# Asymmetric neurogenic commitment of retinal progenitors involves Notch through the endocytic pathway

1. Max Planck Institute of Molecular Cell Biology and Genetics, Germany
2. Instituto Gulbenkian de Ciência, Portugal
Research Article

## Abstract

During brain development, progenitor cells need to balanceproliferation and differentiation in order to generate different neurons in the correct numbers and proportions. Currently, the patterns of multipotent progenitor divisions that lead to neurogenic entry and the factors that regulate them are not fully understood. We here use the zebrafish retina to address this gap, exploiting its suitability for quantitative live-imaging. We show that early neurogenic progenitors arise from asymmetric divisions. Notch regulates this asymmetry, as when inhibited, symmetric divisions producing two neurogenic progenitors occur. Surprisingly however, Notch does not act through an apicobasal activity gradient as previously suggested, but through asymmetric inheritance of Sara-positive endosomes. Further, the resulting neurogenic progenitors show cell biological features different from multipotent progenitors, raising the possibility that an intermediate progenitor state exists in the retina. Our study thus reveals new insights into the regulation of proliferative and differentiative events during central nervous system development.

## Introduction

During central nervous system (CNS) development in vertebrates, the balance between proliferation and differentiation is crucial to generate the different types of neurons in the correct numbers and proportions. This in turn ensures functionality of the neural tissue as it is the prerequisite for correct neuronal connectivity and circuit formation. Multipotent progenitors are key to maintain this balance. They first increase in number through a phase of active proliferation (Fish et al., 2006; Homem et al., 2015) which is followed by neurogenesis onset. At neurogenesis onset, subsets of progenitors leave the cell cycle and differentiate into neurons (G?tz and Huttner, 2005; Agathocleous and Harris, 2009). Expectedly, impairment of the intricate balance between proliferation and differentiation can lead to severe defects of brain development and is associated with numerous human diseases, including microcephaly and tumour formation (Fish et al., 2006; Thornton and Woods, 2009). Thus, understanding the parameters that differ between cells primed for continued proliferation versus cells that enter the neurogenic program is of utmost importance to comprehend brain development in health and disease.

Previous studies using the mammalian neocortex showed that at the onset of neurogenesis multipotent progenitors start dividing asymmetrically, generating one radial glial cell and one neuron or a basal progenitor and thereby balancing the ratio between differentiation and proliferation at early neurogenesis stages (Chenn and McConnell, 1995; Huttner and Brand, 1997; G?tz and Huttner, 2005; Shen et al., 2006; Shitamukai and Matsuzaki, 2012). Once a sufficient number of progenitors is produced, final symmetric consumptive neuronal divisions deplete the progenitor pool (Miyata et al., 2004; Noctor et al., 2004). Similarly to the mammalian cortex, asymmetric divisions have also been shown to be important in the fore- and hindbrain of the teleost Danio rerio (zebrafish) to ensure the maintenance of the progenitor pool before its final depletion (Alexandre et al., 2010; Dong et al., 2012). Thus, while some questions remain, the division modes responsible for the generation of the first neurons in these parts of the nervous system are relatively well explored. What is much less understood, however, is how neurogenic progenitors arise from the multipotent progenitor pool in the first place. Especially in the retina, the tissue that upon maturation enables the perception of visual information, this question has not yet gathered sufficient attention. Here, division patterns have mainly been investigated at later stages when the emergence of the seven different neurons that ensure retinal functionality has already commenced (Poggi et al., 2005; He et al., 2012; Kechad et al., 2012; Suzuki et al., 2013). It was shown that these late neurogenic divisions can be asymmetric, generating two neurons of different fate. However, how multipotent progenitors enter the neurogenic path in the first place and what type of division patterns and regulatory pathways these initial fate decisions entail is not yet well explored.

In many parts of the CNS, including the retina, asymmetric divisions leading to the formation of neurons have often been linked to the asymmetric inheritance of distinct cellular components. Examples of differentially inherited components are the basal process (Saito et al., 2003; Okamoto et al., 2013), the apical domain (Bultje et al., 2009; Alexandre et al., 2010) or polarity components (Paolini et al., 2015; Zhao et al., 2020). In addition, signalling pathways including Wnt (Hirabayashi et al., 2004) and very prominently Notch (Dorsky et al., 1995;?Dorsky et al., 1997; Cayouette et al., 2001; Ohtsuka et al., 2001; Dong et al., 2012) have been shown to be involved. Notch, for example, can regulate the asymmetry of fate decisions through the asymmetric inheritance of Notch signalling endosomes, as shown in the zebrafish spinal cord and forebrain (Kressmann et al., 2015; Zhao et al., 2020). In the vertebrate retina, Notch signalling maintains progenitors in a proliferative state through lateral inhibition (Henrique et al., 1997; Jadhav et al., 2006). It was previously suggested that here Notch regulates neurogenic commitment of multipotent progenitors via an extracellular gradient along the apicobasal axis of the elongated neuroepithelial cells (Murciano et al., 2002; Del Bene et al., 2008). It was postulated that when nuclei travel along the apicobasal axis during the cell cycle (a phenomenon termed interkinetic nuclear migration, IKNM [Sauer, 1935; Baye and Link, 2007; Norden et al., 2009]), they are exposed to different Notch signalling levels, that in turn influence the progenitors’ probability to enter the neurogenic program (Murciano et al., 2002; Baye and Link, 2007; Del Bene et al., 2008). However, the division patterns of retinal progenitors that lead to neurogenic entry and their possible links to Notch signalling were not fully explored.

We here investigate the division patterns that give rise to the neurogenic progenitors, one cell cycle before the generation of post-mitotic neurons. We find that divisions generating neurogenic progenitors are almost always asymmetric and that sister cells do not enter the same lineage. While Notch is involved in the generation of this asymmetry, it does not act via an apicobasal activity gradient. Instead, we provide evidence that Notch signalling acts through the asymmetric inheritance of Sara-positive endosomes. We further show that these first neurogenic progenitors differ in their cell biological features from multipotent progenitors, indicating that they are an intermediate cell type en route to differentiation.

## Results

### Ath5 positive neurogenic progenitors arise from asymmetric divisions

To understand how neurogenic progenitors emerge from multipotent progenitors in the developing zebrafish retina, we asked how the earliest of these neurogenic progenitors arise. Possible division patterns were symmetric, asymmetric or a mixture of the two. To follow these earliest neurogenic progenitors, we made use of the well-established fact that they express the transcription factor ath5 (atoh7) (Poggi et al., 2005; Zolessi et al., 2006) and traced their origin by mosaically labelling cells in zebrafish embryos with an Ath5-driven reporter construct (Icha et al., 2016a). We followed the onset of the Ath5 signal using light-sheet imaging from the beginning of neurogenesis at 28 hpf (hours post-fertilisation; Hu and Easter, 1999; Schmitt and Dowling, 1999; Martinez-Morales et al., 2005) as previously described (Icha et al., 2016b). We backtracked progenitors that began to express Ath5 (from here on referred to as ‘Ath5 positive progenitors’; Figure 1A, Figure 1—figure supplement 1A) by labelling the plasma membrane and/or the nucleus of the same cells using hsp70:mKate2-ras or hsp70:H2B-RFP constructs (Figure 1B). Sister cells of divisions giving rise to at least one Ath5 positive neurogenic progenitor were backtracked and their division pattern was analysed (Figure 1C,C’).

Figure 1 with 3 supplements see all

Backtracking of 38 multipotent progenitor divisions from nine embryos (Figure 1B, Figure 1—video 1) revealed that 92% of these divisions were asymmetric, generating one Ath5 positive (Ath5+) progenitor and one Ath5 negative (Ath5-) progenitor (Figure 1C,C’,D). Only 8% of divisions were symmetric, producing two Ath5+ cells (Figure 1C’,D). Following Ath5- sister cells for another cell cycle showed that all cells divided again, but none of the daughter cells started expressing Ath5 (n?=?5 divisions, N?=?2 embryos; Figure 1—figure supplement 1B,C, Figure 1—video 1). This suggests that Ath5- sister cells enter a different lineage than their Ath5+ counterpart. Together, these data provide first direct evidence that neurogenic progenitors in the retina emerge from asymmetric divisions, indicating that cellular asymmetry plays an important role in the selection of progenitors that enter the neurogenic program.

### Inhibition of Notch signalling affects the pattern of divisions generating Ath5 positive progenitors

To understand the asymmetry of divisions producing Ath5+ progenitors at the molecular level, we set out to investigate the signalling factors involved. One promising candidate that drives asymmetric divisions in many neurogenic lineages is the Notch pathway (Bultje et al., 2009; Egger et al., 2010; Kechad et al., 2012; Kressmann et al., 2015). In the retina, Notch signalling has been shown to maintain progenitors in the proliferative state thus inhibiting neurogenesis (Bao and Cepko, 1997; Jadhav et al., 2006; Kechad et al., 2012; Maurer et al., 2014; Sato et al., 2016). We therefore tested whether and how Notch depletion influenced the asymmetry of divisions generating Ath5+ progenitors using two established Notch inhibitors, the γ-secretase inhibitors LY411575 (Figure 1E,F) and DAPT (Figure 1—figure supplement 1D; Ninov et al., 2012; MacDonald et al., 2015). In line with the reports that Notch signalling suppresses neurogenesis and thereby Ath5 expression onset (Maurer et al., 2014), we found that both Notch inhibitors led to an increase in the number of Ath5+ cells at 36 and 48 hours post fertilisation (hpf) (Figure 1E,F). LY411575 treatment further led to a decrease in the number of EdU-positive proliferative cells (Figure 1—figure supplement 1E,F). The increase of Ath5+ cells was accompanied by a striking reduction in tissue thickness (Figure 1G,G’) and retinal diameter (Figure 1G,G’’) compared to controls at 48 hpf. This retinal size reduction showed that indeed the balance between proliferation and differentiation in the retina was disturbed upon Notch inhibition, leading to tissue-wide consequences.

The increase in the number of Ath5+ progenitors observed upon Notch inhibition could arise from two not mutually exclusive scenarios: Ath5+ progenitors could emerge prematurely by asymmetric divisions that absorb multipotent progenitors too early, or an increase in symmetric divisions giving rise to two Ath5+ progenitors could occur that would result in less progenitors entering another cell cycle, thereby also depleting the pool of remaining progenitors prematurely (Figure 1C,C’). To explore these possibilities, live imaging of LY411575 treated embryos was performed (Figure 1H). Compared to controls, Notch inhibition led to a significant increase in symmetric neurogenic divisions that generated two Ath5+ progenitors (48% compared to 8% in controls, Figure 1I, Figure 1—video 2). Thus, Notch signalling regulates the asymmetry of the division that produces neurogenic Ath5+ progenitors. Interference with Notch and thereby this division asymmetry?influences neurogenesis patterns and tissue maturation in the developing retina.

### Ath5 positive progenitors generally translocate more basally than Ath5 negative sister cells but a spatial bias is not observed at the cell population level

The selection of the sister cell that enters the neurogenic program could be regulated by differential exposure to the extracellular environment that hosts different fate determinants as shown for other systems (Yamashita and Fuller, 2005; Knoblich, 2008). It was previously reported that Notch RNA is distributed in a graded manner along the cell’s apicobasal axis with more Notch RNA detected towards apical positions in the retinal neuroepithelium (Murciano et al., 2002; Del Bene et al., 2008). This outcome led to the idea that distinct Notch signalling milieus exist along the apicobasal axis of the retinal neuroepithelium and that these could influence neurogenic outcome due to nuclear movement along this axis during IKNM (Murciano et al., 2002; Baye and Link, 2007; Del Bene et al., 2008). Based on these studies, we asked whether the observed cell division asymmetry that generates Ath5+ progenitors (Figure 1C) could be explained by differential translocation depths between Ath5+ and Ath5- sister cells. To this end, we compared trajectories of Ath5+ and Ath5- sister cells after division (Figure 2A). We found that indeed one sister cell soma always translocated more basally than the other (Figure 2B), with an average difference of 10.11?±?7.292 μm between the maximal basal positions (n?=?35 divisions, N?=?9 embryos; Figure 2C). In 78% of sister cells, the more basally translocating sister cell started expressing Ath5 (27/35 divisions; Figure 2D) while in 22% of divisions the Ath5- cell positioned the soma more basally (Figure 2E). In the latter case, however, the difference between maximum basal position of sister cells was on average less prominent than in the reverse scenario (difference of 6.048?±?3.559 μm when Ath5- soma is most basal compared to 11.31?±?7.712 μm when Ath5+ soma is most basal; Figure 2F).

Figure 2 with 2 supplements see all

To probe whether a basal positioning threshold could be defined that predicts the onset of Ath5 expression, all trajectories were pooled (Figure 2G,H) To our surprise, this analysis showed a significant overlap between trajectories and maximum basal positions for Ath5+ and Ath5- somas without clearly distinct Ath5+ and Ath5- zones (Figure 2G,H,I, Figure 2—figure supplement 1A). The measurement of the average depth occupied by somas of Ath5+ and Ath5- cells further confirmed that the two populations did not occupy distinguishable basal positions (integral mean, Figure 2—figure supplement 1B).

These results show that basal translocation differs between sister cells and that the cell becoming Ath5+ usually translocated to more basal positions. Nevertheless, basal position was not a clear-cut readout for Ath5 expression onset as an overlap zone existed between Ath5+ and Ath5- cells.

### Asymmetric neurogenic commitment is independent of maximum basal position

The large overlap detected for basal positioning of Ath5+ and Ath5- sister cell soma argued against a direct link between basal translocation and neurogenic commitment. To show this more directly, we set out to equalise the basal translocation pattern after apical divisions of sister cells. To achieve this, we used a heat shock inducible dominant negative version of dynactin (Norden et al., 2009), as dynactin downregulation has been shown to induce increased basal translocation of retinal progenitor cell soma (Del Bene et al., 2008; Norden et al., 2009; Figure 2J). Heat shock expression of the hsp70:DNdynactin-mKate2 construct led to the translocation of both sister cell somas to greater and almost identical basal positions compared to controls (Figure 2K, Figure 2—video 1, Figure 2—figure supplement 1E). However, despite the fact that both sister cell somas reached very basal positions, only one became Ath5+ while the other remained Ath5- (n?=?10, N?=?5) (Figure 2L, Figure 2—video 1).

These results suggest?that the asymmetry of neurogenic commitment is not linked to differential somal translocation of sister cells along the apicobasal axis.

### Notch pathway components are uniformly distributed along the apicobasal axis of the retinal neuroepithelium

Our results so far argue against a graded effect of Notch signalling on the asymmetry of divisions producing Ath5+ progenitors. Thus, we probed whether the previously reported Notch gradient at the RNA level (Del Bene et al., 2008; data not shown) was also found at the level of the signalling proteins.

We performed immunostaining against two components of the Notch pathway, the receptor Notch1b and the ligand DeltaC. We found that both Notch pathway components appeared as puncta that were uniformly distributed along the apicobasal axis of the retinal neuroepithelium (Figure 3A,A’,A’’ and Figure 3B,B’). The specificity of the Notch1b and DeltaC antibodies was confirmed by somite labelling (Figure 3—figure supplement 1C). Further, antibody staining in notch1b and deltaC morphant embryos showed a disappearance respectively a strong reduction of labelling along the apicobasal axis of the neuroepithelium (Figure 3—figure supplement 1A,B). We further analysed the distribution of Notch signalling activity using two previously established reporter lines: Tg(her4.1:mcherry-CreERT2), a marker for the gene her4, a direct Notch target (Takke et al., 1999), and Tg(Tp1bglob:H2BmCherry), a marker for a secondary Notch target (Souilhol et al., 2006). This analysis confirmed that Ath5- cells that display active Notch can occupy diverse positions along the apicobasal axis of the neuroepithelium (Figure 3C,D and Figure 3—video 1). Interestingly, Ath5+ progenitors never seemed to express Tp1 or her4 (Figure 3C,D, arrowheads), while Ath5- cells were seen to express both of these markers. Quantification of cells with Tp1 expression confirmed this notion, as we found that 210/215 of Ath5+ cells showed no Tp1 expression while Tp1 expression was very weak in the remaining 5 cells (n?=?215 cells, N?=?4 embryos; Figure 3—figure supplement 2A,A’). Thus, Ath5+ and Ath5- sister cells experience different levels of Notch signalling activation, independently of their position along the apicobasal axis.

Figure 3 with 3 supplements see all

Overall, these results, combined with the fact that asymmetric Ath5 expression onset persisted even when both sister cell somas translocated very basally, argue that Notch signalling does not act via an apicobasal signalling gradient for setting up the asymmetry of divisions leading to neurogenic commitment.

### Basal somal translocation of Ath5 positive progenitors depends on a combination of basal process inheritance and stabilised microtubules

While we show that somal positioning along the apicobasal axis does not influence the asymmetry of divisions producing Ath5+ progenitors, we nevertheless found differences in basal translocation potential between Ath5+ and Ath5- sister cells. In the majority of cases, the cell that later starts expressing Ath5 (Ath5+ progenitor) translocated more basally (Figure 2D). We thus asked what other factors could facilitate this enhanced basal translocation pattern of Ath5+ progenitors.

To this end, we compared basal translocation kinetics of Ath5+ and Ath5- progenitor soma and performed mean square displacement (MSD) analysis. This analysis revealed that Ath5+ basal somal translocation is more directed than that of Ath5- soma (Figure 4B), a finding confirmed by directionality ratio analysis (Figure 2—figure supplement 1C). Furthermore, Ath5+ sister cell soma showed a higher average velocity during basal translocation (Figure 2—figure supplement 1D), indicating that Ath5+ progenitors translocate more efficiently towards the basal side.

Figure 4 with 1 supplement see all

Basal process inheritance is known to facilitate efficient basal translocation of progenitors and emerging neurons in neocortex and retinal neuroepithelia (Saito et al., 2003; Icha et al., 2016a). We thus asked whether basal process inheritance was correlated with faster basal displacement of Ath5+ soma (Figure 4A). Indeed, in 15 out of 16 cell pairs analysed, the cell that inherited the basal process initially translocated its soma more basally (Figure 4C). However, only in 10/16 cases (60%), did that cell also become the Ath5+ progenitor (Figure 4D). Thus, it was possible that additional factors were responsible for the more efficient basal somal translocation of Ath5+ cells. It was previously shown that stabilised microtubules emanating from the apically positioned centrosome facilitate somal translocation of emerging retinal ganglion cells, the first neurons born in the retina (Icha et al., 2016a). To test whether stabilised microtubules could also drive translocation of Ath5+ progenitors, we stained for acetylated tubulin in Tg(ath5:gap-GFP/RFP) embryos (Zolessi et al., 2006) at stages before and during Ath5+ progenitors emergence, 24, 28, 32, and 36 hpf (Figure 4E). This revealed that at 24 hpf, the developmental stage prior to neurogenesis onset, acetylated tubulin was seen only in the primary cilium of progenitor cells (Figure 4E, top left). However, once neurogenesis started, acetylated tubulin was also observed in the apical process of progenitor cells (Figure 4E top right and bottom left, 28 and 32 hpf) as well as in retinal ganglion cells at 36 hpf (Figure 4E bottom right), as previously reported (Icha et al., 2016a).

To examine whether these stabilised microtubules were directly involved in the basal somal translocation of neurogenic progenitors, we backtracked Ath5+ progenitors overexpressing the microtubule-destabilising protein Stathmin-1 using an hsp70:Stathmin1-mKate2 construct (Figure 4F, Figure 4—video 1; Icha et al., 2016a). The soma of Ath5+ progenitors expressing Stathmin-1 did not translocate as basally as the soma of control cells (Figure 4I,K). In contrast, the basal somal translocation of Ath5- progenitors was not affected (Figure 4H,J), indicating that only progenitors that enter the neurogenic state use stabilised microtubules for basal somal translocation. This experiment further confirmed that Ath5 expression onset is independent of somal positioning along the apicobasal axis. While both sister cells remained at very apical positions after division, only one became an Ath5+ progenitor (Figure 4G, 10/10 divisions). This is the opposite scenario to the dominant negative dynactin condition in which cell nuclei translocated more basally (Figure 4J,K). Thus, the asymmetry of Ath5 expression onset persisted independently of sister cell basal or apical positions (Figure 2L, Figure 2—video 1, Figure 4—video 1).

Overall, these experiments showed that the greater basal somal translocation of Ath5+ progenitors is driven by a combination of basal process inheritance and the presence of apically enriched stabilised microtubules. This reflects other examples of neuronal basal somal translocation in CNS development (Yuasa et al., 1996; Umeshima et al., 2007; Cooper, 2013; Sakakibara et al., 2014; Icha et al., 2016a).

### Notch signalling affects neurogenic commitment through Sara endosomes

We conclude that the greater basal translocation of progenitors that start expressing Ath5 was driven by a combination of basal process inheritance and a stabilised microtubule network. However, these findings do not yet explain the involvement of Notch signalling in the asymmetry of neurogenic commitment independently of a Notch activity gradient. It was shown that Notch signalling components can segregate asymmetrically during division of the mother cell in other systems (Le Borgne and Schweisguth, 2003; Dong et al., 2012; Kechad et al., 2012; Coumailleau et al., 2009), which leads to the inheritance of different levels of Notch by the two daughter cells. In particular, Sara-positive endosomes (Sara endosomes) that carry active Notch via the endocytosis of Delta receptors and Notch ligands were shown to be involved in asymmetric neurogenic fate decisions in the Drosophila sensory organ precursor and the zebrafish spinal cord (Coumailleau et al., 2009; Kressmann et al., 2015).

To test whether Sara?endosomes could also be involved in divisions generating Ath5 positive progenitors, we assessed whether Sara-?endosomes are present in retinal progenitors at stages at which Ath5+ progenitors emerge. Embryos were injected with mRFP-Sara mRNA and stained for Notch1b and DeltaC. We found that these components indeed co-localise in retinal progenitors (Figure 5A,A’ and Figure 5B,B’). To understand whether Sara?endosomes are asymmetrically distributed between progenitors upon division as suggested elsewhere (Coumailleau et al., 2009; Kressmann et al., 2015), we injected mRFP-Sara RNA into membrane labelled Tg(actb1:HRAS-EGFP) embryos. At all developmental stages analysed, 24, 28 and 36 hpf, asymmetric distribution of Sara?endosomes was observed in dividing cells (Figure 5C). Cells mainly contained 1–3 endosomes and rarely more than that (Figure 5D, n?=?25 dividing cells, N?=?7 embryos). In most cases, one of the two sister cells inherited all endosomes (Figure 5D’),suggesting that the majority of Sara endosome related Notch signalling is dispatched to only one sister cell. We further confirmed that Tp1-expressing cells carry Sara endosomes (Figure 3—figure supplement 2B) meaning that they feature active Notch signalling through the endosomal pathway at 36 hpf. Furthermore, light-sheet live imaging confirmed asymmetric distribution of Sara-?endosomes during multipotent retinal progenitor divisions (Figure 5E,E’). These experiments indicated that Notch signalling is asymmetrically partitioned through Sara-positive endosomes. Thus, Sara-endosomes inheritance could mediate asymmetric neurogenic commitment in the retina. Our finding that Ath5+ progenitors never displayed active Notch signalling (Figure 3C,D, Figure 3—figure supplement 2A) is in agreement with this notion. Combined, these findings argue that the depletion of Sara endosomes is linked to Ath5+ fate. We propose that Sara endosomes are mostly inherited by the Ath5- cell, which retains high Notch signalling and thus stays in the proliferative state (see working model in Figure 5G and cell in Figure 5—figure supplement 1E, Figure 5—video 1).

Figure 5 with 2 supplements see all

To further test the involvement of Sara endosomes in asymmetric neurogenic commitment, we depleted the Sara protein using an established Sara splice morpholino (Kressmann et al., 2015). We found that the number of Ath5+ cells was greatly reduced compared to controls upon Sara inhibition, leading to the presence of fewer neurons (Figure 5F, Figure 5—figure supplement 1A). At the same time, the number of Tp1-positive cells increased (Figure 5F’) suggesting an increase in active Notch signalling in both daughter cells. This is the opposite scenario of what is observed upon Notch signalling inhibition which led to the generation of more neurogenic progenitors through an increase in symmetric neurogenic divisions (Figure 1E,H,I). We excluded the possibility that this decrease in Ath5+ progenitor emergence is due to cells not being able to divide and progress through the cell cycle before neurogenesis, as pH3 staining showed similar amounts of mitotic cells in Sara morphants as in control embryos (Figure 5—figure supplement 1B). Furthermore, Sara morphants showed a higher number of EdU-positive cells than embryos injected with control morpholino, suggesting that indeed more cells remained proliferative (Figure 5—figure supplement 1C,D).

We thus speculate that depletion of Sara led to more symmetric proliferative divisions generating two Ath5- cells. This finding differs from an earlier study in which Sara depletion led to an increase of neuronal production (Kressmann et al., 2015). However, in this study a different neuronal lineage was investigated, raising the possibility that diverse mechanisms of Sara-mediated regulation of Notch exist that are dependent on tissue and lineage context (see discussion). Overall, these results suggest that Sara-mediated intracellular Notch signalling is necessary to create asymmetries in Notch activity among progenitors, thereby selecting the progenitors that enter the neurogenic path.

## Discussion

We here used long-term live imaging to explore the emergence of neurogenic progenitors in the vertebrate retina. Our study reveals that early neurogenic commitment arises via asymmetric divisions of multipotent retinal progenitors that generate one Ath5+ neurogenic progenitor and one Ath5- progenitor that will enter a different lineage. This asymmetry is regulated by Notch signalling and is important to ensure the balance between proliferating cells and cells entering differentiation routes at early neurogenesis stages. Interference with the division asymmetry leads to premature or delayed differentiation. We further provide evidence that Notch does not act through an apicobasal activity gradient as previously proposed but via the endocytic pathway. Figure 5G summarises our main findings and presents a working model.

Our quantitative imaging approach showed that progenitor divisions that generate the earliest retinal neurogenic progenitors, Ath5 positive progenitors, are reproducibly asymmetric. Notch inhibition leads to more symmetric neurogenic divisions producing two neurogenic progenitors. This explains the increase in the total number of neurons at the lineage level, adding to previous findings that investigated this issue at the tissue level (Austin et al., 1995; Ahmad et al., 1997). This increase in symmetric divisions eventually leads to the premature exhaustion of multipotent progenitors.

In line with our findings, it has been shown earlier that Notch is involved in asymmetric divisions that generate two different neurons or one neuron and one proliferative cell in other neural systems. including the vertebrate spinal cord and mammalian neocortex (Mizutani et al., 2007; Nelson et al., 2013; Kressmann et al., 2015). For the early neurogenic progenitor divisions in the retina investigated here, it was speculated that Notch acts via a signalling gradient along the apicobasal axis of the neuroepithelium (Murciano et al., 2002; Del Bene et al., 2008). It was further proposed that more basally translocating progenitors are more likely to become neurogenic, generating two neurons after the next division (Murciano et al., 2002; Baye and Link, 2007). This led to the hypothesis that Notch signalling acts mainly at the apical side of the neuroepithelium where it suppresses neurogenic fate. However, due to technical challenges at the time, these analyses could only take a small number of unrelated cells into consideration. It was reported in one of these previous studies that tissue wide treatment of chick retinae with cytochalasin B led to increased premature neurogenesis (Murciano et al., 2002). However, no tracking of nuclei was performed and the tissue wide effects that such drug treatment entail, including cell delamination, were not further investigated. We thus aimed to manipulate nuclear position affecting single cells, to?have?the possibility to track the basal translocation patterns of large cohorts of sister cells. This generated a more quantitative and direct assessment of the effects of nuclear positioning during IKNM on neurogenesis in the context of the first asymmetric neurogenic divisions. We indeed found that the more basally translocating sister cell is more likely to enter the Ath5+ lineage. However, when the entirety of Ath5+ cells was taken into account, a significant overlap of basal positioning of Ath5+ and Ath5- cohorts was observed. In addition, the asymmetry of neurogenic commitment persisted when the translocation position of sister cells was levelled towards either the basal or the apical side. Together, this shows that somal position along the apicobasal axis of the retinal neuroepithelium does not influence asymmetric neurogenic commitment.

Instead, we present evidence that it is the asymmetric partitioning of Sara endosomes that drives the asymmetry of retinal progenitor divisions to enter the neurogenic pathway. A similar idea was previously proposed for the Drosophila sensory organ precursor and the zebrafish spinal cord (Coumailleau et al., 2009; Kressmann et al., 2015). Ideally, we would have directly correlated Sara-positive endosomes inheritance with neurogenic or proliferative fate occurrence in numerous cells. However, this correlation could not be systematically quantified over several instances of asymmetric divisions in our system, due to its complexity and long timeframes between division and fate readout. Nevertheless, we could follow one division in which Sara endosomes were asymmetrically inherited and assess the fate of the sister cells. As expected, the cell inheriting all endosomes became the Ath5- progenitor and thus remained proliferative (Figure 5—figure supplement 1E, Figure 5—video 1), while the cell that did not inherit Sara?endosomes started Ath5 expression and thereby entered the neurogenic lineage (see working model Figure 5G and Figure 5—figure supplement 1E, Figure 5—video 1). This outcome is in line with the finding that Ath5+ cells do not show active Notch signalling. These observations, combined with the finding that asymmetric inheritance is observed at stages at which asymmetric Ath5 decisions occur and that Sara knockdown impairs the balance between proliferation and neurogenesis in the Ath5 lineage, strongly argue that Notch signalling via Sara-positive endosomes is important for lineage asymmetry in the retina. The asymmetric distribution of Sara-positive endosomes most likely leads to the downregulation of Notch in one sister cell. When Sara is depleted however, more progenitors seem to express active Notch signalling, leading to a reduced amount of neurogenic Ath5+ cells at early neurogenic stages. This finding is different from what was presented in an earlier study that investigated the role of Sara endosomes on neurogenesis in the zebrafish spinal cord (Kressmann et al., 2015). Here, Sara depletion (mutant and morphant) had the same effect as direct inhibition of Notch signalling, namely the premature production of neurons. This argues that Sara can fine-tune Notch activity differently in different ttissues?and neuronal lineage contexts. We speculate that in the context of the early retinal lineages we studied, asymmetric neurogenic decisions are influenced and regulated by Sara protein that guides the asymmetric dispatch of these endosomes (Coumailleau et al., 2009), consequently downregulating Notch in one of the sister cells and promoting Notch in the other. When Sara is not present however, Notch continues to be active. In this case, Notch distribution between sister cells is not regulated any more through the endosomal pathway, so it cannot be depleted from one sister cell during division, and as a consequence more cells maintain high Notch activity. These additional cells then also stay in the proliferative state, leading to a decrease in the number of neurogenic cells. The exact mechanism of how Sara fine-tunes lineage decisions in different contexts will be interesting to explore in future studies. One factor could be the commitment state of neurogenic progenitors which differs between spinal cord and retina, but this point will need further investigation.

It should be noted that our systematic analysis not only enhances the understanding of Notch signalling in this particular neurogenesis event, but also shows the importance of quantitative approaches in developmental systems. Only due to the analysis of large cohorts of sister cells combined with a backtracking approach could we achieve the depth of analysis needed to exclude the role of a possible Notch gradient, which led to the investigation of the Notch endocytic pathway. The advance of imaging techniques, image analysis algorithms, and the possibility of performing backtracking studies in the retina are exciting emerging tools to follow similar questions in future studies. Such sophisticated studies will in turn promote our understanding of reproducibility and/or plasticity of lineage decisions at diverse developmental stages in different tissues of the CNS (Zechner et al., 2020).

Despite the fact that basal position is not directly linked to Ath5+ fate commitment, we find that the majority of progenitors entering the Ath5 lineage nevertheless translocate their somas more basally than their Ath5- sister cell. This observation argues that the more basal translocation of neurogenic Ath5+ progenitors is a consequence of neurogenic commitment, rather than a cause for it. We find that basal translocation of Ath5+ progenitors is linked to apically localised stabilised microtubules and basal process inheritance. This is remarkably similar to what has been shown for somal translocation events of retinal ganglion cells (Icha et al., 2016a). By contrast, this is different to nuclear and thereby somal translocation during IKNM in multipotent progenitors, which has been shown to be caused by passive nuclear displacements (Norden et al., 2009; Kosodo et al., 2011; Leung et al., 2011). Thus, an interesting interpretation of these findings could be that Ath5+ progenitors constitute an intermediate cell type placed between multipotent progenitors and differentiating neurons at the level of cellular arrangements including cytoskeletal components. This is reminiscent of intermediate progenitors found in the mammalian neocortex (Haubensak et al., 2004; Miyata et al., 2004; Noctor et al., 2004). To our knowledge, the existence of such intermediate progenitors has not been thoroughly investigated in the retina. However, a recent study performing single cell RNA sequencing of mouse retinae (Wu et al., 2020) hints that?a transitional state of retinal progenitors that express atoh7 and other neurogenic genes?could?exist. Therefore, it will be interesting to probe how cytoskeletal changes could be linked to these cell identity changes in the developing retina and beyond. Such studies will lead to new insights into the biology of fate decisions downstream of transcription profiles and towards the cell biological scale.

Asymmetric progenitor divisions are an important mean to maintain the balance between proliferation and differentiation in developing organs. To understand the regulation of these types of divisions and thereby embryogenesis as a whole, their investigation needs to span scales from the molecular to the cellular to the tissue level. Furthermore, investigations need to combine knowledge of cell type specific transcription factor profiles, the signalling factors involved in guiding fate decisions and the cell biological consequences that differentiation entails. The emerging possibilities of quantitative in vivo studies mentioned above in combination with the advent of single-cell sequencing data that is becoming more and more extensive (Wagner and Klein, 2020; Xu et al., 2020), opens new perspectives for an understanding of these phenomena at a previously unimaginable level. This, in combination with the continued use of the zebrafish as a model for quantitative developmental biology, has the potential to reveal general principles that can then be extended to other, often less accessible systems.

## Materials and methods

Key resources table

### Zebrafish husbandry

Request a detailed protocol

Wild-type zebrafish were maintained at 26°C. Embryos were raised at 21, 28.5, or 32°C in E3 medium, which was changed daily. To prevent pigmentation, E3 medium was supplemented with 0.2 mM 1-phenyl-2-thiourea (Sigma-Aldrich) from 8 hpf. Animals were staged in hpf according to Kimmel et al., 1995. All animal work was performed in accordance with European Union directive 2010/63/EU, as well as the German Animal Welfare Act.

### Transgenic lines

Request a detailed protocol

To visualise Ath5 neurogenic progenitors, the Tg(ath5:gap-GFP) and Tg(ath5:gap-RFP) (Zolessi et al., 2006) lines were used. To visualise Notch activity, the Notch reporter lines Tg(her4.1:mcherry-CreERT2) (Kroehne et al., 2011), Tg(her4:EGFP) (Yeo et al., 2007), and Tg(Tp1bglob:H2BmCherry) (Ninov et al., 2012) were used. To label cell membranes, the Tg(actb1:HRAS-EGFP) (Cooper et al., 2005) line was used. To visualise progenitors nuclei, Tg(hsp70:H2B-RFP) line was used (Dzafic et al., 2015).

### DNA and RNA injections

Request a detailed protocol

To mosaically label cells in the zebrafish retina, DNA constructs were injected into one cell stage embryos. Injected volumes ranged from 0.5 to 1 nl. The amount of injected DNA ranged from 10 to 20 pg per embryo and did not exceed 40 pg even when multiple constructs were injected. See Table 1 for details. pCS2 expression vectors containing CFP-Sara or mRFP-Sara were kindly provided by Marcos Gonzáles-Gaitán and published previously (Kressmann et al., 2015). The Sara mRNA was synthesised using the Ambion mMessage mMachine kit. To visualise SARA endosomes, 100 pg of CFP-SARA or mRFP-SARA where injected into one cell stage embryos for ubiquitous expression.

Table 1

### Heat shock of embryos

Request a detailed protocol

To induce expression of the heat shock promoter (hsp70)-driven constructs, embryos were incubated at 24 or 28 hpf in water bath at different temperatures (See Table 1 for details). Imaging always started 3-4 hr after heat shock.

### Morpholino experiments

Request a detailed protocol

To knockdown gene function, the following amounts of morpholinos (Gene Tools) were injected into the yolk at one cell stage: 2,5 ng of a splice morpholino for the sara gene (5’-TGAACTAGAGACTTTACCTTGCCAC-3’) (Kressmann et al., 2015), 2 ng of p53 MO, (5’-GCGCCATTGCTTTGCAAGAATTG-3’ [Robu et al., 2007]), 1 ng of standard control morpholino (5’-CCTCTTACCTCAGTTACAATTTATA-3’), 1 ng of a translation blocking morpholino for the notch1b gene (5’-GTATTGCATTCTCCTCTCCCGTCTG-3’), 0.41 ng of a translation blocking morpholino for the gene deltaC (MO1, 5’-AGCCATCTTTGCCTTCT TGTCTGCT-3’) (Zhang et al., 2008), and 2 ng of another translation blocking morpholino for the gene deltaC (MO2, 5’-CGATAGCAGACTGTGAGAGTAGTCC-3’) (Holley et al., 2002; Okigawa et al., 2014).

### Blastomere transplantation

Request a detailed protocol

Donor embryos from the double transgenic line Tg(hsp70:H2B-RFP),Tg(her4:EGFP) and acceptor were dechorionated in pronase. At the high stage, cells from the animal pole of donors were transplanted into the acceptor wild type embryos. Transplanted embryos were grown at 32°C to recover for 4 hours, then moved to 28°C. The E3 medium was supplemented with 100 U of penicillin and streptomycin (Thermo Fisher Scientific). Transplanted embryos were heat shocked at 24 hpf for 15 min at 39°C and imaged from 28 hpf.

### Drug treatments

Request a detailed protocol

Notch inhibitors DAPT (Sigma-Aldrich) and LY411575 (Sigma-Aldrich) were dissolved in DMSO and used at 50 and 10 μM concentration respectively and compared to DMSO only controls. Embryos were dechorionated and treated with the drugs at 28°C in E3 medium, either in a 24 multiwell plate or the light sheet microscope chamber for live imaging experiments. All treatments started at 24 hpf. For live imaging experiments, fish were preincubated with the drug for 4 hours and imaging started at 28 hpf.

### In vivo labelling of proliferating cells

Request a detailed protocol

To label proliferating cells in the Notch inhibition and Sara knock down experiments, the EdU Click-iT-Alexa 488 or 647 fluorophore kit (Invitrogen) was used. Embryos were incubated at 4°C in E3 medium supplemented with 500 μM of EdU in 10% DMSO for 1 hour before fixation. Embryos were then washed twice with E3 and fixed overnight in 4% PFA. After whole mount staining, EdU was detected using the Click-iT-Alexa 488 or 647 fluorophore kit (Invitrogen) according to manufacturer’s protocol. Embryos were then stored in Phosphate buffered saline (PBS).

### Immunofluorescence

Request a detailed protocol

All immunostainings were performed on whole-mount embryos fixed in 4% paraformaldehyde (Sigma-Aldrich) in PBS as previously described (Icha et al., 2016a). In brief, embryos were permeabilised with trypsin, blocked with goat or donkey serum 10% in PBS-Triton 0.8% and incubated with the primary antibody for 72 hours. The Notch1b (GTX48505, GeneTex), DeltaC (zdc2, Abcam) and GFP (50430–2-AP, Proteintech) antibodies were used at 1:100 dilution. The acetylated tubulin antibody (T6793, Sigma-Aldrich) was used at 1:250 dilution. The phospho-histone H3 antibody (ab10543, abcam) was used at 1:500 dilution. Embryos were incubated for 48 hours with an appropriate fluorescently labeled secondary antibody (Molecular Probes) at 1:500 dilution and DAPI at 1:1000 dilution (Thermo Fisher Scientific).

### Image acquisition

#### Laser scanning confocal microscopy

Request a detailed protocol

Fixed samples from immunostaining experiments were imaged with the Zeiss LSM 880 inverted point-scanning confocal microscope (Carl Zeiss Microscopy) using the 40x/1.2 C-Apochromat water immersion objective (Zeiss). Samples were mounted in 0.6% agarose in glass-bottom dishes (MatTek Corporation) and imaged at room temperature. The microscope was operated with the ZEN 2011 (black edition) software (Zeiss).

#### Spinning disk confocal microscopy (SDCM)

Request a detailed protocol

Confocal stacks for fixed samples for Notch inhibition and Sara knockdown experiments were taken using an Andor SDCM system (Andor), consisting of an Andor IX 83 stand equipped with a CSU-W1 scan head (Yokogawa) with Borealis upgrade and an Andor iXon Ultra 888 Monochrome EMCCD camera. 40x/1.25 or 60x/1.3 silicone objectives were used to acquire 50–80 μm z-stacks were recorded with 1 μm optical sectioning. Fixed samples were mounted in 0.6% agarose in glass-bottom dishes (MatTek Corporation) and imaged at room temperature. The microscope was operated by Andor iQ 3.6 software.

#### In vivo light sheet fluorescent imaging

Request a detailed protocol

Light sheet imaging started between 28 and 30 hpf. Embryos were manually dechorionated and mounted in glass capillaries in 0.6% low-melting-point agarose as previously described (Icha et al., 2016b). The sample chamber was filled with E3 medium containing 0.01% MS-222 (Sigma-Aldrich) to immobilise embryos and 0.2 mM N-phenylthiourea (Sigma-Aldrich) to prevent pigmentation. Imaging was performed on a Zeiss Lightsheet Z.1 microscope (Carl Zeiss Microscopy) equipped with two PCO.Edge 5.5 sCMOS cameras and using a 20x/1.2 Zeiss Plan-Apochromat water-immersion objective at 28.5°C. Z-stacks spanning the entire retinal epithelium (70–100 μm) were recorded with 1 μm optical sectioning every 5 min for 15–20 hr using the double-sided illumination mode. Sara endosomes inheritance during cell division was imaged with 1 min time resolution and 0.5 μm optical sectioning. The system was operated by the ZEN 2014 software (black edition).

### Image analysis

Images from live imaging experiments were cropped and averaged in ZEN Black and/or Fiji (Schindelin et al., 2012) and were corrected for drift using a Fiji plugin (Manual drift correction) created by Benoit Lombardot (Scientific Computing Facility, Max Planck Institute of Molecular Cell Biology and Genetics, Dresden, Germany).

#### Analysis of sister cell somal translocation

Request a detailed protocol

Ath5 progenitor divisions were identified and cells were backtracked to the previous cell division in order to identify cells giving rise to Ath5+ progenitors and the sister cell arising from the division. From the point after division, the translocation of sister cells was tracked in 2D in maximum projected sub-stacks by following the center of the cell body in Fiji using the semi-automated ImageJ plugin MTrackJ (Meijering et al., 2012). MSDs and directionality ratios were calculated in the DiPer program (Gorelik and Gautreau, 2014), executed as a macro in Excel (Microsoft). To calculate the average instantaneous velocities, a custom-made script was written using Python 3 and can be found here (Nerli, 2020).

#### Integral mean calculations

Request a detailed protocol

The integral mean $d^$ was computed using Python 3 (https://git.mpi-cbg.de/nerli/nerli-et-al-2020-scripts;?Nerli, 2020) as follows:

(1) $\text{d?}=\frac{I}{T}$

where T is the time length of each cellular trajectory and I is the area below the trajectory curve, calculated as:

(2) $I={\int }_{0}^{T}\mathrm{depth}?\left(t\right)?dt$

where depth(t)?>0.

#### Retinal size measurements

Request a detailed protocol

Retinal size was manually measured using 48 hpf retinas treated with LY411575 10 μM or equal volume of DMSO, as illustrated in Figure 1G. Thickness and diameter were measured in the central z plane of the nasal, central and temporal region of the retina, using the Fiji line tool.

#### Quantification of the number of EdU-positive cells

Request a detailed protocol

Prior to counting, a Gaussian blur of 0.5 was applied to all the images. The number of EdU positive cells was manually counted in the central part of the retinal neuroepithelium in Fiji, using a region of interest (ROI) that spans the entire length of the apicobasal axis (6400 μm2). For each embryo, three Z slices per central region were analysed.

#### Notch1b and DeltaC spatial distribution

Request a detailed protocol

Prior to analysis, a Gaussian blur of 0.5 was applied to all the images. Apical, central and basal ROIs of the same dimensions (560 μm2) were set for each image, dividing the length of the apicobasal axis by 3. Three central slices per embryo were analysed in the nasal, central and temporal regions (Figure 3A’’). SpotCounter plugin for Fiji (Spot counter, Nico Stuurman, ValelabUtils package) was used to measure the number of spots in each ROIs. The number of spots in each ROIs was then normalised by the total number of spots in the 3 ROIs of a slice.

#### Sara-positive endosome count

Request a detailed protocol

Prior to analysis, a Gaussian blur of 0.5 was applied to all the images. Confocal z-stack of retinal progenitors were reconstructed in 3D using the volume renderer ClearVolume (Royer et al., 2015) for Fiji. The number of Sara-positive endosomes was manually counted in dividing cells from 3D reconstructions.

### Statistical analysis

Request a detailed protocol

All statistical tests used are indicated in the figure legend, as well as the definitions of error bars. All tests used were two?sided and 95% confidence intervals were considered. P values and sample sizes are noted in figure panels or figure legends. Data was analysed using GraphPad Prism 6 or Python 3. Statistical analysis was performed using GraphPad Prism 6 and MATLAB.

## References

1. 1
2. 2
3. 3
4. 4
Vertebrate retinal ganglion cells are selected from competent progenitors by the action of notch
(1995)
Development 121:3637–3650.
5. 5
6. 6
7. 7
8. 8
9. 9
10. 10
11. 11
12. 12
13. 13
14. 14
15. 15
16. 16
17. 17
18. 18
19. 19
20. 20
21. 21
The cell biology of neurogenesis (2005)
Nature Reviews Molecular Cell Biology 6:777–788.
https://doi.org/10.1038/nrm1739
22. 22
23. 23
24. 24
25. 25
26. 26
her1 and the notch pathway function within the oscillator mechanism that regulates zebrafish somitogenesis
(2002)
Development 129:1175–1183.
27. 27
28. 28
29. 29
30. 30
31. 31
32. 32
33. 33
34. 34
35. 35
36. 36
37. 37
38. 38
39. 39
40. 40
41. 41
42. 42
43. 43
44. 44
45. 45
46. 46
47. 47
48. 48
49. 49
50. 50
51. 51
52. 52
53. 53
54. 54
55. 55
56. 56
57. 57
58. 58
59. 59
60. 60
61. 61
62. 62
63. 63
64. 64
Mitosis in the neural tube (1935)
The Journal of Comparative Neurology 62:377–405.
https://doi.org/10.1002/cne.900620207
65. 65
66. 66
67. 67
68. 68
69. 69
70. 70
71. 71
72. 72
4, a zebrafish homologue of the Drosophila neurogenic gene E(spl), is a target of NOTCH signalling
(1999)
Development 126:1811–1821.
73. 73
74. 74
75. 75
76. 76
77. 77
78. 78
79. 79
80. 80
81. 81
82. 82
83. 83
84. 84
85. 85

## Decision letter

1. Tatjana Piotrowski
Reviewing Editor; Stowers Institute for Medical Research, United States
2. Marianne E Bronner
Senior Editor; California Institute of Technology, United States

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

Acceptance summary:

This work nicely demonstrates the existence of asymmetric neurogenic divisions in the developing retina and rules out, in a clean and convincing way, previous hypotheses on graded Notch signaling along the apicobasal axis. It makes an important contribution to our understanding of retinal development.

Decision letter after peer review:

Thank you for submitting your article "Asymmetric neurogenic commitment of retinal progenitors is regulated via the Notch endocytic pathway" for consideration by eLife. Your article has been reviewed by two peer reviewers, one of whom is a member of our Board of Reviewing Editors, and the evaluation has been overseen by Marianne Bronner as the Senior Editor. The reviewers have opted to remain anonymous.

The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.

As the editors have judged that your manuscript is of interest, but as described below that additional experiments are required before it is published, we would like to draw your attention to changes in our revision policy that we have made in response to COVID-19 (http://www.asadoresteatre.com/articles/57162). First, because many researchers have temporarily lost access to the labs, we will give authors as much time as they need to submit revised manuscripts. We are also offering, if you choose, to post the manuscript to bioRxiv (if it is not already there) along with this decision letter and a formal designation that the manuscript is "in revision at eLife". Please let us know if you would like to pursue this option. (If your work is more suitable for medRxiv, you will need to post the preprint yourself, as the mechanisms for us to do so are still in development.)

Summary:

The paper by Nerli et al. addresses the molecular control of the first neurogenic divisions in the developing zebrafish retina. Progenitors fated to become neurons express the transcription factor Ath5. Using Ath5-driven transgenic lines in live light-sheet microscopy to retrospectively track neurogenic progenitors, the authors demonstrate that these progenitors originate from divisions that are asymmetric in fate (one daughter cell becoming Ath5+) and in Notch signaling (the Ath5- daughter is Notch signaling high, as revealed using her4- or TP1-driven transgenes), the latter being necessary for the Ath5- fate. The remainder of the paper then aims to identify the mechanism by which a Notch signaling asymmetry is generated between daughter cells at the onset of neurogenic divisions. The work reported specifically addresses and challenges the conclusions made by previous studies that interkinetic nuclear migration combined with an apicobasal Notch gradient exposes progenitor nuclei to different Notch signaling levels. This hypothesis is ruled out here using a series of elegant experiments tracing individual nuclei and modifying their position by means of manipulating dynactin or stathmin. The authors then address the alternative hypothesis of intra-lineage signaling with an implication of Sara endosomes, as previously shown in Drosophila (Coumailleau et al., 2009) and zebrafish neural progenitors of the embryonic spinal cord (Kressmann et al., 2015). The authors observe (using overexpression of fluo-Sara mRNA) Sara endosomes carrying Notch1b and DeltaC proteins in retinal progenitors, and an asymmetric distribution of Sara endosomes in daughter cells. Knocking-down Sara (using a Sara MO) generates fewer neurons and an increased number of TP1-positive cells. The authors conclude that neurogenic asymmetry is driven by the asymmetric segregation of Sara endosomes, driving asymmetric Notch signaling in retinal progenitors.

This is a largely well-executed and nicely written piece of work. It convincingly rules out the previous hypothesis of apicobasal nuclear positioning, which had driven the field for years although solely based on in situ hybridization data. This in itself is important. However, the "positive" mechanistic data implying Sara endosomes remains correlative and describes a mechanism that has been previously reported. It is also at odds with previously published work (Kressmann et al., 2015). Key experiments directly implicating Sara endosomes in division asymmetry, asymmetrical Notch signaling and asymmetrical fate are lacking and should be provided to support the conclusions.

Essential revisions:

1) The authors report one case (=1 division) where Sara endosomes are concentrated in the Ath5- daughter cell. Attempts should be made to confirm this result. Live imaging may not be doable, but fixed material, where daughter cells are mosaically labeled could be analyzed for Ath5 and TP1 expression. This is crucial, as Sara is associated with the progenitor fate in the spinal cord (Kressmann et al), is postulated (but not shown) to have the same association here, but when depleted leads to more progenitors (Figure 5E, F and see point 3 below).

2) The effects of the Sara MO are suggestive of increased Notch signaling and symmetric proliferative, as opposed to asymmetric neurogenic, divisions. However, these hypotheses are not demonstrated directly. It could be possible, for example, that progenitors are blocked at the pre-division phase. The authors should directly show, using live cell tracking, that divisions become symmetric proliferative, generating two Ath5- (or TP1+) cells in the absence of Sara.

3) The results of the Sara MO experiments are strikingly different from those of Kressmann et al. In the present report, Sara knock-down leads to increased Notch signaling; in the spinal cord, Kressmann et al. showed that Sara knock-down led to decreased Notch signaling and a loss of the progenitor fate. Nerli et al. state "We propose that Sara endosomes are mostly inherited by the Ath5- cell, which retains high Notch signaling and thus stays in the proliferative state" and, under Sara knock-down: "there is an increase in active Notch signaling in both daughter cells as postulated by our working model". These two statements, which a appear contradictory, and the discrepancies with the work of Kressmann et al., should absolutely be discussed. What is the authors' hypothesis on how Sara endosomes would associate with the future progenitor cell and increased Notch signaling, but lead to more progenitor cells and more Notch when functionally depleted?

4) Figure 3: The Notch1b and DeltaC fluorescent antibody stainings have a lot of background and the brain is equally stained. Proper negative controls are needed for the Notch1b antibody. This means, controls should be performed using specimen lacking Notch1b (e.g. mutants or morphants), rather than specimen where the primary antibody is omitted (Figure 3—figure supplement 1). Also, could colorimetric in situs be performed? Given that the del Bene paper argued that it was notch1a, deltaB and deltaC RNA that was graded along the apicobasal axis of the retina, these in situ hybridization experiments should be repeated, in addition to the antibody stainings.

5) How do the authors reconcile that they did not observe graded DeltaC expression, whereas in the del Bene paper the RNA and antibody of DeltaC looks indeed graded? Murciano et al. (Figure 1C), also show clear Notch1 accumulation basally in the chicken retina, which should be discussed.

6) The author's state: “Heat shock expression of the hsp70:DNdynactin-mKate2 construct led to the translocation of both sister cell somas to greater and almost identical basal positions compared to controls. However, despite the fact that both sister cell somas reached very basal positions, only one became Ath5+.”

Murciano et al., also pharmacologically inhibited internuclear migration but saw and effect on neurogenesis. This finding should be discussed.

[Editors' note: further revisions were suggested prior to acceptance, as described below.]

Thank you for resubmitting your work entitled "Asymmetric neurogenic commitment of retinal progenitors is regulated via the Notch endocytic pathway" for further consideration by eLife. Your revised article has been evaluated by Marianne Bronner (Senior Editor) and a Reviewing Editor.

The manuscript has been significantly improved but there are some remaining issues that need to be addressed before acceptance, as outlined below:

Overall, this work is important in that it demonstrates the existence of asymmetric neurogenic divisions in the developing retina and rules out, in a clean and convincing way, previous hypotheses on graded Notch signaling along the apicobasal axis. The authors took the reviewer's comments to heart and adequately addressed critiques 1-6. However, the manuscript still shows a weakness in identifying the mechanism controlling this asymmetry. One key aspect that remains undemonstrated is the fact that Sara endosomes are involved in the asymmetry of progenitor division (requested revisions point 2). It is unclear why the authors state that live imaging is not possible to monitor symmetric divisions in Sara morphants, while the same method was used in the case of LY411575 application (Figure 1I). The absence of experiments showing that cell divisions become symmetric, diminishes the Sara-related message.

Overall, the reviewers are enthusiastic about the work but believe that in the absence of live imaging of cell divisions in Sara morphants, it is important not to overstate the function of Sara, irrespective of how logical it may seem.

The authors already changed the text to state that they speculate that cell divisions become symmetric after Sara MO but key titles and sentences were not correspondingly changed. For example: (i) title of paragraph "Asymmetric inheritance of Sara-positive endosomes affects the asymmetry of neurogenic commitment" – the MO manipulation does not affect the asymmetric inheritance of Sara endosome but their overall presence, and the effect on asymmetrical neurogenic commitment is not shown; (ii) title of Figure 5 "Sara-positive endosomes affect the asymmetry of divisions generating Ath5+ progenitors", and.… (iii) the title of the paper "Asymmetric neurogenic commitment of retinal progenitors is regulated via the Notch endocytic pathway".

https://doi.org/10.7554/eLife.60462.sa1

## Author response

Essential revisions:

1) The authors report one case (=1 division) where Sara endosomes are concentrated in the Ath5- daughter cell. Attempts should be made to confirm this result. Live imaging may not be doable, but fixed material, where daughter cells are mosaically labeled could be analyzed for Ath5 and TP1 expression. This is crucial, as Sara is associated with the progenitor fate in the spinal cord (Kressmann et al), is postulated (but not shown) to have the same association here, but when depleted leads to more progenitors (Figure 5E,F and see point 3 below).

We now expanded our analysis of the lack of Tp1, Ath5 co-localization. We analyzed 215 cells expressing Ath5 from 4 different embryos and find that only in 5 cases weak expression of Tp1 with Ath5 is seen, while all other cells show no expression (text in subsection “Notch pathway components are uniformly distributed along the apicobasal axis of the retinal neuroepithelium” and Figure 3—figure supplement 2A, A’). This strongly argues that indeed Notch signaling is not active in cells starting to express Ath5 and our finding that Sara depletion increases the number of Tp1 positive cells and decreases the number of Ath5 positive cells argues that indeed Sara is responsible for a balanced distribution of this signaling pathway component (see also our interpretation on the Sara related pathway in point 3). In addition, we confirmed that Tp1 positive cells indeed carry Sara endosomes (Figure 3—figure supplement 2B).

2) The effects of the Sara MO are suggestive of increased Notch signaling and symmetric proliferative, as opposed to asymmetric neurogenic, divisions. However, these hypotheses are not demonstrated directly. It could be possible, for example, that progenitors are blocked at the pre-division phase. The authors should directly show, using live cell tracking, that divisions become symmetric proliferative, generating two Ath5- (or TP1+) cells in the absence of Sara.

This is an important point brought up by the reviewers and we previously had this idea ourselves. However, when we started to outline the actual experiments, we noted that we made a thinking mistake as this experiment is close to impossible to carry out. The reason is that at the stage at which the Ath5+ progenitors arise, most other progenitors still undergo symmetric proliferative divisions, generating two Ath5- cells, also in the control situation. Thus, finding symmetric proliferative Ath5- divisions would be the norm, not the exception, and it would not be feasible to detect an increase in symmetric proliferative divisions upon Sara knockdown with the cohort size that is possible in live-imaging experiments. We believe however that the experiments in response to Q1 and Q3 are convincing to substantiate our general hypothesis.

Nonetheless, to further underline this notion, we excluded the possibility that progenitors are stalled before or during division using pH3 staining. We find a similar amount of divisions in Sara morphants as in controls. This data in now shown in Figure 5—figure supplement 1B. Moreover, consistent with the hypothesis that Sara knockdown leads to more symmetric proliferative divisions, we demonstrate that morphant embryos feature a higher level of proliferation, an expected consequence for an increase in the number of symmetric proliferative divisions, shown by quantification of EdU labelling in controls versus Sara morphants (Figure 5—figure supplement 1B, C, D) at stages at which in control progenitors already leave the cell cycle and become Ath5+ (Figure 5—figure supplement 1C, control embryos, and Figure 1 E, F, control embryos). These experiments confirm that the balance between proliferation and differentiation is changed in Sara morphants.

Nevertheless, as we do understand the reviewer’s point, we tuned down our statement to “We thus speculate that depletion of Sara led to more symmetric proliferative divisions generating two Ath5- cells.”.

We also added a statement and compare our findings to previous findings and added a paragraph in the Discussion (see also point 3).

3) The results of the Sara MO experiments are strikingly different from those of Kressmann et al. In the present report, Sara knock-down leads to increased Notch signaling; in the spinal cord, Kressmann et al. showed that Sara knock-down led to decreased Notch signaling and a loss of the progenitor fate. Nerli et al. state "We propose that Sara endosomes are mostly inherited by the Ath5- cell, which retains high Notch signaling and thus stays in the proliferative state" and, under Sara knock-down: "there is an increase in active Notch signaling in both daughter cells as postulated by our working model". These two statements, which a appear contradictory, and the discrepancies with the work of Kressmann et al., should absolutely be discussed. What is the authors' hypothesis on how Sara endosomes would associate with the future progenitor cell and increased Notch signaling, but lead to more progenitor cells and more Notch when functionally depleted?

While these two statements might on first sight sound slightly counterintuitive, we do not believe that they contradict each other. Our interpretation is the following: In the control scenario, in which the Sara protein is expressed, Sara-positive endosomes are linked to Notch signaling and can be unequally distributed during cell divisions generating imbalances/asymmetries/differences in Notch activity among progenitor cells. While the cell inheriting the endosomes and high notch activity remains a progenitor, its sister cell is depleted of Notch and enters neurogenesis. However, when Sara is depleted from all the cells by morpholino knockdown, Notch continues to be active. In this case, Notch distribution between sister cells is not regulated any more through the endosomal pathway, so it cannot be depleted from one sister cell during division. Consequently, more cells maintain high Notch activity (Figure 5F), leading to increased proliferation and decreased neurogenesis (Figure 5—figure supplement 1C). This argues that Sara is responsible for controlled Notch distribution between sister cells.

Thus, the differences in outcome between retina and spinal cord imply that the fine-tuning of Notch depends on the specific neuronal system and lineage in which it is acting. One of these differences could be different commitment states in the lineages as seen between the spinal cord and the retina. The exact mechanisms of how this fine-tuning of Notch signaling through Sara-positive endosomes occurs will be interesting to explore in future studies.

We now discuss this point more explicitly in the Discussion.

4) Figure 3: The Notch1b and DeltaC fluorescent antibody stainings have a lot of background and the brain is equally stained. Proper negative controls are needed for the Notch1b antibody. This means, controls should be performed using specimen lacking Notch1b (e.g. mutants or morphants), rather than specimen where the primary antibody is omitted (Figure 3—figure supplement 1). Also, could colorimetric in situs be performed? Given that the del Bene paper argued that it was notch1a, deltaB and deltaC RNA that was graded along the apicobasal axis of the retina, these in situ hybridization experiments should be repeated, in addition to the antibody stainings.

We now added further controls and used the Notch1b and DeltaC antibodies on respective morphants (we used established morpholinos as the timeframe of the revisions and COVID situation did not allow for importing mutants).

In both cases, especially strong in the Notch scenario, we see disappearance or strong reduction of antibody staining compared to the control situation. Thus, we are confident that the antibody staining of controls is genuine in both scenarios (Figure 3—figure supplement 1A, B).

We further performed in situ hybridization for notch1a and deltaC, similar to experiments done in del Bene et al., 2008. We show the results in Author response image 1 as we do not believe that they significantly add to the manuscript. We note that our staining looks very similar to what was reported by del Bene (their Figure 5 A-C). However, as these are RNA probes, we believe that the antibody experiments are more meaningful as they reflect the protein localization. Despite the fact that there might be some graded RNA distribution, which is hard to tell at this resolution, our findings that there is no correlation between apicobasal nuclear positioning and neurogenic fate determination shows that such an RNA gradient, if existent, does not influence neurogenic potential.

Author response image 1

5) How do the authors reconcile that they did not observe graded DeltaC expression, whereas in the del Bene paper the RNA and antibody of DeltaC looks indeed graded? Murciano et al. (Figure 1C), also show clear Notch1 accumulation basally in the chicken retina, which should be discussed.

See our response to point 4. Furthermore, in the Del Bene paper staining was performed using a different DeltaC antibody. We here used a DeltaC antibody that has been shown to be specific in zebrafish (see Wright et al., 2011 Development, Giudicelli, 2007 Plos Biology and others). It is not uncommon that RNA distribution and protein distribution slightly differ in tissues (shown in Spirov et al., 2009 for the bicoid mRNA and protein, and in Toyoda et al., 2010 for Fgf8 in the mouse telencephalon). However, as noted above, the fact that apicobasal nuclear positioning does not influence neurogenic fate determination shows that apicobasal graded distribution of Notch/Delta, if existent at the RNA level, is independent of neurogenic potential.

6) The author's state: “Heat shock expression of the hsp70:DNdynactin-mKate2 construct led to the translocation of both sister cell somas to greater and almost identical basal positions compared to controls. However, despite the fact that both sister cell somas reached very basal positions, only one became Ath5+.”

Murciano et al., also pharmacologically inhibited internuclear migration but saw and effect on neurogenesis. This finding should be discussed.

It is hard to interpret the differences between our study and the previous study in which nuclear movement was blocked by a tissue wide cytochalasin B (CCB) treatment (Figure 5 in Murciano et al., 2002) due to the fact that CCB not only blocks IKNM (which was not explicitly shown as only fixed samples were presented) but also interferes with diverse other cellular processes such as cytokinesis (shown by the increase of mitotic figures in Figure 5B) and can lead to cell delamination (which could explain the non-apical mitotic cells observed in Figure 5B). It was not investigated which, if not all, of these effects apply in the previously presented scenario. Thus, it is hard to decipher in hindsight, the exact contributions that led to the increase in neurogenesis as division patterns (e.g. symmetric versus asymmetric) were not explored in this study.

In contrast, for our experiments we used genetic interference in single cells in an otherwise untreated neuroepithelium. Neither of the two independent genetic conditions, one that keeps cell soma at more apical positions and another one that sends cell soma more basally, had an influence on the asymmetric division patterns of cells producing one Ath5 positive sister cell. This fact makes us confident that the extent of basal migration does not influence the asymmetry of divisions leading to one neurogenic Ath5+ daughter cell.

We now added a brief statement discussing this item in the Discussion.

[Editors' note: further revisions were suggested prior to acceptance, as described below.]

Overall, the reviewers are enthusiastic about the work but believe that in the absence of live imaging of cell divisions in Sara morphants, it is important not to overstate the function of Sara, irrespective of how logical it may seem.

The authors already changed the text to state that they speculate that cell divisions become symmetric after Sara MO but key titles and sentences were not correspondingly changed. For example: (i) title of paragraph "Asymmetric inheritance of Sara-positive endosomes affects the asymmetry of neurogenic commitment" – the MO manipulation does not affect the asymmetric inheritance of Sara endosome but their overall presence, and the effect on asymmetrical neurogenic commitment is not shown; (ii) title of Figure 5 "Sara-positive endosomes affect the asymmetry of divisions generating Ath5+ progenitors", and.… (iii) the title of the paper "Asymmetric neurogenic commitment of retinal progenitors is regulated via the Notch endocytic pathway".

In particular, we toned down the statements about the involvement of the Sara endocytic pathway in asymmetric neurogenic commitment. We changed:

– The title of the paper to “Asymmetric neurogenic commitment of retinal progenitors involves Notch through the endocytic pathway”;

– The title of the subsection to “Notch signalling affects neurogenic commitment of progenitors through Sara endosomes;”

– The title of Figure 5 to “Notch signalling affects neurogenic commitment of progenitors through Sara endosomes”.

We also like to take this opportunity to further clarify why we cannot monitor symmetric division with live imaging in Sara morphants as we did upon Notch inhibition. In the case of Notch inhibition using the LY411575 inhibitor, we showed that more symmetric neurogenic divisions occur, and more cells express Ath5 (Figure 1E, H). This experiment was possible as in this condition, both sister cells start expressing Ath5+, while in control conditions only one sister expresses Ath5 (Figure 1B) meaning we could use the onset of fluorescence as a readout. In Sara morphants however, fewer Ath5+ cells are present (Figure 5F) and we would expect to observe more symmetric proliferative divisions, in which both progenitors do not express Ath5.

However, as this type of symmetric proliferative divisions normally occurs also for the majority of cells in controls, we do not have a positive readout for cells that would have turned on Ath5 but do not do so in the Sara morphant condition. What we did show is an increase in the number of proliferative cells (Figure 5—figure supplement 1C) and a decrease in the number of Ath5+ cells (Figure 5F, Figure 5—figure supplement 1C), which in our opinion can only be explained by an increase of symmetric proliferative divisions generating two Ath5- progenitors at the expenses of asymmetric neurogenic divisions. We hope this clarifies this issue, but we agree that tuning down our statements further improves the manuscript.

https://doi.org/10.7554/eLife.60462.sa2

## Article and author information

### Author details

1. #### Elisa Nerli

1. Max Planck Institute of Molecular Cell Biology and Genetics, Dresden, Germany
2. Instituto Gulbenkian de Ciência, Oeiras, Portugal
##### Contribution
Conceptualization, Data curation, Formal analysis, Validation, Investigation, Visualization, Methodology, Writing - original draft
##### Competing interests
No competing interests declared
2. #### Mauricio Rocha-Martins

1. Max Planck Institute of Molecular Cell Biology and Genetics, Dresden, Germany
2. Instituto Gulbenkian de Ciência, Oeiras, Portugal
##### Contribution
Conceptualization, Data curation, Supervision, Investigation, Methodology, Writing - original draft
##### For correspondence
mrmartins@igc.gulbenkian.pt
##### Competing interests
No competing interests declared
3. #### Caren Norden

1. Max Planck Institute of Molecular Cell Biology and Genetics, Dresden, Germany
2. Instituto Gulbenkian de Ciência, Oeiras, Portugal
##### Contribution
Conceptualization, Resources, Supervision, Funding acquisition, Writing - original draft, Project administration
##### For correspondence
cnorden@igc.gulbenkian.pt
##### Competing interests
No competing interests declared

### Funding

• Caren Norden

• Caren Norden

#### Max-Planck-Gesellschaft

• Caren Norden

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

### Acknowledgements

We thank the Norden Lab for fruitful project discussions. We are grateful to William A Harris, Julien Vermont and Iskra Yanakieva for the helpful comments on the manuscript. Christoph Zechner and Tommaso Bianucci are thanked for help with data analysis. We thank Sylvia Kaufmann, the Computer Department, Light Microscopy Facility, Scientific Computing, and Fish facility of the Max Planck Institute of Molecular Cell Biology and Genetics for experimental support. We also thank Marcos Gonzáles-Gaitán for sharing the CFP-Sara and mRFP-Sara constructs, Nikolay Ninov for sharing the Tp1bglob:H2BmCherry line, Nadine Vastenhow’s lab, Wieland Huttner’s lab and Michael Brand’s lab for sharing probes and antibodies and help with experiments. EN is a member of the IMPRS-CellDevoSys PhD program and the IBB-Integrative Biology and Biomedicine PhD program and is supported by the MPI-CBG. CN is supported by MPI-CBG, the FCG-IGC, the German Research Foundation (NO 1069/5–1) and an ERC consolidator grant (H2020 ERC-2018-CoG-81904).

### Ethics

Animal experimentation: All animal work in this study was performed in accordance with European Union directive 2010/63/EU, as well as the German Animal Welfare Act.

### Senior Editor

1. Marianne E Bronner, California Institute of Technology, United States

### Reviewing Editor

1. Tatjana Piotrowski, Stowers Institute for Medical Research, United States

### Publication history

2. Accepted: November 2, 2020
3. Accepted Manuscript published: November 3, 2020 (version 1)
4. Version of Record published: November 20, 2020 (version 2)

? 2020, Nerli et al.

## Metrics

• 802
Page views
• 161
• 0
Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

A two-part list of links to download the article, or parts of the article, in various formats.

### Open citations (links to open the citations from this article in various online reference manager services)

1. Cell Biology
2. Developmental Biology

# Zebrafish Posterior Lateral Line primordium migration requires interactions between a superficial sheath of motile cells and the skin

Damian E Dalle Nogare et al.
Research Article

The Zebrafish Posterior Lateral Line primordium migrates in a channel between the skin and somites. Its migration depends on the coordinated movement of its mesenchymal-like leading cells and trailing cells, which form epithelial rosettes, or protoneuromasts. We describe a superficial population of flat primordium cells that wrap around deeper epithelialized cells and extend polarized lamellipodia to migrate apposed to the overlying skin. Polarization of lamellipodia extended by both superficial and deeper protoneuromast-forming cells depends on Fgf signaling. Removal of the overlying skin has similar effects on superficial and deep cells: lamellipodia are lost, blebs appear instead, and collective migration fails. When skinned embryos are embedded in Matrigel, basal and superficial lamellipodia are recovered; however, only the directionality of basal protrusions is recovered, and migration is not rescued. These observations support a key role played by superficial primordium cells and the skin in directed migration of the Posterior Lateral Line primordium.

1. Developmental Biology

# Wnt3 distribution in the zebrafish brain is determined by expression, diffusion and multiple molecular interactions

Sapthaswaran Veerapathiran et al.
Research Article

Wnt3 proteins are lipidated and glycosylated, secreted signaling molecules that play an important role in zebrafish neural patterning and brain development. However, the transport mechanism of lipid-modified Wnts through the hydrophilic extracellular environment for long-range action remains unresolved. Here, we determine how Wnt3 accomplishes long-range distribution in the zebrafish brain. First, we characterize the Wnt3-producing source and Wnt3-receiving target regions. Subsequently, we analyze Wnt3 mobility at different length scales by fluorescence correlation spectroscopy and fluorescence recovery after photobleaching. We demonstrate that Wnt3 spreads extracellularly and interacts with heparan sulfate proteoglycans (HSPG). We then determine the binding affinity of Wnt3 to its receptor, Frizzled1 (Fzd1), using fluorescence cross-correlation spectroscopy, and show that the co-receptor, low-density lipoprotein receptor-related protein 5 (Lrp5), is required for Wnt3-Fzd1 interaction. Our results are consistent with the extracellular distribution of Wnt3 by a diffusive mechanism that is modified by tissue morphology, interactions with HSPG and Lrp5-mediated receptor binding, to regulate zebrafish brain development.