Unravelling differential Hes1 dynamics during axis elongation of mouse embryos through single-cell tracking

During embryonic development, tightly regulated cell fate decisions rely on intercellular communication. Recent insights indicate that cells encode information not only through the presence but also the dynamics of signalling pathways (Sonnen and Janda, 2021). Oscillations in signalling pathways and target genes are widespread across tissues and developmental stages. Prominent among these are members of the hairy and enhancer of split (Hes) family of transcriptional repressors. Hes proteins, which are basic helix-loop-helix (bHLH) transcription factors, modulate gene expression by interacting with DNA as dimers (Sasai et al., 1992). Several of the seven Hes family members in mammalian cells are dynamically expressed in embryonic development (Ohtsuka et al., 1999; Palmeirim et al., 1997; Sasai et al., 1992; Wu et al., 2003; Ohtsuka et al., 2001; Kobayashi and Kageyama, 2010).

During vertebrate development, neuromesodermal progenitors (NMPs) in the tailbud (TB) give rise to both the mesodermal and neural lineages, namely the presomitic mesoderm (PSM) and the neural tube (NT), respectively (Cambray and Wilson, 2002; 2007). Hes dynamics play a crucial role in the differentiation of both PSM and NT progenitors (Shimojo et al., 2016; 2008; Hirata et al., 2002; Ohtsuka et al., 1999). The periodic segmentation of the PSM into somites, known as somitogenesis, is driven by oscillatory signalling pathways, including FGF (fibroblast growth factor), Wnt (wingless and Int-1) and Notch signalling (Bosman and Sonnen, 2022; Dequeant et al., 2006; Matsuda et al., 2020; Palmeirim et al., 1997; Niwa et al., 2011; Aulehla et al., 2003). Whereas the exact composition of the segmentation clock differs by species, Notch signalling oscillations are found in all studied vertebrates, and several Notch target genes display oscillatory expression in PSM cells (Riedel-Kruse et al., 2007; Palmeirim et al., 1997; Soza-Ried et al., 2014; Oginuma et al., 2010; Niwa et al., 2011; Venzin and Oates, 2020), including several members of the Hes gene family (Holley et al., 2002; Palmeirim et al., 1997; Bessho et al., 2001; Hirata et al., 2002).

NMPs transitioning from the TB to the NT pass through the preneural tube (pre-NT), surrounded by oscillating PSM (Delfino-Machin et al., 2005). FGF signalling prevents premature differentiation in the pre-NT, while Notch signalling supports cell proliferation (Akai et al., 2005; Semprich et al., 2022). In the anterior NT, progenitor cells continue to proliferate and differentiate into future spinal cord neurons. Whether NT progenitor cells proliferate or differentiate has been linked to oscillations in Hes1 and Hes5 that are reminiscent of their function in somitogenesis (Shimojo et al., 2008, 2016). Hes genes oscillate in progenitor cells to maintain a proliferative state (Shimojo et al., 2008; Manning et al., 2019; Biga et al., 2021; Imayoshi and Kageyama, 2014; Shimojo et al., 2016), while sustained expression has been linked to quiescence (Marinopoulou et al., 2021; Sueda et al., 2019). Despite research into Hes dynamics in the anterior NT, investigations into Hes dynamics in NMPs and pre-NT are lacking.

In this study, we delineate the dynamics of Hes gene expression in mouse development and elucidate the signalling pathways orchestrating its expression during the differentiation of NMPs along neural and mesodermal trajectories. Leveraging a newly developed homozygously viable Hes1 reporter mouse line, generated through the endogenous tagging of the gene locus with a rapidly maturing yellow fluorescent protein, we thoroughly quantify Hes1 dynamics in the PSM, TB and pre-NT regions of the embryonic tail. Through single-cell tracking, we discern distinct Hes1 dynamics between the PSM and pre-NT. Further investigation into Hes1 dynamics reveals disparate responses of PSM and pre-NT cells to signalling pathway perturbations. Thus, our new endogenous reporter mouse line, combined with recently available single-cell tracking software, allows us to uncover previously unquantified dynamics in single cells of the mouse embryonic tail.

To investigate Hes dynamics in the developing mouse embryo, we first clarified which Hes genes were expressed in the PSM, TB and, in particular, the pre-NT. The last is defined as the tube-like Pax6-negative region emanating from the TB, which lies next to the PSM (Fig. 1A,B) (Semprich et al., 2022; Akai et al., 2005). To this end, we performed in situ hybridization and hybridization chain reactions (Choi et al., 2018) (Fig. S1). Expression of Hes7 and Lfng (lunatic fringe), key components of the segmentation clock (Morales et al., 2002; Lázaro et al., 2023; Harima et al., 2013; Bessho et al., 2003, 2001), was restricted to the PSM and TB (for Hes7) or PSM (for Lfng) (Fig. S1A,B,D). In contrast, both Hes1 and Hes5 were present in the TB, the PSM and the pre-NT, with highest expression levels in the pre-NT and TB (Figs 1C and S1C,D) (Hatakeyama et al., 2004) (also see MAMEP database, http://mamep.molgen.mpg.de). Thus, Hes1 and Hes5 are not limited to PSM, but are also expressed in the TB and pre-NT of the embryonic tail.

For further analysis into the dynamics of Hes genes, we then focused on Hes1, which shows a more localized expression to the neural lineage and PSM of the developing embryo than the ubiquitously expressed Hes5 (MAMEP database). To be able to quantify expression levels over time, we generated a new Hes1 mouse line (Fig. 1D-J, Fig. S2). Building on previous approaches (Yoshioka-Kobayashi et al., 2020; Niwa et al., 2007; Masamizu et al., 2006; Marinopoulou et al., 2021), we endogenously tagged Hes1 with Achilles, a fast-maturing fluorescent protein (Yoshioka-Kobayashi et al., 2020) (Fig. 1D). We placed Achilles at the C-terminal side of Hes1 opposite the DNA-binding bHLH domain of Hes1 (Sasai et al., 1992). We separated Achilles and Hes1 by a flexible linker, reasoning that this should allow interaction of Hes1 via its C-terminal WRPW motif with the Groucho/TLE family of transcriptional repressors (Jennings et al., 2006; Fisher et al., 1996). Structure of the fusion protein was assessed for accessibility of the WRPW motif and bHLH domain using Alphafold (Jumper et al., 2021; Varadi et al. 2024) (Fig. S2J,K).

While Hes1 knockout mice die directly after birth (Suzuki et al., 2005; Jensen et al., 2000), homozygous Hes1-Achilles mice were born at Mendelian ratio (offspring of eight heterozygous breedings: 17 wild type, 34 heterozygous and 16 homozygous), and were viable and fertile. Moreover, Hes1-knockout mice show premature differentiation of Paneth cells in the intestine (Suzuki et al., 2005). However, we did not detect differences in intestinal morphology or Paneth cell number in the small intestine when comparing wild-type and Hes1-Achilles homozygous mice (Fig. S2L,M). We next confirmed that Hes1 dynamics and, in particular, protein turnover were not altered by tagging. To this end, we compared the half-life of tagged versus untagged Hes1 protein in embryonic stem cells (ESCs) heterozygous for Hes1-Achilles (Fig. 1E) upon translation inhibition and found these to be comparable (Figs 1F,G and S2N,O). Finally, we investigated the expression pattern of Hes1-Achilles. When comparing mRNA staining of Hes1 in wild-type embryos and Achilles in Hes1-Achilles embryos, the expression patterns corresponded (Figs 1H,I and S2P,Q). At the protein level, Hes1-Achilles expression had the expected pattern, with high expression levels in the pre-NT and stripes in the forming somites (Fig. 1J). Thus, we successfully generated an endogenous Hes1-Achilles reporter mouse line with maintained Hes1 functionality and dynamics.

We next addressed whether Hes1 expression was solely dependent on Notch signalling or also on other pathways, as suggested previously for Hes7 (Niwa et al., 2007). To test this, we had to culture E10.5 embryonic tails ex vivo. However, standard embryo culture protocols (Lauschke et al., 2013) needed to be optimized for survival of the NT by testing different media compositions (Fig. 2A; see Materials and Methods). In standard embryo culture medium, cleaved caspase 3 expression levels were significantly higher than in uncultured embryos at embryonic day E11.5, especially in the pre-NT and NT (Fig. 2A,B). The tube was often filled with dead cells or disintegrated entirely. In contrast, in embryonic tails cultured in neurobasal medium (see further details in the Materials and Methods), the number of apoptotic cells was not increased compared with uncultured tails. Furthermore, the NT maintained its morphology and proliferative state (Fig. 2A,C). This suggests that the pre-NT is viable in the neurobasal medium. Even though somites were less organized in neurobasal medium than in control samples (Fig. 2A), we did not observe differences in the well-studied dynamics of the segmentation clock using the segmentation clock reporter line LuVeLu (Aulehla et al., 2008). Oscillation dynamics as well as the regression of the oscillating field, a sign of differentiation, were similar between embryonic tails cultured in the different media (Fig. S3A-D). Thus, the dynamics of the segmentation clock were maintained. The optimized culture conditions of embryonic tails now allows the investigation of signalling dynamics in PSM, TB and pre-NT ex vivo.

We initially tested how pathway perturbation affected Hes1-Achilles expression levels. Notch inhibition using the γ-secretase inhibitor DAPT led to a drop in Hes1-Achilles expression levels within 3 h (Fig. 2D,E), which was comparable with the effect of Notch inhibition on the reporter Achilles-Hes7 (Yoshioka-Kobayashi et al., 2020) (Fig. S3E,F). Besides Notch signalling, we also focused on Wnt and FGF signalling, as gradients of these morphogens, including the respective downstream targets Axin2 and Dusp4, are present in the embryonic tail and pre-NT (Fig. S3G,H). We therefore perturbed embryonic tails with small molecules inhibiting these pathways, using the porcupine inhibitor IWP2 to inhibit Wnt signalling (Fig. S3I-K) and SU5402, a small molecule inhibitor against the FGF receptor, to inhibit FGF signalling (Fig. S3L-N). As expected, both inhibitors lead to a rapid regression of the oscillating field in the PSM (Fig. S3K-N). Inhibiting either did not lead to an immediate (IWP2) or consistent (SU5402) drop in signal, suggesting that these pathways do not have a direct effect on Hes1 expression levels. Thus, Hes1 expression in the posterior embryonic tail is driven by Notch signalling.

Subsequently, we carefully quantified Hes1-Achilles dynamics in the different regions of the embryonic tail (TB, PSM and pre-NT) (Figs 3A-G and S4, Movie 1). We quantified dynamics in the TB region (Fig. 3D), which includes the NMPs but presumably also some posterior PSM cells. Expression levels differed between all regions, in particular the pre-NT showed higher levels compared with the PSM region (Fig. 3H). In both the PSM and TB region, we detected oscillations over time (Fig. 3C,E). In contrast, Hes1-Achilles expression in the pre-NT region was dynamic, albeit noisier and showed less apparent oscillations over time (Fig. 3G). When analysing the period using wavelet transform (Schmal et al., 2022), there was a slight increase in the TB region compared with the PSM region (Fig. 3I). A trend towards the elevated period was also observed for the pre-NT region. Furthermore, the amplitude of the PSM region was significantly higher than in the pre-NT and TB regions (Fig. 3J). Throughout the study, we have quantified the amplitude after normalization to the mean intensity of a track, i.e. the amplitude should be considered relative to the absolute intensity. Finally, whereas kymographs revealed travelling waves of Hes1-Achilles expression levels in the PSM region (see asterisk in kymograph), no clear population-wide wave patterns were detected in the pre-NT region (Figs 3K and S3O). Overall, the data suggest that Hes1 is dynamic in the different regions of the mouse embryonic tail. However, Hes1 expression in the pre-NT seem to be noisier and less oscillatory compared with the PSM.

There are two possible reasons for not finding clear population-wide Hes1 oscillations or travelling waves in the pre-NT region: (1) there are no oscillations; (2) single cells oscillate but are not synchronized with each other. To test the latter, we performed a 2D spread-out experiment (Lauschke et al., 2013) combined with mosaic labelling of nuclei that allowed us to track cells and quantify Hes1-Achilles expression in single cells (Figs 4A-E, S5 and S6; Movie 2). When posterior embryonic tail-tips are cultured on fibronectin-coated dishes, tissue spreads and cells grow in a quasi-monolayer (Lauschke et al., 2013), which facilitates single-cell tracking. Staining for neural and mesodermal markers confirmed the presence and survival of PSM and neural cell types, when culturing in neurobasal medium (Fig. S5A,B). PSM and pre-NT regions displayed similar Hes1 dynamics when analysed at population level (Fig. S5C,D), as was observed for the respective region in growing embryonic tails (Figs 1 and 3). Hes1-Achilles expression was higher in pre-NT than PSM cells, the latter of which slightly increased in intensity upon differentiation to somites (Fig. 4F). Interestingly, oscillatory Hes1-Achilles dynamics were found in both PSM and pre-NT cells (Fig. 4B-E). Wavelet analysis indicated similar wavelet power distributions for PSM and pre-NT cells (Fig. S6B). Likewise, the periods and amplitudes were similar, even though the amplitudes in pre-NT cells were slightly reduced (Fig. 4G,H). The discrepancy between amplitude at population (Fig. 3J) and single-cell level (Fig. 4H) supports the hypothesis that Hes1 dynamics are less synchronized between neighbouring pre-NT cells.

Using Fourier transform to analyse the dynamics, similar periods were obtained (Figs 4I,J and Fig. S5E,F). Comparison of Fourier spectra of individual cells indicated that Hes1-Achilles dynamics were noisier in pre-NT than in PSM cells, as PSM cells mostly showed one main peak at 2.5 h and pre-NT cells showed multiple peaks around 2-4 h (Figs 4I,J and S6C). As a measure of variability, we determined the full width at half maximum (FWHM) for the period quantification (Fig. S5G), which was 0.5 for the PSM and 1.6 for the pre-NT. The coefficient of variation for periods was 0.1 and 0.2 for PSM and pre-NT cells, respectively. Finally, to quantify synchronization between cells of a tissue, we calculated the Kuramoto order parameter (Sanchez et al., 2022) between oscillations of single cells per embryonic tissue, which was approx. 0.8 for PSM and 0.45 for pre-NT cells (Fig. 4K). Thus, like PSM cells, single cells in the pre-NT show Hes1 oscillations, albeit with higher variability and lower synchronization.

As pre-NT and PSM both arise from the NMPs, we next sought to quantify Hes1 dynamics in NMPs. The 2D spread-out experiments (Fig. 4) did not allow us to unequivocally identify NMPs. Therefore, we analysed Hes1-Achilles dynamics in NMPs by in vitro differentiation from ESCs derived from Hes1-Achilles mice, applying a previously published protocol (Fig. 5A) (Gouti et al., 2014). To identify embryonic and neural fate, while quantifying Hes1-Achilles expression levels over time, we added a Sox2-H2B-iRFP reporter (‘Sox2-iRFP’) into Hes1-Achilles cells by endogenously tagging Sox2. Based on immunostainings and previous literature, we defined the NMP window between 56 and 72 h (Figs 5A and Fig. S7A) (Metzis et al., 2018; Gouti et al., 2014). When analysing Hes1-Achilles dynamics at the population level, oscillations resembled the appearance and amplitude of the TB region in ex vivo embryonic tails, albeit with a reduced period (Figs 5B-E and 3D,E). We next tracked single cells under NMP conditions and quantified Hes1-Achilles dynamics (Fig. 5F-I). Hes1-Achilles expression was dynamic in NMPs with a period of around 2.5 h (Fig. 5H). The oscillations were variable with low amplitude (Fig. 5G,I). When perturbing Notch signalling using the inhibitor DAPT, we detected a dose-dependent decrease of Hes1-Achilles expression levels, but no apparent effect on period or amplitude of Hes1-Achilles dynamics (Fig. S7B-E). When we differentiated NMPs further towards the mesodermal or neural lineage (Fig. S7F-P), cells displayed Notch-dependant Hes1 dynamics, in particular period and amplitude, at population level corresponding to those found ex vivo (Fig. 3I,J). Thus, cells differentiated towards NMPs show variable oscillations in Hes1 expression levels that are Notch dependent.

Previous studies have highlighted the function of Hes dynamics in the anterior NT (Sueda et al., 2019; Bansod et al., 2017; Hirata et al., 2002; Ohtsuka et al., 2001; Biga et al., 2021; Manning et al., 2019). Therefore, we asked how Hes1 dynamics changed upon induced differentiation. As FGF inhibition induces differentiation of pre-NT to NT (Semprich et al., 2022; Delfino-Machin et al., 2005; Patel et al., 2013) and PSM to somite (Dubrulle and Pourquie, 2004; Dubrulle et al., 2001), we inhibited FGF activity in 2D spread-out cultures using SU5402 (Figs 6, S8 and S9). As expected, FGF inhibition led to a rapid regression of the oscillating field in the PSM, which indicated induced differentiation and confirmed effectiveness of the inhibitor (Fig. S8A). In agreement with changes of Hes1 expression levels from ‘posterior’ to ‘anterior’ PSM (Fig. 4F), Hes1-Achilles expression levels at the population level were increased in the PSM region, while the amplitude and period remained unchanged (Fig. S8B-D). In contrast, Hes1-Achilles expression levels were slightly decreased in the pre-NT region upon FGF inhibition with the period and amplitude also being unaffected (Fig. S8E-G).

In single PSM cells, FGF inhibition had a similar effect on Hes1-Achilles dynamics at the population level, with increased absolute expression levels, while the amplitude and period were maintained (Fig. S8H-J). In contrast, FGF signalling inhibition had a different effect on Hes1-Achilles dynamics in pre-NT cells: absolute expression levels and the oscillation period remained similar (Figs 6A-D and S8K,L), whereas the amplitude increased (Figs 6E and Fig. S8M).

It has been shown recently that expression of the Notch ligand Delta-like 1 (Dll1) begins to oscillate in the NT, adjacent to the last-formed somites (Shimojo et al., 2016). To test whether differentiation induced by FGF inhibition might have led to changes in Dll1 expression in 2D cultures, we quantified Dll1 by immunostaining. However, we did not detect changes in Dll1 expression levels in the Sox2-positive cell population upon FGF inhibition (Figs 6F,G and S8N). Thus, this suggests that high FGF signalling levels in pre-NT cells have a dampening effect on Hes1 oscillations and that differentiation leads to an increase in oscillation amplitude, which is not dependent on overall Dll1 protein expression levels.

We next addressed whether a common mechanism drives Hes1 oscillations in cells of the different tissue types. In general, a delayed negative-feedback loop has been suggested to drive oscillations via auto-repression of Hes genes (Goodwin, 1965; Monk, 2003; Takashima et al., 2011; Hirata et al., 2002; Freeman, 2000;Harima et al., 2013; Bessho et al., 2001). In addition to this Hes-driven feedback, the Notch ligand Dll1 is dynamically expressed in the PSM (Shimojo et al., 2016; Takagi et al., 2020), which has been proposed to drive a delayed coupling mechanism between neighbouring cells, resulting in kinematic waves (Takagi et al., 2020; Murray et al., 2013; Morelli et al., 2009; Yoshioka-Kobayashi et al., 2020). In neural tissue, however, Dll1 oscillations were not observed in the pre-NT adjacent to the PSM, but only in the NT adjacent to the last formed somites, where single cells started to oscillate in a salt-and-pepper-like pattern (Shimojo et al., 2016).

To clarify the role of Notch signalling in driving Hes1 oscillations in the embryonic tail, we first analysed the expression of Notch signalling components in the different regions of the embryonic tail, especially the pre-NT. We found Notch ligands Dll1, Dll3 and Dll4, and the Notch receptors to be expressed in the PSM, TB and pre-NT by hybridization chain reaction (Choi et al., 2018) (Fig. S10A,B). We next performed immunostaining to visualize ligand protein expression in the embryonic tail. In agreement with previous publications (Shimojo et al., 2016), Dll1 expression levels were elevated in the PSM compared with the pre-NT, and staining patterns varied between embryonic tails, indicating dynamic expression (Fig. 7A). However, Dll1 – just like Dll4 – was not only detected in the PSM but also in the pre-NT and TB, albeit at lower levels and without indications of dynamics (Fig. 7A). Together with previous findings on Dll1 oscillations in the embryonic tail (Shimojo et al., 2016), this suggests that Dll1 and Dll4 are present at low expression levels in pre-NT cells, whereas Dll1 expression is higher and oscillates in PSM cells, as shown before (Fig. 7B).

To compare the influence of Notch signalling on Hes1 protein dynamics in PSM and pre-NT cells, we inhibited Notch signalling with different doses of DAPT and quantified Hes1-Achilles dynamics within PSM and pre-NT at population level and in single cells (Figs 7, S10 and S11). Analysing dynamics in PSM at population level, we found that Hes1-Achilles expression levels decreased to a similar extent for all DAPT concentrations used (Fig. S10C). The oscillation period was slightly increased at 1 µM DAPT, whereas the amplitude was strongly reduced independently of the DAPT concentration used (Fig. S10D,E). This indicates that, at population level, Notch inhibition leads to a loss of Hes1-Achilles oscillations in the PSM, even at low concentrations, which could be consistent with a loss in synchrony between neighbouring cells (Riedel-Kruse et al., 2007). In contrast, in pre-NT (Fig. S10F-H), Notch inhibition led to dose-dependent decrease in Hes1-Achilles expression levels (Fig. S10F) and amplitude (Fig. S10H), whereas no significant change in period was observed (Fig. S10G).

Subsequently, we performed single-cell tracking and quantified Hes1-Achilles expression levels in pre-NT in the different conditions. Also at single-cell level, Notch inhibition led to a dose-dependent decrease of Hes1-Achilles expression levels in pre-NT cells (Figs 7C-E and S10I), whereas the amplitude and period remained unchanged (Figs 7F,G and S10J K). Conversely, in PSM cells Notch inhibition led to a decrease in both Hes1-Achilles expression levels and amplitude (Figs 7C-G and S10L-N). Likewise, the Hes1-Achilles oscillation period was decreased with high DAPT concentration (Fig. 7F). Thus, Notch inhibition has differential effects on Hes1 oscillations in the PSM and pre-NT, which implies that the mechanism driving Hes1 dynamics differs between the two cell types.

Although the essential function of Notch signalling in somitogenesis has been studied extensively (Ferjentsik et al., 2009; Ozbudak and Lewis, 2008; Horikawa et al., 2006; Evrard et al., 1998), its role in the pre-NT especially with regards to signalling dynamics is less clear. Notch signalling has been implicated in proliferation control, as overexpression of a dominant-negative form of the Notch ligand Delta1 prevents proliferation (Akai et al., 2005). In other regions of the nervous system, Hes1 oscillations also regulate proliferation with high, sustained Hes1 expression levels promoting quiescence (Marinopoulou et al., 2021; Sueda et al., 2019; Maeda et al., 2023). We therefore addressed how Notch signalling inhibition and a corresponding modulation of Hes1-Achilles dynamics affected pre-NT proliferation. To this end, we inhibited Notch signalling in 2D spread-out cultures expressing the Fucci cell cycle reporter (Sakaue-Sawano et al., 2008). To combine this with mosaic labelling for single-cell tracking, we used a mouse line expressing the inducible nuclear marker iRFPnucmem in the Rosa locus. Comparing control and Notch inhibition, we did not detect differences in the timing of cell cycle, G1 phase, S/G2/M phase or mitosis (Fig. 7H-K). When quantifying the percentage of phospho-histone H3-positive mitotic cells (mitotic index) in the Sox2-positive region, we observed a slight increase in the mitotic index, even at low inhibitor concentrations after 21 h (Fig. S10P,Q). This suggests that lowering Notch signalling per se, which results in decreased Hes1 expression levels and oscillation amplitude but unchanged period, does not affect proliferation in the pre-NT and could indicate that it is instead the dynamics that promote proliferation.

Here, we have addressed how Hes dynamics change in the context of the developing embryonic tail, in particular focussing on the tissues derived from NMPs (Fig. 7L). To enable this, we have generated a new endogenous Hes1 reporter mouse line. It contains the fast-maturing fluorescent protein Achilles, which is placed C-terminally to the Hes1 gene. Importantly, the line is homozygously viable and fertile, indicating functionality of the fusion protein, and allows the investigation of Hes1 dynamics at single-cell resolution in various tissue types (Weterings et al., 2024). Using this reporter, we compared Hes1 dynamics in the PSM and pre-NT at population and single-cell level.

To make this feasible, we had to optimize the embryonic tail culture protocol to ensure survival of NT cells during ex vivo culture. In contrast to the minimal DMEM-F-12 medium, which is optimal for PSM growth and differentiation (Sonnen et al., 2018), we made use of neurobasal medium that is optimized for culture of neural tissues. The latter is not only supplemented with additional nutrients and vitamins, but also contains proteins such as insulin. Which of these components are required for NT and pre-NT survival must be addressed in future studies. Although this medium allows NT survival, we do observe morphological defects in somite formation, which are presumably due to the presence of retinoic acid in the medium. Despite these morphological defects, the dynamics of the segmentation clock are maintained. This medium therefore allowed us to culture embryonic tails ex vivo for the simultaneous quantification of signalling dynamics in both tissues of an embryonic tail.

In pre-NT cells, Hes1 oscillated at high expression levels with low amplitude, whereas in PSM cells, it oscillated at low expression levels with high amplitude (Fig. 7L). Notch inhibition led to a decrease in Hes1 expression levels and a proportional decrease in the amplitude in pre-NT cells. In contrast, in PSM cells a decrease in amplitude was induced, over-proportional compared with the decrease in Hes1 intensity, both when analysing Hes dynamics at population and single-cell level. This implies that different mechanisms drive Hes1 dynamics in the two tissue types. Although the effect of Notch inhibition in pre-NT cells is consistent with a delayed negative-feedback loop downstream of constant Notch signalling (Goodwin, 1965; Monk, 2003), oscillations in PSM cells appear to be further amplified in a Notch-dependent manner.

In fact, the Notch ligand Dll1 has been shown to oscillate in various cell types, including the PSM, which results in periodic activation of proneural gene expression (Seymour et al., 2020; Shimojo et al., 2016; Zhang et al., 2021). Previously, Dll1 oscillations were not detected in the pre-NT, but only in the NT adjacent to the somites (Shimojo et al., 2016). We have shown here that even in the pre-NT, both Dll1 and Dll4 are present at low expression levels. Interestingly, in a study using cultured cancer cell lines, it was shown that Dll1 expression induces dynamic expression, whereas Dll4 leads to more sustained expression of downstream target genes (Nandagopal et al., 2018). How the presence of each ligand and the dynamics of Dll1 exactly influence Hes1 dynamics in PSM and pre-NT cells has to be addressed in future studies.

In the PSM, oscillatory Dll1 expression has been suggested to lead to kinematic waves travelling through the PSM (Herrgen et al., 2010; Morelli et al., 2009; Shimojo et al., 2016). Loss of Notch-dependant intercellular coupling could therefore explain the loss of population-wide dynamics, which we observed already at low Notch inhibitor concentrations. However, the oscillation amplitude was also reduced in single-cell oscillations upon Notch inhibition, arguing against there being a single effect produced by synchronicity between oscillators. The fact that high Notch inhibitor levels also led to a decrease in the Hes1 oscillation amplitude in the pre-NT at population level suggests that Notch signalling might induce some form of synchronization between pre-NT cells, as suggested recently for Hes5 in the NT (Hawley et al., 2022). Although our mosaic labelling approach for single-cell tracking allows us to track cells and their offspring, it does not allow the quantification of Hes1 dynamics in all neighbours of a given cell. Therefore, the intercellular synchronization between single-cell oscillators has to be addressed in future investigations.

Hes1 dynamics have been found to regulate proliferation and differentiation in the anterior developing NT (Maeda et al., 2023; Kobayashi et al., 2009; Hirata et al., 2002). In undifferentiated cells, Hes1 oscillates in an alternating fashion with proneural genes, while Hes1 is switched off in differentiating cells. Moreover, high absolute expression levels regulate the switch between quiescence and proliferation in adult stem cells, with high expression levels maintaining a quiescent state (Maeda et al., 2023; Marinopoulou et al., 2021; Sueda et al., 2019). It is therefore interesting to note that we find Hes1 expression levels highest in the pre-NT and that expression decreases towards the NT. Moreover, proliferation in the pre-NT requires Notch signalling in chick embryos (Akai et al., 2005). In our experiments, we found that inhibiting Notch signalling, and thereby decreasing Hes1 expression levels and oscillation amplitude, did not change cell cycle timing after 24 h of culture, and perturbation even led to a slight increase in mitotic index. Multiple reasons could explain the apparent discrepancy of the role of Notch signalling in the pre-NT. There might be species-specific differences in the requirement for Notch signalling (Cooper et al., 2024). Moreover, based on the remaining Hes1 expression we had detected, we presumably did not inhibit Notch signalling entirely, which might be in contrast to previous studies (Akai et al., 2005). In any case, our data suggest that it might not be the absolute signalling level but the dynamics mediating proliferation control in the pre-NT, as previously suggested for other regions (Maeda et al., 2023; Marinopoulou et al., 2021; Sueda et al., 2019). Although it is compelling to speculate that the Hes1 dynamics remaining upon Notch inhibition are sufficient to allow proliferation, other independent effectors and signals cannot be ruled out.

Not only is expression of Hes1 highest in the pre-NT, but FGF and Wnt signalling are also elevated. High FGF signalling prevents differentiation of the pre-NT to NT (Semprich et al., 2022; Patel et al., 2013; Delfino-Machin et al., 2005; Dubrulle and Pourquie, 2004; Dubrulle et al., 2001). By quantifying Hes1 dynamics in pre-NT cells upon inhibition of FGF signalling, we found that the amplitude of Hes1 dynamics was increased (Fig. 7H). This indicates that FGF has a dampening effect on Hes1 dynamics in the pre-NT and that reduced FGF signalling results in Hes1 dynamics with higher amplitude. Whether FGF signalling directly affects Hes1 dynamics or functions indirectly via induction of differentiation has to be disentangled in the future.

Together, this supports a hypothesis in which cells in the pre-NT are maintained in an undifferentiated, anticipating phase until they become part of the NT, where cells differentiate further in conjunction with the neighbouring somites.

We have quantified Hes1 oscillations in single PSM, NMP and pre-NT cells. In both NMP and pre-NT cells, Hes1 oscillations are noisy at high expression levels with a low amplitude. In contrast, in the PSM single-cell oscillations are less noisy at low expression levels with a higher amplitude (Fig. 7H). In addition to these single-cell oscillations, it is well known that oscillations between neighbouring PSM cells are coupled, leading to travelling waves of signalling along the tissue (Masamizu et al., 2006; Aulehla et al., 2008; Rohde et al., 2021 preprint). These coupled, highly synchronized oscillations ensure that information for proper segmentation can be transmitted from posterior to anterior in every segmentation cycle.

Even though the pre-NT lies directly adjacent to this oscillating field, the tissues are spatially separated by epithelialization and extracellular material. In the TB and pre-NT, we detect different expression patterns of several signalling components and downstream targets of the Wnt and Notch signalling pathway compared with the PSM (this study and MAMEP database). In correlation with this, Hes1 dynamics in pre-NT cells and NMPs are not only noisy but also less synchronized between neighbouring cells. The absence of synchronized oscillations or travelling waves implies that information in the pre-NT is not transmitted along the anterior-posterior axis via these dynamics. Conversely, Hes1 presumably has a more local function in regulating cellular behaviour on a single-cell basis, as has been suggested in other tissues (Seymour et al., 2020; Zhang et al., 2021).

Notably, once the pre-NT has differentiated to the NT, it has been shown that signalling between somite and NT does regulate further patterning and segmentation of the developing NT in conjunction with the differentiating somite (Kuan et al., 2004). This ensures that spinal cord, nerves and vertebrae form one unit. Thus, proper patterning of the spinal axis of the embryo would be dependent on proper somitogenesis. Based on this, somitogenesis would therefore have to be tightly regulated and coupled to axial elongation, which necessitates highly accurate information transmission. In contrast, differentiation and segmentation of the NT would be downstream of somite formation, which suggests that information could be kept local with the sole aim of allowing proliferation and preventing faulty or premature differentiation of single cells in the pre-NT.

In summary, Notch-dependent Hes1 dynamics in single cells of the developing embryonic tail differ in expression levels, amplitude and noise, which has implications for information transmission in the PSM and NT. The detailed molecular mechanisms through which Hes dynamics are generated and what information is stored in these dynamics in different cell types during embryonic development are important questions for future studies.

All animals were housed and bred according to institutional guidelines, and procedures were performed in compliance with Standards for Care and Use of Laboratory Animals with approval from the Hubrecht Institute ethical review board. All animal experiments were approved by the Animal Experimentation Committee (DEC) of the Royal Netherlands Academy of Arts and Sciences.

The Hes1-Achilles knock-in reporter line (Hes1^{tm1(Link-Achilles)Kfs}) was generated by employing CRISPR/Cas9 using IB10 ESCs. To generate Hes1-Achilles alleles, we targeted the stop codon of the endogenous Hes1 locus with a reporter cassette encoding a linked fluorophore (Hes1-linker-Achilles) using the Hes1 gRNA (CGCTCACTTCGGACTCCATG). The reporter cassette is linked to the Hes1 gene via a flexible GSAGS sequence and includes a selection cassette. Briefly, chimeric mice were obtained by C57BL/6 blastocyst injection and then outbred to establish the line through germline transmission. After generation of the knock-in reporter line, the selection cassette was removed by Cre-mediated excision to yield the final Hes1-Achilles allele, using the PGK-Cre mouse line (Lallemand et al., 1998).

The inducible iRFPnucmem mouse line [Gt(ROSA)26Sor^{tm1(CAG-iRFPnucmem)Aau}] was generated by the lab of Alexander Aulehla (EMBL Heidelberg, Germany). iRFPnucmem (Okita et al., 2004; Reimann et al., 2023) was amplified by PCR using the following primers: forward, 5′AGGCCGGCCGCCACCATGGCGG3′; reverse, 5′TGGCCGGCCTTAGTTTTCC3′. The resulting amplicon was then cloned into the Rosa26 targeting vector Ai9 (Madisen et al., 2010) using the FseI restriction enzyme to generate the loxP-stop-loxP-IRFP713-importin alpha1 construct for conditional expression (Shioi et al., 2011; Abe et al., 2011). The inducible iRFPnucmem mouse line was generated by standard gene targeting techniques using R1 embryonic stem cells. Briefly, chimeric mice were obtained by C57BL/6 blastocyst injection and then outbred to establish the line through germline transmission.

The LuVeLu (Aulehla et al., 2008) and Achilles-Hes7 (Yoshioka-Kobayashi et al., 2020) line were published previously. The R26R-H2B-mCherry (Abe et al., 2011) (‘H2B-mCherry’) and Fucci (Sakaue-Sawano et al., 2008) lines (http://www.clst.riken.jp/arg/reporter_mice.html) were provided by the Riken Center for Biosystems Dynamics Research and Riken BioResource Research Center, respectively (accession numbers CDB0204K, RBRC02706 and RBRC02707). The Tg (T-cre/ERT2)1Lwd/J (Anderson et al., 2013) (‘T-Cre/ERT2’) line was obtained from The Jackson Laboratory (accession number 025520). The Tg (Pgk1-cre)1Lni/CrsJ (Lallemand et al., 1998) (‘PGK-Cre’) line was obtained from the Jackson Laboratory (accession number 020811). The R26-CreERT2 (Ventura et al., 2007) (‘RosaCreERT’) line was obtained from the Jackson Laboratory (accession number 008463).

For the generation of Hes1-Achilles_RosaCreERT ESCs, E3.5 blastocysts were isolated from pregnant Hes1-Achilles_RosaCreERT females. Each blastocyst was transferred to a single well of a 24-well plate seeded with feeder MEFs. After 6 days, cells were trypsinized and expanded into a 12-well plate with feeders. Four days later, cells were expanded into six-well plates, and simultaneously used for extraction of genomic DNA followed by PCR to determine the genotype. After confirmation of genotype, cells were further expanded for one more passage before freezing down. ESCs were cultured in N2B27-2i medium, as previously described (Silva et al., 2008).

The Sox2-H2B-iRFP (‘Sox2-iRFP’) knock-in reporter line was generated employing CRISPaint (Schmid-Burgk et al., 2016) gene targeting techniques using Lipofectamine 3000 (ThermoFisher, L3000001) and Hes1-Achilles_RosaCreERT ESCs. To generate Sox2-iRFP alleles, we targeted a PAM site before the stop codon of the endogenous Sox2 locus with a reporter cassette coding for a separated H2B gene linked to a fluorophore (Sox2-T2A-H2B-linker-iRFP) using the Sox2 gRNA (CACTGGAGCTGTCCTTAAATAGG). The reporter cassette was separated from the Sox2 gene via a small 2A sequence. Furthermore, the fluorophore was linked to the H2B gene via a flexible GSAGS sequence. The selection cassette was also separated from the reporter cassette via a small 2A sequence. Selection occurred 2 days after transfection and lasted 3 days. Colonies were picked and transferred to a single well of a 96-well plate seeded with irradiated feeder MEFs. Colonies were then expanded, genotyped and frozen as described above.

Hes1-Achilles mice were genotyped by PCR using the primers Hes1-C-forward (GACCTCGGTGGGTCCTAACGC), linker-Achilles-reverse (CTTGCCGGTGGTGCAGATCAG) and Hes1-N-reverse (GAGGTGGGCTAGGGACTTTACGG). iRFPnucmem mice were genotyped using Rosa-wt-forward (GAGCTGCAGTGGAGTAGGCG), CAG promoter reverse (CTCGACCATGGTAATAGCGA), Rosa-wt-forward (TTGGAGGCAGGAAGCACTTG) and Rosa-wt-reverse (CCGACAAAACCGAAAATCTGTG) primers resulting in the following bands: wild type, 358 bp; 580 bp, transgene.

For all experiments (unless stated otherwise) female mice were euthanized at 10.5 dpc and embryos dissected. Dissections were performed in PBS supplemented with 1% bovine serum albumin (Biowest, P6154), 200 μg penicillin-streptomycin (Gibco, 15140122), 1× GlutaMAX (Gibco, 35050038) and 10% glucose solution 45 (Sigma-Aldrich, G8769). For sparse labelling of mice with H2B-mCherry, iRFPnucmem and Fucci reporters, tamoxifen (Sigma-Aldrich, T5648, dissolved in corn oil) was injected intraperitoneally into pregnant females on day E9.5. For H2B-mCherry and iRFPnucmem, 1 mg and 0.1 mg were used, respectively.

Ex vivo culturing was performed as described previously (Sonnen et al., 2018; Lauschke et al., 2013; Aulehla et al., 2008). The neurobasal medium used to culture ex vivo samples consists Special Advanced DMEM/F-12 (Cell Culture Technologies) and Neurobasal medium (Gibco, 12348017) (1:1), 1× N2 (Gibco, 17502048), 1× B27 (Gibco, 12587010), 1× GlutaMAX (Gibco, 35050038), 0.004% bovine serum albumin (Biowest, P6154), 0.1 mM 2-mercaptoethanol and 200 μg penicillin-streptomycin (Gibco, 15140122). The following small molecule inhibitors dissolved in DMSO were used at the concentration indicated in the text and figure legends: FGF-basic (‘bFGF’, Peprotech, 450-33), CHIR99021 (‘Chiron’, Sigma-Aldrich, SML1046), DAPT (Sigma-Aldrich, D5942), IWP-2 (Sigma-Aldrich, I0536), retinoic acid (RA; Sigma-Aldrich, R2625-50MG) and SU5402 (Sigma-Aldrich, SML0443). Samples were imaged in eight-well Chambered Coverglass w/non-removable wells (ThermoFisher, 155411PK). Details on resources used can be found in Table S1.

ESCs were differentiated to NMP, mesodermal or neural fate as described previously (Gouti et al., 2014). To minimize fluorescent background, advanced DMEM/F-12 (Cell Culture Technologies) without Phenol Red was used. Cells were differentiated and imaged in µ-Dish 35 mm Quad dishes (Ibidi, 80416).

To compare protein half-lives of Hes1 wild-type and Hes1-Achilles protein, mESCs heterozygous for Hes1-Achilles were maintained on feeder cells. Two days before the experiments, feeder cells were removed and ESCs cultured in N2B27-2i medium, after which 50 µg/ml cycloheximide (Sigma-Aldrich) or DMSO was added for the indicated amount of time. Cells were harvested and lysed in 2× Laemmli buffer [125 mM Tris-HCl (pH 6.8), 4% SDS, 10% glycerol, 0.2% Bromophenol Blue and 0.05 mM DTT] at 95°Cfor 5 min. Cell numbers were counted to load equal amounts. Proteins were separated in a 12% gel by denaturing polyacrylamide gel electrophoresis (SDS-PAGE) and blotted onto a PVDF membrane as described previously (Sonnen et al., 2013). Western blots were analysed by staining for Hes1 (anti-Hes1 rabbit, clone D6P2U, Cell Signalling Technology, 11988) and alpha-Tubulin (anti-alpha-Tubulin mouse, clone DM1A, Sigma-Aldrich, T6199). Band intensities were quantified using Fiji (Schindelin et al., 2012) and Hes1 levels were quantified by normalizing to the corresponding alpha-Tubulin bands and then to the DMSO control using Python.

To confirm Hes1 antibody specificity in western blotting, knockdown was performed in ESCs. ESCs were cultured on feeder cells. To initiate the experiment, feeder cells were removed and ESCs cultured in N2B27-2i medium. After 1 day of adherent culture, mESCs were transfected with 25 pmol siRNA (Hes1 siRNA #1 GGGAAGUUUAUUUGAGAAAUU, Hes1 siRNA #2 GGAUUGCGCCUUUGUAUUAUU, Hes1 siRNA #3 AGAUCAACGCCAUGACCUAUU; control: GL2 siRNA CGUACGCGGAAUACUUCGAUU) using Lipofectamine RNAiMAX (ThermoFisher, 13778075) according to the manufacturer's instructions. Cells were harvested 72 h after transfection and analysed by western blotting (as described above).

For immunostaining, samples were fixed in 4% formaldehyde for 30 min and kept in PBS at 4°C. Commercial antibodies were used, including anti-Dll1 rat (MABN2284, Sigma-Aldrich), anti-Dll4 goat (AF1389-SP, R&D systems), anti-cleaved caspase 3 rabbit (9661S, CST), anti-phospho Histone H3 mouse (05-806, Sigma-Aldrich), Alexa Fluor 555 Phalloidin (A34055, ThermoFisher), anti-Lyz1 (A0099, DAKO), anti-Tbx6 rabbit (ab38883, Abcam), anti-Sox2 mouse (ab79351, Abcam), anti-Sox2 rabbit (ab92494, Abcam), anti-T human/mouse (AF2085-SP, R&D Systems), anti-Pax6 recombinant (ab195045, Abcam), goat anti-rat Alexa Fluor 555 (A-21434, ThermoFisher), donkey anti-rabbit Alexa Fluor 488 (A-21206, ThermoFisher), donkey anti-rabbit Alexa Fluor 568 (A10042 ThermoFisher), donkey anti-rabbit Alexa Fluor 405 (A48258, ThermoFisher), donkey anti-mouse Alexa Fluor 647 (A-31571, ThermoFisher), donkey anti-rat Alexa Fluor 555 (A78945, ThermoFisher), donkey anti-goat Alexa Fluor 647 (A32849, ThermoFisher), donkey anti-mouse Alexa Fluor 568 (A10037, ThermoFisher) and donkey anti-mouse Alexa Fluor 647 (A-31571, ThermoFisher). We stained samples with primary and secondary antibodies overnight at a 1:1000 dilution in PBST (0.1% Tween 20) containing 1% bovine serum albumin (Biowest, P6154). For imaging, a z-stack of 10-12 planes at 10 µm distance was scanned in 1024×1024 pixels (pixel size 0,758 µm) using a Leica TCS SP8 MP confocal microscope.

For the characterization of adult tissue in the Hes1-Achilles mice, two males and two virgin females of 8 weeks from both Hes1-linker-Achilles (homozygous) and C57BL/6 background each were used. Small intestine was isolated (from beneath the stomach until the caecum) and immediately flushed with formalin 4%. The tissue was rolled and incubated in formalin 4% for 1 day at room temperature. This was followed by dehydration using a series of ethanol solutions (30%, 50% and 70%). Next, the slides were soaked in xylene, embedded in paraffin wax and sectioned into 5 µm sections. For staining, the slides were soaked in xylene, 100% ethanol, 95% ethanol, 70% ethanol and 50% ethanol, and rinsed with running cold tap water. For antigen retrieval, the slides were cooked in a citrate buffer [sodium citrate (8.2 mM) and citric acid (1.8 mM) at pH 6]. 5% goat serum (ab7481) in 0.3%-Triton X100 PBS was used for blocking and permeabilization. Primary (overnight at 4°C) and secondary antibodies (3 h at room temperature) were used in 0.1%BSA PBS (Lyz1, Dako, A0099, 1:200). After staining, gut rolls were mounted with Vectashield (H1200NB). The samples were imaged using a SP8 confocal microscope (Leica Microsystems, 40× oil objective, numerical aperture=1.3).

Probe generation and in situ hybridization have been described previously (Aulehla et al., 2003). Probes against Axin2, Dusp4, Hes7 and Lfng were used as described in the literature (Aulehla et al., 2003; Mansouri et al., 1997; Saga et al., 1997). Probes against Achilles and Hes1 were generated using the full-length cDNA (Morimoto et al., 2007). Subsequently, samples were dehydrated, embedded in paraffin wax and sectioned. Standard Hematoxylin and Eosin staining was performed.

Hybridization chain reaction was performed as described previously (Choi et al., 2018). Probe sets for the following gene targets: Dll1, Dll3, Dll4, Hes5, Hes7, Meox1, Notch1, Notch2, Notch3, Notch4 and Pax6 were generated (Kuehn et al., 2022). All animals used for hybridization chain reaction were stained once using the same hybridization, washing and amplification time for all gene targets.

Images of in situ hybridization were taken with a Leica M165 FC stereomicroscope using 8:1 zoom. Images of hybridization chain reaction were taken with a Leica TCS SP8 MP confocal microscope using a 10× and 20× objective. Brightness and contrast were adjusted uniformly to the entire image.

Real-time imaging was performed using a Leica TCS SP8 MP confocal microscope featuring an incubator and gas mixer for CO₂, O₂ and temperature control. Samples were excited with a OPSL Laser at a wavelength of 514 nm (Hes1-Achilles, Achilles-Hes7 and LuVeLu), 568 nm (H2B-mCherry) or 638 nm (Sox2-iRFP and iRFPnucmem) through a HC PL APO CS2 20×/0.75 DRY objective. For ex vivo culture experiments, a z-stack of 10-12 planes at 5-10 µm distance was scanned every 10 min. For in vitro differentiation experiments, a z-stack of three or four planes at 5 µm distance was scanned every 5 min. Multiple samples were recorded using a motorized stage during each experiment. Movies were recorded in 512×512 pixels (pixel size 1517 µm).

For quantification of mean intensity or number of positive cells in fixed sample imaging, Fiji (Schindelin et al., 2012) was used. For quantification of mitotic cells in spread-out cultures, phospho-histone H3-positive and Sox2-positive cells were counted in the Sox2-positive region using the ‘Analyze Particles’ function in Fiji and the percentage determined. For quantification of mitotic cells in the embryonic tails, a maximum intensity projection of slices containing the neural tube was made and a region of interest (ROI) outlining the neural tube manually placed. Positive cells were quantified using the ‘Analyze Particles’ function in Fiji using standard settings (size 50-400 µm² and circularity 0.0-1.0). Numbers are displayed as cells per area. Kymographs were generated along the posterior-anterior axis as described previously (Sonnen et al., 2018). For kymographs of embryonic tails, movies were smoothened (gaussian blur, radius 5), a maximum intensity projection was made of slices containing the tissue of interest and movies were registered using the ‘Correct 3D drift’ function in Fiji. Kymographs were then generated along a 25 pixel-wide line. For quantification of mean intensity over time, ROImanager (quantification of intensity of cell populations), Mastodon (quantification of spot centre intensities for single-cell tracking in spread-out cultures) or manual tracking (quantification of a ROI with 5-15 µm diameter for 2D in vitro differentiations) were used.

Data analysis was performed using Julia unless stated otherwise. Quantification of Hes1 intensity was performed as follows: time-courses of cell populations in embryonic tails upon signalling perturbations (Fig. 2 and Fig. S3) were normalized to the starting point of each sample. For quantification of Hes1 intensity in unperturbed embryonic tails or spread-out cultures (Figs 3 and 4), absolute intensity was quantified. Hes1 intensity in spread-out cultures upon signalling perturbations (Figs 6 and 7) was determined by normalizing to the mean of all DMSO samples of each experiment.

Quantification of oscillation dynamics was performed as follows: each time series track was normalized by dividing by its own mean. Tracks were smoothened using rolling average (window size 5). Detrending was performed by determining the trend using rolling average (window size 15) and subtracting this from the smoothened signal. Quantification of oscillation dynamics were performed using pyBOAT (Schmal et al., 2022) implemented in Julia [Settings for wavelet analysis were as follows: global period cutoff for sync-smoothing 225 min, window size for amplitude envelope 450 min, period range from 50 to 350 min (step size 1), window size for ridge detection 45.] For further analysis of period and amplitude, the mean of each track was determined. Fourier transform was obtained from the pyBOAT analysis. For further analysis, peaks with highest power were used (percentile>98). The coefficient of variation was calculated based on these Fourier peaks using the formula cov=std (FourierPeriods)/mean (FourierPeriods).

To quantify synchrony between PSM or NT cells, we calculated the Kuramoto order parameter using python (Sanchez et al., 2022). We first obtained the instantaneous phase of Hes1 dynamics of single cells using pyBOAT (Schmal et al., 2022). The instantaneous phases were used to calculate the instantaneous Kuramoto order parameter of PSM and NT cells per embryo 2D spread-out culture independently. All timepoints were considered, when tracks of at least two cells were available. Last, we extracted the median instantaneous Kuramoto order parameter per tail-tip and cell type as a measure for synchronicity within single tail-tips. Significance was computed using an unpaired Student's t-test.

Gut slices were analysed using Fiji (Schindelin et al., 2012), Lyz1-positive cells per crypt were counted manually and graphs were plotted with Prism. To visualize data, Tukey style boxplots were used (bar indicates the median, boxes indicate the 25th and 75th percentiles, and whiskers indicate 1.5×IQR). For statistical comparisons between groups, a Mann–Whitney U test was used.

To quantify cell cycle duration in the NT, we tracked single cells carrying Fucci reporters and a mosaically expressed iRFPnucmem. Cell tracks start at a mitotic event and finish at the next mitotic event. Track lengths were used to calculate cell cycle duration. The start of the track until the peak of Cdt1-mKO2 was considered as G1 phase duration, while the peak of Cdt1-mKO2 until the end of the track was considered as S-G2-M duration. To smoothen reporter quantifications, we used a moving average with a window size of five timepoints. In addition, all tracks were min/max normalized between 0 and 1.

For all experiments at least three independent experiments were performed. To visualize data, Tukey style boxplots were used (bar indicates the median, boxes indicate the 25th and 75th percentiles, and whiskers indicate 1.5×IQR). Data displayed as violin plots present data from max to min with bar indicating the median. For statistical comparisons between groups, Mann–Whitney U tests were computed in Julia, unless stated otherwise.

We thank all members of the Sonnen, Bothma and Garaycoechea labs for feedback and discussions. We are thankful to Juan Garaycoechea and Jacques Bothma for feedback on the manuscript. We thank Jan-Daniël de Leede for help with data analysis, Valerie Wilson, Mina Gouti and Jana Wolf for useful discussions, and Andreas Sonnen for help with Alphafold. We thank Jeroen Korving, Stieneke van den Brink, Benaissa el Haddouti, Joris Geigenmüller, Tjerk Swinkels and Mariska Dijkers for technical help. This work was supported by the Hubrecht imaging and animal facilities. The iRFPnucmem mouse line was generated with support from the EMBL Transgenic Core facility (Yvonne Petersen) and mouse facility. We thank Alexander Aulehla for support and providing mouse lines, in particular the inducible iRFPnucmem line. The H2B-mCherry mouse line was kindly provided by the Riken Center for Biosystems Dynamics Research and the Fucci line by the Riken BioResource Research Center. The T-Cre/ERT2, RosaCreERT and PGK-Cre lines were kindly provided by Jackson Laboratories. The LuVeLu reporter line was kindly provided by Olivier Pourquié, the Achilles-Hes7 by Ryoichiro Kageyama and the Achilles construct by Atsushi Miyawaki (Riken Center for Brain Science). The iRFPnucmem construct was kindly provided by Timm Schroeder.

The peer review history is available online at https://journals.biologists.com/dev/lookup/doi/10.1242/dev.202936.reviewer-comments.pdf