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.

  • Article
  • Published:

A patterned human neural tube model using microfluidic gradients

Nature volume 628, pages 391–399 (2024)Cite this article

Subjects

Abstract

The human nervous system is a highly complex but organized organ. The foundation of its complexity and organization is laid down during regional patterning of the neural tube, the embryonic precursor to the human nervous system. Historically, studies of neural tube patterning have relied on animal models to uncover underlying principles. Recently, models of neurodevelopment based on human pluripotent stem cells, including neural organoids1,2,3,4,5 and bioengineered neural tube development models6,7,8,9,10, have emerged. However, such models fail to recapitulate neural patterning along both rostral–caudal and dorsal–ventral axes in a three-dimensional tubular geometry, a hallmark of neural tube development. Here we report a human pluripotent stem cell-based, microfluidic neural tube-like structure, the development of which recapitulates several crucial aspects of neural patterning in brain and spinal cord regions and along rostral–caudal and dorsal–ventral axes. This structure was utilized for studying neuronal lineage development, which revealed pre-patterning of axial identities of neural crest progenitors and functional roles of neuromesodermal progenitors and the caudal gene CDX2 in spinal cord and trunk neural crest development. We further developed dorsal–ventral patterned microfluidic forebrain-like structures with spatially segregated dorsal and ventral regions and layered apicobasal cellular organizations that mimic development of the human forebrain pallium and subpallium, respectively. Together, these microfluidics-based neurodevelopment models provide three-dimensional lumenal tissue architectures with in vivo-like spatiotemporal cell differentiation and organization, which will facilitate the study of human neurodevelopment and disease.

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

Access options

Buy this article

  • Purchase on Springer Link
  • Instant access to full article PDF
Buy now

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

Fig. 1: An R–C patterned human NT model using microfluidics.
Fig. 2: R–C and D–V patterning of µNTLS.
Fig. 3: Single-cell transcriptome analysis of µNTLS.
Fig. 4: NC development in µNTLS.
Fig. 5: Development of D–V patterned μFBLS.

Similar content being viewed by others

Data availability

Data supporting findings of this study are available within the article and its Supplementary Information. scRNA-seq data supporting this results of this study have been deposited into the Gene Expression Omnibus (GEO) database with accession number GSE194225. Mouse embryo scRNA-seq data are from the GEO (GSE87038 and GSE119945). Human embryo scRNA-seq data are from GEO (GSE157329). Developing human brain scRNA-seq data are from the Linnarsson Laboratory GitHub site. Source data are provided with this paper.

Code availability

Custom R, Python and Matlab scripts are used in this work. They are not central to the conclusions of the paper. These codes are available from the corresponding author upon request.

References

  1. Lancaster, M. A. et al. Cerebral organoids model human brain development and microcephaly. Nature 501, 373–379 (2013).

    Article  CAS  PubMed  ADS  Google Scholar 

  2. Qian, X. et al. Brain-region-specific organoids using mini-bioreactors for modeling ZIKV exposure. Cell 165, 1238–1254 (2016).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  3. ctipBirey, F. et al. Assembly of functionally integrated human forebrain spheroids. Nature 545, 54–59 (2017).

    Article  Google Scholar 

  4. Velasco, S. et al. Individual brain organoids reproducibly form cell diversity of the human cerebral cortex. Nature 570, 523–527 (2019).

    Article  CAS  PubMed  PubMed Central  ADS  Google Scholar 

  5. Faustino Martins, J.-M. et al. Self-organizing 3D human trunk neuromuscular organoids. Cell Stem Cell 26, 172–186.e176 (2020).

    Article  CAS  PubMed  Google Scholar 

  6. Meinhardt, A. et al. 3D reconstitution of the patterned neural tube from embryonic stem cells. Stem Cell Reports 3, 987–999 (2014).

    Article  PubMed  PubMed Central  Google Scholar 

  7. Haremaki, T. et al. Self-organizing neuruloids model developmental aspects of Huntington’s disease in the ectodermal compartment. Nat. Biotechnol. 37, 1198–1208 (2019).

    Article  CAS  PubMed  Google Scholar 

  8. Zheng, Y. et al. Dorsal–ventral patterned neural cyst from human pluripotent stem cells in a neurogenic niche. Sci. Adv. 5, eaax5933 (2019).

    Article  CAS  PubMed  PubMed Central  ADS  Google Scholar 

  9. Rifes, P. et al. Modeling neural tube development by differentiation of human embryonic stem cells in a microfluidic WNT gradient. Nat. Biotechnol. 38, 1265–1273 (2020).

    Article  CAS  PubMed  Google Scholar 

  10. Karzbrun, E. et al. Human neural tube morphogenesis in vitro by geometric constraints. Nature 599, 268–272 (2021).

    Article  CAS  PubMed  PubMed Central  ADS  Google Scholar 

  11. Müller, F. & O’Rahilly, R. The development of the human brain, the closure of the caudal neuropore, and the beginning of secondary neurulation at stage 12. Anat. Embryol. 176, 413–430 (1987).

    Article  Google Scholar 

  12. Storey, K. G. et al. Early posterior neural tissue is induced by FGF in the chick embryo. Development 125, 473–484 (1998).

    Article  CAS  PubMed  Google Scholar 

  13. Niederreither, K., Subbarayan, V., Dollé, P. & Chambon, P. Embryonic retinoic acid synthesis is essential for early mouse post-implantation development. Nat. Genet. 21, 444–448 (1999).

    Article  CAS  PubMed  Google Scholar 

  14. Nordström, U., Jessell, T. M. & Edlund, T. Progressive induction of caudal neural character by graded Wnt signaling. Nat. Neurosci. 5, 525–532 (2002).

    Article  PubMed  Google Scholar 

  15. O’Shea, K. S. Differential deposition of basement membrane components during formation of the caudal neural tube in the mouse embryo. Development 99, 509–519 (1987).

    Article  PubMed  Google Scholar 

  16. Chambers, S. M. et al. Highly efficient neural conversion of human ES and iPS cells by dual inhibition of SMAD signaling. Nat. Biotechnol. 27, 275–280 (2009).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  17. Xue, X. et al. Mechanics-guided embryonic patterning of neuroectoderm tissue from human pluripotent stem cells. Nat. Mater. 17, 633–641 (2018).

    Article  CAS  PubMed  PubMed Central  ADS  Google Scholar 

  18. Philippidou, P. & Dasen, J. S. Hox genes: choreographers in neural development, architects of circuit organization. Neuron 80, 12–34 (2013).

    Article  CAS  PubMed  Google Scholar 

  19. Metzis, V. et al. Nervous system regionalization entails axial allocation before neural differentiation. Cell 175, 1105–1118.e17 (2018).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  20. Deschamps, J. et al. Initiation, establishment and maintenance of Hox gene expression patterns in the mouse. Int. J. Dev. Biol. 43, 635–650 (1999).

    CAS  PubMed  Google Scholar 

  21. Bel-Vialar, S., Itasaki, N. & Krumlauf, R. Initiating Hox gene expression: in the early chick neural tube differential sensitivity to FGF and RA signaling subdivides the HoxB genes in two distinct groups. Development 129, 5103–5115 (2002).

    Article  CAS  PubMed  Google Scholar 

  22. Duband, J.-L., Blavet, C., Jarov, A. & Fournier-Thibault, C. Spatio-temporal control of neural epithelial cell migration and epithelium-to-mesenchyme transition during avian neural tube development. Dev. Growth Differ. 51, 25–44 (2009).

    Article  PubMed  Google Scholar 

  23. Wymeersch, F. J. et al. Position-dependent plasticity of distinct progenitor types in the primitive streak. eLife 5, e10042 (2016).

    Article  PubMed  PubMed Central  Google Scholar 

  24. Tzouanacou, E., Wegener, A., Wymeersch, F. J., Wilson, V. & Nicolas, J.-F. Redefining the progression of lineage segregations during mammalian embryogenesis by clonal analysis. Dev. Cell 17, 365–376 (2009).

    Article  CAS  PubMed  Google Scholar 

  25. Henrique, D., Abranches, E., Verrier, L. & Storey, K. G. Neuromesodermal progenitors and the making of the spinal cord. Development 142, 2864–2875 (2015).

    Article  CAS  PubMed  Google Scholar 

  26. Diaz-Cuadros, M. et al. In vitro characterization of the human segmentation clock. Nature 580, 113–118 (2020).

    Article  CAS  PubMed  PubMed Central  ADS  Google Scholar 

  27. Gouti, M. et al. In vitro generation of neuromesodermal progenitors reveals distinct roles for Wnt signalling in the specification of spinal cord and paraxial mesoderm identity. PLoS Biol. 12, e1001937 (2014).

    Article  PubMed  PubMed Central  Google Scholar 

  28. Wurst, W. & Bally-Cuif, L. Neural plate patterning: upstream and downstream of the isthmic organizer. Nat. Rev. Neurosci. 2, 99–108 (2001).

    Article  CAS  PubMed  Google Scholar 

  29. Duester, G. Retinoic acid synthesis and signaling during early organogenesis. Cell 134, 921–931 (2008).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  30. Sakai, Y. et al. The retinoic acid-inactivating enzyme CYP26 is essential for establishing an uneven distribution of retinoic acid along the anterio-posterior axis within the mouse embryo. Genes Dev. 15, 213–225 (2001).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  31. Chawengsaksophak, K., de Graaff, W., Rossant, J., Deschamps, J. & Beck, F. Cdx2 is essential for axial elongation in mouse development. Proc. Natl Acad. Sci. USA 101, 7641–7645 (2004).

    Article  CAS  PubMed  PubMed Central  ADS  Google Scholar 

  32. Amin, S. et al. Cdx and T brachyury co-activate growth signaling in the embryonic axial progenitor niche. Cell Rep. 17, 3165–3177 (2016).

    Article  CAS  PubMed  Google Scholar 

  33. Liem, K. F., Tremml, G. & Jessell, T. M. A role for the roof plate and its resident TGFβ-related proteins in neuronal patterning in the dorsal spinal cord. Cell 91, 127–138 (1997).

    Article  CAS  PubMed  Google Scholar 

  34. Muroyama, Y., Fujihara, M., Ikeya, M., Kondoh, H. & Takada, S. Wnt signaling plays an essential role in neuronal specification of the dorsal spinal cord. Genes Dev. 16, 548–553 (2002).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  35. Dessaud, E. et al. Interpretation of the sonic hedgehog morphogen gradient by a temporal adaptation mechanism. Nature 450, 717–720 (2007).

    Article  CAS  PubMed  ADS  Google Scholar 

  36. Wilson, L., Gale, E., Chambers, D. & Maden, M. Retinoic acid and the control of dorsoventral patterning in the avian spinal cord. Dev. Biol. 269, 433–446 (2004).

    Article  CAS  PubMed  Google Scholar 

  37. Wilson, S. W. & Rubenstein, J. L. R. Induction and dorsoventral patterning of the telencephalon. Neuron 28, 641–651 (2000).

    Article  CAS  PubMed  Google Scholar 

  38. Sagner, A. & Briscoe, J. Establishing neuronal diversity in the spinal cord: a time and a place. Development 146, dev182154 (2019).

    Article  CAS  PubMed  Google Scholar 

  39. Sauka-Spengler, T. & Bronner-Fraser, M. A gene regulatory network orchestrates neural crest formation. Nat. Rev. Mol. Cell Biol. 9, 557–568 (2008).

    Article  CAS  PubMed  Google Scholar 

  40. Jeong, J. & McMahon, A. P. Growth and pattern of the mammalian neural tube are governed by partially overlapping feedback activities of the hedgehog antagonists patched 1 and Hhip1. Development 132, 143–154 (2005).

    Article  CAS  PubMed  Google Scholar 

  41. Kiecker, C. & Lumsden, A. Compartments and their boundaries in vertebrate brain development. Nat. Rev. Neurosci. 6, 553–564 (2005).

    Article  CAS  PubMed  Google Scholar 

  42. Xu, Y. et al. A single-cell transcriptome atlas profiles early organogenesis in human embryos. Nat. Cell Biol. 25, 604–615 (2023).

    Article  PubMed  Google Scholar 

  43. Cao, J. et al. The single-cell transcriptional landscape of mammalian organogenesis. Nature 566, 496–502 (2019).

    Article  CAS  PubMed  PubMed Central  ADS  Google Scholar 

  44. Pijuan-Sala, B. et al. A single-cell molecular map of mouse gastrulation and early organogenesis. Nature 566, 490–495 (2019).

    Article  CAS  PubMed  PubMed Central  ADS  Google Scholar 

  45. Le Douarin, N. M., Creuzet, S., Couly, G. R. & Dupin, E. Neural crest cell plasticity and its limits. Development 131, 4637–4650 (2004).

    Article  PubMed  Google Scholar 

  46. Soldatov, R. et al. Spatiotemporal structure of cell fate decisions in murine neural crest. Science 364, eaas9536 (2019).

    Article  CAS  PubMed  Google Scholar 

  47. Frith, T. J. R. et al. Human axial progenitors generate trunk neural crest cells in vitro. eLife 7, e35786 (2018).

    Article  PubMed  PubMed Central  Google Scholar 

  48. Eze, U. C., Bhaduri, A., Haeussler, M., Nowakowski, T. J. & Kriegstein, A. R. Single-cell atlas of early human brain development highlights heterogeneity of human neuroepithelial cells and early radial glia. Nat. Neurosci. 24, 584–594 (2021).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  49. Onorati, M. et al. Molecular and functional definition of the developing human striatum. Nat. Neurosci. 17, 1804–1815 (2014).

    Article  CAS  PubMed  Google Scholar 

  50. Braun, E. et al. Comprehensive cell atlas of the first-trimester developing human brain. Science 382, eadf1226 (2023).

    Article  CAS  PubMed  Google Scholar 

  51. Muncie, J. M. et al. Mechanical tension promotes formation of gastrulation-like nodes and patterns mesoderm specification in human embryonic stem cells. Dev. Cell 55, 679–694 (2020).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  52. Kumar, N. et al. The lineage-specific transcription factor CDX2 navigates dynamic chromatin to control distinct stages of intestine development. Development 146, dev172189 (2019).

    Article  PubMed  PubMed Central  Google Scholar 

  53. Bayerl, J. et al. Principles of signaling pathway modulation for enhancing human naive pluripotency induction. Cell Stem Cell 28, 1549–1565 (2021).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  54. Cong, L. et al. Multiplex genome engineering using CRISPR/Cas systems. Science 339, 819–823 (2013).

    Article  CAS  PubMed  PubMed Central  ADS  Google Scholar 

  55. Chen, Y. et al. Engineering human stem cell lines with inducible gene knockout using CRISPR/Cas9. Cell Stem Cell 17, 233–244 (2015).

    Article  PubMed  PubMed Central  Google Scholar 

  56. Hermann, M. et al. Binary recombinase systems for high-resolution conditional mutagenesis. Nucleic Acids Res. 42, 3894–3907 (2014).

    Article  CAS  PubMed  PubMed Central  ADS  Google Scholar 

  57. Chu, L.-F. et al. Single-cell RNA-seq reveals novel regulators of human embryonic stem cell differentiation to definitive endoderm. Genome Biol. 17, 173 (2016).

    Article  PubMed  PubMed Central  Google Scholar 

  58. Cudré-Mauroux, C. et al. Lentivector-mediated transfer of Bmi-1 and telomerase in muscle satellite cells yields a Duchenne myoblast cell line with long-term genotypic and phenotypic stability. Hum. Gene Ther. 14, 1525–1533 (2003).

    Article  PubMed  Google Scholar 

  59. Li, X. et al. Desktop aligner for fabrication of multilayer microfluidic devices. Rev. Sci. Instrum. 86, 075008 (2015).

    Article  PubMed  PubMed Central  ADS  Google Scholar 

  60. Watanabe, K. et al. A ROCK inhibitor permits survival of dissociated human embryonic stem cells. Nat. Biotechnol. 25, 681–686 (2007).

    Article  CAS  PubMed  Google Scholar 

  61. Serra, D. et al. Self-organization and symmetry breaking in intestinal organoid development. Nature 569, 66–72 (2019).

    Article  CAS  PubMed  PubMed Central  ADS  Google Scholar 

  62. Butler, A., Hoffman, P., Smibert, P., Papalexi, E. & Satija, R. Integrating single-cell transcriptomic data across different conditions, technologies, and species. Nat. Biotechnol. 36, 411–420 (2018).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  63. La Manno, G. et al. RNA velocity of single cells. Nature 560, 494–498 (2018).

    Article  PubMed  PubMed Central  ADS  Google Scholar 

  64. Jin, S. et al. Inference and analysis of cell–cell communication using CellChat. Nat. Commun. 12, 1088 (2021).

    Article  CAS  PubMed  PubMed Central  ADS  Google Scholar 

  65. Aibar, S. et al. SCENIC: single-cell regulatory network inference and clustering. Nat. Methods 14, 1083–1086 (2017).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  66. Street, K. et al. Slingshot: cell lineage and pseudotime inference for single-cell transcriptomics. BMC Genomics 19, 477 (2018).

    Article  PubMed  PubMed Central  Google Scholar 

  67. Delile, J. et al. Single cell transcriptomics reveals spatial and temporal dynamics of gene expression in the developing mouse spinal cord. Development 146, dev173807 (2019).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

Download references

Acknowledgements

We thank A. Tsakiridis, V. Wilson, A. Martinez Arias, B. Allen, J. M. Parent, S. O’Shea, D. M. Wellik, A. M. Tidball, D. M. Martin and Y. Zhai for their comments; J. N. Lakins and V. M. Weaver for providing the Brachyury–mNeonGreen human ES cell reporter line; M. Povolotski for helping with PCR genotyping; and staff at the Michigan Medicine Microscopy Core for training and support in microscopy imaging, the Michigan Orthopaedic Research Laboratories Histology Core for support in cryosectioning, the Michigan Advanced Genomics Core for scRNA-seq service, and the Michigan Lurie Nanofabrication Facility for support in microfabrication. This work is supported by the Michigan-Cambridge Collaboration Initiative (J.F.), the University of Michigan Mcubed Fund (J.F.), the 21st Century Jobs Trust Fund received through the Michigan Strategic Fund from the State of Michigan (Grant CASE-315037; J.F.), a University of Michigan Mid-career Biosciences Faculty Achievement Recognition Award (J.F.), the Dr. Miriam and Sheldon G. Adelson Medical Research Foundation (G.-L.M.), the National Science Foundation of the United States (I-Corps 2112458 and CBET 1901718 to J.F.), and the National Institutes of Health of the United States (R21 NS113518, R01 GM143297 and R01 NS129850 to J.F.; R21 NS127983 to J.F. and O.R.; R35 NS097370 and RF1 MH123979 to G.-L.M.; and R35 NS116843 to H.S.). N.K. is partially supported by the Uehara Memorial Foundation and International Medical Research Foundation of Japan. O.R. is the incumbent of the Berstein–Mason Professorial Chair of Neurochemistry at the Weizmann Institute of Science.

Author information

Authors and Affiliations

  1. Department of Mechanical Engineering, University of Michigan, Ann Arbor, MI, USA

    Xufeng Xue, Yung Su Kim, Robin Zhexuan Yan, Norio Kobayashi, Shiyu Sun, Yi Zheng, Yue Liu & Jianping Fu

  2. Department of Molecular Genetics, Weizmann Institute of Science, Rehovot, Israel

    Alfredo-Isaac Ponce-Arias, Rami Yair Tshuva & Orly Reiner

  3. Department of Molecular Neuroscience, Weizmann Institute of Science, Rehovot, Israel

    Alfredo-Isaac Ponce-Arias, Richard O’Laughlin, Rami Yair Tshuva & Orly Reiner

  4. Department of Neuroscience and Mahoney Institute for Neurosciences, Perelman School of Medicine, University of Pennsylvania, Philadelphia, PA, USA

    Hongjun Song & Guo-Li Ming

  5. Department of Internal Medicine, Division of Gastroenterology, University of Michigan Medical School, Ann Arbor, MI, USA

    Yu-Hwai Tsai & Jason R. Spence

  6. Wellcome Trust/Cancer Research UK Gurdon Institute, University of Cambridge, Cambridge, UK

    Frederick C. K. Wong, Azim Surani & Jason R. Spence

  7. Department of Physiology, Development and Neuroscience, University of Cambridge, Cambridge, UK

    Azim Surani & Jason R. Spence

  8. Department of Cell and Developmental Biology, University of Michigan Medical School, Ann Arbor, MI, USA

    Jianping Fu

  9. Department of Biomedical Engineering, University of Michigan, Ann Arbor, MI, USA

    Jianping Fu

  10. Department of Cell and Developmental Biology, Perelman School of Medicine, University of Pennsylvania, Philadelphia, PA, USA

    Hongjun Song & Guo-Li Ming

  11. Institute for Regenerative Medicine, University of Pennsylvania, Philadelphia, PA, USA

    Hongjun Song & Guo-Li Ming

  12. The Epigenetics Institute, Perelman School of Medicine, University of Pennsylvania, Philadelphia, PA, USA

    Hongjun Song

  13. Department of Psychiatry, Perelman School of Medicine, University of Pennsylvania, Philadelphia, PA, USA

    Guo-Li Ming

Authors
  1. Xufeng Xue
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Yung Su Kim
    View author publications

    You can also search for this author in PubMed Google Scholar

  3. Alfredo-Isaac Ponce-Arias
    View author publications

    You can also search for this author in PubMed Google Scholar

  4. Richard O’Laughlin
    View author publications

    You can also search for this author in PubMed Google Scholar

  5. Robin Zhexuan Yan
    View author publications

    You can also search for this author in PubMed Google Scholar

  6. Norio Kobayashi
    View author publications

    You can also search for this author in PubMed Google Scholar

  7. Rami Yair Tshuva
    View author publications

    You can also search for this author in PubMed Google Scholar

  8. Yu-Hwai Tsai
    View author publications

    You can also search for this author in PubMed Google Scholar

  9. Shiyu Sun
    View author publications

    You can also search for this author in PubMed Google Scholar

  10. Yi Zheng
    View author publications

    You can also search for this author in PubMed Google Scholar

  11. Yue Liu
    View author publications

    You can also search for this author in PubMed Google Scholar

  12. Frederick C. K. Wong
    View author publications

    You can also search for this author in PubMed Google Scholar

  13. Azim Surani
    View author publications

    You can also search for this author in PubMed Google Scholar

  14. Jason R. Spence
    View author publications

    You can also search for this author in PubMed Google Scholar

  15. Hongjun Song
    View author publications

    You can also search for this author in PubMed Google Scholar

  16. Guo-Li Ming
    View author publications

    You can also search for this author in PubMed Google Scholar

  17. Orly Reiner
    View author publications

    You can also search for this author in PubMed Google Scholar

  18. Jianping Fu
    View author publications

    You can also search for this author in PubMed Google Scholar

Contributions

X.X. and J.F. conceived and initiated the project. X.X. designed, performed and quantified most experiments, including scRNA-seq data analyses and interpretation. Y.S.K. generated the TBXT::T2A-Cre lineage tracing human ES cell line. A.-I.P.-A. generated the TBXT KO human ES cell line. R.O. assisted in developing the μFBLS culture protocol and performed iDISCO whole-mount immunostaining of μFBLS. N.K. and R.Y.T. generated H2B-GFP CDX2 KO human ES cells. Y.-H.T. and J.R.S. provided the CDX2 KO human ES cell line. R.Z.Y. developed Matlab scripts for image processing and independently repeated experiments. Y.Z. helped with scRNA-seq data analyses. S.S. helped with chemical perturbation assays. Y.L., F.C.K.W., A.S., G.-L.M., H.S. and O.R. helped with data interpretation and experimental designs. X.X. and J.F. wrote the manuscript. J.F. supervised the study. All authors edited and approved the manuscript.

Corresponding author

Correspondence to Jianping Fu.

Ethics declarations

Competing interests

The University of Michigan, Ann Arbor, has filed a patent application describing microfluidic devices and methods for the development of NT-like tissues and neural spheroids (PCT/US2021/058090), with J.F. and X.X. as co-inventors. The other authors declare no competing interests.

Peer review

Peer review information

Nature thanks Madeline Lancaster and the other, anonymous, reviewer(s) for their contribution to the peer review of this work.

Additional information

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

Extended data figures and tables

Extended Data Fig. 1 Development of R-C patterned µNTLS.

a. (Left) Side view of a Carnegie Stage (CS) 12 human embryo, showing its head-to-tail length of about 4 mm. Image reproduced from ref. 42, Springer Nature Limited. During human embryogenesis, both rostral and caudal neuropores close at around CS12, leading to a completely closed NT structure at around CS1211. (Right) Transverse sectional image of a CS12 human embryo through somite 3, marked by the dashed red line shown on the left. Image reproduced from the Endowment for the Human Development website (https://www.ehd.org), with permission from R. F. Gasser (1975), all rights reserved. Height of NT along the D-V axis is about 200 µm. b. Schematic of microfluidic device, containing top, central, and bottom microchannels, with both ends of each channel connected with medium reservoirs. In the device center marked by a red rectangle, the three channels are separated by two linear arrays of circular support posts, which defines a patterning region in the central channel marked by a dashed red rectangle. Within the patterning region, stable gradients of chemical signals are established along the length of the central channel (R-C axis) by supplementing different concentrations of chemical factors in the two reservoirs of the central channel. Similarly, through passive diffusion from the top and bottom channels, stable gradients of dorsalizing and ventralizing factors are established perpendicular to the central channel (D-V axis). In the schematic, an array of rectangular colonies of human PS cells is formed in the patterning region. c. Microfluidic device design. All microchannels have a height of 150 µm. Central channel has a width of 4 mm. Circular support posts have a diameter of 100 µm and an edge-to-edge distance of 50 µm. Patterning region in the central channel is defined by a 4 mm × 4 mm square as indicated in b. d. Photograph showing microfluidic devices generated through batch fabrication. e. Schematics and brightfield and confocal images showing microcontact printing to generate rectangular Geltrex adhesive islands, microfluidic device assembly, cell and gel loading into the device, and lumenogenesis of human PS cell colonies to form µNTLS. Specifically, rectangular Geltrex adhesive islands (length: 4 mm; width: 100 µm) are printed onto a coverslip using microcontact printing with a polydimethylsiloxane (PDMS) stamp. A PDMS structural layer is then attached onto the coverslip with Geltrex islands aligned with the patterning region of the central channel. On day 0, dissociated single human PS cells are loaded into the central channel and allowed to adhere to Geltrex islands. One hour after cell seeding, floating human PS cells not attached to Geltrex islands are flushed away gently. On day 1, 100% Geltrex is loaded into the central channel, and a neural induction medium (NIM), comprising basal medium and dual SMAD inhibitors (DSi; see Methods), is added into the two medium reservoirs of the central channel. Colonies of human PS cells self-organize and undergo lumenogenesis, with small lumens, demarcated by ZO-1, emerging on day 2 (see Supplementary Video 1). These lumens grow over time and coalesce with each other. By day 3, human PS cell colonies, which still express OCT4, form an elongated tubular structure containing a single continuous, central apical lumen. Zoom-in views of some marked regions are provided. Arrowheads mark small apical lumens demarcated by ZO-1. f. Protocol for generating R-C patterned µNTLS. Human PS cells are seeded into the central channel on day 0 using mTeSR (Step 1). After gel loading on Day 1, culture medium in the central channel is switched to NIM (Step 2). From day 2 to day 5, CHIR99021 (CHIR, 3 µM), FGF8 (200 ng mL−1) and retinoic acid (RA, 500 nM) are added into the right reservoir of the central channel in addition to NIM (Step 3). From day 5 to day 7, all caudalizing factors are removed, and only NIM is added into the two medium reservoirs of the central channel (Step 4). g. Representative brightfield images showing a single µNTLS on different days as indicated. Zoom-in views of a marked region are provided. h. Representative stitched confocal images showing an array of R-C patterned µNTLS on day 7 from a single microfluidic device stained for HOXB1, HOXB4, and HOXC9. i. Intensity maps showing relative mean expression levels of indicated markers as a function of relative R-C position in R-C patterned µNTLS on day 7. nOTX2 = 12, nHOXB1 = 22, nHOXB4 = 22, nHOXC9 = 12, and nexperiment = 3. j. (Left) Protocol for generating µNTLS with a default dorsal forebrain identity. human PS cells are seeded into the central channel on day 0 using mTeSR (Step 1). After gel loading on Day 1, culture medium is switched to NIM from Day 1 onwards (Step 2). µNTLS are analyzed on Day 7. (Right) Representative stitched confocal micrographs showing µNTLS on Day 7 stained for ZO-1, PAX6, OTX2, and EdU, as indicated. Micrographs on the right show y-z planes of selected regions in µNTLS. k. Representative confocal micrographs showing x-y and y-z planes of selected regions in R-C patterned µNTLS on day 7 stained for ZO-1, ADP-ribosylation factor-like protein 13B (ARL 13B), EdU, and phospho-histone H3 (pH3), respectively. Arrowheads mark ARL 13B-enriched cilia on µNTLS apical surfaces. l. Schematic showing dissection of R-C patterned µNTLS on day 7 using a surgical scissor into four tissue segments of equal lengths for downstream RT-qPCR analysis. m. Dorsal view of human NT and expression pattern of HOX family genes in HB and SC. Color coding of HOX family genes represents their expression domains along the R-C axis of NT. n. Heatmaps showing normalized expression of HOX family genes as a function of the four segments of day 7 R-C patterned µNTLS. n = 3 experiments. In e, h, j and k, nuclei were counterstained with DAPI. In e, g, h, j and k, experiments are repeated three times with similar results. Scale bars, 1 mm (side view image in a), 200 µm (transverse section image in a), 15 mm (d), 1 mm (whole µNTLS array images in e), 50 µm (zoom-in images in e), 800 µm (g and h), 400 µm (j), and 150 µm (k).

Source Data

Extended Data Fig. 2 Effect of morphogen dosages on R-C patterning of µNTLS.

a. Representative stitched confocal micrographs showing R-C patterned µNTLS on day 7 under different concentrations of CHIR as indicated, stained for OTX2, HOXB1, HOXB4, and HOXC9. Concentrations of RA and FGF8 were kept the same as 500 nM and 200 ng mL−1, respectively. Zoom-in views of boxed regions are shown on the right. Intensity maps depict relative mean values of indicated markers as a function of relative R-C position in µNTLS. For 1.5 µM CHIR, nOTX2 = 12, nHOXB1 = 30, nHOXB4 = 30, nHOXC9 = 18, and nexperiment = 3; For 3 µM CHIR, nOTX2 = 12, nHOXB1 = 22, nHOXB4 = 22, nHOXC9 = 12, and nexperiment = 3; For 4.5 µM CHIR, nOTX2 = 14, nHOXB1 = 28, nHOXB4 = 28, nHOXC9 = 14, and nexperiment = 3. b. Representative stitched confocal micrographs showing R-C patterned µNTLS on day 7 under different concentrations of FGF8 as indicated, stained for OTX2, HOXB1, HOXB4, and HOXC9. Concentrations of CHIR and RA were kept the same as 3 µM and 500 nM, respectively. Zoom-in views of boxed regions are shown on the right. Intensity maps depict relative mean values of indicated markers as a function of relative R-C position in µNTLS. For without FGF8, nOTX2 = 13, nHOXB1 = 25, nHOXB4 = 25, nHOXC9 = 12, and nexperiment = 3; For 200 ng mL−1 FGF8, nOTX2 = 12, nHOXB1 = 22, nHOXB4 = 22, nHOXC9 = 12, and nexperiment = 3; For 400 ng mL−1 FGF8, nOTX2 = 10, nHOXB1 = 20, nHOXB4 = 20, nHOXC9 = 10, and nexperiment = 2. c. Representative stitched confocal micrographs showing R-C patterned µNTLS on day 7 treated with or without RA as indicated, stained for OTX2, HOXB1, HOXB4, and HOXC9. Concentrations of CHIR and FGF8 were kept the same as 3 µM and 200 ng mL−1, respectively. Zoom-in views of boxed regions are shown on the right. Intensity maps depict relative mean values of indicated markers as a function of relative R-C position in µNTLS. For without RA, nOTX2 = 14, nHOXB1 = 39, nHOXB4 = 39, nHOXC9 = 25, and nexperiment = 3; For 500 nM RA, nOTX2 = 12, nHOXB1 = 22, nHOXB4 = 22, nHOXC9 = 12, and nexperiment = 3. d. Plot showing relative R-C positions of OTX2+, HOXB1+, HOXB4+, and HOXC9+ domains in R-C patterned µNTLS on day 7 under indicated conditions. Rostral and caudal ends of µNTLS are designated as 0 and 1, respectively. Error bars represent mean ± s.e.m. n values are provided in a-c. In a-c, nuclei were counterstained with DAPI. Scale bars, 400 µm.

Source Data

Extended Data Fig. 3 Dynamic HOX gene expression in R-C patterned µNTLS.

a. Representative stitched confocal micrographs showing µNTLS on different days developed under default caudalizing condition (3 μM CHIR, 500 nM RA, and 200 ng mL−1 FGF8) stained for OTX2 and HOXB1, HOXB1, HOXB4, and HOXC9, and OCT4 and SOX2, respectively. Zoom-in views of boxed regions are shown on the right. Arrowheads mark nuclear staining of HOXB4 in day 6 µNTLS. Intensity maps depict relative mean expression levels of indicated markers as a function of relative R-C position in µNTLS. For intensity maps, on day 3, nOTX2 = 15, nHOXB1 = 15, and nexperiment = 3; on day 4, nOTX2 = 15, nHOXB1 = 15, and nexperiment = 3; on day 5, nOTX2 = 10, nHOXB1 = 31, nHOXC9 = 21, and nexperiment = 3; on day 6, nOTX2 = 13, nHOXB1 = 23, nHOXB4 = 11, nHOXC9 = 27, and nexperiment = 3; on day 7, nOTX2 = 12, nHOXB1 = 22, nHOXB4 = 22, nHOXC9 = 12, and nexperiment = 3. b. Plots showing expression levels of OCT4, NANOG, SOX1/2, and PAX6 in R-C patterned µNTLS as a function of time. Error bars represent mean ± s.e.m. nexperiment = 4 for OCT4, NANOG, SOX1/2, and PAX6. One-way ANOVA tests were performed, followed by Tukey’s multiple comparison tests to calculate p values. c. Representative stitched confocal micrographs showing µNTLS development under high RA condition (3 μM CHIR, 2,000 nM RA, and 200 ng mL−1 FGF8) on different days stained for HOXB1, HOXB4, and HOXC9. Zoom-in views of boxed regions are shown on the right. Intensity maps depict relative mean expression levels of indicated markers as a function of relative R-C position in µNTLS. For intensity maps, on day 4, nHOXB1 = 14 and nexperiment = 3; on day 5, nHOXB1 = 13, nHOXB4 = 13, and nexperiment = 3; on day 6, nHOXB1 = 9, nHOXB4 = 9, nHOXC9 = 9, and nexperiment = 3; on day 7, nHOXB1 = 9, nHOXB4 = 12, nHOXC9 = 15, and nexperiment = 3. d. Plots showing relative R-C positions of OTX2+, HOXB1+, HOXB4+, and HOXC9+ domains in R-C patterned µNTLS under default (left) and high RA (right) conditions as a function of time. Rostral and caudal ends of µNTLS are designated as 0 and 1, respectively. Error bars represent mean ± s.e.m. n values are provided in a & c. In a and c, nuclei were counterstained with DAPI. Scale bars, 400 µm.

Source Data

Extended Data Fig. 4 Development of neuromesodermal progenitors (NMPs) and secondary organizers in R-C patterned µNTLS.

a. Representative stitched confocal micrographs showing µNTLS on different days stained for SOX2, CDX2, and BRACHYURY (BRA). Zoom-in views of µNTLS caudal ends are shown on the right, with arrowheads marking SOX2+BRA+ NMPs. Intensity maps depict relative mean expression levels of indicated markers as a function of relative R-C position in µNTLS. ncolony = 15, 20, 14, and 14 for day 3, day 4, day 5, and day 6, respectively. nexperiment = 3. b. Plot showing relative R-C positions of CDX2+ and BRA+ domains in µNTLS as a function of time. Rostral and caudal ends of µNTLS are designated as 0 and 1, respectively. Error bars represent mean ± s.e.m. n values are provided in a. c. Live imaging with a BRACHYURY-mNeonGreen human ES cell reporter line to track dynamic BRACHYURY expression at µNTLS caudal ends. Arrowheads mark BRACHYURY-mNeonGreen+ cells. Experiments were repeated three times with similar results. See Supplementary Video 3. d. Protocols for deriving presomitic mesodermal (PSM) and motor neuron progenitor (pMN) cells from cells isolated from caudal ends of day 4 µNTLS. Caudal regions of day 4 µNTLS were physically dissected using a surgical scissor and were re-plated and cultured under PSM or pMN differentiation protocols for another 4 days as indicated. For PSM differentiation, basal medium (BM) is supplemented with CHIR (3 µM) and LDN (500 nM). For pMN differentiation, BM is supplemented with Smoothened Agonist (SAG, SHH agonist; 1 µM) and RA (1 µM). e. Representative confocal micrographs showing cell colonies after 4 days of culture under PSM or pMN differentiation protocols as indicated. Cells were stained for PSM marker TBX6 and pMN marker OLIG2. Experiments were repeated three times with similar results. f. RT-qPCR analysis of caudal regions of day 4 R-C patterned µNTLS, which contain NMPs, and cells cultured for 4 days under either PSM differentiation protocol (PSM protocol) or pMN differentiation protocol (pMN protocol). Heatmaps show normalized expression of NMP, PSM, and pMN markers as indicated. n = 4 experiments. g. Schematic showing generation of a TBXT::T2A-Cre lineage tracer hESC line. h. (Top) Protocol for deriving NMPs from TBXT::T2A-Cre lineage tracer cells followed by pMN induction. TBXT::T2A-Cre lineage tracer cells were seeded as single cells onto culture dishes at a density of 1.5 × 104 cells cm−2 in mTeSR containing Y27632 (10 μM). Culture medium was switched to basal medium supplemented with CHIR (3 µM) and FGF8 (200 ng mL−1) from day 1 to day 3 to promote NMP differentiation. From Day 3 to Day 5, culture medium was switched to basal medium supplemented with RA (500 nM) and SAG (500 nM) to induce pMN differentiation. (Bottom) Representative confocal micrographs showing cells on Day 3 stained for BRA and SOX2 and on Day 5 stained for BRA and OLIG2, respectively. Experiments were repeated three times with similar results. i. Live imaging with the TBXT::T2A-Cre lineage tracer to track dynamic NMP development at µNTLS caudal ends between day 3 and day 7. Experiments were repeated three times with similar results. See Supplementary Video 4. j. Representative confocal micrographs showing caudal ends of day 7 R-C patterned µNTLS generated from the TBXT::T2A-Cre lineage tracer, stained for HOXC9. White arrowheads mark ZsGreen+HOXC9+ cells. Experiments were repeated three times with similar results. k. Plot showing relative length of ZsGreen+ domains in R-C patterned µNTLS as a function of time. Length of ZsGreen+ domain is normalized by the total length of µNTLS. Error bars represent mean ± s.e.m. ncolony = 9 and nexperiment = 3. One-way ANOVA tests were performed, followed by Tukey’s multiple comparison tests to calculate p values. l. Schematic showing dissection of day 7 R-C patterned µNTLS using a surgical scissor into four tissue segments of equal lengths for downstream RT-qPCR analysis. m. Heatmaps showing normalized expression of FB, MB, HB, and MB-HB boundary markers and genes related to RA signaling, as a function of the four segments of day 7 µNTLS. n = 3 experiments. In a, e, h and j, nuclei were counterstained with DAPI. Scale bars, 400 µm (a), 100 µm (c and e), and 50 µm (h, i and j).

Source Data

Extended Data Fig. 5 Roles of CDX2 and TBXT in R-C patterning of µNTLS.

a. Representative stitched confocal micrographs showing R-C patterned µNTLS on day 4, generated from wild type (WT) or CDX2-knockout (CDX2-KO) H9 human ES cell lines as indicated, stained for SOX2, CDX2, and BRACHYURY (BRA). Zoom-in views of boxed regions are shown on the right. Intensity maps depict relative mean values of indicated markers as a function of relative R-C position in µNTLS. White arrowheads mark SOX2+BRA+ NMPs. n ≥ 2 experiments for each condition. For WT, ncolony = 20 and nexperiment = 3; for CDX2-KO, ncolony = 14 and nexperiment = 2. b. Representative stitched confocal micrographs showing R-C patterned µNTLS on day 7, generated from WT or CDX2-KO H9 human ES cell lines as indicated, stained for HOXB1, HOXB4, and HOXC9. Zoom-in views of boxed regions are shown on the right. Intensity maps depict relative mean values of indicated markers as a function of relative R-C position in µNTLS. n ≥ 2 experiments for each condition. For WT, nHOXB1 = 22, nHOXB4 = 22, nHOXC9 = 12, and nexperiment = 3; for CDX2-KO, nHOXB1 = nHOXB4 = nHOXC9 = 13, and nexperiment = 2. c. Plot showing relative R-C positions of BRA+ domains in day 4 R-C patterned µNTLS and HOXB1+, HOXB4+, and HOXC9+ domains in day 7 R-C patterned µNTLS, generated from either WT or CDX2-KO human ES cells as indicated. Rostral and caudal ends of µNTLS are designated as 0 and 1, respectively. Error bars represent mean ± s.e.m. n values are provided in a & b. Two-sided Student’s t-tests were performed to calculate p values. d. (Left) Cartoon illustrating definition of normalized contour length, calculated as the ratio between contour length and region length. (Right) Box-and-whisker plot showing normalized contour length of µNTLS generated from WT and CDX2-KO human ES cells in indicated R-C regions (box: 25–75%; bar-in-box: median; whiskers: 1.5 × interquartile range). For WT, ncolony = 10 and nexperiment = 2; For CDX2-KO, ncolony = 11 and nexperiment = 2; Two-sided Student’s t-tests were performed to calculate p values. e. (Left) Protocol for clonal growth assay. Single H2B-GFP human ES cells are mixed with non-fluorescent human ES cells during cell seeding at a ratio of 1:200. Length of H2B-GFP cell colonies is recorded daily. (Right) Representative brightfield and fluorescence images showing clonal growth of single WT and CDX2-KO H2B-GFP human ES cells in µNTLS from day 1 to day 6. Experiments were repeated three times with similar results. f. Box-and-whisker plot showing length of H2B-GFP cell colonies in different R-C regions of day 6 µNTLS generated from WT and CDX2-KO human ES cells (box: 25–75%; bar-in-box: median; whiskers: 1.5 × interquartile range). Only clonal growth from a single H2B-GFP human ES cell is included for quantification. For WT, ncolony = 10, 11, 10, and 10 for Region 1, 2, 3, and 4, respectively, and nexperiment = 2; For CDX2-KO, ncolony = 12, 11, 14, and 14 for Region 1, 2, 3, and 4, respectively, and nexperiment = 2. Two-sided Student’s t-tests were performed to calculate p values. g. Representative stitched confocal micrographs showing day 4 R-C patterned µNTLS generated from WT or TBXT-KO WIBR3 human ES cells as indicated, stained for SOX2, CDX2, and BRA. Zoom-in views of boxed regions are shown on the right. Arrowheads mark SOX2+BRA+ NMPs. Intensity maps depict relative mean values of indicated markers as a function of relative R-C position in µNTLS. For WT, ncolony = 20 and nexperiment = 3; For TBXT-KO, ncolony = 18 and nexperiment = 3. h. Representative stitched confocal micrographs showing day 7 R-C patterned µNTLS generated from WT or TBXT-KO WIBR3 human ES cells as indicated, stained for HOXB1, HOXB4, and HOXC9. Zoom-in views of boxed regions are shown on the right. Intensity maps depict relative mean values of indicated markers as a function of relative R-C position in µNTLS. For WT, ncolony = 14 and nexperiment = 2; for TBXT-KO, ncolony = 12 and nexperiment = 2. i. Plot showing relative R-C positions of CDX2+ domain in day 4 R-C patterned µNTLS and HOXB1+, HOXB4+, and HOXC9+ domains in day 7 R-C patterned µNTLS, generated from either WT or TBXT-KO human ES cells. Rostral and caudal ends of µNTLS are designated as 0 and 1, respectively. Error bars represent mean ± s.e.m. n values are provided in g & h. Two-sided Student’s t-tests were performed to calculate p values. In a, b, g and h, nuclei were counterstained with DAPI. Scale bars, 400 µm (a, b, g and h) and 200 µm (e).

Source Data

Extended Data Fig. 6 Development of region-specific µNS.

a. Protocol for generating region-specific µNS. Human PS cells are seeded into the central channel on day 0 using mTeSR (Step 1). After gel loading on Day 1, culture medium of the central channel is switched to NIM (Step 2). From Day 2 to Day 5, CHIR (3 µM), FGF8 (200 ng mL−1), and RA (500 nM) are added into the right reservoir of the central channel in addition to NIM, to induce caudalization of µNS (Step 3). From day 5 to day 7, all caudalizing factors are removed, and only NIM is added into the two medium reservoirs of the central channel (Step 4). Tissues are analyzed at different time points as indicated. b. Representative stitched brightfield (left) and confocal (right) micrographs showing a regular array of µNS in the patterning region on indicated days. µNS on day 2 and day 4 was stained for OCT4, SOX2, and ZO-1. On day 7, they were stained for OTX2, HOXB1, and HOXB4. Zoom-in views of boxed regions are shown on the right. Experiments were repeated three times with similar results. c. Representative confocal micrographs showing µNS on day 7 stained for PAX6, OTX2, HOXB1, HOXB4, and HOXC9 as indicated. Experiments were repeated three times with similar results. d. Pie charts showing percentages of different types of µNS at different locations of the patterning region on day 7. n = 3 experiments. In b and c, nuclei were counterstained with DAPI. Scale bars, 400 µm (b) and 100 µm (c).

Source Data

Extended Data Fig. 7 Development and characterization of R-C and D-V patterned µNTLS.

a. Protocol for generating R-C and D-V patterned µNTLS. Human PS cells are seeded into the central channel on day 0 using mTeSR (Step 1). After gel loading on day 1, culture medium in the central channel is switched to NIM (Step 2). From day 2 to day 5, CHIR (3 µM), FGF8 (200 ng mL−1), and RA (500 nM) are added into the right reservoir of the central channel in addition to NIM, to induce caudalization and R-C patterning of µNTLS (Step 3). From day 5 to day 9, BMP4 (25 ng mL−1) and RA (500 nM) / smoothened agonist (SAG, 500 nM), unless otherwise specified, are added into the top and bottom channels, respectively, to induce D-V patterning of µNTLS. b. Representative brightfield images showing R-C and D-V patterned µNTLS on day 9 and day 21 as indicated. Experiments were repeated five times with similar results. c. Representative stitched confocal micrographs showing R-C and D-V patterned µNTLS on day 9 stained for OTX2, HOXB1, HOXB4, and HOXC9. Zoom-in views of boxed regions are shown on the bottom. Experiments were repeated five times with similar results. d. Representative confocal micrographs showing transverse sections of rostral (d’) and caudal (d”) SC regions of day 9 R-C and D-V patterned µNTLS as indicated, stained for HOXB1, HOXB4, HOXC9, SOX10, PAX3, OLIG2, NKX2.2, and FOXA2. Zoom-in views of boxed regions are included. Intensity maps depict relative mean values of indicated markers as a function of relative D-V position in µNTLS. For rostral SC regions, nPAX3 = 20, nOLIG2 = 24, and nexperiment = 5. For caudal SC regions, nPAX3 = 22, nOLIG2 = 38, and nexperiment = 5. e. Plot showing percentages of µNTLS containing a single lumen in their rostral and caudal regions on day 9 and day 21 as indicated. Error bars represent mean ± SEM. nexperiment = 3. f. Representative confocal micrographs showing transverse sections of rostral (f’) and caudal (f”) regions of day 21 R-C and D-V patterned µNTLS cultured under different conditions as indicated, stained for ZO-1, PAX6, DLX2, NKX2-1, SOX10, PAX3, OLIG2, NKX2-2, and FOXA2. From day 2 to day 5, CHIR (3 µM), FGF8 (200 ng mL−1) and RA (500 nM) are added into the right reservoir of the central channel to induce R-C patterning of µNTLS. From day 5 to day 9, BMP4 (25 ng mL−1) and RA (500 nM or 2,000 nM) / smoothened agonist (SAG, 500 nM or 2,000 nM) are added into the top and bottom channels, respectively, to induce D-V patterning of µNTLS. Zoom-in views of boxed regions are included. Intensity maps depict relative mean values of indicated markers as a function of relative D-V position in µNTLS. For 2,000 nM RA / 2,000 nM SAG condition, nPAX6-rostral = 20, nDLX2 = 20, nNKX2-1 = 20, nPAX6-caudal = 21, nOLIG2 = 40, nNKX2-2 = 19, nFOXA2 = 19, and nexperiment = 5; for 500 nM RA / 500 nM SAG condition, nPAX3 = 20, nOLIG2 = 41, nNKX2-2 = 21, nFOXA2 = 21, and nexperiment = 5. In c, d, and f, nuclei were counterstained with DAPI. Scale bars, 400 µm (b and c) and 100 µm (d and f).

Source Data

Extended Data Fig. 8 Characterization of µNTLS using single-cell RNA-sequencing (scRNA-seq).

a. scRNA-seq assay design. R-C patterned µNTLS on day 4 and R-C and D-V patterned µNTLS on day 9 and day 21 were dissociated into single cells before the cells were sequenced using 10× Genomics and Illumina HiSeq 6000 (see Methods). Single-cell transcriptome data of day 4, day 9 and day 21 µNTLS were then integrated and analyzed. The number of cells from each sample after data filtering, average UMI counts, and average detected genes are listed. b. UMAP of integrated single-cell transcriptome data of day 4, day 9 and day 21 µNTLS, color-coded according to cell identity annotations (left) or time points (right). n indicates total cell number combined from all three time points. FB, forebrain; MB, midbrain; IsO, isthmic organizer; HB, hindbrain; SC, spinal cord; NMP, neuromesodermal progenitors; RP, roof plate; FP, floor plate; NC, neural crest. c. UMAP of single-cell transcriptome data of day 4, day 9, and day 21 µNTLS, separated from integrated UMAP plots in b. n indicates cell numbers at each time point. d. Alluvial plot showing percentages of cells for each cell cluster in µNTLS on day 4, day 9, and day 21. e. Heatmap of relative expression (Z-score) of top-20 gene signatures distinguishing each cell cluster. All genes are listed in Supplementary Table 1. f. Heatmap of average relative expression (Z-score) of selected genes among indicated cell clusters. g. Feature plots showing expression of selected genes associated with indicated cell cluster annotations in UMAP plots of integrated datasets from day 4, day 9, and day 21 µNTLS.

Source Data

Extended Data Fig. 9 Transcriptomic comparison between µNTLS and neural cells from human embryonic tissues.

a. (Top left) UMAP showing integrated data from day 9 µNTLS and neural cells in Carnegie Stage (CS) 12 human embryos42. (Top middle & right) UMAP plots of day 9 µNTLS data and CS12 human neural cell data, separated from the integrated UMAP plot, as indicated. n indicates cell numbers. (Bottom) Feature plots comparing expression of key marker genes between day 9 µNTLS and neural cells in CS12 human embryos. b. (Top) UMAP plots of Neuron clusters in day 9 µNTLS and CS12 human embryo datasets. Neuron cluster in day 9 µNTLS dataset is projected onto the Neuron cluster in CS12 human embryo dataset and is annotated following the human reference data42 (see Methods). n indicates cell numbers. (Bottom) Feature Plots comparing expression of key marker genes between Neuron clusters in day 9 µNTLS and CS12 human embryos. c. (Top left) UMAP showing integrated data from day 21 µNTLS and neural cells in CS15-16 human embryos42. (Top middle & right) UMAP plots of day 21 µNTLS data and CS15-16 human neural cells, separated from the integrated UMAP plot, as indicated. n indicates cell numbers. (Bottom) Feature plots compare expression of key marker genes between day 21 µNTLS and neural cells in CS15-16 embryos. d. (Top) UMAP plots of Neuron clusters in day 21 µNTLS and CS15-16 human embryo datasets. Neuron cluster in day 21 µNTLS dataset is projected onto the Neuron cluster in CS15-16 human embryo dataset and is annotated following the human reference data42 (see Methods). n indicates cell numbers. (Bottom) Feature Plots comparing expression of key marker genes between Neuron clusters in day 21 µNTLS and CS15-16 human embryos. e. Pearson’s correlation analysis of cell clusters in day 9 and day 21 µNTLS with neural clusters in CS12-16 human embryo datasets42. Black boxes highlight the highest correlation coefficients in each column. Correlation coefficients between indicated µNTLS and human clusters are calculated based on variable genes identified from human neural cell clusters (Supplementary Table 1). f. Pearson’s correlation analysis of Neuron subclusters in day 9 and day 21 µNTLS with Neuron subclusters in CS12-16 human embryo datasets42. Black boxes highlight the highest correlation coefficients in each column. Correlation coefficients between indicated µNTLS and human clusters are calculated based on variable genes identified from human neural cell clusters. g. Alluvial plots showing percentages of cells for each cell cluster in day 9 and day 21 µNTLS (top) and human embryos at CS12, CS13-14 and CS15-16 (bottom). h. day 4, day 9, and day 21 µNTLS show closest transcriptome similarities with human neural cells at CS8-9 (day 17–20), CS12 (day 26–30), and CS15-16 (day 35–42), respectively42. Note that corresponding human embryo stage of day 4 μNTLS is inferred from the comparison between day 4 μNTLS and mouse data44 (See Supplementary Figs. 8–10).

Source Data

Extended Data Fig. 10 Analysis of Forebrain (FB) cluster and Spinal cord (SC)-related cells in µNTLS.

a. D-V patterning of FB, leading to formation of dorsal pallium and ventral subpallium domains. Pallium is marked by PAX6 and EMX2 expression, whereas subpallium is marked by NKX2-1. b. UMAP plots of FB clusters in CS15-16 human embryo data (left) and day 21 µNTLS data (right). FB cluster in day 21 µNTLS data is annotated following human reference data42 (see Methods). c. Stacked bar plots showing percentages of cells for each FB subcluster in CS15-16 human FB cluster and day 21 µNTLS FB cluster. d. Dot plot showing expression of key marker genes across different cell subclusters in day 21 µNTLS FB cluster and CS15-16 human embryo FB cluster as indicated. Dot sizes and colors indicate proportions of cells expressing corresponding genes and their averaged scaled values of log-transformed expression, respectively. e. Heatmap showing relative expression (Z-score) of top-30 gene signatures calculated from CS15-16 human FB cells in both CS15-16 human FB cluster and day 21 µNTLS FB cluster as indicated. For gene signature information, see Supplementary Table 1. f. Enrichment of Gene Ontology (GO) terms in DEGs upregulated in day 21 µNTLS FB cells or CS15-16 human FB cells as indicated. For detailed information about DEGs and GO terms, see Supplementary Table 1. g. In vivo, dorsal SC gives rise to roof plate (RP) and six neuronal progenitor domains (dp1-dp6), whereas ventral SC generates floor plate (FP) and five neuronal progenitor domains (pMN and p0-p3). These domains express distinct combinations of transcription factors. Markers listed on the right for each domain are identified from scRNA-seq data of mouse SC between E9.5 - E13.567. h. UMAP of SC-related cells from integrated dataset of day 4, day 9, and day 21 µNTLS, color-coded according to time points (left) or subcluster identity annotations (right). n indicates cell number. Note that cells from SC cluster as well as those from RP and FP clusters that express HOX4-13 genes have been included in analyses in h-k. i. Dot plot showing expression of key marker genes across subclusters of SC-related cells as indicated. Dot sizes and colors indicate proportions of cells expressing corresponding genes and their averaged scaled values of log-transformed expression, respectively. j. Feature plots showing average expression of indicated markers associated with each SC progenitor domain in UMAP plots of SC-related cells. k. Heatmap of average relative expression (Z-score) of top-20 gene signatures distinguishing each cell subcluster in SC-related cells. For gene signature information, see Supplementary Table 1.

Source Data

Extended Data Fig. 11 Characterization of neural crest (NC) development in µNTLS.

a. Representative confocal micrographs showing NC cells from R-C and D-V patterned µNTLS on day 11 stained for PAX7, SNAI2, and SOX10. Dashed lines mark dorsal boundaries of µNTLS. Experiments were repeated twice with similar results. b. Representative confocal micrographs showing NC cells and their derivatives in rostral and caudal regions of R-C and D-V patterned µNTLS on day 11, stained for SOX10, TWIST1, PHOX2B, ISL1, and S100B. Dashed lines mark dorsal boundaries of µNTLS. Experiments were repeated three times with similar results. c. Plots showing percentages of TWIST1+, PHOX2B+, ISL1+, and S100B+ cells among all disseminated cells in rostral and caudal halves of µNTLS. Error bar represents mean ± s.e.m. nTWIST1 = nPHOX2B = nISL1 = nS100B = 8 and nexperiment = 2. Two-sided Student’s t-tests were performed to calculate p values. d. Protocol for generating control R-C and D-V patterned µNTLS and ectopic caudalization of µNTLS. From day 2 to day 5, CHIR (3 µM), FGF8 (200 ng mL−1), and RA (500 nM) are supplemented into the right reservoir of the central channel in addition to NIM to induce caudalization and R-C patterning of µNTLS. From day 5 to day 9, BMP4 (25 ng mL−1) and RA (500 nM) / smoothened agonist (SAG, 500 nM) are supplemented into the top and bottom channels, respectively, to induce D-V patterning of µNTLS. For ectopic caudalization of µNTLS, caudalizing factors CHIR (3 µM), FGF8 (200 ng mL−1), and RA (500 nM) are added together into the left reservoir of the center channel from day 5 to day 9. Culture medium in all reservoirs is then switched back to fresh basal medium (BM) from day 9 to day 11 to allow for further development of NC cells. e. Representative stitched confocal micrographs showing day 5 µNTLS cultured following the protocol in d, stained for OTX2, HOXB1, and SOX10. Zoom-in views of boxed regions are shown on the right. Experiments were repeated three times with similar results. f. Live imaging with the TBXT::T2A-Cre lineage tracer to track progenies of NMPs during the development of R-C and D-V patterned µNTLS from day 5 to day 11 (see Supplementary Video 6). Only caudal ends of µNTLS are monitored as indicated. Dashed lines mark dorsal boundaries of µNTLS. Zoom-in views of boxed regions are shown. Experiments were repeated three times with similar results. g. Representative confocal micrographs showing caudal ends of R-C and D-V patterned µNTLS generated from the TBXT::T2A-Cre lineage tracer, stained on day 11 for SOX10, HOXC9, ISL1, and PHOX2B. Dashed lines mark dorsal boundaries of µNTLS. White arrowheads mark ZsGreen+ NC cells, and yellow arrowheads mark ZsGreen− NC cells. Experiments were repeated twice with similar results. h. Plot showing percentages of ZsGreen+ cells among all HOXC9+ trunk NC cells, ISL1+ sensory neurons, or PHOX2B+ sympathetic neurons. nHOXC9 = 11, nISL1 = 9, nPHOX2B = 11, and nexperiment = 2. Error bars represent mean ± s.e.m. i. Representative confocal micrographs showing NC cells in caudal regions of R-C and D-V patterned µNTLS on day 11 generated from wild type (WT) and CDX2-KO human ES cell lines, stained for indicated markers. Dashed lines mark dorsal boundaries of µNTLS. Zoom-in views of boxed regions are shown. Experiments were repeated twice with similar results. In a, b, e, g and i, nuclei were counterstained with DAPI. Scale bars, 100 µm (a, b, and i), 200 µm (e and f), and 50 µm (g).

Source Data

Extended Data Fig. 12 Subclustering analysis and trajectory inference of NC cluster in µNTLS.

a. UMAP of NC cluster from integrated dataset of day 4, day 9, and day 11 µNTLS, color-coded according to time points (left) or cell subcluster identity annotations (right). Seven cell subclusters are identified, including Premigratory NC (Cluster 0), Delaminating NC (Cluster 1), Schwann cell (Cluster 2), Mesenchymal cell (Cluster 3), Sensory neuron (Cluster 4), Sympathetic neuron (Cluster 5), and Melanoblast (Cluster 6). n indicates cell number. b. Dot plot showing expression of key marker genes across all NC subclusters as indicated. Dot sizes and colors indicate proportions of cells expressing corresponding genes and their averaged scaled values of log-transformed expression, respectively. c. Feature plots showing expression of selected markers associated with indicated cell subclusters in UMAP plots of NC cluster. d. Heatmap of average relative expression (Z-score) of top-20 gene signatures distinguishing each cell subcluster in NC cluster. For gene signature information, see Supplementary Table 1. e. Dot plot showing expression of HOX genes in HOX− cranial NC, HOX+ cranial NC, vagal NC, and trunk NC. HOX− cranial NC doesn’t express any HOX genes. HOX+ cranial NC expresses HOX paralogous group (PG) 1-2 but not HOX PG 3-13. Vagal NC expresses HOX PG 3-5 but not HOX PG 6-13. Trunk NC expresses HOX PG 6-9 but not HOX PG 10-13. Dot sizes and colors indicate proportions of cells expressing corresponding genes and their averaged scaled values of log-transformed expression, respectively. f. Stacked bar plot showing cellular compositions in HOX- cranial, HOX+ cranial, vagal, and trunk NC cells, as indicated. g. Dot plot comparing expression of key marker genes in HOX- cranial NC, HOX+ cranial NC, vagal NC, and trunk NC across different NC clusters as indicated. Dot sizes and colors indicate proportions of cells expressing corresponding genes and their averaged scaled values of log-transformed expression, respectively. h. UMAP of HOX- cranial NC cells separated from NC cluster shown in a, color-coded according to time points (top) or cell subcluster identity annotations (bottom). Six cell subclusters are identified, including Premigratory NC (Cluster 0), Delaminating NC (Cluster 1), Schwann cell (Cluster 2), Mesenchymal cell (Cluster 3), Sensory neuron (Cluster 4), and Melanoblast (Cluster 6). n indicates cell number. i. UMAP of cell subclusters of HOX- cranial NC cells associated with lineage developments of Schwann cell, Mesenchymal cell, Sensory neuron, and Melanoblast, color-coded according to pseudotime. Solid lines represent principal curves of each lineage. Pseudotime values are computed by projecting each single cell onto principal curves. j. Heatmap of smoothened expression of all differentially expressed genes (DEGs) along pseudotime of Schwann cell, Mesenchymal cell, Sensory neuron and Melanoblast lineage development trajectories in HOX- cranial NC cells. Selected genes are highlighted. A gene is considered significant when adjusted p-value based on FDR is <0.05 (see Methods). For DEG information, see Supplementary Table 1.

Source Data

Extended Data Fig. 13 Development of microfluidic forebrain-like structure (μFBLS).

a. Transverse view of forebrain, with dorsal pallium and ventral subpallium. Both pallium and subpallium can be divided into ventricular zone (VZ), subventricular zone (SVZ) and marginal zone (MZ) from the apical to basal surface. Different domains in pallium and subpallium express distinct combinations of transcription factors as indicated. b. Protocol for generating µFBLS. Human PS cells are seeded into the central channel on day 0 using mTeSR, and Geltrex is loaded into the central channel on day 1 to provide 3D culture environment. After gel loading on day 1, culture medium is switched to NIM from day 1 to day 5 to induce neural differentiation. From day 5 to day 9, to generate dorsal µFBLS, NIM are added into all reservoirs connecting the three channels; to generate ventral µFBLS, smoothened agonist (SAG, 500 nM) is supplemented into NIM in all reservoirs; to generate D-V patterned µFBLS, BMP4 (25 ng mL−1) and SAG (500 nM) are supplemented into basal medium in the top and bottom channels, respectively. After regionalization of µFBLS in the microfluidic device, PDMS structural layers are detached manually from the coverslip on day 9, with µFBLS remaining on PDMS structural layers. µFBLS on PDMS structural layers are continuously cultured in basal medium supplemented with insulin (2.5 µg mL−1) and 1% Matrigel till day 40 (See Methods). c. (Top) Representative brightfield images showing dorsal, ventral and D-V patterned µFBLS on day 40 as indicated. (Bottom) Box-and-whisker plots showing projected areas of dorsal (left), ventral (middle) and D-V patterned (right) µFBLS on indicated days (box: 25–75%; bar-in-box: median; whiskers: 1.5 × interquartile range). ndorsal = 12, 12, 11, and 12 on day 22, day 28, day 34, and day 40, respectively; nventral = 16, 20, 11, and 13 on day 22, day 28, day 34, and day 40, respectively; nD-V = 32, 19, 20, and 15 on day 22, day 28, day 34, and day 40, respectively. nexperiment = 3. d. Plot showing percentages of dorsal, ventral, and D-V patterned µFBLS on day 40 with a single lumen. nexperiment = 4, 5, and 5 for dorsal, ventral, and D-V patterned µFBLS, respectively. Error bars represent mean ± s.e.m. e. Representative confocal micrographs showing sections of dorsal µFBLS on day 40, stained for FOXG1, REELIN, ZO-1, PAX6, and DLX2. Experiments were repeated three times with similar results. f. Representative confocal micrographs showing sections of dorsal µFBLS on day 40 stained for ZO-1, PAX6, TBR1/2, CTIP2, and SATB2. Experiments were repeated three times with similar results. g. Representative confocal micrographs showing sections of ventral µFBLS on day 40, stained for FOXG1, ZO-1, ASCL1, MEIS2, and CTIP2. Experiments were repeated three times with similar results. h. Representative confocal micrographs showing sections of ventral µFBLS on day 40, stained for GSX2, DLX2, CTIP2, ASCL1, and MEIS2. Experiments were repeated three times with similar results. i. Representative confocal micrographs showing sections of D-V patterned µFBLS on day 40, stained for ZO-1, PAX6, TBR1, DLX2, ASCL1, CTIP2 and MEIS2. Zoom-in views of boxed regions are shown on the right. Intensity maps depict relative mean values of indicated markers as a function of relative D-V position in µFBLS. Experiments were repeated three times with similar results. In e-i, nuclei were counterstained with DAPI. Scale bars, 200 µm (c, e and i (left)), 50 µm (f and i (zoom-in images)), 100 µm (g), and 30 µm (h).

Source Data

Extended Data Fig. 14 Single-cell transcriptome analysis of D-V patterned µFBLS.

a. UMAP of single-cell transcriptome data of day 40 D-V patterned µFBLS, color-coded according to cell identity annotations. Nine cell clusters are identified, including dorsal radial glia (dRG), ventral radial glia (vRG), outer radial glia (oRG), excitatory intermediate progenitor(IP-EN), inhibitory intermediate progenitor (IP-IN), newborn excitatory neuron (nEN), and newborn inhibitory neuron 1 / 2 / 3 (nIN1, nIN2 and nIN3). n indicates cell number. b. Feature plots showing expression of selected markers associated with indicated cell identities in UMAP plots of single-cell transcriptome data of day 40 D-V patterned µFBLS. Zoom-in views of boxed regions are shown. Arrowheads mark cortical hem-like cells showing notable expression of RSPO2 and LMX1A. c. Principal component analysis (PCA) of pallium cells in day 40 D-V patterned µFBLS relative to published data of pallium cells in human forebrain at different timepoints as indicated50. Pallium cells in µFBLS include dRG, IP-EN, and nEN clusters isolated from D-V patterned µFBLS dataset. Reference cortical clusters expressing EMX1 are isolated from human brain datasets50. d. Pearson’s correlation analysis of pallium cells (dRG, IP-EN, and nEN clusters) in day 40 D-V patterned µFBLS with pallium cells in PCW11.5 human brain dataset50. Black boxes highlight the highest correlation coefficients in each column. Original annotations of human brain cells from the reference dataset are used here (RG or radial glia / IPC or intermediate progenitor cell / Neuroblast / EN or excitatory neuron). Correlation coefficients between paired µFBLS and human clusters are calculated based on variable genes identified from human pallium cell clusters. e. Dot plot comparing expression of key marker genes in different cell clusters of day 40 µFBLS and PCW11.5 human pallium cells as indicated. Dot sizes and colors indicate proportions of cells expressing corresponding genes and their averaged scaled values of log-transformed expression, respectively. f. Venn diagram of differentially expressed genes (DEGs) between intermediate neural progenitor cells (IPC) from day 40 µFBLS and PCW11.5 human pallium cells, with 192 shared genes including some commonly used IPC markers (TBR2, NEUROD4 and NEUROG1). Enriched Gene Ontology (GO) terms in each compartment of the Venn diagram are shown on the right. For information about DEGs and GO terms, see Supplementary Table 1. g. Principal component analysis (PCA) of inhibitory neurons in day 40 D-V patterned µFBLS relative to published data of inhibitory neurons from human forebrain at different timepoints as indicated50. Inhibitory neurons in µFBLS include nIN1, nIN2 and nIN3 clusters isolated from D-V patterned µFBLS dataset. Inhibitory neurons from human forebrain include telencephalic clusters expressing DLX2 and are extracted from original human brain datasets50. h. Pearson’s correlation analysis of inhibitory neuron clusters in day 40 D-V patterned µFBLS with those in PCW12 human brain dataset50. Original annotations of human brain cells from the references are used here (LGE or lateral ganglionic eminence; MGE or medial ganglionic eminence; CGE or caudal ganglionic eminence). Correlation coefficients between paired µFBLS and human cell clusters are calculated based on variable genes identified from human inhibitory neuron clusters. i. (Left) UMAP projection of scRNA-seq data of inhibitory neurons from day 40 D-V patterned µFBLS (nIN1 / 2 / 3), with cell identity annotations indicated. j. Feature plots showing expression of LGE-, MGE- and CGE-associated inhibitory neuron markers in inhibitory neurons from day 40 D-V patterned µFBLS. k. UMAP of single-cell transcriptome data of day 40 D-V patterned µFBLS, color-coded according to cell identity annotations. Grey arrows indicate predicted future cell states calculated using RNA velocity algorithm scVelo. Black arrows indicate lineage trajectories constructed using Slingshot. One excitatory neuron trajectory (dRG → IP-EN → nEN) and three inhibitory neuron trajectories (vRG → IP-IN → nIN1, vRG → IP-IN → nIN2, and vRG → IP-IN → nIN3) are constructed. l. Heatmap of smoothened expression of all differentially expressed genes (DEGs) along the pseudotime of nEN and nIN1 lineage development trajectories. Selected genes are listed on the left. A gene is considered significant when adjusted p-value based on FDR is <0.05 (see Methods). For DEG information, see Supplementary Table 1. m. Heatmap of smoothened expression of top-100 DEGs along the pseudotime of nEN, nIN1, nIN2, and nIN3 lineage development trajectories. A gene is considered significant when adjusted p-value based on FDR is <0.05 (see Methods). For DEG information, see Supplementary Table 1.

Source Data

Supplementary information

Supplementary Information

Supplementary Notes, legends for Supplementary Videos 1–7, legends for Supplementary Tables 1–4, Supplementary Figs. 1–13 (including legends) and Supplementary References.

Reporting Summary

Supplementary Table 1

List of top differentially expressed genes and GO analysis for cell clusters identified by scRNA-seq.

Supplementary Table 2

List of CRISPR–Cas9 guide RNAs and cloning primers to generate TBXT::T2A-Cre lineage tracer human ES cell line.

Supplementary Table 3

List of primers used in RT–qPCR.

Supplementary Table 4

List of antibodies used in immunocytochemistry (ICC) and immunohistochemistry (IHC).

Supplementary Video 1

Dynamic formation of microfluidic human neural tube-like structures (µNTLS). See Supplementary Information for full legend.

Supplementary Video 2

Clonal growth of H2B-GFP human ES cells in µNTLS. See Supplementary Information for full legend.

Supplementary Video 3

Development of neuromesodermal progenitors (NMPs) in µNTLS. See Supplementary Information for full legend.

Supplementary Video 4

Contribution of NMPs to caudal spinal cord development. See Supplementary Information for full legend.

Supplementary Video 5

Dynamic growth of R–C and D–V patterned µNTLS. See Supplementary Information for full legend.

Supplementary Video 6

Contribution of NMPs to trunk neural crest development. See Supplementary Information for full legend.

Supplementary Video 7

3D view of D–V patterned human forebrain-like structure (µFBLS). See Supplementary Information for full legend.

Source data

Rights and permissions

Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Xue, X., Kim, Y.S., Ponce-Arias, AI. et al. A patterned human neural tube model using microfluidic gradients. Nature 628, 391–399 (2024). https://doi.org/10.1038/s41586-024-07204-7

Download citation

  • Received:

  • Accepted:

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1038/s41586-024-07204-7

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

Advanced 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