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

The intricate dynamics of Hes expression across diverse cell types in the developing vertebrate embryonic tail have remained elusive. To address this, we developed an endogenously tagged Hes1-Achilles mouse line, enabling precise quantification of dynamics at the single-cell resolution across various tissues. Our findings reveal striking disparities in Hes1 dynamics between presomitic mesoderm (PSM) and preneural tube (pre-NT) cells. While pre-NT cells display variable, low-amplitude oscillations, PSM cells exhibit synchronized, high-amplitude oscillations. Upon the induction of differentiation, the oscillation amplitude increases in pre-NT cells. Additionally, our study of Notch inhibition on Hes1 oscillations unveiled distinct responses in PSM and pre-NT cells, corresponding to differential Notch ligand expression dynamics. These findings suggest the involvement of separate mechanisms driving Hes1 oscillations. Thus, Hes1 demonstrates dynamic behaviour across adjacent tissues of the embryonic tail, yet the varying oscillation parameters give rise to differences in the information that can be conveyed by these dynamics.


Introduction
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 1 .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.
NMPs transitioning from the TB to the NT pass through the preneural tube (pre-NT), surrounded by oscillating PSM 25 .FGF signalling prevents premature differentiation in the pre-NT, while Notch signalling supports cell proliferation 26,27 .In the anterior NT, neural progenitor cells continue to proliferate and differentiate into future spinal cord neurons.Whether neural progenitor cells proliferate or differentiate has been linked to oscillations in Hes1 and Hes5, reminiscent of their function in somitogenesis 11,12 .Hes genes oscillate in progenitor cells to maintain a proliferative state 11,12,[28][29][30] , while sustained expression has been linked to quiescence 31,32 .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 neuromesodermal progenitors (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 presomitic mesoderm (PSM), tailbud (TB), and preneural tube (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 revealed disparate responses of PSM and pre-NT cells to signalling pathway perturbations.This also correlated with changes in proliferation.Thus, our new endogenous reporter mouse line, combined with recently available single-cell tracking software, allowed us to uncover previously unquantified dynamics in single cells of the mouse embryonic tail.

Generation of a Hes1-Achilles mouse line to study signalling dynamics in the posterior embryonic tail
To investigate Hes dynamics in the developing mouse embryo, we first clarified which Hes genes were expressed in the PSM, NMP and in particular the pre-NT.The latter being defined as the tube-like Pax6-negative region emanating from the TB, which lies next to the PSM (Fig. 1A, B) 26,27 .To this end, we performed in situ hybridization (ISH) and hybridization chain reactions (HCR).Expression of Hes7 and Lfng (lunatic fringe), key components of the segmentation clock 24,[33][34][35][36] , was restricted to the PSM and TB or PSM, respectively (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 (Fig. 1C, S1C, D) (also see MAMEP database, http://mamep.molgen.mpg.de).Thus, both Hes1 and Hes5 are expressed in the different regions 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 Hes1 expression levels over time, we generated a new endogenously tagged Hes1 mouse line (Fig. 1D-H).
Building on previous approaches 31,[37][38][39][40] , we endogenously tagged Hes1 with Achilles, a fastmaturing fluorescent protein 38 (Fig. 1D, S2).We placed Achilles at the C-terminal side of Hes1 since this lies opposite to the DNA-binding bHLH (basic helix-loop-helix) domain of Hes1 2 .In addition, we separated Hes1 and Achilles by a flexible linker, because we reasoned that this should allow interaction of Hes1 via its C-terminal WRPW motif (four amino acid proteinprotein interaction domain) with the Groucho/TLE family of transcriptional repressors 41,42 .
Structure of the fusion protein was assessed for accessibility of the WRPW motif and bHLH domain by Alphafold 43 (data not shown).Homozygous mice were born at Mendelian ratio (offspring of 8 heterozygous x heterozygous breedings: 17 wildtype, 34 heterozygous, 16 homozygous), viable and fertile.When comparing mRNA staining of Hes1 in wildtype embryos and Achilles in Hes1-Achilles embryos, the expression patterns corresponded (Fig. 1F, G).At protein level, Hes1-Achilles expression indicated the expected pattern with high expression levels in the pre-NT and stripes in forming somites (Fig. 1C, H).Thus, we successfully generated a homozygously viable, fertile and endogenous Hes1-Achilles reporter mouse line.

Hes1 expression is dependent on Notch signalling in the posterior embryonic tail
We next addressed whether Hes1 expression was solely dependent on Notch signalling or also on other pathways, as suggested previously for Hes7 39 .To be able to test this, we had to culture E10.5 embryonic tails ex vivo to quantify Hes1-Achilles dynamics upon pathway perturbation.However, standard embryo culture protocols 44,45 needed to be optimized for survival of the NT.Therefore, we optimised the protocol by testing different media to improve pre-NT and NT survival (Fig. 2A and data not shown).In standard embryo culture medium, cleaved caspase-3 expression levels especially in the pre-NT and NT were significantly higher than in uncultured embryos at embryonic day E11.5 (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 Methods), the number of apoptotic cells was not increased compared to 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.Despite this, somites were less well organized in neurobasal medium than in control samples (Fig. 2A).To find out whether culture conditions influenced the well-studied dynamics of the segmentation clock, we compared embryonic tails expressing LuVeLu 46 cultured in either embryo culture medium or neurobasal medium (Fig. S3A-D).Wave and 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, B).Accordingly, period and amplitude remained unchanged (Fig S3C , D).Thus, dynamics of the segmentation clock were maintained, and the optimized culture conditions of embryonic tails now allow the investigation of signalling dynamics in PSM, NMP and pre-NT ex vivo.
Using these culture conditions, we initially tested how pathway perturbation affected Hes1-Achilles expression levels.Notch inhibition, using the gamma-secretase inhibitor DAPT, led to a drop in Hes1-Achilles expression levels within 3 h (Fig. 2D, E).The dynamics and signal loss were comparable to the effect of Notch inhibition on the reporter Achilles-Hes7 38 (Fig. S3F, G).Besides Notch signalling, we also focused on Wnt (Fig. S3H-J) and FGF (Fig. S3K-M) signalling, as gradients of these morphogens are present in the embryonic tail including the pre-NT (Fig. S3E).Accordingly, we found Dusp4 and Axin2, downstream targets of FGF and Wnt signalling, respectively, to be highly expressed in the pre-NT (Fig. S3H, K).We therefore perturbed embryonic tails with small molecules inhibiting these pathways: Wnt signalling was inhibited using the porcupine inhibitor IWP2 and FGF signalling was inhibited using SU5402, a small molecule inhibitor against the FGF receptor.Inhibiting either did not lead to a direct (IWP2) or consistent (SU5402) drop in signal, suggesting that these pathways might have more indirect effects on Hes1 expression levels.Thus, Hes1 expression in the posterior embryonic tail is driven by Notch signalling.

Population-wide Hes1 dynamics differ between TB, pre-NT and PSM
Subsequently, we carefully quantified Hes1-Achilles dynamics in the different regions of the embryonic tail (TB, PSM and pre-NT) (Fig. 3A-G, movie 1).We quantified dynamics in the TB region, which includes the NMPs but presumably also some posterior PSM cells.Every region significantly differed in expression level, in particular the pre-NT region showed higher Hes1-Achilles expression levels compared to 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 prominent oscillations over time compared to the other regions (Fig. 3G).When analysing the period using wavelet transform 47 , there was a slight increase in periods in the TB region compared to PSM region (Fig. 3I).A similar trend towards an elevated period was observed for the pre-NT region.Furthermore, we found that the amplitude of the PSM region was significantly higher than in the pre-NT and TB regions (Fig. 3J).Finally, whereas kymographs revealed travelling waves of Hes1-Achilles expression levels in the PSM region (see * in kymograph), no clear tissuewide pattern was detected in the pre-NT region (Fig. 3K).Overall, the data suggests that Hes1 is dynamic in the different regions of the mouse embryonic tail.However, Hes1 expression levels in the pre-NT are noisier and less oscillatory compared to the PSM.

Hes1 is oscillatory in single cells of the pre-NT
The fact that we did not find clear population-wide Hes1 oscillations or travelling waves in the pre-NT region could have different reasons: (1) There are no oscillations.(2) Single cells oscillate but are not synchronized with each other to induce tissue-wide oscillation dynamics.
To test this, we performed a 2D spread-out experiment 45 combined with mosaic labelling of nuclei that allowed us to track cells individually and quantify Hes1-Achilles expression levels in single cells (Fig. 4A-E, movie 2).When posterior embryonic tail tips are cultured on fibronectin-coated dishes, the tissue spreads and cells, including those of the pre-NT, grow in a quasi-monolayer, which facilitates single-cell tracking 45 .Immunostaining and HCR for neural and mesodermal markers in 2D cultures confirmed presence and survival of PSM and neural cell types, when culturing in neurobasal medium (Fig. S4A, B).As was observed in the corresponding regions (Fig. 1,3), pre-NT cells showed higher Hes1-Achilles expression levels than PSM cells, the latter of which slightly increased in intensity upon differentiation to somites (Fig. 4F).Interestingly, oscillatory dynamics in Hes1-Achilles 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 (data not shown).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. 3) and single-cell level (Fig. 4H) supports the hypothesis that Hes1 dynamics are not synchronized between neighbouring pre-NT cells.
Using Fourier transform to analyse the dynamics, similar periods were obtained (Fig. 4I, J,    S4C-E).Comparison of Fourier spectra of individual cells indicated that Hes1-Achilles dynamics were noisier in pre-NT than PSM cells, since PSM cells mostly showed one main peak at 2.5 h and pre-NT cells multiple peaks around 2-4 h (Fig. 4I, J).As a measure of variability, we determined the full width at half maximum (FWHM) for the period quantification (Fig. S4E), which was 0.5 for the PSM and 1.6 for the pre-NT.The coefficient of variation was 0.1 and 0.2 for PSM and pre-NT cells, respectively.Thus, like PSM cells, single cells in the pre-NT show Hes1 oscillations.However, oscillations are more variable and not synchronized between neighbouring cells.

Hes1 shows oscillations in NMPs that are driven by Notch signalling
After the discovery of Hes1 dynamics in both PSM and pre-NT cells, we next sought to quantify Hes1 dynamics in NMPs, from which both PSM and pre-NT arise.The 2D spread-out experiments (Figure 4) did not allow us to unequivocally identify NMPs.Therefore, we analysed Hes1-Achilles dynamics in NMPs by in vitro differentiation.To this end, we applied a previously published NMP differentiation protocol (Fig. 5A) 48 to generate NMPs from embryonic stem cells (ESCs) derived from Hes1-Achilles mice (Fig. 1E).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 using CRISPaint 49 .Based on previous literature and immunostainings, we defined the NMP phase between 56 and 72 hours (Fig. 5A, S5A) 48,50 .When analysing Hes1-Achilles dynamics at population level, oscillations resembled in appearance and amplitude that of the TB region of the ex vivo embryonic tail, albeit with a reduced period (Fig. 5B-E, 3D, E).To find out if single cells show Hes1 oscillations, we next tracked single NMP cells 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 and had a low amplitude (Fig. 5G, I).When perturbing Notch signalling using the inhibitor DAPT, we detected a dosedependent decrease of Hes1-Achilles expression levels (Fig. S5B-E).This inhibition did not show an apparent effect on the period or amplitude of Hes1-Achilles dynamics.Furthermore, when we differentiated NMPs further towards the mesodermal or neural lineage (Fig. S5F-P), cells displayed dynamics corresponding to those found ex vivo (Fig. 3).Thus, NMPs show variable oscillations in Hes1 expression levels that are dependent on Notch signalling.

Induced differentiation to neural tube by FGF inhibition results in Hes1 dynamics with increased amplitude in the pre-NT
Previous studies have highlighted the function of Hes dynamics in the anterior NT 6,13,28,29,32,51 .
Therefore, we asked how Hes1 dynamics changed upon induced differentiation.Since FGF inhibition induces differentiation of pre-NT to NT 25,27,52 and PSM to somite 53,54 , we inhibited FGF activity in 2D spread-out cultures using SU5402 (Fig. 6).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. S6A).In agreement with changes of Hes1 expression levels from 'posterior' to 'anterior' PSM (Fig. 4F), Hes1-Achilles expression levels were increased in the PSM region upon FGF inhibition, while period and amplitude remained unchanged (Fig. S6B-D).Conversely, we did detect a decrease in Hes1-Achilles amplitude in the pre-NT region upon FGF inhibition (Fig. S6E-G).
To investigate the effect of FGF inhibition on single cell oscillations, we then performed singlecell tracking.In PSM cells FGF inhibition had a similar effect on Hes1-Achilles dynamics as at population level with increased absolute expression levels and a corresponding increase in amplitude, i.e. the relative amplitude remained unchanged, while the period was maintained (Fig. S6H-J).In contrast, FGF signalling inhibition had a different effect on Hes1-Achilles dynamics in pre-NT cells: absolute expression levels of Hes1-Achilles and the oscillation period remained similar (Fig. 6A-D 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.

Changing Hes1 dynamics by modulating Notch signalling correlates with an increase in cell proliferation
We next addressed if 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 such oscillations via repression of Hes genes by Hes proteins themselves 13,24,35,[55][56][57][58] .In addition to a Hes-driven feedback loop, the Notch ligand Dll1 is dynamically expressed in the PSM 11,59 , which has been proposed to drive a delayed coupling mechanism of segmentation clock oscillations resulting in kinematic waves in the PSM 38,[59][60][61] .In neural tissue, however, Dll1 oscillations were not observed in the pre-NT, but only in the NT adjacent to the last formed somites, where single cells started to oscillate in a salt-and-pepper like pattern 11 .Thus, 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 Delta-like (Dll) Notch ligands Dll1, Dll3 and Dll4 and the Notch receptors to be expressed in the PSM, TB and pre-NT by HCR 62 (Fig. S7A,    B).We next performed immunostainings to visualize ligand protein expression levels.In agreement with previous publications 11 , Dll1 expression levels were elevated in the PSM compared to the pre-NT and staining patterns varied between embryonic tails indicating dynamic expression (Fig. 7A and data not shown).However, Dll1 -just like Dll4 -was not only detected in the PSM but also in the pre-NT and TB, albeit at lower expression levels and without indications of dynamics (Fig. 7A).Together with previous findings on Dll1 oscillations in the embryonic tail 11 , this suggests that Dll1 and Dll4 are present at low expression levels in pre-NT cells but not oscillatory, whereas Dll1 expression levels oscillate in PSM cells (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 (Fig. 7, S7).Analysing dynamics in PSM at population level, we found that Hes1-Achilles levels decreased to a similar extent for all DAPT concentrations used (Fig. S7C).The oscillation period was slightly increased at 1 µM DAPT, however the amplitude was strongly reduced independent of the DAPT concentration used (Fig. S7D, 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.In contrast, in pre-NT (Fig. S7F-H) Notch inhibition led to dose-dependent decrease in Hes1-Achilles levels (Fig. S7F) and amplitude, which was only significant at 10 µM DAPT (Fig. S7H).
To clarify how Notch inhibition affects single-cell oscillations, we performed single-cell tracking and quantified Hes1-Achilles levels in PSM in the different conditions.Also, at single-cell level, Notch inhibition led to a dose-dependent decrease of Hes1-Achilles levels in pre-NT cells (Fig. 7C-E, S7I-K).This correlated with a corresponding decrease in oscillation amplitude with the relative amplitude remaining unchanged (Fig. 7G, S7K).However, no significant change in period was observed (Fig. 7F, S7J).Conversely, in PSM cells Notch inhibition led to a decrease in both Hes1-Achilles expression levels and the relative amplitude (Fig. 7C-G, S7L-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.
While the essential function of Notch signalling in somitogenesis has been studied extensively [63][64][65][66] , its role in the pre-NT especially with regards to signalling dynamics is less clear, even though it has been implicated in cell proliferation control 26 .In other regions of the nervous system, Hes1 oscillations also regulate proliferation with high, sustained Hes1 expression levels promoting quiescence 31,32,67 .We therefore addressed how Notch signalling inhibition and a corresponding modulation of Hes1-Achilles dynamics affected pre-NT proliferation.Interestingly, when quantifying the percentage of phospho-histone H3-positive mitotic cells in the Sox2-positive region (mitotic index), we found that Notch inhibition led to an increase in the mitotic index (Fig 7F , G, S7O).This suggests that lowering Notch signalling and thereby Hes1 expression levels and oscillation amplitude, while leaving the period unchanged, promotes proliferation in the pre-NT.

Discussion
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.To enable this, we have generated a new and endogenous Hes1 reporter mouse line.It contains the fastmaturing fluorescent protein Achilles, which is placed C-terminally to the Hes1 gene.Importantly, the line is homozygously viable and fertile and allows the investigation of Hes1 dynamics at single-cell resolution in various tissue types (Sonnen, unpublished).Using this new reporter, we compared Hes1 dynamics in the PSM and pre-NT at population and singlecell 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 68 , we made use of neurobasal medium which 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.While 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, dynamics of the segmentation clock are maintained.This medium therefore allowed us to culture embryonic tails ex vivo for the quantification of signalling dynamics in both tissue types in the same culture.

Differential mechanisms driving Hes1 dynamics in the PSM and pre-NT
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 (Figure 7H).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 an over-proportional decrease in amplitude was induced, i.e. the relative amplitude decreased, both when analysing Hes dynamics at population and single-cell level.This implies that different mechanisms drive Hes1 dynamics in the two tissue types.While the effect of Notch inhibition on Hes1 dynamics in pre-NT cells is consistent with a delayed negative feedback loop downstream of constant Notch signalling 55,56 , 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 11,69,70 .Previously, Dll1 oscillations were not detected in the pre-NT, but only in the NT adjacent to the somites 11 .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, while Dll4 leads to more sustained expression of downstream target genes 71 .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 11,61,72 .Loss of Notch-dependant intercellular coupling could therefore explain the loss of population-wide dynamics we observed already at low Notch inhibitor concentrations.However, the oscillation amplitude was also reduced in single cell oscillations upon Notch inhibition arguing against a sole effect on synchronicity between oscillators.The fact that high Notch inhibitor levels also led to a decrease in the relative 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 73 .Our mosaic labelling approach for single-cell tracking does not allow us to quantify Hes1 dynamics in neighbouring cells, which might be studied further in the future.

Implications for the role of Hes1 dynamics in the preneural tube
Hes1 dynamics have been found to regulate proliferation and differentiation in the anterior developing NT 13,67,74 .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 31,32,67 .It is therefore interesting to note that we find Hes1 expression levels highest in the pre-NT and expression decreases towards the NT.When inducing a decrease in Hes1 expression levels and oscillation amplitude by inhibiting Notch signalling, the mitotic index was indeed increased.Similar to morphogen gradients in the PSM, FGF and Wnt signalling are high in the pre-NT.High FGF signalling prevents differentiation of the pre-NT to NT 25,27,[52][53][54] .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, waiting phase until they become part of the NT, where cells differentiate further in conjunction with the neighbouring somites.

Information transmission by noisy Hes1 oscillations
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 level with low amplitude.In contrast, in the PSM single-cell oscillations are less noisy at low expression level with 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 40,46,75 .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 to 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 anteriorposterior 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 69,70 .
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 76 .This ensures that spinal cord, nerves and vertebrae form one unit.Thus, proper patterning of the embryo's spinal axis 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 to allow proliferation and prevent 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 mechanism of how 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.plates, and simultaneously used for extraction of genomic DNA followed by polymerase chain reaction (PCR) to determine the genotype.After confirmation of genotype, cells were further expanded for one more passage prior to freezing down.ESCs were cultured in N2B27-2i medium as previously described 81 .
The Sox2-H2B-iRFP ("Sox2-iRFP") knock-in reporter line was generated employing CRISPaint 49 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).The reporter cassette is separated from the Sox2 gene via a small 2A sequence.Furthermore, the fluorophore is linked to the H2B gene via a flexible GSAGS sequence.The selection cassette is also separated from the reporter cassette via a small 2A sequence.Selection occurred two days after transfection and lasted three 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 explained above.
Ex vivo culture was performed as described previously 45,46,68

In vitro differentiation
ESCs were differentiated to NMP, mesodermal or neural fate as described previously 48,50 .To

In situ hybridization and hybridization chain reaction
Probe generation and in situ hybridization were described previously 18 .Probes against Axin2, Dusp4, Hes7 and Lfng were used as described in the literature 18,82,83 .Probes against Achilles and Hes1 were generated using the full-length cDNA 84 .Subsequently samples were dehydrated, paraffin embedded, and sectioned.Standard H&E staining was performed.
Hybridization chain reaction was performed as described previously 62 .Probe sets for the following gene targets: Dll1, Dll3, Dll4, Hes5, Hes7, Meox1, Notch1, Notch2, Notch3, Notch4 and Pax6 were used.All animals for HCR were stained once using the same hybridization, washing and amplification time for all gene targets.
Images of ISHs were taken with a Leica M165 FC stereo-microscope using 8:1 zoom.Images of HCR were taken with a Leica TCS SP8 MP confocal microscope using a 10x and 20x objective (details below).Brightness and contrast were adjusted uniformly to the entire image.

Confocal microscopy
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) or 568 nm (H2B-mCherry) or 638 nm (Sox2-iRFP) through a HC PL APO CS2 20x/0.75DRY objective.
For ex vivo culture experiments, a z-stack of 10-12 planes at 10 µm distance was scanned every 10 min.For in vitro differentiation experiments, a z-stack of 3-4 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 x 512 pixels (pixel size 1,517 µm).For imaging of immunostaining and HCR samples, a z-stack of 10-12 planes at 10 µm distance was scanned in 1024 x 1024 pixels (pixel size 0,758 µm).

Image and data processing
For quantification of mean intensity or number of positive cells in fixed sample imaging, Fiji 85 was used.Kymographs were generated as described previously 68 .For quantification of mean intensity over time, ROImanager, Mastodon or manual tracking were used.Quantification of oscillation dynamics were performed using pyBOAT 47 and Julia.For determination of main period and amplitude, sample tracks were averaged and normalised (either by mean or DMSO, as stated in the legends) using Julia.Furthermore, all graphs and heatmaps are plotted using Julia.

Statistical analysis
, S6K, L), whereas the relative amplitude of Hes1-Achilles dynamics increased (Fig. 6E, S6M).It has been shown recently that expression of the Notch ligand Dll1 begins to oscillate in the NT, adjacent to the last formed somites 11 .To test if differentiation induced by FGF inhibition might have led to changes in Dll1 expression in 2D cultures, we quantified Dll1 by immunostainings.However, we did not detect changes in Dll1 expression levels in the Sox2-positive cell population upon FGF inhibition (Fig. 6F, G, S6N).

Figure 1 .
Figure 1.Generation of a Hes1-Achilles mouse line to study Hes dynamics in the posterior embryonic tail.A Schematic representation of the posterior mouse embryonic tail.Tailbud (green) differentiate into PSM (blue), which gives rise to somites (light blue), and the pre-NT (yellow), which differentiates into the NT (light yellow).B Representative image of a E10.5 embryonic tail stained by HCR for the mesodermal marker Meox1 and neural marker Pax6.Scale bar is 100 µm.C 5 µm section of an ISH of an E10.5 embryonic tail for Hes1 (as shown in Fig. S1F).Scale bar is 100 µm.D Schematic representation of the tagging strategy for the Hes1-Achilles mouse line.Hes1 was endogenously tagged with linker-Achilles.E Representative image of a mouse ESC colony expressing the Hes1-Achilles construct.Note the variable expression in the different cell types.Scale bar is 10 µm.F, G Representative images of ISH of an E10.5 embryo for Hes1 (F) (as shown in Fig. S1F) and Achilles (G).Right panel shows magnification of the posterior embryonic tail.Arrows highlight the expression

Figure S2 .
Figure S2.Characterization of Hes1-Achilles reporter mouse line.A, B Genotyping for positive clones using primers (forward (FW) and reverse (RV)) against Achilles and against a product that crosses the homology arm (HA).In A a representative gel and in B a scheme of the expected gene locus are shown.C Integration of exactly one copy of Achilles was confirmed by qPCR (heterozygous (Het) and homozygous (Hom)).D Example of sequencing result of a correctly integrated clone.E Chimeric mice were generated by blastocyst injection.F, G Genotyping of positive mice is performed using primers to detect integration wildtype (WT) and mutated alleles.In F representative gels and in G a scheme of the expected gene locus are shown.H, I Representative images of E10.5 wildtype (WT) (H) and Hes1-Achilles homozygous embryos (I) are represented.Scale bar is 500 µm.

Figure 2 .
Figure 2. Inhibition of Notch signalling results in a decrease in Hes1-Achilles levels in ex vivo cultured E10.5 embryonic tails.A-C Comparison of different culture media to optimize for the survival of the pre-NT ex vivo.Embryonic tails were cultured ex vivo for 18 h in embryo culture medium (ECM) or neurobasal (NB) medium and compared to uncultured (in vivo) E11.5 embryos.Immunostaining against cleaved-caspase 3 (C-Caspase 3, apoptosis) phospho-histone H3 (P-histone H3, mitosis) and counterstaining with Phalloidin was performed.In A representative images are shown.In B the intensity of cleaved caspase 3 signal in the embryonic tail and in C the number of mitotic cells in the pre-NT were quantified.D, E E10.5 embryonic tails were cultured ex vivo and incubated with DMSO control or the gamma-secretase inhibitor DAPT.Hes1-Achilles intensity was measured over time.In D representative images and in E the quantification of Hes1-Achilles intensity in the embryonic tail are shown.Scale bar is 100 µm.Dots in boxplots represent individual data points.* is p<0.05,** is p<0.01,*** is p<0.001,**** is p<0.0001.

Figure S3 .
Figure S3.Differential effect of signalling perturbation on ex vivo cultured E10.5 embryonic tails.A-D Embryonic tails were cultured ex vivo in either ECM or NB medium.Intensity of the dynamic signalling reporters LuVeLu was measured.A Kymographs generated along the PSM from the posterior (P) to the anterior (A) of a growing embryonic tail cultured in ECM (left panel) or NB medium (right panel) were generated.Note the wave dynamics visible in both.B Heatmap of samples illustrating signalling dynamics in the different culture conditions using LuVeLu as reporter line.C, D Quantification of period (C) and amplitude (D) of the signalling dynamics shown in B. ECM n=8 tails, NB n=7 tails E-M Effect of signalling inhibition on ex vivo cultured embryonic tails.E Schematic representation of embryonic tail highlighting the gradients of Wnt and FGF along the tissue.F, G E10.5 embryonic tails were

Figure 3 .K
Figure 3. Pre-NT and tailbud region in ex vivo cultured E10.5 embryonic tails show low amplitude oscillations at population level.A Embryonic tails were cultured ex vivo and Hes1-Achilles intensity measured by fluorescence real-time imaging (corresponding to correctly positioned control samples from Figure 2 and S3).Regions of interest were selected manually and interpolated (IP) between timepoints.Hes1-Achilles dynamics were then

Figure 4 .
Figure 4.Both PSM and pre-NT cells show Hes1 oscillations at single-cell level.A Sparse labelling with H2B-mCherry was induced by injection of 1 mg Tamoxifen (TAM).Embryonic tail tips were cultured ex vivo on fibronectin-coated dishes.Hes1-Achilles and H2B-mCherry intensity were measured by fluorescence real-time imaging.Single cells were tracked semi-automatically to quantify Hes1-Achilles dynamics.B, C Quantification of Hes1-Achilles dynamics in PSM cells.Representative snapshots are shown in B. Representative timeseries data (mean-normalised) of a single PSM cell (blue) is shown in C. Differentiated PSM cell is indicated by light blue colour.D, E Quantification of Hes1-Achilles dynamics in pre-NT cells.Representative snapshots are shown in D. Representative timeseries data

Figure S4 .
Figure S4.Characterization of 2D ex vivo cultures of E10.5 embryonic tail tips.A, B Cultures were fixed after 18 h and stained.A Representative image of immunostaining for PSM marker Tbx6 and neural marker Sox2.B Representative image of HCR for Meox1 and counterstaining for nuclei (DAPI).Scale bar is 100 µm.C, D Periods were quantified using wavelet transform or Fourier transform for PSM cells (C) or pre-NT cells (D).In E periods were quantified using Fourier transform for PSM cells and pre-NT cells.The coefficient of variation is 0.1 for PSM cells and 0.2 for pre-NT cells.* is p<0.05,** is p<0.01,*** is p<0.001,**** is p<0.0001.PSM cells: n=39; pre-NT cells: n=40.

Figure 5 .
Figure 5. Hes1 shows noisy oscillations in in vitro differentiated NMPs.A Mouse ESCs were differentiated in vitro to NMPs.Hes1-Achilles intensity was measured by fluorescence real-time imaging from 56 -76 h of differentiation.B-E Quantification of Hes1-Achilles dynamics in whole NMP colonies (n=8).Representative snapshots are shown in B. Timeseries data (mean-normalised) of representative NMP colony is shown in C. D, E Quantification of the period (D) and amplitude (E) by wavelet transform.Scale bar 100 µm.F-I Quantification of Hes1-Achilles dynamics in single NMP cells (n=46).Single cells were tracked manually to quantify Hes1-Achilles dynamics.Representative snapshots are shown in F. Representative timeseries data (mean-normalised) of NMP cells is shown in G. H, I Quantification of the period (H) and amplitude (I) by wavelet transform.Scale bar 50 µm.Note that data for control corresponds to control data in Figure S5.

Figure S5 .
Figure S5.Population-wide Hes1 dynamics in in vitro differentiated cells.A Mouse ESCs expressing Hes1-Achilles and Sox2-iRFP were differentiated in vitro to NMPs.Cells were fixed at the indicated timepoints and immunostaining against T was performed.Representative images are shown.B-E Quantification of Hes1-Achilles dynamics in whole colonies.Representative snapshots are shown in B. In vitro differentiated NMPs were incubated with DMSO control or the gamma-secretase inhibitor DAPT.Hes1-Achilles intensity was measured over time.In C individual tracks (dotted lines), mean (solid line) and standard deviation (area)

Figure S6 .
Figure S6.FGF inhibition leads to an increase in Hes1 oscillation amplitude in pre-NT cells.Further analysis of data shown in Figure 6.A Representative kymographs for PSM region upon DMSO and FGF inhibition (SU5402) are shown.Hes1-Achilles wave is indicated

Figure 7 .
Figure 7. Notch inhibition has a differential effect on Hes1 oscillations in PSM and pre-NT cells.A Representative image of immunostaining of an E10.5 embryonic tail for Dll1, Dll4 and the neural marker Sox2.Scale bar of top panel is 100 µm and bottom panel is 50 µm.B Schematic illustrating that Dll1 oscillates at high levels in the PSM 11 , whereas Dll1 levels are