DDX17 helicase promotes resolution of R-loop-mediated transcription-replication conflicts in human cells

Abstract

R-loops are three-stranded nucleic acid structures composed of an RNA:DNA hybrid and displaced DNA strand. These structures can halt DNA replication when formed co-transcriptionally in the opposite orientation to replication fork progression. A recent study has shown that replication forks stalled by co-transcriptional R-loops can be restarted by a mechanism involving fork cleavage by MUS81 endonuclease, followed by ELL-dependent reactivation of transcription, and fork religation by the DNA ligase IV (LIG4)/XRCC4 complex. However, how R-loops are eliminated to allow the sequential restart of transcription and replication in this pathway remains elusive. Here, we identified the human DDX17 helicase as a factor that associates with R-loops and counteracts R-loop-mediated replication stress to preserve genome stability. We show that DDX17 unwinds R-loops in vitro and promotes MUS81-dependent restart of R-loop-stalled forks in human cells in a manner dependent on its helicase activity. Loss of DDX17 helicase induces accumulation of R-loops and the formation of R-loop-dependent anaphase bridges and micronuclei. These findings establish DDX17 as a component of the MUS81-LIG4-ELL pathway for resolution of R-loop-mediated transcription-replication conflicts, which may be involved in R-loop unwinding.

Figure 1.

BioID-based proteomic analysis of R-loop interactome at sites of transcription–replication conflicts. (A) Schematic representation of BioID screen of proteins associated with R-loops. The expression of RNH1(D210N)-BioID2-HA fusion protein in U-2 OS T-REx RNH1(D210N)-BioID2-HA cells is induced by doxycycline. CPT (camptothecin, 100 nM) promotes the formation of co-transcriptional R-loops. (B) Top panel: Experimental workflow of protein biotinylation by RNH1(D210N)-BioID2-HA in living cells. Bottom panel: Volcano plots showing significantly differentially abundant proteins for comparison of data from RNH1(D210N)-BioID2-HA-expressing cells treated with vehicle alone (DMSO) or 100 nM CPT for 90 min to control cells in which expression of RNH1(D210N)-BioID2-HA was not induced (control). The -log₁₀ (permutation-based FDR corrected P value) is plotted against the log₂ (fold change: DMSO/control) or log₂ (fold change: CPT/control). Each point represents the difference in abundance of particular protein between the two conditions plotted against the level of statistical significance. Data were processed using the Perseus program with false discovery rate (FDR) of 0.1 (black lines). Significantly enriched proteins are in the right upper quadrant of the volcano plot, depicted in black. DEAD/DEAH box helicases DDX17, DDX5, DHX15, DHX9 and DDX3X helicases are depicted in pink. (C) Venn diagram showing overlap of significantly enriched proteins in DMSO and CPT samples. (D) Significantly enriched proteins from BioID screen clustered into individual classes according to gene ontology molecular function (GO MF) annotations and protein domains. The x-axis indicates statistical significance of overrepresentation.

Figure 2.

DDX17 promotes replication restart following R-loop-mediated fork stalling. (A) Western blot analysis of the lysates of U-2 OS cells transfected with indicated siRNAs. Note that DDX17 exists in two isoforms, p72 and p82, which result from alternative translation initiation sites. (B) Left panel: Experimental workflow of DNA fiber assays. U-2 OS cells were transfected with indicated siRNAs for 72 h followed by pulse-labeling with CldU and IdU. During the IdU labeling, cells were treated with 100 nM CPT or DMSO. Where indicated, 10 μM PARP inhibitor olaparib (PARPi) was added 2 h before labeling and was also present during the labeling. Right panel: Representative immunofluorescence images of replication tracts on DNA fibers of U-2 OS cells transfected with control siRNA (siLuc) and treated as indicated. Scale bar, 10 μm. (C) Effect of the depletion of DDX17, DHX9, DDX5, DHX15 and DDX3X respectively, on the rescue of CPT-induced replication fork slowing by PARP inhibition (PARPi). Scatter plot of the values of IdU/CldU tract length ratio obtained for indicated conditions (n ≥ 300) is shown. Red horizontal lines indicate mean value. ns, non-significant (P > 0.05); **P ≤ 0.01; ****P ≤ 0.0001 (Mann–Whitney test). (D) Effect of the depletion of DDX17, DHX9, DDX5, DHX15 and DDX3X, respectively, on replication fork velocity in U-2 OS cells (n ≥ 300). Red horizontal lines indicate mean value. ns, non-significant (P > 0.05); **P ≤ 0.01; ****P ≤ 0.0001 (Mann–Whitney test). (E) Western blot analysis of the lysates of U-2 OS cells transfected with indicated siRNAs. (F) DDX17 is required for the rescue of CPT-induced replication fork slowing by ZRANB3 depletion in U-2 OS cells. Top panel: Experimental workflow of DNA fiber assay. U-2 OS cells were transfected with indicated siRNAs for 72 h followed by pulse-labeling with CldU and IdU. During the second labeling, cells were treated with 100 nM CPT or DMSO. Bottom panel: Scatter plot of the values of IdU/CldU tract length ratio obtained for indicated conditions (n ≥ 300). Red horizontal lines indicate mean value. ns, non-significant (P > 0.05); ****P ≤ 0.0001 (Mann–Whitney test). (G) Western blot analysis of the lysates of U-2 OS T-REx [RNH1(WT)-GFP] cells transfected with indicated siRNAs. (H) Experimental workflow of DNA fiber assays with U-2 OS T-REx [RNH1(WT)-GFP] cells for (I) and (J). Expression of RNH1(WT)-GFP was induced by addition of doxycycline (DOX; 1 ng/ml) to eliminate R-loops. (I) Loss of DDX17 induces R-loop-dependent sister fork asymmetry. Top panel: representative images of symmetric and asymmetric replication tracts of sister forks. Scale bar, 10 μm. Bottom panel: scatter plot of the values of sister IdU tract length ratio (sister fork ratio; shorter IdU tract/longer IdU tract) measured for the indicated conditions (n ≥ 125). Red horizontal lines indicate mean value. ns, non-significant (P > 0.05); ****P ≤ 0.0001 (Mann–Whitney test). (J) Loss of DDX17 increases the frequency of stalled replication forks in an R-loop-dependent manner. Top panel: representative immunofluorescence images of replication tracts corresponding to ongoing (CldU tract is followed IdU tract) and stalled (CldU tract is not followed IdU tract) replication. Scale bar, 10 μm. Bottom panel: quantification of the replication fork stalling events. Data are mean ± SD, n = 5. ns, non-significant (P > 0.05); *P ≤ 0.05 (one-way ANOVA).

Figure 3.

Loss of DDX17 induces accumulation of R-loops during S-phase. (A) Loss of DDX17 induces accumulation of R-loops. Top panel: schematic representation of experimental workflow. U-2 OS T-REx [RNH1(D210N)-GFP] cells grown on coverslips were transfected with indicated siRNAs for 72 h and treated with doxycycline (1 ng/ml) for last 24 h. Prior to fixation and staining with anti-PCNA antibody, cells were pre-extracted to eliminate unbound RNase H1. Bottom panel: Representative images. Nuclear foci of RNH1(D210N)-GFP indicate sites of R-loop formation. Scale bar, 10 μm. (B) Quantification of the number of RNH1(D210N)-GFP nuclear foci in PCNA-positive and PCNA-negative cells for indicated conditions (n ≥ 1700). Red horizontal lines indicate mean value. ****P ≤ 0.0001 (Mann–Whitney test). (C) Western blot analysis of extracts of U-2 OS T-REx [RNH1(D210N)-GFP] cells transfected with indicated siRNAs. (D) RNH1(D210N)-GFP foci accumulate in late S and G2 phase of the cell cycle in DDX17-depleted cells. Scatter plots of PCNA (y-axis) and DAPI (x-axis) intensities in individual DDX17-depleted (right) or mock-depleted (left) cells (n ∼ 450) to determine the cell cycle phase. The color code indicates the number of RNH1(D210N)-GFP foci per nucleus. (E) RNA:DNA slot blot analysis of genomic DNA isolated form U-2 OS cells transfected with indicated siRNAs. Genomic DNA was treated or not with RNase H. Blots were immunostained using S9.6 and anti-ssDNA antibodies. (F) Quantification of S9.6 signal on slot blots represented in (E). For each sample, S9.6 signal intensity was normalized to ssDNA intensity. The resulting values are normalized to the control siRNA (siLuc) and represent mean ± SD, n = 3. ns, non-significant (P > 0.05); * P ≤ 0.05 (One-way ANOVA). (G) Loss of DDX17 induces accumulation of R-loops at R-loop-prone loci in U-2 OS cells. DRIP-qPCR analyses were performed using S9.6 monoclonal antibody and genomic DNA isolated from U-2 OS cells transfected with indicated siRNAs. The R-loop-prone regions of the APOE, BTBD19, GADD45A and RPL13A genes were examined. An R-loop-free SNRPN gene locus was used as a negative control. Where indicated, the genomic DNA was treated with RNase H to remove RNA:DNA hybrids. Data are plotted as a percentage of the input, and represent the mean ± SD, n = 3. (H,I) DDX17 associates with R-loops. (H) U-2 OS cells were pre-extracted and sonicated to generate chromatin fragments with DNA size ranging from 800–200 bp. After centrifugation, sonicated cell lysates were immunoprecipitated with control IgG or S9.6 antibody followed by western blot analysis. (I) as in (H) except that, before immunoprecipitation, cell lysates were supplemented with 3 mM MgCl₂ and incubated with RNaseH for 45 min at 37°C as indicated.

Figure 4.

Loss of DDX17 induces the formation of R-loop-dependent anaphase bridges and micronuclei. (A, B) Loss of DDX17 induces formation of bulky anaphase bridges. (A) Top panel: Schematic representation of the experimental workflow of bulky anaphase bridge assay. U-2 OS T-REx [RNH1(WT)-GFP] cells were transfected with indicated siRNAs for 72 h and treated or not with doxycycline for the last 24 h to induce RNH1(WT)-GFP expression. Cells were synchronized at G2/M transition by treatment with RO-3306 for 16 h and subsequently released for 90 min into fresh medium, followed by fixation and DAPI staining. Bottom panel: Representative images of anaphase cells for indicated conditions. Red arrow points to a bulky anaphase bridge. Scale bar, 10 μm. (B) Quantitative analysis. Data are mean ± SD, n = 3. At least 25 anaphase cells were analyzed per experiment for each condition. ns, non-significant (P > 0.05); * P ≤ 0.05; ** P ≤ 0.01 (one-way ANOVA). (C,D) Loss of DDX17 induces formation of micronuclei. (C) Top panel: Schematic representation of the experimental workflow of micronucleus assay. U-2 OS T-REx [RNH1(WT)-GFP] cells were transfected with indicated siRNAs for 72 hours and treated or not with doxycycline for last 24 h. The last 16 h, cells were incubated with cytochalasin B (2 μg/ml) to inhibit cytokinesis, and subsequently fixed and stained with DAPI. Bottom panel: Representative images of binucleated cells for indicated conditions. Red arrow points to a micronucleus. Scale bar, 10 μm. (D) Quantitative analysis. Data are mean ± SD, n = 4. At least 150 binucleated cells were analyzed per experiment for each condition. ns, non-significant (P > 0.05); **** P ≤ 0.0001 (one-way ANOVA).

Figure 5.

DDX17 acts in a common pathway with MUS81 to suppress R-loop-mediated replication stress. (A) Western blot analysis of the lysates of wild-type and MUS81 knockout HeLa Kyoto cells transfected with indicated siRNAs. (B) Scatter plot of the values of sister IdU tract length ratio (sister fork ratio; shorter IdU tract/longer IdU tract) measured for the indicated conditions (n ≥ 120). Red horizontal lines indicate mean value. ns, non-significant (P > 0.05); ****P ≤ 0.0001 (Mann–Whitney test). (C) Quantification of the replication fork stalling events (CldU tracts without an IdU tract) for indicated conditions. Data are mean ± SD, n = 3. ns, non-significant (P > 0.05); **P ≤ 0.01; ****P ≤ 0.0001 (one-way ANOVA). (D) Quantification of micronucleation events for indicated conditions. Data are mean ± SD, n = 5. ns, non-significant (P > 0.05); **P ≤ 0.01; ***P ≤ 0.001 (one-way ANOVA).

Figure 6.

DDX17 helicase unwinds RNA:DNA hybrids in vitro. (A) Purified DDX17(WT) and DDX17(K142R) subjected to SDS-PAGE and Coomassie Blue staining. (B, C) DDX17 unwinds RNA:DNA duplexes but not DNA:DNA duplexes. 25-bp RNA:DNA (B) or DNA:DNA (C) duplexes with 3′ or 5′ 20-nt overhang were subjected to helicase assay. Reactions containing 5 nM oligonucleotide substrate and 60 nM DDX17, wild type (WT) or K142R mutant, were carried out at 37°C for 20 min. Reactions were initiated by adding ATP to a final concentration of 5 mM. Reaction products were analyzed on native polyacrylamide gels. Schemes of oligonucleotide substrates are shown on the top. The position of FAM label is indicated (green ball). Lane 1, reaction without protein; lane 4, heat denatured substrate. (D) Unwinding of 25-bp RNA:DNA duplexes with overhangs of various lengths (5 nM) were incubated with increasing concentrations of DDX17 (0–60 nM). Unwinding reactions were carried at 37°C for 20 min and analyzed as in (B). Left panel: Representative images of gels. Right panel: Quantification of gels represented in the left panel. Data are mean ± SD, n = 3. (E) ATPase activity of DDX17 in the presence of RNA or DNA oligonucleotides, or without any nucleic acid (no NA). Reactions containing 1 nM oligonucleotide, 40 nM DDX17 (WT or K142R) and 2 mM ATP were carried out at 37°C for 60 min. The concentration of inorganic phosphate released by ATP hydrolysis was estimated by malachite green assay. Data are mean ± SD, n = 3. ns, non-significant (p > 0.05); **P ≤ 0.01; ****P ≤ 0.0001 (one-way ANOVA).

Figure 7.

DDX17 helicase preferentially unwinds R-loops in vitro. (A) Purified DDX17 unwinds R-loop but not D-loop structures. 5 nM R-loop (left panel) and D-loop (right panel) substrates were incubated with 60 nM wild-type (WT) or K142R mutant of DDX17 at 37°C for 10 min. Reactions were initiated by adding ATP to a final concentration of 5 mM. Reaction products were analysed on native polyacrylamide gels. Schemes of substrates are indicated next to the bands. The position of FAM label is indicated as a green ball in the illustration. Lane 1, reaction without protein; lane 4, heat denatured substrate; lane 5, tailed duplex used as a marker. (B) Unwinding of R-loop and ssDNA-tailed RNA:DNA hybrid by DDX17 helicase. 5 nM R-loop or ssDNA-tailed RNA:DNA hybrid substrates were incubated with increasing concentrations of DDX17 helicase (0–60 nM) as indicated. Unwinding reactions were initiated by adding ATP to a final concentration of 5 mM and carried out at 37°C for 10 min. Reaction products were analysed as in (A). Right panel: Quantification of gels represented in the left panel. Data are mean ± SD, n = 3.

Figure 8.

Helicase activity of DDX17 is required for the restart of R-loop-stalled forks. (A) Top panel: Experimental workflow. U-2 OS T-REx cells carrying DDX17(WT)-FLAG or DDX17(K142R)-FLAG transgenes were transfected with siRNA targeting endogenous DDX17 for 72 h. After 48 h, cells were treated with doxycycline to induce expression of the transgene. Bottom panel: Western blot analysis of extracts of the above cells. (B) PARP inhibition does not rescue CPT-induced replication fork slowing in cells expressing DDX17(K142R). Scatter plot of the values of IdU/CldU tract length ratio obtained for indicated conditions is shown (n ≥ 300). Red horizontal lines indicate mean value. ns, non-significant (P > 0.05); ****P ≤ 0.0001 (Mann–Whitney test). (C) Cells expressing DDX17(K142R) display a reduced replication fork velocity. Scatter plot of the values of velocity (kb/min) of individual forks obtained for indicated conditions is shown (n ≥ 300). Experimental workflow of DNA fiber assay used is shown in Figure 2H. Red horizontal lines indicate mean value. ns, non-significant (P > 0.05); ***P ≤ 0.001; ****P ≤ 0.0001 (Mann–Whitney test). (D) Cells expressing DDX17(K142R) display sister fork asymmetry. Scatter plot of the values of sister IdU tract length ratio (sister fork ratio; shorter IdU tract/longer IdU tract) measured for the indicated conditions is shown (n ≥ 124). Red horizontal lines indicate mean value. ns, non-significant (P > 0.05); ****P < 0.0001 (Mann–Whitney test). (E) Cells expressing DDX17(K142R) show increased frequency of stalled replication forks determined as in Figure 2J. Data are mean ± SD, n = 3. ns, non-significant (P > 0.05); *P ≤ 0.05; **P ≤ 0.01 (One-way ANOVA). (F) Cells expressing DDX17(K142R) show increased micronucleation after mitosis determined as in Figure 4C,D. Data are mean ± SD, n = 3. ns, non-significant (P > 0.05); * P ≤ 0.05; **P ≤ 0.01 (one-way ANOVA).