DEAD-box RNA helicases belong to a large group of RNA-processing factors and play vital roles unwinding RNA helices and in ribosomal RNA biogenesis. Emerging evidence indicates that RNA helicases are associated with genome stability, yet the mechanisms behind this association remain poorly understood. In this study, we performed a comprehensive analysis of RNA helicases using multiplatform proteogenomic databases. More than 50% (28/49) of detected RNA helicases were highly expressed in multiple tumor tissues, and more than 60% (17/28) of tumor-associated members were directly involved in DNA damage repair (DDR). Analysis of repair dynamics revealed that these RNA helicases are engaged in an extensively broad range of DDR pathways. Among these factors is DDX21, which was prominently upregulated in colorectal cancer. The high expression of DDX21 gave rise to frequent chromosome exchange and increased genome fragmentation. Mechanistically, aberrantly high expression of DDX21 triggered inappropriate repair processes by delaying homologous recombination repair and increasing replication stress, leading to genome instability and tumorigenesis. Treatment with distinct chemotherapeutic drugs caused higher lethality to cancer cells with genome fragility induced by DDX21, providing a perspective for treatment of tumors with high DDX21 expression. This study revealed the role of RNA helicases in DNA damage and their associations with cancer, which could expand therapeutic strategies and improve precision treatments for cancer patients with high expression of RNA helicases.

Significance:

The involvement of the majority of tumor-associated RNA helicases in the DNA damage repair process suggests a new mechanism of tumorigenesis and offers potential alternative therapeutic strategies for cancer.

Genome instability characterized by an increased mutation burden has been suggested to be a key driver of tumorigenesis (1). DNA damage response (DDR) factors play critical roles in maintaining genome stability. The aberrantly high expression of these proteins impairs the homeostasis of DDR, leading to the accumulation of DNA damage and posing a threat to genome integrity. For instance, defects in the DNA mismatch repair machinery caused by the inactivation of mismatch repair proteins lead to microsatellite instability (MSI) in the genome and give rise to colorectal and gastric cancers (2, 3). Microsatellite stability, which is classified as microsatellite stability (MSS) and MSI, has been associated with drug response in patients with stage II colorectal cancer. Patients with MSS show a better response to 5-fluorouracil (5-FU) than those with MSI. In contrast, patients with MSI are reported to be more sensitive to immune therapy (4). Therefore, it is assumed that a genome with MSI harbors a high frequency of somatic mutations, consequently resulting in adequate numbers of neoantigens that can be recognized by the immune system (5–7). Moreover, recent studies have revealed that cancer cells with MSI display selective vulnerability to WRN inhibition (8, 9). In addition, BRCA mutations, which cause a deficiency in homologous recombination (HR), lead to vulnerability to breast and ovarian cancers. PARP inhibitors have emerged as a promising targeted therapy for BRCAness HR-deficient (HRD) tumors (10, 11). Therefore, the exploration of abnormalities in DDR process and associated proteins not only enriches the understanding of tumorigenesis but also provides new insight into precision therapies.

Accumulating evidence demonstrates that RNA-processing factors, which are engaged in conserved functions in RNA metabolism, play unexpected roles in DDR. In addition to RNA processing, these factors also execute roles in DDR by cooperating with well-recognized repair proteins, such as BRCA1 and RPA, or being modified by DNA damage enzymes, such as ATM, ATR, and PARP (12, 13). The initial recognition of RNA-processing factors in DDR originates from transcription-coupled repairs. DNA/RNA hybrids, which are transcriptional derivatives, are reported to facilitate the recruitment of DDR factors in the early stage; however, their persistent existence would impede further repair progression (14). The formation and resolution of DNA/RNA hybrids during repair processes rely on a series of RNA-processing factors (15, 16). In addition to DNA/RNA hybrids, some noncoding RNAs and their binding factors have also been revealed to be involved in DDR and genome maintenance (17–19). Therefore, due to their dual functions in both DDR and transcriptional control, some RNA-processing factors play a critical role in genome maintenance, and their dysfunctions contribute to tumorigenesis (20).

The DEAD/H-box RNA helicase SF2 (superfamily 2) family, DDXs and DHXs, are the most abundant eukaryotic RNA processing factors that function in unwinding RNA helices (21). DDXs and DHXs share a conserved core helicase region, named the DEAD box or DEAH box, which is necessary for ATP binding and helicase activity (22). Interestingly, due to their diverse C- and N-terminal domains, DDXs and DHXs have multifaceted functions beyond RNA processing. The functions of RNA helicases attributed to R-loops mediation during transcription bring up a new insight into the roles towards genome maintenance (23–25). DDX21 is revealed to resolve genomic R loops and safeguard genome stability in cooperation with SIRT7 during the transcription (23). Extensive evidence indicates that the aberrant expression or dysfunction of RNA helicases is associated with human diseases, such as neurodevelopmental disorders. However, the cancer-associated functions of DDXs and DHXs remain controversial. It is unclear how RNA helicases are required for genome integrity in response to stresses given that mutations and genomic instability are frequently occurred in cancers in response to metabolic and replicative stresses. Thus, more comprehensive investigations are required to understand the precise functions of DDX and DHX family members in distinct cancers, especially with an insight of functions in genome instability in response to stresses.

To obtain more insights into the roles of DExD/H box family members in tumorigenesis and genome maintenance, we carried out a large-scale analysis of RNA helicases using proteogenomic databases and then investigated their roles in DDR. We found that the expression levels of 28 DDX and DHX family members were consistently increased in tissues from multiple types of tumors and that more than 60% exhibited a DNA damage response. Among them, DDX21 was the most prominently upregulated protein in colorectal cancers. We found that the aberrantly high expression of DDX21 increased replication stress and delayed HR repair, which eventually triggered genome instability and induced tumorigenesis. We also found that the high expression of DDX21 induced higher sensitivity to some chemotherapeutic drugs, providing a new precision treatment strategy for colorectal cancer with high DDX21 expression.

Cell culture and clinical patient samples

U2OS, COCA2, and 293T cells were cultured in DMEM (Gibco, #8120506) and supplemented with 10% FBS (Biological Industries, #18224477). RKO, SW480, SW620, LOVO, HCT8 and HCT116 cells were cultured in RPMI1640 (Gibco, #8120303), supplemented with 10% FBS. All cell lines were maintained and passaged less than 20 times at 37°C in an atmosphere containing 5% CO2 and tested negatively for Mycoplasma contamination. All the colon cancer cell lines were identified by Shanghai Genesky Biotechnologies Ins. Samples of human colon cancer tissues and matched normal adjacent tissues from patients were collected from the Xiangya Hospital Central South University (Changsha, China). The study was approved by the Animal Ethics Committee of Xiangya Hospital Central South University (Changsha, China).

405 nm Laser microirradiation

The method of 405 nm laser microirradiation plus light sensitizer 8-methoxypsoralen (8-MOP) to induce DNA damage has been described previously (26). A Leica DM6500 confocal microscopy 405 nm laser system was used for microirradiation and image acquisition. U2OS cells were seeded on a 35-mm glass bottom dish (MatTek) and transfected with indicated plasmids. Before 405 nm laser irradiation, cells were treated with 100 μmol/L 8-MOP in the medium for 2 to 5 minutes. During microirradiation and image acquisition, cells were maintained at 37°C using a heated chamber control. FRAP mode was used to measure dynamics of recruitment, and the output of the 405 nm laser (50 mW) was set the maximum power continue to 500 ms. Images were acquired in three Z planes with 0.8-μm plane spacing. At least 15 cells were irradiated in each experiment. The three Z planes for every time point were combined using average intensity projection to compensate for cell movement and the images from all time points were registered. We measured the average intensity ratios of the sites of microirradiation to a nearby region, which indicate the ultimately recruitment intensity of proteins. The kinetics of recruitment of the protein was fitted to Weibull distribution equation extracted from SigmaPlot (f = if(x⇐x0-b*((c-1)/c)⁁(1/c), 0, a*((c-1)/c)⁁((1-c)/c)*(abs((x-x0)/b+((c-1)/c)⁁(1/c))⁁(c-1))*exp(-abs((x-x0)/b+((c-1)/c)⁁(1/c))⁁c+(c-1)/c))). Overall half-times of recruitment and removal of the proteins were obtained from their fitted curves.

Immunofluorescence

RKO control and DDX21-stable cells were seeded on 14-mm slides and treated with indicated drugs for 24 hours (Fig. 4), and then released into fresh medium. Cells were collected at the indicated time point and fixed in 4% formaldehyde (PFA) solution for 15 minutes at room temperature, followed by permeabilization with 0.2% Triton X-100 in PBS for 10 minutes. Cells were blocked with 2% BSA for 30 minutes and then incubated with primary antibodies [anti-RPA pS33 (Sigma, #PLA0310, 1:500); anti-RAD51 (Abcam, #ab63801, RRID: AB_1142428 1:400); anti-γH2AX (Millipore, #05–636, RRID: AB_309864, 1:1000); anti-Flag (Sigma, #F1804, RRID: AB_262044, 1:1000)] at 4°C overnight. After being washed three times with PBST (0.05% Tween-20 in PBS), cells were incubated with the secondary antibody dilution (Invitrogen, 1:500) for 1 hour at room temperature in the dark. Slides were stained with DAPI and sealed in mounting medium (Boster, #AR1109) for image acquisition.

For endogenous DDX21 and γH2AX staining, RKO and U2OS cells were seeded on a 35-mm glass bottom dish. Cells were treated with 100 μmol/L 8-MOP in medium for 2 to 5 minutes before using 405 nm laser microirradiation as described above. At least 100 cells were irradiated in each experiment. After microirradiation, cells were fixed in 4% PFA for 15 minutes immediately after 405 nm laser microirradiation, then followed by permeabilization and blocking as described above. Primary antibodies [anti-DDX21 (ProteinTech, #10528–1-AP, 1:50); anti-γH2AX] were applied to incubate the cells overnight at 4°C. Alexa Fluor 488 anti-mouse or Alexa Fluor 594 anti-rabbit antibodies were used as secondary antibodies. Images were acquired using a Leica TCS SP8 confocal laser microscopy system.

Analysis of efficiency of HR and NHEJ

DR-GFP and EJ5-GFP reporter containing U2OS cells were used to measure the homologous recombination (HR) and nonhomologous end joining (NHEJ)-mediated double strand break (DSB) repair efficiency, respectively. DR-GFP or EJ5-GFP cells were transfected with the siRNA indicated using DharmaFECT (Dharmacon, #T-2001–03) according to the manufacturer's instructions. After 6-hour siRNA transfection, the cells were subject to be transfected with pCMV-I-SceI plasmid together with RFP plasmid. For DDX21 rescue assay, DR-GFP U2OS cells were transfected with siRNA as described above, and then were transfected with pCMV-I-SceI plasmid together with RFP or siRNA-resistant RFP-DDX21 plasmid 6 hours after siRNA transfection. Forty-eight hours after all transfection, the cells are harvested and the percentage of GFP-positive cells was analyzed by flow cytometry. HR and NHEJ repair activity were assessed by quantification of the percentages of RFP+ and GFP+ cells. For the dynamic of HR efficiency analysis, DR-GFP U2OS cells were transfected with pCMV-I-SceI together with RFP or RFP-DDX21 plasmids. The cells were harvested and fixed with fixation buffer at 8, 12, 24, 36, 48, 60, and 72 hours after transfection (BioLegend, #420801) at 4°C. FACS analyses were performed after the cells at each time points were fully collected.

Xenograft experiment

The animal experiments were approved by the Animal Ethics Committee of Xiangya Hospital Central South University (Changsha, China; ethics number: 2019030536). 1.5 × 106 cells RKO control and DDX21 stable cells were collected, mixed with 50 μL Matrigel, and injected subcutaneously into the flank of the nude mice aging from 4–6 weeks. When tumor volumes reached approximately 50–60 mm3, mice hosted different cell line based on tumors were randomly divided into two groups and subsequently treated with indicated drugs (Fig. 6A; i.p.). Animal body weights and tumor sizes were measured and recorded every day. Tumor volumes were calculated using tumor volume = 1/2 (length × width2).

Data availability

The colon and lung cancer proteomics databases used in this study are available and obtained from data portal websites (https://cptac-data-portal.georgetown.edu/cptac/s/S045; https://cptac-data-portal.georgetown.edu/cptac/s/S022; https://cptac-data-portal.georgetown.edu/study-summary/S056), CPTAC data used in this publication were generated by the Clinical Proteomic Tumor Analysis Consortium (NCI/NIH); the liver cancer proteomics databases used in this study are available on the NODE (https://www.biosino.org/node), accession number (OEP000321). The mRNA databases used in this study are available on the TCGA (https://tcga-data.nci.nih.gov/tcga/). The CNA data matrices used in this study are available on the LinkedOmics (http://www.linkedomics.org). The data generated in this study are available within the article and its Supplementary Data files. Additional resources and data related to this article can be requested from the corresponding authors.

Comprehensive investigation of the expression of DExD/H box family members in multiple tumors

To investigate the expression of DExD/H box family members (DBF) in tumors, we performed multiplatform proteomic and genomic analyses of all DExD/H box RNA helicases in colon, lung, and liver tumors and their normal adjacent tissues (NAT; refs. 27–30). In colon cancer, TMT global proteomics revealed 47 DBFs in total, and 74.5% of DDX and DHX proteins (35 proteins, 35/47) were significantly increased in tumor tissues compared with NATs (Fig. 1A, D, and E; Supplementary Fig. S1A and S1B; Supplementary Tables S1A and S1B). Similar results were obtained in lung cancer from the CPTAC proteomics database and liver cancer database from Fan and colleagues (29, 30). Analysis of the proteomics database for lung cancer and liver cancer revealed 45 and 43 DBFs, respectively; 73.3% (33/45) of proteins were prominently enriched in lung carcinoma, and 86% (37/43) were enriched in liver cancer (Fig. 1B and C; Supplementary Fig. S1C; Supplementary Table S1A and S1B). Combined analysis of three proteomics databases revealed significant overlap in the upregulated proteins, and 28 DHX and DDX proteins were concomitantly elevated in all tumors (Fig. 1E). We hypothesized that the high expression of these tumor-associated proteins could be affected by transcription and/or translation processes. To further investigate the main drivers of the high expression of these proteins, we assessed their correlation of proteins with paired mRNAs and copy-number alterations (CNA) across all tumors. The results showed that most of the DExD/H box family members displayed comparative consistency at the protein and mRNA levels, represented by statistically significantly positive Pearson correlations between the protein and mRNA (Fig. 1F; Supplementary Fig. S1E), indicating the high expression of most DExD/H box family proteins that are highly correlated with mRNA may be less affected by translational regulations. Thus, we next examined the amplification of these genes (reflected by CNAs) at the genome level, which has a great impact on both mRNA and protein expression at the cis and trans levels (27, 29). Interestingly, we found highly positive correlations between the protein expression and CNAs levels in all 28 family members (Fig. 1G; Supplementary Table. S1C and S1D). In this regard, the increased level of DExD/H box proteins in tumors could be attributed to the consequences of amplification at the gene level to some extent.

Figure 1.

Comprehensive investigation of the expression of DExD/H box family members in multiple tumors. A–C, Heatmaps showing expression of DExD/H box family proteins analyzed in tumor and NATs in three independent proteomics databases [A, 2019 TMT global proteomics database in colorectal cancer (27); B, 2020 CPTAC proteomics database in lung cancer (29); C, 2019 proteomics database in liver cancer (30)]. Color of each cell indicates Z score (log2 of relative abundance scaled by row) of protein expression in each sample. Red, increase; blue, decrease. D, Volcano plot showing the ratio of expression of DExD/H box family proteins by comparing tumors to NATs in 2019 colorectal cancer proteomics database (27). Significantly differentially expressed proteins are highlighted in red or blue [P < 0.05, log2 (fold change (FC)) > 0.25, two-tailed Student t test]. E, Venn diagram depicting the overlap of upregulated DExD/H box family proteins in the three cancer types (P < 0.05 and log2 FC > 0). F, Pearson correlations assessing for 28 upregulated proteins between mRNA and protein abundance in the three cancer types. G, The correlations of CNAs to proteins abundance of DExD/H box in colon cancer and lung cancer database (27, 29). Blue and red represent positive and negative Spearman correlations, respectively.

Figure 1.

Comprehensive investigation of the expression of DExD/H box family members in multiple tumors. A–C, Heatmaps showing expression of DExD/H box family proteins analyzed in tumor and NATs in three independent proteomics databases [A, 2019 TMT global proteomics database in colorectal cancer (27); B, 2020 CPTAC proteomics database in lung cancer (29); C, 2019 proteomics database in liver cancer (30)]. Color of each cell indicates Z score (log2 of relative abundance scaled by row) of protein expression in each sample. Red, increase; blue, decrease. D, Volcano plot showing the ratio of expression of DExD/H box family proteins by comparing tumors to NATs in 2019 colorectal cancer proteomics database (27). Significantly differentially expressed proteins are highlighted in red or blue [P < 0.05, log2 (fold change (FC)) > 0.25, two-tailed Student t test]. E, Venn diagram depicting the overlap of upregulated DExD/H box family proteins in the three cancer types (P < 0.05 and log2 FC > 0). F, Pearson correlations assessing for 28 upregulated proteins between mRNA and protein abundance in the three cancer types. G, The correlations of CNAs to proteins abundance of DExD/H box in colon cancer and lung cancer database (27, 29). Blue and red represent positive and negative Spearman correlations, respectively.

Close modal

Some DExD/H box family members are involved in the DNA damage response

The genomic aberrations that are accumulated by inadequate physiologic or pathologic processes contribute to tumorigenesis. DExD/H box family members have been identified as a group of RNA-binding proteins with physiologic roles in RNA metabolism. However, the functions of DExD/H box family members in genome integrity in a pathologic context, particularly during DNA damage, remain elusive. Large-scale genomic and proteomic analyses have revealed that some RNA-processing factors are involved in the regulation of the DDR pathways (31–34). To evaluate whether DBFs play a direct role in the DNA damage response, we constructed 49 EGFP-tagged DExD/H-box family proteins (Supplementary Table S1E) and verified their involvement in the DNA damage response using 405-nm sensitized microirradiation (Fig. 2A; Supplementary Fig. S2A and S2B; Supplementary Table S1F; ref. 26). Intriguingly, we found that 20 DExD/H box proteins were recruited at DNA damage sites after microirradiation (Fig. 2A; Supplementary Fig. S2A and S2B; Supplementary Table S1F). A large number of these proteins have never been reported to be involved in DDR processes before. To obtain kinetics on the recruitment or removal of these proteins in response to DNA lesions, we generated pulsed microirradiation in a well-defined point region and sequentially monitored the fluorescence intensity of the proteins at damage sites. The half-times of the recruitments and disassociations were measured by calculating the intensity curve (Fig. 2B and C; Supplementary Fig. S2C). The half-time for the recruitments of DBFs in response to DNA damage varied from 3 seconds to half a minute. Moreover, the half-time for removal from the damage sites ranged from 2.5 minutes to more than 10 minutes, implicating that these DBFs are comprehensively involved in different DDR pathways or in different steps of DDR pathway. To further determine the pathways that these proteins may be involved in, we compared the recruitment and removal curves of DBFs to the well-recognized DNA damage proteins PARP1, RFC1, and BARD1, which display distinct repair dynamics and possess functions ranging from single-strand break repair to DSB repair (Fig. 2BD). We showed that DBFs could be grouped into three classes based on their association with the dynamics of PARP1, RFC1, and BARD1, suggesting that these DBFs are comprehensively involved in different DDR pathways or individual steps of DNA damage repair. We then verified the tumor expression of helicases that display DDR. Interestingly, 85% (17/20) of RNA helicases with DDR were highly expressed in multiple tumors, accounting for 60% (17/28) of all tumor-associated RNA helicases, further implying that the functions of RNA helicases in genome maintenance play a vital role in tumorigenesis (Fig. 2E).

Figure 2.

Some DExD/H box family members are involved in DNA damage response. A, Schematic diagram of DNA damage induced by laser microirradiation (left). Representative images of damage response of DDX1 and DHX37 in U2OS cells expressing EGFP-tagged DDX1 or DHX37 before and after laser microirradiation. Arrows, laser lines induced by microirradiation. B, Normalized kinetics of 16 DExD/H box family proteins (DDX5, DDX18, DDX21, DDX24, DDX27, DDX31, DDX49, DDX51, DDX54, DDX10, DDX23, DDX40, DDX47, DDX52, DDX16, and DHX37) and RFC1, PARP1, BARD1. The DNA damage responses of 4 members (DDX28, DHX35, DHX36, and DDX50) were too weak to measuring kinetics. C, The half-times of recruitment and removal of DNA damage response in 16 DExD/H box family proteins and RFC1, PARP1, BARD1. D, Scatter plot of half-times of recruitment and removal of DNA damage response in DExD/H box family proteins and RFC1, PARP1, and BARD1. E, Venn diagram depicting the overlap of all cloned 49 DExD/H box family proteins (DDX/DHX, gray; DDX39A failed to clone), 27 tumor-associated upregulated proteins (red), and 20 members involved in DDR (yellow).

Figure 2.

Some DExD/H box family members are involved in DNA damage response. A, Schematic diagram of DNA damage induced by laser microirradiation (left). Representative images of damage response of DDX1 and DHX37 in U2OS cells expressing EGFP-tagged DDX1 or DHX37 before and after laser microirradiation. Arrows, laser lines induced by microirradiation. B, Normalized kinetics of 16 DExD/H box family proteins (DDX5, DDX18, DDX21, DDX24, DDX27, DDX31, DDX49, DDX51, DDX54, DDX10, DDX23, DDX40, DDX47, DDX52, DDX16, and DHX37) and RFC1, PARP1, BARD1. The DNA damage responses of 4 members (DDX28, DHX35, DHX36, and DDX50) were too weak to measuring kinetics. C, The half-times of recruitment and removal of DNA damage response in 16 DExD/H box family proteins and RFC1, PARP1, BARD1. D, Scatter plot of half-times of recruitment and removal of DNA damage response in DExD/H box family proteins and RFC1, PARP1, and BARD1. E, Venn diagram depicting the overlap of all cloned 49 DExD/H box family proteins (DDX/DHX, gray; DDX39A failed to clone), 27 tumor-associated upregulated proteins (red), and 20 members involved in DDR (yellow).

Close modal

DDX21 is highly expressed in colon cancers and possesses an oncogenic activity

Among the DBFs investigated, we found that DDX21 was consistently high expressed in all three types of cancers (Fig. 1AC; Supplementary Fig. S1A). Notably, DDX21 displayed prominently elevated expression in tumor tissues compared with NATs at both the protein and mRNA levels in colorectal cancer (Figs. 1A and D, 3A and B; Supplementary Figs. S1A, S1B, and S3A). To further evaluate the expression of DDX21 in clinical samples, we collected fresh colon cancer tissues and paraffin-coated specimens from 50 patients with tumor tissues and paired NATs. Western blots showed that DDX21 was remarkably upregulated in tumor tissues compared with NATs (Fig. 3C). Consistent with these results, IHC staining showed that DDX21 expressions were significantly higher in tumor tissues (Fig. 3D). To further explore the relevance of DDX21 in tumor progression, we established two stable colorectal cancer cell lines with DDX21 knockdown (RKO and HCT116) using short hairpin RNAs (shRNA). As expected, RKO cells deprived of DDX21 exhibited decreased colony formation and cell proliferation abilities (Fig. 3E; Supplementary Fig. S3C), and similar results were obtained in HCT116 cells (Supplementary Fig. S3D). To gain more insight into the high expression of DDX21 in genome integrity, we performed assays to assess genomic variations. Chromosome spreading assays showed that sister chromatid exchanges (SCE) were significantly increased in cells with 2–3 folds overexpression of DDX21 compared with control cells (Fig. 3F; Supplementary Fig. S3B). In addition, the numbers of DNA fragments sheared by irinotecan treatments were greatly increased in DDX21-overexpressing RKO cells compared with control cells (Fig. 3G). We previously demonstrated that GFP-tagged DDX21 is recruited to the DNA damage sites. We further showed that endogenous DDX21 was accumulated at DSB sites after microirradiation and AsiSI-induced damaged chromosomes, suggesting it directly influences genome integrity under DNA damage (Fig. 3H and I; Supplementary Fig. S3E). We then employed two independent DDR assays, HR and NHEJ assays to detect the influence of DDX21 on the two DNA DSB repair pathways. We observed that cells with knockdown of DDX21 by siRNA exhibited reduced HR, but not NHEJ (Fig. 3J; Supplementary Fig. S3F). DDX21-deficient cells that were rescued by DDX21 recovered HR (Fig. 3K). However, excessive expression of DDX21 by overexpression greatly suppressed HR at the early phase and then gradually recovered in later phase (Fig. 3L), indicating DDX21 facilitates DSB repair in an appropriate expression, and on the contrast, aberrantly high expression triggers inappropriate DDR and genome instability.

Figure 3.

DDX21 is highly expressed in colon cancers and possesses an oncogenic activity. A, Protein abundance of DDX21 in tumors (n = 97) and NATs (n = 100) analyzed in 2019 colorectal database (ref. 27; ***, P < 0.001, Wilcoxon sighed-rank test). B, mRNA expression of DDX21 in tumors (n = 471) and NATs (n = 41) analyzed TCGA-COAD project (***, P < 0.001, Wilcoxon signed-rank test). C, Western blot showing the expression of DDX21 in colorectal tumors and paired NATs samples. D, Left, representative IHC staining of DDX21 in 50 tumors and 50 NATs. Right, statistical analysis of DDX21 expressions. E, Colony formation assay of inducible shCtrl and two shDDX21 stable RKO cell lines with or without doxycycline. Inset, DDX21 Western blot. F, SCEs frequency in RKO-Ctrl and RKO DDX21–overexpressed cells (right) and representative images of metaphase chromosomes (left). Arrows, chromosomes with exchanges. G, Top, representative neutral comet assay showing the tail moment of RKO cells expressing Flag or Flag-DDX21 treated with 10 μmol/L irinotecan for 24 hours. Bottom, quantification. H, Representative immunofluorescence images of endogenous DDX21 (red) and γH2AX (green) at laser microirradiation–induced DNA damage sites in U2OS and RKO cells. Scale bar, 5 μm. I, DIvA -ER-AsiSI U2OS cells were treated with 4-hydroxytamoxifen for 4 hours, and then chromatin immunoprecipitation assays were performed using IgG, anti-DDX21, or anti-γH2AX antibodies as indicated. Pull-down DNA was analyzed by qPCR using primers proximal to two AsiSI-induced DSBs (Chr1 and Chr6) or distal to AsiSI site located on Chr21. The means and SDs from three independent experiments are plotted. J, Effects of DDX21 knockdown on HR (left) and NHEJ (right) in U2OS cells containing DR-GFP and EJ5-GFP reporters. K, Effects of DDX21 knockdown complemented with the RFP-DDX21 on the efficiency of HR in U2OS cells containing DR-GFP. L, HR efficiencies were assessed in DR-GFP that contained U2OS cells transfected with pCMV-I-SceI and RFP or RFP-DDX21 at indicated times. D–L, Bars represent the average of n = 3 biologically independent experiments; error bars, SEM. NS, not significant; *, P < 0.05; **, P < 0.01; ***, P < 0.001, two-sided, unpaired t test.

Figure 3.

DDX21 is highly expressed in colon cancers and possesses an oncogenic activity. A, Protein abundance of DDX21 in tumors (n = 97) and NATs (n = 100) analyzed in 2019 colorectal database (ref. 27; ***, P < 0.001, Wilcoxon sighed-rank test). B, mRNA expression of DDX21 in tumors (n = 471) and NATs (n = 41) analyzed TCGA-COAD project (***, P < 0.001, Wilcoxon signed-rank test). C, Western blot showing the expression of DDX21 in colorectal tumors and paired NATs samples. D, Left, representative IHC staining of DDX21 in 50 tumors and 50 NATs. Right, statistical analysis of DDX21 expressions. E, Colony formation assay of inducible shCtrl and two shDDX21 stable RKO cell lines with or without doxycycline. Inset, DDX21 Western blot. F, SCEs frequency in RKO-Ctrl and RKO DDX21–overexpressed cells (right) and representative images of metaphase chromosomes (left). Arrows, chromosomes with exchanges. G, Top, representative neutral comet assay showing the tail moment of RKO cells expressing Flag or Flag-DDX21 treated with 10 μmol/L irinotecan for 24 hours. Bottom, quantification. H, Representative immunofluorescence images of endogenous DDX21 (red) and γH2AX (green) at laser microirradiation–induced DNA damage sites in U2OS and RKO cells. Scale bar, 5 μm. I, DIvA -ER-AsiSI U2OS cells were treated with 4-hydroxytamoxifen for 4 hours, and then chromatin immunoprecipitation assays were performed using IgG, anti-DDX21, or anti-γH2AX antibodies as indicated. Pull-down DNA was analyzed by qPCR using primers proximal to two AsiSI-induced DSBs (Chr1 and Chr6) or distal to AsiSI site located on Chr21. The means and SDs from three independent experiments are plotted. J, Effects of DDX21 knockdown on HR (left) and NHEJ (right) in U2OS cells containing DR-GFP and EJ5-GFP reporters. K, Effects of DDX21 knockdown complemented with the RFP-DDX21 on the efficiency of HR in U2OS cells containing DR-GFP. L, HR efficiencies were assessed in DR-GFP that contained U2OS cells transfected with pCMV-I-SceI and RFP or RFP-DDX21 at indicated times. D–L, Bars represent the average of n = 3 biologically independent experiments; error bars, SEM. NS, not significant; *, P < 0.05; **, P < 0.01; ***, P < 0.001, two-sided, unpaired t test.

Close modal

Overexpression of DDX21 delays the completion of HR repair

γH2AX is an early DNA damage marker that initiates and activates subsequent DDR cascades. We investigated the intensity of DNA damage and repair kinetics by measuring the activation and clearance of γH2AX. Cells with or without DDX21 overexpression were treated with DNA-damaging drugs and then stained with anti-γH2AX. We observed a higher percentage of DDX21-overexpressing cells with positive γH2AX foci immediately after drug withdrawal (Fig. 4A). Moreover, DDX21-overexpressing cells exhibited a substantially delayed disappearance of γH2AX foci, as indicated by sustained immunofluorescence signals, implying that the aberrantly high expression of DDX21 triggers DNA damage and delays the efficiency of DDR (Fig. 4B and C). Because the increased expression of DDX21 resulted in accumulated SCE, a phenotype that suggests excessive HR (Fig. 3F), we then evaluated the recruitment and disassociation of two major HR factors, RPA and Rad51, in response to chemotherapy drugs. Consistent with the γH2AX foci, DDX21-overexpressing cells had a high percentage of RPA pS33 foci and Rad51 foci (Fig. 4D and E). In addition, by measuring the dynamics of RPA pS33 and Rad51 foci, the results showed that both RPA and Rad51 were trapped at DNA damage sites and exhibited slower disassociations in DDX21-overexpressing cells with either irinotecan or etoposide treatments, indicating inappropriate HR repair in cells with high DDX21 expression and showing consistency with the γH2AX results (Fig. 4FI). We get the similar results when the cells were exposed to short time of high concentrations of irinotecan (Supplementary Fig. S4A–S4E). Taken together, these results suggest that the aberrantly high expression of DDX21 greatly impedes DSB repair by impairing HR.

Figure 4.

Overexpression of DDX21 delays the completion of HR repair. A, Control and overexpressed DDX21 RKO cells were treated with irinotecan (5 μmol/L) or etoposide (5 μmol/L) for 24 hours. Representative images of γH2AX foci are shown immediately after drug withdrawal. Right, the percentage of cells with positive γH2AX foci (>5) in at least 100 cells was calculated. B and C, Graphs represent the fold change of cells with γH2AX foci (>5) in RKO control and overexpressed DDX21 cells treated with 5 μmol/L irinotecan (B) or 5 μmol/L etoposide (C) for 24 hours and then released into fresh medium as indicated time points. The fold changes were calculated by normalized the percentages of cells at all the time point to 0 hour. D, Control and overexpressed DDX21 RKO cells treated with irinotecan (10 μmol/L) or etoposide (10 μmol/L) for 24 hours. Representative images of RPA pS33 foci are shown immediately after drug withdrawal. Right, the percentage of cells with positive RPA pS33 foci (>5) in at least 100 cells. E, Control and overexpressed DDX21 RKO cells were treated with irinotecan (20 μmol/L) or etoposide (10 μmol/L) for 24 hours. Representative images of Rad51 foci are shown immediately after drug withdrawal. Right, the percentage of cells with positive Rad51 foci (>5) in at least 100 cells. F and G, Graphs represent the fold changes of cells with RPA pS33 foci in RKO control and overexpressed DDX21 cells treated with 10 μmol/L irinotecan (F) and 10 μmol/L etoposide (G) for 24 hours and then released into fresh medium as indicated time points. The fold changes were calculated by normalizing the percentages of cells at all the time points to 0 hour. H and I, Graphs represent the fold changes of cells with RAD51 foci (>5) in RKO control and overexpressed DDX21 cells treated with 20 μmol/L irinotecan (H) and 10 μmol/L etoposide (I) for 24 hours and then released into fresh medium at indicated time points. The fold changes were calculated by normalizing the percentages of cells at all the time points to 0 hour. A–I, Bars represent the average of n = 3 biologically independent experiments; error bars, SEM. NS, not significant; *, P < 0.05; **, P < 0.01; ***, P < 0.001, two-sided, unpaired t test. For A, D, and E, scale bar, 5 μm.

Figure 4.

Overexpression of DDX21 delays the completion of HR repair. A, Control and overexpressed DDX21 RKO cells were treated with irinotecan (5 μmol/L) or etoposide (5 μmol/L) for 24 hours. Representative images of γH2AX foci are shown immediately after drug withdrawal. Right, the percentage of cells with positive γH2AX foci (>5) in at least 100 cells was calculated. B and C, Graphs represent the fold change of cells with γH2AX foci (>5) in RKO control and overexpressed DDX21 cells treated with 5 μmol/L irinotecan (B) or 5 μmol/L etoposide (C) for 24 hours and then released into fresh medium as indicated time points. The fold changes were calculated by normalized the percentages of cells at all the time point to 0 hour. D, Control and overexpressed DDX21 RKO cells treated with irinotecan (10 μmol/L) or etoposide (10 μmol/L) for 24 hours. Representative images of RPA pS33 foci are shown immediately after drug withdrawal. Right, the percentage of cells with positive RPA pS33 foci (>5) in at least 100 cells. E, Control and overexpressed DDX21 RKO cells were treated with irinotecan (20 μmol/L) or etoposide (10 μmol/L) for 24 hours. Representative images of Rad51 foci are shown immediately after drug withdrawal. Right, the percentage of cells with positive Rad51 foci (>5) in at least 100 cells. F and G, Graphs represent the fold changes of cells with RPA pS33 foci in RKO control and overexpressed DDX21 cells treated with 10 μmol/L irinotecan (F) and 10 μmol/L etoposide (G) for 24 hours and then released into fresh medium as indicated time points. The fold changes were calculated by normalizing the percentages of cells at all the time points to 0 hour. H and I, Graphs represent the fold changes of cells with RAD51 foci (>5) in RKO control and overexpressed DDX21 cells treated with 20 μmol/L irinotecan (H) and 10 μmol/L etoposide (I) for 24 hours and then released into fresh medium at indicated time points. The fold changes were calculated by normalizing the percentages of cells at all the time points to 0 hour. A–I, Bars represent the average of n = 3 biologically independent experiments; error bars, SEM. NS, not significant; *, P < 0.05; **, P < 0.01; ***, P < 0.001, two-sided, unpaired t test. For A, D, and E, scale bar, 5 μm.

Close modal

High expression of DDX21 induces a differential response to DNA-damaging drugs and persistent replication stress

Genome instability induced by aberrant DDR may create various vulnerabilities to DNA-damaging drugs. To test whether high DDX21 expression of induces selective sensitivity to DNA-damaging drugs, we examined the response of DDX21-overexpressing cells to different genotoxic drugs, some of which have been approved and applied for first-line clinical chemotherapy in colon cancer. As expected, cells with high DDX21 expression exhibited hypersensitivity to irinotecan, oxaliplatin, etoposide and hydroxyurea (HU; Fig. 5AD; Supplementary Fig. S4E), indicating that the overexpression of DDX21 disrupts the homeostasis of DDR, which produces higher sensitivity with specific drugs. However, DDX21-overexpressing cells were not highly sensitive to 5-FU, methyl methanesulfonate, or cisplatin (Supplementary Fig. S5A–S5C). In this regard, the diverse responses to genotoxic drugs also implicate that DDX21 may be alternatively engaged in specific DNA damage pathways. Irinotecan and etoposide, which are topoisomerase inhibitors, preferentially introduce DNA breaks during replication (35). The application of HU also attenuates the replication speed and induces intensive replication stress by inhibiting ribonucleoside diphosphate reductase (36, 37). We then evaluated replication stress under treatment with irinotecan, etoposide and HU in control and DDX21-overexpressing cells. The results showed that the phosphorylation of Chk1 and RPA2 was more strongly activated after treatment with HU, irinotecan and etoposide in DDX21-overexpressing cells than in control cells, clearly indicating that replication stress is greatly accumulated under DDX21 overexpression (Fig. 5EG). Moreover, overexpression of DDX21 did not obviously alter the cell cycle distributions (Supplementary Fig. S5D–S5F). These results further emphasize that the high expression of DDX21 causes genome instability.

Figure 5.

High expression of DDX21 induces a differential response to DNA-damaging drugs and persistent replication stress. Viability assays of control or DDX21-overexpressing SW480 and RKO cells treated with indicated drugs [irinotecan (A), oxaliplatin (B), etoposide (C), HU (D)] for 48 hours. *, P < 0.05; **, P < 0.01; ***, P < 0.001, two-sided, unpaired t test. E–G, Western blotting showing phosphorylation level of RPA2 and Chk1 and proteins as indicated in control or DDX21-overexpressing cells treated with HU (E), irinotecan (F), and etoposide (G).

Figure 5.

High expression of DDX21 induces a differential response to DNA-damaging drugs and persistent replication stress. Viability assays of control or DDX21-overexpressing SW480 and RKO cells treated with indicated drugs [irinotecan (A), oxaliplatin (B), etoposide (C), HU (D)] for 48 hours. *, P < 0.05; **, P < 0.01; ***, P < 0.001, two-sided, unpaired t test. E–G, Western blotting showing phosphorylation level of RPA2 and Chk1 and proteins as indicated in control or DDX21-overexpressing cells treated with HU (E), irinotecan (F), and etoposide (G).

Close modal

DDX21 can be used as a potential marker for drug response

To examine the synergistic role of DDX21 with DNA damaging drugs in killing cancer cells in vivo, we established a xenograft mouse model using stable DDX21-overexpressing RKO cells or control cells (Fig. 6A). The mice bearing tumors were randomized into four groups according to different treatment strategies: control groups with or without irinotecan treatments and DDX21-overexpressing groups with or without irinotecan treatments (Fig. 6A). Without the drug treatments, we observed comparatively similar tumor growth between the control and DDX21-overexpressing groups, indicating a redundant role of DDX21 in tumor growth. In the groups treated with irinotecan, the tumor sizes were significantly controlled, without further progression in the DDX21-overexpressing groups compared to the control groups, recapitulating that DDX21 sensitizes the drug response in these tumors (Fig. 6B and C; Supplementary Fig. S6A). Furthermore, we detected the proliferation marker Ki67, DNA damage marker γH2AX and apoptosis marker cleaved Caspase 3 in mouse tumor samples and found that Ki67 was decreased, and comparatively, γH2AX was increased in DDX21-overexpressing tumors after irinotecan treatment, further supporting that the DDX21-induced higher lethality to drugs is due to unrepaired DNA damage in vitro and in vivo (Fig. 6D and E). Even though the mice in the treatment groups exhibited a slight decrease in body weight, this can be accounted by the side effects of irinotecan, similar to most other chemotherapeutic drugs (Supplementary Fig. S6B).

Figure 6.

DDX21 can be used as a potential marker for drug response. A, Schematic diagram of nude mice implant model. Time schedule and concentration of treated irinotecan as indicated. B, Growth curves of xenograft tumors derived from control and overexpressed DDX21 RKO treated with DMSO or irinotecan. The mice with tumors were randomly divided into four groups treated with DMSO or irinotecan (10 mg/5 mg per kg, i.p.) after the tumor volumes reached to 50–60 mm3. Tumor volumes were measured at indicated time points. n = 7, error bars, SEM; *, P < 0.05; **, P < 0.01; ***, P < 0.001, two-sided, unpaired t test. C, Tumors isolated from individual mice. D, Representative images of IHC staining with Flag, cell proliferation marker Ki67, DNA damage maker γH2AX, and apoptosis marker cleaved caspase-3 in tumor samples derived from xenografts in nude mice. Scale bar, 50 μm. E, Quantitation of the histologic score (H score) of indicated markers from seven individual mice in different treatment groups as indicated.

Figure 6.

DDX21 can be used as a potential marker for drug response. A, Schematic diagram of nude mice implant model. Time schedule and concentration of treated irinotecan as indicated. B, Growth curves of xenograft tumors derived from control and overexpressed DDX21 RKO treated with DMSO or irinotecan. The mice with tumors were randomly divided into four groups treated with DMSO or irinotecan (10 mg/5 mg per kg, i.p.) after the tumor volumes reached to 50–60 mm3. Tumor volumes were measured at indicated time points. n = 7, error bars, SEM; *, P < 0.05; **, P < 0.01; ***, P < 0.001, two-sided, unpaired t test. C, Tumors isolated from individual mice. D, Representative images of IHC staining with Flag, cell proliferation marker Ki67, DNA damage maker γH2AX, and apoptosis marker cleaved caspase-3 in tumor samples derived from xenografts in nude mice. Scale bar, 50 μm. E, Quantitation of the histologic score (H score) of indicated markers from seven individual mice in different treatment groups as indicated.

Close modal

With a better understanding of DDR processes, the factors involved in DDR are no longer limited to DNA-associated proteins. Several RNA binding factors have been found to play a role in DDR. In our study, we carried out large-scale screening and revealed that approximately 40% of DExD/H box RNA helicases are directly involved in DDR. More importantly, substantial numbers of them are preferentially upregulated in tumor tissues. The oncogenic roles of RNA helicases in regulating DDR are further exemplified by DDX21, which showed predominant expression in colorectal cancers. The high expression of DDX21 leads to genome fragility by impairing double-strand repair and delaying HR repair (Fig. 7). Thus, we provide new insight that the high expression of RNA helicases provokes genome instability via dual dysfunctions in both transcription and DDR. Most importantly, the genome instability induced by RNA helicases provides alternative sensitivity to chemotherapy drugs, shedding light on precision treatments for RNA helicase-induced cancers.

Figure 7.

Illustration showing large-scale screening and high expression of DDX21 induces tumorigenesis and leads to differential chemotherapy sensitivity by increasing genome instability.

Figure 7.

Illustration showing large-scale screening and high expression of DDX21 induces tumorigenesis and leads to differential chemotherapy sensitivity by increasing genome instability.

Close modal

Extensive efforts have been made to elucidate the alternative roles of RNA binding proteins that go far beyond their functions in transcriptional regulation (31, 32, 38). Mass spectrometry-based proteomics and affinity purification analyses have revealed that a number of RNA processing factors interact with DDR proteins or are post translationally modified by DDR enzymes, accelerating the study of RNA processing factors in DDR (38, 39). In our study, we investigated the involvement of RNA helicases in the DNA damage response by employing laser microirradiation. We found over 40% of RNA helicases that directly participate in DNA damage processes. A recent mathematical analysis of several DDR factors in live images advocated that protein have similar damage-induced recruitment and removal kinetics coordinate in the same cascade (40). RNA processing is suggested to be involved in transcription-coupled repair, which displays comparatively slow dynamics due to its spatiotemporal dependency on transcription (24, 41). By comparing the dynamics of helicases in the DNA damage response to those of several well-recognized DNA damage proteins that have substantially broad half-times of recruitment to and disassociation from damage sites, we observed that these proteins participate in distinct pathways or in the different steps of DDR pathway, further extending the possibility that RNA binding factors act more comprehensive roles in DDR pathways beyond transcriptional-coupled repair. However, the regulatory mechanisms of these RNA helicases in various DDR pathways require further investigation.

DExD/H box helicase family members, which contain the evolutionarily conserved core DEAD-box and DEAH box helicase domains, belong to a large RNA binding cluster of RNA processing factors (42). Previous studies have shown that the abnormal expression of RNA helicases changes RNA structures and transcription profiles, which subsequently induces positive or negative alterations in some oncogenes, leading to complex and contradictory consequences in cancers (42). Our study revealed that the expression levels of RNA helicases that function in DDR were preferentially elevated in tumors, potentially indicating the contribution of aberrant DDR and genome instability to tumor progression is independent of transcriptional regulation. Thus, we hypothesize that the positive and negative roles of RNA helicases in tumor progression are an integrative consequence of transcription activity, DNA damage intensity and the corresponding repair burden in various tissues. Furthermore, the dual regulation of transcription and DDR further reveals the paradoxical roles of RNA helicases in tumorigenesis.

Compared to other RNA helicases, DDX21 showed pronounced expression in colorectal cancers. DDX21 is a nuclear protein that is essential for pre-RNA processing and ribosome biosynthesis (43, 44). DDX21 associates with Pol I in nucleoli and Pol II in the nucleoplasm to sequentially regulate rDNA transcription and ribosome assembly (44). The locations of DDX21 are reported to be mediated by polymerase activity. Polymerase I and II inhibition induces the nucleolar exclusion of DDX21. However, energy deprivation either by starvation or metabolic inhibition, which indirectly attenuates transcription, elicits no effects on the nucleolar location of DDX21 (44). Thus, insufficient evidence supports the idea that transcription regulates the different functions of DDX21 in the nucleoli and nucleoplasm. Recently, DDX21 was revealed to translocate to the nucleoplasm in response to rDNA damage (45). In our microirradiation screening, we discovered that DDX21 is directly involved in the genomic DNA damage response. Thus, DDX21 possesses a function that is distinct from its function in transcription under stresses in the nucleoplasm. Both the inhibition of transcription and rDNA damage are major resources that provoke replication stress. However, replication is compromised in cells with inadequate energy, leading to a senescent state without overt replication stress. Analysis of the landscape of repair dynamics revealed that DDX21 was automatically grouped into a repair pathway with RFC1, indicating the involvement of DDX21 in replication-stress-induced repair processes. DDX21 is involved in DDR, and aberrantly high expression of DDX21 induces inappropriate HR in early phase, retains Rad51 and RPA associations, which induce high frequency of chromosome exchanges, and subsequently increases damage burden with long-term culture. DDX21 was also revealed to unwind R-loop by association with SIRT7 (23). The loss of DDX21 leads to R-loops accumulation and increases DSB. Thus, either deficient or over abundant expression of DDX21 interferences with genome stability through processes of transcription and DDR, further emphasizing that precise regulation of DDX21 is necessarily required for appropriate physiological processes. Moreover, increasing evidences indicate long noncoding RNA (lncRNA) or snoRNA serves as an essential component to activate DDX21 in rDNA synthesis and ribosome biogenesis (44, 46, 47). Noncoding RNAs can be generated in response to DNA damage and are increasingly realized to directly contribute to DDR (17–19, 48). The damage-induced long noncoding RNAs (dilncRNAs) yield a possibility that the involvement of DDX21 in DDR may require lncRNAs. Thus, the fine regulation of DDX21, which precisely governs transcription and DDR, remains to be further explored.

As the high expression of RNA helicases plays a critical role in tumorigenesis, targeting RNA helicases should be a therapeutic strategy for cancers. Unfortunately, studies on both shRNAs and CRISPR have revealed that most RNA helicases are essential for cell survival, which raises a challenge for directly targeting RNA helicases as therapeutic targets (9, 49). To address this issue, we screened a variety of DNA-damaging drugs that increase DNA damage burden via genome fragility. As expected, we observed that cells overexpressing DDX21 manifested sensitivity to distinct DNA-damaging drugs that induce DSBs. Thus, the genome instability induced by the excessive expression of DDX21 can be viewed as a two-edged sword. On the one hand, this genome instability leads to mutations that result in tumorigenesis; on the other hand, it causes more lethal to some chemotherapy drugs via genome fragility to achieve the desired therapeutic effects. This finding sheds light on the prospect that DDX21 could be used as a potential biomarker to guide colorectal cancer treatments. Thus, our study revealed the involvement of RNA helicases in DNA damage and their associations with cancer. These new mechanisms extensively expand therapeutic strategies and improve precision treatments for patients with cancer with high expression of RNA helicases.

No disclosures were reported.

J. Xie: Conceptualization, resources, data curation, formal analysis, supervision, funding acquisition, validation, investigation, visualization, methodology, writing–original draft, project administration, writing–review and editing. M. Wen: Resources, data curation, software, formal analysis, funding acquisition, validation, investigation, visualization, methodology, writing–original draft, writing–review and editing. J. Zhang: Resources, data curation, software, formal analysis, funding acquisition, validation, investigation, methodology, writing–original draft, writing–review and editing. Z. Wang: Data curation, software, formal analysis. M. Wang: Data curation, software, formal analysis, visualization. Y. Qiu: Data curation, validation, visualization. W. Zhao: Resources, data curation, validation, visualization. F. Zhu: Resources, visualization. M. Yao: Resources, visualization. Z. Rong: Resources, formal analysis, visualization. W. Hu: Resources, data curation, formal analysis, visualization. Q. Pei: Resources, data curation. X. Sun: Resources, data curation. J. Li: Resources, data curation, software. Z. Mao: Data curation. L. Sun: Conceptualization, resources, supervision, funding acquisition, writing–review and editing. R. Tan: Conceptualization, resources, formal analysis, supervision, funding acquisition, writing–original draft, project administration, writing–review and editing.

The authors are grateful to Dr. Legube and members of Gaëlle Legube Laboratory for providing the DlvA-ER-AsiSI U2OS cell, and thank Zhiyong Mao for providing the HR and NHEJ reporter plasmids. They acknowledge usage of data from The Cancer Genome Atlas (TCGA), LinkedOmics, and The Clinical Proteomic Tumor Analysis Consortium (CPTAC). TCGA Data, Resources, and Materials were originally published by the NCI. CPTAC data used in this publication were generated by the Clinical Proteomic Tumor Analysis Consortium (NCI/NIH). This work was supported by the National Natural Science Foundation of China (81974451, 81902888, 82030087), the Natural Science Foundation of Hunan Province for Outstanding Young Scholars (2020JJ3062), Innovation Foundation of Central South University for Outstanding Young Scholars (2019CX038), Award of Hunan Province for 2021 Hu Xiang Talent (2020RC3061) Youth Research Foundation of Xiangya Hospital, Central South University (2018Q07) and Open Project Program of the State Key Laboratory of Proteomics (SKLP-O202002).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1.
Negrini
S
,
Gorgoulis
VG
,
Halazonetis
TD
.
Genomic instability–an evolving hallmark of cancer
.
Nat Rev Mol Cell Biol
2010
;
11
:
220
8
.
2.
Kim
TM
,
Laird
PW
,
Park
PJ
.
The landscape of microsatellite instability in colorectal and endometrial cancer genomes
.
Cell
2013
;
155
:
858
68
.
3.
Cancer Genome Atlas Research Network
.
Comprehensive molecular characterization of gastric adenocarcinoma
.
Nature
2014
;
513
:
202
9
.
4.
Zhou
F
,
Liu
Y
,
Rohde
C
,
Pauli
C
,
Gerloff
D
,
Kohn
M
, et al
.
AML1-ETO requires enhanced C/D box snoRNA/RNP formation to induce self-renewal and leukaemia
.
Nat Cell Biol
2017
;
19
:
844
55
.
5.
Gubin
MM
,
Zhang
X
,
Schuster
H
,
Caron
E
,
Ward
JP
,
Noguchi
T
, et al
.
Checkpoint blockade cancer immunotherapy targets tumour-specific mutant antigens
.
Nature
2014
;
515
:
577
81
.
6.
Schumacher
TN
,
Schreiber
RD
.
Neoantigens in cancer immunotherapy
.
Science
2015
;
348
:
69
74
.
7.
Ward
JP
,
Gubin
MM
,
Schreiber
RD
.
The role of neoantigens in naturally occurring and therapeutically induced immune responses to cancer
.
Adv Immunol
2016
;
130
:
25
74
.
8.
Chan
EM
,
Shibue
T
,
McFarland
JM
,
Gaeta
B
,
Ghandi
M
,
Dumont
N
, et al
.
WRN helicase is a synthetic lethal target in microsatellite unstable cancers
.
Nature
2019
;
568
:
551
6
.
9.
Behan
FM
,
Iorio
F
,
Picco
G
,
Goncalves
E
,
Beaver
CM
,
Migliardi
G
, et al
.
Prioritization of cancer therapeutic targets using CRISPR-Cas9 screens
.
Nature
2019
;
568
:
511
6
.
10.
Sun
C
,
Yin
J
,
Fang
Y
,
Chen
J
,
Jeong
KJ
,
Chen
X
, et al
.
BRD4 inhibition is synthetic lethal with PARP inhibitors through the induction of homologous recombination deficiency
.
Cancer Cell
2018
;
33
:
401
16
.
11.
Parvin
S
,
Ramirez-Labrada
A
,
Aumann
S
,
Lu
X
,
Weich
N
,
Santiago
G
, et al
.
LMO2 confers synthetic lethality to PARP inhibition in DLBCL
.
Cancer Cell
2019
;
36
:
237
49
.
12.
Bhatia
V
,
Barroso
SI
,
Garcia-Rubio
ML
,
Tumini
E
,
Herrera-Moyano
E
,
Aguilera
A
.
BRCA2 prevents R-loop accumulation and associates with TREX-2 mRNA export factor PCID2
.
Nature
2014
;
511
:
362
5
.
13.
Wickramasinghe
VO
,
Venkitaraman
AR
.
RNA processing and genome stability: Cause and consequence
.
Mol Cell
2016
;
61
:
496
505
.
14.
Ohle
C
,
Tesorero
R
,
Schermann
G
,
Dobrev
N
,
Sinning
I
,
Fischer
T
.
Transient RNA-DNA hybrids are required for efficient double-strand break repair
.
Cell
2016
;
167
:
1001
13
.
15.
Garcia-Muse
T
,
Aguilera
A
.
R loops: From physiological to pathological roles
.
Cell
2019
;
179
:
604
18
.
16.
Aguilera
A
,
Garcia-Muse
T
.
R loops: from transcription byproducts to threats to genome stability
.
Mol Cell
2012
;
46
:
115
24
.
17.
Michelini
F
,
Pitchiaya
S
,
Vitelli
V
,
Sharma
S
,
Gioia
U
,
Pessina
F
, et al
.
Damage-induced lncRNAs control the DNA damage response through interaction with DDRNAs at individual double-strand breaks
.
Nat Cell Biol
2017
;
19
:
1400
11
.
18.
Francia
S
,
Michelini
F
,
Saxena
A
,
Tang
D
,
de Hoon
M
,
Anelli
V
, et al
.
Site-specific DICER and DROSHA RNA products control the DNA-damage response
.
Nature
2012
;
488
:
231
5
.
19.
Thapar
R
,
Wang
JL
,
Hammel
M
,
Ye
R
,
Liang
K
,
Sun
C
, et al
.
Mechanism of efficient double-strand break repair by a long non-coding RNA
.
Nucleic Acids Res
2021
;
49
:
1199
200
.
20.
Zhong
P
,
Peng
J
,
Bian
Z
,
Huang
H
.
The role of cold inducible RNA-binding protein in cardiac physiology and diseases
.
Front Pharmacol
2021
;
12
:
610792
.
21.
Fairman-Williams
ME
,
Guenther
UP
,
Jankowsky
E
.
SF1 and SF2 helicases: family matters
.
Curr Opin Struct Biol
2010
;
20
:
313
24
.
22.
De Bortoli
F
,
Espinosa
S
,
Zhao
R
.
DEAH-box RNA helicases in pre-mRNA splicing
.
Trends Biochem Sci
2021
;
46
:
225
38
.
23.
Song
C
,
Hotz-Wagenblatt
A
,
Voit
R
,
Grummt
I
.
SIRT7 and the DEAD-box helicase DDX21 cooperate to resolve genomic R loops and safeguard genome stability
.
Genes Dev
2017
;
31
:
1370
81
.
24.
Chakraborty
P
,
Huang
JTJ
,
Hiom
K
.
DHX9 helicase promotes R-loop formation in cells with impaired RNA splicing
.
Nat Commun
2018
;
9
:
4346
.
25.
Kim
S
,
Kang
N
,
Park
SH
,
Wells
J
,
Hwang
T
,
Ryu
E
, et al
.
ATAD5 restricts R-loop formation through PCNA unloading and RNA helicase maintenance at the replication fork
.
Nucleic Acids Res
2020
;
48
:
7218
38
.
26.
Lan
L
,
Ui
A
,
Nakajima
S
,
Hatakeyama
K
,
Hoshi
M
,
Watanabe
R
, et al
.
The ACF1 complex is required for DNA double-strand break repair in human cells
.
Mol Cell
2010
;
40
:
976
87
.
27.
Vasaikar
S
,
Huang
C
,
Wang
X
,
Petyuk
VA
,
Savage
SR
,
Wen
B
, et al
.
Proteogenomic analysis of human colon cancer reveals new therapeutic opportunities
.
Cell
2019
;
177
:
1035
49
.
28.
Zhang
B
,
Wang
J
,
Wang
X
,
Zhu
J
,
Liu
Q
,
Shi
Z
, et al
.
Proteogenomic characterization of human colon and rectal cancer
.
Nature
2014
;
513
:
382
7
.
29.
Gillette
MA
,
Satpathy
S
,
Cao
S
,
Dhanasekaran
SM
,
Vasaikar
SV
,
Krug
K
, et al
.
Proteogenomic characterization reveals therapeutic vulnerabilities in lung adenocarcinoma
.
Cell
2020
;
182
:
200
25
.
30.
Gao
Q
,
Zhu
H
,
Dong
L
,
Shi
W
,
Chen
R
,
Song
Z
, et al
.
Integrated proteogenomic characterization of HBV-related hepatocellular carcinoma
.
Cell
2019
;
179
:
561
77
.
31.
Adamson
B
,
Smogorzewska
A
,
Sigoillot
FD
,
King
RW
,
Elledge
SJ
.
A genome-wide homologous recombination screen identifies the RNA-binding protein RBMX as a component of the DNA-damage response
.
Nat Cell Biol
2012
;
14
:
318
28
.
32.
Beli
P
,
Lukashchuk
N
,
Wagner
SA
,
Weinert
BT
,
Olsen
JV
,
Baskcomb
L
, et al
.
Proteomic investigations reveal a role for RNA processing factor THRAP3 in the DNA damage response
.
Mol Cell
2012
;
46
:
212
25
.
33.
Blasius
M
,
Wagner
SA
,
Choudhary
C
,
Bartek
J
,
Jackson
SP
.
A quantitative 14–3-3 interaction screen connects the nuclear exosome targeting complex to the DNA damage response
.
Genes Dev
2014
;
28
:
1977
82
.
34.
Matsuoka
S
,
Ballif
BA
,
Smogorzewska
A
,
McDonald
ER
, 3rd,
Hurov
KE
,
Luo
J
, et al
.
ATM and ATR substrate analysis reveals extensive protein networks responsive to DNA damage
.
Science
2007
;
316
:
1160
6
.
35.
Zhao
H
,
Rybak
P
,
Dobrucki
J
,
Traganos
F
,
Darzynkiewicz
Z
.
Relationship of DNA damage signaling to DNA replication following treatment with DNA topoisomerase inhibitors camptothecin/topotecan, mitoxantrone, or etoposide
.
Cytometry A
2012
;
81
:
45
51
.
36.
Eklund
H
,
Uhlin
U
,
Farnegardh
M
,
Logan
DT
,
Nordlund
P
.
Structure and function of the radical enzyme ribonucleotide reductase
.
Prog Biophys Mol Biol
2001
;
77
:
177
268
.
37.
Biswas
D
,
Takahata
S
,
Xin
H
,
Dutta-Biswas
R
,
Yu
Y
,
Formosa
T
, et al
.
A role for Chd1 and Set2 in negatively regulating DNA replication in Saccharomyces cerevisiae
.
Genetics
2008
;
178
:
649
59
.
38.
Van Nostrand
EL
,
Freese
P
,
Pratt
GA
,
Wang
X
,
Wei
X
,
Xiao
R
, et al
.
A large-scale binding and functional map of human RNA-binding proteins
.
Nature
2020
;
583
:
711
9
.
39.
Beckmann
BM
,
Castello
A
,
Medenbach
J
.
The expanding universe of ribonucleoproteins: of novel RNA-binding proteins and unconventional interactions
.
Pflugers Arch
2016
;
468
:
1029
40
.
40.
Aleksandrov
R
,
Dotchev
A
,
Poser
I
,
Krastev
D
,
Georgiev
G
,
Panova
G
, et al
.
Protein dynamics in complex DNA lesions
.
Mol Cell
2018
;
69
:
1046
61
.
41.
Nakazawa
Y
,
Hara
Y
,
Oka
Y
,
Komine
O
,
van den Heuvel
D
,
Guo
C
, et al
.
Ubiquitination of DNA damage-stalled RNAPII promotes transcription-coupled repair
.
Cell
2020
;
180
:
1228
44
.
42.
Fuller-Pace
FV
.
DEAD box RNA helicase functions in cancer
.
RNA Biol
2013
;
10
:
121
32
.
43.
Valdez
BC
,
Perlaky
L
,
Henning
D
.
Expression, cellular localization, and enzymatic activities of RNA helicase II/Gu(beta)
.
Exp Cell Res
2002
;
276
:
249
63
.
44.
Calo
E
,
Flynn
RA
,
Martin
L
,
Spitale
RC
,
Chang
HY
,
Wysocka
J
.
RNA helicase DDX21 coordinates transcription and ribosomal RNA processing
.
Nature
2015
;
518
:
249
53
.
45.
Calo
E
,
Gu
B
,
Bowen
ME
,
Aryan
F
,
Zalc
A
,
Liang
J
, et al
.
Tissue-selective effects of nucleolar stress and rDNA damage in developmental disorders
.
Nature
2018
;
554
:
112
7
.
46.
Xing
YH
,
Yao
RW
,
Zhang
Y
,
Guo
CJ
,
Jiang
S
,
Xu
G
, et al
.
SLERT regulates DDX21 rings associated with Pol I transcription
.
Cell
2017
;
169
:
664
78
.
47.
Kim
DS
,
Camacho
CV
,
Nagari
A
,
Malladi
VS
,
Challa
S
,
Kraus
WL
.
Activation of PARP-1 by snoRNAs controls ribosome biogenesis and cell growth via the RNA helicase DDX21
.
Mol Cell
2019
;
75
:
1270
85
.
48.
Fijen
C
,
Rothenberg
E
.
The evolving complexity of DNA damage foci: RNA, condensates and chromatin in DNA double-strand break repair
.
DNA Repair
2021
;
105
:
103170
.
49.
Chang
L
,
Ruiz
P
,
Ito
T
,
Sellers
WR
.
Targeting pan-essential genes in cancer: Challenges and opportunities
.
Cancer Cell
2021
;
39
:
466
79
.

Supplementary data