RNA helicase DDX21 coordinates transcription and ribosomal RNA processing

Abstract

DEAD-box RNA helicases are vital for the regulation of various aspects of the RNA life cycle, but the molecular underpinnings of their involvement, particularly in mammalian cells, remain poorly understood. Here we show that the DEAD-box RNA helicase DDX21 can sense the transcriptional status of both RNA polymerase (Pol) I and II to control multiple steps of ribosome biogenesis in human cells. We demonstrate that DDX21 widely associates with Pol I- and Pol II-transcribed genes and with diverse species of RNA, most prominently with non-coding RNAs involved in the formation of ribonucleoprotein complexes, including ribosomal RNA, small nucleolar RNAs (snoRNAs) and 7SK RNA. Although broad, these molecular interactions, both at the chromatin and RNA level, exhibit remarkable specificity for the regulation of ribosomal genes. In the nucleolus, DDX21 occupies the transcribed rDNA locus, directly contacts both rRNA and snoRNAs, and promotes rRNA transcription, processing and modification. In the nucleoplasm, DDX21 binds 7SK RNA and, as a component of the 7SK small nuclear ribonucleoprotein (snRNP) complex, is recruited to the promoters of Pol II-transcribed genes encoding ribosomal proteins and snoRNAs. Promoter-bound DDX21 facilitates the release of the positive transcription elongation factor b (P-TEFb) from the 7SK snRNP in a manner that is dependent on its helicase activity, thereby promoting transcription of its target genes. Our results uncover the multifaceted role of DDX21 in multiple steps of ribosome biogenesis, and provide evidence implicating a mammalian RNA helicase in RNA modification and Pol II elongation control.

Extended Data Figure 1. DDX21 associates with non- and protein-coding ribosomal genes

a, MEME analysis of DDX21-bound regions defined by DDX21 ChIP-seq. Motif logo, annotated transcription factor, number of motif instances within the ChIP-seq regions, Z score, and P value for each motif are shown. b, DDX21 ChIP-qPCR from HeLa cell chromatin extracts with primers spanning a representative number of loci found to be enriched in the DDX21 ChIP-seq analyses from HEK293 cells. Data are mean and s.d. of three independent experiments. c, Comparison of DDX21 (this study) and H3K4me3 (publically available data, see Methods for accession numbers) ChIP-seq-bound regions. 2,863 regions are common between the data sets, 505 regions are unique to DDX21, and 11,403 regions are unique to the H3K4me3 data set. d, Gene Ontology terms for H3K4me3 regions that are either DDX21-bound (left) or not bound by DDX21 (right). e, Box plots representing the expression levels of snoRNA-host genes whose promoter regions are either bound or not by DDX21. As shown, snoRNA-host gene promoters bound by DDX21 are, on average, more highly expressed than those not occupied by DDX21. Fragments per kilobase of exon per million mapped reads (FPKM) values were taken from publically available HEK293 RNA-seq data (see Methods for accession number). The P value (P ≤ 0.05) was calculated using the Wilcoxon signed-rank test. f, g, UCSC genome browser tracks depicting DDX21 ChIP-seq and iCLIP-seq, and RNA-seq enrichment profiles at differentially expressed snoRNA-host genes in HEK293 cells.

Extended Data Figure 2. DDX21 positively regulates transcription of Pol I-and Pol II-dependent ribosomal genes

a, siRNA-mediated knockdown of the DDX21 antibody used for ChIP. We transfected HEK293 cells with two different sets of siRNAs targeting endogenous DDX21 mRNA (siRNA1 and siRNA2 (3′ UTR)) and performed western blots with the indicated antibodies. As shown, the DDX21-specific band is diminished in cells transfected with DDX21-targeting siRNAs, but not with control siRNAs. Actin was used as a loading control for this experiment. b, RT–qPCR analysis assessing the RNA expression levels of the same genes analysed in Fig. 1h upon DDX21 knockdown by a second siRNA that targets the 3′ UTR of DDX21 mRNA. Data are mean and s.d. of three independent experiments. For DDX21-target genes the difference between control and DDX21 siRNA is significant, P ≤ 0.05 (Student’s t-test). c, Diagram of DDX21 protein domains. The two conserved RecA-like (A and B) domains and the GUCT domains are shown in green and blue, respectively. Amino acids targeted for mutation to convert DDX21^WT into DDX21^SAT, the ATP-hydrolysis mutant, are indicated with red and purple lines in the diagram. Specific amino acid changes are displayed below. d, qRT–PCR analysis assessing nascent unspliced mRNA levels from additional DDX21-target and DDX21-non-target promoters. For a detailed description see Fig. 1j. Data are mean and s.d. of three biological replicates. e, Nuclear rRNA abundance analysis by RNA BioAnalyzer of HEK293 cells depleted of DDX21 and rescued with DDX21^WT, DDX21^SAT, or DDX21^DEV. For each analysis, total RNA was isolated from 1,500,000 nuclei. Total nanogram amounts are shown for each of the two large rRNA subunits.

Extended Data Figure 3. Selective inhibition of Pol I alters DDX21 nuclear localization and chromatin association

a, b, Immunofluorescence images of methanol-fixed HEK293 cells after 1 h incubation with either DMSO or 2 μM of the specific Pol I inhibitor CX-5461. DDX21 (a) and fibrillarin (b) immuno-labellings are shown. Scale bars, 10 μm. c, d, ChIP-qPCR analyses from HEK293 sampling DDX21 genomic occupancy, at the rDNA locus (c) and at a representative panel of Pol II-regulated gene promoters (d), after treatment with DMSO or CX-5461. Data are mean and s.d. of three independent experiments. As displayed, inhibition of Pol I alters DDX21 nuclear localization and this coincides with nearly complete eviction of DDX21 from Pol I- and Pol II-regulated genes. e, f, ChIP-qPCR analyses from HEK293 cells treated with 50 ng ml⁻¹ of actinomycin-D for 1 h. Binding of the transcriptional repressor CTCF across the rDNA locus (e) and the c-MYC insulator element (MINE) (f) demonstrates that actinomycin-D treatment does not effect CTCF binding to chromatin. Red arrow indicates relative location of the CTCF DNA-binding site (DBS) at the rDNA locus. Data are mean and s.d. of three independent experiments.

Extended Data Figure 4. DDX21 nuclear re-localization is preferentially sensitive to acute transcriptional inhibition over other cellular stressors

Immunofluorescence analyses of HEK293 cells after targeting different metabolic pathways. For inhibition of mitogen, cells were starved for 16 h in the absence of serum. For cellular respiration inhibition, cells were treated for with either oligomycin (100 μM) or 2-deoxy-D-glucose (10 mM) for 1 h. To inhibit the mTOR pathway, cells were treated with 250 nM of either Torin 1 or rapamycin for 2 h.

Extended Data Figure 5. Tandem affinity iCLIP of FH–DDX21^WT

a, FH–DDX21^WT iCLIP ³²P-autoradiogram and western blots. All samples were loaded with constant input lysate amounts (actin loading). FH–DDX21^WT was isolated from HEK293 cells induced to express the transgene for 24 h and crosslinked with ultraviolet light (top panel same as Fig. 3a). b, Schematic of the modified iCLIP procedure. To achieve high stringency and specificity Flag–HA–DDX21^WT is first purified on anti-Flag–M2 agarose beads, washed with 1 M NaCl, 1% Triton X-100 and 1% sodium deoxycholate. Complexes are specifically eluted with Flag peptide and recaptured with anti-HA agarose. Standard iCLIP steps were performed thereafter to generate deep sequencing libraries. c–e, Scatter plot analysis of iCLIP reverse transcription stops on snoRNAs, rRNA and mRNAs within the FH–DDX21^WT (this study) and hnRNP-C (ref. ; publically available data) data sets. Little concordance between the data sets is evident, suggesting specific transcriptome targets of these two RNA binding proteins (RBPs). f, DDX21^WT ultraviolet RNA immunoprecipitation qRT–PCR of FH–DDX21^WT was performed in three conditions: native HEK293 cells crosslinked with ultraviolet light; FH–DDX21^WT HEK293 cells without crosslinking; and FH–DDX21^WT HEK293 cells with ultraviolet crosslinking. snoRNAs, scaRNAs and TERC were validated targets identified in the sequencing data. Each experiment was performed in biological duplicates (rep1 and rep2) and error bars represent s.d. of technical triplicates.

Extended Data Figure 6. Tandem affinity iCLIP of FH–DDX21^SAT

a, FH–DDX21^SAT was isolated from HEK293 cells induced to express the transgene for 24 h, at which point we did not observe significant dominant negative effects. iCLIP was performed as described for FH–DDX21^WT, and biological duplicates of FH–DDX21^SAT iCLIP ³²P-autoradiogram and western blots (lanes 2 and 3) are shown. All samples were loaded with constant input lysate amounts (actin loading). FH–DDX21^WT was loaded as a control. WB, western blot. b, Left, DDX21^SAT iCLIP reads annotated to known repetitive (rRNA and snRNAs) and non-repetitive (hg19 genome build: mRNAs and snoRNAs) regions of the human genome. Categories are notes with their respective percentage of the total iCLIP experiment. Right, enriched Gene Ontology and KEGG pathway terms from DDX21^SAT-bound mRNAs obtained using the DAVID tool. The x axis values (in log scale) correspond to the negative Benjamini P value. c, Distribution of all DDX21^SAT-bound snoRNAs, representing C/D box, H/ACA box and scaRNAs. The number (n) and fraction (per cent) of each snoRNA type is displayed. d, Comparison of the snoRNAs bound by DDX21^WT and DDX21^SAT, revealing significant overlap between the active and catalytically inactive DDX21. e, DDX21^SAT iCLIP reads mapped to the transcribed region of the rDNA. f, DDX21^WT (left) and DDX21^SAT (right) iCLIP reads mapped to the repetitive U3 snoRNA. Binding is represented as reverse transcription stops per nucleotide normalized to the total number of reverse transcription stops mapping to the U3 snoRNA. Two strong binding sites are evident between bases 25–40 and 175–185 of U3 in DDX21^WT iCLIP, whereas the 5′ binding site is reduced in DDX21^SAT. nts, nucleotides. g, qRT–PCR analysis assessing the expression levels of several snoRNAs 24 h after expression of either DDX21^WT or DDX21^SAT. This experiment was performed in biological duplicates. Data are mean and s.d. of technical triplicates.

Extended Data Figure 7. DDX21 functionally interacts with snoRNAs and the snoRNP

a, UCSC genome browser view of DDX21^WT iCLIP reads across the snorD66 snoRNA. The C box [C] and D box [D] regions are highlighted in red. b, Same visualization as in a but showing the snorA67 snoRNA with the H box [H] and ACA box [ACA] regions highlighted. c, Immunoprecipitation of NOP58 from HEK293 nuclear extracts confirms DDX21 as a protein member of the snoRNP machinery. As a control for this experiment we performed western blots against FBL, a well-known NOP58-interacting partner and an essential factor of the snoRNP machinery. d, DDX21 interacts with XRN2, a 5′–3′ exoribonuclease required for maturation and processing of snoRNAs. The DDX21–XRN2 interaction appears to be bridged by RNA, as treatment of the nuclear lysates with RNaseA abolishes the interaction. e, Schematic of the site-directed RNaseH cleavage of RNA sensitive to 2′-Ome. RNA of interest is hybridized to a 2′-Ome/DNA chimaeric oligonucleotide in which the DNA nucleotides specifically target the ability of RNaseH to interrogate the 2′-Ome status of a single nucleotide. 2′-Ome will inhibit RNaseH and leave intact RNA, while unmethylated RNA will be cleaved. f, UCSC genome browser view of DDX21^WT and DDX21^SAT iCLIP reads across the snoRNAs responsible for guiding the modifications tested in Fig. 3f.

Extended Data Figure 8. Association of DDX21 with the RNA and protein components of the 7SK snRNP

a, Comparison of DDX21^WT ChIP-seq targets to DDX21^WT iCLIP-bound mRNAs. The numbers of unique and common genes are represented, revealing that most ChIP-seq-bound genes are not immunoprecipitated by iCLIP but that some are recovered in both assays. b, iCLIP read distribution of DDX21^WT-target mRNAs categorized by the regions within mRNAs that were bound. Most iCLIP reads fell outside the 5′ UTR. c, DDX21^WT iCLIP reads mapping to short repetitive RNAs of the human genome. Percentages of the top four short repetitive RNAs are shown. d, Secondary structure model of the 7SK snRNA annotated with iCLIP reverse transcription stops identified from the DDX21^WT (blue) and DDX21^SAT (orange) experiments. Nucleotides commonly crosslinked are labelled in green. Known RNA binding protein sites: HEXIM1/2 is highlighted in purple; P-TEFb is highlighted in red; and other P-TEFb ‘release’ factors in the centre are highlighted in green. e, Co-immunoprecipitation analysis of HEXIM1 as assayed by western blotting for DDX21^WT and 7SK snRNP components (CDK9 and LARP7). f, Immunoprecipitation of Flag–HA–DDX21^WT from HEK293 nuclear extracts confirms DDX21 as a protein component of the 7SK snRNP through co-recovery of LARP7. The abundant protein actin, which is not part of the 7SK snRNP, was not recovered.

Extended Data Figure 9. Binding of the DDX21–7SK snRNP at ribosomal gene promoters

a, b, ChIP-qPCR of CDK9 (a) and HEXIM1 (b) in HEK293 cells at representative Pol II-regulated, DDX21-target and -non-target gene promoters, negative control regions, and the rDNA locus. c, ChIP-qPCR of DDX21 in control or 3′-7SK-ASO-treated HEK293 cells at representative Pol II-regulated, DDX21-target gene promoters, negative control regions, and the rDNA locus. For the promoter-associated genes, P ≤ 0.05 (Student’s t-test) when compared to control ASO. d, Gene Ontology molecular function and cellular component analysis of publically available CDK9 ChIP-seq data. e, ChIP-qPCR of total Pol II in control or DDX21-targeting siRNA-treated HEK293 cells at representative TSSs of Pol II-regulated, DDX21-target gene promoters and negative control regions. Data are mean and s.d. of three independent experiments.

Extended Data Figure 10. Catalytically inactive DDX21 is still incorporated into the 7SK snRNP

a, ChIP-qPCR of DDX21^WT (black) and DDX21^SAT (green) in HEK293 cells at representative TSSs of Pol II-regulated, DDX21-target gene promoters and negative control regions. Data are mean and s.d. of three independent experiments. b, Immunoprecipitation of DDX21^WT, DDX21^DEV or DDX21^SAT from HEK293 nuclear extracts confirms DDX21 interacts with CDK9 (P-TEFb) regardless of its catalytic activity. c, DDX21^WT (blue) and DDX21^SAT (orange) annotated iCLIP reads mapped across the 7SK snRNA. The four annotated stem–loops are marked below the graph. d, Model of multi-level control of ribosomal pathway by DDX21. In the nucleolus, DDX21 associates with the chromatin across the transcribed region of the rDNA and is a component of the snoRNP. Furthermore, DDX21 functionally interacts with the rRNA, snoRNAs and snoRNP to control 2′-Ome deposition on the rRNA in a helicase activity-dependent manner. In the nucleoplasm, DDX21 is bound to the promoter regions of ribosomal Pol II-transcribed genes, many of which contain precursor snoRNA transcripts. Mechanistically, DDX21 activates transcription of its target genes through the 7SK–P-TEFb axis. As part of the 7SK snRNP, DDX21 can facilitate the release of P-TEFb from the inhibitory complex in a manner dependent on ATP hydrolysis, leading to productive Pol II elongation and increased phosphorylation of Ser 2. Efficient transcription of its target genes enforces high expression of both snoRNAs and other ribosomal proteins critical for the rRNA maturation process, placing DDX21 as a central operator of the ribosomal pathway.

Figure 1. DDX21 associates with actively transcribed ribosomal genes

a, DDX21 ChIP-seq reads were mapped to a custom annotation file of the human rDNA locus and compared to input reads. IGS, intergenic spacer. b, Distribution of DDX21 ChIP-seq peaks over known genomic features. TSS, transcription start site; TTS, transcription termination site. c, d, Average ChIP-seq signal profiles from publically available data sets (see Methods) were generated for Pol II (c) and the indicated histone modifications around the centre of DDX21-bound regions (d). bp, base pairs. e, Genomic regions enrichment of annotations tool (GREAT) analyses of DDX21-bound regions. The x axis corresponds to the negative binomial P values. GO, Gene Ontology. f, University of California Santa Cruz (UCSC) genome browser tracks of DDX21 ChIP-seq at ribosomal genes containing intronic snoRNAs. g, Quantitative assessment of DDX21 binding to the promoters of snoRNA host genes. h, qRT–PCR analysis of a representative panel of DDX21-bound genes upon DDX21 knockdown. Bars represent the average of three independent experiments. For DDX21-target genes, expression difference P < 0.05 (Student’s t-test). i, qRT–PCR analysis of DDX21 levels after transfecting control or DDX21 siRNA and/or subsequent overexpression of siRNA-resistant DDX21^WT or DDX21^SAT. j, qRT–PCR analysis assessing the synthesis of newly made, unspliced transcripts upon DDX21 knockdown and/or reconstitution with siRNA-resistant DDX21^WT or DDX21^SAT (see Methods). Data are mean and s.d. of three independent biological replicates. DMSO, dimethylsulphoxide; FL, flavopiridol.

Figure 2. DDX21 chromatin association is sensitive to Pol I and Pol II transcriptional status

a, Representative immunofluorescence images of methanol-fixed HEK293 cells after 1-h incubation with DMSO, actinomycin-D (50 ng ml⁻¹) or flavopiridol (1 μM). Scale bars, 10 μm. DAPI, 4′,6-diamidino-2-phenylindole. b, c, ChIP-qPCR analysis in HEK293 cells sampling DDX21 genomic occupancy at the rDNA locus and at a representative panel of Pol II-regulated, DDX21-target promoters, upon treatment with DMSO, actinomycin-D or flavopiridol. Data are mean and s.d. of three independent experiments. Neg, negative controls.

Figure 3. DDX21 binds rRNA and snoRNAs and facilitates rRNA modification

a, Autoradiogram of ³²P-labelled RNA crosslinked to Flag–HA–DDX21^WT. RNA–protein complexes are seen in the purifications from Flag–HA–DDX21^WT cells but not from control cells. kDa, kilodaltons; UV, ultraviolet crosslinking. b, Left, DDX21 iCLIP reads annotated to known repetitive (rRNA and snRNAs) and non-repetitive regions of the human genome with percentage of the total iCLIP reads shown. Right, enriched Gene Ontology and KEGG pathway terms from DDX21-bound mRNAs obtained using the DAVID tool. The x axis values correspond to the negative Benjamini P value. c, DDX21^WT iCLIP reads mapped to the transcribed region of the rDNA. The mature portions of the 18S, 5.8S and 28S rRNAs are highlighted (grey). Known 2′-Ome (*) and Ψ (↑) modification sites are marked below the iCLIP reads. RT, reverse transcription. ETS, external transcribed spacer. d, Distribution of all DDX21^WT-bound snoRNAs, representing C/D box, H/ACA box, and scaRNAs. The number (n) and fraction of each snoRNA type are displayed. e, Co-immunoprecipitation analysis of Flag–HA–DDX21^WT and components of the snoRNPs. Conditions with/without RNaseA treatment are shown. f, Site-directed cleavage of rRNA by RNaseH at three sites. Values from three biological replicates are shown. Student’s t-test was performed and only DDX21^SAT was significantly reduced (P < 0.05) relative to control. NS, not significant.

Figure 4. DDX21 promotes release of P-TEFb from the 7SK snRNP

a, DDX21^WT iCLIP reads mapped to the 7SK snRNA. The four stem–loops (SL1–4) are marked below the iCLIP reads and are shadowed on the cartoon in c. b, ChIP-qPCR of DDX21 in control or 5′-7SK-antisense oligonucleotide (ASO)-treated HEK293 cells at representative Pol II-regulated, DDX21-target gene promoters, negative control regions (Neg), and the rDNA locus. Data are mean and s.d. For DDX21-target genes the difference between control ASO and 5′ ASO is significant at P < 0.05 (Student’s t-test). c, d, Schematic (left) and western blot analysis (right) of a P-TEFb release assay as a function of increasing amounts of DDX21^WT. e, ChIP-qPCR of Pol II Ser2p of representative Pol II-regulated genes at their transcriptional termination sites (TTS) in control or DDX21-siRNA-treated cells. Data are mean and s.d. of three independent experiments. f, Western blot analyses of a P-TEFb release assay with either DDX21^WT or catalytically impaired DDX21^DEV and DDX21^SAT.