Fig 2: TDRD3 interacts with DHX9. (A) Tandem affinity purification of the TOP3B protein complex (TOP3B.com) from HEK293 cells. The eluted protein complex was separated by SDS-PAGE and silver-stained (left panel). The number of unique peptides from the top four purified proteins is shown (right panel). (B) The interaction of DHX9 with TDRD3 and TOP3B was confirmed by co-immunoprecipitation (co-IP). MCF7 cells were immunoprecipitated with control IgG and αTOP3B antibodies. The eluted protein samples were detected by western blot analysis using αDHX9, αTDRD3 and αTOP3B antibodies. (C) The interaction of DHX9 with TDRD3 and TOP3B was detected by co-IP in wild type (WT) and TDRD3 knockout (KO) MCF7 cells using two different αTDRD3 antibodies (Ab1 and Ab2). TDRD3, TOP3B and DHX9 were detected in the input samples by western blot analysis. (D) TOP3B is dispensable for the TDRD3-DHX9 interaction. Both WT and TOP3B KO MCF7 cells were transfected with Flag-TDRD3. The interaction of TDRD3 with DHX9 and TOP3B was detected by co-IP. (E) TDRD3 is essential for the TOP3B-DHX9 interaction. Both WT and TDRD3 KO MCF7 cells were transfected with Flag-TOP3B. The interaction of TOP3B with DHX9 and TDRD3 was detected by co-IP. (F) TDRD3 directly interacts with DHX9. Flag-DHX9 recombinant proteins were purified from HEK293 cells. GST, GST-TDRD3, and GST-TOP3B recombinant proteins were purified from E. coli. The interactions of GST-tagged recombinant proteins with Flag-DHX9 were assessed by GST pull-down and western blot analysis using an αFlag antibody. The amount of proteins used in the binding was visualized by the Ponceau S staining of the PVDF membrane.

Fig 3: Characterization of the interaction between DHX9 and TDRD3. (A) Mapping the region of TDRD3 that interacts with DHX9. GST-tagged TDRD3 truncation constructs were generated, containing the oligonucleotide/oligosaccharide-binding (OB)-fold, the ubiquitin-associated domain (UBA; wild type or with L324A mutation), and the Tudor domain (Tudor; wild type or with E691K mutation), respectively. A graphic summary of their interactions with DHX9 is shown. A GST pull-down assay was performed by incubating the recombinant GST-fusion proteins with MCF7 cell lysates. The pull-down samples were detected by western blot analysis using an αDHX9 antibody. The GST-fusion proteins were visualized by Ponceau staining. (B) Mapping the region of DHX9 that interacts with TDRD3. GFP-tagged full-length or truncations of DHX9 constructs were generated. The locations of the N-terminal double-stranded RNA binding domain (dsRBD), the helicase domain (HD), and the C-terminal RGG-containing domain (RGG) are indicated (upper panel). A graphic summary of their interactions with TDRD3 is shown. A GST pull-down assay was performed by incubating the recombinant GST-Tudor with MCF7 lysates that were transfected with different GFP-DHX9 fusion vectors. Both the input and pull-down samples were detected by western blot analysis using an αGFP antibody. The GST-Tudor recombinant protein used in the binding was visualized by Ponceau staining. (C) The interaction of TDRD3 with DHX9 requires a functional Tudor domain. A co-IP assay was performed to assess the interaction of Flag-tagged WT and methylarginine binding-deficient (E691K) TDRD3 with endogenous DHX9 in MCF7 cells. The input and αFlag antibody-immunoprecipitated samples were analyzed by western blot using indicated antibodies. (D) DHX9 interacts with TDRD3 in an arginine methylation-dependent manner. The interaction of DHX9 with TDRD3 was detected by co-IP in MCF7 cells treated with vehicle (−) or with the type I PRMT inhibitor MS023 (+). The level of cellular ADMA was detected by western blot using a pan-ADMA antibody (ASYM26). The DHX9-immunoprecipitated samples were probed for methylation and interactions with TDRD3 using the ASYM26 and αTDRD3 antibodies.

Fig 4: TDRD3 promotes DHX9 helicase activity in R-loop resolution. (A) DHX9 resolves R-loops in a helicase activity-dependent manner. Coomassie Blue staining of recombinant wild type (WT) and helicase activity-deficient (K417R) DHX9 purified from HEK293 cells (left). The helicase assay on R-loops was performed by incubating increasing amounts of recombinant WT or K417R DHX9 with 5′ 6-FAM-labeled R-loop substrates (5 nM) at 37°C for 10 min. The reaction products were analyzed by gel electrophoresis and fluorescence imaging (right). The open triangle indicates the recombinant proteins. (B) TDRD3 stimulates the helicase activity of WT DHX9, but not K417R mutant DHX9, in R-loop resolution. Coomassie Blue staining of recombinant WT, K417R mutant DHX9, and TDRD3 purified from HEK293 cells (left). The helicase assay on R-loops was performed by incubating 5′ 6-FAM-labeled R-loop substrates with constant amounts of either WT or K417R mutant DHX9 (2 nM) and increasing amounts (40 and 60 nM) of TDRD3 (right). The open triangles indicate the recombinant proteins. (C) The OB-fold of TDRD3 binds single-stranded DNA (ssDNA). A graphic summary of the interactions of truncated TDRD3 fragments with ssDNA is shown (upper panel). Coomassie Blue staining of recombinant GST-fusion proteins of TDRD3, including its OB-fold, UBA domain, and Tudor domain, purified from E. coli (lower left panel). The electrophoretic mobility shift assay (EMSA) was performed by incubating a 5′ 6-FAM-labeled ssDNA oligonucleotide (5 nM) with increasing amounts (25 and 50 nM) of the recombinant TDRD3 proteins (lower right panel). The solid triangle indicates the protein–nucleotide complex. (D) The OB-fold is essential for the interaction of TDRD3 with ssDNA. A graphic summary of the WT, methylarginine binding-deficient (E691K), and OB-fold-truncated (ΔOB) TDRD3 interaction with ssDNA is shown (upper panel). Coomassie Blue staining of all three recombinant TDRD3 proteins purified from HEK293 cells (lower left panel). EMSA was performed to detect the binding of increasing amounts (20, 40 and 60 nM) of recombinant proteins with a 5′ 6-FAM-labeled ssDNA oligonucleotide (5 nM) (lower right panel). The open triangle indicates the recombinant proteins. The solid triangle indicates the protein-nucleotide complex. (E) TDRD3 interacts with R-loops, but not DNA/RNA hybrids. EMSA was performed by incubating increasing amounts (40 and 60 nM) of recombinant WT and OB-fold-truncated (ΔOB) TDRD3 with 5′ 6-FAM-labeled R-loop or DNA/RNA hybrid oligonucleotide (5 nM). The open triangle indicates the recombinant proteins. The solid triangle indicates the protein-nucleotide complex. (F) Both the OB-fold and the functional Tudor domain are required for TDRD3 to stimulate the helicase activity of DHX9 in R-loop resolution. The helicase assay on R-loops was performed by incubating 5′ 6-FAM-labeled R-loop substrates (5 nM) with a constant amount of DHX9 (2 nM) and increasing amounts (20, 40 and 60 nM) of WT, methylarginine binding-deficient (E691K), and ssDNA binding-deficient (ΔOB) TDRD3 (left). Helicase activity was quantified by measuring the percentage of unwound substrates under the indicated assay conditions (right). Statistical analysis was performed using Student's t-tests of data from three independent experiments. *** P < 0.001.

Fig 5: The TDRD3-DHX9 protein complex resolves promoter-associated R-loops. (A) Venn diagram of TDRD3 target genes and genes that form R-loops at their promoters, as identified by DRIPc-Seq (15). (B) TDRD3 knockout (KO) increases the R-loop levels at the promoters of its target genes. DRIP-qPCR analysis was performed to compare R-loop levels at the promoters of the TDRD3 target genes in wild type (WT) and TDRD3 KO MCF7 cells. Samples treated with RNase H, which disrupts R-loops, served as negative controls. (C) DHX9 knockdown does not affect TDRD3 protein expression. DHX9 and TDRD3 were detected by western blot analysis in MCF7 cells transfected with control siRNA (siControl) or DHX9-specific siRNA (siDHX9). ACTIN was used as a loading control. (D) DHX9 knockdown increases R-loop levels at the promoters of TDRD3 target genes. DRIP-qPCR analysis was performed to compare R-loop levels at the promoters of the TDRD3 target genes in control and DHX9 knockdown MCF7 cells. Samples treated with RNase H served as negative controls. (E) MCF7 cells were transfected with control siRNA (siControl) or DHX9-specific siRNA (siDHX9). After 24 h, the siDHX9-transfected cells were transfected with either an empty vector, WT DHX9 or helicase activity-deficient (K417R) DHX9 for an additional 48 h. DHX9 and TDRD3 were detected in these cells by western blot analysis. The anti-ACTIN was used as a loading control. (F) The helicase activity of DHX9 is essential for resolving R-loops at TDRD3 target gene promoters. DRIP-qPCR analysis was performed to compare R-loop levels at the promoters of the TDRD3 target genes in control (siControl) or DHX9 knockdown (siDHX9) MCF7 cells or DHX9 knockdown MCF7 cells with re-expression of either WT DHX9 (siDHX9 + WT) or helicase activity-deficient DHX9 (siDHX9 + K417R), as described in (E). Samples treated with RNase H served as negative controls. (G) TDRD3 KO and DHX9 knockdown cause RNAPII accumulation at gene promoters. RNAPII ChIP assays were performed in WT and TDRD3 KO, as well as control and DHX9 knockdown MCF7 cells. Normal rabbit IgG was used as a negative control. (H) Impact of TDRD3 KO and DHX9 knockdown on the nascent RNA transcription. Diagram demonstration of RHOB and RAD23A gene locus. The bars at the bottom of the diagrams indicate the amplicon locations for nascent RNA transcripts (red: promoter; blue: gene body and 3′end). Chromatin RNA immunoprecipitation was performed in WT and TDRD3 KO, as well as control and DHX9 knockdown MCF7 cells. The levels of nascent RNA were quantified by RT-qPCR. Experiments were performed independently three times. Data are represented as mean ± standard deviation of three technical qPCR replicates. Statistical analysis was performed using Student's t-tests. * P < 0.05; ** P < 0.01; *** P < 0.001.