Letter | Published:

Parasitic helminths induce fetal-like reversion in the intestinal stem cell niche

Naturevolume 559pages109113 (2018) | Download Citation

Abstract

Epithelial surfaces form critical barriers to the outside world and are continuously renewed by adult stem cells1. Whereas dynamics of epithelial stem cells during homeostasis are increasingly well understood, how stem cells are redirected from a tissue-maintenance program to initiate repair after injury remains unclear. Here we examined infection by Heligmosomoides polygyrus, a co-evolved pathosymbiont of mice, to assess the epithelial response to disruption of the mucosal barrier. H. polygyrus disrupts tissue integrity by penetrating the duodenal mucosa, where it develops while surrounded by a multicellular granulomatous infiltrate2. Crypts overlying larvae-associated granulomas did not express intestinal stem cell markers, including Lgr53, in spite of continued epithelial proliferation. Granuloma-associated Lgr5 crypt epithelium activated an interferon-gamma (IFN-γ)-dependent transcriptional program, highlighted by Sca-1 expression, and IFN-γ-producing immune cells were found in granulomas. A similar epithelial response accompanied systemic activation of immune cells, intestinal irradiation, or ablation of Lgr5+ intestinal stem cells. When cultured in vitro, granuloma-associated crypt cells formed spheroids similar to those formed by fetal epithelium, and a sub-population of H. polygyrus-induced cells activated a fetal-like transcriptional program, demonstrating that adult intestinal tissues can repurpose aspects of fetal development. Therefore, re-initiation of the developmental program represents a fundamental mechanism by which the intestinal crypt can remodel itself to sustain function after injury.

Access optionsAccess options

Rent or Buy article

Get time limited or full article access on ReadCube.

from $8.99

All prices are NET prices.

Additional information

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

Change history

  • Correction 16 July 2018

    In this Letter, the received date should have been 23 March 2017 instead of 13 April 2018. Authors R.M.K. and O.D.K. were incorrectly denoted as ‘equally contributing’ authors. The labels for ‘control’ and ‘IFNγ’ in Extended Data Fig. 4g were reversed. These have been corrected online.

References

  1. 1.

    Karin, M. & Clevers, H. Reparative inflammation takes charge of tissue regeneration. Nature 529, 307–315 (2016).

  2. 2.

    Maizels, R. M. et al. Immune modulation and modulators in Heligmosomoides polygyrus infection. Exp. Parasitol. 132, 76–89 (2012).

  3. 3.

    Barker, N. et al. Identification of stem cells in small intestine and colon by marker gene Lgr5. Nature 449, 1003–1007 (2007).

  4. 4.

    Tian, H. et al. A reserve stem cell population in small intestine renders Lgr5-positive cells dispensable. Nature 478, 255–259 (2011).

  5. 5.

    Ferguson, A. & Jarrett, E. E. Hypersensitivity reactions in small intestine. I Thymus dependence of experimental ‘partial villous atrophy’. Gut 16, 114–117 (1975).

  6. 6.

    Kamal, M. et al. Paneth and intermediate cell hyperplasia induced in mice by helminth infections. Parasitology 125, 275–281 (2002).

  7. 7.

    VanDussen, K. L. et al. Notch signaling modulates proliferation and differentiation of intestinal crypt base columnar stem cells. Development 139, 488–497 (2012).

  8. 8.

    Sato, T. et al. Paneth cells constitute the niche for Lgr5 stem cells in intestinal crypts. Nature 469, 415–418 (2011).

  9. 9.

    Muñoz, J. et al. The Lgr5 intestinal stem cell signature: robust expression of proposed quiescent ‘+4’ cell markers. EMBO J. 31, 3079–3091 (2012).

  10. 10.

    Flanagan, K. et al. Intestinal epithelial cell up-regulation of LY6 molecules during colitis results in enhanced chemokine secretion. J. Immunol. 180, 3874–3881 (2008).

  11. 11.

    Grencis, R. K. Immunity to helminths: Resistance, regulation, and susceptibility to gastrointestinal nematodes. Annu. Rev. Immunol. 33, 201–225 (2015).

  12. 12.

    Amulic, B., Cazalet, C., Hayes, G. L., Metzler, K. D. & Zychlinsky, A. Neutrophil function: From mechanisms to disease. Annu. Rev. Immunol. 30, 459–489 (2012).

  13. 13.

    Miura, N. et al. Anti-CD3 induces bi-phasic apoptosis in murine intestinal epithelial cells: possible involvement of the Fas/Fas ligand system in different T cell compartments. Int. Immunol. 17, 513–522 (2005).

  14. 14.

    Sollid, L. M. & Jabri, B. Triggers and drivers of autoimmunity: lessons from coeliac disease. Nat. Rev. Immunol. 13, 294–302 (2013).

  15. 15.

    Zhou, P., Streutker, C., Borojevic, R., Wang, Y. & Croitoru, K. IL-10 modulates intestinal damage and epithelial cell apoptosis in T cell-mediated enteropathy. Am. J. Physiol. Gastrointest. Liver Physiol. 287, G599–G604 (2004).

  16. 16.

    Yan, K. S. et al. The intestinal stem cell markers Bmi1 and Lgr5 identify two functionally distinct populations. Proc. Natl Acad. Sci. USA 109, 466–471 (2011).

  17. 17.

    Metcalfe, C., Kljavin, N. M., Ybarra, R. & de Sauvage, F. J. Lgr5+ stem cells are indispensable for radiation-induced intestinal regeneration. Cell Stem Cell 14, 149–159 (2014).

  18. 18.

    Melo, F. de S. E. et al. A distinct role for Lgr5+stem cells in primary and metastatic colon cancer. Nature 543, 676–680 (2017).

  19. 19.

    Mustata, R. C. et al. Identification of Lgr5-independent spheroid-generating progenitors of the mouse fetal intestinal epithelium. Cell Reports 5, 421–432 (2013).

  20. 20.

    Fordham, R. P. et al. Transplantation of expanded fetal intestinal progenitors contributes to colon regeneration after injury. Cell Stem Cell 13, 734–744 (2013).

  21. 21.

    Haber, A. L. et al. A single-cell survey of the small intestinal epithelium. Nature 551, 333–339 (2017).

  22. 22.

    Asfaha, S. et al. Krt19+/Lgr5 cells are radioresistant cancer-initiating stem cells in the colon and intestine. Cell Stem Cell 16, 627–638 (2015).

  23. 23.

    van Es, J. H. et al. Dll1+ secretory progenitor cells revert to stem cells upon crypt damage. Nat. Cell Biol. 14, 1099–1104 (2012).

  24. 24.

    Tetteh, P. W. et al. Replacement of lost Lgr5-positive stem cells through plasticity of their enterocyte-lineage daughters. Cell Stem Cell 18, 203–213 (2016).

  25. 25.

    Buczacki, S. J. A. et al. Intestinal label-retaining cells are secretory precursors expressing Lgr5. Nature 495, 65–69 (2013).

  26. 26.

    Fernandez Vallone, V. et al. Trop2 marks transient gastric fetal epithelium and adult regenerating cells after epithelial damage. Development 143, 1452–1463 (2016).

  27. 27.

    Gadye, L. et al. Injury activates transient olfactory stem cell states with diverse lineage capacities. Cell Stem Cell 21, 775–790 (2017).

  28. 28.

    Lin, B. et al. Injury induces endogenous reprogramming and dedifferentiation of neuronal progenitors to multipotency. Cell Stem Cell 21, 761–774 (2017).

  29. 29.

    Yui, S. et al. Yap/Taz-dependent reprogramming of colonic epithelium links ECM remodeling to tissue regeneration. Cell Stem Cell 22, 35–49 (2018).

  30. 30.

    Goodyear, A. W., Kumar, A., Dow, S. & Ryan, E. P. Optimization of murine small intestine leukocyte isolation for global immune phenotype analysis. J. Immunol. Methods 405, 97–108 (2014).

  31. 31.

    Gregorieff, A. & Clevers, H. in Current Protocols in Stem Cell Biology (John Wiley & Sons, 2015).

  32. 32.

    Subramanian, A. et al. Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles. Proc. Natl Acad. Sci. USA 102, 15545–15550 (2005).

  33. 33.

    Liberzon, A. et al. The molecular signatures database (MSigDB) hallmark gene set collection. Cell Syst. 1, 417–425 (2015).

  34. 34.

    Macosko, E. Z. et al. Highly parallel genome-wide expression profiling of individual cells using nanoliter droplets. Cell 161, 1202–1214 (2015).

  35. 35.

    Satija, R., Farrell, J. A., Gennert, D., Schier, A. F. & Regev, A. Spatial reconstruction of single-cell gene expression data. Nat. Biotechnol. 33, 495–502 (2015).

  36. 36.

    Mahe, M. M. et al. Establishment of gastrointestinal epithelial organoids. Curr. Protoc. Mouse Biol. 3, 217–240 (2013).

  37. 37.

    Halbritter, F., Vaidya, H. J. & Tomlinson, S. R. GeneProf: analysis of high-throughput sequencing experiments. Nat. Methods 9, 7–8 (2011).

Download references

Acknowledgements

We thank M. Consengco, R. D’Urso, J. Ming, A. Rathnayake, N. Wang and Z. Wang for technical expertise, the UCSF Institute of Human Genetics Core and Functional Genomics Core for performing the RNA-seq experiments, members of the Klein and Locksley labs for discussions, and K. Lindquist for advice on GSEA analysis. This work was supported by the National Institutes of Health (AI026918, AI030663 and U01DK103147 from the Intestinal Stem Cell Consortium—a collaborative research project funded by the National Institute of Diabetes and Digestive and Kidney Diseases and the National Institute of Allergy and Infectious Diseases), the Howard Hughes Medical Institute (HHMI), the California Institute for Regenerative Medicine (RN3-06525) and the Sandler Asthma Basic Research Center at the University of California, San Francisco. A.K.S. is an HHMI Fellow. Y.M.N. was awarded a Genentech Graduate Fellowship in 2014.

Reviewer information

Nature thanks A. Hanash, T. Sato, S. Takashima and the other anonymous reviewer(s) for their contribution to the peer review of this work.

Author information

Author notes

  1. These authors contributed equally: Ysbrand M. Nusse, Adam K. Savage.

Affiliations

  1. Biomedical Sciences Graduate Program, University of California, San Francisco, CA, USA

    • Ysbrand M. Nusse
  2. Program in Craniofacial Biology and Department of Orofacial Sciences, University of California, San Francisco, CA, USA

    • Ysbrand M. Nusse
    • , Pauline Marangoni
    • , Axel K. M. Rosendahl-Huber
    • , Tyler A. Landman
    •  & Ophir D. Klein
  3. Howard Hughes Medical Institute and Departments of Medicine and Microbiology & Immunology, University of California, San Francisco, CA, USA

    • Adam K. Savage
    •  & Richard M. Locksley
  4. Department of Molecular Oncology, Genentech Inc., South San Francisco, CA, USA

    • Frederic J. de Sauvage
  5. Department of Pediatrics and Institute for Human Genetics, University of California, San Francisco, CA, USA

    • Ophir D. Klein

Authors

  1. Search for Ysbrand M. Nusse in:

  2. Search for Adam K. Savage in:

  3. Search for Pauline Marangoni in:

  4. Search for Axel K. M. Rosendahl-Huber in:

  5. Search for Tyler A. Landman in:

  6. Search for Frederic J. de Sauvage in:

  7. Search for Richard M. Locksley in:

  8. Search for Ophir D. Klein in:

Contributions

Y.M.N. and A.K.S. jointly conceived the study and designed, performed and interpreted histology, cytometry, transcriptomic and culture experiments. P.M. performed computational analysis of the single-cell transcriptomic experiment. A.K.M.R.-H. and T.A.L. performed and interpreted histology experiments. F.J.d.S. contributed essential reagents. O.D.K. and R.M.L. directed the study and wrote the manuscript with Y.M.N. and A.K.S.

Competing interests

F.J.d.S. is an employee of Genentech and owns shares in Roche.

Corresponding authors

Correspondence to Richard M. Locksley or Ophir D. Klein.

Extended data figures and tables

  1. Extended Data Fig. 1 Helminth infection alters the crypt and intestinal stem cell niche.

    Day 6 of H. polygyrus infection. a, Flow cytometry of CD44+ epithelium from non-granuloma or granuloma biopsies. b, Representative image of Lgr5–GFP staining in the duodenum. Granulomas are indicated by the dashed brackets. In some granulomas, the helminth larva is recognizable by autofluorescence. The presence of rare Lgr5–GFP negative crypts is likely to be a sectioning artefact. c, d, MMP7 and MUC2 staining in normal duodenum or duodenum from mice infected with H. polygyrus. Gr, granuloma. n = 3 (b), or 5 mice (a, c, d). Statistics represent all biological replicates, and all experiments were replicated at least twice. Graphs show mean ± s.d. (a). **P < 0.01 by unpaired, two-tailed Mann–Whitney test. Scale bars: 1 mm (b), 200 µm (c, d). Source Data

  2. Extended Data Fig. 2 RNA-seq analysis of granuloma-associated crypt epithelium.

    a, Representative gating example of epithelia, crypt cells, Lgr5–GFP and Sca-1 in biopsied tissue six days after H. polygyrus infection. Unfractionated tissue preps (as in Extended Data Fig. 3a) were gated similarly. be, Crypt epithelium was sorted from granuloma and non-granuloma biopsies and subjected to RNA-seq analysis as indicated in the Methods. b, The data were filtered for ≥100 reads average in either group, FDR ≤ 0.05, and fold-change comparison of ≥ 2. The 277 genes that passed were compiled into a heat map demonstrating high (red) and low (blue) relative expression. c, GSEA for Lgr5+ signature genes9. FDR <0.01. ES, enrichment score. d, Lgr5+ intestinal stem cell signature genes9 were cross-referenced to the RNA-seq dataset. Data were filtered as in (b) except no fold-change requirement was applied. Clca4 is also known as Clca3b e, The unfiltered RNA-seq dataset was analysed for upstream regulators using Ingenuity Pathways Analysis. The activation Z score indicates the extent of enrichment of targets within the RNA-seq dataset downstream of the indicated regulator, with a positive score indicating enrichment. IFN-related pathways are highlighted in orange. n = 4 independently sorted samples (be, granuloma, 20 mice total), or 5 independently sorted samples (be, non-granuloma, 25 mice total).

  3. Extended Data Fig. 3 Sca-1 is expressed in granuloma crypt epithelium and IFN-γ is present in granulomas.

    a, Representative flow cytometry of Lgr5–GFP and Sca-1 in crypt cells from unfractionated duodenum preps of Lgr5–GFP mice after H. polygyrus infection. b, Flow cytometry of Sca-1 in crypt cells from biopsies from mice 6 or 10 days after infection with H. polygyrus. c, Flow cytometry of Sca-1 in crypt cells from unfractionated duodenum preps of mice 4 days after infection with N. brasiliensis (Nippo). d, Non-granuloma or granuloma biopsies from wild-type mice were analysed by qPCR for Ifng transcript. e, Fold change and read counts of IFN and IFN receptor genes from RNA-seq performed as in Extended Data Fig. 2b with no filter applied. ‘NA’ results from division by zero. n = 4 (a, days 2, 4, 6, b, day 6), 5 (a, day 0, b, day 10), 7 (c, controls), or 8 mice (c, Nippo, d). Statistics represent all biological replicates, and all experiments were replicated at least twice. Graphs show mean ± s.d. (bd). *P < 0.05, **P < 0.01 by unpaired, two-tailed Mann–Whitney test. Source Data

  4. Extended Data Fig. 4 IFN-γ produced by H. polygyrus-responsive immune cells drives the granuloma gene signature.

    ag, Mice were infected with H. polygyrus and analysed at day 6, unless otherwise indicated. a, b, Representative gating example of neutrophils (a) and natural killer (NK) cells, ILC1, ILC2/3, αβ T cells, and γδ T cells (b). c, Neutrophils were enumerated by flow cytometry from non-granuloma (non-gran) or granuloma (gran) biopsies. d, Ifng reporter mice were untreated (uninfected) or infected (gran or non-gran) with H. polygyrus and analysed by flow cytometry 5–6 days later for haematopoietic (CD45+) populations: NK cells, ILC1, ILC2/3, αβ T cells and γδ T cells. No reporter signal was seen in non-lymphoid populations. e, Crypt cells were sorted from granuloma biopsies of IFN-γ-knockout (KO) mice and analysed by qPCR for the indicated transcripts. f, Lgr5–GFP mice were bred with IFN-γ-knockout (KO) mice and offspring were analysed by flow cytometry for Lgr5–GFP expression in crypt epithelia from granuloma biopsies. g, Ifngr1loxp/loxp mice were bred with Vil1-Cre mice and analysed by flow cytometry for Sca-1 expression in crypt epithelia from granuloma biopsies. h, Wild type organoids were treated with 5 ng ml−1 IFN-γ for 24 h and analysed by qPCR for the indicated transcripts. n = 5 mice (d, uninfected, f, KO, g, Ifngr1loxp/loxp), 6 mice (c, d, infected, e, f, heterozygous), 7 mice (g, Ifngr1loxp/loxp;Vil1-Cre), or 7 cultures (h). Statistics represent all biological replicates, and all experiments were replicated at least twice. Graphs show mean ± s.d. (ch). *P < 0.05, **P < 0.01, ***P < 0.001 by unpaired, two-tailed Mann–Whitney test. Source Data

  5. Extended Data Fig. 5 Inflammation via immune cell activation and irradiation induces granuloma-like epithelial responses.

    ad, Mice were treated with 20 µg isotype antibody or anti-TCRβ antibody (clone H57) and analysed 24 h later. a, Unfractionated tissue analysed by qPCR for Ifng transcript. b, Representative flow cytometry of CD44 and Sca-1 in total epithelium. c, Epithelium was assessed for crypt size by flow cytometry using frequency of CD44. d, Crypt cells were sorted and analysed by qPCR for the indicated transcripts. eg, Mice were untreated or subjected to 10 Gy irradiation and analysed three days later. e, Representative flow cytometry of CD44 and Sca-1 expression on total epithelium. f, Crypt cells were sorted and analysed by qPCR for the indicated transcripts. g, Flow cytometry of the frequency of CD44+ crypt cells among total epithelium. n = 3 (a, isotype), 6 (a, anti-TCRβ, b, c, eg), or 7 mice (d). Statistics represent all biological replicates, and all experiments were replicated at least twice. Graphs show mean ±s.d. (a, c, d, f, g). *P < 0.05, **P < 0.01, ***P < 0.001 by unpaired, two-tailed Mann–Whitney test. Source Data

  6. Extended Data Fig. 6 Granuloma crypt epithelium arises from pre-existing Lgr5+ cells but does not require Lgr5+ cells.

    ac, Lgr5DTRGFP/+ (Lgr5) or wild-type (B6) mice were treated with diphtheria toxin (DT) and analysed a day later by flow cytometry for Lgr5–GFP (a), Sca-1 (b), or frequency of crypt cells among total epithelium (c). d, e, Representative images of lineage tracing of Lgr5+ precursors and Sca-1 staining in crypts overlying (gran) and adjacent to (non-gran) H. polygyrus (Hp) granulomas. Lgr5GFP-creERT2/+ Rosa26RFP/+ mice were administered 2.5 mg tamoxifen (Tmx) either immediately before (d) or three days after (e) infection with H. polygyrus. Mice were analysed at day 6. fk, Lgr5DTRGFP/+ mice were treated with diphtheria toxin immediately before infection with H. polygyrus and analysed by flow cytometry at day 1 for Lgr5–GFP (f), or at day 6 for CD44 (g), Sca-1 (h) and EdU (i) in epithelial cells from granuloma biopsies. j, k, Representative images of Sca-1 and EdU detection at day 6. n = 2 (a, b, e), 3 (d, f, j, k), or 4 mice (c, gi). Experiments were replicated at least twice, except the experiment in a and b, which was performed once. Graphs show mean ± s.d. (c, fi). Scale bars: main, 200 µm; insets 50 µm (d, e), 100 µm. (j, k). Source Data

  7. Extended Data Fig. 7 Granuloma crypt epithelium contributes to epithelial turnover.

    a, b, Representative images of Sca-1 and EdU staining. Wild-type mice were injected with EdU at day 5 of infection and analysed after 1 h (a) or 24 h (b) to localize labelled cells within villi, and within crypts overlying (gran) or adjacent to (non-gran) H. polygyrus granulomas. n = 4 (a) or 6 mice (b). Experiments were replicated at least twice. Scale bars: left, 200 µm; insets, 50 µm.

  8. Extended Data Fig. 8 Granuloma crypt epithelium activates a fetal-like program and exhibits altered differentiation.

    a, Material from the cultures described in Fig. 4a, b was analysed by qPCR for Axin2 transcript. b, Sca-1+ or Sca-1 crypt cells were sorted from mice infected with H. polygyrus for the indicated times and analysed by qPCR for fetal transcripts. c, Representative image of whole-mount e15.5 fetal intestine stained for Sca-1 and E-cadherin. d, Bulk RNA-seq data (as in Fig. 1e) were analysed by GSEA for intestinal epithelial signature genes21. Enrichment score (ES) is indicated and all analyses have FDR <10−3. n = 3 mice (a), 3 fetuses (c), 4 mice (b, day 2, 4, 6), 4 independently sorted samples (d, granuloma, 20 mice total), 5 mice (b, day 0), or 5 independently sorted samples (d, non-granuloma, 25 mice total). Statistics represent all biological replicates, and all experiments were replicated at least twice. Graphs show mean ± s.d. (a, b). *P < 0.05 by unpaired, two-tailed Mann–Whitney test. Scale bar: 1 mm (c). Source Data

  9. Extended Data Fig. 9 Markers of adult intestinal cell types in single-cell RNA-seq of Sca-1+ and Sca-1 crypt epithelium.

    Clusters identified by unsupervised hierarchical clustering were arranged per the unsupervised dendrogram of cluster relatedness (top) and normalized expression values for intestinal cell type gene signatures21 were displayed as a heat map in each cluster (middle). The total number of cells in each cluster is shown (bottom). Source Data

Supplementary information

  1. Reporting Summary

  2. Supplementary Table 1

    Granuloma vs. Non-granuloma Bulk RNAseq. This table shows RNAseq data comparing Granuloma (Gr) to Non-Granuloma (NonGr) crypt epithelium from mice infected with H. polygyrus for six days, filtered for expression greater than 100 reads, False Discovery Rate (FDR) < 0.05, Fold Change (FC) > 2, ranked by fold change. FC is unlogged fold change, raw p is adjusted p-value calculated by Holm–Bonferroni method, and FDR is the multiple comparisons adjusted p-value. n=4 (granuloma), and 5 (non-granuloma) samples, from 20 and 25 mice, respectively

  3. Supplementary Table 2

    Gene Set Enrichment Analysis. This table shows Gene Set Enrichment Analysis (GSEA) of bulk Granuloma vs Non-granuloma RNAseq from mice infected with H. polygyrus for six days. GSEA reports an enrichment score (ES), a weighted Kolmogorov-Smirnov statistic, and false discovery rate (FDR), an estimated probability of a false positive finding adjusted for multiple comparisons. n=4 (granuloma), and 5 (non-granuloma) samples, from 20 and 25 mice, respectively

  4. Supplementary Table 3

    Single Cell RNAseq Cluster Mean Gene Expression. This table shows the mean gene expression values for each cluster identified by single cell RNAseq analysis of n=19,754 Sca-1 and n=6,669 Sca-1+ individually sorted crypt cells from one mouse infected with H. polygyrus for six days

  5. Supplementary Table 4

    Single Cell RNAseq Cell Type Signatures. This table shows the mean expression values of intestinal cell type signature genes for each cluster identified by single cell RNAseq analysis of n=19,754 Sca-1 and n=6,669 Sca-1+ individually sorted crypt cells from one mouse infected with H. polygyrus for six days. Genes are ranked by mean expression across all clusters

  6. Supplementary Table 5

    Fetal Signature Single Cell Hypergeometric Distribution Test. This table shows the hypergeometric distribution test of enrichment for fetal signature genes within each cluster identified by single cell RNAseq analysis of n=19,754 Sca-1– and n=6,669 Sca-1+ individually sorted crypt cells from one mouse infected with H. polygyrus for six daysFetal Signature Single Cell Hypergeometric Distribution Test. This table shows the hypergeometric distribution test of enrichment for fetal signature genes within each cluster identified by single cell RNAseq analysis of n=19,754 Sca-1 and n=6,669 Sca-1+ individually sorted crypt cells from one mouse infected with H. polygyrus for six days

  7. Supplementary Table 6

    qPCR Primers. This table lists the sequences of primers used for qPCR analysis throughout the study

Source Data

About this article

Publication history

Received

Accepted

Published

DOI

https://doi.org/10.1038/s41586-018-0257-1

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.