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.

Fundamental properties of unperturbed haematopoiesis from stem cells in vivo

Abstract

Haematopoietic stem cells (HSCs) are widely studied by HSC transplantation into immune- and blood-cell-depleted recipients. Single HSCs can rebuild the system after transplantation1,2,3,4,5. Chromosomal marking6, viral integration7,8,9 and barcoding10,11,12 of transplanted HSCs suggest that very low numbers of HSCs perpetuate a continuous stream of differentiating cells. However, the numbers of productive HSCs during normal haematopoiesis, and the flux of differentiating progeny remain unknown. Here we devise a mouse model allowing inducible genetic labelling of the most primitive Tie2+ HSCs in bone marrow, and quantify label progression along haematopoietic development by limiting dilution analysis and data-driven modelling. During maintenance of the haematopoietic system, at least 30% or 5,000 HSCs are productive in the adult mouse after label induction. However, the time to approach equilibrium between labelled HSCs and their progeny is surprisingly long, a time scale that would exceed the mouse’s life. Indeed, we find that adult haematopoiesis is largely sustained by previously designated ‘short-term’ stem cells downstream of HSCs that nearly fully self-renew, and receive rare but polyclonal HSC input. By contrast, in fetal and early postnatal life, HSCs are rapidly used to establish the immune and blood system. In the adult mouse, 5-fluoruracil-induced leukopenia enhances the output of HSCs and of downstream compartments, thus accelerating haematopoietic flux. Label tracing also identifies a strong lineage bias in adult mice, with several-hundred-fold larger myeloid than lymphoid output, which is only marginally accentuated with age. Finally, we show that transplantation imposes severe constraints on HSC engraftment, consistent with the previously observed oligoclonal HSC activity under these conditions. Thus, we uncover fundamental differences between the normal maintenance of the haematopoietic system, its regulation by challenge, and its re-establishment after transplantation. HSC fate mapping and its linked modelling provide a quantitative framework for studying in situ the regulation of haematopoiesis in health and disease.

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

Access options

Rent or buy this article

Prices vary by article type

from$1.95

to$39.95

Prices may be subject to local taxes which are calculated during checkout

Figure 1: Inducible HSC labelling in Tie2MCM/+RosaYFP mice and frequency estimates on HSC output.
Figure 2: Label progression through the haematopoietic system during adult maintenance and fetal development.
Figure 3: Inference of stem and progenitor cell differentiation, and proliferation from label progression.
Figure 4: In situ HSC response to 5-FU challenge.
Figure 5: Fate of in-situ-labelled HSCs after bone marrow transplantation.

References

  1. Smith, L. G., Weissman, I. L. & Heimfeld, S. Clonal analysis of hematopoietic stem-cell differentiation in vivo. Proc. Natl Acad. Sci. USA 88, 2788–2792 (1991)

    Article  ADS  CAS  Google Scholar 

  2. Osawa, M., Hanada, K., Hamada, H. & Nakauchi, H. Long-term lymphohematopoietic reconstitution by a single CD34-low/negative hematopoietic stem cell. Science 273, 242–245 (1996)

    Article  ADS  CAS  Google Scholar 

  3. Kiel, M. J., Yilmaz, O. H., Iwashita, T., Terhorst, C. & Morrison, S. J. SLAM family receptors distinguish hematopoietic stem and progenitor cells and reveal endothelial niches for stem cells. Cell 121, 1109–1121 (2005)

    Article  CAS  Google Scholar 

  4. Sieburg, H. B. et al. The hematopoietic stem compartment consists of a limited number of discrete stem cell subsets. Blood 107, 2311–2316 (2006)

    Article  CAS  Google Scholar 

  5. Dykstra, B. et al. Long-term propagation of distinct hematopoietic differentiation programs in vivo. Cell Stem Cell 1, 218–229 (2007)

    Article  CAS  Google Scholar 

  6. Abramson, S., Miller, R. G. & Phillips, R. A. The identification in adult bone marrow of pluripotent and restricted stem cells of the myeloid and lymphoid systems. J. Exp. Med. 145, 1567–1579 (1977)

    Article  CAS  Google Scholar 

  7. Keller, G., Paige, C., Gilboa, E. & Wagner, E. F. Expression of a foreign gene in myeloid and lymphoid cells derived from multipotent haematopoietic precursors. Nature 318, 149–154 (1985)

    Article  ADS  CAS  Google Scholar 

  8. Lemischka, I. R., Raulet, D. H. & Mulligan, R. C. Developmental potential and dynamic behavior of hematopoietic stem cells. Cell 45, 917–927 (1986)

    Article  CAS  Google Scholar 

  9. Dick, J. E., Magli, M. C., Huszar, D., Phillips, R. A. & Bernstein, A. Introduction of a selectable gene into primitive stem cells capable of long-term reconstitution of the hemopoietic system of W/Wv mice. Cell 42, 71–79 (1985)

    Article  CAS  Google Scholar 

  10. Gerrits, A. et al. Cellular barcoding tool for clonal analysis in the hematopoietic system. Blood 115, 2610–2618 (2010)

    Article  CAS  Google Scholar 

  11. Lu, R., Neff, N. F., Quake, S. R. & Weissman, I. L. Tracking single hematopoietic stem cells in vivo using high-throughput sequencing in conjunction with viral genetic barcoding. Nature Biotechnol. 29, 928–933 (2011)

    Article  CAS  Google Scholar 

  12. Naik, S. H. et al. Diverse and heritable lineage imprinting of early haematopoietic progenitors. Nature 496, 229–232 (2013)

    Article  ADS  CAS  Google Scholar 

  13. Harrison, D. E., Lerner, C., Hoppe, P. C., Carlson, G. A. & Alling, D. Large numbers of primitive stem cells are active simultaneously in aggregated embryo chimeric mice. Blood 69, 773–777 (1987)

    Article  CAS  Google Scholar 

  14. Yano, M. et al. Expression and function of murine receptor tyrosine kinases, TIE and TEK, in hematopoietic stem cells. Blood 89, 4317–4326 (1997)

    Article  CAS  Google Scholar 

  15. Hsu, H. C. et al. Hematopoietic stem cells express Tie-2 receptor in the murine fetal liver. Blood 96, 3757–3762 (2000)

    Article  CAS  Google Scholar 

  16. Zhang, Y. et al. Inducible site-directed recombination in mouse embryonic stem cells. Nucleic Acids Res. 24, 543–548 (1996)

    Article  CAS  Google Scholar 

  17. Weissman, I. L. Stem cells: units of development, units of regeneration, and units in evolution. Cell 100, 157–168 (2000)

    Article  CAS  Google Scholar 

  18. Oguro, H., Ding, L. & Morrison, S. J. SLAM family markers resolve functionally distinct subpopulations of hematopoietic stem cells and multipotent progenitors. Cell Stem Cell 13, 102–116 (2013)

    Article  CAS  Google Scholar 

  19. Waskow, C. et al. Hematopoietic stem cell transplantation without irradiation. Nature Methods 6, 267–269 (2009)

    Article  CAS  Google Scholar 

  20. Boggs, D. R. The total marrow mass of the mouse: a simplified method of measurement. Am. J. Hematol. 16, 277–286 (1984)

    Article  CAS  Google Scholar 

  21. Gomez Perdiguero, E. et al. Tissue-resident macrophages originate from yolk sac-derived erythro-myeloid progenitors. Nature http://dx.doi.org/10.1038/nature13989 (2014)

  22. Geiger, H., de Haan, G. & Florian, M. C. The ageing haematopoietic stem cell compartment. Nature Rev. Immunol. 13, 376–389 (2013)

    Article  CAS  Google Scholar 

  23. Wilson, A. et al. Hematopoietic stem cells reversibly switch from dormancy to self-renewal during homeostasis and repair. Cell 135, 1118–1129 (2008)

    Article  CAS  Google Scholar 

  24. Sun, J. et al. Clonal dynamics of native haematopoiesis. Nature 514, 322–327 (2014)

    Article  ADS  CAS  Google Scholar 

  25. Kouskoff, V., Fehling, H. J., Lemeur, M., Benoist, C. & Mathis, D. A vector driving the expression of foreign cDNAs in the MHC class II-positive cells of transgenic mice. J. Immunol. Methods 166, 287–291 (1993)

    Article  CAS  Google Scholar 

  26. Shimshek, D. R. et al. Codon-improved Cre recombinase (iCre) expression in the mouse. Genesis 32, 19–26 (2002)

    Article  CAS  Google Scholar 

  27. Verrou, C., Zhang, Y., Zurn, C., Schamel, W. W. & Reth, M. Comparison of the tamoxifen regulated chimeric Cre recombinases MerCreMer and CreMer. Biol. Chem. 380, 1435–1438 (1999)

    Article  CAS  Google Scholar 

  28. Casanova, E. et al. ER-based double iCre fusion protein allows partial recombination in forebrain. Genesis 34, 208–214 (2002)

    Article  CAS  Google Scholar 

  29. Dumont, D. J. et al. Dominant-negative and targeted null mutations in the endothelial receptor tyrosine kinase, tek, reveal a critical role in vasculogenesis of the embryo. Genes Dev. 8, 1897–1909 (1994)

    Article  CAS  Google Scholar 

  30. Sato, T. N. et al. Distinct roles of the receptor tyrosine kinases Tie-1 and Tie-2 in blood vessel formation. Nature 376, 70–74 (1995)

    Article  ADS  CAS  Google Scholar 

  31. Srinivas, S. et al. Cre reporter strains produced by targeted insertion of EYFP and ECFP into the ROSA26 locus. BMC Dev. Biol. 1, 4 (2001)

    Article  CAS  Google Scholar 

  32. Bertrand, J. Y. et al. Three pathways to mature macrophages in the early mouse yolk sac. Blood 106, 3004–3011 (2005)

    Article  CAS  Google Scholar 

  33. Luche, H., Weber, O., Nageswara Rao, T., Blum, C. & Fehling, H. J. Faithful activation of an extra-bright red fluorescent protein in “knock-in” Cre-reporter mice ideally suited for lineage tracing studies. Eur. J. Immunol. 37, 43–53 (2007)

    Article  CAS  Google Scholar 

  34. Schwenk, F., Baron, U. & Rajewsky, K. A cre-transgenic mouse strain for the ubiquitous deletion of loxP-flanked gene segments including deletion in germ cells. Nucleic Acids Res. 23, 5080–5081 (1995)

    Article  CAS  Google Scholar 

Download references

Acknowledgements

We thank C. Blum for blastocyst injections, T. Arnsperger, N. Maltry and S. Schäfer for technical assistance, and the animal facility at the DKFZ for expert mouse husbandry. We are grateful to N. Dietlein, H. J. Fehling, T. Feyerabend, C. Opitz, J. Rodewald, A. Roers, K. Rajewsky and A. Trumpp for discussions. K.B. was initially funded through the International Graduate School in Molecular Medicine, Ulm. M.R. was supported by the DFG through EXC294, TRR130 and SFB746. T.H. is a member of CellNetworks, and was supported by BMBF e:Bio grant T-Sys (Fkz 031614) and EU-FP7 Marie Curie ITN Quantitative T cell immunology (QuanTI). H.-R.R. was supported by DFG-SFB 873-project B11, DFG-SFB 938 project L, ERC Advanced Grant no. 233074, and the Helmholtz Alliance on preclinical cancer models (PCCC).

Author information

Authors and Affiliations

Authors

Contributions

K.B. designed and performed experiments, K.K. and S.M.S. performed experiments, M.B. and M.F. performed theoretical modelling, T.H.-L. made initial bioinformatic analyses, M.R. contributed to the tamoxifen-regulated Cre construct, T.H. developed mathematical models and wrote the paper, and H.-R.R. conceived and supervised the study, and wrote the paper.

Corresponding authors

Correspondence to Thomas Höfer or Hans-Reimer Rodewald.

Ethics declarations

Competing interests

The authors declare no competing financial interests.

Extended data figures and tables

Extended Data Figure 1 Generation of the Tie2MCM allele, labelling of HSCs in tamoxifen-treated Tie2MCM/+RosaYFP mice, and groups for limiting dilution analysis.

a, The endogenous Tie2 locus, the gene targeting vector and the targeted allele with (Tie2MCMNeo) and without (Tie2MCMΔNeo) neomycin are depicted. Oligonucleotides and DNA probe for genotyping, and restriction sites used for Southern blotting are indicated (not drawn to scale). b, PCR verification of the targeted Tie2MCM allele in embryonic stem cells before and after neomycin deletion. c, Southern blot analysis of Tie2MCMNeo (−Flp) and Tie2MCMΔNeo (+Flp), and wild-type embryonic stem-cell clones. d, Principle of inducible fate mapping. In the absence of tamoxifen MCM is inactive (reporteroff). Tamoxifen treatment activates the reporter (reporteron). After tamoxifen treatment, labelled cells and their progeny remain marked (reporteron). e, Summary of HSC labelling frequencies of tamoxifen-treated Tie2MCM/+RosaYFP mice (n = 112; 5 times on 5 consecutive days) analysed between 1 and 34 weeks after label induction. These data are the basis for the kinetic analysis (Fig. 2a–f) and for the mathematical modelling (Fig. 3). Each dot represents an individual mouse. Bar indicates mean (1.041 ± 0.8013 s.d.). f, A single YFP+ LSK CD150+CD48 HSC from a tamoxifen-treated Tie2MCM/+RosaYFP mouse was transplanted into a Rag2−/− γc−/− KitW/Wv recipient mouse (1° transfer). Donor cells were identified by YFP expression, and analysed 16 weeks after transplantation in bone marrow (BM), thymus and spleen using the markers shown. YFP+Kit+ donor bone marrow cells were re-transplanted into a secondary Rag2−/− γc−/− KitW/Wv recipient (2° transfer), and analysed as described for the primary transfer. gj, HSC labelling frequencies in tamoxifen-treated Tie2MCM/+RosaYFP mice analysed 6 weeks onwards after label induction were used for limiting dilution analysis of CD45+ output, granulocytes (n = 60) (g, h), pro B cells and double-positive thymocytes (n = 79) (i, j). Each dot represents an individual mouse. Mice grouped together are highlighted in black or white (groups I–IV). Mathematical calculations are shown in the tables (h, j). In g, data shown represent the aggregate of labelling frequencies below 1% shown in e, plus data obtained in mice receiving only a single tamoxifen injection.

Extended Data Figure 2 Characterization of Tie2MCM/+ mice.

ad, Numbers of haematopoietic cell subsets isolated from bone marrow (a, b), thymus (c) and spleen (d) of Tie2MCM/+ (n = 10; black) and Tie2+/+ (n = 10; white) littermates were determined by flow cytometry. Each dot represents an individual mouse. Bars indicate mean. e, f, Flow cytometric analysis of a representative Tie2MCM/+ (top) and Tie2+/+ (bottom) mouse gated on Lin cells, and analysed for expression of Kit versus Sca-1. Further analysis of LinKit+Sca-1+ (LSK) cells for CD150 and CD48 revealed comparable marker distributions (f). Percentages of LSK cells among the Lin fraction (left), and of HSCs, ST-HSCs and MPPs among the LSK fraction (right) in the bone marrow of three independent Tie2MCM/+ (black) and Tie2+/+ (white) mice. Data are mean ± s.d. g, Proliferation rates in surface receptor Tie2-positive and Tie2-negative HSCs in the bone marrow of Tie2MCM/+ (black) and Tie2+/+ (white) mice 24 h after EdU administration. Data represent mean ± s.d. from two independent experiments of FACS-sorted populations from Tie2MCM/+ (nExp1 = 5; nExp2 = 3) and Tie2+/+ (nExp1 = 3; nExp2 = 2) mice. h, i, Rag2−/− γc−/− KitW/Wv recipients (n = 30; for analysis of B and T cells in the spleen n = 28) were injected with equivalent numbers of Tie2MCM/+ and Tie2+/+ HSCs (500–1,000 of each), and analysed after at least 16 weeks. The percentages of Tie2MCM/+ HSC-derived haematopoietic cells in bone marrow, spleen and thymus are shown. Each dot represents an individual mouse. Bars indicate mean.

Extended Data Figure 3 Kinetics of YFP label emergence after label induction in total bone marrow cells in adult mice, and in fetal liver and bone marrow cells in fetal and early postnatal mice.

a, Percentages of YFP+ cells among total non-lineage-depleted bone marrow cells of tamoxifen-treated Tie2MCM/+RosaYFP mice (n = 47). Time point 0 corresponds to the time of tamoxifen treatment of adult mice (all of which were at least 6 weeks). b, Tie2MCM/+RosaYFP mice (n = 32) were treated with tamoxifen on E10.5 (time point 0). Subsequently, percentages of YFP+ cells were determined among total fetal liver on E12.5 and E15.5, and in bone marrow at birth and 1 week of age. Each dot represents an individual mouse.

Extended Data Figure 4 Donor-derived myeloid cells disappear within 6 weeks after adoptive transfer.

a, Rag2−/− γc−/− KitW/Wv recipient mice (n = 8) received CMPs and GMPs (together 5 × 104 per mouse) from panRFP mice together with CLPs (0.5 × 104 per mouse) from panYFP mice. Peripheral blood samples were measured 7, 14, 21 and 32 days after transplantation. b, Analysis of donor chimaerism of CMP/GMP-derived (Gr-1+CD11b+ granulocytes, red) and CLP-derived (CD19+ B cells, yellow) progeny. Each line represents an individual mouse.

Extended Data Figure 5 Labelling of HSCs, but not erythromyeloid progenitors, in Tie2MCM/+RosaYFP embryos treated on E10.5 in utero by tamoxifen.

a, Tie2MCM/+RosaYFP embryos were treated with tamoxifen on E7.5 or E10.5, and analysed at E12.5. be, Fetal liver cells from representative Tie2MCM/+RosaYFP embryos labelled on E10.5 (b) or E7.5 (d) were analysed by flow cytometry for YFP labelling in HSCs (LSK CD150+CD48) and erythromyeloid progenitors (EMP) (LinKit+CD45+/low) (see Methods for references). HSCs were marked in mice labelled at E7.5 and E10.5, but erythromyeloid progenitors were not marked in mice labelled on E10.5. c, d, Summary of experiments depicted in b (n = 8) (c) and d (n = 6) (e). Each dot represents an individual embryo. Bars indicate mean.

Extended Data Figure 6 Details of model analysis.

a, Ratio of compartment sizes for stem and progenitor cell compartments. Compartment sizes were determined by cell counting of cell suspensions from bone marrow, spleen and thymus in a Neubauer chamber combined with phenotypic definition of each population by flow cytometry. Mean and s.e.m. for n = 10 mice are shown. b, Confidence bounds for the model parameters. Profile likelihoods (blue lines) for the inverse residence times, k = 1/τ, of cells in the ST-HSC, MPP, CMP and CLP compartments (units: per day). The profile likelihoods have been computed as described in Supplementary Methods, with the experimental data from Fig. 3c. The red lines indicate the 95% confidence level. Except for the CMPs, for which the data are consistent with a broad range of k (that is, the CMP residence time is low), the residence times are accurately determined by the label propagation data alone. In particular, the very low k for the ST-HSCs shows that this compartment operates near self-renewal. c, Profile likelihoods (blue lines) and 95% confidence level (red lines) for the net proliferation rates (β) and differentiation rates (α), computed with the data of Fig. 3c and Extended Data Fig. 4 (unit: per day). The differentiation rate from HSCs (αHSCST-HSC) has the same profile likelihood as the net proliferation of HSC (βHSC) and is therefore not shown. αHSCST-HSC, αST-HSCMPP, αMPPCLP, βHSC and βST-HSC are accurately determined by the data. Moreover, both αMPPCMP and βMPP have a lower bound that is one and two orders of magnitude larger than the respective parameters for ST-HSCs and HSCs, showing that MPPs have significantly higher proliferation and differentiation activities than the preceding compartments. Note that (βMPP) and αMPPCMP are strongly correlated (not shown). d, Label progression data and fit of the mathematical model for further downstream myeloid precursors (CMPs to GMPs or MEPs) and lymphoid precursors (CLPs to pro B cells) in the bone marrow. Data were measured up to 238 days after label induction (blue points, with s.e.m., as in Fig. 2l) have been used for the model fit (red lines, best fit and grey shades, 95% confidence bands). The parameter values are as in Supplementary Table 1 and, in addition, αCMPGMP = 2 (0.04, 4) day−1, αCMPMEP = 3 (0.1, 4) day−1, αCLPpro B = 2 (0.8, 4) day−1, βCMP = 4 (−1, 4) day−1, βCLP = 3 (0.4, 4) day−1, τGMP = 0.12 (0.12, 33) days, τMEP = 0.13 (0.13, 22) days, τproB = 54 (6, 141) days (in brackets: 95% confidence bounds). For further details see Supplementary Information.

Extended Data Figure 7 Consequences of lymphoid–myeloid branch points at distinct stem or progenitor stages.

ac, Lymphoid–myeloid branch points were considered at the MPP stage (a) or at the earlier CD229+ ST-HSC subset (b). Each population is phenotypically defined as shown in c. MPPs have also been termed HPC-1, and the CD229+ ST-HSC subset has also been termed MPP2/3 (ref. 18). Cell flux rates per day are shown from the branch points to CMPs or CLPs (right panels in a and b, with 95% confidence intervals). In both scenarios, the production of CMPs is several-hundred-fold larger than that of CLPs. Assuming the branch point at the CD229+ ST-HSC stage, biased myeloid differentiation is still evident (5-fold), but the large uneven production is mainly achieved by flux amplification downstream from the bifurcation. d, Label progression data and fit of the mathematical model. Data measured up to 311 days after label induction in the ST-HSC CD229 and ST-HSC CD229+ compartments (orange points, with s.e.m.; groups of mice: 57 days n = 7; 122 days n = 4; 311 days n = 8) and in CLP, MPP and CMP compartments data measured up to 238 days (orange points, with s.e.m.; groups of adult mice as in Fig. 2l) have been used for the model fit (red lines, best fit; grey shades, 95% confidence bands). The resulting parameters are αHSCCD229–ST-HSC = 0.0001 (0.00001–0.00016) day−1, αCD229–ST-HSCCD229+ST-HSC = 4 (2–4) day−1, αCD229+ST-HSCMPP = 0.03 (0.02–0.06) day−1, αCD229+ST-HSCCLP = 0.007 (0.004–0.012) day−1, βCD229–ST-HSC = 4 (2–4) day−1, βCD229+ST-HSC = 0.0 (−0.06–0.0016) day−1 and βMPP = 4 (0.3–4) day−1 (in brackets: 95% confidence intervals). For further details see Supplementary Methods.

Extended Data Figure 8 Limiting dilution analysis of in-situ-labelled transplanted HSCs.

a, Tie2MCM/+RosaYFP mice were injected with tamoxifen (donor bone marrow; see Fig. 5a). Numbers of YFP+ HSCs in donor bone marrow transplanted into 38 recipient mice are displayed. Each dot is an individual recipient. Mice grouped together are highlighted in black or white (groups I–IV). b, Data underlying limiting dilution analysis. c, Numbers of injected YFP+ donor HSCs are plotted against the fraction of YFP-negative recipients on a logarithmic scale. d, Histogram showing the extent of self-renewing proliferation for engrafted YFP+ HSCs in mice labelled ‘within sampling error’ and ‘overrepresented’ in Fig. 5d (n = 18). Proliferation is shown as the number of generations per engrafted HSC needed to achieve the measured frequencies of YFP+ HSCs in the bone marrow 4 months after transplantation. For further details see Supplementary Discussion.

Extended Data Figure 9 Simulated effects of ablation of HSCs, ST-HSCs or MPPs.

In the mathematical model used to fit the label progression data (Fig. 3c), numbers of HSCs (a), HSCs and ST-HSCs (b), and HSCs, ST-HSCs and MPPs (c) were reduced from its steady state value to zero at t = 0, and subsequently held there. The predicted responses in the downstream compartments are shown. The horizontal black lines at 1/e = 37% indicate the residence times of the compartments that control the time scale of the response. Note that these simulations assume that there are no homeostatic mechanisms present within the compartments that could maintain their size independent of input from upstream compartments.

Extended Data Table 1 Transplantations of in-situ-labelled HSC

Supplementary information

Supplementary Information

This file contains Supplementary Methods, Supplementary Table 1, Supplementary Discussion and Supplementary References. (PDF 166 kb)

PowerPoint slides

Rights and permissions

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Busch, K., Klapproth, K., Barile, M. et al. Fundamental properties of unperturbed haematopoiesis from stem cells in vivo. Nature 518, 542–546 (2015). https://doi.org/10.1038/nature14242

Download citation

  • Received:

  • Accepted:

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1038/nature14242

This article is cited by

Comments

By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.

Search

Quick links

Nature Briefing

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing