RTK signalling promotes epithelial columnar cell shape and apical junction maintenance in human lung progenitor cells

The lung undergoes branching morphogenesis to build a tree-like structure during development. In the mouse embryonic lung, a SOX9⁺ID2⁺ epithelial population located at the distal branching tips is a multipotent progenitor population that gives rise to all lung epithelial lineages (Alanis et al., 2014; Rawlins et al., 2009). Recent work in the developing human lung has identified similar distal tip epithelial progenitors that are SOX9⁺SOX2⁺ (Danopoulos et al., 2019; Miller et al., 2018; Nikolic et al., 2017). Human epithelial progenitor cells derived from the pseudoglandular stage (∼5 to 17 post-conception weeks, pcw) have been cultured as self-renewing organoids in the presence of EGF, FGF and Wnt activation, and TGFB and BMP inhibition (Miller et al., 2018; Nikolic et al., 2017).

FGF10 plays a crucial role in the emergence of the mouse lung during development (Arman et al., 1999; Min et al., 1998; Sekine et al., 1999). Both in vivo and in vitro studies have shown that FGF10 promotes mouse lung branching at the pseudoglandular stage (Abler et al., 2009; El Agha et al., 2017; Bellusci et al., 1997; Taghizadeh et al., 2020; Volckaert et al., 2013; Weaver et al., 2000; Yin and Ornitz, 2020). Mesenchymal FGF10 acts via epithelial FGFR2, which activates MAPK/ERK signalling and promotes SOX9 expression and morphogenesis (Abler et al., 2009; Chang et al., 2013; Tang et al., 2011; Yin and Ornitz, 2020). However, recent studies on developing human lungs have not supported a crucial role for FGF10 (Danopoulos et al., 2019; Nikolic et al., 2017). In the mouse, FGF7 deficiency does not result in severe lung defects (Guo et al., 1996) and its role in development is thought to be secondary (Post et al., 1996; Scott et al., 1995). However, FGF7 is a key component in sustaining the in vitro culture of human tip epithelial organoids (Miller et al., 2018; Nikolic et al., 2017) and therefore considered likely to contribute to tip progenitor maintenance in vivo. FGF ligands are widely expressed in the developing human lung (He et al., 2022) and neither FGF7 nor FGF10 shows regionalized distribution (Danopoulos et al., 2019).

EGFR is also implicated in lung development. Egfr knockout mice show defects in lung branching and alveolar formation (Kheradmand et al., 2002; Miettinen et al., 1997). EGFR is expressed in human embryonic lungs undergoing alveolar differentiation (Klein et al., 1995). However, the function of EGF and EGFR in human lung development remains largely unknown.

We find that EGF, FGF7 and FGF10 have distinct functions in human lung tip progenitor cell shape maintenance. FGF7 promotes columnar cell shape in primary tip progenitor cells, whereas EGF and FGF10 cannot. Only FGF7 induces sustained MAPK/ERK and PI3K/AKT pathway activity when assayed using kinase translocation reporters. Moreover, inhibition experiments show that both downstream pathways are required for cell shape maintenance. The MAPK/ERK and PI3K/AKT pathways regulate proliferation, promote columnar cell shape and maintain junction organisation. We show that integrin signalling is a key pathway downstream of MAPK/ERK in the tip progenitor cells. Disrupting integrin alters cell polarity, cell adhesion and tight junction assembly. This work provides a convenient platform for dissecting RTK ligand function in epithelial cells, and will facilitate future studies of cell shape and cell fate coordination in organoid-based research.

We characterised the cell shape of tip and stalk epithelial cells in the developing human lung at the early pseudoglandular stage (7 to 13 pcw) and compared in vivo cell shape and arrangement with the tip cells expanded as self-renewing (SN) organoids. Tip progenitor cells (SOX9⁺) in vivo and cells in a SN organoid were columnar and apically constricted (Fig. 1A,B; Fig. S1A). Both in vivo tip progenitors and SN organoids are highly proliferative and contain rounded (proliferating) cells at the basal side (Fig. 1B; Fig. S1A, arrowheads). A 3D reconstructed SN organoid is partially reminiscent of the conformation of an extending tip in vivo (Movie 1). By contrast, in vivo stalk epithelial cells (SOX9⁻, adjacent to the tip) were more cuboidal (Fig. 1A). We quantified lateral, apical and basal lengths of the tip and stalk cells, and organoids (Fig. 1C; Fig. S1B-E), showing an overall conservation of tip cell shape in vitro, particularly of lateral cell length.

The seven-factor organoid self-renewing (SN) medium promotes proliferation, columnar cell morphology and organoid budding (Nikolic et al., 2017). To identify the specific components responsible for these effects, we removed medium components and examined organoid survival and morphology during continued passaging (Fig. 1D). When the Wnt agonist Chir99021 (Chir) or the SMAD inhibitors Noggin and SB431542 (SB) were removed, organoid growth terminated within 7 days (Fig. 1E). By contrast, after removal of all RTK ligands (EGF, FGF7 and FGF10) organoids survived for longer, suggesting that exogenous RTK inputs are not required for immediate cell survival (Fig. 1E,F). Removal of EGF alone, or FGF7/10, permitted organoid culture for slightly longer than removal of all three RTK ligands, although neither medium could sustain long-term culture with regular passaging (Fig. 1E,F). We noticed that organoids gradually became more spherical when FGF cytokines were removed, whereas removing EGF did not cause such clear morphology changes (Fig. 1F). These data suggest a requirement for FGFR2 and EGFR signalling in the long-term maintenance of the SN organoids. We therefore explored the spatial distribution of FGFR and EGFR receptors in vivo.

In agreement with previous findings (Danopoulos et al., 2019), in the developing human lung at the early pseudoglandular stage (7 to 13 pcw) FGFR2 is expressed throughout the tip, stalk and airway epithelium (Fig. 1G; Fig. S1F). Likewise, EGFR localized to tip, stalk and airway epithelial cells, and many mesenchymal cells (Fig. S1G). We further examined the expression of phospho-pan-FGFR and phospho-EGFR, and observed subsets of cells that were responding to RTK signals throughout the epithelial compartment (Fig. 1H; Fig. S1H). These data confirm that human lung distal tip epithelium is responding to FGF and EGF signalling in vivo, and we continued to explore the roles of these signals using the organoid model.

To confirm that the primary human lung epithelial tip progenitors respond to RTK signalling, we micro-dissected tip tissues and enriched for the epithelial cells using magnetic-activated cell sorting (MACS) (Fig. S2A). The EpCAM⁺ single cells organized into both spherical and budding organoids when provided with SN medium (Fig. S2A,B), and these were morphologically similar to the SN organoids previously established by plating whole tips (Nikolic et al., 2017). Similarly, the single tip cells were able to form small spheres in a basal medium containing Chir, RSPO, Noggin and SB only (Fig. S2A,B). These basal spheres could not be maintained long term in the basal medium (Fig. S2C). However, in response to the addition of RTK ligands, the basal spheres proliferated more quickly and underwent morphological changes (Fig. S2D,E).

Based on the plasticity of the basal spheres to respond to RTK ligands, we designed an experiment to identify the effects of individual RTK ligands on tip progenitor cells (Fig. 2A). We tested EGF, FGF7 and FGF10, the three RTK ligands in the SN medium, and FGF9, which is highly enriched in the tip epithelium in vivo (Nikolic et al., 2017). Spheres maintained in the basal medium remained small and spherical, and did not maintain stable SOX9 expression (Fig. 2B-G; Fig. S2F), consistent with SOX9 being downstream of FGFR signalling in the branching mouse lung (Chang et al., 2013). Organoids supplied with EGF became bigger, but remained spherical (Fig. 2B-D; Fig. S2F). Cells in EGF were SOX9 positive and more proliferative than those in the basal medium (Fig. 2F-H; Fig. S2F). Both FGF7 and FGF9 were sufficient to promote organoid budding and increase proliferation (Fig. 2B,C; Fig. S2F). Organoids in these two conditions were largely SOX9 positive, although we occasionally observed FGF7-treated organoids with patchy SOX9 expression (Fig. 2F,H; Fig. S2F,G). FGF7- and FGF9-treated cells also displayed columnar morphology, in contrast to the cuboidal-like cells in the basal spheres (Fig. S2H).

FGF10 did not have obvious effect on the basal spheres in terms of organoid morphology, proliferation or cell shape, although it can induce some SOX9 expression (Fig. 2B-H; Fig. S2F,H). These results suggest that, unlike its crucial function in the developing mouse lung (Abler et al., 2009; Bellusci et al., 1997; el Agha et al., 2014; Li et al., 2018; Ramasamy et al., 2007; Volckaert et al., 2013), FGF10 does not work as the most crucial RTK ligand in the developing human lung epithelial cells, consistent with previous publications (Danopoulos et al., 2019; Miller et al., 2018). Collectively, the results of the basal sphere stimulation experiment show that, in the presence of WNT activators (Chir and RSPO), RTK signalling is an upstream pathway promoting SOX9 expression and contributing to the proliferation of the human lung tip epithelial progenitors, consistent with mouse data (Chang et al., 2013).

Given the distinct organoid morphologies we observed in FGF7- and EGF-treated organoids (Fig. 2B,F), we focused on these conditions to investigate the effects on tip epithelial cells in detail. Significant organoid budding in response to FGF7 treatment, or the size increase after EGF treatment occurred between days 4 and 6 (Fig. S3A). We therefore investigated organoid proliferation and cell shape at days 5 and 10. We found that proliferation of cells in FGF7 and EGF treatments differed at day 5, whereas differences were no longer observable at day 10, possibly due to paracrine secretion of negative regulators of proliferation as cell/organoid density increases (Fig. S3B,C). Together with the observation in vivo that tip epithelial cells were more proliferative than stalk epithelium (Fig. S3D,E), we conclude that a high level of proliferation is correlated with branching in vivo and organoid budding in vitro. However, comparing basal spheres and EGF-treated organoids shows that proliferation alone is not sufficient to drive organoid budding.

To better understand the 3D volume of organoids and to inspect whether proliferating cells have a ‘preferred’ distribution or localization in the organoids, we took advantage of light sheet microscopy and carefully examined organoids of distinct architectures. Reconstructed images illustrated the complex, mostly irregular, structures of FGF7-treated organoids and spherical, or the doughnut-like organoids in the basal medium and after EGF treatment (Fig. 3A). We did not spot regionalized EdU⁺ cells in any of the day 10 organoids in the three medium conditions (Fig. 3A). Rather, we saw generally consistent rates of proliferation in each condition, based on quantitation at day 10 in both single z planes and 3D reconstructions (Fig. 3B).

We next sought to track SOX9 expression over time. The basal spheres largely lost SOX9 expression by day 5 (Fig. 3C). In contrast, FGF7-treated organoids sustained SOX9 over the time course (Fig. 3C). Interestingly, some organoids in the EGF treatment were not SOX9⁺ at day 5, although 100% were SOX9⁺ at day 10 (Figs 3C, 2H). We conclude that both FGF7 and EGF stimulation can re-activate SOX9 expression, which is not maintained by Wnt agonists alone in the basal medium. Moreover, SOX9 re-activation likely requires a persistent high level of signalling input and therefore took several days to occur. FGF7 may have greater capacity to activate signalling pathways upstream of SOX9 than EGF.

FGF7- and EGF-treated organoids were distinguishable by cell shape. Cells in the basal spheres were laterally flattened with a poor arrangement of ZO1 and E-cadherin, suggesting disruptions in tight and adherens junctions (Fig. 3D,E). In FGF7-treated organoids, we observed an increase in lateral cell length and an overall cuboidal-to-columnar transition over the time course (Fig. 3D,E). At day 10, FGF7-treated organoids were largely reminiscent of SN organoids (Fig. 3D). By contrast, although EGF treatment changed cell shape, it did not result in a clear cuboidal-to-columnar transition (Fig. 3D,E). Similarly, cells in the basal and EGF-treated spheres exhibited wider apical and basal surfaces than cells supplemented with FGF7 (Fig. 3C,F,G). Accordingly, we observed varied nucleus shape in some cells in the basal and EGF-treated spheres (Fig. 3C,H; Fig. S3F). The nuclei of these cells were wider (Fig. S3F), which might result from the lateral shortening. These data show that RTK signals control fate (SOX9 expression), cell shape and arrangement of the tip epithelial cells. These observations, that FGF7 promoted the organization of the cell junctions and cell-cell adhesion, are analogous to findings in the mouse lung where FGF10 induces genes that regulate cell adhesion (Jones et al., 2019; Lü et al., 2005).

The drastically different organoid morphologies and cell shapes that we observed in FGF7- and EGF-treated organoids prompted us to investigate the downstream signalling activity. We first detailed the expression pattern of pERK and pAKT in the developing human lung (see Fig. S9D for phospho-antibody validation experiments). In the pseudoglandular mouse lungs, pERK is predominantly expressed in the distal epithelia (Hirashima and Matsuda, 2022 preprint; Jiang et al., 2018; Liu et al., 2004; Tang et al., 2011) and pAKT can be detected in the epithelial cells (Wang et al., 2005; Yin and Ornitz, 2020). There was sample-to-sample variation in pERK and pAKT staining in human embryonic lungs. Nonetheless, a moderately high level of pERK was observed in both the tip and the stalk epithelial cells, whereas pAKT was expressed at a reduced level in the epithelium (Fig. 4A,B; Fig. S4A,B). We did not consistently observe regionalized enrichment of pERK or pAKT between the tip and the stalk epithelia (Fig. 4A,B; Fig. S4A,B).

Next, we stained for the two phospho-antibodies in organoids growing in the basal medium and FGF7 or EGF treatment, and quantified the per-cell expression levels (Fig. 4C-F). The basal spheres retained ERK and AKT activation, although at much lower levels than the organoids supplemented with FGF7 or EGF (Fig. 4C-F). FGF7-treated cells at day 10 had pERK and pAKT staining levels greater than cells that were treated with EGF (Fig. 4C-F). Of note, pERK and pAKT staining were heterogeneous in the FGF7- and EGF-treated organoids (Fig. 4C,E). This is expected, as the staining is a snap-shot of pathway activity at 10 days after ligand treatment.

To reveal the temporal dynamics of ERK and AKT signalling in organoids subjected to different RTK ligands, we adopted the kinase translocation reporter (KTR) system (Fig. 4G). The KTR system for ERK and AKT signalling reports on phosphorylation levels using the cytoplasmic-to-nuclear ratio of a fluorescent protein (Kudo et al., 2018; Maryu et al., 2016; Regot et al., 2014). We first validated the reporters, microscope set-up and image analysis by imaging the SN organoids (which have active MEK and AKT signalling) receiving a MEK or AKT inhibitor in the presence of a nuclear dye (Fig. S5A). The cytoplasm-to-nucleus fluorescence translocation indicated the reduction of ERK or AKT activity (Fig. S5B-F). Similarly, we confirmed that adding SN medium to MEK- or AKT-inhibited reporter cells could efficiently recover cytoplasm-enriched fluorescence localization (Fig. S5G-L). These control experiments confirmed that the KTR reporters functioned as expected in our system. We next imaged dynamic changes to ERK and AKT reporters at 10 min intervals for up to 2 h as MEK- or AKT-inhibited organoids received an FGF7 or EGF supplement from time 0 (T0, Fig. 4H,J). In these time-lapse images (Fig. 4H,J; Figs S6, S7), we could observe that, at T0, when ERK or AKT signalling was inhibited, the fluorescent reporter was located largely in the nuclei. Moreover, we were able to observe fluorescence translocation from the nuclei to the cytoplasm following ligand addition (Fig. 4H,K, arrowheads and dotted lines; Figs S6, S7). Although there was variation between organoids, in general, FGF7 supplementation could readily initiate nucleus-to-cytoplasm translocation of both ERK and AKT reporters (Fig. 4H-K; Fig. S6), whereas EGF could efficiently activate ERK, but not AKT (Fig. S7). We did not notice any spatial dynamics of the ERK and AKT reporter in the conditions tested. However, we could always identify more than one cell that demonstrated fluorescence translocation, which might indicate interactions between neighbouring cells. In summary, FGF7 could efficiently activate ERK and AKT signalling, whereas EGF was only effective on ERK phosphorylation.

We next asked how MAPK/ERK and PI3K/AKT signalling influenced SN organoid maintenance. We applied PD0325901 (a MEK inhibitor, hereafter MEKi) or MK2206 (an AKT inhibitor, hereafter AKTi) to the SN organoids to assess phenotypic changes and explore underlying molecular mechanisms (Fig. 5A; Fig. S8A). Both inhibitors displayed concentration-dependent effects on organoid budding, proliferation and cell shape (Fig. S8B-I). MEKi and AKTi application led to distinct phenotypic changes, with MEKi organoids being less proliferative and AKTi organoids having a bigger lumen (Fig. 5B; Fig. S8B,E,G). These observations suggested that the phenotypes were independently triggered by the disruption of each pathway, although the two pathways are known to interact (Rhim et al., 2016; Toulany et al., 2014). We subsequently characterized cell shape over a time course and found that the morphological differences became obvious at day 4 of inhibitor treatment (Fig. 5B). Although both MEK and AKT inhibition resulted in shortened cells and disrupted tight junctions, MEKi cells were more squamous, whereas AKTi cells were more cuboidal (Fig. 5B-E). At day 7 the MEKi cells, which had the shortest lateral membranes, had the longest basal membranes (Fig. 5B,F). We measured nucleus circularity at day 7 and found clear divergence of nucleus shape in the MEK- and AKT-inhibited organoids compared with SN cells (Figs 5H, 9A). Nuclear deformation was likely a consequence of the lateral shortening (Hatch and Hetzer, 2016; Keeling et al., 2017; Wang et al., 2022).

To investigate cell junctions at high resolution, we performed electron microscopy and observed striking differences in the SN, MEKi and AKTi cells (Fig. 5I; Fig. S9B). SN cells were overall columnar and apically constricted with basally localized nuclei that were rounded or ovoid. Microvilli and intracellular vesicles (yellow arrowheads) were visible at the apical surface of the SN cells and we could identify the apical tight junctions and adherens junctions (Fig. 5I; Fig. S9B, red and blue arrowheads). In contrast, MEKi and AKTi cells were squamous or cuboidal (Fig. 5I; Fig. S9B). Although we did not notice obvious defects in microvilli in MEKi and AKTi cells, the apical surface of these cells was elongated, and the tight junction complexes were either difficult to find or mislocated (Fig. 5I; Fig. S9B). There were no apical vesicles in MEKi and AKTi cells, implying perturbed protein trafficking, and nucleus shape was highly irregular (Fig. 5I,J; Fig. S9B). To summarize, our analysis of cell morphologies revealed that the two signalling pathways are required to maintain the columnar cell shape and cell junctions in the SN cells.

We performed bulk RNA-seq to identify transcriptomic targets of PI3K/AKT and MAPK/ERK pathways in the human lung tip epithelial cells. SN organoids received 4 days of chemical inhibition before lysis and RNA extraction (Fig. 6A, Table S1). Principal component analysis showed that inhibited cells are different from the SN controls, and AKT-inhibited organoids were distinct from MEK- or ERK-inhibited organoid (together referred to as MAPK inhibited, Fig. S10A). We identified >180 differentially expressed genes (DEGs) between AKT-inhibited cells and SN cells, including genes associated with the cytoskeleton and cell adhesion, such as COL1A1, COL17A1, MYO5B, ECM1 and CDH12 (Fig. S10B, log₂FC>1, P<0.05). Pathways that were downregulated after AKT inhibition included PI3K/AKT signalling, HIF-1 signalling and mineral absorption (Fig. S10C). By contrast, >230 DEGs were identified between MAPK-inhibited cells and SN control (Fig. 6B; log₂FC>1, P<0.05). We noticed the downregulation of integrin (ITGA2, ITGA6 and ITGB4) and cell junction-related (ANXA10, JAML, CTNNAL1 and GJB3) genes (Fig. 6B). Moreover, KEGG pathway analysis revealed that genes involved in MAPK/ERK signalling, focal adhesion, actin cytoskeleton regulation and adherens junctions were downregulated after MAPK inhibition (Fig. 6C). These data suggest that MAPK/ERK signalling regulates cell-matrix interactions and further confirm that the signalling maintains the columnar cell shape in the human tip epithelial cells.

It has been established that integrin signalling plays a key role in the interactions between epithelial cells and their ECM niche. We decided to focus specifically on ITGA2, because of its in vivo enrichment in human lung tip epithelia, compared with the stalk, at both transcriptional and protein level (Fig. 6D; Table S1), and its high abundance in organoids (Nikolic et al., 2017). After MEKi treatment, immunostaining indicated that the expression pattern of integrin α2 protein was altered (Fig. S11A). Moreover, cells in basal medium-treated spheres lose lateral localization of integrin α2 (Fig. S11B), further showing that RTK inputs (FGF7, FGF10 and EGF) via MAPK/ERK activation promote integrin localisation of the human lung tip epithelial cells.

Integrins can regulate epithelial cell shape by controlling the cytoskeleton (Domínguez-Giménez et al., 2007; Mateos et al., 2020). To test whether integrin activity is required for maintaining the columnar shape of the lung tip epithelial cells, we focused on integrin α2 for reasons mentioned above, and include integrin β1 as it is the main β-integrin subunit for integrin heterodimers (Takada et al., 2007). In the branching human lungs, epithelial integrin β1 is localized at the basal cell surface (Fig. S11C). We neutralized integrin α2 and β1 with blocking antibodies (Fig. 6E). The majority of the organoids in the isotype control condition showed correct apical-basal polarity and columnar cell shape, with some exceptions that showed inverted polarity due to organoid passaging by fragmentation (Fig. 6F; Fig. S11D, arrowhead). However, blocking antibody treatment increased the number of organoids with inverted polarity, where the cells were apically polarised towards the Matrigel and possessed more than one lumen (Fig. 6F; Fig. S11D). Meanwhile, organoids with correct polarity had laterally shortened cells, suggesting defects in adherens junctions (Fig. 6F; Fig. S11D). We therefore hypothesise that integrin signalling acts downstream of FGFR-MAPK signalling to regulate cell polarity and cell junctions in the tip epithelial cells.

To clarify the functions of integrin signalling in the tip epithelial cells, we turned to our previously published CRISPRi-based knockdown system (Sun et al., 2021) to reduce the expression of ITGB1, the major integrin subunit in the developing human lung (Coraux et al., 1998). SN organoids were sequentially transduced with an inducible KRAB-dCas9 vector and then a constitutive gRNA vector using lentivirus (Fig. S12), and were treated with doxycycline (DOX) and trimethoprim (TMP) to achieve gene knockdown (Fig. 6H). Experiments in four organoid lines showed a moderate, but significant and reproducible, reduction of ITGB1 expression and a similar trend for SOX9, but not for ITGA2 or SOX2 (Fig. 6I). We observed a loss of the lumen in some ITGB1-knockdown (KD) organoids (Fig. S12B), confirmed by immunostaining (Fig. 6J; Fig. S12C). Moreover, unlike non-targeting (NT) control organoids, the ITGB1-KD organoids displayed aberrant cell shape and apical-basal polarity, shown by ZO1 and integrin β1 staining (Fig. 6J; Fig. S12C). This phenotype is partially reminiscent of the integrin antibody blocked organoids (Fig. 6F; Fig. S11C). Together, these data suggest that integrin signalling indeed regulates cell polarity and cell shape of the human lung tip epithelial cells. To test whether integrin signalling is required for the tip epithelial cells to survive, we performed an organoid formation assay by seeding single control or ITGB1-KD cells and treating the cells with DOX and TMP for 10 days (Fig. S12D). Control cells could grow into organoids with a lumen, whereas ITGB1-KD cells could not (Fig. S12E).

Integrins as adhesion receptors bind to extracellular matrix proteins, such as collagens and laminins, to initiate ‘outside-in’ signalling, and MAPK/ERK is one of the downstream pathways (Howe et al., 2002). We examined ERK activity in day 5 ITGB1-KD cells. Western blotting experiments demonstrated some decline in pERK level when ITGB1 was perturbed (Fig. 6K). Together, our analyses reveal that integrins are downstream of MAPK/ERK signalling in the tip epithelial cells and, moreover, that integrin signalling can increase ERK activity. These data suggest intricate crosstalk between integrin and MAPK/ERK signalling in the tip epithelial cells. MAPK/ERK and integrin signalling pathways both regulate the maintenance of columnar shape in the lung tip epithelial cells.

Mesenchymal FGF10 is essential for lung morphogenesis in mice (Abler et al., 2009). However, consistent with the work of others (Danopoulos et al., 2019; Miller et al., 2018), we have shown that FGF10 addition is not sufficient to rescue columnar cell shape, proliferation, organoid budding or SOX9 expression in primary human lung tip spheres grown in basal medium (Fig. 2). Yet FGF10 is required for the establishment and maintenance of human lung bud tip organoids (Nikolic et al., 2017). We therefore sought to confirm whether FGF10 was required for the long-term maintenance of the SN organoids. We removed FGF10 from the SN medium and examined organoid survival and morphology during continued passaging (Fig. S13A-C). We noted patches of SOX2⁺SOX9⁻ cells emerging after five passages in the FGF10-removed condition (Fig. 7A; Fig. S13D,E), reminiscent of the patchy SOX9 expression in the organoids established without FGF10 (Nikolic et al., 2017). The SOX9⁻ cells in these organoids stopped proliferating (Fig. 7A), consistent with our recent findings (Sun et al., 2022) and we could not maintain these organoids for subsequent passages. Moreover, we observed altered E-cadherin patterning and F-actin in the SOX9⁻ cells (Fig. 7B; Fig. S13D), suggesting changes in cell junctions and polarity. Such observations coincided with the observations of disrupted cell junctions and laminin deposition of distal epithelial cells in Sox9 deleted embryonic mouse lungs (Rockich et al., 2013). These data confirm that FGF10 is required for long-term SN organoid maintenance, where it plays a role in maintenance of progenitor cell fate, shape and proliferation.

EGF, FGF7 and FGF10 are widely and ubiquitously expressed in the pseudoglandular human lungs (Danopoulos et al., 2019; He et al., 2022), meaning epithelial cells likely receive multiple, dynamic RTK signalling inputs. We next interrogated the combined function of FGF10 and EGF on the basal spheres (freshly dissected tip epithelial cells established in basal medium; Fig. 2A). Combined treatment of EGF and FGF10 robustly initiated organoid budding (Fig. 7C,D), which is distinct from the phenotype of single EGF or single FGF10 (Fig. 2B). To better understand such combined effects, we grew single SN cells in EGF medium (basal medium plus EGF and ROCK inhibitor), or FGF10 medium (basal medium plus FGF10 and ROCK inhibitor) (Fig. 7E). In both conditions, single SN cells formed spherical organoids with cuboidal cells (Fig. 7F). Subsequently, we added EGF+FGF10 medium (basal medium plus EGF and FGF10) and observed organoid budding, cuboidal-to-columnar transition of cell shape and increased cell proliferation (Fig. 7F-I).

We reasoned that adding both EGF and FGF10 to the culture might result in greater downstream signalling activation than EGF or FGF10 alone. Therefore, we harvested EGF- or FGF10-treated organoids, and those that received both EGF and FGF10, after 10 days and examined the pERK level (Fig. 7J,K), confirming greater pERK activation after combined treatment. Moreover, western blotting suggested that EGF+FGF10 induced ERK phosphorylation to a similar level as FGF7 (Fig. S13E-G). These data together confirm the combined effect of the two ligands. In summary, FGF10 contributes to the maintenance of SN organoids; we conclude that FGF7, FGF10 and EGF, the three RTK inputs in the SN medium, work together to sustain signalling activities in the tip epithelial cells to maintain progenitor identity and morphology.

During the early pseudoglandular stage (∼6 to 13 pcw) of human lung development, tip epithelial progenitor cells are columnar and apically constricted (Fig. 1; Fig. S1). The cells can grow into self-renewing tip epithelial organoids in vitro in the seven-factor SN medium (Nikolic et al., 2017). We have used the SN organoids to study the cell shape maintenance of the tip progenitor cells at the early pseudoglandular stage.

Using primary human embryonic tip epithelial cells, we showed that the RTK inputs (FGF7, FGF10 and EGF) promoted organoid budding, increased cell proliferation and activated the ERK and AKT pathways. Among the three, FGF7 displayed the greatest potential for promoting cell proliferation and maintaining columnar cell shape. By contrast, EGF was less potent, at least partially because of its inefficiency in activating AKT, when compared with FGF7 (Fig. 4; Figs S6, S7). Such a supportive and secondary role of EGF was previously observed in adult mouse lung epithelial organoids (Rabata et al., 2020). However, the effects of other EGF family ligands on the lung tip epithelial progenitor cells remain to be elucidated. Recent research on human intestinal epithelial organoids showed clear differences in organoid morphology and cell fate decisions of organoids receiving EGF or epiregulin stimulation (Childs et al., 2023). We initially did not observe significant effects of FGF10 on the primary human embryonic tip epithelial cells (Fig. 2), suggesting species differences between the mouse and human lung. However, FGF10 is required in our long-term culture, and it exerts combined effects with EGF (Fig. 7). Such observations have led us to reason that the three RTK inputs in the SN medium also show combined effects on the cells. For example, our recent findings have suggested that the three RTK inputs contributed to the expression of ETV4 and ETV5, co-regulators of SOX9 in the self-renewing tip epithelial cells (Sun et al., 2022). Moreover, we speculate that similar crosstalk occurs in vivo where the RTK ligands cooperate and contribute to development.

Understanding the mechanisms by which cells acquire and maintain their shape is a long-standing problem. Here we have shown that both ERK and AKT signalling pathways regulate cell shape maintenance and the assembly of cell junctions in the human lung tip epithelial cells (Fig. 5). The decrease in lateral length and the increase in apical length of a SN cell is a useful proxy to describe cell shape changes and predict disruption in cell junctions. It will be interesting to broaden the cell shape analysis to later human lung developmental stages in the future and further investigate the underlying molecular processes.

We identified disruptions in the expression of integrin genes as the result of chemical inhibition of MAPK/ERK signalling (Fig. 6). Several integrin genes have been implicated in the early development of the mouse lung (Chen and Krasnow, 2012; de Arcangelis et al., 1999; Kreidberg et al., 1996; Wu and Santoro, 1996), and its branching morphogenesis (Chen and Krasnow, 2012; Plosa et al., 2014). Our results provide experimental evidence that clarifies the roles of epithelial integrin signalling in the maintenance of cell shape and cell polarization (Fig. 6). Recent findings have raised a role for FGFR-regulated cell-matrix adhesion during salivary gland branching (Ray and Soriano, 2023) and in neural crest cells (Ray et al., 2020). Future investigation of the crosstalk between RTK and integrin signalling pathways would help further clarify how the niche influences the tip epithelial cells. Taken together, our results show that RTK signalling activates MAPK/ERK and PI3K/AKT signalling to regulate the shape and junctional structure of the human lung epithelial progenitor cells at the early pseudoglandular stage.

Human embryonic and foetal lungs were collected from terminations of pregnancy from the Cambridge University Hospitals NHS Foundation Trust under permission from NHS Research Ethical Committee (96/085) and the Joint MRC/Wellcome Trust Human Developmental Biology Resource [London and Newcastle, University College London (UCL) site REC reference: 18/LO/0822; Newcastle site REC reference: 18/||NE/0290; Project 200454; www.hdbr.org]. Samples used in this study had no known genetic abnormalities.

Human embryonic lung organoids were derived and maintained as previously reported (Nikolic et al., 2017). Briefly, human embryonic lung tissues were incubated in dispase (8 U/ml, ThermoFisher Scientific, 17105041) at room temperature for 2 min for dissociation. Mesenchyme was dissected away using forceps. Epithelial tips were micro-dissected and transferred into 40 μl of Matrigel (Corning, 356231) in one well of a 24-well low-attachment plate (Greiner, M9312-100EA). The plate was incubated for 15 min at 37°C to solidify the Matrigel and 600 μl self-renewal (SN) medium was added. SN medium consists of Advanced DMEM/F12 supplemented with 1× GlutaMax (ThermoFisher Scientific, 35050-061), 1 mM HEPES (ThermoFisher Scientific, 15630-060), Penicillin/Streptomycin (as Adv+++), 1× N2 (ThermoFisher Scientific, 17502–048), 1× B27 (ThermoFisher Scientific, 12587–010), N-acetylcysteine (1.25 mM, Merck, A9165), EGF (50 ng/ml, PeproTech, AF-100-15), FGF10 (100 ng/ml, PeproTech, 100-26), FGF7 (100 ng/ml, PeproTech, 100-19), Noggin (100 ng/ml, PeproTech, 120-10C), R-spondin (5% v/v, Stem Cell Institute, University of Cambridge), CHIR99021 (3 μM, Stem Cell Institute, University of Cambridge) and SB431542 (10 μM, Bio-Techne, 1614). Once formed, SN organoids were maintained in the SN medium and passaged by mechanically breaking using P1000 pipettes every 7-10 days. SN organoids can also be maintained by passaging via enzymatic digestion to single cells; this method was used whenever cell sorting was performed (see next section).

The micro-dissected epithelial tips were incubated in TrypLE Express Enzyme (ThermoFisher Scientific, 12605010) for 10 min at 37°C to dissociate into single cells. After rinsing in cold Adv+++, cells were filtered by 40 μm cell strainer and collected by centrifuge. A CD326 MicroBead kit (Miltenyl Biotec, 130-061-101) was used to enrich EpCAM⁺ cells. The cell pellet was resuspended in 300 μl magnetic-activated cell sorting (MACS) buffer (0.5% BSA, 2 mM EDTA in PBS) and purified following the manufacturer's protocol. Briefly, Fc receptors were blocked using the reagent provided to saturate non-epithelial cells, then cells were incubated with CD326 MicroBeads for 30 min at 4°C. After two MACS buffer washes, the positive cells were harvested through the LS column in the magnetic field in a 15 ml tube and transferred into 40 μl of Matrigel in 24-well low-attachment plate at 5000 cells per well. The cells were supplied with SN medium plus Y27632 (10 μM, Merck, 688000) for the first 48 h, then with SN medium or with basal medium plus 10 μM Y27632 for the first 48 h, and with basal medium thereafter. Basal medium consists of Adv+++, 1× N2, 1× B27, N-acetylcysteine (1.25 mM), Noggin (100 ng/ml, PeproTech, 120-10C), R-spondin (5% v/v, Stem Cell Institute, University of Cambridge), CHIR99021 (3 μM, Stem Cell Institute, University of Cambridge) and SB431542 (10 μM, Bio-Techne, 1614).

Human embryonic lungs were fixed on ice for 1-3 h depending their size in 4% (w/v) paraformaldehyde in 1× PBS, washed in 15, 20 and 30% (w/v) sucrose solutions in PBS for 1 h each at room temperatures before incubating in a 1:1 mix of optimal cutting temperature compound (OCT; Tissue-tek):30% sucrose overnight at 4°C. Lungs were embedded in 100% OCT and stored at −70°C before sectioning. Human embryonic lung cryosections (12 μm) were rinsed with PBS and permeabilised with 0.2% Triton-X/PBS (washing solution). Normal donkey serum (5%; Stratech, 017-000-121-JIR) in washing solution containing 0.5% (w/v) bovine serum albumin (BSA) was used for blocking at room temperature for 1 h. Primary antibodies (Table S2) in blocking solution were incubated at 4°C overnight. After three washes in washing solution, secondary antibodies (Table S3) in 0.2% Triton-X/0.5% BSA/PBS were incubated at 4°C overnight. After three washes, DAPI (100 ng/ml, Sigma, D9542) was added for 30 min at room temperature. Samples were mounted in Fluoromount Aqueous Mounting Medium (Sigma, F4680). Confocal z stacks of single planes were acquired using Leica SP8 at an optical resolution of 1024×1024 at 40×. Images were processed using ImageJ (version 2.1.0).

Organoids were recovered from Matrigel using Corning Matrigel Cell Recovery Solution (Corning, 354253) and fixed with 4% (w/v) paraformaldehyde (PFA) for 20 min on ice. After washing in PBS at least twice, organoids were transferred to CellCarrier-96 Ultra Microplate (PerkinElmer, 6055300) for staining. Permeabilization in 0.5% (v/v) Triton-X/PBS for 30 min was followed by washing in 0.5% (w/v) BSA and 0.2% Triton-X/PBS (washing solution). Normal donkey serum (5%) in washing solution was used for blocking for 1 h at 4°C. Primary antibodies (Table S2) in blocking solution were incubated at 4°C over 2 nights. After three washes, secondary antibodies (Table S3) in the washing solution were incubated at 4°C overnight. After three washes, DAPI (100 ng/ml) was added to the washing solution for 30 min at 4°C. Organoids were mounted in a fructose-glycerol clearing buffer (Dekkers et al., 2019).

Organoids of the ITGB1 inducible knockdown experiment (Fig. 6F,J; Fig. S12C) were immunostained in situ (organoids not recovered from the Matrigel). Cells were grown in the CellCarrier-96 Ultra Microplate or glass bottom microwell dishes (MatTek, P35G-1.5-20-C). For fixation, pre-warmed 4% PFA was added to the dish after removing culture medium and incubation was for 15 min at 37°C. Permeabilization was in 1% (v/v) Triton-X/PBS for 1 h and 1% (v/v) Triton-X was used in all solutions. Primary antibody incubation was over 3 nights and over 2 nights for secondary antibodies. The fructose-glycerol clearing buffer (Dekkers et al., 2019) was used for clearing at least overnight at 4°C. Confocal z stacks of single planes were acquired using Leica SP8 at an optical resolution of 1024×1024 with a 40× objective. Images were processed using ImageJ (version 2.1.0).

After complete removal of the Matrigel, organoids were lysed using RIPA buffer supplemented with 1× Halt Protease and Phosphatase Inhibitor Cocktail (ThermoFisher Scientific, 78440) on ice, with strong vortexing every 5 min six times. Samples were centrifuged for 15 min at 16,200 g, 4°C, supernatant was collected and the protein content was quantified with a BCA kit (ThermoFisher Scientific, 23225). Equal amounts of protein were denatured by mixing with 4× Laemmli buffer (Bio-Rad, 1610747) and incubated for 5 min at 95°C. Samples were separated by SDS-PAGE (Bio-Rad, 1610148) and electro-transferred onto PVDF membranes (Merck, Immobilon-P membrane). Membranes were blocked with 5% skim milk and then incubated with primary antibodies (Table S2) overnight at 4°C. After extensive washes, membranes were incubated with IRDye-conjugated secondary antibodies (Table S3) overnight at 4°C. The protein bands were visualised using the Li-Cor Odyssey system and quantified in ImageJ (version 2.1.0). Histone H3 was used as a loading control. Integrin α2 and integrin β1 were normalized to histone H3. pERK and pAKT were normalized to total ERK and total AKT, respectively.

The ERK-KTR vector was generated based on Addgene 59138 with mClover swapped to mNeonGreen. The AKT KTR vector was generated based on the ERK-mNeonGreen vector with ERK-KTR sequence removed by MluI-HF and SpeI-HF cutting, and swapped to AKT-FOXO3A-KTR. The AKT-FOXO3A-KTR sequence was sub-cloned from human cDNA according to Maryu et al. (2016).

For the ITGB1 knock-down experiment, a doxycycline-inducible CRISPRi vector was used (Sun et al., 2021) to harvest KRAB-dCas9 cells. A gRNA plasmid (Addgene 167936) was linearized with BbsI-HF restriction enzyme for 1 h at 37°C and gel purified with QIAquick Gel Extraction Kit (Qiagen, 28704). Two gRNAs targeting ITGB1 (Horlbeck et al., 2016) were individually subcloned into gRNA vector as follows: gRNA-1, 5′-GAGAGGCCCAGCGGGAGTCG-3′; gRNA-2, 5′-GGGGAGACCGCAGGTGTCAG-3′. The non-targeting control sequence was 5′-GCTGCATGGGGCGCGAATCA-3′.

For the gRNA, dCas9 and KTR vectors, HEK293T cells at 70-80% confluency in 10 cm dishes were transfected with the insert construct plus 3rd generation packaging plasmids: pMD2.G (3 μg, Addgene 12259), psPAX2 (6 μg, Addgene 12260) and pAdVAntage (3 μg, E1711, Promega) using Lipofectamine 2000 Transfection Reagent (11668019, ThermoFisher Scientific) according to the manufacturer's protocol. Medium was changed after 16 h. Lentivirus-containing supernatant was collected and pelleted (5 min at 1000 rpm) on the third day post-transfection. The supernatant was filtered through a 0.45 μm filter then concentrated.

To concentrate the lentiviral supernatant, one volume of Lenti-X Concentrator (TAKARA, 631231) was added to three volumes of supernatant and incubated at 4°C for 1 h. The solution was centrifuged for 45 min at 4°C and the lentiviral pellet re-suspended in ice-cold sterile PBS at 1/100 the original volume, distributed into 10 μl aliquots and stored at −80°C.

Lentiviral transduction of organoids was performed on a single cell solution of the organoids after TryPLE dissociation. 2.5 μl concentrated lentivirus was added to 50,000 single cells suspended in fresh SN medium supplemented with 10 μM Y27632 and incubated at 37°C overnight. Cells were collected and embedded in Matrigel at 50,000 per well the next morning and grown in SN medium supplemented with 10 μM Y27632 for 24 h. Virally transduced cells were cultured for 7-10 days to grow into organoids. To isolate a pure population of transduced cells, single cells were collected by dissociating organoids with TryPLE and sorted based on fluorescence colours using Sony SH800S cell sorter.

Organoids were collected from the Matrigel and lysed. RNA extraction was performed according to the manufacturer's protocol (Qiagen, 74004). RNA concentrations were measured by NanoDrop. cDNA synthesis was performed using MultiScribe Reverse Transcriptase (Applied Biosystem, 4308228). The mix was left at 25°C for 5 min, 50°C for 50 min then 15 min at 70°C. cDNA was diluted 1:10 and 1 μl was used for each qPCR reaction with SYBR Green assays (PowerUp SYBR Green Master Mix, Applied Biosystem, 100029284). Relative Cp values for target genes were standardized against GAPDH expression. Relative gene expression was calculated using ΔΔCp method. P-values were obtained using an unpaired two-tailed Student's t-test with unequal variance.

Primers were as follows: GAPDH-F, 5′-TGCCCTCAACGACCACTTTG-3′; GAPDH-R, 5′-GGGTCTCTCTCTTCCTCTTGTGCT-3′; ITGB1-F, 5′-TTCAAGGGCAAACGTGTGAG-3′; ITGB1-R, 5′-GGACACAGGATCAGGTTGGA-3′; ITGA2-F, 5′-AGAAAGCCGAAGTACCAACAG GAGT-3′; ITGA2-R, 5′-TGCAGGTAGGTCTGCTGGTTCA-3′; SOX2-F, 5′-TACAGCATGTCCTACTCGCAG-3′; SOX2-R, 5′-GAGGAAGAGGTAACCACAGGG-3′; SOX9-F, 5′-AGCACTGGGAACAACCCGTCT-3′; SOX9-R, 5′-TAGGATCATCTCGGCCATCTTCGC-3′.

For bulk RNA-seq, RNA quality was validated using an Agilent 2200 Tapestation with High Sensitivity RNA Screen Tape (Agilent, 5067-5579). RNA samples were sent for Eukaryotic RNA-seq library preparation and sequencing (PE150) at Novogene (Cambridge, UK). mRNA was purified from total RNA using poly-T oligo-attached magnetic beads. After fragmentation, the first-strand cDNA was synthesized using random hexamer primers, followed by the second strand cDNA synthesis. The library was checked with Qubit and real-time PCR for quantification and bioanalyzer for size distribution. Quantified libraries were pooled and sequenced on Illumina platforms, and paired-end reads were generated. Raw sequencing data were pre-analysed by Novogene. Raw data (raw reads) of fastq format were first processed through Novogene in-house perl scripts. In this step, clean data (clean reads) were obtained by removing reads containing the adapter, reads containing ploy-N and low-quality reads from raw data. The index of the reference genome was built and paired-end clean reads were aligned to the reference genome using Hisat2 (2.0.5). To count the reads numbers mapped to each gene, featureCounts (1.5.0-p3) was used and fragments per kilobase of transcript sequence per millions base pairs (FPKM) of gene were calculated based on the length of the gene and reads count mapped to this gene. Differential expression analysis of two conditions was performed using the DESeq2 R package (1.20.0). The resulting P-values were adjusted using the Benjamini and Hochberg's approach for controlling the false discovery rate. KEGG pathway analysis was performed using DAVID (Huang et al., 2009). RNA-seq data have been deposited in GEO under accession number GSE211308.

ERK-KTR and AKT-FOXO3A-KTR expressing organoids were cultured in CellCarrier-96 Ultra Microplate and imaged on a Zeiss 880 Airyscan inverted confocal microscope equipped with an incubation chamber and CO₂ supply to maintain 37°C and 5% CO₂. NucRed Live 647 (ThermoFisher Scientific, 2146834) was added 1 h before live-imaging experiments to allow cell nucleus imaging. Multiple z stacks (5 μm step) at each time point of the KTR-mNeonGreen and the nuclear marker were simultaneously captured through a 25×0.8 N.A. water objective.

ImageJ and CellProfiler (Stirling et al., 2021) were used to process the images. To determine the nuclear and cytoplasmic fluorescence intensities shown in Fig. 4 and Figs S5-S7, we referred to previous reports with custom changes (Kudo et al., 2018; Regot et al., 2014). We manually selected at least three non-consecutive z planes at each time point (same relative z position) in ImageJ. In CellProfiler, the nuclear region of each cell was segmented based on the NucRed 647 channel. The nuclear segmentation was used as a mask and a ring of 5 pixels width around the nucleus from the nuclear segmentation was used to define the cytoplasmic region. Fluorescence intensity for mNeonGreen of the nuclear region and cytoplasmic region of each cell was measured. The cytoplasmic to nuclear ratio (cytoplasm/nucleus ratio) of the KTR-mNeonGreen of each measured cell was calculated for each time point by dividing the mean cytoplasmic intensity by the mean nuclear intensity of a cell. This ratio is normalized to time 0 (as 100%) and is shown as a percentage in Fig. 4 and Figs S5-S7.

The organoid samples were fixed in 2% formaldehyde/2% glutaraldehyde in 0.05 M sodium cacodylate buffer (NaCAC) (pH 7.4) containing 2 mM calcium chloride (Merck, C27902) overnight at 4°C. After washing in 0.05 M NaCAC at pH 7.4, the samples were osmicated for 3 days at 4°C. After washing in deionised water (DIW), the samples were treated twice with 0.1% (w/v) thiocarbohydrazide (Merck, 223220) in DIW for 20 min each time and then left at1 h at room temperature in the dark, followed by block-staining with uranyl acetate [2% uranyl acetate in 0.05 M maleate buffer (pH 5.5)] for 3 days at 4°C. The samples were then dehydrated in a graded series of ethanol (50%/70%/95%/100%/100% dry), 100% dry acetone and 100% dry acetonitrile, three times in each for at least 5 min. Next, the samples were infiltrated with a 50:50 mixture of 100% dry acetonitrile/Quetol resin (TAAB, Q005) without BDMA (TAAB, B008) overnight, followed by 3 days in 100% Quetol without BDMA. The sample was infiltrated for 5 days in 100% Quetol resin with BDMA, exchanging the resin each day. The Quetol resin mixture is: 12 g Quetol 651, 15.7 g NSA (TAAB, N020), 5.7 g MNA (TAAB, M012) and 0.5 g BDMA. Samples were placed in embedding moulds and cured at 60°C for 3 days. Thin sections were cut using an Ultracut E ultramicrotome (Leica) and mounted on melinex plastic coverslips. The coverslips were mounted on aluminium SEM stubs using conductive carbon tabs and the edges of the slides were painted with conductive silver paint. The samples were then sputter coated with 30 nm carbon using a Quorum Q150 T E carbon coater and imaged in a Verios 460 scanning electron microscope (FEI, ThermoFisher Scientific) at 4 keV accelerating voltage and 0.2 nA probe current in backscatter mode using the concentric backscatter detector in immersion mode at a working distance of 3.5-4 mm; 1536×1024 pixel resolution, 3 μs dwell time, 4 line integrations. Stitched maps were acquired using FEI MAPS software using the default stitching profile and 10% image overlap.

Whole-mount immunostained organoids were embedded in an embedding solution as previously described (Dekkers et al., 2019) with light-sheet glass capillaries. Briefly, 0.4 g of low-melting point agarose (Bio-Rad, 1613111) was fully dissolved in 10 ml of water and 10 ml of fructose-glycerol clearing solution added and mixed well to obtain a clear solution. A Zeiss Z1 light sheet microscope was used for imaging. Samples were imaged by placing the sample-containing capillary in the light sheet chamber filled with fructose-glycerol clearing solution and pushing down the embedded sample from the capillary to be exposed for imaging. A 20× detection objective (clearing immersion N.A.=1.0) was used to acquire whole z stack images with 1024×1024 frame size. Arivis Vision4D was used to process the z planes and generate 3D-rendering images that exported as .tiff files.

Integrin α2 and β1 blocking experiment (Fig. 6E-G) was performed by culturing SN organoids in CellCarrier-96 Ultra Microplate. 50 μl of Matrigel containing organoids was seeded in each well. Organoids recovered from cell seeding were treated with 10 ng/ml recombinant rabbit integrin α2 (Abcam, ab181548) and 10 ng/ml mouse IgG1 integrin β1 (Abcam, ab30394) or 10 ng/ml mouse monoclonal 2C11 IgG1 isotype control antibody (Abcam, ab1927) for 6 days with culture medium and antibodies being replenished every 2 days. Organoids were fixed in situ and stained for F-actin (488 ReadyProbes Reagent, ThermoFisher Scientific, R37110) according to the manufacturer's instructions.

Click-iT EdU Imaging Kit (ThermoFisher Scientific, C10338) was used to assay organoids shown in Fig. 2F according to manufacturer's instructions. Briefly, 10 μg/ml EdU was added to organoids in different conditions for 6 h. Afterwards, EdU-containing media were washed off and organoids were recovered from Matrigel and fixed with 4% PFA for Click-iT assay.

SN organoids were sequentially transfected with inducible KRAB-dCas9 vector and then gRNA vector or non-targeting control vector (Fig. S12). Purified cells were cultured into organoids and treated with 2 μg/ml doxycycline (DOX, Merck, D9891) and 10 μM trimethoprim (TMP, Merck, 92131) for 5 days. Samples were then collected for qRT-PCR and western blotting, or fixed using pre-warmed 4% PFA for 15 min at 37°C in situ for immunostaining.

Cell shape quantitation was performed in ImageJ (version 2.1.0). Images of cryosections and organoids were manually scored by drawing lines and measuring the lengths, based on ZO1, fibronectin and E-cadherin staining or F-actin staining. Cells not integrated into the epithelial sheet were not included, because cells round up during division.

Nucleus circularity (DAPI staining based) was measured and calculated in ImageJ (version 2.1.0) by manually tracing the outline of individual cell nuclei following their DAPI signal or circling the nucleus of SEM images. Quantitation of EdU assay was performed in ImageJ (version 2.1.0) with 2D images acquired by Leica SP8 microscope. The Fiji plug-in OAK (Organoid Analysis Kit available at https://github.com/gurdon-institute/OAK/releases/tag/1.7.1) was used to score the EdU-positive cell numbers and total cell numbers. Similarly, KI67 quantitation was achieved with the same procedure. Arivis Vision4D was used to quantify EdU-positive cells with images acquired by Zeiss Z1 light sheet microscope by scoring cell numbers from the EdU channel and DAPI channel. Quantitation of spherical or budding organoids (Fig. 2C) was performed by manually counting organoid numbers of each phenotype.

Quantitation of the area of the organoids (Fig. 2D) was performed by using a custom script for Fiji (https://github.com/gurdon-institute/OrganoidArea/blob/main/OrganoidArea.py) to segment and measure organoids in images taken on Zeiss Axiophot compound microscope. Quantitation of SOX9 expression of organoids (Fig. 2H) was performed by sampling 25 SOX9-stained organoids from three biological replicates and manually scoring the numbers of SOX9⁺, partially SOX9⁺ and SOX9⁻ organoids. Quantitation of pERK/pAKT intensity per cell (Fig. 3A) was performed in CellProfiler. Nuclear region of each cell was segmented based on DAPI staining. The nuclear segmentation was used as a mask and a ring of 5 pixels width around the nucleus from the nuclear segmentation was used to define the cytoplasmic region. Fluorescence intensity for pERK or pAKT of the nuclear region and cytoplasmic region of each cell was measured and added up as per cell intensity. Quantitation for the integrin blocking experiments was performed by sampling F-actin stained images of isotype control and integrin blocking condition, and the numbers of organoids showing correct apical-basal polarity, inverted polarity or shortened cell height (but with correct polarity) were counted.

Data are mean±s.d. or mean±s.e.m. as stated in the figure legends. Statistical significance was evaluated by one-way ANOVA, Mann–Whitney U-test or unpaired Student's t-test; n.s., not significant, *P<0.05.

We acknowledge the Gurdon Institute Imaging Facility and Dr Karin Mueller of Cambridge Advanced Imaging Centre for microscopy support. Core funding to the Gurdon Institute comes from the Wellcome Trust (203144/Z/16/Z) and Cancer Research UK (C6946/A24843).

The peer review history is available online at https://journals.biologists.com/dev/lookup/doi/10.1242/dev.201284.reviewer-comments.pdf.