot as in (e) for indicated conditions (n ≥ 1097). i HU-induced co-localization between PCNA and elongating form of RNAPII depends on PRDX2. Top: Experimental workflow. Bottom: Scatter plot as in (b) for indicated conditions (n ≥ 558). A representative plot from two independent experiments yielding similar results is shown. j, k TIMELESS depletion induces the formation of RNH1(D210N)-GFP foci in S-phase nuclei of U2OS T-REx [RNH1(D210N)-GFP]. j as in (e) for indicated conditions (n ≥ 370). k as in (g) for indicated conditions. b, c, e, h, i, j Horizontal lines indicate the median; p values were calculated by Kruskal-Wallis test followed by Dunn’s multiple comparisons test. Source data are provided as a Source Data file.

Fig. 4. MUS81-LIG4-PRIMPOL axis mediates replication restart following ROS-induced fork stalling.

a Western blot analysis of the extracts of U2OS cells transfected with indicated siRNAs. b ZRANB3 depletion restores replication fork progression in HU-treated U2OS cells in a manner dependent on PRIMPOL and the proteins required for restarting R-loop-stalled forks. Top: Experimental workflow. Bottom: Scatter plot of the values of IdU/CldU tract length ratio obtained from two independent experiments for indicated conditions (n ≥ 211). c, d Restart of DNA synthesis following replication arrest by HU depends on the same set of proteins as in b. c Experimental workflow of replication restart assays and representative images of replication tracts corresponding to restarted and stalled forks, respectively. Scale bar, 10 μm. d Quantification of the replication fork stalling events in cells depleted of indicated proteins. e–g MUS81 and PRIMPOL act in the same replication restart pathway. e Experimental workflow of replication restart assays with wild-type and MUS81 knockout (KO) HeLa Kyoto cells. f Western blot analysis of the extracts of cells in e transfected with indicated siRNAs. g Quantification of replication fork stalling events for the indicated conditions. h HU-induced fork stalling in MUS81-depleted U2OS cells depends on ROS. Top: Experimental workflow of replication restart assays. NAC, N-acetyl cysteine (5 mM). Bottom: Quantification of the replication fork stalling events for indicated conditions. d, g, h Data are presented as mean ± SD, n = 3; p values were calculated by one-way ANOVA followed by Tukey’s test. i Western blot analysis of the extracts of U2OS T-REx [RNH1(WT)-GFP] cells transfected with indicated siRNAs. j Restoration of replication fork progression in HU-treated cells by RNase H1 overexpression depends on MUS81 and PRIMPOL, but not RECQ1. Top: Experimental workflow of DNA fiber assays with U2OS T-REx [RNH1(WT)-GFP] cells. Doxycycline (Dox; 1 ng/ml) was added to induce RNase H1 (RNH1) expression. Bottom: Scatter plot of the values of IdU/CldU tract length ratio obtained from three independent experiments for indicated conditions (n ≥ 406). k Sensitivity of DNA replication tracts to S1 nuclease upon indicated conditions. Top: Experimental workflow of DNA fiber assays. TRP, triptolide (1 μM). Bottom: Scatter plot of the values of IdU/CldU tract length ratio obtained from three independent experiments for indicated conditions (n ≥ 417). b, j, k Red horizontal lines indicate the median; p values were calculated by Kruskal-Wallis test followed by Dunn’s multiple comparisons test. Source data are provided as a Source Data file.

Fig. 5. R-loop-dependent fork stalling is induced by partial inhibition of DNA synthesis with aphidicolin.

a–c RNase H1 overexpression rescues replication fork stalling induced by a low dose of aphidicolin (APH). a Experimental workflow of DNA fiber assays with U2OS T-REx [RNH1(WT)-GFP] cells. Doxycycline (Dox; 1 ng/ml) was added to induce the expression of RNase H1 (RNH1). b Scatter plot of the values of sister fork ratio obtained from three independent experiments for indicated conditions (n ≥ 153). c Scatter plot of the values of IdU/CldU tract length ratio obtained from three independent experiments for indicated conditions (n ≥ 402). d, e Co-localization of PCNA and elongating form of RNAPII in S-phase nuclei of U2OS cells after 1 h treatment with 0.2 μM APH, as revealed by proximity ligation assay (PLA) and EdU-pulse labeling. d Top: Experimental workflow. EdU, 5-ethynyl-2′-deoxyuridine (10 μM). Bottom: Representative galleries of PLA signal in cell nuclei exported from ScanR analysis program. Scale bar, 10 μm. e Top: Scatter plot of the number of PLA foci in individual cells for indicated conditions (n ≥ 2064). Gray bars indicate the median; p values were calculated by Mann-Whitney test. Bottom: Scatter plot of total DAPI (x-axis) and mean EdU (y-axis) intensities in individual cells. Colors indicate the number of PLA foci and correspond to the upper plot. Representative plots from three independent experiments yielding similar results are shown. NT non-treated. f Low but not high doses of APH induce nascent DNA degradation in BRCA2-depleted U2OS cells. Top: Experimental workflow of DNA fiber assays. A thymidine (400 μM) chase was included for all conditions to stop IdU incorporation. Bottom: Scatter plot of the values of IdU/CldU tract length ratio obtained from two independent experiments for indicated conditions (n ≥ 332). b, c, f Red horizontal lines indicate the median; p values were calculated by Kruskal-Wallis test followed by Dunn’s multiple comparisons test. Source data are provided as a Source Data file.

Fig. 6. Replication slowdown leads to R-loop-dependent micronucleation and accumulation of 53BP1 nuclear bodies in nascent G1 cells.

a, b HU induces R-loop-dependent micronucleation. a Top: Experimental workflow. U2OS T-REx [RNH1(WT)-GFP] cells were treated with 50 μM HU for 16 h. RNase H1 expression was induced by doxycycline (Dox; 1 ng/ml) 24 h before HU addition. Cytochalasin B (2 μg/ml; 16 h) was added to block cells in cytokinesis. Bottom: Representative images of binucleated cells without or with (white arrow) a micronucleus. Scale bar, 10 μm. b Quantification of the frequency of micronuclei for indicated conditions. c, d R-loop-dependent formation of G1-specific 53BP1 nuclear bodies in U2OS T-REx [RNH1(WT)-GFP] cells treated with 50 μM HU for 16 h. c, Top: Experimental workflow. Bottom: Representative images of DAPI, cyclin A and 53BP1 channels. Scale bar, 10 μm. d Percentages of cyclin A-negative cells (G1) with > 3 53BP1 nuclear bodies for indicated conditions. e TIMELESS depletion induces R-loop-dependent micronucleation. Percentage of binucleated U2OS T-REx [RNH1(WT)-GFP] cells with micronuclei is plotted for indicated conditions. siTIM or siLuc were transfected for 24 h. Cytochalasin B (2 μg/ml) was added for the last 16 h before fixation to block cells in cytokinesis. RNH1 expression was induced with doxycycline (1 ng/ml) 24 h before siRNA transfection. f TIMELESS depletion does not further exacerbate the micronucleation phenotype induced by 50 μM HU in U2OS cells. Top: Experimental workflow. Bottom: Quantification of the frequency of micronuclei for indicated conditions. g HU-induced micronucleation is suppressed by PRDX2 depletion. Top: Experimental workflow. Bottom: Quantification of the frequency of micronuclei for indicated conditions. b, d, e, f, g Data are presented as mean ± SD, n = 4; p values were calculated by one-way ANOVA followed by Tukey’s test. b, e–g For each condition, at least 150 binucleated cells were examined for the presence of micronuclei in each experiment. d For each condition, at least 300 cyclin A-negative cells were analyzed in each experiment to determine the number of 53BP1 nuclear bodies. Source data are provided as a Source Data file.

Fig. 7. Replication arrest due to dNTP shortage leads to R-loop-independent DNA breakage.

a Scatter plots showing the mean intensity of RPA2 (x-axis) versus the mean intensity of RPA2 pS4/8 (y-axis) in individual U2OS cells for indicated conditions, as measured by QIBC. Representative plots from three independent experiments yielding similar results are shown. At least 1000 cells were analyzed for each condition in each experiment. Cells were treated with 4 mM HU for 4 h in the absence or presence of N-acetyl cysteine (NAC; 5 mM) or exogenous nucleosides (NUC; 20 μM each). Individual cells are colored based on RPA and RPA2 pS4/8 intensities and classified as follows: RPA-negative (green), RPA-positive (blue) and RPA2 pS4/8-positive (red; dashed box). The numerical values inside the plots represent the percentage of RPA2 pS4/8-positive cells in the population. NT, non-treated; A.U., arbitraty units. b Plot of the percentage of RPA2 pS4/8-positive cells from the QIBC analysis represented in (a). c Plot of the mean intensity of RPA2 from the QIBC analysis represented in (a). Data are normalized to the values obtained for NT. d–f The same as in (a–c) for U2OS T-REx [RNH1(WT)-GFP] cells. Representative plots from four independent experiments yielding similar results are shown. RNase H1 (RNH1) expression was induced by doxycycline (1 ng/ml) 24 h before the addition of HU. g A model of consequences of replication fork slowdown and replication arrest. ROS levels are frequently elevated in cancer cells by oncogenes such as HRasV12 or as a consequence of alterations in the cellular metabolism. ROS can be also generated during cancer therapy by ionizing radiation or chemotherapeutics such as mitomycin C or cisplatin. b, c, e, f Data are presented as mean ± SEM; p values were calculated by one-way ANOVA followed by Tukey’s test. Source data are provided as a Source Data file.