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:

Quantitative lineage analysis identifies a hepato-pancreato-biliary progenitor niche

Nature volume 597, pages 87–91 (2021)Cite this article

Subjects

Abstract

Studies based on single cells have revealed vast cellular heterogeneity in stem cell and progenitor compartments, suggesting continuous differentiation trajectories with intermixing of cells at various states of lineage commitment and notable degrees of plasticity during organogenesis1,2,3,4,5. The hepato-pancreato-biliary organ system relies on a small endoderm progenitor compartment that gives rise to a variety of different adult tissues, including the liver, pancreas, gall bladder and extra-hepatic bile ducts6,7. Experimental manipulation of various developmental signals in the mouse embryo has underscored important cellular plasticity in this embryonic territory6. This is reflected in the existence of human genetic syndromes as well as congenital malformations featuring multi-organ phenotypes in liver, pancreas and gall bladder6. Nevertheless, the precise lineage hierarchy and succession of events leading to the segregation of an endoderm progenitor compartment into hepatic, biliary and pancreatic structures have not yet been established. Here we combine computational modelling approaches with genetic lineage tracing to accurately reconstruct the hepato-pancreato-biliary lineage tree. We show that a multipotent progenitor subpopulation persists in the pancreato-biliary organ rudiment, contributing cells not only to the pancreas and gall bladder but also to the liver. Moreover, using single-cell RNA sequencing and functional experiments we define a specialized niche that supports this subpopulation in a multipotent state for an extended time during development. Together these findings indicate sustained plasticity underlying hepato-pancreato-biliary development that might also explain the rapid expansion of the liver while attenuating pancreato-biliary growth.

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: Modelling of tissue dynamics in hepato-pancreatic organ rudiments.
Fig. 2: Lineage tracing of hepato-pancreato-biliary progenitor populations.
Fig. 3: scRNA-seq identifies a distinct progenitor subpopulation in the PB bud.
Fig. 4: Unique signalling signature defines the IMP population.

Similar content being viewed by others

Data availability

The single-cell RNA-sequencing data in this study have been deposited in the Gene Expression Omnibus under accession code GSE144103. Source data are provided with this paper.

Code availability

The single-cell RNA-sequencing data were analysed using publicly available R packages. Model D is available at https://www.ebi.ac.uk/biomodels/MODEL2105030001.

References

  1. Etzrodt, M., Endele, M. & Schroeder, T. Quantitative single-cell approaches to stem cell research. Cell Stem Cell 15, 546–558 (2014).

    Article  CAS  PubMed  Google Scholar 

  2. Fischer, D. S. et al. Inferring population dynamics from single-cell RNA-sequencing time series data. Nat. Biotechnol. 37, 461–468 (2019).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  3. Haas, S., Trumpp, A. & Milsom, M. D. Causes and consequences of hematopoietic stem cell heterogeneity. Cell Stem Cell 22, 627–638 (2018).

    Article  CAS  PubMed  Google Scholar 

  4. Laurenti, E. & Göttgens, B. From haematopoietic stem cells to complex differentiation landscapes. Nature 553, 418–426 (2018).

    Article  ADS  CAS  PubMed  PubMed Central  Google Scholar 

  5. Nowotschin, S. et al. The emergent landscape of the mouse gut endoderm at single-cell resolution. Nature 569, 361–367 (2019).

    Article  ADS  CAS  PubMed  PubMed Central  Google Scholar 

  6. Zaret, K. S. & Grompe, M. Generation and regeneration of cells of the liver and pancreas. Science 322, 1490–1494 (2008).

    Article  ADS  CAS  PubMed  PubMed Central  Google Scholar 

  7. Zorn, A. M. & Wells, J. M. Vertebrate endoderm development and organ formation. Annu. Rev. Cell Dev. Biol. 25, 221–251 (2009).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  8. Slack, J. M. W. Origin of stem cells in organogenesis. Science 322, 1498–1501 (2008).

    Article  ADS  CAS  PubMed  Google Scholar 

  9. Larsen, H. L. et al. Stochastic priming and spatial cues orchestrate heterogeneous clonal contribution to mouse pancreas organogenesis. Nat. Commun. 8, 605 (2017).

    Article  ADS  PubMed  PubMed Central  CAS  Google Scholar 

  10. Treutlein, B. et al. Reconstructing lineage hierarchies of the distal lung epithelium using single-cell RNA-seq. Nature 509, 371–375 (2014).

    Article  ADS  CAS  PubMed  PubMed Central  Google Scholar 

  11. Baron, C. S. & van Oudenaarden, A. Unravelling cellular relationships during development and regeneration using genetic lineage tracing. Nat. Rev. Mol. Cell Biol. 20, 753–765 (2019).

    Article  CAS  PubMed  Google Scholar 

  12. Pei, W. et al. Polylox barcoding reveals haematopoietic stem cell fates realized in vivo. Nature 548, 456–460 (2017).

    Article  ADS  CAS  PubMed  PubMed Central  Google Scholar 

  13. Rulands, S. & Simons, B. D. Tracing cellular dynamics in tissue development, maintenance and disease. Curr. Opin. Cell Biol. 43, 38–45 (2016).

    Article  CAS  PubMed  Google Scholar 

  14. Prior, N. et al. Lgr5+ stem and progenitor cells reside at the apex of a heterogeneous embryonic hepatoblast pool. Development 146, dev174557 (2019).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  15. Tremblay, K. D. & Zaret, K. S. Distinct populations of endoderm cells converge to generate the embryonic liver bud and ventral foregut tissues. Dev. Biol. 280, 87–99 (2005).

    Article  CAS  PubMed  Google Scholar 

  16. Li, L.-C. et al. Single-cell transcriptomic analyses reveal distinct dorsal/ventral pancreatic programs. EMBO Rep. 19, e46148 (2018).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  17. Spence, J. R. et al. Sox17 regulates organ lineage segregation of ventral foregut progenitor cells. Dev. Cell 17, 62–74 (2009).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  18. Zhou, Q. et al. A multipotent progenitor domain guides pancreatic organogenesis. Dev. Cell 13, 103–114 (2007).

    Article  CAS  PubMed  Google Scholar 

  19. Sosa-Pineda, B., Wigle, J. T. & Oliver, G. Hepatocyte migration during liver development requires Prox1. Nat. Genet. 25, 254–255 (2000).

    Article  CAS  PubMed  Google Scholar 

  20. Raue, A. et al. Lessons learned from quantitative dynamical modeling in systems biology. PLoS ONE 8, e74335 (2013).

    Article  ADS  CAS  PubMed  PubMed Central  Google Scholar 

  21. Rodríguez-Seguel, E. et al. Mutually exclusive signaling signatures define the hepatic and pancreatic progenitor cell lineage divergence. Genes Dev. 27, 1932–1946 (2013).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  22. Ober, E. A. & Lemaigre, F. P. Development of the liver: Insights into organ and tissue morphogenesis. J. Hepatol. 68, 1049–1062 (2018).

    Article  CAS  PubMed  Google Scholar 

  23. Liu, J. et al. Non-parallel recombination limits Cre-LoxP-based reporters as precise indicators of conditional genetic manipulation. Genesis 51, 436–442 (2013).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  24. Danielian, P. S., Muccino, D., Rowitch, D. H., Michael, S. K. & McMahon, A. P. Modification of gene activity in mouse embryos in utero by a tamoxifen-inducible form of Cre recombinase. Curr. Biol. 8, 1323–1326 (1998).

    Article  CAS  PubMed  Google Scholar 

  25. Lotto, J. et al. Single-cell transcriptomics reveals early emergence of liver parenchymal and non-parenchymal cell lineages. Cell 183, 702–716.e14 (2020).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  26. Hebrok, M. Hedgehog signaling in pancreas development. Mech. Dev. 120, 45–57 (2003).

    Article  CAS  PubMed  Google Scholar 

  27. Bangs, F. & Anderson, K. V. Primary cilia and mammalian hedgehog signaling. Cold Spring Harb. Perspect. Biol. 9, a028175 (2017).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  28. Dessaud, E., McMahon, A. P. & Briscoe, J. Pattern formation in the vertebrate neural tube: a sonic hedgehog morphogen-regulated transcriptional network. Development 135, 2489–2503 (2008).

    Article  CAS  PubMed  Google Scholar 

  29. Martínez-Frías, M. L. et al. Tracheoesophageal fistula, gastrointestinal abnormalities, hypospadias, and prenatal growth deficiency. Am. J. Med. Genet. 44, 352–355 (1992).

    Article  PubMed  Google Scholar 

  30. Mitchell, J. et al. Neonatal diabetes, with hypoplastic pancreas, intestinal atresia and gall bladder hypoplasia: search for the aetiology of a new autosomal recessive syndrome. Diabetologia 47, 2160–2167 (2004).

    Article  CAS  PubMed  Google Scholar 

  31. Heij, H. A. & Niessen, G. J. Annular pancreas associated with congenital absence of the gallbladder. J. Pediatr. Surg. 22, 1033 (1987).

    Article  CAS  PubMed  Google Scholar 

  32. Park, E. J. et al. System for tamoxifen-inducible expression of cre-recombinase from the Foxa2 locus in mice. Dev. Dyn. 237, 447–453 (2008).

    Article  CAS  PubMed  Google Scholar 

  33. Hingorani, S. R. et al. Preinvasive and invasive ductal pancreatic cancer and its early detection in the mouse. Cancer Cell 4, 437–450 (2003).

    Article  CAS  PubMed  Google Scholar 

  34. Perl, A.-K. T., Wert, S. E., Nagy, A., Lobe, C. G. & Whitsett, J. A. Early restriction of peripheral and proximal cell lineages during formation of the lung. Proc. Natl Acad. Sci. USA 99, 10482–10487 (2002).

    Article  ADS  CAS  PubMed  PubMed Central  Google Scholar 

  35. Snippert, H. J. et al. Intestinal crypt homeostasis results from neutral competition between symmetrically dividing Lgr5 stem cells. Cell 143, 134–144 (2010).

    Article  CAS  PubMed  Google Scholar 

  36. Abe, T. et al. Establishment of conditional reporter mouse lines at ROSA26 locus for live cell imaging. Genesis 49, 579–590 (2011).

    Article  CAS  PubMed  Google Scholar 

  37. Gong, S. et al. A gene expression atlas of the central nervous system based on bacterial artificial chromosomes. Nature 425, 917–925 (2003).

    Article  ADS  CAS  PubMed  Google Scholar 

  38. Means, A. L., Xu, Y., Zhao, A., Ray, K. C. & Gu, G. A CK19(CreERT) knockin mouse line allows for conditional DNA recombination in epithelial cells in multiple endodermal organs. Genesis 46, 318–323 (2008).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  39. Gu, G., Dubauskaite, J. & Melton, D. A. Direct evidence for the pancreatic lineage: NGN3+ cells are islet progenitors and are distinct from duct progenitors. Development 129, 2447–2457 (2002).

    Article  CAS  PubMed  Google Scholar 

  40. Madisen, L. et al. A robust and high-throughput Cre reporting and characterization system for the whole mouse brain. Nat. Neurosci. 13, 133–140 (2010).

    Article  CAS  PubMed  Google Scholar 

  41. Gong, S., Yang, X. W., Li, C. & Heintz, N. Highly efficient modification of bacterial artificial chromosomes (BACs) using novel shuttle vectors containing the R6Kgamma origin of replication. Genome Res. 12, 1992–1998 (2002).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  42. Petzold, K. M. & Spagnoli, F. M. A system for ex vivo culturing of embryonic pancreas. J. Vis. Exp. (66), e3979 (2012). https://doi.org/10.3791/3979.

    Article  CAS  Google Scholar 

  43. Preibisch, S., Saalfeld, S. & Tomancak, P. Globally optimal stitching of tiled 3D microscopic image acquisitions. Bioinformatics 25, 1463–1465 (2009).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  44. Dewitz, C. et al. Nuclear organization in the spinal cord depends on motor neuron lamination orchestrated by catenin and afadin function. Cell Rep. 22, 1681–1694 (2018).

    Article  CAS  PubMed  Google Scholar 

  45. Husson, F., Josse, J. & Pagès, J. Principal component methods-hierarchical clustering-partitional clustering: why would we need to choose for visualizing data? Applied Mathematics Department, 1–10 (2010).

  46. Charrad, M., Ghazzali, N., Boiteau, V. & Niknafs, A. NbClust: An R package for determining the relevant number of clusters in a data set. J. Stat. Softw. 61, 1–36 (2014).

    Article  Google Scholar 

  47. Raue, A. et al. Data2Dynamics: a modeling environment tailored to parameter estimation in dynamical systems. Bioinformatics 31, 3558–3560 (2015).

    Article  CAS  PubMed  Google Scholar 

  48. Picelli, S. et al. Smart-seq2 for sensitive full-length transcriptome profiling in single cells. Nat. Methods 10, 1096–1098 (2013).

    Article  CAS  PubMed  Google Scholar 

  49. Lun, A. T. L., McCarthy, D. J. & Marioni, J. C. A step-by-step workflow for low-level analysis of single-cell RNA-seq data with Bioconductor. F1000Res 5, 2122 (2016).

    PubMed  PubMed Central  Google Scholar 

  50. McCarthy, D. J., Campbell, K. R., Lun, A. T. L. & Wills, Q. F. Scater: pre-processing, quality control, normalization and visualization of single-cell RNA-seq data in R. Bioinformatics 33, 1179–1186 (2017).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  51. Lun, A. T., Bach, K. & Marioni, J. C. Pooling across cells to normalize single-cell RNA sequencing data with many zero counts. Genome Biol. 17, 75 (2016).

    Article  PubMed  CAS  Google Scholar 

  52. Amezquita, R. A. et al. Orchestrating single-cell analysis with Bioconductor. Nat. Methods 17, 137–145 (2020).

    Article  CAS  PubMed  Google Scholar 

  53. Trapnell, C. et al. The dynamics and regulators of cell fate decisions are revealed by pseudotemporal ordering of single cells. Nat. Biotechnol. 32, 381–386 (2014).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  54. Escot, S., Willnow, D., Naumann, H., Di Francescantonio, S. & Spagnoli, F. M. Robo signalling controls pancreatic progenitor identity by regulating Tead transcription factors. Nat. Commun. 9, 5082 (2018).

    Article  ADS  PubMed  PubMed Central  CAS  Google Scholar 

  55. Palaria, A., Angelo, J. R., Guertin, T. M., Mager, J. & Tremblay, K. D. Patterning of the hepato-pancreatobiliary boundary by BMP reveals heterogeneity within the murine liver bud. Hepatology 68, 274–288 (2018).

    Article  CAS  PubMed  Google Scholar 

  56. Sanders, E. J., Varedi, M. & French, A. S. Cell proliferation in the gastrulating chick embryo: a study using BrdU incorporation and PCNA localization. Development 118, 389–399 (1993).

    Article  CAS  PubMed  Google Scholar 

  57. van den Berg, G. et al. A caudal proliferating growth center contributes to both poles of the forming heart tube. Circ. Res. 104, 179–188 (2009).

    Article  PubMed  CAS  Google Scholar 

  58. Clevers, H. & Watt, F. M. Defining adult stem cells by function, not by phenotype. Annu. Rev. Biochem. 87, 1015–1027 (2018).

    Article  CAS  PubMed  Google Scholar 

Download references

Acknowledgements

We thank all the members of the Spagnoli laboratory for their useful comments and suggestions on the study; H. Naumann for technical help; G. Pazour for IFT57 antibody; A. Christ and T. Willnow for the GLI2 and SMO antibodies; the MDC Transgenic Unit for technical help in generating the Prox1-rtTA mouse strain; and C. Beisel of the Genomics Facility of D-BSSE for NGS RNA sequencing. Maintenance of the two-photon microscopy setup was supported by the staff of the Advanced Light Microscopy technology platform via funding from the MDC in the Helmholtz Association. This research was supported by funds from the Helmholtz Association (FMS, JW), European Union’s Horizon 2020 Research and Innovation Programme (Pan3DP Grant agreement no. 800981) (FMS, DW); D.W. was a recipient of a BIH (Tr. PhD) fellowship. U.B. was a recipient of an EMBO Short-Term Fellowship (7853).

Author information

Authors and Affiliations

  1. Centre for Stem Cells and Regenerative Medicine, King’s College London, London, UK

    David Willnow, Maria Lillina Vignola, Alessandra Vigilante & Francesca M. Spagnoli

  2. Berlin Institute of Health, Berlin, Germany

    David Willnow

  3. Max-Delbrueck-Center for Molecular Medicine, Berlin, Germany

    Uwe Benary, Anca Margineanu, Fabian Konrath, Igor M. Pongrac & Jana Wolf

  4. Department of Biosystems Science and Engineering (D-BSSE), ETH Zurich, Basel, Switzerland

    Zahra Karimaddini

  5. Swiss Institute of Bioinformatics, Basel, Switzerland

    Zahra Karimaddini

  6. Department of Mathematics and Computer Science, Free University, Berlin, Germany

    Jana Wolf

Authors
  1. David Willnow
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Uwe Benary
    View author publications

    You can also search for this author in PubMed Google Scholar

  3. Anca Margineanu
    View author publications

    You can also search for this author in PubMed Google Scholar

  4. Maria Lillina Vignola
    View author publications

    You can also search for this author in PubMed Google Scholar

  5. Fabian Konrath
    View author publications

    You can also search for this author in PubMed Google Scholar

  6. Igor M. Pongrac
    View author publications

    You can also search for this author in PubMed Google Scholar

  7. Zahra Karimaddini
    View author publications

    You can also search for this author in PubMed Google Scholar

  8. Alessandra Vigilante
    View author publications

    You can also search for this author in PubMed Google Scholar

  9. Jana Wolf
    View author publications

    You can also search for this author in PubMed Google Scholar

  10. Francesca M. Spagnoli
    View author publications

    You can also search for this author in PubMed Google Scholar

Contributions

F.M.S. and D.W. conceived the study, designed the experiments and wrote the manuscript. D.W. performed all the experiments. I.M.P. generated the Prox1-rtTA transgenic strain. A.M. helped with the Confetti lineage-tracing experiments and two-photon image acquisition. U.B., F.K. and J.W. developed the mathematical modelling approach. M.L.V., A.V. and Z.K. performed scRNA-seq data analyses.

Corresponding author

Correspondence to Francesca M. Spagnoli.

Ethics declarations

Competing interests

The authors declare no competing interests.

Additional information

Peer review information Nature thanks Dominic Grun, Meritxell Huch and the other, anonymous, reviewer(s) for their contribution to the peer review of this work.

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 Expression of hepato-pancreato-biliary markers during early cell fate specification.

a, Images of mouse embryos at the indicated somite stages (ss) taken using a Zeiss Stemi 2000 stereomicroscope. Scale bar, 1mm. b, Representative confocal whole‐mount immunofluorescence (IF) images of embryos stained for Prox1. In addition to the ventral foregut region (VFG), Prox1 is expressed in the developing heart tube (HT) from 6ss onwards. Scale bar, 100μm. c–e, Representative confocal IF images of the VFG in E8.5-E8.75 (8-13ss) embryos. VFG shown as whole‐mount in frontal (c) or lateral (d) views or in transverse cryosections (e). Prox1 (red), Pdx1 (blue), and Sox17 (green) mark hepatic (Prox1+) and pancreato-biliary (Prox1+/Pdx1+/Sox17+) progenitors in the VFG (outlined by dotted line). Insets in c, e show higher magnifications of boxed regions in grey scale for Pdx1 channel. Right panels in d show Pdx1 channel in grey scale highlighting the onset of Pdx1 expression at the VFG lip. Hoechst dye was used as nuclear counterstain. Scale bars, 100μm. f, Scatter plot showing the ratio of LV versus PB cell counts (n = 166 embryos). Quantification of cell numbers in LV and PB buds was performed on IF stained cryosections (see Fig. 1b). Data are representative of 3 or more biologically independent experiments with similar results.

Source data

Extended Data Fig. 2 Dynamics of cell proliferation and apoptosis in hepatic, ventral pancreato-biliary, and dorsal pancreatic organ rudiments.

a–c, Representative IF images of cryosections of mouse embryos at indicated somite stages stained for pH3 (green) and indicated markers. In a, Ecad (red) marks epithelial cells of the endoderm and ectoderm at 0ss. In b, Ecad (blue)/Prox1 (red) mark the ventral foregut (VFG) at 6ss. In c, Prox1 (left panel) marks liver (LV) cells; Pdx1/Prox1 (left panel) or Pdx1/Sox17 (right panel) mark pancreato-biliary PB cells. In c, left and right panels show consecutive sections of the same embryos. VFG, LV, and PB are outlined by white dotted lines. Hoechst dye was used as nuclear counterstain. NF, neural fold. Scale bars, 100μm. d, Quantification of pH3+ cells in PB buds on cryosections (n = 26 embryos). Dot plot shows the numbers of pH3+ cells as % of the total cell count in the PB bud from E8.5 to E11.0 (9-41ss). Proliferating PB cells were counted on consecutive sections of the same embryos stained either for Prox1/Pdx1/pH3 or Sox17/Pdx1/pH3. Mean ± s.d. Two-tailed Kruskal–Wallis test with Dunn’s multiple comparisons test. ns, not significant. e, Quantification of pH3+ dorsal (DP) and ventral (VP) pancreatic cells on cryosections (n = 28 embryos). Dot plot shows the % fraction of pH3+ cells relative to the total cell number in each respective organ rudiment (% of total cell count) from E9.0 to E11.0 (15-44ss). VP cells within the PB bud were defined as described in Extended Data Fig. 5. Mean ± s.d. Two-tailed Kruskal–Wallis test with Dunn’s multiple comparisons test. ns. f, Representative whole-mount IF images of embryos at E9.0-E9.75 (16-26 ss) stained for pH3 (green), Prox1 (red) and Pdx1 (blue). LV and PB buds are outlined by dashed white lines. Scale bar, 100μm. g, Measurement of LV and PB bud volumes at the indicated somite stages (n = 129 embryos). 3D organ volumes were reconstructed by measuring the surface area on individual optical sections of confocal z-series of whole-mount IFs (Fig. 1a). Average surface areas in each embryo were then multiplied by tissue thickness. h, Quantification of pH3+ cell numbers in whole-mount IF images of LV and PB buds (n = 71 embryos) normalized to organ bud volume (mm3). pH3 levels were higher in pancreato-biliary rudiments as compared to liver buds at early stages (E8.5; P = 0.0016), while no significant differences between the two organ rudiments were observed at later stages. Mean ± s.d. Two-tailed Kruskal–Wallis test with Dunn’s multiple comparisons test. ns. i, Representative image of 18ss embryo stained for the indicated markers shows BrdU (green) labelled cells following a 4h-labelling period. Prox1 (red) marks LV, Prox1/Pdx1 (blue) marks PB progenitors. LV and PB buds are outlined by a dashed white line. Scale bar, 100μm. j, BrdU incorporation in LV and PB progenitors of E8.5 and E10.5 embryos following the indicated BrdU pulse-chasing periods. Dot plot showing the fraction of BrdU-labelled cells relative to the total cell number in each organ rudiment (% of total cell count). Each dot represents the mean from an individual embryo [E8.5: n(1h) = 4, n(2h) = 6, n(4h) = 8; E10.5: n(0.5h) = 3, n(4h) = 5]. Mean ± s.d. Two-tailed Kruskal–Wallis test with Dunn’s multiple comparisons test. ns. k, Graph illustrating cell proliferation rates as shown by the % of BrdU+ cells in E8.5 and E10.5 LV and PB organ domains at different labelling periods (Extended Data Fig. 2j). The cell cycle length (tc) was estimated based on BrdU incorporation using linear regression56,57. LV (E8.5) slope = 3.8%/h; estimated cell cycle length = 26.6h (range with 95% confidence interval: 44.7h-19h). PB (E8.5) slope = 5.8%/h; estimated cell cycle length = 17.3h (range with 95% confidence interval: 29.6h-12.2h). LV (E10.5) slope = 5.5%/h; estimated cell cycle length = 18.2h (range with 95% confidence interval: 21h-16.1h). PB (E10.5) slope = 4.8%/h; estimated cell cycle length = 21h (range with 95% confidence interval: 33h-15.4h). No statistically significant differences in average cell cycle length were found between LV and PB progenitors at E8.5 (P = 0.14) or E10.5 (P = 0.36). Two-tailed linear regression t-test. l, Fraction of pH3+/BrdU+ cells in LV and PB buds as % of the total BrdU+ cell population at the indicated pulse-chasing period [n(0.5h) = 3, n(4h) = 3]. Mean ± s.d. Two-tailed Kruskal–Wallis test with Dunn’s multiple comparisons test. ns. m, Fraction of BrdU+ early and late mitotic cells in LV and PB buds as % of the total early or late mitotic cell population at the indicated pulse-chasing period [n(0.5h) = 3, n(4h) = 3]. pH3 IF staining intensity was used to identify early (low, punctate pH3 signal) and late (high pH3 signal) mitotic cells. Mean ± s.d. Two-tailed Kruskal–Wallis test with Dunn’s multiple comparisons test. ns. n, Fraction of BrdU+ cells in LV, PB and DP buds of E9.5 embryos following the indicated pulse-chasing period [n(2h) = 2, n(4h) = 5]. Mean ± s.d. Two-tailed Kruskal–Wallis test with Dunn’s multiple comparisons test. ns. o, Fraction of BrdU+ cells in DP, VP and GB of E10.5 embryos following the indicated pulse-chasing period (n = 4). Mean ± s.d. Two-tailed Kruskal–Wallis test with Dunn’s multiple comparisons test. ns. p, q, Immunostaining for apoptotic cell marker cleaved caspase 3 (cCas3) in embryos at the indicated somite stages. Representative IF images of cryosections of E8.0 (5ss) Tg(Prox1-EGFP) embryo (p) stained for cCas3 (blue), Prox1 (red) and GFP (green) to detect VFG cells (Prox1+, GFP+) and E9.5 (22ss) embryo (q) stained for cCas3 (blue), Prox1 (red) and Pdx1 (blue). Insets show higher magnifications of boxed regions in grey scale for the indicated channels. Rare apoptotic cells are found in LV and PB bud (arrowheads) as well as surrounding mesenchyme (arrow). Hoechst, nuclear counterstain. Scale bar, 100μm. r, Quantification of apoptotic (cCas3+) cells in VFG, LV, and PB buds on stained cryosections (n = 19 embryos) as shown in p, q. Dot plot shows the fraction of cCas3+ cells as % of total cell count in the indicated organ domains. No statistically significant differences in apoptosis were found between LV and PB buds at E8.5-E9.0 (12-20ss) or E9.5-E10.0 (21-31ss). Mean ± s.d. Two-tailed Kruskal–Wallis test with Dunn’s multiple comparisons test. ns. Data are representative of 3 or more biologically independent experiments with similar results.

Source data

Extended Data Fig. 3 Extended computational modelling analysis of ventral foregut development.

a, Models A and B fitted to the experimental data. Best fit of models A (black solid line) and B (red solid line) and the respective standard deviations are shown. Cell count data (black dots) collected between E7.5 and E11.5 (0-45ss) were used to estimate the parameter values of the models. b, Models C and D fitted to the experimental cell count data (black dots). Best fit of model C (black solid line) and model D (red solid line) and the respective standard deviations are shown. c–g, Simulations of distinct cell populations (indicated above each panel) in model C (black solid line) and model D (red solid line). h, i, Simulations of model D predict a flux of cells from the pancreato-biliary (PB) population to the hepatic (LV) population (h) throughout the time frame of the model, and a delayed increasing flux in the reverse direction (i). Until the 30ss, the flux of cells from the PB to the LV domain exceeds that of LV cells to the PB domain. Afterwards the direction of the positive net flux is reversed. j, Sum of BEP, BEPp, LV, and LVp cell counts (LV*) simulated with model D (red line) overlaid with the exponential function (black dashed line) that was fitted to the simulated cell counts to estimate the doubling time (11.8h). k, Sum of PB, PBp, VP, VPp, GB, and GBp cell counts (PB*) simulated with model D (red line) overlaid with the exponential function (black dashed line) that was fitted to the simulated cell counts to estimate the doubling time (14.1h). l, Sum of BEP, BEPp, LV, and LVp cell counts (LV*, dark red line) and sum of PB, PBp, VP, VPp, GB, and GBp cell counts (PB*, dark green line) simulated with model D using the nominal parameter values. Light colored lines represent simulations of LV* and PB* cell counts after reducing plasticity by setting parameter k17 and k18 to 1/4th of their nominal parameter values. The overlay shows that a reduction of plasticity reduces LV* cell count and increases PB* cell count. m, Ratio of the sum of BEP, BEPp, LV, and LVp cell counts (LV*) and the sum of PB, PBp, VP, VPp, GB, and GBp cell counts (PB*) simulated with model D (red line) is overlaid with LV*/PB* ratios simulated with model D in which parameters k17 and k18 were both reduced to 1/2, 1/4th and 1/8th of their original values. The overlay shows that a reduction of parameters k17 and k18 (that is, plasticity) reduces LV*/PB* cell count ratio.

Source data

Extended Data Fig. 4 Overview of hepato-pancreato-biliary cell populations and their attributed marker gene expression.

a, Bar graph showing the number of data points that were collected for each somite stage and used for parameterizing the mathematical models. b, Schematic representation of mathematical model C overlaid with overlapping coloured boxes that indicate marker gene expression of various hepato-pancreato-biliary subpopulations (Prox1, red; Sox17, green; Pdx1, blue; pH3, grey). c, d, Eight new cell populations extend model D: LVs, LVsp, PBs, PBsp, GBs, GBsp, VPs, and VPsp that correspond to labelled cells (see Supplementary Note 1). Two different experimental setups were simulated: multicolour Prox1-rtTA;Confetti lineage tracing and Pdx1-CreERT lineage tracing experiments. The two experimental setups, Prox1-rtTA;Confetti lineage tracing (c) and Pdx1-CreERT lineage tracing experiments (d), differ in the sets of cell populations that are affected by the spike-in event. Abbreviation: s, spike-in.

Extended Data Fig. 5 Hepatic and pancreato-biliary organ domains contain cell subpopulations with distinct marker expression profiles.

a–f, Single-cell measurement of fluorescence intensity (FI) of Pdx1 and Sox17 in pancreato-biliary (PB) rudiments at E8.5-E11.5 (9-45ss). Pdx1 and Sox17 FI values for individual cells were plotted. Embryos of similar somite stages were grouped as indicated (n, number of embryos; 9-14ss: n = 27, 6608 cells; 15-20ss: n = 15, 4310 cells; 21-26ss: n = 16, 5125 cells; 27-32ss: n = 16, 6644 cells; 33-38ss: n = 16, 13002 cells; 39-45ss: n = 6, 10588 cells). From E8.5 to E9.5 (9ss-26ss), the majority of cells in the PB organ rudiment co-expressed Sox17 and Pdx1, while at E10.0-E11.5 (27ss-45ss) they acquired either Sox17high or Pdx1high identity. FI was measured using ImageJ/Fiji. Values were corrected by linear normalization within each embryo. Black dashed lines indicate sub-division of the progenitor populations based on Pdx1 and Sox17 FI levels. g, Representative IF images of cryosections of the PB organ domain from E8.5 (11ss) to E11.0 (40ss) embryos. Pdx1 (green) and Sox17 (red) mark PB progenitor cells. After E9.75 (about 27ss), the PB bud segregates into gall bladder (GB; marked by Sox17) and ventral pancreatic bud (VP; marked by Pdx1). Hoechst dye was used as nuclear counterstain. Scale bar, 100μm. h, i, Scatter plots displaying the number of cells in different PB rudiments against developmental time (n = 96 embryos). FI data were used to categorize the subpopulations according to the relative expression levels of Pdx1 and Sox17. Pdx1high subpopulation corresponds to VP progenitors (Pdx1-FI >50, Sox17-FI\(\le \)50) (green dots); Sox17high corresponds to GB progenitors (Pdx1-FI\(\le \)50, Sox17-FI >50) (red dots); Pdx1/Sox17-double positive at low (Pdx1-FI\(\le \)50, Sox17-FI\(\le \)50) or high levels (Pdx1-FI >50, Sox17-FI >50) correspond to PB progenitors (yellow dots). The three subpopulations exhibit distinct propagation kinetics as shown by plotting absolute cell counts in the individual organ domains (h) or fraction of each subpopulation as percentage of the total cell number in the PB bud (i) against somite stage. While the relative fraction of PB progenitors decreased as compared to GB and VP progenitors after E9.75 [26-27ss; (i)], their total cell number remained constant throughout the analysed time period (h). j, k, Representative IF stainings for Prox1 (red), Pdx1 (blue), and albumin (green) on cryosections of embryos at the indicated somite stages. Between 15ss and 19ss (j), a subset of liver (LV) progenitor cells is positive for the hepatic marker albumin (arrowheads). At 36ss (k), the majority of LV cells co-express Prox1 and albumin, but a fraction of albumin- cells is found in the LV (arrows). At no analysed time point, albumin+/Prox1- cells were found in the LV. Right (j) or bottom (k) panels show higher magnifications of the boxed region as merged or single channels. Hoechst was used as nuclear counterstain. Scale bars, 100μm. l, Quantification of albumin+ cells in the LV bud on IF-stained cryosections (n = 23 embryos), shown as % of the Prox1+ LV cells. m, Comparison of LV cell counts measured on cryosections stained for Prox1 and albumin (n = 20 embryos) with LV cell counts predicted from model D [Fig. 1e; predicted LV cell count combines model predictions for the LV and LVp (Extended Data Fig. 3d) populations]. Expression measurements agree well with the model predictions at early somite stages (<30ss), although the LV cell counts based on albumin expression (red dots) or Prox1 expression (black circles) differ slightly at these early time points.

Source data

Extended Data Fig. 6 Extended lineage tracing analysis of hepato-pancreato-biliary progenitor populations.

a, Schematic representation of the transgenic constructs used for the Prox1-rtTA lineage tracing of mouse hepato-pancreato-biliary progenitors. The 2A peptide enabled co-expression of the rtTA and mCherry fluorescent reporter transgene. loxP sites are shown as purple triangles. b, c, Assessment of experimental parameters influencing Prox1-rtTA labelling efficiency. Bar plot in b shows the percentage (%) of embryos with induced label at increasing doxycycline doses [75μg/g body weight (bw), n = 22; 100μg/g bw, n = 29; 150μg/g bw, n = 116]. Graph in c shows the positive correlation between the number of genetically labelled cells in individual embryos and somite stage of labelling for the three indicated doxycycline doses. For all further analyses, pregnant females were injected with 150μg/g bw of doxycycline. d, Plots displaying the number of genetically labelled hepatic (LV) or pancreato-biliary (PB) progenitor cells, as percentage (%) of total labelled cells in both rudiments, plotted against somite stage in Foxa2-CreERT experiments (upper panel) or tracing period in Prox1-rtTA experiments (lower panel). Paired dots vertically aligned correspond to labelled cells (as %) measured in the LV (red dot) or PB (green dot) of the same embryo. Two-tailed linear regression t-test; P = 0.006 (Foxa2-CreERT; n = 83 embryos), P = 0.061 (Prox1-rtTA; n = 81 embryos). e, Schematic representation of the transgenic constructs used for the Foxa2-CreERT lineage tracing of mouse hepato-pancreato-biliary progenitors. loxP sites are shown as purple triangles. IRES, internal ribosome entry site; mER, murine oestrogen receptor. f, Schematic representation of experimental setup for clonal labelling of Tg(Foxa-CreERT;R26R-H2B-GFP) embryos. Pregnant females were intraperitoneally (IP) injected with a single dose of tamoxifen (TAM; 12μg/g bw) at E8.5 and embryos collected at E9.0-E9.5 (n = 15 embryos). g, Representative whole-mount IF of Tg(Foxa2-CreERT;R26R-H2B-GFP) embryos at E9.0-E9.5 (16-21ss) following TAM injection at E8.5. Arrowheads indicate genetically labelled (GFP+) cells in LV and PB buds. In clonal labelling experiments, 2-3 labelled cells were often found in close proximity to one another, suggesting they were descendants of a common labelled progenitor cell. Boxed region is shown at a higher magnification in the right panels; white dotted circles indicate two adjacent GFP+ cells, one Pdx1+ and one Pdx1-, which possibly arose from a common progenitor. Scale bar, 100μm. h, Schematic representation of the R26R-Confetti transgene. Following recombination, genetically labelled cells either express nuclear GFP (nGFP), cytoplasmic RFP or YFP, or membrane-associated CFP (mCFP). i, Representative optical section of two-photon microscopy 3D scans of E11.5 Tg(Prox1-rtTA;R26R-Confetti) embryos. Genetically labelled cells expressing CFP, YFP, or RFP were detected in both ventral (VP) and dorsal pancreatic (DP) buds (circled by dashed white lines). Scale bar, 100μm. j, Representative images of VP and DP buds from E10.5 Tg(Pdx1-Cre;R26R-H2B-GFP) (top) or Tg(Pdx1-Cre;R26R-Confetti) (bottom) embryos. Top panels, IF staining for Pdx1 (blue) and Prox1 (red) mark the VP and DP buds; GFP+ (green) labelled cells are in both VP and DP of Tg(Pdx1-Cre;R26R-H2B-GFP) embryos. Bottom panels, optical sections of two-photon microscopy 3D scans of Tg(Pdx1-Cre;R26R-Confetti) embryos show native reporter fluorescence of CFP, YFP, and RFP in genetically labelled cells. Scale bars, 100μm. k, Quantitative analysis of the contribution of the indicated fluorophores to the total number of labelled cells in the embryos shows characteristic induction efficiencies for each fluorophore that are comparable among different tissues. Mean ± s.d. Brown-Forsythe and Welch ANOVA test with Dunnett’s T3 multiple comparisons test. P < 0.001. l, In silico re-construction of genetically labelled cells in pancreatic buds of Tg(Prox1-rtTA;R26R-Confetti) embryos. Spot detection analysis in Imaris software was used to identify xyz-coordinates of individual genetically labelled cells and to reconstruct the labelled tissues. m, n, Clone size distribution established from Confetti lineage tracing experiments using the Prox1-rtTA (m) or Pdx1-Cre (n) Tg lines. Information on xyz-coordinates of labelled cells in DP and VP was used for clustering individual cells into clonal clusters based on their geometric distance from one another46. Significantly different distribution of clone sizes was measured between VP and DP, with most clusters in the VP being composed of 4-5 cells, while about 8 cells per cluster were found in the DP. (Prox1, n = 38 embryos; Pdx1, n = 16 embryos). Vertical dotted lines indicate mean values; two-tailed Mann–Whitney test. P < 0.001. o, Quantification of total number of cells (grey dots) and GFP-labelled cells (green dots) in DP and VP of E10.5 Tg(Pdx1-Cre;R26R-H2B-GFP) embryos (n = 18 embryos). Total cell count and GFP+ labelled cells are significantly higher in the DP as compared to the VP. Mean ± s.d. Two-tailed Kruskal–Wallis test with Dunn’s multiple comparisons test. P < 0.001. p, Quantification of the GFP+ cell population as percentage (%) of the total cell population in VP (green dots) or DP (blue dots) of Tg(Pdx1-Cre;R26R-H2B-GFP) embryos (n = 18 embryos). No statistically significant differences were found between DP and VP. Mean ± s.d. Two-tailed Mann–Whitney test. ns, not significant. q, Quantification of mitotic H2B-GFP+ cells in LV, DP, and PB organ rudiments on IF-stained cryosections of E11.5 Tg(Prox1-rtTA;R26R-H2B-GFP) (n = 4 embryos) and E10.5 Tg(Pdx1-Cre;R26R-H2B-GFP) embryos (n = 4 embryos). Dot plot shows the fraction of genetically labelled cells in mitosis relative to the total labelled cell number in the respective organ rudiments. Mean ± s.d. Two-tailed Kruskal–Wallis test with Dunn’s multiple comparisons test. ns. r, Simulation of Prox1-rtTA;Confetti lineage tracing experiment (see Supplementary Note 1, Extended Data Fig. 4c). Fraction of labelled cells in all VP* population (ratio of the sum of VPs and VPsp cell counts versus the sum of VPs, VPsp, VP, and VPp cell counts) is shown for different combinations of label induction time points and recombination efficiencies. Grey surface shows the fraction of labelled VP* simulated using model D with nominal parameter values; white-red gradient surface shows the fraction of labelled VP* simulated using model D after reducing plasticity by setting parameter k17 and k18 to 1/8th of their nominal values. The fraction of labelled VP* is always higher in the model with reduced plasticity, supporting the hypothesis that the PB to LV plasticity results in clonal dispersal and reduced cluster sizes in the VP. s, Sum of VP and VPp cell counts obtained from simulations with model D using the nominal k17 and k18 parameter values (red line) or simulations with model D, in which both parameters k17 and k18 were reduced to 1/2 (black line), 1/4th (dark grey line), or 1/8th (light grey line) of their nominal values. The overlay shows that a reduction of parameters k17 and k18 (that is, plasticity) results in increased VP and VPp cell counts.

Source data

Extended Data Fig. 7 Extended lineage tracing analysis of Pdx1+ pancreato-biliary progenitors.

a–c, Spatial representation of Pdx1-Cre lineage tracing experiments. IF images of cryosections from E10.5 (a, b) and E11.0 (c) Tg(Pdx1-Cre;R26R-H2B-GFP) embryos stained for Prox1, Pdx1, and GFP were digitalized to obtain xyz-coordinates for all GFP− (red) and GFP+ (green) cells in the liver (LV) and pancreato-biliary (PB; blue) buds44. Data from individual embryos at E10.5 (a, b; n = 8) and E11.0 (c; n = 10) were combined and cells plotted according to their xy- (b) or yz-coordinates (a, c). d, Representative IF image of Tg(Pdx1-Cre; R26R-H2B-GFP) embryo at E11.0. GFP (green) identifies labelled cells descended from Pdx1+ progenitors. Prox1 (red) marks the LV and gall bladder (GB; outlined by a white dashed line), Prox1/Pdx1 (blue) marks the ventral pancreas (VP; outlined by a white dashed line). Insets show higher magnification of boxed regions as merged and single GFP and Pdx1 channels. Arrows indicate GFP+ LV cells at the border with the VP and GB. Hoechst dye was used as nuclear counterstain. Scale bar, 100μm. e, Representative IF images of newborn Tg(Pdx1-Cre;R26R-H2B-GFP) mouse liver. Glutamine synthetase (GS, red) marks hepatocytes near the central vein, whereas Krt19 (blue) marks cholangiocytes. Co-staining of GFP with GS (arrows) or Krt19 (arrowheads) identifies descendants of Pdx1+ PB cells capable of differentiating into hepatocytes and cholangiocytes. Scale bar, 100μm. f, Representative IF image of Tg(Pdx1-Cre;R26R-H2B-GFP) embryo at E9.5. Right panels show higher magnification of the boxed region. Arrowheads indicate genetically labelled (GFP+) cells in PB bud and surrounding LV bud positive for Krt19. Scale bar, 100μm. g, Quantification of GFP+ cells in LV of genetically labelled Tg(Pdx1-Cre;R26R-H2B-GFP) embryos (E9.0, n = 7; E9.5, n = 10; E10.5, n = 7; E11.0, n = 7) and newborn mice (P1, n = 5) shows a constant labelling index of 0.8% of the total liver cell population. No statistically significant differences were detected between analysed time points. Mean ± s.d. Two-tailed Kruskal–Wallis test. ns, not significant. h, Quantification of labelled PB progenitors in Tg(Pdx1-Cre;R26R-Confetti) embryos (E9.5, n = 4; E10.5, n = 10). Genetically labelled cell populations increase with developmental stage. Mean ± s.d. Two-tailed Mann–Whitney test. P = 0.04. i, Quantification of labelled LV progenitors as % of total labelled cell population in the ventral foregut endoderm of Tg(Pdx1-Cre;R26R-Confetti) embryos (E9.5, n = 4; E10.5, n = 10). Mean ± s.d. Two-tailed Mann–Whitney test. P = 0.03. j, Representative IF image of cryosections from E10.5 Tg(Pdx1-Cre;R26R-Confetti) stained for Prox1 (blue), RFP (red) and GFP (green). Anti-GFP antibody also detects membrane-bound CFP and cytoplasmic YFP. Dashed white line marks PB progenitor cells. Insets show higher magnifications of the boxed regions highlighting genetically labelled descendants of Pdx1+ PB cells expressing YFP (arrowheads) or RFP (arrow) in the LV bud. Scale bar, 100μm. k–n, Spatial representations of clonal Pdx1-Cre lineage tracing experiments in four individual E10.5 Tg(Pdx1-Cre;R26R-Confetti) embryos. IF data were digitalized to obtain xyz-coordinates for individual labelled LV (LV-CFP, LV-RFP, LV-YFP) and PB progenitors (PB-cells)44. Cells were plotted according to their xy- (k-m) or yz-coordinates (n). Individual clones of the same colour are distributed along hepatic chords. Yellow arrows suggest trajectories of the labelled cell clusters arising from Pdx1+ PB progenitors and spreading along hepatic chords.

Source data

Extended Data Fig. 8 Inducible lineage tracing experiments of Pdx1+ pancreato-biliary progenitors.

a, Schematic representation of the Pdx1-CreERT transgene and experimental setup to obtain genetically labelled Tg(Pdx1-CreERT;R26R-H2B-GFP) or Tg(Pdx1-CreERT;R26R-tdTomato) embryos. Pregnant females were intraperitoneally (IP) injected with a single dose of tamoxifen [TAM; 150μg/g (R26R-H2B-GFP) or 25μg/g (R26R-tdTomato) body weight] at E8.5, E9.5, or E10.5. Labelled embryos were collected at E11.5. Based on the reported time-delay between TAM administration and induction for Cre-ERT223,24, Pdx1+ cells may be marked within a 24 h period post-injection. b, Representative IF images of cryosections from E11.5 Tg(Pdx1-CreERT;R26R-H2B-GFP) embryos stained for Prox1 (red), Pdx1 (blue), and GFP (green). GFP+ labelled cells were found in ventral pancreas (VP), gall bladder (GB), extrahepatic bile duct (EHBD), and liver (LV) following TAM administration at E8.5 (left panels) and E9.5 (right panels). Insets show higher magnifications of the boxed regions. GB, VP, and EHBD are outlined by white dashed lines. The EHBD structure is a derivative of the pancreato-biliary (PB) bud and becomes visible beginning at E11.5 in the mouse17. Scale bars, 100μm. c, Quantification showing the percentage (%) of labelled cells in VP of genetically labelled Tg(Pdx1-Cre;R26R-H2B-GFP) (grey dots; data in Extended Data Fig. 6p; n = 18 embryos) or Tg(Pdx1-CreERT;R26R-H2B-GFP) embryos [green dots; n(TAM E8.5) = 22 embryos, n(TAM E9.5) = 24 embryos, n(TAM E10.5) = 13 embryos]. The fraction of labelled VP cells induced by the constitutive Pdx1-Cre Tg was significantly higher compared to the TAM-inducible Pdx1-CreERT Tg, which is in line with previous reports23,39. Mean ± s.d. Two-tailed Kruskal–Wallis test with Dunn’s multiple comparisons test. P < 0.001. ns, not significant. d, Quantification of the fraction of labelled non-VP (that is, in GB, EHBD, and LV) cells as percentage (%) of total labelled cells in Tg(Pdx1-Cre;R26R-H2B-GFP) (grey dots; n = 18 embryos) or Tg(Pdx1-CreERT;R26R-H2B-GFP) embryos [green dots; n(TAM E8.5) = 21 embryos, n(TAM E9.5) = 24 embryos, n(TAM E10.5) = 13 embryos]. The fraction of labelled non-VP cells induced with the Pdx1-Cre Tg was significantly higher compared to the TAM-inducible Pdx1-CreERT one. TAM induction at E8.5 led to a significantly higher fraction of labelled non-VP cells compared to TAM at E10.5. Mean ± s.d. Two-tailed Kruskal–Wallis test with Dunn’s multiple comparisons test. ** P = 0.005; *** P < 0.001. e, Bar chart showing the percentage (%) of embryos with labelled LV cells following constitutive Pdx1-Cre [(18/19), 94.7%] or TAM-inducible Pdx1-CreERT lineage tracing strategies [TAM E8.5: (3/22), 15%; TAM E9.5: (2/24), 8.3%; TAM E10.5: (0/13), 0%]. f, Quantification showing the percentage (%) of labelled cells in VP of genetically labelled Tg(Pdx1-Cre;R26R-tdTomato) (grey dots; n = 5 embryos) or Tg(Pdx1-CreERT;R26R-tdTomato) embryos (red dots; n(TAM E8.5) = 7 embryos). The fraction of labelled VP cells induced by the constitutive Pdx1-Cre Tg was significantly higher compared to the TAM-inducible Pdx1-CreERT Tg, which is in line with previous reports23,39. Mean ± s.d. Two-tailed Mann–Whitney test. P = 0.003. g, Quantification of the fraction of labelled non-VP (that is, in GB, EHBD, and LV) cells as percentage (%) of total labelled cells in Tg(Pdx1-Cre;R26R-tdTomato) (grey dots; n = 5 embryos) or Tg(Pdx1-CreERT;R26R-tdTomato) embryos [red dots; n(TAM E8.5) = 7 embryos]. The fraction of labelled non-VP cells induced with the Pdx1-Cre Tg was significantly higher compared to the TAM-inducible Pdx1-CreERT one. Mean ± s.d. Two-tailed Mann–Whitney test. P = 0.003. h, Bar chart showing the percentage (%) of embryos with labelled LV cells following constitutive Pdx1-Cre [(5/5), 100%] or TAM-inducible Pdx1-CreERT lineage tracing strategies [TAM E8.5: (3/7), 42.9%]. i, Simulation of Pdx1-CreERT lineage tracing experiments (see Supplementary Note 1, Extended Data Fig. 4d). Fraction of labelled cells in LV* population (ratio of the sum of LVs and LVsp cell counts vs. the sum of LVs, LVsp, LV, and LVp cell counts) is shown for different combinations of label induction time point and recombination efficiency. The maximal fraction of labelled LV* is 21%. The fraction of labelled cells increases with higher recombination efficiency, reflecting the difference in labelled LV cells observed in vivo between Pdx1-Cre and Pdx1-CreERT lineage traced embryos (c–h). For any chosen recombination efficiency, the fraction of labelled LV cells decreases at later induction time points, corroborating the results from Pdx1-CreERT lineage tracing experiments (d, e). Specifically, induction time points after 30ss yield a low % of labelled LV cells, independent of the recombination efficiency. j, Simulations of three induction time points that correspond to the experimental setup (a) are shown.

Source data

Extended Data Fig. 9 Distinct marker gene signatures define hepatic and pancreato-biliary subpopulations.

a–g, Violin plots of normalized log-expression values of selected cell-type-specific marker genes (a–d) and genes encoding components of Fgfr2 and Robo-Slit (e), canonical and non-canonical Wnt (f) and Hedgehog signalling (g) pathways in distinct progenitor subpopulations from the sc-RNA-seq data set generated in this study. Data from E8.5 bipotent endoderm progenitors (BEP), E10.5 hepatic progenitors (LV), E10.5 intermediate progenitors (IMP), and E10.5 ventral pancreato-biliary progenitor cells (PB cells) are shown. h–k, Marker gene expression projected on t-SNE plots. Cells are projected into t-SNE space, as in n, but are coloured by the relative expression of indicated hepato-pancreato-biliary marker genes instead of cluster assignment. Colours span a gradient from red (high expression) to grey (low expression). IMP cells are circled by a black line. l, m, Representative IF images of E10.5 embryos stained for Sox17 (red), Pdx1 (blue), Sox9 [green; (l)], and Hnf1β [(green); (m)]. Right panels show higher magnifications of the boxed regions as merge (l) or single channels (m). In l, arrowheads mark cells bordering duodenum (DUO), LV, and PB buds (boxed region) showing low levels of Sox9 (green) and Pdx1 (blue). In (m), Hnf1β is abundant in Sox17+ cells of the PB bud (boxed region 1) but low at the boundary between DUO, LV, and PB (boxed region 2). Scale bars, 100μm. n, Distinct hepatoblast and hepato-mesenchymal gene signatures define subpopulations in the E10.5 liver bud. t-SNE plot visualization of sc-RNA-seq from BEP, E10.5 LV, PB, dorsal pancreatic (DP) and E14.5 pancreatic cells (PAN) (see Fig. 3a). LV cells were found in two clusters, referred to as LV-A (dashed black line) and LV-B (solid black line). o, Heatmap of average expression levels of selected genes in E10.5 LV-A and LV-B subpopulations. LV-A displayed markers for hepatoblast cell type, while LV-B for hepato-mesenchymal hybrid progenitors, recently reported by Lotto et al. 25. Both subpopulations expressed Prox1. p–s, Marker gene expression projected on t-SNE plots. Cells are projected into t-SNE space, as in n, but are coloured by their relative expression of indicated hepato-mesenchymal hybrid progenitor markers. Colours span a gradient from red (high expression) to grey (low expression).

Source data

Extended Data Fig. 10 Sc-RNA-seq data set integration of liver and pancreato-biliary progenitors.

a, t-SNE plot visualization of the publicly available sc-RNA-seq data set from Li et al. 16 of dorsal pancreatic cells [DP; E9.5 (30 cells), E10.5 (84 cells)], ventral pancreato-biliary cells [PB; E9.5 (44 cells), E10.5 (210 cells)], and hepatic progenitors [LV; E10.5 (22 cells)]. Cells are coloured by tissue of origin (upper panel), or embryonic stage (lower panel). b–g, Marker gene expression projected on t-SNE plots. Cells are projected into t-SNE space, as in a, and coloured by the relative expression of the indicated hepato-pancreato-biliary marker genes. Colours span a gradient from red (high expression) to grey (low expression). Intermediate progenitor (IMP) cells, which show an intermediate hepato-pancreato-biliary gene signature and locate between LV and PB clusters, are circled by solid black line. h, t-SNE visualization of Seurat integration of Li et al. 16 (dots) data set and sc-RNA-seq data generated in this study (triangles) (all data sets merged and scaled together). Cells are coloured by cluster identity (left panel; cluster assignment based on similarities in gene expression profiles) or by embryonic stage (right panel). A subset of E10.5 PB cells, from both data sets, shows an intermediate hepato-pancreato-biliary gene signature and locate between LV and PB clusters (see outlined single cells). Abbreviations: EP, endocrine progenitors; PAN, pancreas. i–n, Marker gene expression projected on t-SNE plots. Cells are projected into t-SNE space, as in h, but are coloured by the relative expression of the indicated hepato-pancreato-biliary marker genes. Colours span a gradient from red (high expression) to grey (low expression). Dotted circles indicate IMP cells. o, Dot plot showing the expression of a subset of hepato-pancreato-biliary marker genes across cell clusters identified in the integrated data sets. Dot size represents fraction of cells in each cluster expressing the marker, colour shows mean expression levels. Colours span a gradient from blue (high expression) to red (low expression). p, Biological process GO term analysis for genes enriched in IMP cells compared to LV and PB progenitor populations in the merged data sets. q–s, E8.5 bipotent endoderm progenitor (BEP), E10.5 LV, and PB cells of the two integrated data sets [excluding DP and E14.5 pancreatic (PAN) cells in the original analysis] were used to construct a pseudotime trajectory. Monocle 353 arranged the sequenced cells into a branched trajectory with BEP as starting state and LV or VP/EP as end states. Major cell trajectories are visualized with UMAP but coloured as per cell cluster (q), pseudotime (r), or displaying IMP cells (s). The lines correspond to the principal graph learned by Monocle 3.

Source data

Extended Data Fig. 11 Extended characterization of developmental potential and signalling signatures in hepato-pancreato-biliary subpopulations.

a, Schematic representation of transgenes used for lineage tracing of intermediate progenitor (IMP) cells. b, Representative IF of E10.0 (29ss) cryosections stained for pH3, Pdx1, Prox1 (left), and Krt19 (right). Left and right panels show IFs of consecutive sections of the same embryo. DUO, duodenum; LV, hepatic progenitors (Prox1+); PB, pancreato-biliary progenitors (Prox1+/Pdx1+; left panel, white dashed line); IMP (Pdx1+/Krt19+; right panel, white dashed line). Scale bar, 100μm. c, Quantification of proliferating (pH3+) cells in E10.0 (29-31ss) dorsal pancreas (DP), LV, PB and IMP domains on IF-stained cryosections (b). Graph shows the fraction of pH3+ cells as percentage of the total cell number in each organ rudiment [n(LV, PB, DP) = 4 embryo; n(IMP) = 6 embryos]. Grey lines connect samples in which pH3+ cells were counted on consecutive sections from the same embryo. No statistically significant differences in the size of the proliferative cell fractions were detected. Even though the IMP cells do not proliferate more than the surrounding compartments, it is possible that when they exit from their niche they might be recruited into a putative transit amplifying cell compartment and undergo increased rounds of division, as shown for example in the skin stem cell compartment58. Mean ± s.d. Two-tailed Kruskal–Wallis test with Dunn’s multiple comparisons test, ns. d–h, Spatial representation of Krt19-CreERT lineage tracing experiments. IF images of cryosections from E11.5 Tg(Krt19-CreERT;R26R-H2B-GFP) embryos stained for Prox1, Pdx1, and GFP were digitalized to obtain xyz-coordinates for GFP- (red) and GFP+ (orange) liver (LV) as well as GFP- (blue) and GFP+ (green) PB cells44. Each plot shows data from an individual Tg(Krt19-CreERT;R26R-H2B-GFP) embryo with labelled cells in LV and PB organ domains. Cells are plotted according to their xy- (upper panels) or xz-coordinates (lower panels). Close proximity of labelled cells in different organs, such as in (d, f, g) strongly suggests that cells descended from a common multipotent progenitor at E9.5. i, j, Representative cryosections of E9.5-E10.5 embryos stained for the indicated markers. At E9.5, Flrt2 and Flrt3 co-localize in the PB bud, but mark distinct subpopulations within the PB rudiment at later stages. At E10.0-E10.5, Flrt2 is enriched in the Pdx1+ ventral pancreas (VP), while Flrt3 is enriched in the Pdx1- gall bladder (GB). Krt19 is enriched in intermediate (IMP) subpopulation at the border of DUO, LV, and PB buds. Scale bars, 100μm. k, l, Representative IF images of the E10.5 gut boundary between LV, PB and DUO for the indicated markers. Shh (green) is expressed in the DUO (k). Krt19 staining (green) defines the border zone in l, encompassing IMP cells. Right panels show higher magnifications of the boxed regions as merge or single channels of Shh (k) and Krt19 (l) staining. Dotted line demarcates the border between IMP and DUO. Hoechst dye was used as nuclear counterstain. Scale bars, 100μm. m, Representative images of the E10.5 embryonic neural tube stained for cilia marker Arl13b and Shh pathway components Smo and Gli2. The ventral neural tube region (vnt) served as positive control for active Hedgehog signalling27,28, showing high density of Smo localization at the primary cilia (Arl13b+) together with Gli2 (bottom panel), while it is absent in the dorsal neural tube region (dnt)27,28. Insets show higher magnification of boxed regions 1 and 2, arrowheads indicate colocalization of Gli2 and Arl13b at the tip of a subset of primary cilia in the vnt. n, IF for Arl13b (left) and Smo (right) on E10.5 hepato-pancreato-biliary region. DUO, PB, and IMP are outlined by dashed white lines. Hoechst dye was used as nuclear counterstain. The density of Smo+ primary cilia is high in cells of the DUO and IMP but low in the PB bud. Scale bars, 10μm. o, IF for Arl13b (red, left) and Smo (red, right) shows absence of Hedgehog signalling (that is, no ciliary Smo signal) in the DP, as marked by E-cadherin (Ecad; green). Right panels show higher magnifications of the boxed regions as for the indicated channels in grey scale. Hoechst dye was used as nuclear counterstain. Scale bar, 100μm. p, IF staining shows abundant levels of Shh (green) in the lung, vnt, and notochord (nc), as previously published27,28. Hoechst dye was used as nuclear counterstain. Scale bar, 100μm. q–t, t-SNE plots showing the expression of Flrt2 (q), Flrt3 (r), Robo2 (s), and Ift57 (t) in our sc-RNA-Seq data set. Colours span a gradient from red (high expression) to grey (low expression).

Source data

Extended Data Fig. 12 Establishment of ventral foregut explants as a model system to manipulate the Hedgehog signalling pathway ex vivo.

a, Schematic representation of the experimental workflow to collect and culture ventral foregut explants from E9.5 mouse embryos. White dashed lines indicate incision sites for microdissection of the posterior ventral foregut and anterior midgut (containing the hepatic and pancreato-biliary organ rudiments) for ex vivo culture. b, Representative 3D images of whole-mount ventral foregut explants IF stained for Prox1, Pdx1, and Krt19. Prox1 marks hepatic (LV; red), Prox1/Pdx1 pancreatic (VP; white), and Prox1/Krt19 intermediate progenitor (IMP; green) domains. Explants were treated with 2μM Smoothened agonist (SAG) or 5μM KAAD-cyclopamine for 24h or left untreated (Control). Top panels show raw data in 3D projection mode. In middle panels, Prox1+ tissues were highlighted using the manual surface creation tool in Imaris (Bitplane) to quantify the volume of distinct hepato-pancreato-biliary organ domains. Bottom panels show 3D renderings for distinct organ domains generated and quantified using the automatic surface creation tool with manual thresholding in Imaris (Bitplane). Scale bars, 100μm. c, Representative images of whole-mount IF of a ventral foregut explant for Prox1 (red), Pdx1 (white), and Krt19 (green) (left panel). After antibodies elution, the same sample was re-stained for Pdx1 (white), Sox17 (red), and Cdx2 (green) to mark the gall bladder (GB; Pdx1-/Sox17+; white dotted line) and intestine (Cdx2+) (right panel). d–f, Quantification of the volume of distinct organ domains in ventral foregut explants following SAG treatment (Control, n = 12; 2μM SAG, n = 7). Plots show LV/GB (d), GB/IMP (e), or GB/VP (f) volume ratios. Mean ± s.d. SAG led to a reduced GB/IMP volume ratio [(e); two-tailed Mann–Whitney test; P = 0.01]. g, Schematic representation of the hepato-pancreato-biliary region in the E10.5 mouse embryo. The PB bud contains an IMP domain (green) that borders the LV and the duodenum and receives different signalling cues (green, red, blue arrows) from surrounding epithelial and mesenchymal tissues. Cells within this domain exhibit an extended cell fate plasticity during organogenesis (E8.5-E11.5) and retain the unique ability to contribute (black arrows) to the pancreatic bud (blue), GB (orange), and LV (red). Markers of the different progenitor domains and signalling pathways involved in hepato-pancreato-biliary development are listed on the left.

Source data

Supplementary information

Supplementary Information

This file contains Supplementary Text, Supplementary Note 1 (Description of the mathematical models), Supplementary Note 2 (Generation of the Tg(Prox1-rtTA) transgenic mouse line) and Supplementary References

Reporting Summary

41586_2021_3844_MOESM3_ESM.xlsx

Supplementary Table 1 Description of the cell populations considered in the models. Table including the description of the cell populations considered in the mathematical models, their acronyms, and attributed phenotypic markers related to Fig. 1, Extended Data Fig. 3, Supplementary Note 1.

41586_2021_3844_MOESM4_ESM.xlsx

Supplementary Table 2 Experimental data and sums of cell populations in models A and B. Table showing the experimental data and their corresponding sums of cell populations in models A and B.

41586_2021_3844_MOESM5_ESM.xlsx

Supplementary Table 3 Experimental data and sums of cell populations in models C and D. Table showing the experimental data and their corresponding sums of cell populations in models C and D.

Supplementary Table 4 Rate parameters of the models. Table listing the rate parameters of the models [1/h].

41586_2021_3844_MOESM7_ESM.xlsx

Supplementary Table 5 Initial conditions of the models [cells]. Table listing the initial cell numbers for each population in the different models.

41586_2021_3844_MOESM8_ESM.xlsx

Supplementary Table 6 Initial conditions of the extended model D and redistribution of cell counts at the time point of label induction. Table describing the initial cell numbers for each population in extended model D and the rates at which cells are redistributed to spike in cell populations at the time point of label induction.

Supplementary Table 7 List of antibodies. Table including all primary and secondary antibodies used in the study.

Source data

Rights and permissions

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Willnow, D., Benary, U., Margineanu, A. et al. Quantitative lineage analysis identifies a hepato-pancreato-biliary progenitor niche. Nature 597, 87–91 (2021). https://doi.org/10.1038/s41586-021-03844-1

Download citation

  • Received:

  • Accepted:

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1038/s41586-021-03844-1

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

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