Pore-forming activity and structural autoinhibition of the gasdermin family (original) (raw)

Accession codes

Primary accessions

Protein Data Bank

Data deposits

The atomic coordinates and structure factors of GSDMA3 have been deposited in the Protein Data Bank under the accession code 5B5R.

Change history

An error in the Reviewer Information section was corrected.

An Erratum to this paper has been published: https://doi.org/10.1038/nature20106

References

  1. Jorgensen, I. & Miao, E. A. Pyroptotic cell death defends against intracellular pathogens. Immunol. Rev. 265, 130–142 (2015)
    Article CAS PubMed PubMed Central Google Scholar
  2. Cookson, B. T. & Brennan, M. A. Pro-inflammatory programmed cell death. Trends Microbiol. 9, 113–114 (2001)
    Article CAS PubMed Google Scholar
  3. Shi, J. et al. Inflammatory caspases are innate immune receptors for intracellular LPS. Nature 514, 187–192 (2014)
    Article CAS PubMed ADS Google Scholar
  4. Kayagaki, N. et al. Non-canonical inflammasome activation targets caspase-11. Nature 479, 117–121 (2011)
    Article CAS PubMed ADS Google Scholar
  5. Lamkanfi, M. & Dixit, V. M. Mechanisms and functions of inflammasomes. Cell 157, 1013–1022 (2014)
    Article CAS PubMed Google Scholar
  6. Zhao, Y. & Shao, F. Diverse mechanisms for inflammasome sensing of cytosolic bacteria and bacterial virulence. Curr. Opin. Microbiol. 29, 37–42 (2016)
    Article PubMed Google Scholar
  7. Hagar, J. A., Powell, D. A., Aachoui, Y., Ernst, R. K. & Miao, E. A. Cytoplasmic LPS activates caspase-11: implications in TLR4-independent endotoxic shock. Science 341, 1250–1253 (2013)
    Article CAS PubMed PubMed Central ADS Google Scholar
  8. Kayagaki, N. et al. Noncanonical inflammasome activation by intracellular LPS independent of TLR4. Science 341, 1246–1249 (2013)
    Article CAS PubMed ADS Google Scholar
  9. Yang, J., Zhao, Y. & Shao, F. Non-canonical activation of inflammatory caspases by cytosolic LPS in innate immunity. Curr. Opin. Immunol. 32, 78–83 (2015)
    Article PubMed CAS Google Scholar
  10. Shi, J. et al. Cleavage of GSDMD by inflammatory caspases determines pyroptotic cell death. Nature 526, 660–665 (2015)
    Article CAS PubMed ADS Google Scholar
  11. Kayagaki, N. et al. Caspase-11 cleaves gasdermin D for non-canonical inflammasome signalling. Nature 526, 666–671 (2015)
    Article CAS PubMed ADS Google Scholar
  12. Saeki, N. & Sasaki, H. in Endothelium and Epithelium: Composition, Functions, and Pathology (eds Carrasco, J. & Matheus, M. ) Ch. IX, 193–211 (Nova Science Publishers, 2011)
    Google Scholar
  13. Tanaka, S., Mizushina, Y., Kato, Y., Tamura, M. & Shiroishi, T. Functional conservation of Gsdma cluster genes specifically duplicated in the mouse genome. G3 (Bethesda) 3, 1843–1850 (2013)
    Article CAS Google Scholar
  14. Sato, H. et al. A new mutation Rim3 resembling Re den is mapped close to retinoic acid receptor alpha (Rara) gene on mouse chromosome 11. Mamm. Genome 9, 20–25 (1998)
    Article CAS PubMed Google Scholar
  15. Porter, R. M. et al. Defolliculated (dfl): a dominant mouse mutation leading to poor sebaceous gland differentiation and total elimination of pelage follicles. J. Invest. Dermatol. 119, 32–37 (2002)
    Article CAS PubMed Google Scholar
  16. Runkel, F. et al. The dominant alopecia phenotypes Bareskin, Rex-denuded, and Reduced Coat 2 are caused by mutations in gasdermin 3. Genomics 84, 824–835 (2004)
    Article CAS PubMed Google Scholar
  17. Van Laer, L. et al. Nonsyndromic hearing impairment is associated with a mutation in DFNA5. Nat. Genet. 20, 194–197 (1998)
    Article CAS PubMed Google Scholar
  18. Delmaghani, S. et al. Mutations in the gene encoding pejvakin, a newly identified protein of the afferent auditory pathway, cause DFNB59 auditory neuropathy. Nat. Genet. 38, 770–778 (2006)
    Article CAS PubMed Google Scholar
  19. Shepard, L. A., Shatursky, O., Johnson, A. E. & Tweten, R. K. The mechanism of pore assembly for a cholesterol-dependent cytolysin: formation of a large prepore complex precedes the insertion of the transmembrane beta-hairpins. Biochemistry 39, 10284–10293 (2000)
    Article CAS PubMed Google Scholar
  20. Fink, S. L., Bergsbaken, T. & Cookson, B. T. Anthrax lethal toxin and Salmonella elicit the common cell death pathway of caspase-1-dependent pyroptosis via distinct mechanisms. Proc. Natl Acad. Sci. USA 105, 4312–4317 (2008)
    Article CAS PubMed PubMed Central ADS Google Scholar
  21. Holm, L. & Rosenström, P. Dali server: conservation mapping in 3D. Nucleic Acids Res. 38, W545–W549 (2010)
    Article CAS PubMed PubMed Central Google Scholar
  22. Delmaghani, S. et al. Hypervulnerability to sound exposure through impaired adaptive proliferation of peroxisomes. Cell 163, 894–906 (2015)
    Article CAS PubMed Google Scholar
  23. Van Rossom, S., Op de Beeck, K., Hristovska, V., Winderickx, J. & Van Camp, G. The deafness gene DFNA5 induces programmed cell death through mitochondria and MAPK-related pathways. Front. Cell. Neurosci. 9, 231 (2015)
    Article PubMed PubMed Central CAS Google Scholar
  24. Bischofberger, M., Iacovache, I. & van der Goot, F. G. Pathogenic pore-forming proteins: function and host response. Cell Host Microbe 12, 266–275 (2012)
    Article CAS PubMed Google Scholar
  25. Hildebrand, J. M. et al. Activation of the pseudokinase MLKL unleashes the four-helix bundle domain to induce membrane localization and necroptotic cell death. Proc. Natl Acad. Sci. USA 111, 15072–15077 (2014)
    Article CAS PubMed PubMed Central ADS Google Scholar
  26. Dondelinger, Y. et al. MLKL compromises plasma membrane integrity by binding to phosphatidylinositol phosphates. Cell Reports 7, 971–981 (2014)
    Article CAS PubMed Google Scholar
  27. Wang, H. et al. Mixed lineage kinase domain-like protein MLKL causes necrotic membrane disruption upon phosphorylation by RIP3. Mol. Cell 54, 133–146 (2014)
    Article CAS PubMed Google Scholar
  28. Cai, Z. et al. Plasma membrane translocation of trimerized MLKL protein is required for TNF-induced necroptosis. Nat. Cell Biol. 16, 55–65 (2014)
    Article CAS PubMed Google Scholar
  29. Iacovache, I., Bischofberger, M. & van der Goot, F. G. Structure and assembly of pore-forming proteins. Curr. Opin. Struct. Biol. 20, 241–246 (2010)
    Article CAS PubMed Google Scholar
  30. Reboul, C. F., Whisstock, J. C. & Dunstone, M. A. Giant MACPF/CDC pore forming toxins: A class of their own. Biochim. Biophys. Acta 1858, 475–486 (2016)
    Article CAS PubMed Google Scholar
  31. Zhao, Y. et al. The NLRC4 inflammasome receptors for bacterial flagellin and type III secretion apparatus. Nature 477, 596–600 (2011)
    Article CAS PubMed ADS Google Scholar
  32. Gong, Y.-N. et al. Chemical probing reveals insights into the signaling mechanism of inflammasome activation. Cell Res. 20, 1289–1305 (2010)
    Article CAS PubMed Google Scholar
  33. Wilschut, J. & Papahadjopoulos, D. Ca2+-induced fusion of phospholipid vesicles monitored by mixing of aqueous contents. Nature 281, 690–692 (1979)
    Article CAS PubMed ADS Google Scholar
  34. Tang, G. et al. EMAN2: an extensible image processing suite for electron microscopy. J. Struct. Biol. 157, 38–46 (2007)
    Article CAS PubMed Google Scholar
  35. Scheres, S. H. A Bayesian view on cryo-EM structure determination. J. Mol. Biol. 415, 406–418 (2012)
    Article CAS PubMed PubMed Central Google Scholar
  36. Frank, J. et al. SPIDER and WEB: processing and visualization of images in 3D electron microscopy and related fields. J. Struct. Biol. 116, 190–199 (1996)
    Article CAS PubMed Google Scholar
  37. Otwinowski, Z. & Minor, W. in Methods in Enzymology Vol. 276 (eds Carter, C. W. Jr & Sweet, R. M. ) 307–326 (Academic, 1997)
    Article Google Scholar
  38. Adams, P. D. et al. PHENIX: a comprehensive Python-based system for macromolecular structure solution. Acta Crystallogr. D 66, 213–221 (2010)
    Article CAS PubMed PubMed Central Google Scholar
  39. Emsley, P., Lohkamp, B., Scott, W. G. & Cowtan, K. Features and development of Coot. Acta Crystallogr. D 66, 486–501 (2010)
    Article CAS PubMed PubMed Central Google Scholar
  40. Chen, V. B. et al. MolProbity: all-atom structure validation for macromolecular crystallography. Acta Crystallogr. D 66, 12–21 (2010)
    Article CAS PubMed Google Scholar
  41. Eswar, N. et al. Comparative protein structure modeling using MODELLER. Curr. Protoc. Protein Sci. Chapter 2, Unit 2.9 (2007)
  42. Shen, M. Y. & Sali, A. Statistical potential for assessment and prediction of protein structures. Protein Sci. 15, 2507–2524 (2006)
    Article CAS PubMed PubMed Central Google Scholar

Download references

Acknowledgements

We thank W. Wei for reagents, H. Wang for suggestions on electron microscopy data analysis, and the staff of beamlines BL18U1 and BL19U1 at National Center for Protein Sciences, Shanghai, and Shanghai Synchrotron Radiation Facility for X-ray data collection. This work was supported by grants from the Strategic Priority Research Program of the Chinese Academy of Sciences (XDB08020202), the China National Science Foundation Program for Distinguished Young Scholars (31225002) and Program for International Collaborations (31461143006), and the National Basic Research Program of China 973 Program (2012CB518700 and 2014CB849602) to F.S. The research was also supported in part by an International Early Career Scientist grant from the Howard Hughes Medical Institute and the Beijing Scholar Program to F.S.

Author information

Authors and Affiliations

  1. National Laboratory of Biomacromolecules, Institute of Biophysics, Chinese Academy of Sciences, 100101, Beijing, China
    Jingjin Ding, Yang She, Da-Cheng Wang & Feng Shao
  2. National Institute of Biological Sciences, Beijing, 102206, China
    Jingjin Ding, Kun Wang, Wang Liu, Yang She, Qi Sun, Jianjin Shi, Hanzi Sun & Feng Shao
  3. Foshan University, Guangdong, 528000, China
    Da-Cheng Wang
  4. National Institute of Biological Sciences, Collaborative Innovation Center for Cancer Medicine, Beijing, 102206, China,
    Feng Shao

Authors

  1. Jingjin Ding
    You can also search for this author inPubMed Google Scholar
  2. Kun Wang
    You can also search for this author inPubMed Google Scholar
  3. Wang Liu
    You can also search for this author inPubMed Google Scholar
  4. Yang She
    You can also search for this author inPubMed Google Scholar
  5. Qi Sun
    You can also search for this author inPubMed Google Scholar
  6. Jianjin Shi
    You can also search for this author inPubMed Google Scholar
  7. Hanzi Sun
    You can also search for this author inPubMed Google Scholar
  8. Da-Cheng Wang
    You can also search for this author inPubMed Google Scholar
  9. Feng Shao
    You can also search for this author inPubMed Google Scholar

Contributions

J.D., D.-C.W. and F.S. conceived the study; J.D., together with K.W., designed and performed the majority of the experiments; Y.S. helped with protein purification; W.L. performed the pyroptosis assay; Q.S. assisted J.D. in electron microscopy studies; J.S. provided critical reagents and suggestions; H.S. performed structural modelling; and J.D. and F.S. analysed the data and wrote the manuscript. All authors discussed the results and commented on the manuscript.

Corresponding authors

Correspondence toDa-Cheng Wang or Feng Shao.

Ethics declarations

Competing interests

The authors declare no competing financial interests.

Additional information

Reviewer Information Nature thanks F. Sigworth and the other anonymous reviewer(s) for their contribution to the peer review of this work.

Extended data figures and tables

Extended Data Figure 1 Multiple gasdermin-N domains can induce mammalian cell pyroptosis and also exhibit cytotoxicity in bacteria.

a, b, Full-length (FL) or N-terminal domain regions of different gasdermin-family members were transfected into 293T cells for 20 h. Human GSDMD and mouse GSDMA3 had an N-terminal 3 × Flag tag and human GSDMA, GSDMB, GSDMC and DFNA5 had a C-terminal Flag tag. ATP-based cell viability is expressed as mean ± s.d. from three technical replicates (a). Representative views of cell death morphology are shown in b. c, d, Cytotoxicity of the gasdermin-N domain in bacteria. Indicated gasdermins were cloned into an IPTG-inducible vector for transformation into E. coli. c, Representative agar plates showing transformed E. coli colonies for GSDMD. d, Bacterial colony-forming units (CFU) per transformation for GSDMD and other gasdermins are shown in the logarithmic form (log10) as mean ± s.d. from three technical replicates. All data shown are representative of three independent experiments.

Extended Data Figure 2 Membrane phospholipid binding of the gasdermin-N domain.

a–d, f, Liposomes with indicated lipid compositions (a–d) or prepared using bovine liver or brain-derived polar lipid extracts (f) were incubated with purified gasdermin proteins. After ultracentrifugation, the liposome-free supernatant (S) and liposome pellet (P) were analysed by SDS–PAGE and Coomassie blue staining. e, Noncovalent complex of cleaved GSDMD and GSDMA3 with a Flag tag attached to the end of the gasdermin-N domain or the corresponding uncleaved full-length proteins were incubated with the lipid strips, and the strips were then probed with the anti-Flag antibody. Right, protein loading control. All data shown are representative of three independent experiments.

Extended Data Figure 3 Biomembrane association and lysis by the gasdermin-N domain.

a, b, Subcellular fractionation of the gasdermin-N domains of GSDMD and GSDMA during pyroptosis. Gsdmd_−/− iBMDMs expressing 2 × Flag and haemagglutinin (HA)-tagged GSDMD were untreated or stimulated with LPS electroporation (a). 293T cells expressing PPase-cleavable Flag–GSDMA were untreated or electroporated with purified PPase (b). Homogenized cell extracts were sequentially centrifuged at 700_g, 20,000_g_ and 100,000_g_ to separate membrane fractions (P7, P20 and P100) from the S100 soluble fraction. The factions were immunoblotted as indicated. c, Microscopy of GSDMA3-N domain localization in cells undergoing pyroptosis. The gasdermin-N domain of GSDMA3 (GSDMA3-N(L184D)) fused N-terminally to eGFP was stably expressed in HeLa cells under a tetracycline-inducible promoter. Shown are representative time-lapse cell images (brightfield and fluorescence) taken from 4–5 h after doxycycline addition. Scale bar, 15 μm. For videos of two representative cells, see Supplementary Videos 3 and 4. d, e, Effects of extracellular or intracellular delivery of purified gasdermin proteins on 293T cell viability. Equal amounts of indicated gasdermin proteins or PFO were added directly into cell culture medium (d) or electroporated into the cytosol (e). ATP-based cell viability is expressed as mean ± s.d. from three technical replicates. f, g, Bacterial protoplast lysis by purified gasdermin proteins. Protoplasts of B. megaterium were treated with indicated gasdermin proteins or PFO. Membrane lysis was assessed by measuring the OD600 of the protoplasts. Triton-X 100 treatment was used to achieve 100% lysis of the protoplasts. Time-course measurement of GSDMD treatment is shown in f. Relative protoplast lysis by GSDMD and other gasdermins is expressed as mean ± s.d. from three technical replicates (g). All data shown are representative of three independent experiments.

Extended Data Figure 4 Liposome-leakage-inducing activity of the gasdermin-N domain.

a–c, Liposomes with indicated lipid compositions were treated with purified gasdermin proteins or PFO as indicated. Liposome leakage was monitored by measuring DPA chelating-induced fluorescence of released Tb3+. Time course of relative Tb3+ release is shown. A dose titration of GSDMA proteins is shown in b. CTL, control. All data shown are representative of three independent experiments.

Extended Data Figure 5 Membrane binding-induced oligomerization of and pore formation by the gasdermin-N domain.

a, Gel filtration chromatography of full-length GSDMD, GSDMA and GSDMA3. Indicated gasdermin proteins were loaded on the Superdex G75 column. Arrows indicate elution volume of the molecular mass markers. b, Oligomerization of gasdermin-N domain on the liposome membrane. Indicated gasdermin proteins or PFO were incubated with cardiolipin or cholesterol liposomes, respectively. Intact proteins or proteins associated with the liposomes were mock treated or treated with glutaraldehyde and analysed by SDS-agarose gel electrophoresis and Coomassie blue staining. The gasdermin-C domain migrating at the bottom of the gel was omitted for clarity. c, Oligomerization of the gasdermin-N domain during pyroptosis. To trigger pyroptosis, _Gsdmd_−/− iBMDMs expressing 2 × Flag-HA–GSDMD and HeLa cells expressing the PPase-cleavable Flag–GSDMA were electroporated with LPS and PPase, respectively. The cytosol (S) and membrane (P) fractions from unstimulated and pyroptotic cells were subjected to glutaraldehyde-mediated crosslinking followed by SDS-agarose (top) or SDS–PAGE (bottom) gel electrophoresis. d, Pore-forming activity of the gasdermin-N domain. Liposomes with 80% phosphatidylcholine and 20% PtdIns(4, 5)P2 were treated with indicated gasdermin proteins. Shown are representative negative-stain electron microscopy micrographs of the liposomes (scale bar, 100 nm). All data shown are representative of three independent experiments.

Extended Data Figure 6 Analyses of the gasdermin pore.

a, b, Size distribution of GSDMD and GSDMA3 pores formed on cardiolipin liposomes. The inner diameters of pores were measured and plotted. A total of 200 or 100 pores for 5 μM or 0.5 μM GSDMD/GSDMA3-treated liposome samples, respectively, were randomly selected from the negative-stain electron microscopy micrographs in Fig. 3a. c, Effects of different PEG molecules on lactate dehydrogenase (LDH) release from caspase-1-mediated pyroptotic cells. iBMDMs harbouring a sensitive Nlrp1b allele were treated with indicated mass concentration of different PEG molecules and then stimulated with anthrax lethal toxin or LFn-BsaK to activate the canonical NLRP1B or NAIP2/NLRC4 inflammasomes, respectively. 1.2% PEG200 and 12% PEG2000 (mass concentration) have roughly the same molar concentration. Shown are LDH release expressed as mean ± s.d. from three technical replicates. d, Pores formed by active GSDMD and GSDMA3 on monolayer membranes containing 80% phosphatidylcholine and 20% cardiolipin. Shown are representative negative-stain electron microscopy micrograph images (scale bar, 100 nm). e, Symmetry determination of the gasdermin pore. GSDMA3 pores formed on the monolayer membrane (d) were subjected to 2D reference-free classification. One class of pores with the best particle contrast were subjected to rotational auto-correlation calculation and the inlet electron microscopy image (scale bar, 20 nm) shows the averaged view of the class of pores (242 particles). The analyses revealed 16-fold symmetry. Data shown in a–d are representative of three independent experiments.

Extended Data Figure 7 Crystal structure of GSDMA3 and Dali search results for its gasdermin-N domain.

a, 2_F_o − F_c electron density map (contoured at 1.0_σ) of GSDMA3 gasdermin-N domain (GSDMA3-N) structure. b, Cartoon diagram of GSDMA3-N structure. c, Structural model of GSDMD obtained from homology modelling and the conserved autoinhibitory interactions. Bottom, overall structure of modelled GSDMD; top, comparisons of the hydrophobic core (left) and the second inter-domain contact (right) with the corresponding structures in GSDMA3. Conserved residues involved in the autoinhibitory interactions are labelled and shown as sticks. Cyan, GSDMD-N; orange, GSDMD-C; green, GSDMA3-N; yellow, GSDMA3-C. d, Dali search results for the GSDMA3-N structure.

Extended Data Figure 8 Multiple sequence alignment of gasdermin family members.

GSDMA3 is a mouse protein and sequences of other gasdermins are from human. The secondary structures determined from GSDMA3 are marked along the sequence. The alignment was performed with the ClustalW2 algorithm with structure-based manual adjustment of the α4 region in GSDMA3 and GSDMD. Identical residues are highlighted by dark red background and conserved residues are coloured red. The residues involved in the autoinhibitory interactions are marked underneath the sequences with blue triangle for polar residues, orange rhombus for hydrophobic residues and black rhombus for hydrophobic residues in the second inter-domain interface. The residue number is indicated on the left of the sequence.

Extended Data Figure 9 Mutations in GSDMA3-N affecting lipid binding and pore formation also reduce pyroptosis.

a, Effects of L184D/E14K mutations on pyroptosis-inducing activity of GSDMA3-N (residues 1–284). Full-length GSDMA3 or its gasdermin-N domain (wild type or indicated mutants) was transfected into 293T cells. ATP-based cell viability is expressed as mean ± s.d. from three technical replicates. b, c, Effects of L184D/E14K mutations on lipid-binding and liposome-leakage-inducing activities of GSDMA3-N domain. Liposomes containing 80% phosphatidylcholine and 20% phosphatidylethanolamine, cardiolipin, phosphatidylinositol or PtdIns(4,5)P2 were treated with purified GSDMA3. After ultracentrifugation, the liposome-free supernatant (S) and the liposome pellet (P) were analysed by SDS–PAGE (b). Liposome leakage was monitored by measuring DPA chelating-induced fluorescence of released Tb3+ (c). Triton-X 100 treatment was used to achieve 100% leakage. d, Effects of L184D/E14K mutations on pore formation by GSDMA3-N. Representative electron microscopy images of the pores on the cardiolipin liposome are shown (scale bar, 100 nm). All data shown are representative of three independent experiments.

Extended Data Table 1 Data collection refinement statistics

Full size table

Supplementary information

Supplementary Information

This file contains the uncropped immunoblot images presented in figure 5 and the Extended Data Figures. (PDF 1392 kb)

Membrane targeting of the Gasdermin-N domain of GSDMD during pyroptosis

The Gasdermin-N domain of GSDMD (GSDMD-N L192D) fused N-terminal to EGFP was stably expressed in HeLa cells under a tetracycline-inducible promoter. Pyroptosis was triggered by doxycycline addition to induce GSDMD-N L192D-EGFP expression. Shown are real-time videos of two representative pyroptotic cells. Scale bar, 15 μm. The time-lapse images for Video 1 are in Fig. 1c. (MP4 9991 kb)

Membrane targeting of the Gasdermin-N domain of GSDMD during pyroptosis

The Gasdermin-N domain of GSDMD (GSDMD-N L192D) fused N-terminal to EGFP was stably expressed in HeLa cells under a tetracycline-inducible promoter. Pyroptosis was triggered by doxycycline addition to induce GSDMD-N L192D-EGFP expression. Shown are real-time videos of two representative pyroptotic cells. Scale bar, 15 μm. (MP4 9764 kb)

Membrane targeting of the Gasdermin-N domain of GSDMA during pyroptosis.

The Gasdermin-N domain of GSDMA3 (GSDMA3-N L184D) fused N-terminal to EGFP was stably expressed in HeLa cells under a tetracycline-inducible promoter. Pyroptosis was triggered by doxycycline addition to induce GSDMA3-N L184D-EGFP expression. Shown are real-time videos of two representative pyroptotic cells. Scale bar, 15 μm. The time-lapse images for Video 3 are in Extended Data Fig. 3c. (MP4 9412 kb)

Membrane targeting of the Gasdermin-N domain of GSDMA during pyroptosis.

The Gasdermin-N domain of GSDMA3 (GSDMA3-N L184D) fused N-terminal to EGFP was stably expressed in HeLa cells under a tetracycline-inducible promoter. Pyroptosis was triggered by doxycycline addition to induce GSDMA3-N L184D-EGFP expression. Shown are real-time videos of two representative pyroptotic cells. Scale bar, 15 μm. (MP4 9569 kb)

PowerPoint slides

Rights and permissions

About this article

Cite this article

Ding, J., Wang, K., Liu, W. et al. Pore-forming activity and structural autoinhibition of the gasdermin family.Nature 535, 111–116 (2016). https://doi.org/10.1038/nature18590

Download citation