DHX9 SUMOylation is required for the suppression of R-loop-associated genome instability

Abstract

RNA helicase DHX9 is essential for genome stability by resolving aberrant R-loops. However, its regulatory mechanisms remain unclear. Here we show that SUMOylation at lysine 120 (K120) is crucial for DHX9 function. Preventing SUMOylation at K120 leads to R-loop dysregulation, increased DNA damage, and cell death. Cells expressing DHX9 K120R mutant which cannot be SUMOylated are more sensitive to genotoxic agents and this sensitivity is mitigated by RNase H overexpression. Unlike the mutant, wild-type DHX9 interacts with R-loop-associated proteins such as PARP1 and DDX21 via SUMO-interacting motifs. Fusion of SUMO2 to the DHX9 K120R mutant enhances its association with these proteins, reduces R-loop accumulation, and alleviates survival defects of DHX9 K120R. Our findings highlight the critical role of DHX9 SUMOylation in maintaining genome stability by regulating protein interactions necessary for R-loop balance.

Fig. 1. K120 is a major SUMO conjugation site of DHX9.

a A primary structure of DHX9, with an SFB-tag at its N-terminus, shows the helicase core and the location of five putative SUMO2 conjugation sites. Colored boxes represent specific features: the DExH-box colored in blue, the NLS in yellow, and boxes in magenta (I and II) for dsRBD1 and dsRBD2, respectively. b Sequences of five putative SUMO2 acceptor sites mapped by system-wide studies. c HeLa cells transfected with SFB-DHX9 (8 μg) alone or with HA-SUMO1 (4 μg) or HA-SUMO2 (3 μg) were used for FLAG affinity purification under denaturing conditions. Protein levels and SUMO conjugation on SFB-DHX9 were determined by Western blot with the indicated antibodies. d A His-GST-tagged DHX9 fragment (1–399) purified from E. coli was subjected to reconstituted reactions to examine DHX9 SUMOylation in vitro. Levels of recombinant DHX9^1–399 and SUMO-modified DHX9^1–399 were determined by Western blot using anti-DHX9 antibody. e DHX9 SUMOylation is not regulated by DNA damage. HeLa cells transfected with SFB-DHX9 alone or with HA-SUMO2 were treated with DMSO, HU (4 mM), or CPT (2 μM) for 1 h. SUMOylation of DHX9 and protein levels were determined by Western blot. f, g DHX9 K120 is the major SUMO2 acceptor site. The HA-SUMO2 (f) and endogenous SUMO2/3 (g) conjugation of SFB-DHX9 variants purified by the anti-FLAG magnetic beads was analyzed by Western blot. h Bacterially expressed recombinant DHX9^1–399 (WT or K120R) were subjected to test DHX9 SUMOylation in vitro. DHX9 SUMOylation was assessed by Western blot. i–k SFB-DHX9^WT, but not SFB-DHX9^K120R associated SUMO2/3 in the nucleus. HeLa derivative clones transfected with siDHX9 were used for expressing the vector, wild-type (WT), or K120R DHX9 upon doxycycline induction, followed by PLA analysis using the indicated antibodies. i Representative images of PLA analysis. j Quantification of PLA foci-positive cells (red). The black line indicates the median. The number of PLA foci per nucleus was counted from 150 nuclei under each condition and analyzed by a two-sided Mann–Whitney U test. k Levels of SFB-DHX9 in different stable cell lines were determined by Western blot. Source data are provided as a Source Data file.

Fig. 2. DHX9 SUMOylation is crucial for cell survival.

a, b DHX9 SUMOylation supports cell proliferation in different cell types. As shown in the upper timeline, 1 × 10⁵ of HeLa (a) or U2OS (b) cells were reverse transfected with siControl or siDHX9. 1 d later, the cells were forward transfected with the empty vector, or SFB-DHX9 variants, with doxycycline simultaneously added to the medium. Cell counts were then recorded daily in triplicates. Data are presented as mean ± SEM from n ≥ 3 individual experiments analyzed by two-way ANOVA. c Protein levels of SFB-DHX9^WT and SFB-DHX9^K120R in cells used for cell counts in (a, b) were determined by Western blot. d HeLa derivative clones (Vect-2, WT-2, and K120R-13) transfected with siDHX9 were induced with doxycycline to express SFB-DHX9 variants. Four days after ablating endogenous DHX9, cell counts were quantified. Data are presented as mean ± SEM with dots indicating the results of three independent experiments analyzed by one-way ANOVA. e Same as the condition in (d), cell death of indicated HeLa derivative clones was assessed using Annexin V apoptosis detection kits. Data are presented as mean ± SEM with dots indicating the results of three independent experiments analyzed by one-way ANOVA. f Protein levels of SFB-DHX9 variants in different HeLa derivative clones transfected with control or DHX9 siRNAs were determined by Western blot. Source data are provided as a Source Data file.

Fig. 3. DHX9 SUMOylation prevents the accumulation of DNA damage.

a–e HeLa derivative clones transfected with negative control or DHX9 siRNAs for 1 d were induced to express SFB-DHX9^WT (WT-2) or SFB-DHX9^K120R (K120R-13) by doxycycline for 3 d, followed by treatment with DMSO or CPT (1 μM) for 1 h. The percentage and intensity of γH2AX (red) and FLAG (green) staining were determined. a Representative images of three independent experiments. Scale bar = 5 μm. b The percentage of γH2AX foci-positive cells was quantified. Data are presented as mean ± SD with dots indicating results from three independent experiments (ns = no significance) analyzed by one-way ANOVA. c Intensity of γH2AX foci was quantified in DMSO and CPT-treated HeLa clones. Representative data of three biological replicates analyzed by two-sided t-test. Cells with background staining were colored black. The black line indicates the median. d, e The correlation between SFB-DHX9 expression and γH2AX foci intensity in WT and mutant clone cells treated with DMSO were plotted in 2D, respectively. f Representative Western blot of three independent experiments shows the levels of endogenous and SFB-tagged DHX9 in different HeLa derivative lines. Source data are provided as a Source Data file.

Fig. 4. DHX9 SUMOylation suppresses R-loop accumulation and the impairment of transcription termination.

a HeLa cells expressing DHX9^K120R displayed an accumulation of R-loops. HeLa cells with DHX9 depletion by siRNA were transfected with the empty vector, SFB-DHX9^WT, or SFB-DHX9^K120R. DNA/RNA hybrids isolated from nuclear extracts were treated with or without RNH and analyzed using a slot blot assay with the S9.6 antibody. The relative S9.6 signal was quantified (mean ± SEM of three independent experiments). Top: representative blots; Bottom: quantification of relative S9.6 signal from n = 3 biological replicates analyzed by two-sided t test. b Workflow of nucleic acids extraction and the RNH digestion, followed by DRIP assay. c, d HeLa cells treated with control or DHX9 siRNA were transfected with SFB-DHX9^WT, SFB-DHX9^K120R, or the empty vector. DRIP using the S9.6 antibody was analyzed by RT-qPCR for the abundance of R-loops over the polyA-proximal regions of β-actin and γ-actin genes, respectively. Values are normalized to intron 1 (in1) (mean ± SEM of three independent experiments analyzed by two-sided t test). e, f Ablating K120 SUMOylation resulted in transcription termination deficiency. HeLa cells treated with control or DHX9 siRNA were transfected with SFB-DHX9^WT, SFB-DHX9^K120R, or the empty vector. Extracted nucleic acids were analyzed by RT-qPCR for readthrough transcription of β-actin and γ-actin genes, respectively. Data of three biological individual experiments. Values are normalized to β-actin intron 3 (in3) and γ-actin in1 and presented as mean ± SEM (two-sided t test). Source data are provided as a Source Data file.

Fig. 5. The vulnerability of DHX9^K120R cells to genotoxic stress can be rescued by RNH.

a-d DHX9^K120R cells exhibited increased susceptibility to genotoxic drugs. HeLa inducible clones were transfected with siDHX9 to deplete endogenous DHX9. The expression of SFB-DHX9^WT and SFB-DHX9^K120R was induced with doxycycline. The cells were treated with CPT (a), HU (b), cisplatin (c), and berzosertib (d) at the indicated concentrations for 24 h. Cell viability was assessed using Cell Titer-Glo 2.0 and is presented as mean ± SEM of three independent experiments (two-way ANOVA). e, f Overexpression of RNH mitigated R-loop-associated DNA damage. HeLa inducible clones with DHX9 depletion were transfected with or without RNH. The percentages of γH2AX-positive cells (e) and RNH-positive cells (f) were determined using Flow Cytometry. Data are presented as mean ± SEM of three independent experiments (ns = no significance) analyzed by two-sided t-test. g RNH overexpression rescued the cell survival of DHX9-depleted or DHX9^K120R cells. HeLa inducible clones transfected with siDHX9 for 1 d were introduced with or without RNH for an additional 3 d before cell counting was performed. Data are presented as mean ± SEM of three independent experiments analyzed by two-sided t-test. Source data are provided as a Source Data file.

Fig. 6. DHX9^K120R fails to efficiently associate with R-loops and R-loop interacting factors.

a The association of DHX9 variants with R-loops (S9.6 signal) in DHX9-depleted HeLa cells was assessed by PLA with indicated antibodies. The number of PLA foci (red) per nucleus was quantified by analyzing 150 nuclei under each condition (two-sided Mann–Whitney U test). The black line indicates the median. b SFB-DHX9^K120R exhibited poor binding to R-loops. DRIP was prepared as described in Fig. 4b, with the addition of universal nuclease as a negative control. The association of SFB-DHX9 variants with R-loops was analyzed by Western blot, three biological repeats. c, d Ablating K120 SUMOylation attenuated the interaction between DHX9 and multiple RNA-processing factors. HeLa cell lysates were used for SFB-tag pull-down using Dynabead M-280-streptavidin. Proteins interacting with SFB-DHX9 variants were determined by Western blot, two biological repeats. e The associations between indicated SFB-DHX9 variants and multi-RNA binding proteins were determined by Western blot, three biological repeats. f Similar to (a), the association of different SFB-DHX9 variants with R-loops in DHX9-depleted HeLa cells was assessed by PLA (red) with indicated antibodies (two-sided Mann–Whitney U test). g The SUMOylation of SFB-DHX9 variants purified by the anti-FLAG magnetic beads was analyzed by Western blot The association of SFB-DHX9 variants with R-loops was analyzed by Western blot, two biological repeats. h RNase A treatment showed distinct interaction patterns between DHX9 and RNA binding proteins. HeLa cell lysates treated without or with RNase A were used for SFB-tag pull-down. Proteins interacting with SFB-DHX9 variants were determined by Western blot, two biological repeats. i HeLa cells transfected with the empty vectors, SFB-DHX9, or SFB-DHX9 with WT or mutant RNH-mCherry plasmids were subjected to IP by anti-FLAG beads. The interaction between SFB-DXH9 and indicated proteins was determined by Western blot, two biological repeats. Source data are provided as a Source Data file.

Fig. 7. DHX9 SUMOylation enhances its association with PARP1 and DDX21 via SUMO-SIM interaction.

a Schematic diagrams show predicted SIMs in PARP1, and a previously identified SIM in DDX21, with their locations highlighted in red boxes. b HeLa cells co-transfected with indicated constructs were split into two groups. One for FLAG affinity purification and the other for HA affinity purification. The proteins precipitated were determined by Western blot, and the quantification data are presented as mean ± SD from three separate experiments analyzed by two-sided t test. c, d HeLa cells co-transfected with indicated constructs were used for IP. One group of samples were subjected to FLAG affinity purification (c) and the other group of samples were subjected to HA affinity purification (d). The association of SFB-DHX9 with HA-DDX21 variants was determined by Western blot, and the quantification data are presented as mean ± SD from three separate experiments analyzed by a two-sided t test. e The SUMO pathway modulates protein associations between SFB-DHX9 and multiple RNA-processing factors. HeLa cells transfected with siControl or siUBC9 were transfected with indicated constructs, followed by FLAG affinity purification. Associations between DHX9 and RNA-processing factors were determined by Western blot (n = 3). Source data are provided as a Source Data file.

Fig. 8. Fusion of a SUMO2 to DHX9^K120R bypasses the SUMOylation of DHX9.

a The interactions between different SFB-DHX9 variants and HA-PARP1 constructs were assessed by co-IP. HeLa cells co-transfected with indicated constructs were collected for IP using anti-HA beads. Levels of SFB-DHX9 variants and SUMO2-DHX9^K120R co-precipitated by HA-PARP1 variants were determined and quantified by Western blot. Representative data of two biological replicates. b The association between DHX9 variants and R-loops in DHX9-depleted HeLa cells was assessed by PLA with indicated antibodies. The number of PLA foci (red) per nucleus was quantified by analyzing 150 nuclei in each condition (two-sided Mann-Whitney U test). The black line indicates the median. c The relative R-loop level at the 5’ pause region of β-actin gene was determined by RT-qPCR. The representative bar graph was from three separate experiments. Values are normalized to β-actin in1 and presented as mean ± SEM. d S2-DHX9^K120R improved cell viability. HeLa cells with DHX9 depletion by siRNA were transfected with the indicated constructs for 66 h before collection. Cell counting for each condition in triplicates was obtained. Data are presented as mean ± SEM from three individual experiments analyzed by one-way ANOVA. e A model depicts the role of DHX9 SUMOylation in enhancing the association of DHX9 with R-loops and multiple RNA-processing factors, aiding in R-loop balance and genome stability. Yellow crescent shapes in PARP1 and DDX21 represent SIMs, and circled S symbols represent SUMO modifications. Source data are provided as a Source Data file.