Abstract

Limited DNA end resection is the key to impaired homologous recombination in BRCA1-mutant cancer cells. Here, using a loss-of-function CRISPR screen, we identify DYNLL1 as an inhibitor of DNA end resection. The loss of DYNLL1 enables DNA end resection and restores homologous recombination in BRCA1-mutant cells, thereby inducing resistance to platinum drugs and inhibitors of poly(ADP-ribose) polymerase. Low BRCA1 expression correlates with increased chromosomal aberrations in primary ovarian carcinomas, and the junction sequences of somatic structural variants indicate diminished homologous recombination. Concurrent decreases in DYNLL1 expression in carcinomas with low BRCA1 expression reduced genomic alterations and increased homology at lesions. In cells, DYNLL1 limits nucleolytic degradation of DNA ends by associating with the DNA end-resection machinery (MRN complex, BLM helicase and DNA2 endonuclease). In vitro, DYNLL1 binds directly to MRE11 to limit its end-resection activity. Therefore, we infer that DYNLL1 is an important anti-resection factor that influences genomic stability and responses to DNA-damaging chemotherapy.

Access optionsAccess options

Rent or Buy article

Get time limited or full article access on ReadCube.

from$8.99

All prices are NET prices.

Data availability

All relevant data are included in the paper and/or its Supplementary Information.

Additional information

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

References

  1. 1.

    Bolton, K. L. et al. Association between BRCA1 and BRCA2 mutations and survival in women with invasive epithelial ovarian cancer. JAMA 307, 382–390 (2012).

  2. 2.

    Vencken, P. M. et al. Chemosensitivity and outcome of BRCA1- and BRCA2-associated ovarian cancer patients after first-line chemotherapy compared with sporadic ovarian cancer patients. Ann. Oncol. 22, 1346–1352 (2011).

  3. 3.

    Yang, D. et al. Association of BRCA1 and BRCA2 mutations with survival, chemotherapy sensitivity, and gene mutator phenotype in patients with ovarian cancer. JAMA 306, 1557–1565 (2011).

  4. 4.

    Bouwman, P. et al. 53BP1 loss rescues BRCA1 deficiency and is associated with triple-negative and BRCA-mutated breast cancers. Nat. Struct. Mol. Biol. 17, 688–695 (2010).

  5. 5.

    Bunting, S. F. et al. 53BP1 inhibits homologous recombination in Brca1-deficient cells by blocking resection of DNA breaks. Cell 141, 243–254 (2010).

  6. 6.

    Callen, E. et al. 53BP1 mediates productive and mutagenic DNA repair through distinct phosphoprotein interactions. Cell 153, 1266–1280 (2013).

  7. 7.

    Escribano-Díaz, C. et al. A cell cycle-dependent regulatory circuit composed of 53BP1-RIF1 and BRCA1-CtIP controls DNA repair pathway choice. Mol. Cell 49, 872–883 (2013).

  8. 8.

    Xu, G. et al. REV7 counteracts DNA double-strand break resection and affects PARP inhibition. Nature 521, 541–544 (2015).

  9. 9.

    Gupta, R. et al. DNA repair network analysis reveals shieldin as a key regulator of NHEJ and PARP inhibitor sensitivity. Cell 173, 972–988 (2018).

  10. 10.

    Dev, H. et al. Shieldin complex promotes DNA end-joining and counters homologous recombination in BRCA1-null cells. Nat. Cell Biol. 20, 954–965 (2018).

  11. 11.

    Noordermeer, S. M. et al. The shieldin complex mediates 53BP1-dependent DNA repair. Nature 560, 117–121 (2018).

  12. 12.

    Mirman, Z. et al. 53BP1–RIF1–shieldin counteracts DSB resection through CST- and Polα-dependent fill-in. Nature 560, 112–116 (2018).

  13. 13.

    Ghezraoui, H. et al. 53BP1 cooperation with the REV7–shieldin complex underpins DNA structure-specific NHEJ. Nature 560, 122–127 (2018).

  14. 14.

    Ray Chaudhuri, A. et al. Replication fork stability confers chemoresistance in BRCA-deficient cells. Nature 535, 382–387 (2016).

  15. 15.

    Guillemette, S. et al. Resistance to therapy in BRCA2 mutant cells due to loss of the nucleosome remodeling factor CHD4. Genes Dev. 29, 489–494 (2015).

  16. 16.

    Rondinelli, B. et al. EZH2 promotes degradation of stalled replication forks by recruiting MUS81 through histone H3 trimethylation. Nat. Cell Biol. 19, 1371–1378 (2017).

  17. 17.

    Schlacher, K. et al. Double-strand break repair-independent role for BRCA2 in blocking stalled replication fork degradation by MRE11. Cell 145, 529–542 (2011).

  18. 18.

    Lemaçon, D. et al. MRE11 and EXO1 nucleases degrade reversed forks and elicit MUS81-dependent fork rescue in BRCA2-deficient cells. Nat. Commun. 8, 860 (2017).

  19. 19.

    Shalem, O. et al. Genome-scale CRISPR–Cas9 knockout screening in human cells. Science 343, 84–87 (2014).

  20. 20.

    Doench, J. G. et al. Optimized sgRNA design to maximize activity and minimize off-target effects of CRISPR–Cas9. Nat. Biotechnol. 34, 184–191 (2016).

  21. 21.

    Planells-Cases, R. et al. Subunit composition of VRAC channels determines substrate specificity and cellular resistance to Pt-based anti-cancer drugs. EMBO J. 34, 2993–3008 (2015).

  22. 22.

    Barbar, E. Dynein light chain LC8 is a dimerization hub essential in diverse protein networks. Biochemistry 47, 503–508 (2008).

  23. 23.

    King, S. M. Dynein-independent functions of DYNLL1/LC8: redox state sensing and transcriptional control. Sci. Signal. 1, pe51 (2008).

  24. 24.

    The Cancer Genome Atlas Research Network. Integrated genomic analyses of ovarian carcinoma. Nature 474, 609–615 (2011).

  25. 25.

    Jurado, S. et al. ATM substrate Chk2-interacting Zn2+ finger (ASCIZ) is a bi-functional transcriptional activator and feedback sensor in the regulation of dynein light chain (DYNLL1) expression. J. Biol. Chem. 287, 3156–3164 (2012).

  26. 26.

    Lo, K. W. et al. The 8-kDa dynein light chain binds to p53-binding protein 1 and mediates DNA damage-induced p53 nuclear accumulation. J. Biol. Chem. 280, 8172–8179 (2005).

  27. 27.

    Iacovoni, J. S. et al. High-resolution profiling of γH2AX around DNA double strand breaks in the mammalian genome. EMBO J. 29, 1446–1457 (2010).

  28. 28.

    Zhou, Y., Caron, P., Legube, G. & Paull, T. T. Quantitation of DNA double-strand break resection intermediates in human cells. Nucleic Acids Res. 42, e19 (2014).

  29. 29.

    Cruz-García, A., López-Saavedra, A. & Huertas, P. BRCA1 accelerates CtIP-mediated DNA-end resection. Cell Rep. 9, 451–459 (2014).

  30. 30.

    Patch, A. M. et al. Whole-genome characterization of chemoresistant ovarian cancer. Nature 521, 489–494 (2015).

  31. 31.

    Rayala, S. K. et al. Functional regulation of oestrogen receptor pathway by the dynein light chain 1. EMBO Rep. 6, 538–544 (2005).

  32. 32.

    Mazouzi, A. et al. A comprehensive analysis of the dynamic response to aphidicolin-mediated replication stress uncovers targets for ATM and ATMIN. Cell Rep. 15, 893–908 (2016).

  33. 33.

    Rapali, P. et al. Directed evolution reveals the binding motif preference of the LC8/DYNLL hub protein and predicts large numbers of novel binders in the human proteome. PLoS ONE 6, e18818 (2011).

  34. 34.

    Rapali, P. et al. DYNLL/LC8: a light chain subunit of the dynein motor complex and beyond. FEBS J. 278, 2980–2996 (2011).

  35. 35.

    Puthalakath, H., Huang, D. C., O’Reilly, L. A., King, S. M. & Strasser, A. The proapoptotic activity of the Bcl-2 family member Bim is regulated by interaction with the dynein motor complex. Mol. Cell 3, 287–296 (1999).

  36. 36.

    Jaffrey, S. R. & Snyder, S. H. PIN: an associated protein inhibitor of neuronal nitric oxide synthase. Science 274, 774–777 (1996).

  37. 37.

    Dundr, M. et al. Actin-dependent intranuclear repositioning of an active gene locus in vivo. J. Cell Biol. 179, 1095–1103 (2007).

  38. 38.

    Lightcap, C. M. et al. Biochemical and structural characterization of the Pak1–LC8 interaction. J. Biol. Chem. 283, 27314–27324 (2008).

  39. 39.

    Song, C. et al. Serine 88 phosphorylation of the 8-kDa dynein light chain 1 is a molecular switch for its dimerization status and functions. J. Biol. Chem. 283, 4004–4013 (2008).

  40. 40.

    Jung, Y., Kim, H., Min, S. H., Rhee, S. G. & Jeong, W. Dynein light chain LC8 negatively regulates NF-κB through the redox-dependent interaction with IκBα. J. Biol. Chem. 283, 23863–23871 (2008).

  41. 41.

    Tkác, J. et al. HELB is a feedback inhibitor of DNA end resection. Mol. Cell 61, 405–418 (2016).

  42. 42.

    Buisson, R. et al. Cooperation of breast cancer proteins PALB2 and piccolo BRCA2 in stimulating homologous recombination. Nat. Struct. Mol. Biol. 17, 1247–1254 (2010).

  43. 43.

    Yu, Z. et al. The MRE11 GAR motif regulates DNA double-strand break processing and ATR activation. Cell Res. 22, 305–320 (2012).

  44. 44.

    Boisvert, F.-M., Déry, U., Masson, J.-Y. & Richard, S. Arginine methylation of MRE11 by PRMT1 is required for DNA damage checkpoint control. Genes Dev. 19, 671–676 (2005).

  45. 45.

    Henricksen, L. A., Umbricht, C. B. & Wold, M. S. Recombinant replication protein A expression, complex formation, and functional characterization. J. Biol. Chem. 269, 11121–11132 (1994).

  46. 46.

    Moiani, D. et al. Targeting allostery with avatars to design inhibitors assessed by cell activity: dissecting MRE11 endo- and exonuclease activities. Methods Enzymol. 601, 205–241 (2018).

  47. 47.

    Nieminuszczy, J., Schwab, R. A. & Niedzwiedz, W. The DNA fibre technique - tracking helicases at work. Methods 108, 92–98 (2016).

Download references

Acknowledgements

D.C. is supported by NIH grants R01 CA208244 and R01CA142698, a Leukemia and Lymphoma Society Scholar grant, and the Claudia Adams Barr Program in Innovative Basic Cancer Research. D.C. and P.A.K. are supported by DOD W81XWH-15-0564/OC140632. Y.J.H. is supported by an AACR-AstraZeneca Ovarian Cancer Research Fellowship (17-40-12-HE). J.-Y.M. was supported by a CIHR foundation grant.

Reviewer information

Nature thanks T. Stracker and the anonymous reviewer(s) for their contribution to the peer review of this work.

Author information

Affiliations

  1. Division of Radiation and Genome Stability, Department of Radiation Oncology, Dana-Farber Cancer Institute, Harvard Medical School, Boston, MA, USA

    • Yizhou Joseph He
    • , Khyati Meghani
    • , Chunyu Yang
    • , Jie Bian
    • , Jessica Moore
    • , Joshi Niraj
    • , Alan D. D’Andrea
    • , Pascal Drané
    •  & Dipanjan Chowdhury
  2. Genome Stability Laboratory, CHU de Québec Research Center, HDQ Pavilion, Oncology Axis, Québec City, Québec, Canada

    • Marie-Christine Caron
    • , Daryl A. Ronato
    •  & Jean-Yves Masson
  3. Department of Molecular Biology, Medical Biochemistry and Pathology, Laval University Cancer Research Center, Québec City, Québec, Canada

    • Marie-Christine Caron
    • , Daryl A. Ronato
    •  & Jean-Yves Masson
  4. Rutgers Cancer Institute of New Jersey, New Brunswick, NJ, USA

    • Anchal Sharma
    •  & Subhajyoti De
  5. Department of Medical Oncology, Dana-Farber Cancer Institute, Harvard Medical School, Boston, MA, USA

    • Alexandre Detappe
    •  & Panagiotis A. Konstantinopoulos
  6. Broad Institute of Harvard and MIT, Cambridge, MA, USA

    • John G. Doench
    • , David E. Root
    •  & Dipanjan Chowdhury
  7. LBCMCP, Centre de Biologie Integrative (CBI), CNRS, Université de Toulouse, UT3, Toulouse, France

    • Gaelle Legube
  8. Center for DNA Damage and Repair, Dana-Farber Cancer Institute, Harvard Medical School, Boston, MA, USA

    • Alan D. D’Andrea
  9. Department of Biological Chemistry & Molecular Pharmacology, Harvard Medical School, Boston, MA, USA

    • Dipanjan Chowdhury

Authors

  1. Search for Yizhou Joseph He in:

  2. Search for Khyati Meghani in:

  3. Search for Marie-Christine Caron in:

  4. Search for Chunyu Yang in:

  5. Search for Daryl A. Ronato in:

  6. Search for Jie Bian in:

  7. Search for Anchal Sharma in:

  8. Search for Jessica Moore in:

  9. Search for Joshi Niraj in:

  10. Search for Alexandre Detappe in:

  11. Search for John G. Doench in:

  12. Search for Gaelle Legube in:

  13. Search for David E. Root in:

  14. Search for Alan D. D’Andrea in:

  15. Search for Pascal Drané in:

  16. Search for Subhajyoti De in:

  17. Search for Panagiotis A. Konstantinopoulos in:

  18. Search for Jean-Yves Masson in:

  19. Search for Dipanjan Chowdhury in:

Contributions

Y.J.H. and D.C. designed the study with input from P.A.K. and A.D.D. Y.J.H. and K.M. performed most of the cell-based experiments with assistance from C.Y., J.B., J.M. and P.D. A.D. did the statistical analysis of images. M.-C.C., D.A.R. and J.N. conducted the in vitro studies under J.-Y.M.’s supervision. G.L. provided the guidance and reagents on the AsiSI system. J.G.D. and D.E.R. provided all of the reagents and the analysis of the CRISPR library. A.S. performed statistical and computational analysis of clinical data under S.D.’s guidance. D.C. wrote the manuscript with input from P.A.K. and A.D.D.

Competing interests

The authors declare no competing interests.

Corresponding author

Correspondence to Dipanjan Chowdhury.

Extended data figures and tables

  1. Extended Data Fig. 1 DYNLL1 depletion causes resistance to PARPi and cisplatin in multiple lineages.

    a, Relative guide abundance before and after olaparib and cisplatin treatment in Cov362 cells (data provided in Supplementary Tables 1 and 2). b, Comparison of ATMIN and BRCA1 alterations in ovarian cancer according to the TCGA dataset24 (316 samples) from the cBioPortal. c, Survival assay of RPE1 cells treated with olaparib (left) or cisplatin (right), after transfection with non-targeting control or DYNLL1 siRNA (siCtrl or siDYNLL1). d, Immunoblot of ATMIN and DYNLL1 from Cov362 cells with deletions of ATMIN or DYNLL1 (sgATMIN or sgDYNLL1). Tubulin was used as a loading control. eg, Survival assay of BRCA1-mutant cells UWB1.289 (e), MDA-MB-236 (f) and L56Br-C1 (g) treated with olaparib or cisplatin, and transfected with control or DYNLL1 siRNA. Data are mean ± s.e.m. from three different experiments. h, Immunoblots showing depletion of DYNLL1. Experiments were repeated independently three times with similar results.

  2. Extended Data Fig. 2 Impact of DYNLL1 on PARPi and cisplatin is specific to BRCA1-mutant cells.

    a, b, Survival assay of RPE1 (a) and HeLa (b) cells transfected with BRCA1 siRNA and treated with olaparib (left) or cisplatin (right), and co-transfected with control or DYNLL1 siRNA. c, Immunoblots showing depletion of DYNLL1 and BRCA1. Experiments were repeated independently three times with similar results. #1 and #2 are independent stable clones. d, Immunoblot of tagged DYNLL1 (G, GFP; F, Flag) in Cov362 cells after deletions of ATMIN or DYNLL1 (sgATMIN or sgDYNLL1). e, Survival assay of Cov362 DYNLL1−/− clone expressing tagged DYNLL1, treated with olaparib (left) or cisplatin (right). Data are mean ± s.e.m. from three different experiments. f, Survival assay of the indicated Cov362 clones transfected with KURAMOCHI cells and control or DYNLL1 siRNA, and treated with olaparib or cisplatin. For all panels, data are mean ± s.e.m. from three different experiments. g, PFS of ovarian carcinoma patients with BRCA2 mutation based on above or below median expression values of DYNLL1 (DYNLL1-high n = 14, DYNLL1-low n = 18; source: ovarian cancer, TCGA dataset24). Statistical significance was assessed by the one-sided Mantel–Cox test.

  3. Extended Data Fig. 3 DYNLL1 influences RAD51 foci and RPA32 foci formation in BRCA1-mutant cells.

    a, b, Immunofluorescence and quantification of RAD51 foci (a, b) and RPA32 foci (c) in wild-type and DYNLL1−/− Cov362 cells exposed to 5 Gy ionizing radiation. Staining is 6 h (RAD51) and 4 h (RPA32) after ionizing radiation. In a, n = 105; ***P < 0.0001 (control versus sgDYNLL1 #1), ***P = 0.0003 (control versus sgDYNLL1 #2), P = 0.5679 (sgDYNLL1 #1 versus sgDYNLL1 #2); two-tailed unpaired Student’s t-test. In b, wild-type and DYNLL1−/− Cov362 cells were also transfected with control and BRCA1 siRNA and immunblotting was used to confirm silencing. Data are mean ± s.e.m. from three different experiments (n = 100). In c, n = 100; ***P = 0.0002 (control versus sgDYNLL1 #1), ***P < 0.0001 (control versus sgDYNLL1 #2), P = 0.5679 (sgDYNLL1 #1 versus sgDYNLL1 #2); unpaired two-tailed Student’s t-test. d, Immunoblot of wild-type and DYNLL1−/− Cov362 cells after 10 μM olaparib treatment for 48 h with indicated antibodies. Experiments were repeated independently for three times with similar results. e, Analysis of 53BP1 foci as shown in Fig. 2b from wild-type and DYNLL1−/− Cov362 cells treated with 10 µM olaparib for 24 h. Data are mean ± s.e.m. from three different experiments (n = 102 cells of each genotype). fh, Immunofluorescence (left) and quantification (right) of 53BP1 foci (f; n = 95; P = 0.1019), RPA32 foci (g; n = 100; *P = 0.0238) and RAD51 foci (h; n = 94, P = 0.3161) in RPE1 cells transfected with control or DYNLL1 siRNA exposed to 5 Gy ionizing radiation for 1 h (53BP1, f), 4 h (RPA32, g) or 6 h (RAD51, h). Statistical analyses were by unpaired two-sided Student’s t-test. For all panels, data are mean ± s.e.m. from three different experiments.

  4. Extended Data Fig. 4 DYNLL1 regulates the DNA end-resection machinery.

    a, Samples from the PAEN-AU cohort were grouped into four categories based on combinatorial high (above median) and low (below median) expression levels of BRCA1 and DYNLL1. The frequency of somatic structural variants (left) and the frequency of intrachromosomal structural variants (deletion, duplication, insertion or intrachromosomal translocation) (right) were plotted. b, Samples from the PAEN-AU (32 samples) cohort were grouped into four categories based on combinatorial high (above median) and low (below median) expression levels of BRCA1 and DYNLL1 (left) or of BRCA1 and 53BP1 (right). The frequency of structural variants (deletions, duplication, insertions or intrachromosomal translocation) with indications of homology-directed repair (≥10-bp homology) was plotted. In all box plots, the upper whisker is 1.5 × IQR more than the third quartile, and the lower whisker is 1.5 × IQR lower than the first quartile, respectively, in which the interquartile range (IQR) is the difference between the third and the first quartile (that is, the box length). Circles denote outliers. c, Quantification of mRNA levels of the indicated genes in control and ATMIN−/− (sgATMIN) cells (n = 4). Expression levels were normalized to ACTB. Data are mean ± s.e.m. from four different experiments. *P = 0.0335 and **P = 0.0038 (control versus #1 and #2, respectively, for MRE11); *P = 0.0152 and *P = 0.0257 (control versus #1 and #2, respectively, for NBN), *P = 0.0130 and *P = 0.0203 (control versus #1 and #2, respectively, for BLM), ****P < 0.0001 and *P = 0.0179 (control versus #1 and #2, respectively, for DNA2); unpaired two-sided Student’s t-test. d, Quantification of mRNA levels of the indicated genes in control and DYNLL1−/− (sgDYNLL1) Cov362 cells (n = 6). Expression levels were normalized to ACTB. Data are mean ± s.e.m. from six different experiments. ****P < 0.0001 and ****P < 0.0001 (control versus #1 and #2, respectively, for MRE11); **P = 0.0035 and **P = 0.0007 (control versus #1 and #2, respectively, for RAD50); ****P < 0.0001 and ****P < 0.0001 (control versus #1 and #2, respectively, for NBN); ****P < 0.0001 and ****P < 0.0001 (control versus #1 and #2, respectively, for BLM); ****P = 0.0002 and ****P < 0.0001 (control versus #1 and #2, respectively, for DNA2); unpaired two-sided Student’s t-test. e, Quantification of subcellular fraction of indicated proteins (n = 3) in control and DYNLL1−/− Cov362 cells. Levels of total protein and chromatin-bound protein were normalized to H2AX levels, and levels of indicated proteins in DYNLL1−/− Cov362 cells are graphically represented relative to the control Cov362 cells. Data are mean ± s.e.m. f, Flag immunoprecipitation of Flag–DYNLL1 and immunoblot with indicated antibodies. g, Immunofluorescence and quantification of DYNLL1 and γ-H2AX foci in RPE1 cells transfected with control and DYNLL1 siRNA, 1 h after 5 Gy ionizing radiation (IR). Experiments were repeated independently three times with similar results. h, Immunoblot of DYNLL1 in RPE1 cells exposed to 5 Gy ionizing radiation and subcellular fractionation at indicated times. Experiments were repeated independently three times with similar results.

  5. Extended Data Fig. 5 Separation of the functions of DYNLL1 mutants that influence DNA end resection in vitro.

    a, Structure of DYNLL1 dimer with potentially relevant residues indicated. b, Immunoprecipitation of indicated DYNLL1 mutants with 53BP1 and MRE11. Experiments were repeated independently three times with similar results. c, Immunoblot of tagged wild-type and mutant DYNLL1 in DYNLL1−/− Cov362 cells from Fig. 4b. d, Resection products of wild-type or mutant recombinant DYNLL1 (purified proteins, left panel) with MRN–RPA–BLM–DNA2 and a 32P-labelled linear 2.7-kb dsDNA substrate. Experiments were repeated independently three times with similar results. e, GST pull-down of GST-tagged mutant DYNLL1 (Ser88Asp) incubated with purified human MRE11 or human DNA2, EXO1, BLM or the human RPA trimer (RPA70–RPA32–RPA14). Experiments were repeated independently three times with similar results. f, Recombinant wild-type DYNLL1 protein was incubated with RPA and BLM and with a 32P-labelled linear 2.7-kb dsDNA substrate to monitor DNA unwinding. Experiments were repeated independently three times with similar results.

Supplementary information

  1. Supplementary Figure 1

    This file contains the uncropped scans for immunoblots, with protein sizes indicated in kDa.

  2. Reporting Summary

  3. Supplementary Table 1

    Results of CRISPR screens for suppressors of olaparib sensitivity in BRCA1 mutant ovarian cancer cell line COV362. The data include: gene symbol, number of unique guides for the corresponding gene, ranks of each guide, DNA sequence for each corresponding guide RNA, number of most enriched guides, enrichment score, p-value, FDR, and corrected FDR (q-value). Library targeted 18,080 genes with 64,751 unique guide sequences. Statistics analysis was performed using the STARS software from the Broad Institute, Cambridge, MA. See main text and Methods for details.

  4. Supplementary Table 2

    Results of CRISPR screens for suppressors of cisplatin sensitivity in BRCA1 mutant ovarian cancer cell line COV362. The data include: gene symbol, number of unique guides for the corresponding gene, ranks of each guide, DNA sequence for each corresponding guide RNA, number of most enriched guides, enrichment score, p-value, FDR, and corrected FDR (q-value). Library targeted 18,080 genes with 64,751 unique guide sequences. Statistics analysis was performed using the STARS software from the Broad Institute. Cambridge, MA. See main text and Methods for details.

About this article

Publication history

Received

Accepted

Published

DOI

https://doi.org/10.1038/s41586-018-0670-5

Further reading

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.