Skip to main content

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

  • Letter
  • Published:

DHX9 suppresses RNA processing defects originating from the Alu invasion of the human genome

Nature volume 544, pages 115–119 (2017)Cite this article

Subjects

Abstract

Transposable elements are viewed as ‘selfish genetic elements’, yet they contribute to gene regulation and genome evolution in diverse ways1. More than half of the human genome consists of transposable elements2. Alu elements belong to the short interspersed nuclear element (SINE) family of repetitive elements, and with over 1 million insertions they make up more than 10% of the human genome2. Despite their abundance and the potential evolutionary advantages they confer, Alu elements can be mutagenic to the host as they can act as splice acceptors, inhibit translation of mRNAs and cause genomic instability3. Alu elements are the main targets of the RNA-editing enzyme ADAR4 and the formation of Alu exons is suppressed by the nuclear ribonucleoprotein HNRNPC5, but the broad effect of massive secondary structures formed by inverted-repeat Alu elements on RNA processing in the nucleus remains unknown. Here we show that DHX9, an abundant6 nuclear RNA helicase7, binds specifically to inverted-repeat Alu elements that are transcribed as parts of genes. Loss of DHX9 leads to an increase in the number of circular-RNA-producing genes and amount of circular RNAs, translational repression of reporters containing inverted-repeat Alu elements, and transcriptional rewiring (the creation of mostly nonsensical novel connections between exons) of susceptible loci. Biochemical purifications of DHX9 identify the interferon-inducible isoform of ADAR (p150), but not the constitutively expressed ADAR isoform (p110), as an RNA-independent interaction partner. Co-depletion of ADAR and DHX9 augments the double-stranded RNA accumulation defects, leading to increased circular RNA production, revealing a functional link between these two enzymes. Our work uncovers an evolutionarily conserved function of DHX9. We propose that it acts as a nuclear RNA resolvase that neutralizes the immediate threat posed by transposon insertions and allows these elements to evolve as tools for the post-transcriptional regulation of gene expression.

This is a preview of subscription content, access via your institution

Access options

Buy this article

  • Purchase on Springer Link
  • Instant access to full article PDF
Buy now

Prices may be subject to local taxes which are calculated during checkout

Figure 1: uvCLAP and FLASH reveal DHX9 binding on Alu elements.
Figure 2: Loss of DHX9 protein causes increased circular RNA generation and impaired translation.
Figure 3: Loss of DHX9 leads to RNA processing defects.
Figure 4: RNA-independent interaction of ADARp150 with DHX9 is conserved between mouse and human.

Similar content being viewed by others

References

  1. Kazazian, H. H., Jr. Mobile elements: drivers of genome evolution. Science 303, 1626–1632 (2004)

    Article  CAS  ADS  Google Scholar 

  2. de Koning, A. P. J., Gu, W., Castoe, T. A., Batzer, M. A. & Pollock, D. D. Repetitive elements may comprise over two-thirds of the human genome. PLoS Genet. 7, e1002384 (2011)

    Article  CAS  Google Scholar 

  3. Elbarbary, R. A., Lucas, B. A. & Maquat, L. E. Retrotransposons as regulators of gene expression. Science 351, aac7247 (2016)

  4. Kim, D. D. Y. et al. Widespread RNA editing of embedded alu elements in the human transcriptome. Genome Res. 14, 1719–1725 (2004)

    Article  CAS  Google Scholar 

  5. Zarnack, K. et al. Direct competition between hnRNP C and U2AF65 protects the transcriptome from the exonization of Alu elements. Cell 152, 453–466 (2013)

    Article  CAS  Google Scholar 

  6. Hein, M. Y. et al. A human interactome in three quantitative dimensions organized by stoichiometries and abundances. Cell 163, 712–723 (2015)

    Article  CAS  Google Scholar 

  7. Koh, H. R., Xing, L., Kleiman, L. & Myong, S. Repetitive RNA unwinding by RNA helicase A facilitates RNA annealing. Nucleic Acids Res. 42, 8556–8564 (2014)

    Article  CAS  Google Scholar 

  8. Gantier, M. P. & Williams, B. R. G. The response of mammalian cells to double-stranded RNA. Cytokine Growth Factor Rev. 18, 363–371 (2007)

    Article  CAS  Google Scholar 

  9. Lemaire, P. A., Anderson, E., Lary, J. & Cole, J. L. Mechanism of PKR Activation by dsRNA. J. Mol. Biol. 381, 351–360 (2008)

    Article  CAS  Google Scholar 

  10. Elbarbary, R. A., Li, W., Tian, B. & Maquat, L. E. STAU1 binding 3′ UTR IRAlus complements nuclear retention to protect cells from PKR-mediated translational shutdown. Genes Dev. 27, 1495–1510 (2013)

    Article  CAS  Google Scholar 

  11. Oshiumi, H., Kouwaki, T. & Seya, T. Accessory factors of cytoplasmic viral RNA sensors required for antiviral innate immune response. Front. Immunol. 7, 200 (2016)

    Article  Google Scholar 

  12. Lee, C. G . et al. RNA helicase A is essential for normal gastrulation. Proc. Natl Acad. Sci. USA 95, 13709–13713 (1998)

    Article  CAS  ADS  Google Scholar 

  13. Wang, T. et al. Identification and characterization of essential genes in the human genome. Science 350, 1096–1101 (2015)

    Article  CAS  ADS  Google Scholar 

  14. Tagwerker, C. et al. A tandem affinity tag for two-step purification under fully denaturing conditions: application in ubiquitin profiling and protein complex identification combined with in vivocross-linking. Mol. Cell. Proteomics 5, 737–748 (2006)

    Article  CAS  Google Scholar 

  15. Batzer, M. A. & Deininger, P. L. Alu repeats and human genomic diversity. Nat. Rev. Genet. 3, 370–379 (2002)

    Article  CAS  Google Scholar 

  16. Kim, J., Cantor, A. B., Orkin, S. H. & Wang, J. Use of in vivo biotinylation to study protein-protein and protein-DNA interactions in mouse embryonic stem cells. Nat. Protocols 4, 506–517 (2009)

    Article  CAS  Google Scholar 

  17. Kramerov, D. A. & Vassetzky, N. S. Origin and evolution of SINEs in eukaryotic genomes. Heredity 107, 487–495 (2011)

    Article  CAS  Google Scholar 

  18. Ilik, I. A. et al. Tandem stem-loops in roX RNAs act together to mediate X chromosome dosage compensation in Drosophila. Mol. Cell 51, 156–173 (2013)

    Article  CAS  Google Scholar 

  19. Quinn, J. J. et al. Revealing long noncoding RNA architecture and functions using domain-specific chromatin isolation by RNA purification. Nat. Biotechnol. 32, 933–940 (2014)

    Article  CAS  Google Scholar 

  20. Noé, L . & Kucherov, G. YASS: enhancing the sensitivity of DNA similarity search. Nucleic Acids Res. 33, W540–W543 (2005)

    Article  Google Scholar 

  21. Jeck, W. R. et al. Circular RNAs are abundant, conserved, and associated with ALU repeats. RNA 19, 141–157 (2013)

    Article  CAS  Google Scholar 

  22. Zhang, X.-O. et al. Complementary sequence-mediated exon circularization. Cell 159, 134–147 (2014)

    Article  CAS  Google Scholar 

  23. Memczak, S. et al. Circular RNAs are a large class of animal RNAs with regulatory potency. Nature 495, 333–338 (2013)

    Article  CAS  ADS  Google Scholar 

  24. Lee, C. G., Chang, K. A., Kuroda, M. I. & Hurwitz, J. The NTPase/helicase activities of Drosophila maleless, an essential factor in dosage compensation. EMBO J. 16, 2671–2681 (1997)

    Article  CAS  Google Scholar 

  25. Ricci, E. P. et al. Staufen1 senses overall transcript secondary structure to regulate translation. Nat. Struct. Mol. Biol. 21, 26–35 (2014)

    Article  CAS  Google Scholar 

  26. Svensson, M. & Agace, W. W. Role of CCL25/CCR9 in immune homeostasis and disease. Expert Rev. Clin. Immunol. 2, 759–773 (2006)

    Article  CAS  Google Scholar 

  27. Neumann, B. et al. Phenotypic profiling of the human genome by time-lapse microscopy reveals cell division genes. Nature 464, 721–727 (2010)

    Article  CAS  ADS  Google Scholar 

  28. McCleland, M. L. et al. The highly conserved Ndc80 complex is required for kinetochore assembly, chromosome congression, and spindle checkpoint activity. Genes Dev. 17, 101–114 (2003)

    Article  CAS  Google Scholar 

  29. UniProt Consortium. UniProt: a hub for protein information. Nucleic Acids Res. 43, D204–D212 (2015)

  30. Shin, Y. J. et al. Nanobody-targeted E3-ubiquitin ligase complex degrades nuclear proteins. Sci. Rep. 5, 14269 (2015)

    Article  CAS  ADS  Google Scholar 

  31. Ran, F. A. et al. Genome engineering using the CRISPR-Cas9 system. Nat. Protocols 8, 2281–2308 (2013)

    Article  CAS  Google Scholar 

  32. Chelmicki, T. et al. MOF-associated complexes ensure stem cell identity and Xist repression. eLife 3, e02024 (2014)

    Article  Google Scholar 

  33. Konig, J. et al. iCLIP--transcriptome-wide mapping of protein-RNA interactions with individual nucleotide resolution. J. Vis. Exp. 50, 2638 (2011)

    Google Scholar 

  34. Faircloth, B. C. & Glenn, T. C. Not all sequence tags are created equal: designing and validating sequence identification tags robust to indels. PLoS One 7, e42543 (2012)

    Article  CAS  ADS  Google Scholar 

  35. König, J. et al. iCLIP reveals the function of hnRNP particles in splicing at individual nucleotide resolution. Nat. Struct. Mol. Biol. 17, 909–915 (2010)

    Article  Google Scholar 

  36. Dodt, M., Roehr, J. T., Ahmed, R. & Dieterich, C. FLEXBAR—flexible barcode and adapter processing for next-generation sequencing platforms. Biology (Basel) 1, 895–905 (2012)

    Google Scholar 

  37. Langmead, B. & Salzberg, S. L. Fast gapped-read alignment with Bowtie 2. Nat. Methods 9, 357–359 (2012)

    Article  CAS  Google Scholar 

  38. Ibrahim, M. M., Lacadie, S. A. & Ohler, U. JAMM: a peak finder for joint analysis of NGS replicates. Bioinformatics 31, 48–55 (2015)

    Article  CAS  Google Scholar 

  39. Ramírez, F. et al. deepTools2: a next generation web server for deep-sequencing data analysis. Nucleic Acids Res. 44 (W1), W160–W165 (2016)

    Article  Google Scholar 

  40. Afgan, E. et al. The Galaxy platform for accessible, reproducible and collaborative biomedical analyses: 2016 update. Nucleic Acids Res. 44 (W1), W3–W10 (2016)

    Article  CAS  Google Scholar 

  41. Holmqvist, E. et al. Global RNA recognition patterns of post-transcriptional regulators Hfq and CsrA revealed by UV crosslinking in vivo. EMBO J. 35, 991–1011 (2016)

    Article  CAS  Google Scholar 

  42. Day, D. et al. Estimating enrichment of repetitive elements from high-throughput sequence data. Genome Biol. 11, R69 (2010)

    Article  Google Scholar 

  43. Novák, P., Neumann, P. & Macas, J. Graph-based clustering and characterization of repetitive sequences in next-generation sequencing data. BMC Bioinformatics 11, 378 (2010)

    Article  Google Scholar 

  44. Novák, P., Neumann, P., Pech, J., Steinhaisl, J. & Macas, J. RepeatExplorer: a Galaxy-based web server for genome-wide characterization of eukaryotic repetitive elements from next-generation sequence reads. Bioinformatics 29, 792–793 (2013)

    Article  Google Scholar 

  45. Kim, D. et al. TopHat2: accurate alignment of transcriptomes in the presence of insertions, deletions and gene fusions. Genome Biol. 14, R36 (2013)

    Article  Google Scholar 

  46. Liao, Y., Smyth, G. K. & Shi, W. featureCounts: an efficient general purpose program for assigning sequence reads to genomic features. Bioinformatics 30, 923–930 (2014)

    Article  CAS  Google Scholar 

  47. Robinson, M. D., McCarthy, D. J. & Smyth, G. K. edgeR: a Bioconductor package for differential expression analysis of digital gene expression data. Bioinformatics 26, 139–140 (2010)

    Article  CAS  Google Scholar 

  48. Anders, S., Reyes, A. & Huber, W. Detecting differential usage of exons from RNA-seq data. Genome Res. 22, 2008–2017 (2012)

    Article  CAS  Google Scholar 

  49. Dobin, A. et al. STAR: ultrafast universal RNA-seq aligner. Bioinformatics 29, 15–21 (2013)

    Article  CAS  Google Scholar 

  50. Steigemann, P. et al. Aurora B-mediated abscission checkpoint protects against tetraploidization. Cell 136, 473–484 (2009)

    Article  Google Scholar 

  51. Picardi, E. et al. REDIportal: a comprehensive database of A-to-I RNA editing events in humans. Nucleic Acids Res. 45, D750–D757 (2017)

    Article  CAS  Google Scholar 

Download references

Acknowledgements

We thank members of the Akhtar laboratory especially B. Sheikh, C. Keller and K. Lam for critical reading of the manuscript and helpful discussions. We also thank N. Iovino and R. Sawarkar for critical reading of the manuscript. We also thank the members of the Deep Sequencing, Imaging, FACS and the Proteomics Facilities for their support. We acknowledge the support of the Freiburg Galaxy Team: B. Grüning and T. Houwaart, Bioinformatics, University of Freiburg. This work was supported by CRC 992/2 2016 awarded to A.A., T.M. and R.B., and CRC 746 and CRC 1140 awarded to A.A. R.B. is also funded by BA2168/11-1, SPP 1738 and CRC TRR 167.

Author information

Author notes

  1. Tuğçe Aktaş and İbrahim Avşar Ilık: These authors contributed equally to this work.

Authors and Affiliations

  1. Max Planck Institute of Immunobiology and Epigenetics, Freiburg, Germany

    Tuğçe Aktaş, İbrahim Avşar Ilık, Vivek Bhardwaj, Cecilia Pessoa Rodrigues, Gerhard Mittler, Thomas Manke & Asifa Akhtar

  2. Institute for Informatics, Albert-Ludwigs-University, Freiburg, Germany

    Daniel Maticzka & Rolf Backofen

  3. Faculty of Biology, University of Freiburg, Freiburg, 79104, Germany

    Vivek Bhardwaj & Cecilia Pessoa Rodrigues

Authors
  1. Tuğçe Aktaş
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. İbrahim Avşar Ilık
    View author publications

    You can also search for this author in PubMed Google Scholar

  3. Daniel Maticzka
    View author publications

    You can also search for this author in PubMed Google Scholar

  4. Vivek Bhardwaj
    View author publications

    You can also search for this author in PubMed Google Scholar

  5. Cecilia Pessoa Rodrigues
    View author publications

    You can also search for this author in PubMed Google Scholar

  6. Gerhard Mittler
    View author publications

    You can also search for this author in PubMed Google Scholar

  7. Thomas Manke
    View author publications

    You can also search for this author in PubMed Google Scholar

  8. Rolf Backofen
    View author publications

    You can also search for this author in PubMed Google Scholar

  9. Asifa Akhtar
    View author publications

    You can also search for this author in PubMed Google Scholar

Contributions

T.A., I.A.I. and A.A. designed the experiments, T.A. and I.A.I. performed most of the experiments except live-cell imaging and mass-spectrometry, which were carried out by C.P.R. and G.M., respectively; I.A.I. and T.A. developed uvCLAP and FLASH methods with input from D.M., R.B. and A.A.; D.M. analysed uvCLAP and FLASH data and was supervised by R.B.; V.B. analysed the RNA-seq data, carried out repeat-enrichment and circRNA analysis and was supervised by T.M. and A.A.; I.A.I., T.A. and A.A. wrote the manuscript with input from D.M. and V.B.; R.B. and A.A. acquired funding; A.A. supervised all aspects of the study. All authors reviewed, edited and approved the paper.

Corresponding author

Correspondence to Asifa Akhtar.

Ethics declarations

Competing interests

T.A., I.A.I. and A.A. have filed a patent application related to this work; the rest of the authors declare no competing financial interests.

Extended data figures and tables

Extended Data Figure 1 uvCLAP/FLASH identifies DHX9 targets in humans and mice.

a, The domain organization of Drosophila, human and mouse Mle/DHX9 is shown side by side with the other RNA-binding proteins, EIF4A3, QKI-5/6 used in uvCLAP, and the two other enzymatic-domain-containing dsRNA-binding proteins PKR and ADAR. Blue boxes show dsRNA-binding domains, green boxes show helicase-N and helicase-C domains, orange boxes show KH domains, pink box shows the Z-DNA-interacting domains of ADAR. Orange boxes with vertical lines show the RGG-repeats at the C-term of DHX9/Mle/RHA proteins that interact with single-stranded nucleic acids. b, The schematic representation of the two methods (uvCLAP and FLASH) developed for the identification of in vivo targets of RNA-binding proteins. c, The distance between peaks which do not directly overlap with an Alu repeat to the nearest Alu element for RNA-binding proteins EIF4A3, EIF4A1, HNRNPK, QKI-5 and QKI-6 is significantly further than similar peaks for DHX9. uvCLAP and FLASH data are shown separately which are put together in Fig. 1c. d, The fraction of peaks which are directly overlapping with an Alu element is depicted for DHX9 in comparison to peaks randomly placed in targeted regions within a sequencing library (shown as F1, G1, G2) and for EIF4A3, QKI-6, QKI-5, EIF4A1 and HNRNPK in uvCLAP and FLASH experiments (sequencing library XL1). e, The evolutionary divergence of primate and murine SINEs from the 7SL RNA. Scale bar, genetic distance (arbitrary units). f, The fraction of peaks which are directly overlapping with a SINE B1 repeat is depicted for DHX9 in comparison to peaks randomly placed in targeted regions. g, Fraction of uvCLAP and FLASH peaks in mouse embryonic stem (mES) cells and neural progenitor cells (NPCs) with a distance of at most 100 nucleotides from SINE B repeats is significantly enriched (P < 2.2 × 10−16; Fisher’s exact test) for DHX9 in comparison to shuffled intron controls. h, The distances of DHX9 peaks in mES cells, which do not directly overlap with B1 repeats, to the nearest B1 repeats are shown. The distance is significantly smaller than the corresponding shuffled controls with median distances of 508 (n = 33,686) vs 1,312.5 (n = 38,436) nucleotides for uvCLAP and 514 (n = 44,557) vs 1,108.5 (n = 51,272) nucleotides for FLASH (all P < 2.2 × 10−16; one-tailed Mann–Whitney U-test). n = number of uvCLAP/FLASH peaks.

Extended Data Figure 2 Reproducibility of the replicates for each uvCLAP/FLASH library and enrichment of Alu repeat binding in human cells.

a, Replicates (shown as A and B) in human uvCLAP sequencing libraries F1, G1 and G2 are compared to each other using Spearman correlation and similarity is represented as a heat map. Except for replicates the only other two profiles which correlate are of QKI isoforms. b, Replicates (shown as A and B) in mouse uvCLAP sequencing library F3, which is composed of DHX9 knock-in mES cells and NPCs, and BirA ligase control cell line, are compared to each other using Spearman correlation and similarity is represented as a heat map. Replicates of each experiment cluster together. c, Replicates (shown as A and B) in Drosophila melanogaster uvCLAP sequencing libraries K1 and L1, which are composed of Mle and Eif4a3 profiles, are compared to each other using Spearman correlation and similarity is represented as a heat map. Replicates of each experiment cluster together. d, Replicates (shown as A and B) in human FLASH sequencing library XL1 are compared to each other by using Spearman correlation. DHX9 library replicates as well as different antibody used in two independent library replicates cluster together, whereas IgG control is separated. e, Enrichments of Alu repeat subfamilies in uvCLAP data (see Methods). All subfamilies of Alu elements are highly enriched in the DHX9 uvCLAP experiment, compared to EIF4A3, QKI-5 and QKI-6. f, Enrichments of all human repeat families in uvCLAP data (see Methods). Alu elements are highly enriched in the DHX9 uvCLAP experiment, compared to EIF4A3, QKI-5 and QKI-6. g, Mapping-free clustering (see Methods) of repeated reads within uvCLAP libraries for human DHX9, EIF4A3, QKI-6 and QKI-5 reveals that only the top cluster of DHX9 library is composed of Alu elements. Right, a table view for the identity of the top three clusters (maximum reads) produced by graph based clustering of raw sequencing reads from different uvCLAP experiments. Alu elements bound by DHX9 form two distinct clusters (as depicted in the table) while the top three clusters in other samples (QKI-6, QKI-5 and eIF4A3) contain various small RNA families. h, Right, replicates (shown as A and B) in mouse FLASH sequencing library XL9 are compared to each other using Spearman correlation (see Supplementary Table 3 for mapping statistics and number of peaks). i, Binding frequencies of DHX9, EIF4A3 and QKI isoform crosslinking events on mRNA targets shown for UTRs, exons and introns.

Extended Data Figure 3 Mapping for DHX9-bound RNAs onto repeats reveals an enrichment of SINE B binding in mouse cells.

a, Top, endogenous knock-in strategy is shown for the mouse Dhx9 gene. Right, differentiation snapshots of the tagged mES cell line into NPCs. Bottom, the identity of the tagged protein and its biotinylation is shown by DHX9 and streptavidin blots from Flag immunoprecipitation (IP) carried in nuclear extracts both for mES cells and NPCs. b, Endogenous knock-in strategy is shown for the mouse Msl1 gene. The integrity of the protein and its biotinylation is shown by MSL1 and streptavidin blots from Flag IP carried in nuclear extracts. The interaction ability of the tagged MSL1 protein with its known interaction partners (within MSL complex) is validated by this Flag IPs (performed on soluble nuclear extracts) by showing the co-IP for MOF (MYST1) and MSL2. POL2 (8WG16) is used as a loading control. c, Endogenous knock-in strategy is shown for the mouse Msl2 gene. The integrity of the protein and its biotinylation is shown by MSL2 and streptavidin blots from Flag IP carried in nuclear extracts. The interaction ability of the tagged MSL2 protein with its known interaction partners (within MSL complex) is validated by this Flag IP (performed on soluble nuclear extracts) by showing the co-IP for MOF (MYST1) and MSL1. d, Enrichments of SINE repeat subfamilies in uvCLAP data (see Methods). All SINE B subfamilies are highly enriched in DHX9 uvCLAP experiment, both in mES cells and NPCs, compared to MSL1 and MSL2. e, Mapping-free clustering of reads (see Methods) within uvCLAP libraries (mES cells and NPCs) reveals that only the top cluster of mouse DHX9 library is composed of SINE B repeats. f, Snapshots of uvCLAP and FLASH binding events (both in mES cells and NPCs) within Fam73b gene show that the crosslinking sites of DHX9 reside on SINE B repeats on opposing strands. SINEs on the plus strand are shown with red boxes, SINEs on the minus strand are shown with black boxes. Biological replicates were merged in these representations (also see Extended Data Fig. 2 for reproducibility between biological replicates).

Extended Data Figure 4 uvCLAP and FLASH reveal DHX9 binding on inverted-repeat Alu elements.

a, A snapshot of DHX9 binding in the intron of the QTRT1 gene. Alu elements are depicted with red (plus strand) or black (minus strand) boxes. Dot-plot generated by YASS (see Methods) showing that DHX9 binds to inverted repeats (structure-forming), shown by red lines, but not to direct repeats, shown by blue lines. b, A snapshot of DHX9 binding in the intron of the SLC25A27 gene where there are Alu elements on opposite strands. YASS-generated dot-plots as in a (for simplicity only the diagonal of the YASS dot-plot is shown). c, Genome-wide analysis of uvCLAP DHX9 binding near paired Alu elements shows preferential targeting of Alu elements with nearby binding partners in both gene (dark blue) and outside gene (light blue) bound regions. d, Genome-wide analysis of uvCLAP QKI-5 binding near paired Alu elements (binding enriched in introns) shows no significant difference for closeness to nearest potential binding partner. e, Genome-wide analysis of FLASH DHX9 binding near paired Alu elements (Ab#1) shows a preferential targeting of Alu elements with nearby binding partners in both gene (dark blue) and outside gene (light blue) bound regions. f, Genome-wide analysis of FLASH DHX9 (Ab#2) binding near paired Alu elements shows a preferential targeting of Alu elements with nearby binding partners in both gene (dark blue) and outside gene (light blue) bound regions. g, Genome-wide analysis of uvCLAP QKI-6 binding near paired Alu elements shows no significant difference for closeness to nearest potential binding partner. h, Genome-wide analysis of uvCLAP EIF4A1 binding near paired Alu elements shows no significant difference for closeness to nearest potential binding partner.

Extended Data Figure 5 DHX9 depletion increases circular RNA formation and represses translation of reporters containing inverted-repeat Alu elements within the 3′ UTR.

a, Left, the qPCR approach for detecting linear/circular RNAs with two sets of oligos indicated as divergent and convergent used for the detection of circular and linear RNAs, respectively. Right, quantitative real-time PCR analysis of previously reported circRNAs23 in DHX9-siRNA-treated HEK293FT cells (error bars represent standard deviation between biological quadruplicates). b, Snapshot of a circRNA generated locus, PPP1CB with DHX9 crosslinking sites on inverted Alu repeats. Blue bar depicted as ‘circPCR-PPP1CB’ represents the qPCR scored region in (Fig. 2c). Biological replicates were merged in these representations (also see Extended Data Fig. 2 for reproducibility between biological replicates). c, Top, schematic drawing of pmirGlo insert cloning is shown. Bottom, 3′ UTRs used as inserts in luciferase assays. d, Top, genomic position of tested guide RNAs in CRISPR/Cas9 depletion of DHX9 is shown (see Supplementary Table 2 for guide RNA sequences). Bottom left, efficiency of guide RNA pairs is shown by DHX9 and tubulin western blots. Bottom right, description for the making of the constitutive DHX9-depleted clone. e, Expression check by western blot analysis of rescue cells used on two independent (different days) experiments. f, Luciferase assays show that DHX9-depletion does not alter the expression of 3′ UTR elements without inverted-repeat Alu elements. g, Luciferase assay results are shown carried in HEK293FT and HeLa cells upon siRNA knockdown of DHX9. Knockdown efficiency is validated by western blot analysis for DHX9 and tubulin. Similar to what is shown in Fig. 2d reporter expression from the constructs with an inverted-repeat Alu element in 3′ UTRs are affected upon DHX9 depletion. Error bars represent standard deviation of a total of 20 data points that come from one experiment carried out with biological quadruplicates for each cloned 3′ UTR insert (5 with and 5 without inverted-repeat Alu elements).

Extended Data Figure 6 DHX9 depletion leads to changes in gene expression and exon usage.

a, Reproducibility of gene-expression changes between cells treated with siRNA 1 (green) and siRNA 2 (blue) (both show that DHX9 is the most severely downregulated gene in these datasets). Genes with a significant change in their expression levels in both experiments are shown with yellow points. b, MA plot of genes that show a reproducible change in both RNAi experiments (yellow dots in a). c, The extent of gene expression changes in the reproducibly misregulated genes in b. d, Pathway perturbation analysis using the data in b. e, Differentially spliced exons that show a reproducible change in DHX9-depletion experiments (siRNA 1 and siRNA 2). f, Differentially expressed genes (DEgenes) with respect to their detected splicing changes shows 55% of the mis-spliced genes are downregulated. g, GO term analysis using data in e. h, Tissue expression database (Genevestigator) results for CCL25 gene show its expression is specific to the intestine and thymus tissues. CCL25 is expressed at low levels every other tissue (in total 451 tissues). i, DHX9 binding in the ELAVL1–CCL25 locus is shown together with the poly(A)− and poly(A)+ RNA-seq data. Sashimi plots generated from control and DHX9-siRNA-treated samples show a new exon–exon junction from ELAVL1AS to CCL25 (depicted with red lines). For clarity, only plus strand data are shown. Biological replicates were merged in these representations (also see Extended Data Fig. 2 for reproducibility between biological replicates). Poly(A)+ RNA-seq data show that an anti-sense transcript in the opposite direction to ELAVL1 is now connected via a cryptic splice acceptor site to the first exon of CCL25 upon DHX9 depletion (two independent siRNAs, four biological replicates each). Bottom, qPCR validation of the CCL25 upregulation upon RNA processing defects in DHX9 knockdown samples. We observe that CCL25 is ~60-fold upregulated in DHX9-depleted cells, while the expression of neither ELAVL1 nor its accompanying anti-sense transcript change significantly. The cryptic splice acceptor site is enlarged on the bottom right.

Extended Data Figure 7 Many upregulated genes in DHX9 knockdown show RNA processing defects in their gene locus.

a, Genes that were more than twofold up- or downregulated are marked with red dots. Top 11 upregulated genes, which show signs of transcription bleed-through from upstream genes are highlighted with blue circles. CCL25, SALL4, PLIN2 and NTN5 are further highlighted. b, DHX9 binding in the PLIN2–RP11–146N23.1 gene locus on the nascent RNA originating from the antisense transcription of the DENND4C promoter and splicing defect upon DHX9 knockdown. Sashimi plots depicting the exon junctions of these genes are shown in blue for the DHX9 knockdown, whereas the control sample (shown with black lines) does not display such a joining event (threshold is set to five reads). Biological replicates were merged in these representations (also see Extended Data Fig. 2 for reproducibility between biological replicates). For clarity, only minus strand data are shown. c, DHX9 binding in the NTN5–MAMSTR gene locus on the nascent (bleed-through) RNA and splicing defect upon DHX9 knockdown. Sashimi plots depicting the exon junctions of these genes are shown in blue for the DHX9 knockdown, whereas the control sample (shown with black lines) do not display such a joining event (threshold is set to one read). Biological replicates were merged in these representations (also see Extended Data Fig. 2 for reproducibility between biological replicates). For clarity, only minus strand data are shown. d, DHX9 binding in the SALL4–ZFP64 gene locus on the nascent (bleed-through) RNA and splicing defect upon DHX9 knockdown. Sashimi plots depicting the exon junctions of these genes are shown in blue for the DHX9 knockdown whereas the control sample (shown with black lines) displays a lower frequency of such joining events (threshold is set to 1 reads). Biological replicates were merged in these representations (also see Extended Data Fig. 2 for reproducibility between biological replicates). For clarity, only minus strand data are shown.

Extended Data Figure 8 DHX9 depletion leads to a catastrophic disintegration of the nucleus.

a, CRISPR–Cas9 DHX9-knockout cell (on the right side of the image) displays a multinucleated cell phenotype in comparison to a DHX9-expressing cell (on the left side of the image). Scale bar, 10 μm. b, Three still images from Supplementary Video 1 (control) and Supplementary Video 3 (knockdown) representing the start, middle and end point of 22-h-long live imaging for control or DHX9-siRNA-treated HeLa H2B–mCherry, tubulin–GFP cells. c, Quantification of the number of nuclei at the end point of live cell imaging (~24 h imaging duration). Time points on the graph represent the start point of the imaging. All the imaged fields are taken into consideration and only the cells with more than one nucleus are included in the analysis. Statistical analysis performed with Kruskal–Wallis test shows that only the 72–96 h time point of DHX9-knockdown cells significantly differs from the rest with a P < 0.0001 (Kruskal–Wallis). Also see Supplementary Videos 2 and 7 for full field views of imaging and Supplementary Videos 5 and 6 for a second siRNA knockdown of DHX9. Right, multinucleated cells at the end point of imaging (at 96 h of DHX9 knockdown) in a different field from Supplementary Video 7. d, Left, expression levels for the Mitocheck Consortium Grape phenotype category genes (117 are expressed in the RNA-seq out of the 153 in total, mean counts per million > 0). Right, schematic view of the Ndc80 complex, red arrow points at the SPC24 protein. e, Left, experimental design of the homologous recombination of the GFP tag at the DHX9 locus is shown. Single-cell-derived heterozygously or homozygously GFP-tagged clones successfully deplete the GFP–DHX9 protein upon doxycycline induction of SPOP–GFP. Right, PCR-based screening of the knock-in allele is shown for pool of cells before the colony picking and the homozygous knock-in Alu-bypass SPC24 clone after isolation. Knock-in causes a size shift of the PCR product (from 1.2 to 1.9 kb). Correct insertion of the repair construct into the endogenous locus is further validated by Sanger sequencing (not shown).

Extended Data Figure 9 Tandem affinity purification of DHX9 from crosslinked nuclear extracts or native whole-cell extracts identifies a set of RNA-binding proteins that interact with DHX9.

a, Induction test for the stable FLPinTrex HEK293 cell line with a single-copy C-terminal 3×Flag–6×His–biotin–6×His-tagged GFP or DHX9 protein. b, Experimental set-up for the SILAC quantified formaldehyde crosslinked DHX9 interactome (either loaded on gel or eluted from beads by trypsinization) obtained by tandem affinity purification. c, Fold enrichment values of on bead digested proteins from forward and reverse SILAC experiments (forward: DHX9 cells are grown in heavy isotope labelled SILAC medium and GFP cells are grown in unlabelled medium; reverse: DHX9 cells are grown in unlabelled medium and GFP cells are grown in heavy isotope labelled SILAC medium) for those proteins highlighted in Fig. 4a. d, Experimental set-up for the affinity tag pull-downs employing native whole-cell extracts that were used for in-gel digestion LC–MS. e, Silver-stain gel of native pull-down experiments, where lanes 1 and 2 show the input from GFP and DHX9 cell lines (DHX9 cells are grown in unlabelled medium and GFP cells are grown in heavy isotope labelled SILAC medium), lanes 3 and 4 are the final eluates of His-tag pull-down followed by streptavidin pull-downs from GFP and DHX9 cell lines, and lanes 5 and 6 are RNaseA-treated final eluates. The gel is cut into 10 slices as shown and both GFP and DHX9 lanes are combined (3 and 4 together; 5 and 6 together) prior to mass-spectrometry analysis. f, Fold enrichment values in the native purifications for the proteins highlighted in Fig. 4a, showing a depletion for most of these proteins upon RNase digestion. g, Validation of DHX9 interaction with ADAR is shown in a stable cell line for ADAR that can express both p110 and p150 isoforms. The doublet is detected with a blot for Flag after DHX9 IP in this stable cell line. POL2 and CDK9 blots serve as loading controls. h, The description of the experimental set-up for IFNα induction and DHX9–ADAR co-IP in different fractions of the mouse 3T3 cell line. i, Validation of fractionation with tubulin (lanes 3, 4) and RNA–POL2 blots (lanes 2, 5) and IFNα induction of ADAR(p150) (lanes 5 and 6). j–l, Mouse DHX9 interacts with ADAR(p150) in the whole cell (j, lanes 3 and 6) and nuclear extracts (k, lanes 3 and 6) but not in the cytoplasmic extracts (l, lanes 3 and 6).

Extended Data Figure 10 DHX9 and ADAR cooperatively work on large RNA secondary structures.

a, Anti-DHX9 and J2 (dsRNA) antibody staining are shown for a mixed population of cells treated with control siRNA (DHX9 staining present, green) and DHX9 siRNA (DHX9 staining absent, for example, marked nuclei). DHX9-depleted cells accumulate dsRNA. b, The efficiency of knockdown in replicate experiments for J2 pull-down experiment displayed in Fig. 4e are shown by western blot. c, Left, analysis pipeline for detecting differential editing between control and knockdown RNA-seq libraries (both poly(A)+ and poly(A)− considered). Right, distribution of differential A-to-I editing sites detected in DHX9 knockdown is shown. After filtering for known SNPs from dbSNP and HEK293 genome, 1,244 genes with 1,807 potential A-to-I RNA editing changes in DHX9-siRNA-treated samples were detected which were absent in controls. 48% of these sites have been reported previously51. 77% of genes showing editing changes have these changes within introns. Additionally, we observed that 21% of genes with intronic editing also showed splicing defects and 28% of genes with overall editing showed misregulation (differentially expressed) upon DHX9 knockdown. These genes are a small fraction of all the genes with splicing defects (4.5%) or expression changes (6%) observed in DHX9-depleted cells. d, Effect of DHX9, ADAR or double knockdown of DHX9 and ADAR on circular RNA generation and CCL25 upregulation is shown. ADAR depletion alone does not have an appreciable impact on circular RNA generation, whereas the effect of DHX9 depletion is augmented when it is combined with ADAR depletion. e, A summary model. Alu elements are the most abundant transposable elements making up >10% of our own genomes. These elements are potentially harmful to their host and are depleted entirely from 5′ UTRs of genes and developmentally important loci, such as the HOXA–D clusters but are found in abundance in intronic and intergenic regions. We show that DHX9, a highly conserved, very abundant nuclear RNA helicase, directly interacts with Alu elements in vivo and suppresses circular RNA formation, which is probably a symptom of Alu-mediated splicing defects. We propose that our cells have developed a dependency on DHX9 to remove strong secondary structures, such as the ones originating from Alu insertions in the transcribed parts of our genome. We also show that in mice, DHX9 interacts with SINE B elements that are the murine equivalent of Alu elements, underscoring the flexibility of DHX9-mediated control of retrotransposon toxicity throughout evolution.

Supplementary information

Supplementary Figure 1

This file contains uncropped versions of all western blots presented in the manuscript. Molecular weight markers are indicated. Moreover, in which figure the blot was used in and what portion of the blot was cropped is also indicated. (PDF 4268 kb)

Supplementary Table 1

This file contains the enrichment scores (Heavy:Light and/or Light:Heavy ratios) obtained from crosslinked or native purifications of DHX9, together with peptide counts and fractions from which the peptides are recovered are presented. (XLSX 335 kb)

Supplementary Table 2

This file contains the DNA sequences of oligonucleotides used for PCR, qPCR, guide-RNA and conventional cloning and the sequences of synthesized DNA fragments used as repair-templates are presented. (XLSX 22 kb)

Supplementary Table 3

This file contains the mapping statistics and peak counts for uvCLAP/FLASH profiles and statistics regarding distances of uvCLAP/FLASH peaks to nearest Alu-pair are presented. (XLSX 21 kb)

Close-up view of normally dividing HeLa cells that are transfected with a control siRNA (still images used for Fig. 4b).

Imaging start point is 72 hours post-transfection. Cells are imaged for 22 hours by taking an image every 5 minutes. All the cells in the imaged field undergo complete mitosis in the course of the imaging experiment, indicating the lack of photo-toxicity or problems caused by siRNA transfections. Scale bar shows 20µm, time-stamp (upper left corner) shows hh:mm. Green = Tubulin-GFP, red = H2B-mCherry. (MP4 13239 kb)

Wide-field view of normally dividing HeLa cells that transfected with a control siRNA

Imaging start point is 72 hours post-transfection. Cells are imaged for 22 hours by taking an image every 5 minutes. Almost all the cells in the imaged field undergo a complete mitosis, indicating the lack of photo-toxicity or problems caused by siRNA transfections. Scale bar shows 20µm, time-stamp (upper left corner) shows hh:mm. Green = Tubulin-GFP, red = H2B-mCherry. (MP4 24275 kb)

Close-up view of HeLa cells undergoing defective mitosis upon DHX9 depletion (siRNA#1)

Imaging start point is 72 hours post-transfection. Cells are imaged for 22 hours by taking an image every 5 minutes. Three cells undergo mitosis that lasts more than 10 hours on average and all three are unable to align the chromosomes. The cell on the right exits mitosis by dividing into 4 un-equally followed by the merging of all leading to a cell with multiple and variably sized nuclei. The other two cells follow the first one soon after by exiting mitosis similarly. Scale bar shows 20µm, time-stamp (upper left corner) shows hh:mm. Green = Tubulin-GFP, red = H2B-mCherry. (MP4 11763 kb)

Close-up view of three HeLa cells undergoing a joint mitosis after fusing with each other upon DHX9 depletion with siRNA#1.

Imaging start point is 72 hours post-transfection. Cells are imaged for 22 hours by taking an image every 5 minutes. Three cells in the center merge with each other at 2 hours 30 minutes into the imaging that can be seen more clearly at 7 hours. All three cells undergo mitosis at 10 hours of the imaging (microtubule organizing centers each of the individual cells are seen inside the merged cell). The giant cell (made up of three cells merging) is unable to align properly the chromosomes and hence unable to complete the mitosis in the course of the imaging experiment. Other cells surrounding the giant cell exit mitosis with multiple nuclei and in some cases (the cell that is at the lower right corner of the giant cell) cell death can be observed before the movie ends. Scale bar shows 20µm, time-stamp (upper left corner) shows hh:mm. Green = Tubulin-GFP, red = H2B-mCherry. (MP4 9914 kb)

Close-up view of a HeLa cell undergoing defective mitosis upon DHX9 depletion (siRNA#2).

Imaging start point is 72 hours post-transfection. Cells are imaged for 22 hours by taking an image every 5 minutes. The cell in the center is already in mitosis at imaging start point and takes 10 hours to exit the mitosis without aligning the chromosomes thus eventually without dividing. Instead of complete mitosis, the cell makes many small nuclei. Surrounding cells are also seen to exit mitosis to a similar outcome. Scale bar shows 20µm, time-stamp (upper left corner) shows hh:mm. Green = Tubulin-GFP, red = H2B-mCherry. (MP4 15178 kb)

Close-up view of a HeLa cell undergoing defective mitosis upon DHX9 depletion (siRNA#2).

Imaging start point is 72 hours post-transfection. Cells are imaged for 22 hours by taking an image every 5 minutes. The cell enters mitosis and takes 10 hours to exit the mitosis without aligning the chromosomes and dividing into two multinucleated cells. One of the daughter cells dies after mitosis while the other one is alive until the completion of imaging. Scale bar shows 20µm, time-stamp (upper left corner) shows hh:mm. Green = Tubulin-GFP, red = H2B-mCherry. (MP4 11335 kb)

Wide-field view of HeLa cells undergoing defective mitosis upon DHX9 depletion (siRNA#1).

Imaging start point is 72 hours post-transfection. Cells are imaged for 22 hours by taking an image every 5 minutes. This video shows one field out of 15 imaged on the same day as images shown in Figure 4b. Cells which undergo mitosis are unable to align their chromosomes and are stuck in this phase longer than 10 hours. Most of the cells exit mitosis without a proper division and end up as multi-nucleated cells or they die. Cell death is also observed for the multi-nucleated cells which emerged from incomplete mitosis. Scale bar shows 20µm, time-stamp (upper left corner) shows hh:mm. Green = Tubulin-GFP, red = H2B-mCherry. (MP4 20374 kb)

PowerPoint slides

Rights and permissions

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Aktaş, T., Avşar Ilık, İ., Maticzka, D. et al. DHX9 suppresses RNA processing defects originating from the Alu invasion of the human genome. Nature 544, 115–119 (2017). https://doi.org/10.1038/nature21715

Download citation

  • Received:

  • Accepted:

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1038/nature21715

This article is cited by

Comments

By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.

Search

Advanced search

Quick links

Nature Briefing

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing