Strength of interactions in the Notch gene regulatory network determines patterning and fate in the notochord

Version of Record

Accepted for publication after peer review and revision.

Download
Cite
Share
CommentOpen annotations (there are currently 0 annotations on this page).

822 views

Version of Record published: June 6, 2022 (This version)
Accepted: April 28, 2022
Received: November 11, 2021
Preprint posted: March 25, 2021 (Go to version)

1. Of interest
The Slingshot phosphatase 2 is required for acrosome biogenesis during spermatogenesis in mice

Ke Xu, Xianwei Su ... Hongbin Liu

Research Article Updated Mar 31, 2023
Further reading

Article
Figures and data
Abstract
Editor's evaluation
Introduction
Results
Discussion
Materials and methods
Appendix 1
Data availability
References
Decision letter
Author response
Article and author information
Metrics

Abstract

Development of multicellular organisms requires the generation of gene expression patterns that determines cell fate and organ shape. Groups of genetic interactions known as Gene Regulatory Networks (GRNs) play a key role in the generation of such patterns. However, how the topology and parameters of GRNs determine patterning in vivo remains unclear due to the complexity of most experimental systems. To address this, we use the zebrafish notochord, an organ where coin-shaped precursor cells are initially arranged in a simple unidimensional geometry. These cells then differentiate into vacuolated and sheath cells. Using newly developed transgenic tools together with in vivo imaging, we identify jag1a and her6/her9 as the main components of a Notch GRN that generates a lateral inhibition pattern and determines cell fate. Making use of this experimental system and mathematical modeling we show that lateral inhibition patterning is promoted when ligand-receptor interactions are stronger within the same cell than in neighboring cells. Altogether, we establish the zebrafish notochord as an experimental system to study pattern generation, and identify and characterize how the properties of GRNs determine self-organization of gene patterning and cell fate.

Editor's evaluation

This manuscript presents computational and experimental results to study lateral inhibition patterning in the zebrafish notochord, identifying Jag1a as a crucial ligand and marker for vacuolated cell fate whereas her6 and her9 repress Jag1a. The results are complemented with numerical simulations of lateral induction and lateral inhibition circuits in one-dimensional arrays, together with linear stability analysis. The work is very well done and makes an important contribution to the understanding of notochord development.

https://doi.org/10.7554/eLife.75429.sa0

Introduction

Most of the information necessary to build an organism resides in its genome. The co-regulation of subsets of genes form gene regulatory networks (GRNs) that generate patterns of expression, which ultimately regulate cell fate and organ shape. Different types of GRNs regulate different patterning events. For example, some GRNs work in combination with gradients of morphogens to generate patterns at the embryo or organ scale (Briscoe and Small, 2015). In contrast, other GRNs coordinate short-range interactions, generating self-organized patterns of gene expression at the cellular scale (Schweisguth and Corson, 2019; Sjöqvist and Andersson, 2019). Understanding how different GRN topologies and the strength of their interactions regulate the generation of gene expression patterns constitutes a key challenge in developmental biology. However, research in this direction has been hindered by limited experimental systems that can be accurately modeled mathematically.

GRNs controlling short-range interactions produce diverse patterning events, such as lateral inhibition and lateral induction. Lateral inhibition involves a group of cells actively suppressing the expression of some genes in adjacent cells, thereby inducing them to adopt a different cell fate. In contrast, lateral induction involves cells inducing adjacent cells to adopt the same cell fate. Lateral inhibition and lateral induction patterns are two of the main patterns generated by Notch GRNs: one of the most representative signaling pathways that mediates local communication between cells. The Notch pathway is evolutionarily conserved and generates gene expression patterns that regulate cell fate decisions in a wide variety of organs (Apelqvist et al., 1999; Artavanis-Tsakonas et al., 1999; VanDussen et al., 2012; Wibowo et al., 2011). Signaling is triggered by interaction of a Notch receptor with a Notch ligand. Once they bind, the Notch intracellular domain (NICD) is cleaved inside the signal receiving cell, and in complex with Rbp-Jκ and MAML, translocates to the nucleus, where it activates Notch target genes (Bray, 2016).

The generation of either lateral inhibition or lateral induction patterns downstream of Notch has thus far been associated with different ligands. Lateral inhibition patterning has been described for the Delta-like (Dll) ligands and for Jag2 (Heitzler and Simpson, 1991; Lanford et al., 1999) and generally occurs when Notch signaling activates the expression of a transcriptional repressor of the HES family that in turn inhibits the expression of the ligand in adjacent cells, preventing them from adopting the same cell fate (Simpson, 1990; Sjöqvist and Andersson, 2019; Sternberg, 1988). Mathematical simulations have shown that a lateral inhibition GRN can amplify small levels of noise in gene expression, leading to bi-stability and the generation of alternating patterns (Collier et al., 1996). Lateral induction has been shown for the ligand Jag1, whereby Notch activation by Jag1 triggers the expression of the same ligand in the adjacent cells, promoting the same fate (Hartman et al., 2010; Manderfield et al., 2012; Neves et al., 2011). It remains unknown whether lateral inhibition and lateral induction GRNs are restricted to specific ligands, or whether a given ligand can generate different patterns depending on the cellular and signaling context.

Other important parameters in a GRN are the nature and affinities of the ligand-receptor interactions. In the case of Notch, ligands can also interact with receptors in the same cell (Celis de and Bray, 1997; Klein et al., 1997; Micchelli et al., 1997). This interaction, known as cis-inhibition, mutually inactivates both the ligand and receptor, and mathematical models have shown that it is required for patterning in the absence of cooperative interactions (Formosa-Jordan and Ibañes, 2014; Sprinzak et al., 2010; Sprinzak et al., 2011). Different ligands and receptors bind to each other in cis and trans with different affinities, and these affinities can be modulated by posttranslational modifications (Bray, 2016; Sjöqvist and Andersson, 2019). Altogether, these properties increase the complexity and diversity of Notch GRNs. For this reason, understanding how the topology and interaction parameters of these GRNs lead to pattern generation requires a combination of mathematical models and experimental systems that allow in vivo visualization and perturbation of Notch signaling components.

The notochord constitutes an underappreciated system that is ideal for studying the generation of Notch patterns. Initially, notochord coin-shaped precursor cells are arranged unidimensionally. These simple and well-defined cell-cell contacts greatly facilitate mathematical modeling and theoretical analysis, making it valuable for studying the relationship between GRNs parameters and patterns. In vertebrates, such as zebrafish, notochord precursors give rise to two different cell types (Dale and Topczewski, 2011): vacuolated cells, located in the inner part of the organ, that contain a large vacuole that provides hydrostatic pressure (Adams et al., 1990; Bagwell et al., 2020; Ellis et al., 2013), and sheath cells, which form the surface of the cylindrical structure (Dale and Topczewski, 2011; Grotmol et al., 2003; Figure 1A). The cell fate decision between vacuolated and sheath cells depends on Notch signaling (Yamamoto et al., 2010). Inhibition of the Notch ligands jag1a and jag1b by morpholino (MO) injection leads to an excess of vacuolated cells, while overexpression of NICD promotes sheath cell fate (Yamamoto et al., 2010). However, most of the components and topology of the GRN that coordinates cell fate in the notochord remain unknown.

Figure 1 with 1 supplement see all

Download asset Open asset

*Jag1a* generates a lateral inhibition pattern.

(A) Schematic representation of notochord development. At 18–19 hpf most of the notochord is composed of coin-shaped precursor cells. During the following 8 hours, progressively, in an antero-posterior order, coin-shaped precursor cells begin their differentiation into sheath cells and vacuolated cells. (B) Airyscan confocal section of a zebrafish notochord at 19 hpf using the *rcn3*:lyn-mNeonGreen transgenic line. (C) Transmission electron microscopy of a zebrafish notochord at 19 hpf. (D) Magnification of boxed area in (C). (E) (Top) Schematic representation of the model for a Lateral Induction Network shows a pair of cells where the ligand in one cell activates NICD release in the neighboring cell. NICD activates ligand expression in its own cell. (Bottom) Representative simulation of this network applied to an array of cells unidimensionally arranged. (F) (Top) Schematic representation of the model for a Lateral Inhibition Network shows a pair of cells where the ligand in one cell activates NICD release in the neighboring cell. NICD activates the expression of the repressor, which in turn inhibits ligand expression. (Bottom) Simulation of this network applied to an array of cells unidimensionally arranged. (**G–L**) Maximal intensity projection of Airyscan confocal sections of a zebrafish tail at 22 hpf. (**J–L**) Magnification of boxed area in (G). n = 10 fish. (L) Intensity profile across a horizontal line in panel (L). (M) *jag1a*:mScarlet and tp1:GFP expression levels across a 1 μm thick horizontal line on a single plane of the image shown in J. Scale bars, 1 μm (D) 10 μm (**B, C, J**), 100 μm (G).

Here, we exploit the in vivo imaging and genetic manipulations that the zebrafish model offers to quantitatively study the generation of Notch patterns. We establish the zebrafish notochord as the first unidimensional system to study lateral inhibition patterning. Using this experimental model, we show that jag1a generates a lateral inhibition pattern, a possibility thought to be restricted to the other Notch ligands (Boareto, 2020; Boareto et al., 2015; Sjöqvist and Andersson, 2019). Using a combination of single-cell RNA-Seq analysis and genetic perturbations, we identify her6/her9 and jag1a as the key genes that promote sheath and vacuolated fate. Our computational modeling further reveals that a stronger cis- than trans-inhibition promotes the generation of lateral inhibition patterns. We experimentally validate the role of cis-inhibition in our GRN, finding that jag1a is sufficient to disrupt the expression of Notch-target genes in the cells where it is expressed. Altogether, our results describe and characterize a novel Notch GRN that generates lateral inhibition patterns and determines cell fate.

Results

Jag1a and Notch activity show a lateral inhibition pattern in the zebrafish notochord

Notch signaling generates patterns of gene expression by signaling at cell-cell contacts (Bray, 2006; Shaya et al., 2017). Thus, a prerequisite for the study of Notch patterning in the notochord is the characterization of cell-cell contacts. To describe the contacts between cells, we generated an rcn3:lyn-mNeonGreen transgenic line that labels the plasma membrane of all notochord cells. We observed that notochord precursor cells are coin-shaped and unidimensionally arranged one cell after another (Figure 1B). Using transmission electron microscopy, we confirmed this cell arrangement and observed that coin-shaped notochord cells are isolated from the rest of the tissues by a layer of extracellular matrix (Figure 1C–D). Thus, the contacts of each notochord cell are restricted to the two neighboring cells in the stack. This unidimensional geometry with very well-defined cell-cell contacts makes the notochord an ideal system to study Notch patterning.

Whether Notch signaling generates gene expression patterns in the notochord remains unknown. To understand the expression patterns that may be generated in this organ, we modeled lateral induction and lateral inhibition networks in the unidimensional arrangement of notochord cells. We first modeled a lateral induction network as a two component GRN, where the Notch ligand induces NICD cleavage in the adjacent cells, and NICD in turn induces ligand expression in the cells where it is located. This network gives rise to a homogeneous pattern, where all the cells have both high concentrations of NICD and ligand (Figure 1E and Figure 1—figure supplement 1A, Matsuda et al., 2012; Petrovic et al., 2014). Next, we modeled a lateral inhibition network (Collier et al., 1996). Here, the ligand also induces NICD cleavage in the adjacent cells, but in this case, NICD induces the expression of a repressor that in turn inhibits ligand expression. The result of this model is a NICD-ligand alternating pattern (Figure 1F and Figure 1—figure supplement 1B). These results are in agreement with previous models of lateral induction and lateral inhibition (Collier et al., 1996; Matsuda et al., 2012; Petrovic et al., 2014).

Then, we experimentally evaluated whether one of these two patterns was present in the notochord. The two zebrafish homologs of the mammalian Jag1 – jag1a and jag1b – are the main Notch ligands in the notochord (Yamamoto et al., 2010). Although both jag1 ligands show a non-homogeneous expression pattern, the jag1a one is sharper and can be observed in more immature cells – more posteriorly in the notochord – (Figure 1—figure supplement 1C-F). For this reason, and to explore Notch patterns in high resolution, we generated a stable jag1a:mScarlet BAC transgenic line that recapitulates the endogenous jag1a mRNA expression (Figure 1—figure supplement 1C-E), and crossed it to the tp1:GFP transgenic line (Parsons et al., 2009). The tp1 promoter includes 12 Rbp-Jκ binding sites derived from a viral sequence, making the tp1:GFP line a reporter of Notch activity. Interestingly, we found an alternating pattern (Figure 1G–M, Figure 1—figure supplement 1G, H) that resembles lateral inhibition, a pattern that has never been described for Jag1.

To verify that the observed pattern is generated by lateral inhibition, we injected previously validated (Yamamoto et al., 2010) splicing-jag1a and atg-jag1b MOs into the tp1:GFP;jag1a:mScarlet double transgenic line. By using a splicing-jag1a MO we specifically inhibited endogenous jag1a genes but not our jag1a:mScarlet reporter. First, we observed that when we injected the two MOs simultaneously, the tp1:GFP signal almost completely disappeared in the notochord, but not in the neighboring tissues (Figure 2D), supporting the hypothesis that jag1a and jag1b are the main, if not the only, Notch ligands expressed in the notochord. We also observed an increase in the number of jag1a:mScarlet-positive cells that are directly adjacent to other jag1a:mScarlet-positive cells, suggesting that a lateral inhibition process is disrupted upon inhibition of jag1a and jag1b. This effect was also observed, although to a lower extent, when injecting the jag1a or jag1b MOs separately, indicating that jag1a and jag1b have similar, but not completely redundant roles in the generation of the lateral inhibition pattern (Figure 2A–E).

Figure 2 with 4 supplements see all

Download asset Open asset

*Jag1a*/*jag1b* inhibition disrupts lateral inhibition pattern, and this pattern correlates with fate.

(**A–D**) *jag1a*:mScarlet tp1:GFP 21 hpf fish embryos injected with control MO (A), *jag1a* + control MO (B), *jag1b* + control MO (C), or control MO (D), together with lyn-miRFP mRNA to visualize membranes. (E) Quantification of the average number of *jag1a*-positive cells directly adjacent to each *jag1a*-positive cell. Two-tailed p-value is shown in the plot. (F) Cell fate of cells expressing *jag1a* or the tp1:GFP at the coin-shape stage. Quantifications from images as shown in G (standard deviation *jag1a* = 2.696, tp1 = 2.631; n = 5 fish). (G) Time lapse of optical sections of notochord cells using the tp1:GFP; *jag1a*:mScarlet double transgenic line. First frame corresponds to 24 hpf. Scale bars, 20μm.

Together, our results show that Jag1 is not restricted to the generation of lateral induction patterns as previously thought, but can also generate lateral inhibition patterns.

Jag1a and Notch activity are early markers of notochord cell fate

Finding early markers of differentiation is important to understand cell fate decisions. However, no early marker of notochord cell differentiation has been reported to date. Having identified an alternating tp1-jag1a pattern, we evaluated whether it is associated with vacuolated and sheath cell fates. To test this, we used the tp1:GFP;jag1a:mScarlet double transgenic reporter, and followed notochord cells by time lapse in vivo imaging (Figure 2G and Figure 2—video 1). We found that jag1a-positive cells gave rise to vacuolated cells, while tp1-positive cells differentiated into sheath cells (Figure 2F). Interestingly, at the end of the movie, most of the vacuolated cells are labeled with jag1a:mScarlet, while there are some non-labeled cells at the notochord surface. This suggests that the non-labeled cells at the disc-shape stage are Notch active and will differentiate into sheath cells, but their Notch activity is not strong enough to activate the non-endogenous tp1 promoter.

Having identified jag1a is an early marker for cell fate, we decided to verify if cell fate is determined by a lateral inhibition process. An important characteristic of lateral inhibition is that the cell expressing the ligand prevents the neighboring cells to acquire the same cell fate. To evaluate if this is the case in the notochord, we quantified how often two consecutive coin-shaped cells acquire vacuolated cell fate. To do this, we developed a feedback microscopy pipeline that allowed us to image the notochord cells in high quality over time, even though the fish was simultaneously elongating (Figure 2—figure supplement 1 and Figure 2—videos 2; 3). We found that none of the future vacuolated cells were adjacent to another future vacuolated cell at the disc cell stage (n = 0/51 cells quantified from 4 fish). In contrast, future sheath cells almost always had another future sheath cell next to them (n = 221/222 cells quantified from 4 fish).

Altogether, these results establish jag1a and Notch activity as the first available markers of vacuolated and sheath cell fates, and confirmed that this cell fate decision is mediated by a lateral inhibition process.

her9 and her6 have a complementary expression pattern to jag1a

Having identified that the jag1a-Notch alternating pattern correlates with fate, we aimed to identify which are the components of the GRN that make this pattern possible. Notch lateral inhibition model predicts the presence of a Notch target gene that represses jag1a expression. This gene should have a mutually exclusive pattern with jag1a.

The bHLH genes of the HES/HEY families are good candidates as they are transcriptional repressors often activated by Notch signaling (Kageyama et al., 2007). In the notochord, her9 has been shown to be a Notch target gene (Yamamoto et al., 2010). However, the fact that no notochord phenotype was found for the her9 knockdown zebrafish (Yamamoto et al., 2010) suggests functional redundancy with other genes. To identify in an unbiased manner all the HES/HEY genes that repress jag1a, we analyzed single-cell RNA-Seq data (Wagner et al., 2018). We found that her6 and her9 are the most highly expressed genes of this family in the notochord at 18 and 24 hours post-fertilization (hpf) (Figure 3A, Figure 3—figure supplement 1A-F). To evaluate their expression pattern, we analyzed mRNA expression by fluorescent in situ hybridization based on a hybridization chain reaction (HCR). her6 and her9 were expressed in an alternating pattern with jag1a (Figure 3B–O). Importantly, in the her6/her9 HCR mRNA staining, we did not observe unlabeled cells, as was the case with tp1, highlighting the importance of identifying endogenous Notch target genes. In contrast to her6 and her9 expression, her12, which was expressed at a much lower level according to the RNA-Seq, was not detected in the notochord by HCR (Figure 3—figure supplement 1G-M). The observed alternating patterns suggest that her6 and her9 could repress jag1a expression in the notochord.

Figure 3 with 2 supplements see all

Download asset Open asset

*her9* and *her6* show a complementary pattern to *jag1a*.

(A) Heatmap showing the expression levels of the zebrafish HES/HEY family genes. Values represent average normalized UMIs in all notochord cells at 18 and 24 hpf. (B) Projection of confocal optical sections of 18 hpf zebrafish stained with in situ HCR probes against *her6* (green) and *jag1a* (magenta). Transmitted light is shown in gray scale. (C) Maximal projection of confocal Airyscan optical sections of the boxed area in (B). (D), Intensity profile of *her6* (green) and *jag1a* (magenta) along a 1 μm thick horizontal line on the in situ HCR shown in (C). (**E–G**) Magnified views of boxed area in (C), n = 8. (H) Scatter plot of the intensities shown in D. Each point represents *her6* and *jag1a* intensity in a 1-pixel width times 1 um height square. Pearson correlation and p-value of the correlation is shown in the plot. (**I–O**) Analogous images to (**B–H**) based on the *her9* probe instead of *her6* probe, n = 9. Scale bars, 50 μm (**B, I**), 20 μm (**C, E, J, L**).

To analyze if her6 and her9 could be direct targets of Notch signaling, we analyzed Rbp-Jκ binding sites in a recently published zebrafish CUT & RUN experiment (Ye et al., 2021). Several Rbp-Jκ binding sites were identified in the proximity of her6 and her9 transcription start sites, supporting the hypothesis that these genes are direct Notch targets (Figure 3—figure supplement 1N, Ye et al., 2021).

Aside from the ligand and repressor, the other main component of a lateral inhibition Notch GRN is the Notch receptor. By single-cell RNA-Seq data analysis (Wagner et al., 2018) we found that notch2 was detected in most cells at the highest levels at 18 and 24 hpf (Figure 3—figure supplement 2A-E). notch2 notochord expression was confirmed by fluorescent HCR (Figure 3—figure supplement 2F–G). Altogether, we identified the main components of the lateral inhibition GRN, finding her6 and her9 as candidate genes to repress jag1a expression, and notch2 as the main Notch receptor in the notochord.

her6 and her9 inhibit jag1a expression

To directly assess if her6 and her9 are sufficient to inhibit jag1a expression, we established notochord-specific genetic mosaics. To that end, we aimed at identifying a highly specific notochord promoter to overexpress her6 or her9, while simultaneously labeling the perturbed cells. Making use of the single-cell RNA-Seq dataset (Wagner et al., 2018), we identified emilin3a as the gene that offers the best balance between notochord specificity and high expression levels (Figure 4—figure supplement 1A,B). We cloned a 5 kb promoter upstream of the coding region and showed that it is sufficient to drive gene expression in the notochord, including most of both jag1a:mNeonGreen- and tp1:GFP cells (Figure 4—figure supplement 1C-J). Next, we used this promoter and the p2a system (Kim et al., 2011b) to generate her6 or her9 gain-of-function cells concomitantly with GFP expression, or only-GFP as a control. For each of these constructs, we quantified the level of jag1a:mScarlet expression in the GFP-p2a-her6, GFP-p2a-her9 or only-GFP positive cells in comparison to the rest of the notochord. We found that GFP-p2a-her6 and GFP-p2a-her9 cells had a lower level of jag1a:mScarlet than only-GFP cells, indicating that her6 and her9 repress jag1a expression in a cell autonomous manner (Figure 4A–G). This result was confirmed by quantifying endogenous jag1a mRNA expression by fluorescent HCR (Figure 4—figure supplement 2A-G).

Figure 4 with 2 supplements see all

Download asset Open asset

*her6* and *her9* inhibit *jag1a* expression.

(**A – E**) Airyscan confocal optical sections of live 22 hpf transgenic *jag1a*:mScarlet zebrafish injected with *emilin3a*:GFP (**A and B**), *emilin3a*:GFP-p2a-*her6* (**C and D**) or *emilin3a*:GFP-p2a-*her9* (**E and F**). DNA constructs were injected at the one-cell stage together with I-SceI protein. (**B, D, F**) show the boundary of GFP segmentation in A, C, and E, respectively, and manual outline of the notochord. (G), Quantification of *jag1a*:mScarlet intensity inside GFP-positive cells segmented as exemplified in (**B, D, F**). Values in the plot represent the intensity of *jag1a*:mScarlet inside segmented cells divided by the *jag1a*:mScarlet intensity inside the notochord outside of the segmented cells. Each point represents an individual fish (n = 12 GFP, n = 7 GFP-p2a-*her6*, n = 11 GFP-p2a-*her9*). Two-tailed p-values are shown in the plot. (H), Airyscan confocal sections of embryo at 22 hpf injected with Cas9 together with a control guide (H) or Cas9 together with *her6*/*her9* gRNAs (I). (J) Quantification of the average number of *jag1a*-positive cells directly adjacent to each *jag1a*-positive cell. For each individual fish, we count how many *jag1a*-positive cells are adjacent to each *jag1a*-postive cell, and then calculate the average for that fish. This value would be equal to 2 in case all the cells are *jag1a*-positive, and zero if no *jag1a*-positive cell is adjacent to another *jag1a*-positive cell. Each individual point in the plot represents the average value for an independent fish (n = 7 control, n = 8 *her6*/*her9* KOs). Two-tailed p-value is shown in the plot. Scale bars, 20 μm.

Having identified her6 and her9 as genes sufficient to inhibit jag1a expression, we studied if these genes are necessary for lateral inhibition patterning in the notochord. To this end, we generated her6/her9 double transient knockouts (Figure 3—figure supplement 1N) in a jag1a:mScarlet;rcn3:lyn-mNeonGreen background, and quantified the number of jag1a-positive cells that are found adjacent to each jag1a-positive cell. We found this value to be increased upon her6 and her9 gene deletion, showing that her6 and her9 are necessary for lateral inhibition (Figure 4H–J). Altogether, we show that her6 and her9 are the repressors in the GRN that generate a lateral inhibition pattern in the notochord.

her6/her9 and jag1a determine notochord cell fate

To test if the identified GRN genes are sufficient to determine cell fate, we first expressed GFP-p2a-her6, GFP-p2a-her9 or only-GFP in a mosaic fashion in the notochord cells, and evaluated its effect on cell fate. At 2 days postfertilization (dpf), a stage where vacuolated and sheath cells can be distinguished, we found a higher proportion of sheath cells in GFP-p2a-her6 and GFP-p2a-her9 expressing cells. This result indicates that her6 and her9 are sufficient to determine sheath cell fate (Figure 5A–D).

Figure 5 with 1 supplement see all

Download asset Open asset

*her6*, *her9,* and *jag1a* determine cell fate in the zebrafish notochord.

(**A–C, E–F**) Confocal optical sections of 2 dpf live zebrafish that were injected with the *emilin3a*:GFP (**A, E**) *emilin3a*:GFP-p2a-*her6* (B), *emilin3a*:GFP-p2a-*her9* (C) or *emilin3a*:GFP-p2a-*jag1a* (F) constructs. DNA constructs were injected at the one-cell stage together with I-SceI protein. (**D and G**) Proportion of vacuolated cells at 2 dpf are shown. Proportion of vacuolated cells was calculated by counting the number of vacuolated cells divided by the sum of the number of sheath and vacuolated cells. Each point in D, G represents an independent fish quantified from on z-stack confocal planes (D, n = 9 GFP, n = 10 GFP-her6, n = 10 GFP-her9, G, n = 5 GFP, n = 5 GFP-jag1a). Two-tailed p-values are shown in D and G. Scale bars, 50 μm.

Next, we expressed GFP-p2a-jag1a or only-GFP. Interestingly, we found that the Notch ligand jag1a is sufficient to drive vacuolated cell fate in the same cells where it is expressed (Figure 5E–G). Taken together, our results show that not only the Notch targets her6/her9 drive cell fate, but also the Notch ligand jag1a determines cell fate in the same cell where it is expressed.

Stronger cis than trans interactions are required for lateral inhibition patterning

After observing that jag1a, a Notch ligand, drives vacuolated cell fate on the same cell where it is expressed, we next investigated the mechanism mediating this process. First, we explored a potential signaling role of the ligand intracellular domain. It has been shown that upon Notch-ligand trans-interaction, not only the NICD is cleaved in the receiver cell, but also the intracellular domain of some ligands, including Jag1, is cleaved inside the sender cell, leading to bidirectional signaling (Ikeuchi and Sisodia, 2003; Kim et al., 2011a; Kolev et al., 2005; LaVoie and Selkoe, 2003; Liebler et al., 2012; Metrich et al., 2015). The intracellular domain of jag1a (JICD) would then inhibit Notch signaling in the sender cell (Kim et al., 2011a). Thus, overexpression of the full-length ligand in our experiment would increase the amount of ligand that is available to be cleaved, leading to Notch inhibition and promoting vacuolated cell fate. To test this hypothesis, we expressed mScarlet-p2a-JICD or only-mScarlet in a mosaic fashion in notochord cells. We did not observe any effect of JICD on cell fate (Figure 5—figure supplement 1), showing that JICD signaling is not sufficient to explain the jag1a effect on fate in the notochord.

Next, we considered two different signaling circuits that could explain how jag1a can promote vacuolated cell fate in the cells where it is expressed. First, through trans-interactions with the Notch receptor, jag1a could activate Notch signaling and as a consequence, her6/her9 expression in their neighbors. Her6 and her9 would inhibit jag1a in these neighbors, and this would in turn diminish the amount of Notch signaling that the initial cell receives, promoting vacuolated cell fate. A second possible explanation comes from the observation that when Notch ligands are expressed in the same cell as the Notch receptor, they can mutually inhibit each other through cis-inhibition (Celis de and Bray, 1997; Klein et al., 1997; Micchelli et al., 1997). Thus, overexpression of jag1a would deplete the Notch receptor in a cell-autonomous manner, making this cell non-responsive to Notch signaling and thus promoting vacuolated cell fate (Figure 6A).

Figure 6 with 2 supplements see all

Download asset Open asset

Modeling and experimental results of cis- and trans-interactions in the notochord.

(A) Two possible circuits may explain the effect of *jag1a* on fate of the cell where *jag1a* is overexpressed. Circuit 1 is based on the interaction of ligand and receptor in trans. Circuit 2 is based on a possible role of cis-inhibition of the Notch receptor by the ligand. Cells where we overexpress the ligand are represented as the cell on the left. Adjacent cells are represented on the right. OE, overexpression. (**B – G**), Airyscan confocal planes of fixed 22 hpf transgenic fish injected with *emilin3a*:GFP (**B–D**) or *emilin3a*:GFP-p2a-*jag1a* (**E–G**) constructs. GFP was detected by antibody staining and *her6* and *her9* mRNA by in situ HCR in whole mount embryos. (**C and F**) show the notochord outline manually selected and the outline of GFP-positive cells automatically segmented. (**D and G**) show the outline of the manually selected notochord and the neighborhood to the GFP-positive cells. On the right side of each panel, a magnified view of the boxed region is shown. (**H, I**) Quantification of *her6* and *her9* mRNA expression after GFP-based segmentation as shown in (**C, F**) or (**D, G**), respectively. Values of *her6* and *her9* expression levels inside the segmented area inside the notochord were divided by the expression levels of the same genes in the region outside the segmented area, also inside the notochord. Each point represents a different fish. Two-tailed p-values are shown in the plots. n.s., non-significant. (J) Escape rates from the homogeneous steady state (indicated by Maximum Lyapunov Exponents, or MLE) as a function of K_cis and K_trans parameters. Positive MLE values (red) support patterning, while negative MLE values (blue) do not. Scale bars, 20 μm.

To study which of these genetic circuits is predominant in the notochord, we overexpressed jag1a-GFP or only-GFP in some notochord cells and quantified her6 and her9 expression both within the same cell and in their neighboring cells. We found only a minor or no increase in her6/her9 expression in the neighboring cells (Figure 6B, D, E, G,I), suggesting a small Notch-ligand trans-interaction. On the other hand, we observed a strong reduction of her6/her9 expression in the jag1a-expressing cells (Figure 6C, F and H). Although we cannot rule out that the small effect in the neighboring cells is due to limiting Notch receptor levels, the strong effect observed in the jag1a-expressing cells suggests the main mechanism regulating cell fate in its own cell is cis-inhibition.

It has previously been shown that cis-interactions are necessary for patterning in absence of cooperativity (Formosa-Jordan and Ibañes, 2014; Sprinzak et al., 2010; Sprinzak et al., 2011). However, how relative values of interaction in cis – within the same cell – and in trans – between neighboring cells – affect patterning has not been explored. Our experimental results suggest a key role of cis-interactions. To better understand which interactions are required for patterning, we implemented a mathematical model that includes ligand-receptor interactions both in cis and in trans based on Sprinzak et al., 2010; Figure 6—figure supplement 1. Receptor-ligand cis- and trans-interactions are represented by the K_cis and K_trans parameters, respectively. We next used this model to dissect which combinations of cis- and trans-interactions lead to the lateral inhibition pattern observed experimentally (Figure 1G–M). To do so, we evaluated the stability of the homogeneous steady state (HSS) depending on K_cis and K_trans. The HSS is defined as the steady state where all the cells have identical concentrations of Notch ligand, receptor and repressor. When the HSS is stable, the system remains in this homogenous state and no patterning occurs. HSS stability can be evaluated by performing linear stability analysis to calculate the Maximal Lyapunov Exponent (MLE), which represents the exit speed from the homogeneous steady state. Thus, a positive MLE represents an unstable HSS, and this leads to patterning. We found that in the absence of cooperativity, patterning only occurs in a region of the parameter space where K_cis is higher than K_trans (Figure 6J, Figure 6—figure supplement 2). If some degree of cooperativity is assumed, patterning is also possible without cis-interactions, as previously described (Collier et al., 1996). However, we observed that even in this case, stronger cis- than trans-interactions destabilize the homogeneous state, thus promoting patterning (Figure 6—figure supplement 2). These results of our mathematical modeling are in agreement with our experimental observations where we observe a strong cis-inhibition by jag1a.

In conclusion, our results show that a jag1a/jag1b-her6/her9 network generates a lateral inhibition pattern that determines cell fate in the notochord, and that strong ligand-receptor interactions within cells play a key role in the generation of such patterns.

Discussion

The unidimensional arrangement of cells in the zebrafish notochord, combined with its binary cell fate decisions, make it a unique model to study the properties of the Notch GRN that determines its patterning. One of the most important genetic interactions in a Notch GRN is how the expression of the ligands is regulated by Notch signaling. Previously, it was generally accepted that Notch signaling activates Jag1 expression leading to lateral induction patterns (Boareto, 2020; Boareto et al., 2015; Sjöqvist and Andersson, 2019). Here we show that Notch signaling, through the activation of the transcriptional repressors her6 and her9, inhibits jag1a expression in the notochord, leading to the generation of lateral inhibition patterns. Importantly, Jag1 is expressed in many other tissues apart from the notochord, including heart, inner ear, muscle, and kidney (D’Amato et al., 2016; Leimeister et al., 2003; Lindsell et al., 1995; Murata et al., 2006), suggesting that the identified GRN may be relevant for pattern generation in these other contexts.

Another key part of a Notch GRN that may affect patterning, is whether upon ligand-receptor interaction, there is unidirectional or bidirectional signaling. In the bidirectional signaling situation, not only the cell expressing the receptor would receive a signal, but also the cell expressing the ligand. This signal would be mediated by the intracellular domain (ICD) of the ligand. However, the role of ligand ICDs remains unclear. Previous work showed that the ICD of JAG1 and DLL1 modulate cell differentiation, proliferation, and Notch signaling (Ikeuchi and Sisodia, 2003; Kim et al., 2011a; Kolev et al., 2005; LaVoie and Selkoe, 2003; Metrich et al., 2015). In contrast, other studies found little or no effect of DLL1-ICD, DLL4-ICD, and JAG1-ICD on gene expression and migration in endothelial cells (Liebler et al., 2012). In agreement with the latter, we found no role of the zebrafish jag1a-ICD on cell fate. Further research will be needed to elucidate if the role of ligand ICDs depends on the signaling context, and whether different cell types respond differently to ICDs.

Patterning not only depends on the topology of a GRN, but also on the strength of each of the interactions. Here, using mathematical simulations supported by experimental results, we shed light on which combinations of parameters promote pattern generation. Specifically, we find that a stronger Notch-ligand interaction in cis than in trans is key for pattern generation. Importantly, this does not mean that trans-interactions are not needed. In absence of such interactions, there would be no communication between cells and thus no lateral inhibition patterning.

The strength and signaling efficiency of cis- and trans-interactions in Notch GRNs depend on the specific ligand-receptor pair (Benedito et al., 2009; Luca et al., 2015; Petrovic et al., 2014; Sjöqvist and Andersson, 2019). Some DLLs, such as DLL4, activate Notch signaling in trans more strongly than Jagged ligands (Benedito et al., 2009). On the other hand, the Drosophila homolog of Jagged genes, serrate, inhibits Notch receptors in cis more efficiently than Delta ligands (de Celis and Bray, 2000; del Álamo et al., 2011; Klein et al., 1997; Li and Baker, 2004). The possibilities of imaging and genetic manipulation that the zebrafish offers, together with the unique cell-cell contacts in the notochord, will make this organ a very valuable in vivo system to evaluate the properties not only of endogenous ligands, but also other Notch ligands, to better understand how cis and trans parameters determine pattern generation.

Our results not only explain how Notch drives pattern generation, but also how cell fate is determined during notochord development. We identified Notch activity and its downstream genes her6 and her9 as key determinants of sheath cell fate in the notochord. In some tissues, including skeletal muscle, intestine and neural systems, a higher Notch activity is related to stemness, while a lower Notch activity is related to differentiation (Blanpain et al., 2006; Fre et al., 2005; Imayoshi et al., 2010; Schuster-Gossler et al., 2007; Vasyutina et al., 2007). This raises the interesting hypothesis of whether sheath cells can be considered as only partially differentiated notochord cells. In agreement with this concept is the recent finding that upon vacuolated cell damage, sheath cells develop vacuoles and partially restore notochord structure (Garcia et al., 2017; Lopez-Baez et al., 2018). However, a possible role of Notch signaling during notochord regeneration is yet to be tested.

Several pieces of evidence suggest that the GRN that we have identified is not exclusive to zebrafish. Previous studies based on BAC transgenesis showed that Hes1, the mammalian homolog of her6 and her9, is expressed in the mouse notochord, suggesting it may play a role in the patterning of the mammalian notochord (Klinck et al., 2011). Problems in notochord development have been associated with defects in spine morphogenesis (Bagwell et al., 2020; Gray et al., 2014; Gray et al., 2021; Sun et al., 2020). Interestingly, mutations in JAG1 and NOTCH2 (McDaniell et al., 2006; Oda et al., 1997), the human homologs of the main ligands and receptor in the zebrafish notochord, lead to vertebrae malformations in human Alagille Syndrome. This suggests that spine problems in this human syndrome may be the result of defective Notch patterning during notochord development. Thus, in this study, we describe a GRN that is likely conserved across vertebrates, opening the door to better understand how mutations in JAG1 or NOTCH2 lead to the problems observed in the human disease.

In non-vertebrate chordates such as ascidians, a single cell type performs the two main functions of both sheath cells and vacuolated cells: covering the surface and producing the fluid (Deng et al., 2013; Dong et al., 2009). From an evolutionary perspective, it is plausible that Notch signaling was involved in dividing these possible ancestral functions into two different cell types. We speculate that Notch- or Hes-responsive enhancers were co-opted during vertebrate evolution to control the expression of the key genes necessary for vacuolated and sheath cell functions, making the specialization of the two different cell types possible. Given how frequently Notch signaling determines cell fate across development, Notch could represent a general mechanism that facilitated division of functions between different cells, promoting the evolution of new cell types.

Altogether, we have established the notochord as a new model system to study the principles that determine pattern generation. Using a combination of mathematical modeling, single-cell RNA-Seq analysis and genetic perturbation approaches, we identified jag1a, her6, her9 and notch2 as the key genes that determine cell fate and patterning. We expect that the GRN properties identified in this study will help understand the principles underlying patterning and cell fate decisions across multicellular organisms.

Materials and methods

Key resources table

Reagent type (species) or resource	Designation	Source or reference	Identifiers	Additional information
Antibody	GFP-Booster Alexa Fluor 488. anti-GFP Alpaca/recombinant V_HH domain, monoclonal.	Chromotek	gb2AF488	Used at (1:500) dilution.
Genetic reagent (Danio rerio)	jag1a:mScarlet	This paper		“Animal handling and generation of transgenic lines” Methods section
Genetic reagent (Danio rerio)	jag1a:mNeonGreen	This paper		“Animal handling and generation of transgenic lines” Methods section
Genetic reagent (Danio rerio)	emilin3a:mScarlet	This paper		“Animal handling and generation of transgenic lines” Methods section
Genetic reagent (Danio rerio)	rcn3:lyn-mNeonGreen	This paper		“Animal handling and generation of transgenic lines” Methods section
Genetic reagent (Danio rerio)	tp1:GFP	Parsons et al., 2009		“Animal handling and generation of transgenic lines” Methods section
Sequence-based reagent	Primer 1: pTarBAC_HA1_iTol2_F	This paper	Primer	gcgtaagcggggcacatttcattacctctttctccgcacccgacatagatCCCTGCTCGAGCCGGGCCCAAGTG
Sequence-based reagent	Primer 2: pTarBAC_HA2_iTol2_R	This paper	Primer	gcggggcatgactattggcgcgccggatcgatccttaattaagtctactaATTATGATCCTCTAGATCAGATC
Sequence-based reagent	Primer 3: jag1a_HA1_mScarlet_F	This paper	Primer	gaggcgtgtggcggctgaagtggtagttttcacagcgacagacacacagacagacaaaccACCATGGTGAGCAAGGGC
Sequence-based reagent	Primer 4: jag1a_HA2_FRT_R	This paper	Primer	agcagcacgtgagcggacagcgccgcaaaagttgagctcggtctgagaatGGAGGCTACCATGGAGAAG
Sequence-based reagent	Primer 5: jag1a_HA1_mNG_F	This paper	Primer	gaggcgtgtggcggctgaagtggtagttttcacagcgacagacacacagacagacaaaccACCATGGTGAGCAAGGGC
Sequence-based reagent	Primer 6: Scaffold	Shah et al., 2015	Primer	GATCCGCACCGACTCGGTGCCACTTTTTCAAGTTGATAACGGACTAGCCTTATTTTAACTT GCTATTTCTAGCTCTAAAAC
Sequence-based reagent	Primer 7: her6_guide1	This paper	Primer	taatacgactcactataGGTGGTCGGCGCCCCTCCATgttttagagctagaa
Sequence-based reagent	Primer 8: her6_guide2	This paper	Primer	taatacgactcactataGGGTGGCCATTCTTTGAAGGgttttagagctagaa
Sequence-based reagent	Primer 9: her9_guide1	This paper	Primer	taatacgactcactataGGGTGACTGACAGCCCGCGGgttttagagctagaa
Sequence-based reagent	Primer 10: her9_guide2	This paper	Primer	taatacgactcactataGGGGGAAACCCTGCGGCCGTgttttagagctagaa
Sequence-based reagent	Primer 11: univ_guide	Wierson et al., 2020	Primer	taatacgactcactataGGGAGGCGTTCGGGCCACAGgttttagagctagaa
Sequence-based reagent	Primer 12: her6_F	This paper	Primer	GTTTGCTGTTTCTGAGCGGAG
Sequence-based reagent	Primer 13: her6_R	This paper	Primer	GGGAAGCACGTCTGAGTCTG
Sequence-based reagent	Primer 14: her9_F	This paper	Primer	CCGCGCAGTATGTGAATGC
Sequence-based reagent	Primer 15: her9_R	This paper	Primer	ACCTTCACAGGCTACAGAACC
Sequence-based reagent	Control MO	Yamamoto et al., 2010	Morpholino	CCTCTTACCTCAGTTACAATTTATA
Sequence-based reagent	jag1a_splMO	Yamamoto et al., 2010	Morpholino	AAGCCAAACCCGCACATACCCGCAT
Sequence-based reagent	jag1b_atgMO	Yamamoto et al., 2010	Morpholino	CTGAACTCCGTCGCAGAATCATGCC
Recombinant DNA reagent	mScarlet FRT kan FRT	This paper		Sequence available in Source data 1 file.
Recombinant DNA reagent	mNG FRT kan FRT	This paper		Sequence available in Source data 2 file.
Recombinant DNA reagent	emilin3a mScarlet	This paper		Sequence available in Source data 3 file.
Recombinant DNA reagent	rcn3 lyn mNeonGreen	This paper		Sequence available in Source data 4 file.
Recombinant DNA reagent	SP6 lyn-miRFP-pA	This paper		Sequence available in Source data 5 file.
Recombinant DNA reagent	emilin3a GFP	This paper		Sequence available in Source data 6 file.
Recombinant DNA reagent	emilin3a GFP-p2a-her6	This paper		Sequence available in Source data 7 file.
Recombinant DNA reagent	emilin3a GFP-p2a-her9	This paper		Sequence available in Source data 8 file.
Recombinant DNA reagent	emilin3a mScarlet-p2a-JICD	This paper		Sequence available in Source data 9 file.
Commercial assay or kit	mMESSAGEmMACHINESP6 Transcription Kit	Thermo Fisher Scientific	Cat#:AM1340
Commercial assay or kit	Tricaine (MESAB)	Sigma-Aldrich	Cat#:A5040
Software, algorithm	Code for image analysis and mathematical modeling	This paper		Available in github: https://github.com/hsancheziranzo/notochord-lateral-inhibition (copy archived at swh:1:rev:2e5c5fe15e30ea6bacdc0282e1506b44b05415af, Sánchez-Iranzo, 2022)

Animal handling and generation of transgenic lines

Request a detailed protocol

The construct to generate Tg(jag1a:mScarlet) transgenic line was generated by BAC recombineering using the CH211-21D8 BAC. We first used EL250 (Lee et al., 2001) bacteria to recombine first the iTol2Amp cassette (Suster et al., 2011, primers 1 and 2, Key Resources Table) and substitute the loxP site in the BAC backbone. To recombine the mScarlet sequence into the BAC, we first used Gibson Assembly to substitute mCherry-p2a-CreERT2 by mScarlet in the mCherry-p2a-CreERT2-FRT-kan-FRT plasmid (Sánchez-Iranzo et al., 2018) to generate an mScarlet-FRT-kan-FRT plasmid (Source data 1). Then, we used the primers 3 and 4 (Key Resources Table) to amplify and recombine the mScarlet-FRT-kan-FRT into the ATG of jag1a in the BAC CH211-21D8. Finally, we removed the kanamycin resistance by activating flipase expression in the EL250 bacteria.

Similarly, we generated the jag1a:mNeonGreen BAC by first using Gibson Assembly to generate the mNeon-Green-FRT-kan-FRT plasmid (Source data 2). Next, we used primers 4 and 5 (Key Resources Table) to amplify the mNeonGreen-FRT-kan-FRT into the ATG of the jag1a BAC, followed by kanamycin resistance removal.

To clone the emilin3a:mScarlet plasmid (Source data 3) we selected the 5 kb upstream of the emilin3a ATG and cloned it upstream of mScarlet in a tol2 plasmid. The rcn3:lyn-mNeonGreen construct (Source data 4) was generated by Gibson Assembly using the previously described rcn3 promoter (Ellis et al., 2013).

jag1a:mScarlet, jag1a:mNeonGreen, emilin3a:mScarlet and rcn3:lyn-mNeonGreen were injected at the one cell stage using tol2 transposase. To establish the stable transgenic lines, we crossed the fish by wild type until we found 50% of the progeny transgenic, indicative of a probable single insertion. For the rcn3:mNeonGreen transgenic line, due to the high variability in gene expression between different lines, we selected the most notochord specific line among 5–10 different founders.

As a reporter of Notch activity, we used the tp1:GFP line (Parsons et al., 2009). This line includes six copies of the promoter from the Epstein-Barr Virus terminal protein 1 (TP1), cloned upstream of the rabbit β-globin minimal promoter. Each TP1 copy contains two Rbp-Jκ binding sites.

All experiments were performed on embryos younger than 3 dpf, as is stipulated by the EMBL internal policy 65 (IP65) and European Union Directive 2010/63/EU.

her6 and her9 Knockout

Request a detailed protocol

To generate her6 and her9 transient knockout (crispants), we designed guide RNAs (gRNAs) targeting the beginning and the end of both her6 and her9, resulting in whole gene deletion. Guides were identified using CRISPRscan (Doench et al., 2014; Moreno-Mateos et al., 2015) and synthesized as previously described (Shah et al., 2015; Primers 6–10, Key Resources Table). The injection mix included custom-produced Cas9-GFP at 2.4 mg/mL, KCl 300 mM and the four gRNAs, each of them at 12.5 ng/μL. Only embryos where the antero-posterior axis was shortened were selected for imaging. As a control, we used embryos where a gRNA with no target in the zebrafish genome (Wierson et al., 2020; Primer 11, Key Resources Table) was injected. Primers 12–15 (Table S2) were used for the detection of the deleted allele in all the fish used for imaging. Effective deletion was confirmed by sequencing of two KO her6 and two KO her9 PCR products; only embryos where both a her6 and her9 knockout band was detected by PCR (7/10) were considered for the quantification. Heterozygous embryos for both rcn3:mNeonGreen and jag1a:mScarlet transgenes were used in this experiment. Cells with jag1a:mScarlet intensity lower than 10% of the maximum intensity value in each image were considered negative for jag1a.

jag1a and jag1b MOs

Request a detailed protocol

The injection mix contained 100 ng/uL of lyn-miRFP mRNA and 0.4 mM of MO (Gene Tools). Specifically, the jag1a/jag1b mix contained 0.2 mM jag1a + 0.2 mM jag1b, the jag1a mix contained 0.2 mM jag1a + 0.2 control MO, the jag1b mix contained 0.2 mM jag1b + 0.2 mM control MO, and the control MO mix contained 0.4 mM of control MO. jag1a and jag1b MOs had been described and validated previously (Yamamoto et al., 2010).

mRNA was generated by digestion of the SP6 lyn-miRFP-pA plasmid (Source data 5) with NotI, followed by SP6 mediated transcription (mMessage mMachine SP6, Thermo Fisher Scientific).

The lyn-miRFP (Shcherbakova et al., 2016) mRNA injected, not only allowed membrane labeling, but it is also a control of injection. Few embryos where the infrared membrane signal was not detected were excluded from the analysis. Cells with jag1a:mScarlet intensity lower than 10% of the maximum intensity value in each image were considered negative for jag1a.

Cell fate analysis emilin3a:GFP (Source data 6), emilin3a:mScarlet (Source data 3), emilin3a:GFP-p2a-her6 (Source data 7), emilin3a:GFP-p2a-her9 (Source data 8) or emilin3a:mScarlet-p2a-jag1a (Source data 9) were cloned using Gibson Assembly using as template synthesized her6, her9, and jag1a cDNAs. These plasmids were injected at the one cell stage using Isce-I as previously described (Rembold et al., 2006). GFP fluorescence and transmitted light were imaged in vivo at 2 dpf. Quantifications were made on 3D confocal stacks. Number of cells were manually quantified using the Cell Counter Fiji plugin (Schindelin et al., 2012).

Hybridization chain reaction and immunofluorescence

Request a detailed protocol

First, emilin3a:GFP, emilin3a:GFP-p2a-her6, emilin3a:GFP-p2a-her9 or emilin3a:mScarlet-p2a-jag1a constructs were injected at the one cell stage and fish were fixed at 20–22 hpf. Hybridization chain reaction (Molecular Instruments) was performed following manufacturer instructions. her6, her9, jag1a, jag1b and notch2 probes were produced by Molecular Instruments as 20 probe set sizes. If GFP needed to be detected, after HCR protocol, samples were incubated overnight with anti-GFP nanobody A488 (gb2AF488, Chromotek, 1:500), followed by 5 × 30 min SSCT 5 X washing steps.

Single-cell RNA-Seq analysis

Request a detailed protocol

Single-cell RNA-Seq data was obtained from Wagner et al., 2018 (Wagner et al., 2018). We filtered the raw data and selected the cells labeled as notochord in the original publication, and analyzed them using the Scanpy v1.4.4 (Wolf et al., 2018) python package. UMAP coordinates were calculated using normalized non-logarithmically transformed values and the scanpy.pp.neighbors function with n_neighbors = 20 and n_pcs = 5 parameter values. log(UMI +1) values were represented in the UMAP plots, where log represents natural logarithm. Boxplots and heatmaps were generated using the seaborn python package.

emilin3a was found as the gene with the best balance between notochord enrichment and high expression levels. We did this by selecting the gene with the highest score according to this equation:

s c o r e = \frac{{E x p r_N}^{2}}{E x p r_N N}

where $E x p r_N$ represents the average of normalized UMIs for each gene across all notochord cells at 18 hpf, and $E x p r_N N$ represents the analogous values for the non-notochord cells at the same stage. Genes with the highest score are shown in Table 1.

Interpretation of the data was supported by the extensive data available in ZFIN (Howe et al., 2021).

Table 1

Genes with a highest score for specificity and expression levels in the notochord at 18 hpf.

Expression: Average expression in Notochord cells (normalized UMIs per million). Enrichment: Average expression in notochord cells divided by average expression in the rest of the cells in the fish at 18 hpf. Score: Expression multiplied by enrichment (equivalent to the equation described above).

Gene	Expression	Enrichment	Score
emilin3a	5055.75	1118.16	5653.16
ntd5	5929.62	134.03	794.74
col2a1a	9083.50	67.75	633.59
cmn	1536.94	330.67	508.22
loxl5b	1264.29	330.84	418.28
col9a1b	1511.20	226.72	342.62
ta	1443.23	162.65	234.74
LOC100333762	642.86	303.86	195.34
lgals1l1	2163.88	86.59	187.38
col9a2	1331.86	103.37	137.68
si:ch211-125-g7.4	501.98	272.60	136.84
si:dkey-12l12.1	408.15	304.23	127.17
col9a3	754.32	140.39	105.90
LOC100334188	393.62	205.68	80.96
pmp22b	1001.80	68.64	68.76
si:dkey-99l1.9	499.03	131.45	65.60
si:ch73-23l24.1	317.78	205.60	65.34
lgals2a	1018.17	64.08	65.24
twist2	306.83	167.20	51.60
lect1	616.97	79.24	48.89

Electron microscopy

Request a detailed protocol

For EM imaging, samples were chemically fixed by immersing them in 2.5% glutaraldehyde and 4% paraformaldehyde in 0.1 M PHEM buffer. Sections were post-stained with uranyl acetate for 5 min and with lead citrate for 2 min. The overall EM protocol is similar to previously reported (Schieber et al., 2010).

Microscopy

Request a detailed protocol

Zebrafish embryos were embedded in 0.6% agarose low gelling temperature (A0701, Sigma) with 0.16 mg ml−1 Tricaine in E3 medium. For imaging embryos between 18 and 24 hpf, agarose covering the tail was removed to allow freely development of their tail. Imaging was performed with a Zeiss LSM880 laser scanning confocal microscope, using a 40 x/1.1NA water-immersion objective.

Adaptive feedback microscopy workflow

Request a detailed protocol

The adaptive feedback microscopy workflow was set up on Zeiss LSM880 AiryScan Fast microscope. Automated image analysis and definition of high-zoom tile positions was implemented as a Fiji plugin using previously developed AutoMicTools library (https://git.embl.de/halavaty/AutoMicTools). MyPic VBA macro (Politi et al., 2018) was used as a communication interface between the Fiji plugin and ZenBlack software controlling the microscope.

Both low-zoom and high-zoom images were acquired using AiryFast modality to enable time resolution of 5 min. 488 nm line of the Argon laser was used for excitation, fluorescent signal was detected using 499–553 nm emission filter. Low-zoom images were acquired using lowest possible zoom and rectangular tilescan in the total area 991 by 673 μm with the pixel size 0.835 μm and spacing between slices 5 μm. Each high-zoom tile was acquired in the field of view 83.72 by 83.72 μm with the pixel size 0.108 μm and spacing between slices 2.5 μm. Collected high-zoom tiles were stitched in Fiji using BigStitcher plugin (Hörl et al., 2019) and custom Jython scripts. To show the same region of the notochord independently on the move of the developing zebrafish, we used a custom-made Fiji Macro where the region of interest was manually selected every 10 frames, and interpolated for the rest of the timepoints.

To show the same region of the notochord independently on the move of the developing embryo, we used a custom-made Fiji Macro where the region of interest was manually selected every 10 frames, and the region of interest interpolated for the rest of the timepoints.

Image analysis

Request a detailed protocol

Python 3.7.4 was used for image analysis. First, the intensities of each of the channels was normalized between 0 and 1, where 0 was assigned to the minimum intensity value in the image, and 1 to the maximum value. Then, a gaussian filter was applied to the channel. This was done using the filters.gaussian_filter function of scipy.ndimage package, with a sigma value equal to 3. Then, both adaptive and global single-value segmentation were applied to the GFP channel. For the global single-value segmentation, the value was chosen automatically for each image as 1.5 times the median intensity of the GFP channel. To generate the adaptive segmentation, we calculated the local mean using as a kernel a uniform circle of 120 pixel diameter, and the rank.mean function of the skimage.filters package. Only the pixels with a higher value than both the global and the adaptive thresholds were considered for further analysis (Segmentation 1).

To define the GFP-positive cells, we filled holes in the cells by applying a 5-iteration binary dilation followed by a 9-interation binary erosion (scipy.ndimage python package). A higher erosion than dilation was applied to avoid defining as GFP-positive cells the pixels in the boundaries between cells. Only objects with an area of 3500 squared pixels were defined as cells and considered for further analysis (Segmentation 2).

The neighborhood of GFP cells was defined as follows. We first applied an 8-pixel binary dilation of 8 pixels to the GFP cells as defined in ‘Segmentation 1’ to define the boundary between cells. We then applied a 25-pixel binary dilation to define the neighboring cells. The region generated by the 25-pixel dilatation is the region that we considered as ‘neighboring cells’ (Segmentation 3).

To determine the relative intensity inside the ‘GFP-positive cells’ or the ‘neighboring to GFP cells’ we manually selected the notochord region, and we only considered the pixels inside the manually selected region. Then, we measured the mean value of the different mRNA signals inside the selected cells relative to the value of all the notochord.

In all the analyzed images, the stepsize is 63.7 nm/pixel. Plots were generated using boxplot and swarmplot functions of the seaborn python package.

Statistical analysis

Request a detailed protocol

Statistical analysis was performed using the scipy.stats python package. The specific statistical test used, including sample size and the p-values are indicated in the figures and figure legends.

Data and code availability

Request a detailed protocol

Code is available under the MIT open source license on GitHub at: https://github.com/hsancheziranzo/notochord-lateral-inhibition (copy archived at swh:1:rev:2e5c5fe15e30ea6bacdc0282e1506b44b05415af) (Sánchez-Iranzo et al., 2021; Sánchez-Iranzo, 2022). Images used for image analysis are available in Mendeley Data: https://doi.org/10.17632/fzmk5k982j.1 (CC BY 4.0).

Materials availability

Request a detailed protocol

Requests for experimental resources and reagents should be directed to and will be fulfilled by Alba Diz-Muñoz (diz@embl.de) or Héctor Sánchez-Iranzo (hector.sanchez@kit.edu).

Appendix 1

Description of the theoretical model

Lateral induction model

The lateral induction model was defined as a two-component system, Ligand (L) and Notch Intracellular Domain (NICD, represented as I in the equations). Notch-Ligand interaction in adjacent cells triggers the release of NICD following an increasing Hill function. NICD activates the expression of the ligand in its own cell following an increasing Hill function. The equations that describe the model are:

\frac{{d I}_{i}}{d t} = β_{I} \frac{{〈L_{j}〉}^{h}}{a + {〈L_{j}〉}^{h}} - γ_{I} I_{i}

\frac{{d L}_{i}}{d t} = β_{L} \frac{{b I}_{i}^{h}}{1 + {b I}_{i}^{h}} - γ_{L} L_{i}

$L_{i}$ and $I_{i}$ are the average concentrations of Ligand and NICD inside the cells, respectively. $〈L_{j}〉$ is the average concentration of Ligand in each of the neighboring cells. $β_{I}$ and $β_{L}$ are the production rates of ligand and receptor, respectively. $γ_{L}$ and $γ_{I}$ are the degradation rates of Ligand and NICD, respectively, $a$ and $b$ the affinities, and $h$ is the Hill coefficient.

Lateral inhibition model

This model is based on Collier et al., 1996 and is similar to the lateral induction, with the only difference that the lateral inhibition model assumes that NICD activates the expression of a repressor that in turn inhibits the expression of the ligand. For this reason, the production of ligand is represented as an inhibitory Hill function.

The equations that describe the system are:

\frac{{d I}_{i}}{d t} = β_{I} \frac{{〈L_{j}〉}^{h}}{a + {〈L_{j}〉}^{h}} - γ_{I} I_{i}

\frac{{d L}_{i}}{d t} = β_{L} \frac{1}{1 + {b I}_{i}^{h}} - γ_{L} L_{i}

Lateral inhibition model with mutual inhibition

The equations that describe this model are based on Sprinzak et al., 2010; Sprinzak et al., 2011

\frac{{d N}_{i}}{d t} = β_{N} - k_{t} N_{i} 〈L_{j}〉 - k_{c} N_{i} L_{i} - {γ N}_{i}

\frac{{d L}_{i}}{d t} = β_{L} \frac{1}{1 + R_{i}^{n}} - k_{t} L_{i} 〈N_{j}〉 - k_{c} N_{i} L_{i} - {γ L}_{i}

\frac{{d R}_{i}}{d t} = β_{R} \frac{{{(N}_{i} 〈L_{j}〉)}^{n}}{k_{R S} + {{(N}_{i} 〈L_{j}〉)}^{n}} - {γ_{R} R}_{i}

$N_{i}$ , $L_{i}$ and $R_{i}$ are the average concentrations of Notch Receptor, Ligand and Repressor inside the cells, respectively. $〈L_{j}〉$ and $〈N_{j}〉$ are the average concentrations of ligand in the neighboring cells. $β_{N}$ , $β_{L}$ and $β_{R}$ are the production rates of Notch Receptor, Ligand and Repressor, respectively. $γ$ and $γ_{R}$ are the degradation rates of Notch Receptor and Ligand/Repressor, respectively. $k_{R S}$ is the affinity, and $n$ is the Hill coefficient. $k_{c}$ and $k_{t}$ are the interaction strength between ligand and receptor in cis and trans, respectively. These two constants are referred as $K_{c i s}$ and $K_{t r a n s}$ in the manuscript.

Simulations

All the visual simulations were generated by solving the equations using the Euler method with a step set to 0.01. Simulations were initialized with random values uniformly distributed between 0 and 0.1. To avoid boundary effects, we run simulations on a 100 cell array, where only the 20 central cells are displayed, while the 40 cells in each side buffer the boundary effect.

Linear stability analysis

Linear stability analysis was done as previously described (Sprinzak et al., 2011). A prerequisite for pattern formation is the instability of the homogenous steady state ( $N$ *, $L$ *, $R$ *), where every cell has the same value of $N_{i}$ , $L_{i}$ and $R_{i}$ . We first calculated the homogeneous steady state by making $N_{i}$ and $N_{j}$ equal to $N$ *, $L_{i}$ and $L_{j}$ equal to $L$ *, and $R_{i}$ equal to $R$ *, and solving the following system of equations (Sprinzak et al., 2011):

0 = β_{N} - k_{t} N^{*} L^{*} - k_{c} N^{*} L^{*} - {γ N}^{*}

0 = β_{L} \frac{1}{1 + {{(R}^{*})}^{n}} - k_{t} L^{*} 〈N^{*}〉 - k_{c} N^{*} L^{*} - {γ L}^{*}

0 = β_{R} \frac{{(N^{*} 〈L^{*}〉)}^{n}}{k_{R S} + {(N^{*} 〈L^{*}〉)}^{n}} - γ_{R} R^{*}

We solved these equations for the $R$ *, $L$ * and $N$ * using the fsolve function of the scipy.optimize python package.

The stability analysis requires the computation of the Jacobian matrix, that according to Othmer and Scriven, 1971 can be expressed as $J = I_{k} \otimes H + M \otimes B$ , where $I_{k}$ is the $k x k$ identity matrix, $k$ is the number of cells, ⊗ represents the tensor product, $H_{i j} = \frac{\partial \dot{q}}{\partial q_{j}}$ is the change in production of species $i$ for a change in species $j$ in the same cell, $B_{i j} = \frac{\partial \dot{q}}{\partial 〈q_{j}〉}$ is the change in production of species $i$ for a change in species $j$ in a neighboring cell, and $M$ is the connectivity matrix defined as

M = {\begin{cases} \frac{1}{2} i f i a n d j a r e n e i g h b o u r s \\ 0 o t h e r w i s e \end{cases}

In the specific case of our model, where cells are arranged unidimensionally, $i$ and $j$ are neighbors when $|i| - |j|$ = 1.

The eigenvalues of $J$ Jacobian matrix are the eigenvalues of the various matrices $H + q_{k} B H$ , where $q_{k}$ are the eigenvalues of the connectivity matrix $M$ . For our particular $M$ matrix, $q_{k}$ values are always higher or equal to – 1, meaning that we only need to compute an eigenvalue for the extreme case $q_{k} = - 1$ to determine if the highest eigenvalue (known as the Maximum Lyapunov Exponent, MLE) has a positive real part.

Following this strategy, we computed the MLE value for a grid of $k_{c}$ and $k_{t}$ values logarithmically spaced between 0.001 and 100.

Parameter values

Figure 1, Figure 1—figure supplement 1	a = 0.1 b = 10 h = 2 γ_I = 1 γ_L = 1 β_I = 1 β_L = 1
Figure 6, Figure 6—figure supplement 1	n = 1 k_c = 0.001 to 100 k_t = 0.001 to 100 γ = 1 γ_R = 1 k_RS = 1 β_N = 3 β_L = 10 β_R = 10
Figure 6—figure supplement 2	n = 1 k_c = 0.001 to 100 k_t = 0.001 to 100 γ = 1 γ_R = 1 k_RS = 1 β_N = 3 β_L = 1.5 to 10 β_R = 1.5 to 10

Data availability

Code is available under the MIT open source license on GitHub at: https://github.com/hsancheziranzo/notochord-lateral-inhibition (copy archived at swh:1:rev:2e5c5fe15e30ea6bacdc0282e1506b44b05415af) (Sanchez-Iranzo, 2022). Images used for image analysis are available in Mendeley Data: https://doi.org/10.17632/fzmk5k982j.1 (CC BY 4.0).

The following data sets were generated

1. Sánchez-Iranzo H
(2022) Mendeley Data
Strength of interactions in the Notch gene regulatory network determines patterning and fate in the notochord.
https://doi.org/10.17632/fzmk5k982j.1

The following previously published data sets were used

(2018) NCBI Gene Expression Omnibus
ID GSE112294. Systematic mapping of cell state trajectories, cell lineage, and perturbations in the zebrafish embryo using single cell transcriptomics.

https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE112294

References

(1990) The mechanics of notochord elongation, straightening and stiffening in the embryo of Xenopus laevis
Development 110:115–130.
https://doi.org/10.1242/dev.110.1.115
- PubMed
- Google Scholar
1. Apelqvist A
2. Li H
3. Sommer L
4. Beatus P
5. Anderson DJ
6. Honjo T
7. Hrabe de Angelis M
8. Lendahl U
9. Edlund H
(1999) Notch signalling controls pancreatic cell differentiation
Nature 400:877–881.
https://doi.org/10.1038/23716
- PubMed
- Google Scholar
(1999) Notch signaling: cell fate control and signal integration in development
Science 284:770–776.
https://doi.org/10.1126/science.284.5415.770
- PubMed
- Google Scholar
1. Bagwell J
2. Norman J
3. Ellis K
4. Peskin B
5. Hwang J
6. Ge X
7. Nguyen SV
8. McMenamin SK
9. Stainier DY
10. Bagnat M
(2020) Notochord vacuoles absorb compressive bone growth during zebrafish spine formation
eLife 9:e51221.
https://doi.org/10.7554/eLife.51221
- PubMed
- Google Scholar
1. Benedito R
2. Roca C
3. Sörensen I
4. Adams S
5. Gossler A
6. Fruttiger M
7. Adams RH
(2009) The notch ligands Dll4 and Jagged1 have opposing effects on angiogenesis
Cell 137:1124–1135.
https://doi.org/10.1016/j.cell.2009.03.025
- PubMed
- Google Scholar
(2006) Canonical notch signaling functions as a commitment switch in the epidermal lineage
Genes & Development 20:3022–3035.
https://doi.org/10.1101/gad.1477606
- PubMed
- Google Scholar
1. Boareto M
2. Jolly MK
3. Lu M
4. Onuchic JN
5. Clementi C
6. Ben-Jacob E
(2015) Jagged-Delta asymmetry in Notch signaling can give rise to a Sender/Receiver hybrid phenotype
PNAS 112:E402–E409.
https://doi.org/10.1073/pnas.1416287112
- PubMed
- Google Scholar
1. Boareto M
(2020) Patterning via local cell-cell interactions in developing systems
Developmental Biology 460:77–85.
https://doi.org/10.1016/j.ydbio.2019.12.008
- PubMed
- Google Scholar
1. Bray SJ
(2006) Notch signalling: A simple pathway becomes complex
Nature Reviews. Molecular Cell Biology 7:678–689.
https://doi.org/10.1038/nrm2009
- PubMed
- Google Scholar
1. Bray SJ
(2016) Notch signalling in context
Nature Reviews. Molecular Cell Biology 17:722–735.
https://doi.org/10.1038/nrm.2016.94
- PubMed
- Google Scholar
1. Briscoe J
2. Small S
(2015) Morphogen rules: design principles of gradient-mediated embryo patterning
Development 142:3996–4009.
https://doi.org/10.1242/dev.129452
- PubMed
- Google Scholar
1. Celis de JF
2. Bray S
(1997) Feed-back mechanisms affecting Notch activation at the dorsoventral boundary in the Drosophila wing
Development 124:3241–3251.
https://doi.org/10.1242/dev.124.17.3241
- Google Scholar
1. Collier JR
2. Monk NAM
3. Maini PK
4. Lewis JH
(1996) Pattern formation by lateral inhibition with feedback: A mathematical model of delta-notch intercellular signalling
Journal of Theoretical Biology 183:429–446.
https://doi.org/10.1006/jtbi.1996.0233
- PubMed
- Google Scholar
1. Dale RM
2. Topczewski J
(2011) Identification of an evolutionarily conserved regulatory element of the zebrafish col2a1a gene
Developmental Biology 357:518–531.
https://doi.org/10.1016/j.ydbio.2011.06.020
- PubMed
- Google Scholar
1. de Celis JF
2. Bray SJ
(2000) The Abruptex domain of Notch regulates negative interactions between Notch, its ligands and Fringe
Development 127:1291–1302.
https://doi.org/10.1242/dev.127.6.1291
- Google Scholar
(2011) Mechanism and significance of cis-inhibition in Notch signalling
Current Biology 21:R40–R47.
https://doi.org/10.1016/j.cub.2010.10.034
- PubMed
- Google Scholar
1. Deng W
2. Nies F
3. Feuer A
4. Bocina I
5. Oliver D
6. Jiang D
(2013) Anion translocation through an Slc26 transporter mediates lumen expansion during tubulogenesis
PNAS 110:14972–14977.
https://doi.org/10.1073/pnas.1220884110
- PubMed
- Google Scholar
1. Doench JG
2. Hartenian E
3. Graham DB
4. Tothova Z
5. Hegde M
6. Smith I
7. Sullender M
8. Ebert BL
9. Xavier RJ
10. Root DE
(2014) Rational design of highly active sgRNAs for CRISPR-Cas9-mediated gene inactivation
Nature Biotechnology 32:1262–1267.
https://doi.org/10.1038/nbt.3026
- PubMed
- Google Scholar
1. Dong B
2. Horie T
3. Denker E
4. Kusakabe T
5. Tsuda M
6. Smith WC
7. Jiang D
(2009) Tube formation by complex cellular processes in Ciona intestinalis notochord
Developmental Biology 330:237–249.
https://doi.org/10.1016/j.ydbio.2009.03.015
- PubMed
- Google Scholar
(2016) Sequential Notch activation regulates ventricular chamber development
Nature Cell Biology 18:7–20.
https://doi.org/10.1038/ncb3280
- PubMed
- Google Scholar
(2013) Notochord vacuoles are lysosome-related organelles that function in axis and spine morphogenesis
The Journal of Cell Biology 200:667–679.
https://doi.org/10.1083/jcb.201212095
- PubMed
- Google Scholar
1. Formosa-Jordan P
2. Ibañes M
(2014) Competition in notch signaling with cis enriches cell fate decisions
PLOS ONE 9:e95744.
https://doi.org/10.1371/journal.pone.0095744
- PubMed
- Google Scholar
(2005) Notch signals control the fate of immature progenitor cells in the intestine
Nature 435:964–968.
https://doi.org/10.1038/nature03589
- PubMed
- Google Scholar
1. Garcia J
2. Bagwell J
3. Njaine B
4. Norman J
5. Levic DS
6. Wopat S
7. Miller SE
8. Liu X
9. Locasale JW
10. Stainier DYR
11. Bagnat M
(2017) Sheath Cell Invasion and Trans-differentiation Repair Mechanical Damage Caused by Loss of Caveolae in the Zebrafish Notochord
Current Biology 27:1982–1989.
https://doi.org/10.1016/j.cub.2017.05.035
- PubMed
- Google Scholar
1. Gray RS
2. Wilm TP
3. Smith J
4. Bagnat M
5. Dale RM
6. Topczewski J
7. Johnson SL
8. Solnica-Krezel L
(2014) Loss of col8a1a function during zebrafish embryogenesis results in congenital vertebral malformations
Developmental Biology 386:72–85.
https://doi.org/10.1016/j.ydbio.2013.11.028
- PubMed
- Google Scholar
1. Gray RS
2. Gonzalez R
3. Ackerman SD
4. Minowa R
5. Griest JF
6. Bayrak MN
7. Troutwine B
8. Canter S
9. Monk KR
10. Sepich DS
11. Solnica-Krezel L
(2021) Postembryonic screen for mutations affecting spine development in zebrafish
Developmental Biology 471:18–33.
https://doi.org/10.1016/j.ydbio.2020.11.009
- PubMed
- Google Scholar
(2003) Notochord segmentation may lay down the pathway for the development of the vertebral bodies in the Atlantic salmon
Anatomy and Embryology 207:263–272.
https://doi.org/10.1007/s00429-003-0349-y
- PubMed
- Google Scholar
(2010) Notch signaling specifies prosensory domains via lateral induction in the developing mammalian inner ear
PNAS 107:15792–15797.
https://doi.org/10.1073/pnas.1002827107
- PubMed
- Google Scholar
1. Heitzler P
2. Simpson P
(1991) The choice of cell fate in the epidermis of Drosophila
Cell 64:1083–1092.
https://doi.org/10.1016/0092-8674(91)90263-x
- PubMed
- Google Scholar
1. Hörl D
2. Rojas Rusak F
3. Preusser F
4. Tillberg P
5. Randel N
6. Chhetri RK
7. Cardona A
8. Keller PJ
9. Harz H
10. Leonhardt H
11. Treier M
12. Preibisch S
(2019) BigStitcher: reconstructing high-resolution image datasets of cleared and expanded samples
Nature Methods 16:870–874.
https://doi.org/10.1038/s41592-019-0501-0
- PubMed
- Google Scholar
1. Howe DG
2. Ramachandran S
3. Bradford YM
4. Fashena D
5. Toro S
6. Eagle A
7. Frazer K
8. Kalita P
9. Mani P
10. Martin R
11. Moxon ST
12. Paddock H
13. Pich C
14. Ruzicka L
15. Schaper K
16. Shao X
17. Singer A
18. Van Slyke CE
19. Westerfield M
(2021) The Zebrafish Information Network: major gene page and home page updates
Nucleic Acids Research 49:D1058–D1064.
https://doi.org/10.1093/nar/gkaa1010
- PubMed
- Google Scholar
1. Ikeuchi T
2. Sisodia SS
(2003) The Notch ligands, Delta1 and Jagged2, are substrates for presenilin-dependent “gamma-secretase” cleavage
The Journal of Biological Chemistry 278:7751–7754.
https://doi.org/10.1074/jbc.C200711200
- PubMed
- Google Scholar
(2010) Essential roles of Notch signaling in maintenance of neural stem cells in developing and adult brains
The Journal of Neuroscience 30:3489–3498.
https://doi.org/10.1523/JNEUROSCI.4987-09.2010
- PubMed
- Google Scholar
(2007) The Hes gene family: repressors and oscillators that orchestrate embryogenesis
Development 134:1243–1251.
https://doi.org/10.1242/dev.000786
- PubMed
- Google Scholar
1. Kim MY
2. Jung J
3. Mo JS
4. Ann EJ
5. Ahn JS
6. Yoon JH
7. Park HS
(2011a) The intracellular domain of Jagged-1 interacts with Notch1 intracellular domain and promotes its degradation through Fbw7 E3 ligase
Experimental Cell Research 317:2438–2446.
https://doi.org/10.1016/j.yexcr.2011.07.014
- PubMed
- Google Scholar
1. Kim JH
2. Lee SR
3. Li LH
4. Park HJ
5. Park JH
6. Lee KY
7. Kim MK
8. Shin BA
9. Choi SY
(2011b) High cleavage efficiency of a 2A peptide derived from porcine teschovirus-1 in human cell lines, zebrafish and mice
PLOS ONE 6:e18556.
https://doi.org/10.1371/journal.pone.0018556
- PubMed
- Google Scholar
(1997) An intrinsic dominant negative activity of serrate that is modulated during wing development in Drosophila
Developmental Biology 189:123–134.
https://doi.org/10.1006/dbio.1997.8564
- PubMed
- Google Scholar
(2011) A BAC transgenic Hes1-EGFP reporter reveals novel expression domains in mouse embryos
Gene Expression Patterns 11:415–426.
https://doi.org/10.1016/j.gep.2011.06.004
- PubMed
- Google Scholar
1. Kolev V
2. Kacer D
3. Trifonova R
4. Small D
5. Duarte M
6. Soldi R
7. Graziani I
8. Sideleva O
9. Larman B
10. Maciag T
11. Prudovsky I
(2005) The intracellular domain of Notch ligand Delta1 induces cell growth arrest
FEBS Letters 579:5798–5802.
https://doi.org/10.1016/j.febslet.2005.09.042
- PubMed
- Google Scholar
1. Lanford PJ
2. Lan Y
3. Jiang R
4. Lindsell C
5. Weinmaster G
6. Gridley T
7. Kelley MW
(1999) Notch signalling pathway mediates hair cell development in mammalian cochlea
Nature Genetics 21:289–292.
https://doi.org/10.1038/6804
- PubMed
- Google Scholar
1. LaVoie MJ
2. Selkoe DJ
(2003) The Notch ligands, Jagged and Delta, are sequentially processed by alpha-secretase and presenilin/gamma-secretase and release signaling fragments
The Journal of Biological Chemistry 278:34427–34437.
https://doi.org/10.1074/jbc.M302659200
- PubMed
- Google Scholar
(2001) A highly efficient Escherichia coli-based chromosome engineering system adapted for recombinogenic targeting and subcloning of BAC DNA
Genomics 73:56–65.
https://doi.org/10.1006/geno.2000.6451
- PubMed
- Google Scholar
(2003) Expression of Notch pathway genes in the embryonic mouse metanephros suggests a role in proximal tubule development
Gene Expression Patterns 3:595–598.
https://doi.org/10.1016/s1567-133x(03)00114-5
- PubMed
- Google Scholar
1. Li Y
2. Baker NE
(2004) The roles of cis-inactivation by Notch ligands and of neuralized during eye and bristle patterning in Drosophila
BMC Developmental Biology 4:1–41.
https://doi.org/10.1186/1471-213X-4-5
- PubMed
- Google Scholar
1. Liebler SS
2. Feldner A
3. Adam MG
4. Korff T
5. Augustin HG
6. Fischer A
(2012) No Evidence for a Functional Role of Bi-Directional Notch Signaling during Angiogenesis
PLOS ONE 7:12.
https://doi.org/10.1371/journal.pone.0053074
- Google Scholar
(1995) Jagged: A mammalian ligand that activates Notch1
Cell 80:909–917.
https://doi.org/10.1016/0092-8674(95)90294-5
- PubMed
- Google Scholar
1. Lopez-Baez JC
2. Simpson DJ
3. Forero LLl
4. Zeng Z
5. Brunsdon H
6. Salzano A
7. Brombin A
8. Wyatt C
9. Rybski W
10. Huitema LFA
11. Dale RM
12. Kawakami K
13. Englert C
14. Chandra T
15. Schulte-Merker S
16. Hastie ND
17. Patton EE
(2018) Wilms tumor 1b defines a wound-specific sheath cell subpopulation associated with notochord repair
eLife 7:e30657.
https://doi.org/10.7554/eLife.30657
- Google Scholar
1. Luca VC
2. Jude KM
3. Pierce NW
4. Nachury MV
5. Fischer S
6. Garcia KC
(2015) Structural biology. Structural basis for Notch1 engagement of Delta-like 4
Science 347:847–853.
https://doi.org/10.1126/science.1261093
- PubMed
- Google Scholar
1. Manderfield LJ
2. High FA
3. Engleka KA
4. Liu F
5. Li L
6. Rentschler S
7. Epstein JA
(2012) Notch activation of Jagged1 contributes to the assembly of the arterial wall
Circulation 125:314–323.
https://doi.org/10.1161/CIRCULATIONAHA.111.047159
- PubMed
- Google Scholar
1. Matsuda M
2. Koga M
3. Nishida E
4. Ebisuya M
(2012) Synthetic signal propagation through direct cell-cell interaction
Science Signaling 5:ra31.
https://doi.org/10.1126/scisignal.2002764
- PubMed
- Google Scholar
(2006) NOTCH2 mutations cause Alagille syndrome, a heterogeneous disorder of the notch signaling pathway
American Journal of Human Genetics 79:169–173.
https://doi.org/10.1086/505332
- PubMed
- Google Scholar
(2015) Jagged1 intracellular domain-mediated inhibition of Notch1 signalling regulates cardiac homeostasis in the postnatal heart
Cardiovascular Research 108:74–86.
https://doi.org/10.1093/cvr/cvv209
- PubMed
- Google Scholar
(1997) The function and regulation of cut expression on the wing margin of Drosophila: Notch, Wingless and a dominant negative role for Delta and Serrate
Development 124:1485.
https://doi.org/10.1242/dev.124.8.1485
- Google Scholar
(2015) CRISPRscan: designing highly efficient sgRNAs for CRISPR-Cas9 targeting in vivo
Nature Methods 12:982–988.
https://doi.org/10.1038/nmeth.3543
- PubMed
- Google Scholar
1. Murata J
2. Tokunaga A
3. Okano H
4. Kubo T
(2006) Mapping of notch activation during cochlear development in mice: implications for determination of prosensory domain and cell fate diversification
The Journal of Comparative Neurology 497:502–518.
https://doi.org/10.1002/cne.20997
- PubMed
- Google Scholar
(2011) Jagged 1 regulates the restriction of Sox2 expression in the developing chicken inner ear: A mechanism for sensory organ specification
Development 138:735–744.
https://doi.org/10.1242/dev.060657
- PubMed
- Google Scholar
1. Oda T
2. Elkahloun AG
3. Pike BL
4. Okajima K
5. Krantz ID
6. Genin A
7. Piccoli DA
8. Meltzer PS
9. Spinner NB
10. Collins FS
11. Chandrasekharappa SC
(1997) Mutations in the human Jagged1 gene are responsible for Alagille syndrome
Nature Genetics 16:235–242.
https://doi.org/10.1038/ng0797-235
- PubMed
- Google Scholar
1. Othmer HG
2. Scriven LE
(1971) Instability and dynamic pattern in cellular networks
Journal of Theoretical Biology 32:507–537.
https://doi.org/10.1016/0022-5193(71)90154-8
- PubMed
- Google Scholar
1. Parsons MJ
2. Pisharath H
3. Yusuff S
4. Moore JC
5. Siekmann AF
6. Lawson N
7. Leach SD
(2009) Notch-responsive cells initiate the secondary transition in larval zebrafish pancreas
Mechanisms of Development 126:898–912.
https://doi.org/10.1016/j.mod.2009.07.002
- PubMed
- Google Scholar
(2014) Ligand-dependent Notch signaling strength orchestrates lateral induction and lateral inhibition in the developing inner ear
Development 141:2313–2324.
https://doi.org/10.1242/dev.108100
- PubMed
- Google Scholar
1. Politi AZ
2. Cai Y
3. Walther N
4. Hossain MJ
5. Koch B
6. Wachsmuth M
7. Ellenberg J
(2018) Quantitative mapping of fluorescently tagged cellular proteins using FCS-calibrated four-dimensional imaging
Nature Protocols 13:1445–1464.
https://doi.org/10.1038/nprot.2018.040
- PubMed
- Google Scholar
(2006) Transgenesis in fish: efficient selection of transgenic fish by co-injection with a fluorescent reporter construct
Nature Protocols 1:1133–1139.
https://doi.org/10.1038/nprot.2006.165
- PubMed
- Google Scholar
(2018) Tbx5a lineage tracing shows cardiomyocyte plasticity during zebrafish heart regeneration
Nature Communications 9:428.
https://doi.org/10.1038/s41467-017-02650-6
- Google Scholar
Preprint
(2021) Strength of Interactions in the Notch Gene Regulatory Network Determines Patterning and Fate in the Notochord
bioRxiv.
https://doi.org/10.1101/2021.03.25.436857
- Google Scholar
Software
1. Sánchez-Iranzo H
(2022) Notochord Lateral Inhibition, version swh:1:rev:2e5c5fe15e30ea6bacdc0282e1506b44b05415af
Software Heritage.

https://archive.softwareheritage.org/swh:1:dir:65841b6617aaa1ff230e5c98ced57ea202355d84;origin=https://github.com/hsancheziranzo/notochord-lateral-inhibition;visit=swh:1:snp:c82aa0f6475645c7489c29fba382428008287364;anchor=swh:1:rev:2e5c5fe15e30ea6bacdc0282e1506b44b05415af
(2010) Modern approaches for ultrastructural analysis of the zebrafish embryo
Methods in Cell Biology 96:425–442.
https://doi.org/10.1016/S0091-679X(10)96018-4
- PubMed
- Google Scholar
1. Schindelin J
2. Arganda-Carreras I
3. Frise E
4. Kaynig V
5. Longair M
6. Pietzsch T
7. Preibisch S
8. Rueden C
9. Saalfeld S
10. Schmid B
11. Tinevez JY
12. White DJ
13. Hartenstein V
14. Eliceiri K
15. Tomancak P
16. Cardona A
(2012) Fiji: an open-source platform for biological-image analysis
Nature Methods 9:676–682.
https://doi.org/10.1038/nmeth.2019
- PubMed
- Google Scholar
(2007) Premature myogenic differentiation and depletion of progenitor cells cause severe muscle hypotrophy in Delta1 mutants
PNAS 104:537–542.
https://doi.org/10.1073/pnas.0608281104
- PubMed
- Google Scholar
1. Schweisguth F
2. Corson F
(2019) Self-Organization in Pattern Formation
Developmental Cell 49:659–677.
https://doi.org/10.1016/j.devcel.2019.05.019
- PubMed
- Google Scholar
1. Shah AN
2. Davey CF
3. Whitebirch AC
4. Miller AC
5. Moens CB
(2015) Rapid reverse genetic screening using CRISPR in zebrafish
Nature Methods 12:535–540.
https://doi.org/10.1038/nmeth.3360
- PubMed
- Google Scholar
1. Shaya O
2. Binshtok U
3. Hersch M
4. Rivkin D
5. Weinreb S
6. Amir-Zilberstein L
7. Khamaisi B
8. Oppenheim O
9. Desai RA
10. Goodyear RJ
11. Richardson GP
12. Chen CS
13. Sprinzak D
(2017) Cell-Cell Contact Area Affects Notch Signaling and Notch-Dependent Patterning
Developmental Cell 40:505–511.
https://doi.org/10.1016/j.devcel.2017.02.009
- PubMed
- Google Scholar
(2016) Bright monomeric near-infrared fluorescent proteins as tags and biosensors for multiscale imaging
Nature Communications 7:12405.
https://doi.org/10.1038/ncomms12405
- PubMed
- Google Scholar
1. Simpson P
(1990) Lateral inhibition and the development of the sensory bristles of the adult peripheral nervous system of Drosophila
Development 109:509–519.
https://doi.org/10.1242/dev.109.3.509
- PubMed
- Google Scholar
1. Sjöqvist M
2. Andersson ER
(2019) Do as I say, Not(ch) as I do: Lateral control of cell fate
Developmental Biology 447:58–70.
https://doi.org/10.1016/j.ydbio.2017.09.032
- PubMed
- Google Scholar
(2010) Cis-interactions between Notch and Delta generate mutually exclusive signalling states
Nature 465:86–90.
https://doi.org/10.1038/nature08959
- PubMed
- Google Scholar
(2011) Mutual inactivation of Notch receptors and ligands facilitates developmental patterning
PLOS Computational Biology 7:e1002069.
https://doi.org/10.1371/journal.pcbi.1002069
- PubMed
- Google Scholar
1. Sternberg PW
(1988) Lateral inhibition during vulval induction in Caenorhabditis elegans
Nature 335:551–554.
https://doi.org/10.1038/335551a0
- PubMed
- Google Scholar
1. Sun X
2. Zhou Y
3. Zhang R
4. Wang Z
5. Xu M
6. Zhang D
7. Huang J
8. Luo F
9. Li F
10. Ni Z
11. Zhou S
12. Chen H
13. Chen S
14. Chen L
15. Du X
16. Chen B
17. Huang H
18. Liu P
19. Yin L
20. Chen L
(2020) Dstyk mutation leads to congenital scoliosis-like vertebral malformations in zebrafish via dysregulated mTORC1/TFEB pathway
Nature Communications 11:14169.
https://doi.org/10.1038/s41467-019-14169-z
- Google Scholar
1. Suster ML
2. Abe G
3. Schouw A
4. Kawakami K
(2011) Transposon-mediated BAC transgenesis in zebrafish
Nature Protocols 6:1998–2021.
https://doi.org/10.1038/nprot.2011.416
- Google Scholar
1. VanDussen KL
2. Carulli AJ
3. Keeley TM
4. Patel SR
5. Puthoff BJ
6. Magness ST
7. Tran IT
8. Maillard I
9. Siebel C
10. Kolterud Å
11. Grosse AS
12. Gumucio DL
13. Ernst SA
14. Tsai YH
15. Dempsey PJ
16. Samuelson LC
(2012) Notch signaling modulates proliferation and differentiation of intestinal crypt base columnar stem cells
Development 139:488–497.
https://doi.org/10.1242/dev.070763
- Google Scholar
(2007) RBP-J (Rbpsuh) is essential to maintain muscle progenitor cells and to generate satellite cells
PNAS 104:4443–4448.
https://doi.org/10.1073/pnas.0610647104
- PubMed
- Google Scholar
1. Wagner DE
2. Weinreb C
3. Collins ZM
4. Briggs JA
5. Megason SG
6. Klein AM
(2018) Single-cell mapping of gene expression landscapes and lineage in the zebrafish embryo
Science 360:981–987.
https://doi.org/10.1126/science.aar4362
- PubMed
- Google Scholar
(2011) Compartmentalized Notch signaling sustains epithelial mirror symmetry
Development 138:1143–1152.
https://doi.org/10.1242/dev.060566
- PubMed
- Google Scholar
1. Wierson WA
2. Welker JM
3. Almeida MP
4. Mann CM
5. Webster DA
6. Torrie ME
7. Weiss TJ
8. Kambakam S
9. Vollbrecht MK
10. Lan M
11. McKeighan KC
12. Levey J
13. Ming Z
14. Wehmeier A
15. Mikelson CS
16. Haltom JA
17. Kwan KM
18. Chien CB
19. Balciunas D
20. Ekker SC
21. Clark KJ
22. Webber BR
23. Moriarity BS
24. Solin SL
25. Carlson DF
26. Dobbs DL
27. McGrail M
28. Essner J
(2020) Efficient targeted integration directed by short homology in zebrafish and mammalian cells
eLife 9:e53968.
https://doi.org/10.7554/eLife.53968
- Google Scholar
(2018) SCANPY: large-scale single-cell gene expression data analysis
Genome Biology 19:15.
https://doi.org/10.1186/s13059-017-1382-0
- PubMed
- Google Scholar
1. Yamamoto M
2. Morita R
3. Mizoguchi T
4. Matsuo H
5. Isoda M
6. Ishitani T
7. Chitnis AB
8. Matsumoto K
9. Crump JG
10. Hozumi K
11. Yonemura S
12. Kawakami K
13. Itoh M
(2010) Mib-Jag1-Notch signalling regulates patterning and structural roles of the notochord by controlling cell-fate decisions
Development 137:2527–2537.
https://doi.org/10.1242/dev.051011
- PubMed
- Google Scholar
1. Ye Z
2. Braden CR
3. Wills A
4. Kimelman D
(2021) Identification of in vivo Hox13-binding sites reveals an essential locus controlling zebrafish brachyury expression
Development 148:199408.
https://doi.org/10.1242/dev.199408
- Google Scholar

Decision letter

Marianne E Bronner

Senior and Reviewing Editor; California Institute of Technology, United States

In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.

[Editors' note: this paper was reviewed by Review Commons.]

https://doi.org/10.7554/eLife.75429.sa1

Author response

Reviewer #1

Evidence, reproducibility and clarity:

Summary

In their manuscript, the authors characterize the process of differentiation into vacuolated and sheath cells in the zebrafish Notochord. They convincingly show that this differentiation process is essentially governed by Notch mediated lateral inhibition in one dimension (along the AP axis of the notochord). The authors use imaging of fluorescent reporters, single cell sequencing, and genetic perturbations to identify the key components involved in the lateral inhibition feedback including the receptors, ligands, and the transcriptional repressors. They also use mathematical modeling to guide their experiments and show that strong cis-inhibition between receptors and ligands play an important role in the lateral inhibition process. This is one of the unique cases where lateral inhibition in 1D is identified and characterized. The establishment of the system for studying the process of lateral inhibition is important.

Major comments

Overall, the manuscript is well written and describes a rigorous analysis of the lateral inhibition process in the system. The experiments performed follow a clear logic and conclusions drawn from the results are sound. Mathematical modeling is performed in a rigorous manner and the results are presented in a clear manner.

We thank the reviewer for the review of our manuscript and for stating that our data convincingly show our claims and that our work is important. Below we aim to address her/his concerns.

I do not think that additional experiments are required to support the claims presented, except a control for figure 3 (see below). However, some additional analysis of the data and some expansion of the modeling may provide additional support for the conclusions. These include:

1. The alternating pattern of precursor cells in aim1 indeed seems matching lateral inhibition. However, it would be worth making some clarifications as well as analyzing some aspects expected from a lateral inhibition patterns:

a. The pattern in Figure 1J,N show some dark gaps that are not green nor magenta. Are there cells in this gaps? If so, are these undecided cells? Would be good to clarify this point.

There are indeed cells in the dark gaps in Figure 1J, N. In the in vivo lineage tracing shown in Figure 1N, we can observe that most of these non-labelled cells become sheath cells. As tp1:GFP is not an endogenous gene, this suggests that the cells that are not labeled might not receive enough Notch signaling to activate the reporter. This highlights the importance of using endogenous markers. Indeed, we do not detect such gaps upon her6/her9 + jag1a mRNA staining (Figure 2). To address this important point we have added supporting data and clarified this in the text (Section 2 results, lines 196-200).

b. While figure 1M shows a cross section of fluorescence signal, it does not reveal whether the fluorescence per cell. Are there cells that express both green and magenta? Would be good to present a graph showing the fluorescence per cell in both green and red, The ideal expectation is that most cells would be at either high Jag1/low GFP or vice versa.

We have added an additional plot in in Figure 1 —figure supplement 1G where we show the intensity per cell of a single plane of the image shown in Figure 1J. We observe in the quantification that there are no double positive cells. Moreover, we have added the quantification per pixel, that although noisier, shows a similar result (Figure 1 —figure supplement 1H).

c. In a 1D lateral inhibition pattern it is expected that each high jag1 cell would have only low Jag1 neighbors, but low jag1 cells can still be neighbors (as shown in the simulation). Would be good to show if this is indeed the case here.

Due to the high stability of the fluorescent proteins, quantifying jag1a intensity can lead to erroneous conclusions. Thus, we evaluated cell fate instead of jag1a intensity. Cell fate has the advantage that is binary: vacuolated (jag1a) or sheath (notch), eliminating possible ambiguities.

To that end, we took advantage of our image feedback microscopy set up, where we could follow individual cells back in time and thus link them to their future fate at the disc-shaped stage. We have never observed two consecutive disc-shape cells that become vacuolated cells. In contrast, cells that become sheath cells have adjacent cells that become sheath cells. We now show the results of this quantification in Results section 2 lines 208-211.

2. In Figure 2 the authors claim that her6/9 and jag1a mRNA levels exhibit anti correlations. The authors should perform quantitative analysis to show this.

We now quantitatively show that her6/her9 mRNA anticorrelate with jag1a mRNA in Figures 2H, O.

3. In Figure 3-5 the authors use emillin3a driver to overexpress different genes. Is the expression of emillin3a uniform? It is hard to see from Figure S5C. Also, would be good to show that it is uncorrelated with Jag1a of tp1-GFP.

We now show higher magnification images of the notochord (Figures XX). Furthermore, we quantified how many tp1:GFP and jag1a:mNeonGreen are emilin3a:mScarlet-positive. Although not all the notochord cells have the same intensity of emilin3a:mScarlet, 98±3% of the tp1:GFP and 99±2% of the jag1a:mNeonGreen notochord cells express this marker, independently of these cells being sheath (tp1) or vacuolated (jag1a) cell precursors (Figure 3 —figure supplement 1 J–I).

4. In Figure 3h-J the authors show results with Crispr KO of Her6/9. It is important to show that the KO works and at what level, for example by showing mRNA levels after the KO.

To validate our experiments, we have performed PCR of the same embryos that we used for imaging. In Author response image 1A, B we provide the agarose gel where is possible to see the high efficiency of gene deletion. Based on these data, we have included in our quantification only the embryos where both her6 and her9 knock out bands were detected. Moreover, we show below (Author response image 1C, D) a comparison of our quantification using all the embryos or only those where both the her6 and her9 deletions have been confirmed by PCR. Additionally, we have sequenced the PCR products, confirming the deletion of those genes. Our results remain highly significant and we have updated Figure 3J and the methods section accordingly.

Author response image 1

Download asset Open asset

Validation of her6/her9 transient knock-out generation.

(A, B), Confirmation by PCR of her6 (A) and her9 (B) gene deletion after the injection of the Cas9 protein together with guides targeting her6 and her9 genes. Expected PCR product after deletion: 519 bp for her6 and 225 bp for her9. In some cases, due to repair not exactly as expected, different PCR product size can be detected. Only embryos were a her6 and her9 bands were detected were included in the analysis. (C) Original plot including all the imaged fish. (D) New plot including only the samples labeled in green.

Also, a control showing gRNA for an unrelated gene is needed.

We now use a control guide that was specifically designed to not target any zebrafish gene, named as ‘universal guide’ (1) (Figure 3J).

In Figure 3J it is not clear what the y-axis means. Is this a number or a fraction? The authors should clarify this.

We thank the reviewer for pointing this out. We have now clarified in the legend of Figure 3 and Figure 1 —figure supplement 2 that we quantify the number of adjacent jag1a-positive cells. Because of the geometry of the tissue, the maximum value is 2.

5. Regarding the mathematical modeling showing that patterning occurs when Kcis>Ktrans. It would be important to show that this conclusion is valid when changing other parameters. In particular:

a. Does this stay when cooperativity is assumes (n,m>1)

When we assume cooperativity, for example n, m = 2, patterning is also possible for Kcis < Ktrans. This agrees with published results (2–4). Importantly, we observe that with cooperativity, although a Kcis > Ktrans value is not required for patterning, it promotes patterning by making the homogeneous steady state more unstable. We have now added a new LSA analysis using different parameter values in a new supplementary figure (Figure 5 —figure supplement 2) and described our findings in detail in the last section of the results.

b. Does this conclusion stays when the relative expression of Notch and its ligand vary (if betaN>betaL or vice versa). I suspect that this is the case, but by showing that the conclusions from the model are valid for a larger parameter range would strengthen the claims.

We have also added new LSA plots with different relative values for betaN and betaL (Figure 5 —figure supplement 2), including betaN > betaL, betaN = betaL, and betaN < betaL. Interestingly, patterning is not always possible, but in all the cases, the MLE value is higher in a region where Kcis > Ktrans, indicating that a higher Kcis than Ktrans promotes patterning, independent of the value of other parameters.

6. Some minor comments/questions:

a. Would be nice to check whether Her6/9 direct targets of Notch. While I don't think it's necessary to perform additional experiments to do that, it would be good to check if there binding sites for RBPJ in the regulatory regions of Her6/9.

A recent publication (5) has analyzed RBPJ target genes in the zebrafish tailbud, a region of the fish that includes part of the notochord. We now refer to this publication and show that there are RBPJ binding sites close to the her6/her9 transcription start sites (Figure 2 —figure supplement 1N).

b. Figure S7 is written as S5 by mistake

We appreciate that the reviewer noticed this typo. Moreover, we have now changed the formatting to follow eLife guidelines.

c. Figure 5A: color code mismatch (Red and blue arrows match the legend but not the titles circuit1 and 2)

We have corrected this mistake.

d. It seems there may be a confusion in the text between the reference to Figure 5C, D,F,G,I and to 5E,H,J

We have corrected this mistake.

The data and methods are sufficiently detailed and the statistics and statistical analysis are adequate.

Significance: The main significance of the results in my mind are:

1. Establishment of the notochord as a system for studying lateral inhibition. The authors did an excellent job identifying the components of the circuit and their regulatory relations.

2. In particular, this is one of the relatively rare case showing lateral inhibition in an actual 1D configuration. I don't think there are many other characterized cases of lateral inhibition in 1D. The authors should point that out in the manuscript. I agree with the authors that this system would be a really nice system to quantitatively study lateral inhibition circuits in the future.

3. The observation that stronger cis-inhibition is required for patterning is new and interesting.

Overall the findings are new and the conclusions would certainly be interesting for the general community of developmental biology and in particular for researchers interested in quantitative developmental patterning processes (e.g. systems biology).

This review is based on my expertise in analyzing Notch mediated patterning, both on the experimental and theoretical sides.

Reviewer #2

Evidence, reproducibility and clarity:

The manuscript "Strength of interactions in the Notch gene regulatory network determines patterning and fate in the notochord" by H. Sanchez-Iranzo, A.Halavatvi and A. Diz-Muñoz presents computational and experimental results to study lateral inhibition patterning in the zebrafish notochord. The authors identify Jag1a as a crucial ligand and as a marker for vacuolated cell fate, in opposition to tp1 which the authors identify as a marker of sheath cell fate. The authors also identify her6 and her9 as repressors of jag1a, which are able to drive the sheath cell fate upon overexpression. They show that jag1a overexpression drives repression of her6/9 within the overexpressed cell and not in adjacent cells, suggesting jag1a is mediating cis-inhibition. These results are complemented with numerical simulations of lateral induction and lateral inhibition (with and without cis-inhibition) circuits in one dimensional arrays, together with linear stability analysis.

My major criticism is that I do not agree with the conclusion that the results show that cis-interactions are required to be stronger than trans-interactions. The modeling results of Figure 5B correspond to no cooperativity, as the authors state, and indicate that cis-interactions need to be stronger than transinteractions. Indeed these results are in perfect agreement with previous ones from analogous and from similar models (Ref. 21 and 22; this agreement should be indicated). However, for higher cooperativities cis-interactions do not need to be stronger than trans anymore, as previously published computational results have also shown. Indeed the lateral inhibition circuit without cis-interactions is sufficient to drive patterning if cooperativity is high enough (Ref. 13). Therefore, unless the authors demonstrate that there is no cooperativity participating in the process of cell fate choice in the notochord, it can not be predicted that cis-interactions need to be greater than trans. Therefore, this part of the manuscript should be reformulated (e.g. move the modeling part after the experimental part, such that the modeling proposes that cooperativity is expected to be low). Regarding the experimental data, knockdown experiments of jag1a/b could reinforce the conclusion that cis interactions are stronger in the wt than trans interactions, albeit are not required in my opinion.

We thank the reviewer for the thorough review of our manuscript and for pointing out the importance of identifying a novel system to study lateral inhibition patterning. In the revision below we aim to clarify all caveats and better experimentally link our findings to the modelling framework we propose.

As the reviewer points out, if cooperativity is assumed, patterning is possible with no cis interactions. In our manuscript we show, for the first time, that in the absence of cooperativity, stronger cis than trans interactions are required for patterning. As discussed in R1Q5, we have performed an extensive LSA analysis where we study patterning with different parameter values, including cooperativity. We observe that, regardless of cooperativity, a Kcis > Ktrans value promotes patterning by making the homogeneous steady state more unstable. We have included this new data in the manuscript (Figure 5 —figure supplement 2), and reformulated our claims to ‘stronger cis than trans interactions promote patterning’. Moreover, as suggested by the reviewer, we have changed the order of the experimental and modeling part.

Other minor issues:

a. In my opinion the role of jag1b should be further clarified. Despite its pattern of expression is more continuous than that of jag1a (as also indicated in Ref.28), results from Ref.28 showed that only when both jag1a and jag1b are knockdown, there are phenotypes of cell switch change. This may reflect that jag1b can take the role of jag1a in its absence or that- both are required in the wt for patterning. Hence, these hypotheses should be further investigated despite the more continuous pattern of expression. Alternatively, the manuscript should explicitly indicate that jag1b may be a relevant player that can not be completely discarded.

To study the role of these genes on patterning we have used previously validated jag1a and jag1b morpholinos (6). We have observed the strongest effect when both morpholinos are injected simultaneously (Figure 2D, E). In this experiment, tp1:GFP signal almost completely disappears specifically in the notochord, further supporting the fact that jag1a and jag1b are the most (if not the only) relevant Notch ligands in the notochord. In addition, we observe a spread of jag1a:mScarlet reporter to almost all the cells in the notochord (Figure 2D). This is probably the result of decreased Notch signaling, decreased her6/her9 expression, and as a result no repression of jag1a regulatory regions. Altogether, these results show the importance of jag1a/jag1b in patterning the notochord.

When we injected jag1a or jag1b morpholinos individually, we observed a similar effect, but to a lower extent (Figure 2B, C and E). These results suggest that jag1a and jag1b have similar roles on notochord patterning, but neither can compensate the loss of the other. We now specifically state that in lines 174-179.

b. It is well known that lateral inhibition drives alternated cell fates, while lateral induction does not. Therefore, computational results like those of Figure 1F are very well known (e.g. very similar can be found in Ref. 13). Those of Figure 1E are also known (Boareto et al. J R Soc Interface 2016; Matsuda et al. Science Signaling (2012), Ref. 49). Hence, the computational results from Figure 1E and F seem unnecessary to be included. The authors could just state the conclusions by referencing the literature. Yet, if they are preserved for clarity, references to the literature and emphasis that it is already known what each of these circuits drive has to be included.

We think that Figure 1 E and F are helpful to follow the results. Thus, we have rephrased this paragraph to make clearer that lateral inhibition and lateral induction networks have already been described in previous publications.

c. The authors analyse single cell RNA-seq data. How is the expression of jag1a, jag1b and tp1 in the cells where her6 and her9 are expressed and in those where emilin3a is expressed but not her6 nor her9?

The single-cell RNA-Seq data that we used is very helpful to identify genes that are expressed in the notochord when we group the cells. However, the depth of sequencing is too low to allow us to make reliable conclusions about specific cells. To address this reviewer’s comment, we have included new panels where we show her6, her9, jag1a, jag1b and emilin3a expression in each of the cells (Figure 2 —figure supplement 1A – F). In these plots, we can observe that emilin3a can be detected in most of the cells selected as notochord. However, it is difficult to make conclusions about the rest of the genes that are expressed at lower levels.

On the other hand, it is worth noting that tp1 is not an endogenous gene, but a regulatory sequence that has been shown to be responsive to Notch signaling. Thus, its expression cannot be evaluated in the single-cell RNA-Seq. We now further clarify this in the text (Section 1 results lines 162-163 and Methods lines 512-515).

d. Figure 1N: at which hpf?

We have added the stage in the figure legend (24 hpf).

e. Figure 3A-F: why when her6 is overexpressed there seems to be in total more cells expressing jag1? (Scarlet colour is much more present in images D and F than in B)

We thank the reviewer for pointing this out. Indeed, the image shown in Figure 3B was not very representative. We have replaced it by a more representative image.

f. Figure 5C-H: in embryos where jag1 is overexpressed, there seems to be less overall expression of her6 and her9 compared to control (images G and H show much less red and cyan intensity than D and E). How the authors explain this?

We thank the reviewer for pointing out this artefact. This was caused by the non-optimal adjustment of channel intensities. We have now adjusted better the intensities. Nevertheless, it is worth noting that the images used for quantification were adjusted automatically within the python pipeline already in the original version of the manuscript. Specifically, the intensities of each of the channels were normalized between 0 and 1, where 0 was assigned to the minimum intensity value in the image, and 1 to the maximum value.

g. The first time tp1 appears mentioned in the manuscript, it should be introduced: i.e. to indicate what it is and what it is known about it.

We have now added a sentence introducing the tp1 promoter (Section 1 results lines 162-163 and Methods lines 512-515).

h. In section 1 of results, where it indicates "We first modelled a lateral inhibition network as " should be changed to "We first modelled a lateral induction network as ".

We thank the reviewer for pointing out this typo. We have now corrected it in the manuscript.

Significance:

My expertise is on modeling. The experimental results of the manuscript are novel to my knowledge and provide a rather complete framework to characterize and understand binary cell fate choices (vacuolated and sheath cell fates) in the notochord of zebrafish embryos. As the authors indicate, ref. 28 (from 2010), established that Notch signaling by Jag1 (1a and 1b), and through her9, control the cell fate switch between vacuolated and sheath cell fates. In this manuscript, the authors add another HER/HES actor, her6, and show its complementary pattern with jag1a and that it can repress it. In addition, the authors conceptualize their finding within the framework of lateral inhibition and propose jag1a as the relevant player. The manuscript is very well organized and written. The results seem appropriately analyzed and the modeling framework and results seem correct. Yet, the computational results are less novel (see report above). Overall I find it a clear manuscript that clarifies and sets a new system to study lateral inhibition patterning. It can be very well suited to a developmental biology audience.

Reviewer #3

Evidence, reproducibility and clarity:

Summary:

The paper proposes a new model for the specification of vacuolated vs. sheath cell fate in the zebrafish notochord. In doing so, it further establishes a model system for understanding the molecular mechanisms underpinning cell fate decision making in vertebrate development, and to explore the regulatory relationships of the Notch signalling pathway. In particular, the study reveals how Jag1 can not only function in the context of a lateral induction mechanism, but also in producing lateraly inhibition when inhibited by her6 and her9. The authors also clone an emilin3a promotor element that is sufficient to drive robust expression in the notochord of zebrafish embryos.

We thank the reviewer for the review of our manuscript and for highlighting the importance of establishing the notochord as a model system to better explore the mechanism of action of Notch signaling in a vertebrate context. Below we aim to address her/his concerns.

Major comments:

– Are the key conclusions convincing? The conclusions are convincing and well supported by experimental data.

– Should the authors qualify some of their claims as preliminary or speculative, or remove them altogether? All claims are valid and well supported, assuming suitable n numbers are presented for the functional experiments in Figures 3 and 4.

– Would additional experiments be essential to support the claims of the paper? Request additional experiments only where necessary for the paper as it is, and do not ask authors to open new lines of experimentation. The existing conclusions of the paper would be significantly strengthened by loss-offunction experiments for Jag1.

As described in the answer to point “a” from reviewer 2:

“To study the role of these genes on patterning we have used previously validated jag1a and jag1b morpholinos (6). We have observed the strongest effect when both morpholinos are injected simultaneously (Figure 2D, E). In this experiment, tp1:GFP signal almost completely disappears specifically in the notochord, further supporting the fact that jag1a and jag1b are the most (if not the only) relevant Notch ligands in the notochord. In addition, we observe a spread of jag1a:mScarlet reporter to almost all the cells in the notochord (Figure 2D). This is probably the result of decreased Notch signaling, decreased her6/her9 expression, and as a result no repression of jag1a regulatory regions. Altogether, these results show the importance of jag1a/jag1b in patterning the notochord.

– Are the suggested experiments realistic in terms of time and resources? It would help if you could add an estimated cost and time investment for substantial experiments. Approximately 3 months would be required to generate morpholinos against Jag1, and to perform experiments alongside appropriate controls to demonstrate phenotypic rescue.

– Are the data and the methods presented in such a way that they can be reproduced? Yes, the results are very well presented with an adequate description of the methods.

– Are the experiments adequately replicated and statistical analysis adequate? This is not clear for Figures 3 and 4. The n numbers need to be clearly written in the figure legends, as for the previous figures.

We have now explicitly mentioned the numbers of fish for Figures 3 and 4 in the figure legends, as well as their corresponding supplementary figures.

Minor comments:

– Specific experimental issues that are easily addressable. N/A

– Are prior studies referenced appropriately? Yes

– Are the text and figures clear and accurate? Yes.

– Do you have suggestions that would help the authors improve the presentation of their data and conclusions?

1. Results section 4: It would be helpful to describe more details about her6 and her 9 gene deletions.

We describe in more detail the design of her6 and her9 deletions in a new panel in Figure 2 —figure supplement 1N.

2. Results section 5, paragraph 1. The primary evidence is a loss of vaculated cell fate, with an assumed increase in sheath cells. The conclusion should be re-written accordingly.

We thank the reviewer for pointing this out and agree that our conclusions were misleading. We have clarified this in the text and in the figure legend (Figure 4) by defining the proportion of vacuolated cells as the number of vacuolated cells divided by the sum of sheath and vacuolated cells. Thus, a lower proportion of vacuolated cells is equivalent to a higher proportion of sheath cells.

3. Results section 6, paragraph 1. The overexpression of full length ligand assumes that Notch levels are not limiting in the system to be able to observe and effect. Have the authors considered this alternative explanation for an absence in phenotype?

We indeed cannot exclude the possibility that Notch levels are limiting. We now discuss this possibility in the text (Section 6 results, lines 326-327). However, it is worth noting that both the results of our modelling and prior knowledge from the mammalian and Drosophila homologs (Jag1/Serrate have been shown to be weak Notch activators, but strong cis-inhibitors) support our conclusion that stronger cis than trans interactions is a relevant mechanism of action of Notch signaling in vertebrates.

Significance

– Describe the nature and significance of the advance (e.g. conceptual, technical, clinical) for the field: The Notch pathway is an essential signalling pathway in multiple contexts and it is therefore very important to increase our understanding of its mechanism of action. This work presents interesting findings in this regard, and further establishes a model system to better explore its mechanism of action in a vertebrate context.

– Place the work in the context of the existing literature (provide references, where appropriate): Notch function has been studied in the context of notochord development here:

Norman, J., Sorrell, E.L., Hu, Y., Siripurapu, V., Garcia, J., Bagwell, J., Charbonneau, P., Lubkin, S.R., and Bagnat, M. (2018). Tissue self-organization underlies morphogenesis of the notochord. Philos. Trans. R. Soc. B Biol. Sci. 373.

Yamamoto M, Morita R, Mizoguchi T, Matsuo H, Isoda M, Ishitani T, Chitnis AB, Matsumoto K, Crump JG, Hozumi K, Yonemura S, Kawakami K, Itoh M. Mib-Jag1-Notch signalling regulates patterning and structural roles of the notochord by controlling cell-fate decisions. Development. 2010 Aug 1;137(15):2527-37. doi: 10.1242/dev.051011. Epub 2010 Jun 23. PMID: 20573700; PMCID: PMC2927672.

– State what audience might be interested in and influenced by the reported findings: Zebrafish developmental biologists, anyone interested in Notch signalling.

– Define your field of expertise with a few keywords to help the authors contextualize your point of view. Indicate if there are any parts of the paper that you do not have sufficient expertise to evaluate: Zebrafish developmental biology and notochord development.

References

1. W. A. Wierson, et al., Efficient targeted integration directed by short homology in zebrafish and mammalian cells. ELife 9, 1–25 (2020).

2. D. Sprinzak, et al., Cis-interactions between Notch and Δ generate mutually exclusive signalling states. Nature 465, 86–90 (2010).

3. D. Sprinzak, A. Lakhanpal, L. LeBon, J. Garcia-Ojalvo, M. B. Elowitz, Mutual inactivation of Notch receptors and ligands facilitates developmental patterning. PLoS Comput. Biol. 7 (2011).

4. P. Formosa-Jordan, M. Ibañes, Competition in notch signaling with cis enriches cell fate decisions. PLoS One 9 (2014).

5. Z. Ye, C. R. Braden, A. Wills, D. Kimelman, Identification of in vivo Hox13-binding sites reveals an essential locus controlling zebrafish brachyury expression. Dev. 148 (2021).

6. M. Yamamoto, et al., Mib-Jag1-Notch signalling regulates patterning and structural roles of the notochord by controlling cell-fate decisions. Development 137, 2527–2537 (2010).

https://doi.org/10.7554/eLife.75429.sa2

Article and author information

Author details

Héctor Sánchez-Iranzo

Cell Biology and Biophysics Unit, European Molecular Biology Laboratory, Heidelberg, Germany

Present address
Institute of Biological and Chemical Systems – Biological Information Processing, Karlsruhe Institute of Technology, Eggenstein-Leopoldshafen, Germany

Contribution
Conceptualization, Funding acquisition, Investigation, Methodology, Software, Visualization, Writing – original draft, Writing – review and editing

For correspondence
hector.sanchez@kit.edu

Competing interests
No competing interests declared

"This ORCID iD identifies the author of this article:" 0000-0003-2032-0231
Aliaksandr Halavatyi

Advanced Light Microscopy Facility, European Molecular Biology Laboratory, Heidelberg, Germany

Contribution
Methodology, Software, Visualization

Competing interests
No competing interests declared
Alba Diz-Muñoz

Cell Biology and Biophysics Unit, European Molecular Biology Laboratory, Heidelberg, Germany

Contribution
Conceptualization, Funding acquisition, Supervision, Writing – original draft, Writing – review and editing

For correspondence
diz@embl.de

Competing interests
No competing interests declared

"This ORCID iD identifies the author of this article:" 0000-0001-6864-8901

Funding

EMBO (ALTF 306-2018)

Héctor Sánchez-Iranzo

Deutsche Forschungsgemeinschaft (DI 2205/3-1)

Alba Diz-Muñoz

Deutsche Forschungsgemeinschaft (DI 2205/2-1)

Alba Diz-Muñoz

European Molecular Biology Laboratory

Alba Diz-Muñoz

Joachim Herz Stiftung (Add-on Fellowship for Interdisciplinary Life Science)

Héctor Sánchez-Iranzo

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

We thank Anna Erzberger, Aissam Ikmi and Stefano de Renzis for critical reading of the manuscript. We thank Jonas Hartmann for discussion on the project and training on image analysis. We are grateful to the EMBL EM core facility (EMCF), and in particular to Rachel Mellwig and Yannick Schwab, for the EM experiments. We thank the EMBL Fish Facility, and in particular to Sabine Görgens. We thank the EMBL Advanced Light Microscopy Facility, and especially Christian Tischer and Stefan Terjung for image analysis and microscopy support. We thank Alexander Ernst for the development of the custom-made ImageJ macro used to generate some of the movies of the paper. We thank the Life Science Editors for editorial support. Funding: This study was funded by the European Molecular Biology Laboratory (EMBL) and Deutsche Forschungsgemeinschaft (DFG) grants DI 2205/3–1 and DI 2205/2–1 to AD-M. H S-I was funded by the EMBO fellowship (ALTF 306–2018) and the Joachim Herz Stiftung Add-on Fellowship for Interdisciplinary Science.

Ethics

All experiments were performed on embryos younger than 3 dpf, as is stipulated by the EMBL internal policy 65 (IP65) and the European Union Directive 2010/63/EU.

Senior and Reviewing Editor

Marianne E Bronner, California Institute of Technology, United States

Publication history

Preprint posted: March 25, 2021 (view preprint)
Received: November 11, 2021
Accepted: April 28, 2022
Version of Record published: June 6, 2022 (version 1)

Copyright

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.