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.

Aprataxin resolves adenylated RNA–DNA junctions to maintain genome integrity

Abstract

Faithful maintenance and propagation of eukaryotic genomes is ensured by three-step DNA ligation reactions used by ATP-dependent DNA ligases1,2. Paradoxically, when DNA ligases encounter nicked DNA structures with abnormal DNA termini, DNA ligase catalytic activity can generate and/or exacerbate DNA damage through abortive ligation that produces chemically adducted, toxic 5′-adenylated (5′-AMP) DNA lesions3,4,5,6. Aprataxin (APTX) reverses DNA adenylation but the context for deadenylation repair is unclear. Here we examine the importance of APTX to RNase-H2-dependent excision repair (RER) of a lesion that is very frequently introduced into DNA, a ribonucleotide. We show that ligases generate adenylated 5′ ends containing a ribose characteristic of RNase H2 incision. APTX efficiently repairs adenylated RNA–DNA, and acting in an RNA–DNA damage response (RDDR), promotes cellular survival and prevents S-phase checkpoint activation in budding yeast undergoing RER. Structure–function studies of human APTX–RNA–DNA–AMP–Zn complexes define a mechanism for detecting and reversing adenylation at RNA–DNA junctions. This involves A-form RNA binding, proper protein folding and conformational changes, all of which are affected by heritable APTX mutations in ataxia with oculomotor apraxia 1. Together, these results indicate that accumulation of adenylated RNA–DNA may contribute to neurological disease.

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

Relevant articles

Open Access articles citing this article.

Access options

Buy article

Get time limited or full article access on ReadCube.

$32.00

All prices are NET prices.

Figure 1: Abortive ligation at RNA–DNA junctions is resolved by APTX.
Figure 2: Yeast Hnt3Aptx is critical for resolving abortive ligation intermediates that arise after incision at genomic ribonucleotides by RNase H2.
Figure 3: Recognition of adenylated RNA–DNA junctions by human APTX.
Figure 4: RNA–DNA deadenylation reaction mechanism and APTX inactivation in AOA1.

Accession codes

Accessions

Protein Data Bank

Data deposits

Molecular coordinates and structure factors for X-ray structures reported here have been deposited in the RCSB Protein Data Bank under accession codes 4NDF (human APTX–RNA–DNA–AMP–Zn complex), 4NDG (human APTX–RNA–DNA–adenosine–vanadate–Zn complex), 4NDH (human APTX–DNA–AMP–Zn complex) and 4NDI ( human APTX(K197Q)–RNA–DNA–AMP–Zn complex).

References

  1. Pascal, J. M., O’Brien, P. J., Tomkinson, A. E. & Ellenberger, T. Human DNA ligase I completely encircles and partially unwinds nicked DNA. Nature 432, 473–478 (2004)

    Article  ADS  CAS  Google Scholar 

  2. Ellenberger, T. & Tomkinson, A. E. Eukaryotic DNA ligases: structural and functional insights. Annu. Rev. Biochem. 77, 313–338 (2008)

    Article  CAS  Google Scholar 

  3. Tumbale, P. et al. Structure of an aprataxin-DNA complex with insights into AOA1 neurodegenerative disease. Nature Struct. Mol. Biol. 18, 1189–1195 (2011)

    Article  CAS  Google Scholar 

  4. Ahel, I. et al. The neurodegenerative disease protein aprataxin resolves abortive DNA ligation intermediates. Nature 443, 713–716 (2006)

    Article  ADS  CAS  Google Scholar 

  5. Rass, U., Ahel, I. & West, S. C. Actions of aprataxin in multiple DNA repair pathways. J. Biol. Chem. 282, 9469–9474 (2007)

    Article  CAS  Google Scholar 

  6. Harris, J. L. et al. Aprataxin, poly-ADP ribose polymerase 1 (PARP-1) and apurinic endonuclease 1 (APE1) function together to protect the genome against oxidative damage. Hum. Mol. Genet. 18, 4102–4117 (2009)

    Article  CAS  Google Scholar 

  7. El-Khamisy, S. F. et al. Synergistic decrease of DNA single-strand break repair rates in mouse neural cells lacking both Tdp1 and aprataxin. DNA Repair 8, 760–766 (2009)

    Article  CAS  Google Scholar 

  8. Daley, J. M., Wilson, T. E. & Ramotar, D. Genetic interactions between HNT3/Aprataxin and RAD27/FEN1 suggest parallel pathways for 5′ end processing during base excision repair. DNA Repair 9, 690–699 (2010)

    Article  CAS  Google Scholar 

  9. Sparks, J. L. et al. RNase H2-initiated ribonucleotide excision repair. Mol. Cell 47, 980–986 (2012)

    Article  CAS  Google Scholar 

  10. Reijns, M. A. et al. Enzymatic removal of ribonucleotides from DNA is essential for mammalian genome integrity and development. Cell 149, 1008–1022 (2012)

    Article  CAS  Google Scholar 

  11. Nick McElhinny, S. A. et al. Abundant ribonucleotide incorporation into DNA by yeast replicative polymerases. Proc. Natl Acad. Sci. USA 107, 4949–4954 (2010)

    Article  ADS  CAS  Google Scholar 

  12. Williams, J. S. et al. Topoisomerase 1-mediated removal of ribonucleotides from nascent leading-strand DNA. Mol. Cell 49, 1010–1015 (2013)

    Article  CAS  Google Scholar 

  13. Nick McElhinny, S. A. et al. Genome instability due to ribonucleotide incorporation into DNA. Nature Chem. Biol. 6, 774–781 (2010)

    Article  CAS  Google Scholar 

  14. Rumbaugh, J. A., Murante, R. S., Shi, S. & Bambara, R. A. Creation and removal of embedded ribonucleotides in chromosomal DNA during mammalian Okazaki fragment processing. J. Biol. Chem. 272, 22591–22599 (1997)

    Article  CAS  Google Scholar 

  15. Date, H. et al. Early-onset ataxia with ocular motor apraxia and hypoalbuminemia is caused by mutations in a new HIT superfamily gene. Nature Genet. 29, 184–188 (2001)

    Article  CAS  Google Scholar 

  16. Moreira, M. C. et al. The gene mutated in ataxia-ocular apraxia 1 encodes the new HIT/Zn-finger protein aprataxin. Nature Genet. 29, 189–193 (2001)

    Article  CAS  Google Scholar 

  17. Tranchant, C., Fleury, M., Moreira, M. C., Koenig, M. & Warter, J. M. Phenotypic variability of aprataxin gene mutations. Neurology 60, 868–870 (2003)

    Article  CAS  Google Scholar 

  18. Kijas, A. W., Harris, J. L., Harris, J. M. & Lavin, M. F. Aprataxin forms a discrete branch in the HIT (histidine triad) superfamily of proteins with both DNA/RNA binding and nucleotide hydrolase activities. J. Biol. Chem. 281, 13939–13948 (2006)

    Article  CAS  Google Scholar 

  19. Lujan, S. A. et al. Mismatch repair balances leading and lagging strand DNA replication fidelity. PLoS Genet. 8, e1003016 (2012)

    Article  CAS  Google Scholar 

  20. Davidson, M. B. et al. Endogenous DNA replication stress results in expansion of dNTP pools and a mutator phenotype. EMBO J. 31, 895–907 (2012)

    Article  CAS  Google Scholar 

  21. Lima, C. D., Klein, M. G. & Hendrickson, W. A. Structure-based analysis of catalysis and substrate definition in the HIT protein family. Science 278, 286–290 (1997)

    Article  CAS  Google Scholar 

  22. Rass, U., Ahel, I. & West, S. C. Molecular mechanism of DNA deadenylation by the neurological disease protein aprataxin. J. Biol. Chem. 283, 33994–34001 (2008)

    Article  CAS  Google Scholar 

  23. Nakamura, T., Zhao, Y., Yamagata, Y., Hua, Y. J. & Yang, W. Watching DNA polymerase eta make a phosphodiester bond. Nature 487, 196–201 (2012)

    Article  ADS  CAS  Google Scholar 

  24. Sykora, P., Croteau, D. L., Bohr, V. A. & Wilson, D. M., III Aprataxin localizes to mitochondria and preserves mitochondrial function. Proc. Natl Acad. Sci. USA 108, 7437–7442 (2011)

    Article  ADS  Google Scholar 

  25. Kasiviswanathan, R. & Copeland, W. C. Ribonucleotide discrimination and reverse transcription by the human mitochondrial DNA polymerase. J. Biol. Chem. 286, 31490–31500 (2011)

    Article  CAS  Google Scholar 

  26. Yang, M. Y. et al. Biased incorporation of ribonucleotides on the mitochondrial L-strand accounts for apparent strand-asymmetric DNA replication. Cell 111, 495–505 (2002)

    Article  CAS  Google Scholar 

  27. McDonald, J. P., Vaisman, A., Kuban, W., Goodman, M. F. & Woodgate, R. Mechanisms employed by Escherichia coli to prevent ribonucleotide incorporation into genomic DNA by Pol V. PLoS Genet. 8, e1003030 (2012)

    Article  CAS  Google Scholar 

  28. Nick McElhinny, S. A. & Ramsden, D. A. Polymerase mu is a DNA-directed DNA/RNA polymerase. Mol. Cell. Biol. 23, 2309–2315 (2003)

    Article  CAS  Google Scholar 

  29. Chabes, A. et al. Survival of DNA damage in yeast directly depends on increased dNTP levels allowed by relaxed feedback inhibition of ribonucleotide reductase. Cell 112, 391–401 (2003)

    Article  CAS  Google Scholar 

  30. Ferraro, P., Franzolin, E., Pontarin, G., Reichard, P. & Bianchi, V. Quantitation of cellular deoxynucleoside triphosphates. Nucleic Acids Res. 38, e85 (2010)

    Article  Google Scholar 

  31. Cotner-Gohara, E. et al. Human DNA ligase III recognizes DNA ends by dynamic switching between two DNA-bound states. Biochemistry 49, 6165–6176 (2010)

    Article  CAS  Google Scholar 

  32. Otwinowski, Z. & Minor, W. in Methods in Enzymology Vol. 276 Macromolecular Crystallography, Part A (eds Carter, C. W. & Sweets, R. M. ) 307–326 (Academic, 1997)

    Google Scholar 

  33. Pavlov, Y. I., Shcherbakova, P. V. & Kunkel, T. A. In vivo consequences of putative active site mutations in yeast DNA polymerases alpha, epsilon, delta, and zeta. Genetics 159, 47–64 (2001)

    CAS  PubMed  PubMed Central  Google Scholar 

  34. McCoy, A. J. et al. Phaser crystallographic software. J. Appl. Crystallogr. 40, 658–674 (2007)

    Article  CAS  Google Scholar 

  35. Terwilliger, T. C. Maximum-likelihood density modification. Acta Crystallogr. D 56, 965–972 (2000)

    Article  CAS  Google Scholar 

  36. Adams, P. D. et al. PHENIX: a comprehensive Python-based system for macromolecular structure solution. Acta Crystallogr. D 66, 213–221 (2010)

    Article  CAS  Google Scholar 

  37. Murshudov, G. N., Vagin, A. A. & Dodson, E. J. Refinement of macromolecular structures by the maximum-likelihood method. Acta Crystallogr. D 53, 240–255 (1997)

    Article  CAS  Google Scholar 

  38. Emsley, P., Lohkamp, B., Scott, W. G. & Cowtan, K. Features and development of Coot. Acta Crystallogr. D 66, 486–501 (2010)

    Article  CAS  Google Scholar 

Download references

Acknowledgements

This work was supported by the intramural research program of the US National Institutes of Health (NIH), National Institute of Environmental Health Sciences (NIEHS) projects 1Z01ES102765 to R.S.W. and Z01ES065070 to T.A.K. X-ray diffraction data were collected at Southeast Regional Collaborative Access Team (SER-CAT) 22-ID (or 22-BM) beamline at the Advanced Photon Source, Argonne National Laboratory. Use of the Advanced Photon Source was supported by the US Department of Energy, Office of Science, Office of Basic Energy Sciences, under Contract number W-31-109-Eng-38. We thank L. Pedersen of the NIEHS collaborative crystallography group, and the Advanced Photon Source (APS) Southeast Regional Collaborative Access Team (SER-CAT) staff for assistance with crystallographic data collection. We thank L. Pedersen and B. Wallace for critical reading of the manuscript, J. Krahn for assistance with movies, and T. Ellenberger and P. O’Brien for DNA ligase expression vectors.

Author information

Authors and Affiliations

Authors

Contributions

P.T. performed biochemical studies and crystallization. M.J.S. and R.S.W. solved and refined X-ray structures. J.S.W. performed S. cerevisiae experiments. All authors contributed to experimental design, data analysis and preparation of the manuscript.

Corresponding author

Correspondence to R. Scott Williams.

Ethics declarations

Competing interests

The authors declare no competing financial interests.

Extended data figures and tables

Extended Data Figure 1 APTX homologues efficiently process products of abortive DNA ligation at RNA–DNA junctions.

a, Abortive DNA ligation reaction time course. DNA ligation reactions for human DNA ligase I were monitored at the indicated time points. The per cent ligation was measured by quantifying the total amount of 39-nucleoitde ligation product evolved as a percentage of total unprocessed substrate (Fig. 1b, 5′-P species). The per cent adenylation was measured by quantifying the total amount of 5′-AMP product evolved as a percentage of total unprocessed substrate. Substrates used are those described in Fig. 1b. Mean ± s.d. (n = 2 technical replicates) is displayed. b, Recombinant human APTX specifically processes the adenylated product of DNA ligase I abortive ligation at an incised RNA–DNA junction. DNA ligation reactions (see Fig. 1b) on 5′-RNA substrates were stopped with addition of 5 mM EDTA at 240 min. Recombinant human APTX(165–342) catalytic domain was added after abortive ligation (lane 4). Human APTX specifically processed the ‘5′-AMPRNA–DNA’ species generated by abortive ligation at the RNA–DNA junction, but does not process the 19-nucleotide 5′-P substrate, or the 39-nucleotide ligation product, confirming the ‘5′-AMPRNA–DNA’ species is a bona fide adenylation product. c, Kinetic parameters of human APTX processing of oxidative DNA damage and RNA–DNA-junction-derived adenylated substrates. Mean ± s.e.m. (n = 2 technical replicates) is displayed. d, 5′-AMPRNA–DNA deadenylation activity is an evolutionarily conserved function of fission and budding yeast Aptx/Hnt3. Reaction conditions used were as for Fig. 1d. e, Kinetic parameters of S. cerevisiae Aptx (Hnt3) deadenylation processing. Kinetic experiments were performed as described in the Methods. Mean ± s.e.m. (n = 2 technical replicates) is displayed.

Extended Data Figure 2 Deletion of HNT3 in the pol2-M644G strain impairs growth and causes genotoxin sensitivity.

a, Tetrad analysis of the HNT3/hnt3::natMX4 diploid in the pol2-M644G strain background reveals the severe growth defect of the haploid pol2-M644G hnt3Δ mutant. The plate was scanned after 6 days growth on rich medium. b, c, Deleting HNT3 in the pol2-M644G strain impairs growth and causes genotoxin sensitivity that is reduced on deletion of RNH201. b, Serial (tenfold) dilutions of cells were plated on rich medium with or without 100 mM HU and photographed after 3 days of growth at 30 °C. c, A quantitative HU-survival assay was performed as in ref. 12. Data are displayed as the mean ± s.d. (n = 3 independent experiments) with per cent survival calculated as the percentage of surviving cells (grown on YPDA agar + 100 mM HU) compared to the untreated control. *P < 0.0002 (two-tailed t-test).

Extended Data Figure 3 Human APTX domain mapping and structure determination.

a, APTX protein domain schematic. b, Protease stable domains in human APTX were identified with limited tryptic and chymotryptic digests. Left: chymotryptic proteolysis of full-length human APTX produces clustered cut sites (C1 through C3) flanking the N terminus of the human APTX histidine triad (HIT) domain helix. Right: tryptic proteolysis of full-length human APTX produces a major cut site (T1) near the N terminus of the HIT domain. This analysis reveals a meta-stable HIT-Znf domain encompassing residues 152–342, and a chymotrypsin resilient core bounding residues 172–342. c, A comparison of the DNA 5′-deadenylation activities of the chymotryptic C2 fragment (amino acids 168–342), a smaller tryptic fragment (T1, 180–342) and two designed recombinant fragments (HIT-Znf1, 155–342, and HIT-Znf2, 165–342) shows that the HIT-Znf2 fragment, but not the smaller C2 and T1 fragments, retain robust deadenylation activity. Furthermore, the N-terminal boundary of the minimal catalytic domain in the absence of damaged DNA substrate is accessible to proteolytic digest. The HIT Znf2 (165–342) fragment was used for crystallization studies. d, Table of crystallization and assay oligonucleotides. e, The product complex crystallographic asymmetric unit contains two copies of the APTX–RNA–DNA–AMP–Zn complex. An omit (the RNA–DNA duplex and AMP were excluded from the model for electron density map calculation) σ-A weighted 1.95 Å Fo − Fc map is displayed contoured at 3.0σ overlaid upon the AMP and RNA–DNA duplex for complex 1. DNA unwinding of the terminal base pair also facilitates crystallization and formation of a one-nucleotide base pair between the two complexes. f. An omit (the RNA–DNA and AMP were excluded from the model for electron density map calculation) σ-A weighted 1.95 Å Fo − Fc map is displayed contoured at 3.5σ overlaid upon the terminal 5′-rG, showing clearly the location of the 2′-OH and the C3′-endo sugar pucker of the 5′-terminal nucleotide. g, Assembly of the APTX–RNA–DNA–adenosine–vanadate transition state mimic complex. Soaking of an adenosine RNA–DNA-bound crystal form with orthovanadate facilitates an in crystallo reaction, and formation of the APTX–RNA–DNA–adenosine–vanadate covalent complex. h, Adenosine and orthovanadate addition in the presence of a 5′-phosphorylated RNA–DNA duplex inhibits APTX activity on 5′-AMPRNA–DNA, suggesting formation of a specific APTX–RNA–DNA–adenosine–vanadate complex in solution. i. Model phased anomalous difference fourier for the transition state mimic complex is calculated from a 2.85 Å data set for the transition state mimic complex crystal collected on the NIEHS rotating anode home source (λ = 1.5418 Å). A 3σ peak marks the position of the vanadium. j. Unbiased electron density for the enzyme transition state mimic covalent complex. The σ-A weighted 2Fo − Fc electron density 2.5 Å map (contoured at 1.0σ) is calculated using a model in which the displayed atoms were not included, and before refinement.

Extended Data Figure 4 APTX structure-based sequence alignment.

The positions of AOA1 mutations are marked by red text corresponding to the missense single amino acid substitutions or nonsense (marked by red X) coding truncating mutations.

Extended Data Figure 5 Human APTX RNA–DNA interactions.

a, Structural distortions of the human APTX-bound RNA–DNA. A cartoon representation of the duplex from the human APTX–RNA–DNA–AMP–Zn reaction product complex (magenta and green) is shown superimposed on an ideal B-form DNA duplex (grey), showing distortion of the 5′-terminal nucleotides from B-form geometry. b, Electrostatic potential representation of the human APTX HIT–Znf DNA interaction interface is displayed with electropositive (blue), electronegative (red) and hydrophobic (white) surfaces. An extended positively charged surface of the Znf mediates sequence nonspecific RNA–DNA contacts. Hydrophobic base stacking stabilizes the exposed ribonucleotide base. c, Trp 167 and Tyr 195 anchor the terminal ribonucleotide (green), and envelop the adenylate lesion (yellow). An omit σ-A weighted 1.95 Å FoFc map is displayed contoured at 3.0σ overlaid upon the AMP and RNA–DNA duplex product complex. The RNA–DNA duplex and AMP were excluded from the model for electron density map calculation. d, Human APTX protein RNA–DNA and AMP lesion binding contacts are displayed schematically. e, The HIT (tan) and Znf (blue) surfaces mould a contiguous RNA–DNA damage interacting surface. f, Molecular details of the Znf structure-specific DNA damage binding interface. DNA–protein contacts are mediated by four basic side chains (Lys 276, Lys 277, His 278, Lys 314) of Znf helicies α3 and α5. Additional sugar-phosphate backbone contacts from Pro 330 and Ile 329, as well as the electropositive helix dipole of helix α6, also engage the undamaged strand. g, Zn-binding by the human APTX C2H2 zinc finger subdomain. The Zn is coordinated with tetrahedral geometry by four zinc-binding residues (Cys 319, Cys 322, His 335 and His 339).

Extended Data Figure 6 Recognition of RNA–DNA and DNA by human APTX.

a, Structural overlays of DNA-bound (blue–grey) and RNA–DNA-bound (brown) human APTX complex structures. Inset: interactions at the 5′ terminus reveal similar modes of engagement of RNA–DNA and DNA-bound substrates. The β2-β3 loop residues Tyr 195 (Y195) and Lys 197 (K197) orient the 5′ terminus for catalysis. b, Effects of β2-β3 loop mutants on APTX deadenylation activity. Tenfold dilutions of APTX mutant proteins were tested for DNA adenylation activity on 5′-AMPRNA–DNA or 5′-AMPSSB substrates. Mean ± s.d. (n = 2 technical replicates) is displayed. Fold increase of protein to reach 50% activity is relative to the wild-type human APTX for each substrate.

Extended Data Figure 7 DNA end binding by α1 bridges active site conformations to DNA end sensing status.

a, Structural overlays of three states reported in this study. b, Conformational rearrangements localize to the HIT N-terminal helix and the HIT HΦHΦH loop. HIT α1 is found in variable conformations, and movement of α1 is linked to rearrangement of the active site HΦHΦH loop. The conformation of Met 256, His 260 and His 258 is modulated by van der Waals interactions with Leu 171 and Trp 167 from α1 that together flank the HΦHΦH loop. In the disassembled active site conformer of the product complex, α1 migrates 4 Å away from HΦHΦH, with concomitant morphing of His 260 conformation into an inactive state. c, Transition state complex σ-A weighted 2.55 Å 2Fo − Fc electron density map is displayed contoured at 1.0σ for the active site. A yellow sphere, ‘V’, marks the position of the vanadium covalently bonded to His 260, the 5′-phosphate of 5′-rG and adenosine. d, Product complex (assembled active site) σ-A weighted 1.95 Å 2Fo − Fc electron density map is displayed contoured at 1.5σ for the active site. e, Product complex (disassembled active site) σ-A weighted 1.95 Å 2FoFc electron density map is displayed contoured at 1.5σ.

Extended Data Figure 8 Ataxia oculomotor apraxia (AOA1) mutations.

a, The positions of APTX mutations found in AOA1 are mapped onto the structure of human APTX. b, Omit 2Fo − Fc electron density for the K197Q variant. A solvent molecule occupies the position of the Lys 97 epsilon amino group of the wild-type protein, and Gln 197 rotates away from the active site pocket, into the protein core. c, Surface representation of the K197Q mutant (pink) overlaid upon wild-type APTX (grey). In K197Q, the active site pocket is distorted, and rearrangements in the protein core proximal to Gln 197 are also observed, and involve Leu 273.

Extended Data Table 1 Saccharomyces cerevisiae strains
Extended Data Table 2 X-ray data collection and refinement statistics

Supplementary information

Supplementary Information

This file contains a Supplementary Discussion. (PDF 171 kb)

Global hAptx conformational changes

Structural interpolations between hAptx RNA-DNA bound states reveals conformational rearrangements that link DNA end binding by the HIT domain N-terminal (mobile yellow cylinder) to the active site loop (shown in teal). Positioning of the active site nucleophile, His260 (teal with transparent surface) is modulated by HIT helix α1. RNA is shown in green, with DNA (pink) and AMP lesion (yellow). (MOV 11421 kb)

Conformational changes in the hAptx active site

Structural interpolations between hAptx RNA-DNA bound states shows how close molecular interactions between the hAptx HIT domain α1 helix (mobile yellow helical cylinder) dictates conformations of the active site HΦHΦH loop. The active site nucleophile (His260, teal with transparent surface) switches between catalytically competent and incompetent states, and the HΦHΦH loop rearrangements are tied to HIT-α1 that engages the HΦHΦH loop via close van Der Waals interactions. RNA is shown in green, with DNA (pink) and AMP lesion (yellow). (MOV 14392 kb)

PowerPoint slides

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Tumbale, P., Williams, J., Schellenberg, M. et al. Aprataxin resolves adenylated RNA–DNA junctions to maintain genome integrity. Nature 506, 111–115 (2014). https://doi.org/10.1038/nature12824

Download citation

  • Received:

  • Accepted:

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1038/nature12824

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

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