Main

The intestine forms from the pseudostratified gut tube, which becomes patterned during late fetal development into villi and a continuous intervillus region formed by LGR5− and LGR5+ cells, respectively10 (Fig. 1a, Extended Data Fig. 1a–c). The continuous intervillus region is the major site of proliferation in the developing intestine (Extended Data Fig. 1d–f), and crypts subsequently form from this region postnatally11. Despite the apparent transcriptional similarity between fetal and adult LGR5+ cells4, it remains unclear how the fetal immature intestine transitions into the mature structure and how this is orchestrated at the cellular level. In particular, it is not known whether a specialized subset of fetal cells becomes adult intestinal stem cells or whether stem-cell identity is an induced property.

Fig. 1: Fetal LGR5 progeny contribute to the adult intestinal epithelium, but are insufficient to sustain intestinal growth during development.
figure 1

a, Detection of LGR5–eGFP (green) and DAPI (blue) at the indicated time points. Scale bars, 100 μm. Representative images of n = 3 biologically independent samples at each time point are shown. b, Detection of E-cadherin (E-cad, cyan), GFP (green) and RFP (red) in tissue whole mounts from proximal small intestine isolated from Rosa26-lsl-Confetti;Lgr5-eGFP-ires-creERT2 mice at P0 (n = 3 mice), P5 (n = 3 mice), P11 (n = 3 mice) and adulthood (n = 6 mice) following induction at E16.5 by the administration of 4-hydroxytamoxifen. White arrowheads indicate the clones depicted in the white dashed boxes at higher magnifications. Scale bars, 250 μm. c, Relative volume (projected) of clones from b and epithelium based on E-cadherin (P0, n = 9; P5, n = 9; P11, n = 9; adult, n = 9 mice). Dots indicate independent samples; data are mean ± s.e.m. d, e, Model based on LGR5+-cell-driven morphogenesis and assessment of the observed and predicted clonal expansion (experiment clones LGR5+: P5, n = 3; adult, n = 6; tissue: P5, n = 9; adult, n = 9). Data are mean ± s.e.m.

Source data

Full size image

To investigate the role of fetal LGR5+ cells in the establishment of the adult intestinal stem cell population, we performed lineage tracing on this population from embryonic day (E)16.5. Focusing on the proximal part of the small intestine, we observed that, consistent with previous reports3,4,12, progeny of the LGR5-expressing population were maintained into adulthood and thereby contributed to the adult intestinal stem-cell compartment (Fig. 1b). Most of the clones observed at postnatal day (P)0 were, as expected, located in the intervillus regions (Extended Data Fig. 2a). Moreover, it was not until P11 that clones extended as ribbons from the base of crypts to the tips of villi (Supplementary Video 1).

The quantitative contribution from LGR5+ progeny, labelled at E16.5, was slightly greater than the overall degree of tissue expansion (Fig. 1c, Extended Data Fig. 2b–e). This confirmed that LGR5+ cells were an important source of tissue growth. However, given that LGR5+ cells constituted only a small fraction of the cells (fraction of LGR5+ cells/total cells, f = 7.0% ± 0.9%, mean ± s.e.m.) in the proximal part of the small intestine at the time of labelling (Extended data Fig. 2f–h), we reasoned that if LGR5+ cells were the main source of adult epithelium (Fig. 1d) they would have to expand by a ratio 1/f greater than overall tissue to fuel growth and replace cells outside the intervillus regions. Thus, LGR5+ clones should expand 130-fold from P5 to adulthood, nearly an order of magnitude higher than the measured value (Fig. 1e). Expansion of LGR5+ progeny was thus insufficient to explain tissue growth.

To resolve the cellular diversity in the epithelium at E16.5, we performed single-cell RNA sequencing (scRNA-seq) analysis. Consistent with our characterization of LGR5–eGFP, Lgr5 was detected in 7% of the 3,509 cells analysed and—despite detecting only goblet cells by immunostaining—we identified other differentiated cell types, including Paneth cells (Lyz1), entero-endocrine cells (Chga) and enterocytes (Alpi) (Extended Data Fig. 3a, b). In the adult epithelium, the differentiated villi compartment can be separated into at least five transcriptionally distinct populations13. In the fetal intestine, these largely collapse into two populations, and a gene signature for crypt proliferation was detected beyond the LGR5+ compartment, including cells expressing differentiation markers14 (Extended Data Fig. 3c–e). This strongly supported the conclusion that cells in the fetal intestinal epithelium were distinct from their adult counterparts, and that cells expressing differentiation markers had not completed their differentiation program.

To test experimentally how cells outside the intervillus region contributed to tissue growth, we performed fate mapping using a ubiquitously expressed keratin 19 (Krt19)-driven Cre model (Fig. 2a; Extended Data Fig. 3f–h). Although the scRNA-seq data revealed that 49% of Krt19-expressing cells at E16.5 score positive for the proliferation signature, the expansion of clones closely mirrored the overall growth of the tissue (Fig. 2b; Extended Data Fig. 3i), which confirms that Krt19-expressing cells were representative of the tissue. However, we found that both the long-term persistence (defined as the fraction of surviving clones) and size of KRT19-labelled clones were very similar to their LGR5-labelled counterparts (Fig. 2c, d, Supplementary Information, ‘Supporting clonal data’). Several independent measurements confirmed that KRT19 marked a population of cells distributed randomly along the villus–intervillus axis (Extended Data Fig. 3j, k, Supplementary Video 2). Moreover, apoptotic cells at the tips of villi appeared only from P7; this means that KRT19+ clones cannot be lost before P5, and therefore that both P0 and P5 can be used for normalizing clonal persistence (Extended Data Fig. 3l). This strongly suggests that fetal epithelial cells, irrespective of location, contain long-term self-renewal potential and possess the same likelihood of contributing to the adult intestinal stem-cell compartment (Fig. 2e). Of note, although the expansion of the LGR5-targeted clones was slightly higher than that of tissue and KRT19-targeted clones, the difference was small (Fig. 2f). This small difference is likely to reflect a transient proliferative advantage due to their favourable location at the time of induction. Together, these findings indicated that all fetal cells have a similar capacity to contribute to the growth of the intestinal epithelium during development and give rise to adult intestinal stem cells, suggesting that cell fate potential is not fixed in the fetal epithelium.

Fig. 2: Random distribution of intestinal stem cell precursors in the fetal epithelium.
figure 2

a, Detection of E-cadherin (cyan), GFP (green) and RFP (red) in tissue whole mounts from proximal small intestine isolated from Rosa26-lsl-Confetti;Krt19-creERT mice at P0 (n = 1 mouse), P5 (n = 3 mice), P11 (n = 6 mice) and adulthood (n = 3 mice) following induction at E16.5 by the administration of 4-hydroxytamoxifen. White arrowheads indicate the clones depicted in the white dashed boxes at higher magnifications. Scale bars, 250 μm. b, Volume (projected) of clones from the Krt19-creERT induction (from a). Each dot represents a mouse. Data are mean ± s.e.m. c, Relative number of clones (projected persistence). Each dot represents an independent biological sample at the indicated time point (from a and Fig. 1b). Data are mean ± s.e.m. d, Volume of individual clones (Krt19-creERT: P0, n = 94; P5, n = 244; P11, n = 103; P36 (adult), n = 42; Lgr5-eGFP-ires-creERT2: P0, n = 28; P5, n = 39; P11, n = 15; adult, n = 18). Each dot represents an independent biological sample and the line indicates the mean. e, Model based on morphogenesis relying on equipotent stem cells randomly distributed in the tissue. f, Assessment of the observed and predicted clonal expansion (experiment clones KRT19+: P5, n = 3; adult, n = 3; experiment clones LGR5+: P5, n = 3; adult, n = 6; tissue: P5, n = 9; adult, n = 9). Data are mean ± s.e.m.

Source data

Full size image

This finding, however, raised a conundrum, as this implied that cells on villi—which are usually considered as committed to differentiation—contribute substantially in the long term to intestinal growth. Given the influence of gut epithelial geometry on cell fate3,15, we hypothesized that extensive cell rearrangements linked to continuous ‘villification’ could explain this paradox. Between E16.5 and birth, the total number of villi increased approximately sixfold, before reaching the final number characterizing the adult epithelium by P5 (Extended Data Fig. 4a, b). Therefore, we posited that global 3D epithelial sheet movements during villification could provide a mechanism that enables relocation of villus cells to the proliferative intervillus regions and de novo intervillus formation. Detailed 3D reconstruction of the E16.5 intestine revealed that around 30% of villi presented irregular structures of doublet form, with shared mesenchymal clusters within a specific interval of villus heights (Fig. 3a, Extended Data Fig. 4c–e, Supplementary Videos 3, 4). We speculated that this could be a sign of a single villus undergoing fission. Of note, cells found at the juncture between these structures were LGR5− (Extended Data Fig. 4f) and often proliferative (68% of junctures analysed; Fig. 3b, Extended Data Fig. 4g–h, Supplementary Video 5), even though the fission points were distant from the proliferative intervillus regions (Extended Data Fig. 4i). To challenge these observations, we turned to explant cultures, which—despite the intestine not expanding in length—revealed fission events evident from the top of single villi (Fig. 3c, Extended Data Fig. 4j, Supplementary Videos 6, 7). Thus, villi underwent active remodelling, causing pronounced reorganization of the epithelial cell layer.

Fig. 3: Villus fission is required to explain epithelial expansion.
figure 3

a, Detection of E-cadherin (red) and PDGFRA (yellow) in intestinal whole mount during villus formation. A representative image at E16.5 is shown (n = 3 mice). Arrowheads indicate villi with shared mesenchyme. Scale bar, 250 μm. b, Labelling with 5-ethynyl-2′-deoxyuridine (EdU, green) and PDGFRA (yellow) and E-cadherin (red) immunofluorescence in intestinal whole mounts following a 1-h EdU pulse at E16.5. Arrowheads indicate proliferative cells at the edge of a putative branching villus (n = 3 mice). Scale bar, 100 μm. c, Image showing the start and end points from live imaging of a villus undergoing fission (red star) (Supplementary Video 5). White circles indicate reference villi. Three mice were analysed. Scale bar, 50 μm. d, Model for cell repositioning based on villi emerging from the intervillus regions or through villus fission. Clones from villi can recolonize the intervillus region and clones from the intervillus region can recolonize villi. e–g, Simulation of the clonal dynamics (derived from LGR5+ and KRT19+ cells) using the cell-repositioning model to predict the fraction of clones (rootedness) rooted in the intervillus region (e), clonal persistence (experimental data are also shown in Fig. 2c) (f) and mean clone size (g). (LGR5 data: P0, n = 3; P5, n = 3; P11, n = 3; P36 (adult), n = 6 mice; KRT19 data: P0, n = 1; P5, n = 3; P11, n = 6; adult, n = 3 mice). Data are mean ± s.e.m. h, Detection of keratin 20 (KRT20, magenta), GFP (Lgr5-DTR-eGFP) and DAPI (cyan) at E16.5 in proximal small intestine. A representative image of n = 3 independent biological samples is shown. i, Detection of E-cadherin (cyan), GFP (green) and RFP (red) in tissue whole mounts from the proximal small intestine isolated from Rosa26-lsl-Confetti;Krt20-cre-T2A-creERT2 mice at P5 (n = 4 mice), P11 (n = 3 mice), and adulthood (n = 5 mice) following induction at E16.5 by the administration of 4-hydroxytamoxifen. White arrowheads indicate the clones depicted in the white dashed boxes at higher magnifications. j–l, Simulation of the clonal dynamics (KRT20-derived) using the cell-repositioning model to predict the fraction of clones rooted in the intervillus region (j), clonal persistence (experimental data for LGR5 are shown in Fig. 2c) (k) and mean clone size (l). LGR5: P0, n = 3; P5, n = 3; P11, n = 3; adult, n = 6 mice analysed; KRT20: P0, n = 3; P5, n = 4; P11, n = 3; adult, n = 5 mice. Data are mean ± s.e.m.

Source data

Full size image

On the basis of these findings, we turned to a biophysical modelling approach to address whether a minimal model of villus fission could be sufficient to explain the observed long-term equipotency of epithelial cells at E16.5. We modelled both the neutral drift dynamics of proliferative cells restricted to intervillus regions, and the bottom-up influx of cells driving villus elongation, using experimentally derived cell kinetics and geometrical parameters (Supplementary Information ‘Supplementary theory’, Extended Data Fig. 5). Additionally, we incorporated stochastic villus fission (between E16.5 and P4), upon which cells underwent positional changes and a fraction of cells from the parent villus became assigned to the new intervillus region (Fig. 3d). To inform the model, we used the experimentally inferred growth rate of villi number (Extended Data Fig. 6a), so that villus duplication and rapid underlying cell proliferation had a role in fuelling epithelial growth. Within this simplified 2D framework, which takes into account the key features of a more complex 3D process, fission events resulted in the reversible transfer of cells between villus and intervillus regions. We found that, with the experimentally inferred fission rate, just a few days were sufficient to fully erase the positional advantage of progeny from LGR5+ cells, resulting in long-term equipotency among all epithelial cells (Fig. 3e–g). In addition to good reproduction of the global evolution of mean clone size, clonal persistence and rootedness (Fig. 3e–g), the model explained several non-trivial aspects of the observed short-term clonal dynamics. In particular, the model predicted that KRT19+ clones should initially be larger and more dispersed than LGR5+ clones because, even though KRT19+ cells were predominantly induced in non-proliferative regions, they participated more readily in villi fission (Extended Data Fig. 6b–e). Furthermore, the model predicted that these KRT19+ clones should still be less rooted in intervillus regions than LGR5+ clones in the short term (Fig. 3e), with the two converging at later time points. These predictions showed consistently good agreement with the data (detailed statistics in Supplementary Information, ‘Supplementary theory’).

We also performed sensitivity controls on our simulations. First, we assessed the clonal dynamics, only enabling fetal fission from E16.5 to P0. This result showed that fetal remodelling was sufficient to explain the bulk of equipotency between LGR5+ and KRT19+ clones (Extended Data Fig. 7a–c). By contrast, when the model was applied—either in the absence of de novo villus formation or with villus fission without epithelial reshuffling—a satisfactory fit to the data could not be obtained (Extended Data Fig. 7d–i, Supplementary Information, ‘Supplementary theory’). Moreover, a biophysical model based on the current model for villification, which involves successive rounds of de novo villus formation initiating from the intervillus regions16, provided a poor fit to the data as this model does not allow the relocation of villus cells to intervillus regions (Extended Data Fig. 7j–o, Supplementary Information, ‘Supplementary theory’). Finally, our biophysical model predicted that the contribution from the KRT19+ population should be markedly reduced once villi fission slows down; this was corroborated by fate-mapping experiments starting at P0 rather than E16.5 (Extended Data Fig. 8a–c). This strongly suggested that villus remodelling events are required for recruitment of villus cells to the future stem-cell compartment.

To validate independently that cells specifically located on villi contributed to the adult intestinal epithelium, similarly to LGR5+-derived cells, we identified KRT20 as a marker of fetal and adult villi (Fig. 3h, Extended Data Fig. 9a–c). We observed that 27% of KRT20+ cells scored positive for the proliferation signature, even though these cells were confined to villi. Using a Krt20-driven Cre construct, it was evident that labelled cells in the adult epithelium were invariably lost, as reported for other differentiated cell types5,6 (Extended Data Fig. 9b). However, in the fetal epithelium, although the majority of labelled cells were initially detected in villi, clones persisted and generated adult intestinal stem cells (Fig. 3i, Extended Data Fig. 9d, Supplementary Information, ‘Supporting clonal data’). Quantitative analysis confirmed the predictions of the model for rootedness, persistence and size of Krt20 clones, demonstrating once again that intervillus and villus cells display similar behavioural dynamics (Fig. 3j–l). In sum, these findings confirmed that the long-term self-renewal potential of intestinal cells becomes sensitive to location soon after birth17—that is, after major structural changes of villi have ceased and cell migration streams are established.

To further investigate mechanistically the unexpected plasticity of cells from villus (LGR5−CD44−) and intervillus (LGR5+CD44+) regions (Extended Data Fig. 10a–d), we took advantage of a 3D culture system that allows cells to self-organize into villus and crypt domains18,19. When sorted from E16.5 proximal small intestine, both populations grew as spheroids, mimicking the immature nature of the fetal small intestine, with comparable plating efficiencies (Fig. 4a, Extended Data Fig. 10e, f). As reported for bulk cultures19, stimulation with WNT3A was sufficient to convert spheres to budding intestinal organoids, irrespective of their cell-of-origin (Fig. 4b). Notably, CD44+ and CD44− cells isolated from intestine in gestational week 8–10 human fetuses were capable of forming spheroids (Fig. 4c). Finally, to challenge the mouse cultures from the two cell populations, the mouse-derived spheroids were transplanted into conditioned mice20 (Extended Data Fig. 10g–j). Again, both cell populations engrafted with similar efficiencies and established CD44+ adult stem-cell niches at the bottom of crypts and gave rise to appropriate differentiated lineages including Paneth cells (LYZ1+), goblet cells (MUC2+) and enterocytes (ALPI+) (Fig. 4d, e, Extended Data Fig. 10k). Altogether, these findings demonstrated that fetal villi and intervillus cells had similar potential to grow in vivo and in vitro, and exhibited the same regenerative potential.

Fig. 4: Villus and intervillus cells are equipotent and have the same regenerative potential.
figure 4

a, Formation of spheroids from DAPI−EpCAM+CD44+ and DAPI−EpCAM+CD44– epithelial cells isolated from the proximal half of the small intestine of mice at E16.5. A representative image of n = 3 independent samples is shown. b, Spheroids following treatment with WNT3A. A representative image of n = 3 independent samples is shown. c, Formation of spheroids from DAPI−CD44+ and DAPI−CD44− epithelial cells isolated from the proximal half of the human fetal small intestine (eight weeks of gestation). A representative picture of n = 2 independent samples is shown. d, Engraftment of spheroids derived from DAPI−EpCAM+CD44+ and DAPI−EpCAM+CD44− epithelial cells obtained from E16.5 mT/mG mice. Four out of six mice were successfully engrafted with CD44−-derived cells and five out of five mice were engrafted with CD44+-derived cells. Scale bars, 10 mm. A representative image of independent engraftments is shown (CD44−, n = 4; CD44+, n = 5). e, Detection of CD44v6, LYZ1, MUC2 and ALPI in serial section of the engrafted patches (red). Scale bars, 50 μm. A representative image of independent engraftments is shown (CD44−, n = 4; CD44+, n = 5).

Full size image

These results of in vivo lineage tracing in combination with biophysical modelling and in vitro analysis of mouse and human fetal samples demonstrate that fetal intestinal cells—irrespective of their location and marker expression—represent equipotent precursors of adult intestinal stem cells (Extended Data Fig. 10l). We propose that the process of villi fission and tissue bending during intestinal morphogenesis provides a biophysical mechanism to remodel the local microenvironment, allowing cells to switch reversibly between LGR5− and LGR5+ states. This behaviour mirrors findings from various adult injury models, in which Lgr5+ cells lost through damage were subsequently replenished through recolonization by LGR5− cells5,6,8,9,21,22. Notably, the fission mechanism resembled the proposed model for villification in avians, in which longitudinal epithelial ridges generate a zig-zag pattern that breaks up into individual villi15 and villus geometry was proposed to define gradients of growth factors, thereby establishing distinct cell fates3. We propose that these growth-factor gradients impose specific expression patterns without irreversibly changing cell fates. This is evident from the profound cellular plasticity observed among cells in villi and intervillus regions with respect to their long-term self-renewal potential. This could be a generic property of developmental systems, ensuring the robust coordination of morphogenesis and fate specification23.

These findings have key implications for the improvement of differentiation protocols to, for example, generate intestinal stem cells from human induced pluripotent stem cells24,25. Consistent with our findings, fetal-like intestinal cells derived from human induced pluripotent stem cells generate cells with characteristics of adult stem cells only upon transplantation26,27,28,29. In this sense, the plastic fetal precursor state, which retains the capacity to be induced into an adult stem cell state, represents an attainable target for our strategies to direct differentiation. Understanding how these fetal precursors are induced to become adult stem cells will provide insight into how the process can be orchestrated in vitro. As fetal progenitors have been identified in multiple tissues, and all appear to mature in a similar manner upon transplantation26,27,28,29, the precursor state is likely to be a general intermediate during tissue maturation. It is therefore tempting to speculate that the plasticity observed during tissue regeneration could be explained via the equipotent progenitor state observed during fetal development.

METHODS

Mice

Unless otherwise specified, C57BL/6J mice (purchased from Taconic) were used for all the experiments. Transgenic murine lines used in the experiments have been previously described: Rosa26-mT/mG (ref. 30), Lgr5-eGFP-ires-creERT2 (ref. 2), Krt19-creERT (ref. 31), RAG2−/− (Taconic), Rosa26-lsl-Confetti (ref. 32), Lgr5-DTR-eGFP (ref. 22), Villin-cre33, Krt20-T2A-creERT2 (Jackson laboratories, stock no. 030600), and Rosa26-lsl-tdTomato34. CD1 mice (Janvier) were used as foster mothers in the lineage-tracing experiments.

None of the animals used in these studies had been subjected to prior procedures and were drug- and test-naive. All animals were housed in specific-pathogen-free animal facilities, in either open or individually ventilated cages always with companion mice, and cages were placed under a 12-h light–dark cycle. Food and water were provided ad libitum. Randomized cohorts including both male and female animals were distributed in an unblinded manner into the experimental time points for analysis. Sample sizes were selected to provide sufficient statistical power for analysis. The National Animal Ethics Committee in Denmark reviewed and approved all animal procedures (Permit numbers 2013-15-2934-00756 and 2013-15-2934-00927).

Lineage-tracing experiments and quantification

Pregnant females carrying embryos at E16.5 were intraperitoneally injected with the indicated amount of 4-hydroxytamoxifen (4OHT, Sigma) and progesterone (50 μg, Sigma) in corn oil (Sigma). Pregnant Lgr5-eGFP-ires-creERT2 and Krt20-T2A-creERT2 mice were injected with 1 mg 4OHT; pregnant Krt19-creRT2 mice were injected with 100 μg 4OHT. Pregnant Krt19-creERT;Rosa26mT/mG mice were injected with 1 mg 4OHT. Newborn Krt19-creERT;Rosa26-lsl-Confetti and Lgr5-eGFP-ires-creERT2;Rosa26-lsl-Confetti P0 mice were injected with 10 μg 4OHT. Lgr5-eGFP-ires-creERT2 mice were delivered by caesarean section at E18.5 and newborns were transferred to CD1 foster mothers. The proximal half of the small intestine was collected at the time points indicated in the figures. In Krt19-creERT;Rosa26-lsl-Confetti and Lgr5-eGFP-ires-creERT2;Rosa26-lsl-Confetti quantifications, only red clones were considered; for Krt20-T2A-creERT2, red and YFP clones were considered. The projections of volume, number of villi and number of clones (persistence) at the indicated time points were calculated by multiplying the volume/unit area, number of villi/unit area and number of clones/unit area by the total surface area of the proximal half of the intestine at each time point analysed.

Transplantation

Transplantation was performed using RAG2−/− mice as recipients, essentially as previously described20. In brief, colitis was induced in RAG2−/− mice (3–6 months old) by administration of 3.3% DSS (MP Biomedicals; molecular mass 36,000–50,000) in the drinking water for 5 days followed by normal water. Mice were subsequently infused with cultured intestinal epithelial cells derived from Rosa26mT/mG mice on day 8 and day 11 after the beginning of DSS treatment. Cells were cultured in Matrigel as described below. At passage 5, epithelial cells were released from matrix, and mechanically dissociated into epithelial sheets. After washing, cell fragments from approximately 600 organoids were resuspended in 300 µl 5% Matrigel in PBS. Under general anaesthesia (isoflurane; Piramal Healthcare), a flexible catheter was inserted into the colon of mice and the cell suspension was subsequently infused slowly into the colonic lumen. The anus was sealed with surgical histoacryl glue (B. Brawn) for 3 h to reduce flow in the colon. The animals were carefully monitored during the colitis and following transplantation. Two weeks after the second transplantation, recipient mice were killed and colons were collected and processed for analysis.

Human fetal samples

Human fetal tissue collection was approved by the regional ethics committee (permit number H-1-2012-007 and 258206). The adult patients gave their informed written and oral consent. None of the terminations were for reasons of fetal abnormality and the fetal tissue was available following elective termination of pregnancy during first trimester.

Imaging and histology

Whole-mount immunostaining

The proximal half of the small intestine was fixed with 4% paraformaldehyde (PFA) from 3 h to overnight at 4 °C. The samples were then dissected into smaller fragments (approximately 4 mm × 4 mm) followed by dehydration in methanol. Samples were stored in methanol at −20 °C. Then, the tissue was rehydrated in a PBS series before staining with all steps being performed at 4 °C. In brief, samples were blocked and permeabilized in 1% bovine serum albumin (Sigma), 0.5% Triton X-100 (Triton X-100 Surfact-Amps Detergent Solution, Thermo Fisher) in PBS for 24 h. Primary antibodies (indicated in Supplementary Table 1) were incubated in 1% bovine serum albumin (Sigma), 0.5% Triton X-100 in PBS for 48 h. Tissue was subsequently washed in 0.5% Triton X-100 in PBS overnight. Secondary antibodies (indicated in Supplementary Table 1) were incubated in 1% bovine serum albumin (Sigma), 0.5% Triton X-100 in PBS for 48 h and the tissue was subsequently washed overnight with 0.5% Triton X-100 in PBS. Diamidino-2-phenylindole dihydrochloride (DAPI; 1 µM; Sigma) was used to counterstain nuclei in the indicated experiments. Samples were then dehydrated in methanol and kept at −20 °C. Samples were cleared using 1:2 benzyl alcohol:benzyl benzoate (BABB) (Sigma). In brief, 5 changes (1 min each) of BABB:methanol 1:1, then 5 changes (2 min each) of BABB. The samples were subsequently mounted within a Fast well (FW20-FastWells 20-mm diameter × 1.0-mm depth or 25 × 25 mm, Grace-Biolabs). Z-stack images were acquired using laser-scanning confocal microscopy (Leica TSC SP8). Three-dimensional reconstructions were done using Fiji software and clone volume was assessed using Volocity v.6.3.0 (Perkin Elmer).

EdU staining

Mice were intraperitoneally injected with EdU (pregnant females carrying embryos at E16.5 with 250 μg; P0 with 25 μg; P5 with 31.25 μg, P11 with 37.5 μg and adult with 250 μg) and processed according to the manufacturer’s instructions (Click-iT Plus EdU Alexa Fluor 488 Flow Cytometry Assay Kit or Click-iT EdU Alexa Fluor 555 Imaging Kit, Thermo Fisher). Following a 1-h chase, the tissue was processed as described in the previous section. Before (2D staining) or after (whole-mount immunostaining) primary and secondary antibody staining, EdU was developed using Click-iT technology. DAPI (1 µM; Sigma) was used to counterstain nuclei in the indicated experiments

Two-dimensional immunostaining

Proximal halves of the small intestine were fixed with 4% PFA from 3 h to overnight at 4 °C and following dehydration, embedded in paraffin. Sections (5–8-µm thick) were prepared with a microtome.

The luminal perimeter was measured at the indicated time points in sections of small intestines stained with Haematoxylin QS (Vector) and Eosin Solution Aqueous (Sigma). Alkaline phosphatase staining was performed using Vector Red Substrate Kit (Vector). Images were acquired using NDP Zoomer Digital Pathology (Hamamatsu) and subsequently analysed in NDP.view2 software.

To perform immunohistochemistry and immunofluorescence on paraffin-embedded tissues, sections were rehydrated before antigen retrieval. Diva Decloaker solution (Biocare Medical) and Retriever 2100 (Aptum) were used at this step. Then, blocking and permeabilization was performed in 10% adult bovine serum (Sigma), 0.3% Triton X-100 in PBS for at least 1 h at room temperature. Primary antibodies (indicated in Supplementary Table 1) were incubated overnight in 10% adult bovine serum (Sigma) at 4 °C. Secondary antibodies (indicated in Supplementary Table 1) were incubated for 1–2 h at room temperature in 0.5% bovine serum albumin (Sigma) or in the ready-to-use solutions (Vector), when using horseradish peroxidase-conjugated secondary antibodies. For immunofluorescence, DAPI (1 µM; Sigma) was used to counterstain nuclei in the indicated experiments. For immunofluorescence, Alexa-Fluor-conjugated secondary antibodies were used (detailed in Supplementary Table 1), except with anti-CD44 antibody (indicated in Supplementary Table 1), when TSA Plus Cyanine 3 and Fluorescein System (Perkin Elmer) was used. For immunohistochemistry, secondary antibodies were developed using DAB (ImmPACT DAB Peroxidase (HRP) Substrate, Vector) and nuclei were subsequently visualized with Haematoxylin QS (Vector). Fluorescent images were acquired using laser-scanning confocal microscopes (Leica TSC SP8), images of engrafted patches were acquired using a Zeiss Imager M2 microscope equipped with ORCA-R2 Digital CCD camera C10600 (Hamamatsu), and immunohistochemistry samples were imaged using NDP Zoomer Digital Pathology (Hamamatsu). All images were subsequently analysed in NDP.view2 software, Fiji and Adobe Photoshop CS6.

Sequential sectioning of grafted epithelial patches

Two weeks after the second transplantation, recipient RAG2−/− mice were killed by cervical dislocation and colons were dissected. After washing with PBS, colons were opened longitudinally and imaged by stereomicroscopy (Leica M165FC). Engrafted patches were dissected from the colon and fixed with 4% PFA for 3 h followed by incubation in 20% sucrose solution in PBS overnight. Tissue fragments were embedded in OCT. Eight-micrometre sequential frozen sections were analysed histologically as described above.

Intestinal live imaging of explant cultures

The proximal half of small intestine was isolated from E16.5 embryos. The intestine was subsequently cut open longitudinally and attached from both ends of the intestine to a glass-bottomed plate using surgical histoacryl glue (B. Brawn). Explant cultures were covered in BGJb Medium (Thermo Fisher Scientific) supplemented with 0.1 mg/ml ascorbic acid (Sigma) and 1% penicillin–streptomycin (Life technologies, GIBCO) as previously described35. Samples were imaged using a Delta Vision deconvolution microscope (GE Healthcare) and Widefield Zeiss AxioImager microscope.

Isolation of epithelial cells from the fetal epithelium and purification by flow cytometry

Fetal small intestine from mouse or human fetuses was dissected and incubated with collagenase solution (125 µg/ml in PBS) for 45 min at 37 °C and subjected to vigorous pipetting every 15 min using a P1000 pipette. Released cells were pelleted and resuspended in PBS supplemented with 1% BSA and incubated with fluorescent-conjugated primary antibodies (indicated in Supplementary Table 1) for 30 min on ice. After washing, DAPI (Sigma; 1 µM) was added to the cell suspension to facilitate exclusion of dead cells by flow cytometry. Purified cell populations were subsequently isolated using a FACSAria I or III (BD Bioscience). After isolation, cells were embedded in growth factor reduced Matrigel (Corning) in the presence of human EGF (Peprotech; 50 ng/ml), murine noggin (Peprotech; 100 ng/ml), human noggin (Peprotech; 100 ng/ml), mouse R-spondin1 (R&D Systems; 500 ng/ml) as previously reported20. In some experiments Wnt3a-conditioned medium (50% in volume) was added as a supplement as indicated. Prostaglandin E2 (2.5 μM, Sigma) was added to the human fetal cultures. In all experiments using mouse organoids, cells were collected from wild-type C57/BL6 mice (Taconic). Cells were cultured in a dome-shaped 25-µl extracellular matrix droplet and 250 µl culture medium was added to each well in a 48-well plate (flat bottom; Corning). Medium was subsequently changed every 2–3 days. For human samples we observed no discernable effect of whether the pregnant mother was a smoker or not.

Single-cell sequencing

Cells were isolated from a pool of six proximal small intestines (1 cm in length each) at E16.5 using collagenase treatment and cell sorting as described in the previous section. DAPI−CD31−CD45−EpCAM+ cells (20,000 cells) were processed using 10x Genomics Chromium protocols (version 2 chemistry).

RNA-sequencing data analysis

Reads were de-multiplexed, aligned and counted using the 10x Genomics Cellranger v.2.1.0 software36. The refdata-cellranger-mm10-1.2.0 reference was downloaded from the 10x Genomics website (https://support.10xgenomics.com/). The resulting filtered count matrix was loaded into Pagoda 2 (ref. 37) for data exploration and Seurat 2.3 (ref. 38) for visualization purposes. Genes that were expressed in fewer than 5 cells were removed. Cells expressing less than 200 genes, more than 6,000 genes or more than 10% mitochondrial genes were removed. Before filtering, the dataset consisted of 3,770 cells. After quality control, 3,509 cells remained.

Scores for signatures for adult epithelium single-cell zones 1–5 and the proliferation signature were calculated using Seurat’s AddModuleScore function. The corresponding gene lists were extracted from Supplementary Table 114,15.

Cells were considered as positive for a particular zone signature13 when their expression was higher than mean + 1 s.d. for that particular signature. Cells were considered as positive for the proliferation signature when their scores were >0 in Seurat 2.3’s normalized expression value. Similarly, cells were considered as positive for individual genes when their scores were >0 in Seurat’s normalized expression value.

Statistics

The number of biological and technical replicates and the number of animals are indicated in figure legends and text. All tested animals were included. Sample size was not predetermined. For all experiments with error bars, the s.e.m. or s.d. was calculated to indicate the variation within each experiment or sample. Statistical t-tests were used to assess significance in the indicated experiments.

Reporting Summary

Further information on research design is available in the Nature Research Reporting Summary linked to this paper.