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.

Dissection of DNA double-strand-break repair using novel single-molecule forceps

Abstract

Repairing DNA double-strand breaks (DSBs) by nonhomologous end joining (NHEJ) requires multiple proteins to recognize and bind DNA ends, process them for compatibility, and ligate them together. We constructed novel DNA substrates for single-molecule nanomanipulation, allowing us to mechanically detect, probe, and rupture in real-time DSB synapsis by specific human NHEJ components. DNA-PKcs and Ku allow DNA end synapsis on the 100 ms timescale, and the addition of PAXX extends this lifetime to ~2 s. Further addition of XRCC4, XLF and ligase IV results in minute-scale synapsis and leads to robust repair of both strands of the nanomanipulated DNA. The energetic contribution of the different components to synaptic stability is typically on the scale of a few kilocalories per mole. Our results define assembly rules for NHEJ machinery and unveil the importance of weak interactions, rapidly ruptured even at sub-picoNewton forces, in regulating this multicomponent chemomechanical system for genome integrity.

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.

Fig. 1: DSB repair by NHEJ proteins at single-molecule resolution.
Fig. 2: Ku, DNA-PKcs and PAXX are necessary to stabilize DNA-end synapsis.
Fig. 3: DNA synapsis on the 2-s timescale in the presence of Ku, DNA-PKcs, and PAXX (control experiments).
Fig. 4: Stabilization of DNA end synapsis by Ku, DNA-PKcs, PAXX, XLF, XRCC4, and ligase IV.
Fig. 5: Model for multivalent stabilization of DNA-end synapsis by the NHEJ machinery.

References

  1. Walker, J. R., Corpina, R. A. & Goldberg, J. Structure of the Ku heterodimer bound to DNA and its implications for double-strand break repair. Nature 412, 607–614 (2001).

    Article  PubMed  CAS  Google Scholar 

  2. Smith, G. C. & Jackson, S. P. The DNA-dependent protein kinase. Genes Dev. 13, 916–934 (1999).

    Article  PubMed  CAS  Google Scholar 

  3. Dvir, A., Stein, L. Y., Calore, B. L. & Dynan, W. S. Purification and characterization of a template-associated protein kinase that phosphorylates RNA polymerase II. J. Biol. Chem. 268, 10440–10447 (1993).

    PubMed  CAS  Google Scholar 

  4. Carter, T., Vancurová, I., Sun, I., Lou, W. & DeLeon, S. A DNA-activated protein kinase from HeLa cell nuclei. Mol. Cell. Biol. 10, 6460–6471 (1990).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  5. Sibanda, B. L., Chirgadze, D. Y. & Blundell, T. L. Crystal structure of DNA-PKcs reveals a large open-ring cradle comprised of HEAT repeats. Nature 463, 118–121 (2010).

    Article  PubMed  CAS  Google Scholar 

  6. Hammel, M. et al. Ku and DNA-dependent protein kinase dynamic conformations and assembly regulate DNA binding and the initial non-homologous end joining complex. J. Biol. Chem. 285, 1414–1423 (2010).

    Article  PubMed  CAS  Google Scholar 

  7. Jiang, W. et al. Differential phosphorylation of DNA-PKcs regulates the interplay between end-processing and end-ligation during nonhomologous end-joining. Mol. Cell 58, 172–185 (2015).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  8. Sibanda, B. L., Chirgadze, D. Y., Ascher, D. B. & Blundell, T. L. DNA-PKcs structure suggests an allosteric mechanism modulating DNA double-strand break repair. Science 355, 520–524 (2017).

    Article  PubMed  CAS  Google Scholar 

  9. Moshous, D. et al. Artemis, a novel DNA double-strand break repair/V(D)J recombination protein, is mutated in human severe combined immune deficiency. Cell 105, 177–186 (2001).

    Article  PubMed  CAS  Google Scholar 

  10. Ma, Y., Pannicke, U., Schwarz, K. & Lieber, M. R. Hairpin opening and overhang processing by an Artemis/DNA-dependent protein kinase complex in nonhomologous end joining and V(D)J recombination. Cell 108, 781–794 (2002).

    Article  PubMed  CAS  Google Scholar 

  11. Ochi, T. et al. DNA repair. PAXX, a paralog of XRCC4 and XLF, interacts with Ku to promote DNA double-strand break repair. Science 347, 185–188 (2015).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  12. Xing, M. et al. Interactome analysis identifies a new paralogue of XRCC4 in non-homologous end joining DNA repair pathway. Nat. Commun. 6, 6233 (2015).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  13. Craxton, A. et al. XLS (c9orf142) is a new component of mammalian DNA double-stranded break repair. Cell Death Differ. 22, 890–897 (2015).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  14. Tadi, S. K. et al. PAXX is an accessory c-NHEJ factor that associates with Ku70 and has overlapping functions with XLF. Cell Reports 17, 541–555 (2016).

    Article  PubMed  CAS  Google Scholar 

  15. Ropars, V. et al. Structural characterization of filaments formed by human Xrcc4-Cernunnos/XLF complex involved in nonhomologous DNA end-joining. Proc. Natl Acad. Sci. USA 108, 12663–12668 (2011).

    Article  PubMed  PubMed Central  Google Scholar 

  16. Hammel, M. et al. XRCC4 protein interactions with XRCC4-like factor (XLF) create an extended grooved scaffold for DNA ligation and double strand break repair. J. Biol. Chem. 286, 32638–32650 (2011).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  17. Wu, Q. et al. Non-homologous end-joining partners in a helical dance: structural studies of XLF-XRCC4 interactions. Biochem. Soc. Trans. 39, 1387–1392 (2011). 2, 1392.

    Article  PubMed  CAS  Google Scholar 

  18. Andres, S. N. et al. A human XRCC4-XLF complex bridges DNA. Nucleic Acids Res. 40, 1868–1878 (2012).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  19. Brouwer, I. et al. Sliding sleeves of XRCC4-XLF bridge DNA and connect fragments of broken DNA. Nature 535, 566–569 (2016).

    Article  PubMed  CAS  Google Scholar 

  20. Ahnesorg, P., Smith, P. & Jackson, S. P. XLF interacts with the XRCC4-DNA ligase IV complex to promote DNA nonhomologous end-joining. Cell 124, 301–313 (2006).

    Article  PubMed  CAS  Google Scholar 

  21. Mahaney, B. L., Hammel, M., Meek, K., Tainer, J. A. & Lees-Miller, S. P. XRCC4 and XLF form long helical protein filaments suitable for DNA end protection and alignment to facilitate DNA double strand break repair. Biochem. Cell Biol. 91, 31–41 (2013).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  22. Lieber, M. R. The mechanism of double-strand DNA break repair by the nonhomologous DNA end-joining pathway. Annu. Rev. Biochem. 79, 181–211 (2010).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  23. Liang, S. et al. Achieving selectivity in space and time with DNA double-strand-break response and repair: molecular stages and scaffolds come with strings attached. Struct. Chem. 28, 161–171 (2017).

    Article  CAS  Google Scholar 

  24. Yaneva, M., Kowalewski, T. & Lieber, M. R. Interaction of DNA-dependent protein kinase with DNA and with Ku: biochemical and atomic-force microscopy studies. EMBO J. 16, 5098–5112 (1997).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  25. West, R. B., Yaneva, M. & Lieber, M. R. Productive and nonproductive complexes of Ku and DNA-dependent protein kinase at DNA termini. Mol. Cell. Biol. 18, 5908–5920 (1998).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  26. Strick, T. R., Allemand, J. F., Bensimon, D., Bensimon, A. & Croquette, V. The elasticity of a single supercoiled DNA molecule. Science 271, 1835–1837 (1996).

    Article  PubMed  CAS  Google Scholar 

  27. Bouchiat, C. et al. Estimating the persistence length of a worm-like chain molecule from force-extension measurements. Biophys. J. 76, 409–413 (1999).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  28. Yano, K. et al. Ku recruits XLF to DNA double-strand breaks. EMBO Rep. 9, 91–96 (2008).

    Article  PubMed  CAS  Google Scholar 

  29. Grundy, G. J. et al. The Ku-binding motif is a conserved module for recruitment and stimulation of non-homologous end-joining proteins. Nat. Commun. 7, 11242 (2016).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  30. Hammel, M., Yu, Y., Fang, S., Lees-Miller, S. P. & Tainer, J. A. XLF regulates filament architecture of the XRCC4-Ligase IV complex. Structure 18, 1431–1442 (2010).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  31. Balmus, G. et al. Synthetic lethality between PAXX and XLF in mammalian development. Genes Dev. 30, 2152–2157 (2016).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  32. Liu, X., Shao, Z., Jiang, W., Lee, B. J. & Zha, S. PAXX promotes KU accumulation at DNA breaks and is essential for end-joining in XLF-deficient mice. Nat. Commun. 8, 13816 (2017).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  33. Walther, A. P., Gomes, X. V., Lao, Y., Lee, C. G. & Wold, M. S. Replication protein A interactions with DNA. 1. Functions of the DNA-binding and zinc-finger domains of the 70-kDa subunit. Biochemistry 38, 3963–3973 (1999).

    Article  PubMed  CAS  Google Scholar 

  34. Hammel, M. et al. An intrinsically disordered APLF links Ku, DNA-PKcs and XRCC4-DNA ligase IV in an extended flexible non-homologous end joining complex. J. Biol. Chem. 291, 26987–27006 (2016).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  35. Lees-Miller, S. P., Beattie, T. L. & Tainer, J. A. Noncoding RNA joins Ku and DNA-PKcs for DNA-break resistance in breast cancer. Nat. Struct. Mol. Biol. 23, 509–510 (2016).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  36. Graves, E. T. et al. A dynamic DNA-repair complex observed by correlative single-molecule nanomanipulation and fluorescence. Nat. Struct. Mol. Biol. 22, 452–457 (2015).

    Article  PubMed  CAS  Google Scholar 

  37. Bouchiat, C. & Mezard, M. Elasticity model of a supercoiled DNA molecule. Phys. Rev. Lett. 80, 1556 (1998).

    Article  CAS  Google Scholar 

  38. Hammarsten, O. & Chu, G. DNA-dependent protein kinase: DNA binding and activation in the absence of Ku. Proc. Natl Acad. Sci. USA 95, 525–530 (1998).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  39. Duboc, C., Fan, J., Graves, E. T. & Strick, T. R. Preparation of DNA substrates and functionalized glass surfaces for correlative nanomanipulation and colocalization (NanoCOSM) of single molecules. Methods Enzymol. 582, 275–296 (2017).

    Article  PubMed  CAS  Google Scholar 

  40. Peränen, J., Rikkonen, M., Hyvönen, M. & Kääriäinen, L. T7 vectors with modified T7lac promoter for expression of proteins in Escherichia coli. Anal. Biochem. 236, 371–373 (1996).

    Article  PubMed  Google Scholar 

  41. Ochi, T., Wu, Q., Chirgadze, D. Y., Grossmann, J. G., Bolanos-Garcia, V.-M. & Blundell, T. L. Structural insights into the role of domain flexibility in human DNA ligase IV. Structure 20, 1212–1222 (2012).

    Article  PubMed  CAS  Google Scholar 

  42. Li, Y. et al. Crystal structure of human XLF/Cernunnos reveals unexpected differences from XRCC4 with implications for NHEJ. EMBO J. 27, 290–300 (2008).

    Article  PubMed  CAS  Google Scholar 

  43. Murray, J. E. et al. Mutations in the NHEJ component XRCC4 cause primordial dwarfism. Am. J. Hum. Genet. 96, 412–424 (2015).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

Download references

Acknowledgements

J.L.W. is supported by a PhD scholarship from the China Scholarship Council; C.D. is supported by a PhD scholarship from the University of Paris V and the Frontieres du Vivant doctoral program. For funding and support, we thank the Ligue Nationale Contre le Cancer “Equipes Labellisées” program and PSL University (NanoRep grant), as well as the University of Paris VII, the CNRS, and the Ecole normale supérieure (to T.R.S.); the National Institute of Health (P01CA92584 to S.P.L.-M. and J.A.T. and R01GM110387 to S.E.T.); and the Wellcome Trust for support for T.O., Q.W., and S.L. (093167MA and 200814/Z/16/Z to T.L.B.). J.A.T. is partly supported by a Robert A. Welch Chemistry Chair, the Cancer Prevention and Research Institute of Texas, and the University of Texas System Science and Technology Acquisition and Retention.

Author information

Authors and Affiliations

Authors

Contributions

J.L.W., C.D., Q.W., T.O., S.L., S.E.T., S.P.L.-M., M.N., J.A.T., T.L.B., and T.R.S. designed experiments; J.L.W., C.D., Q.W., T.O., S.L., and T.R.S. prepared reagents; J.L.W., C.D., Q.W., and T.R.S. carried out experiments; J.L.W., C.D. and T.R.S. analyzed data; J.L.W.., C.D., Q.W., T.O., S.L., S.E.T., S.P.L-M., M.N., J.A.T., T.L.B. and T.R.S. wrote the paper.

Corresponding author

Correspondence to Terence R. Strick.

Ethics declarations

Competing interests

The authors declare no competing interests.

Additional information

Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Integrated Supplementary Information

Supplementary Figure 1 Experimental reagents.

(A) Denaturing PAGE gel showing results of the oligo click reaction described in the Methods. Lane order is, left to right: DBCO oligo, Azido oligo, and DBCO and Azido oligos combined, for each of the two combinations of oligos used in the construct. (B) Schematic workflow for building construct from the Click-Oligos purified as in (A) (see Methods for details). (C) SDS-PAGE analysis of purified NHEJ components. 2 μg of each protein was loaded in each lane. “M” indicates protein ladder (kDa). (D) DNA ligation assays with ligase IV-XRCC4 and its catalytically dead mutant. 200 ng of linearized pHAT4 plasmid (double digested by XhoI and NcoI to give 4-bp overhangs) was incubated with 25 nM of ligase IV-XRCC4 and ligase IV(K273A)-XRCC4 (catalytically dead mutant). Agarose gel was stained with SYBR gold. See Methods for details.

Supplementary Figure 2 Ligation of both DNA strands at blunt DNA ends by the NHEJ system and efficiency of ligation by T4 DNA ligase.

(A) (Bottom panel): DNA extension vs. time in the presence of the full complement of NHEJ components. (Middle panel): Supercoiling pattern. (Top panel): Force modulation pattern. The experiment has four phases. (1, light cyan) We calibrate the DNA extension for forces ranging from 1.4 to 0.04 pN. (2, blue) At F~0.4 pN the DNA is at first insensitive to supercoiling as the single covalent bonds joining the bridge and the main segments of the construct act as free swivels. (3, red) Force modulation reveals DNA ligation takes place shortly after injecting the full set of NHEJ components (2000s mark). (4, blue) Now at F~0.4 pN the DNA extension responds to supercoiling indicating both strands of the DNA have been ligated. (B) Time-trace of DNA extension as a function of time upon ligation and cleavage of DNA ends. Force modulation pattern is shown in red. DNA is prepared with 4 bp complementary overhangs using XmaI digest. Addition of T4 DNA ligase (black uparrow) and incubation of DNA at low force rapidly leads to the shortened DNA state which is a hallmark of repair. Addition of SmaI (red uparrow) leads to cleavage of the ligation site and restoration of the DNA extension. (C) Histogram of ∆l values is fit (red line) to a Gaussian distribution with a maximum at 166 ± 10 nm (SD, n = 57 cleavage events). (D) DNA ligation probability per traction cycle for DNA prepared with a 4-base overhang (XmaI digest, blue) or blunt ends (SmaI digest, magenta). In the XmaI-based assay 55/93 DNA molecules underwent ligation. In the SmaI-based assay only 7/84 DNA molecules underwent ligation. Error bars: S.D.

Supplementary Figure 3 DNA end synapsis in NHEJ reactions lacking one or two components or employing a mutant component.

The stated component is withheld from a “complete” reaction. Panels display (left) a time-trace, (middle) an amplitude distribution with Gaussian fit to the peak (red), and (right) a lifetime distribution for synaptic events with exponential fit (red). The number of events in the lifetime fits is equal to the number of events in the Gaussian fit and is a subset of the total number of events observed. DNA ends are phosphorylated (P + ) or unphosphorylated (P-). Errors are SEM. For (C, F) the coiled-coil domain of XRCC4 is present. (a) Reaction lacking ligase IV (P-). 46 events were observed, with likeliest amplitude 161 ± 4 nm and mean lifetime 4.4 ± 1.6 s (n = 25). (b) Reaction lacking XLF (P-). 179 events were observed, with likeliest amplitude 160 ± 2 nm and mean lifetime 3.2 ± 0.6 s (n = 74). (c) Reaction lacking XRCC4 NTD/CTD (P-). 137 events were observed, with likeliest amplitude 165 ± 2 nm and mean lifetime 2.1 ± 0.4 s (n = 80). (d) Reaction lacking ligase IV (P+). 79 events were observed, with likeliest amplitude 157 ± 2 nm and mean lifetime 3.3 ± 0.7 s (n = 47). (e) Reaction lacking XLF (P+). 135 events were observed, with likeliest amplitude 159 ± 2 nm and mean lifetime 3 ± 0.5 s (n = 78). (f) Reaction lacking XRCC4 NTD/CTD (P+). 94 events were observed, with likeliest amplitude 158 ± 2 nm and mean lifetime 2.7 ± 0.7 s (n = 49). (g) Reaction with catalytically dead ligase IV (P+). 181 events were observed, with likeliest amplitude 163 ± 2 nm and mean lifetime 34.5 ± 6.9 s (n = 82). (h) Reaction with mutant XLF (P-). 132 events were observed, with likeliest amplitude 164 ± 1 nm and mean lifetime 44 ± 12.6 s (n = 84). (i) Reaction lacking DNA-PKcs (P+). Time-trace shows no synapsis in these conditions. (j) Reaction lacking XLF and PAXX (P-). Time-trace shows no synapsis in these conditions.

Supplementary Figure 4 Effect of downstream components alone.

(A) XRCC4, XLF and ligase IV on their own do not affect the DNA scaffold in standard experimental conditions. (B, C) Low and high concentrations of XRCC4 and XLF do not affect the DNA scaffold in standard experimental conditions. (D, E) Low and high concentrations of XRCC4 and XLF do not affect the DNA scaffold in “high salt” buffer composed of 20 mM Tris-HCl pH 7.6, 150 mM KCl, 1 mM DTT, 0.05% Tween-20 and 0.5 mg/ml BSA. (F, G) Low and high concentrations of XRCC4 and XLF provoke nonspecific compaction of the DNA scaffold in “low salt” buffer composed of 20 mM Tris-HCl pH 7.6, 25 mM KCl, 1 mM DTT, 0.05% Tween-20 and 0.5 mg/ml BSA. (H, I) Low and high concentrations of XRCC4 and XLF provoke nonspecific compaction of regular DNA molecules in buffer used for (F,G). (J,K) The quaternary combination of PAXX, XRCC4, XLF and ligase IV results in non-specific compaction of the DNA scaffold in standard experimental conditions. The absence of such non-specific compaction when Ku is additionally present (i.e. when all components are present but for DNA-PKcs, see Supp. Figure 3I) underscores the role of Ku in regulating non-specific interactions between downstream components and DNA.

Supplementary Figure 5 Complete reaction lacking PAXX (A-D) or including a PAXX mutant deficient in interactions with Ku (E-G).

(A) Representative time-trace showing rupture events observed when a complete reaction is assembled in the absence of PAXX. (B) Histogram of observed rupture amplitudes with Gaussian fit to the main peak. A total of 165 events is observed. The maximum of the Gaussian is located at 160 ± 2 nm (SEM; σ = 17 nm; n = 77) (red line). (C) Histogram of synaptic event lifetimes (i.e. with an amplitude contained within three standard deviations of the Gaussian peak) is fitted by a single-exponential distribution (red line), giving a mean duration of 9 ± 1.8 s (SEM, n = 77). (D) Force-dependence of synaptic lifetime in these conditions indicates the lifetime is only very weakly force-dependent. Data were collected using a different preparation of components as in (A-C) but are the same within error. Error bars: SEM (n = 62, n = 63 and n = 53 for low, medium and high force, respectively). (E) Representative time-trace showing rupture events observed when a complete reaction is assembled using the V199A/F201A mutant of PAXX instead of wild-type PAXX. (F) Histogram of observed rupture amplitudes with Gaussian fit to the main peak. A total of 77 events is observed. The maximum of the Gaussian is located at 166 ± 2 nm (SEM; σ = 9 nm; n = 39) (red line). (G) Histogram of synaptic event lifetimes (i.e. with an amplitude contained within three standard deviations of Gaussian peak) is fitted by a single-exponential distribution (red line), giving a mean duration of 5.4 ± 1.3 s (SEM, n = 39). Experiments were carried out on DNA ends lacking phosphate groups.

Supplementary Figure 6 Detection of DNA end synapsis by DNA-PKcs and Ku.

The force was switched with 50 s period. (A) Time-trace showing rupture events obtained using 100 pM DNA-PKcs and 10 nM Ku, absent ATP. Red circles and expanded views below highlight transient synaptic events. (B) Amplitude distribution containing 101 events with (red) Gaussian fit to the peak giving the likeliest amplitude 177 ± 8 nm (SD, n = 56). (C) Lifetime distribution of synaptic events (i.e. with an amplitude contained within three standard deviations of the Gaussian peak) follows single-exponential statistics (red line) with mean lifetime 0.1 ± 0.02 s (SEM, n = 56). (D) Time-trace showing rupture events obtained in the presence of 10 nM Ku, absent ATP. Red circles highlight transient, non-specific events. (E) Amplitude distribution containing 133 events with (black) Gaussian fit to the main peak (likeliest amplitude 98 ± 28 nm SD, n = 93) and (red) Gaussian fit to a minor peak (likeliest amplitude 166 ± 6 nm SD, n = 18). Events in the main peak likely correspond to non-specific (tip-scaffold or scaffold-scaffold) interactions; events in the minor peak appear to take place specifically between the DNA ends. Identical scaling to panel B permits direct comparison of results to those obtained when Ku and DNA-PKcs are both present. (F) Single-exponential fit to the lifetime distribution of non-specific events (black line) gives a mean lifetime of 1.2 ± 0.2 s (SEM, n = 93); single-exponential fit to the lifetime distribution of specific events (red line) gives a mean lifetime of 0.7 ± 0.3 s (SEM, n = 18). We conclude that interactions based on Ku alone are distinct from those based on Ku + DNA-PKcs and are for the most part non-specific in nature.

Supplementary information

Supplementary Text and Figures

Supplementary Figures 1–6 and Supplementary Tables 1 and 2

Reporting Summary

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Wang, J.L., Duboc, C., Wu, Q. et al. Dissection of DNA double-strand-break repair using novel single-molecule forceps. Nat Struct Mol Biol 25, 482–487 (2018). https://doi.org/10.1038/s41594-018-0065-1

Download citation

  • Received:

  • Accepted:

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1038/s41594-018-0065-1

This article is cited by

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