Notch signaling is maintained in prospective neurons

Following the completion of mitosis, prospective neurons remain attached to the ventricular surface for a transition period of up to 20 h before they eventually retract their apical endfoot as they start expressing the early differentiation marker class III β-tubulin (Tuj1) [1, 27]. While it is accepted that Notch activity is switched off in differentiated cells, the state of signaling during the transition period that precedes has never been explored. We decided to address this point in a chicken transgenic line carrying a fluorescent reporter of Notch activity. A transgene containing the promoter of the Hairy and Enhancer of Split 5 (Hes5) gene (a target of the Notch pathway) upstream of a destabilized nuclear Venus coding sequence (Venus-NLS-PEST [VNP]) [28] was inserted into the chicken genome (Fig 1A, and see Materials and methods). We first investigated the intensity of the VNP signal through immunostaining (the native VNP signal does not allow direct visualization) in normal conditions. Hes5-VNP distribution was consistent with the endogenous chicken Hairy and Enhancer of Split 5.1 (cHes5.1) expression at embryonic day (E) 4 (S1A Fig, [29]), while nuclear localization of the VNP signal provided a better cellular resolution. Transverse sections of the spinal cord were analyzed during early neurogenesis (E3 and E4), and VNP signal intensity was compared between progenitors and neurons (Fig 1B, the red line delimits the boundary of the differentiated zone in the color code panel). While progenitors displayed a wide spectrum of VNP intensities, all neurons (identified by the expression of the neuron-specific RNA-binding proteins HuC and HuD [HuCD]) showed low VNP levels. This is consistent with data obtained in the mouse cortex using a Hes1 reporter suggesting that Notch target gene expression oscillates in progenitors and is switched off during differentiation [3].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Notch signaling is maintained in prospective neurons.

(A) Schematic representation of the Hes5-VNP sequence that was inserted in the Notch reporter transgenic chick line. (B) Left: Transverse sections of the NT of the Hes5-VNP transgenic line at E3 and E4 immunostained for Venus (green) and HuCD (red) to label neurons. Middle: Color coded map of Hes5-VNP intensity. The red line separates HuCD⁻ from HuCD⁺ cells. The black dotted lines delineate the ventral limit of the roof plate and dorsal limit of the motor neuron domain. Right: Distribution of the Hes5-VNP signal intensity in HuCD⁻ and HuCD⁺ cells. Note that cells within the limits of the black dotted lines of the color code panel were labeled in black in the HuCD⁻ population. (C) Top: Time course of the protocol. Bottom: Distribution of the Hes5-VNP signal intensity in FT⁺/HuCD⁻ cells. This population is then divided into EdU⁺ (blue) and EdU⁻ (magenta) cells. A minimum of 58 cells collected from four embryos were analyzed for each group. (D) Left: Transverse sections of the dorsal NT in the Hes5-VNP transgenic line at E4 immunostained for Venus (green), Neurog2 (red), and HuCD (blue). Bottom: Enlarged view of the boxed area showing representative examples of Neurog2+ cells. Right: Distribution of the Hes5-VNP signal intensity in Neurog2⁻ and Neurog2⁺ cells. The latter population was divided based on Neurog2⁺ signal intensity. A minimum of 75 cells collected from six embryos were analyzed for each group. Horizontal bars correspond to medians. ns, p > 0.05; **p < 0.01, ***p < 0.001 (Kruskal-Wallis test). Underlying data are provided in S1 Data. Scale bar represents 25 μm. See also S1 and S2 Figs. E, embryonic day; EdU, 5-ethynyl-2′-deoxyuridine; FT, FlashTag; Hes5, Hairy and Enhancer of Split 5; HuCD, neuron-specific RNA-binding proteins HuC and HuD; Neurog2, Neurogenin 2; ns, nonsignificant; NT, neural tube; VNP, Venus-NLS-PEST.

https://doi.org/10.1371/journal.pbio.2004162.g001

We next assessed whether VNP intensities would reliably reflect perturbations of Notch signaling activity. Notch gain-of-function through overexpression of the Notch intracellular domain (NICD) resulted in a 6-fold increase in VNP intensities as well as a blockade of differentiation (S1B Fig). Conversely, incubation of NT explants with the Notch signaling inhibitor N-(3,5-difluorophenylacetyl-L-alanyl)-S-phenylglycine t-ButylEster (DAPT) led to a rapid reduction of the VNP signal, suggesting a half-life of the reporter of less than 4 h, reaching down to the background level measured in neurons within 6 h of incubation (S1C Fig). Thus, the Hes5-VNP chicken line appears as an excellent tool to monitor the dynamics of Notch signaling in the embryonic spinal cord. Progenitors located in the roof plate region, and ventrally up to the dorsal limit of the motor neuron progenitor domain (delimited by the black dotted lines in Fig 1B, Middle) displayed a lower Notch activity compared to the rest of the ventricular zone (VZ) (Fig 1B, Right; cells in those regions are represented by gray dots). This pattern is consistent with previous reports that floor and roof plates are signaling centers displaying low Notch activity, while the reduced Notch levels measured in the motor neuron progenitor domain may be associated with the early and massive motor neuron differentiation process [30].

Then, we sought to characterize the level of Notch activity in prospective neurons. To this end, we first took advantage of the FlashTag (FT) technique, based on the ability of the cell-permeant dye carboxyfluorescein succinimidyl ester (CFSE) to fluorescently label intracellular proteins. Previous experiments in the mouse developing cortex have shown that upon injection in the ventricles, FT dyes preferentially enter progenitor cells undergoing mitosis, offering a convenient means to synchronously label a cohort of dividing cells and track their progeny over different time periods [31]. To validate the technique and calibrate its dynamics in the chick spinal cord, FT was injected in the NT at E2.75 and fluorescence was monitored at different time points. Fifteen minutes after injection, FT⁺ cells’ nuclei were exclusively located near the ventricular surface, and many were positive for phospho-Histone H3, consistent with the preferential labeling of cells undergoing mitosis (S2A Fig). Increasing incubation times (1 h, 4 h) correlated with FT⁺ nuclei being located at progressively more basal positions and no longer in mitotic cells. This indicates that incorporation into mitotic cells was restricted to a short time period after FT injection, allowing the labeling of a cohort of cells that collectively undergo mitosis in a very narrow time window. We then asked whether the progeny of mitotic cells labeled with FT entered S phase or exited the cell cycle and differentiated. Embryos were injected with the FT dye at E2 or E2.75 (respectively before and after the onset of neurogenic divisions). EdU (5-ethynyl-2′-deoxyuridine) was injected 3 h after FT injection and then every 4 h in order to cumulatively label the whole population of cycling cells (S2B Fig). Embryos were harvested at different time points after FT injection and labeled for EdU incorporation and HuCD expression (S2C and S2D Fig). In both conditions, the number of FT⁺/EdU⁺ cells reached a plateau by 12 h after FT injection, indicating a saturating labeling of cycling progenitors with EdU (S2C and S2D Fig). Consistent with the fact that virtually all progenitors undergo symmetric proliferative divisions at E2 (excluding the motor neuron domain, which differentiates earlier than the rest of the NT and was excluded from the analysis), the plateau of FT⁺/ EdU⁺ was close to 100% in embryos injected at E2 (S2D Fig), and no FT⁺/HuCD⁺ cells were found. By contrast, in embryos injected at E2.75, the plateau of FT⁺/EdU⁺ cells remained below 65% (S2D Fig). Thus, about one third of FT⁺ cells remained EdU⁻. Within this population, the proportion of HuCD⁺ neurons increased between 12 h and 16 h (S2C and S2D Fig). Therefore, three populations could be discriminated based on EdU incorporation and HuCD expression: cycling progenitors (EdU⁺/HuCD⁻), prospective neurons (EdU⁻/HuCD⁻), and neurons (EdU⁻/HuCD⁺). We then investigated the level of Notch signaling in these three populations using FT injection in the Hes5-VNP chicken line. Strikingly, levels of Notch activity in EdU⁻/HuCD⁻ prospective neurons remained elevated 12 h after mitosis (with a median of 0.62 and a mean of 0.80 ± 0.09, the average VNP intensities measured in HuCD⁻ and HuCD⁺ cells being normalized to 1 and 0, respectively [Fig 1C]). This sustained activity is not due to inertia of the Venus reporter fluorescence, because DAPT treatment of NT explants results in complete loss of Venus fluorescence within 6 h (S1C Fig). Hence, prospective neurons maintain high Notch signaling activity up to 12 h after they exit the cell cycle and until they enter differentiation.

To strengthen these results, we sought to identify the population of prospective neurons by another means. As proneural genes promote cell cycle exit and neural commitment [8], they are likely to be expressed at high levels in prospective neurons. We focused on the proneural gene Neurogenin 2 (Neurog2), which is widely expressed in the chick spinal cord [32]. Neurog2 was strongly expressed at the basal limit of the VZ but also in scattered cells within the VZ, albeit at lower levels (Fig 1D). Cumulative EdU incorporation and HuCD staining indicated that these two populations, referred to as Neurog2^Low and Neurog2^High, had mostly exited the cell cycle (S2E Fig), while only a fraction had differentiated (S2F Fig). Thus, the vast majority of Neurog2⁺/HuCD⁻ cells correspond to prospective neurons, amongst which, Neurog2^High cells are likely to be closer to differentiation (as twice more Neurog2^High than Neurog2^Low have started to express the differentiation marker HuCD [S2F Fig]). We compared the level of Notch activity in Neurog2⁻/HuCD⁻ cells (which closely match the progenitor population), Neurog2^Low/HuCD⁻ and Neurog2^High/HuCD⁻ cells (prospective neurons), and HuCD⁺ neurons. Remarkably, Neurog2 negative, Low, and High populations of HuCD⁻ cells exhibited progressively lower Notch activity but remained above the level measured in the HuCD⁺ neuronal population (Fig 1D).

Taken together, these results indicate that Notch signaling is maintained in prospective neurons until they eventually differentiate. This raises the question of the importance of maintaining Notch activity during the events preceding differentiation.

Maintenance of Notch signaling is required for proper neuronal delamination

A hallmark of neuronal differentiation is the withdrawal of the apical attachment from the ventricular surface [1, 2, 27, 33]. To gain insight into the cellular events that accompany this delamination process, we investigated three parameters in parallel: the size of the apical area, the level of N-Cadherin at apical junctions, and the expression of the early differentiation marker Tuj1 at the apical surface (i.e., in nascent neurons that are still attached). These parameters were analyzed at 6, 12, 18, and 24 hours after electroporation (hae), focusing on single electroporated cells surrounded by non-transfected neighbors (the latter were used as a reference for measurements; see Materials and methods). Electroporation targets a mix of cycling progenitors and apically attached prospective neurons. While early time points (6 h, 12 h) will still retain many progenitors, these will eventually divide and appear as pairs that will be discarded from the analysis, such that at later time points (18 h, 24 h), the selected population will be enriched in prospective neurons. In cells transfected with a ZO1-GFP control vector alone, a decrease in the apical area was apparent at 18 hae and was further enhanced at 24 hae (Fig 2A, top panel). We also observed a modest decrease of N-Cadherin levels at 24 hae, which was not significant when considering the whole population (Fig 2A, top panel). However, when “small” (below the median) and “large” (above the median) areas were discriminated at 24 hae, we observed a significantly lower level of N-Cadherin in cells with a small apical area (Fig 2B). Moreover, the differentiation marker Tuj1 was almost exclusively expressed in this population (Fig 2C). These results are consistent with a differentiation process, as apical constriction and N-Cadherin reduction are features associated with neuronal delamination [1, 2, 27, 33].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Sequence of events leading to neuron delamination.

(A) Left: Apical view of the NT electroporated at E2 with ZO1-GFP/iRFP (green), along with the constructs indicated on the left, and harvested at different hae, followed by an immunostaining for N-Cadherin. The boxed area indicates the cell of interest. Right: Quantification of the apical area ratio (ratio of the area of one transfected cell versus the mean area of four of its close non-transfected neighbors) and N-Cadherin level ratio (ratio of the average pixel intensity within the apical circumference of one transfected cell corrected by the background versus the mean of average pixel intensity of four of its close non-transfected neighbors) at different hae. Data represent mean + SEM. (B) N-Cadherin intensity ratio as a function of apical area ratio at 24 hae. Data were taken from (A). The “median” used as a threshold to discriminate between small and large apical areas corresponds to the median of the control (0.62). **p < 0.01; ***p < 0.001 (one-way ANOVA). (C) Top: Apical view of the NT transfected with ZO1-iRFP (green) along with the indicated constructs and immunostained for Tuj1 (red) and Par3 (blue). Bottom: Three-dimensional view of the cell represented above but showing only the ZO1-iRFP and Tuj1 stainings. Right: Scatterplot of the mean apical area ratio for Tuj1⁺ cells. Each point represents one apical area ratio calculated as in (A). n = 49, 66, 51 cells collected from five embryos were analyzed for control, Neurog2, and ΔMaml1, respectively. ns, p > 0.05; ***p < 0.001 (Kruskal-Wallis test). Horizontal bars correspond to means. Underlying data are provided in S1 Data. Scale bar represents 2 μm. See also S3 Fig. ΔMaml1, dominant-negative Mastermind-like 1; E, embryonic day; EP, electroporation; GFP, green fluorescent protein; hae, hour after electroporation; iRFP, infrared fluorescent protein; Neurog2, Neurogenin 2; ns, nonsignificant; NT, neural tube; Par3, Partition defective protein 3; ZO1, Zonula Occludens 1.

https://doi.org/10.1371/journal.pbio.2004162.g002

To characterize the evolution of these parameters over time more specifically in cells committing to differentiation, we sought to synchronize the differentiation process. To this end, cells were transfected with the proneural gene Neurog2. As previously reported [34–36], Neurog2 repressed the expression of the progenitor marker Paired box gene 6 (Pax6) (S3A Fig), induced cell cycle exit after 24 h (S3B Fig), and strongly increased the differentiation rate at 48 hae (S3D Fig). In this case, we observed a strong reduction of the apical area from 12 hae, while reduction of N-Cadherin was detected only at 24 hae (Fig 2A, middle panel). In addition, reduction of N-Cadherin levels and expression of Tuj1 were observed almost exclusively in cells with a “small” area 24 hae (Fig 2B and 2C). Thus, control and Neurog2-induced differentiating cells appear to follow a similar sequence of events: constriction of the apical area precedes the reduction of N-Cadherin levels at apical junctions and the expression of Tuj1. Importantly, 24 h after Neurog2 electroporation, when most electroporated cells have exited the cell cycle (S3B Fig), the majority of Neurog2⁺/HuCD⁻ cells correspond to prospective neurons, and accordingly, these cells retained high levels of Notch reporter expression (S3C Fig).

We next wanted to assess the role of Notch signaling in this context. To this end, we measured these same parameters in cells transfected with a dominant negative version of the Notch pathway transcriptional coactivator Mastermind-like 1 (ΔMaml1) [37, 38]. In contrast to Neurog2, ΔMaml1 directly inhibits Notch transcriptional targets. Consistent with this, transfection of ΔMaml1 induced a massive decrease of Notch activity at 24 hae in the HuCD⁻ population and pushed cells to differentiate faster than Neurog2 (S3E and S3F Fig). These cells reduced their apical area ratio and their N-Cadherin level earlier than in the Neurog2 situation (Fig 2A, bottom panels). However, while the constriction of the apical surface appeared earlier (6 hae), at later time points the average apical surface remained significantly larger than in the Neurog2 case (Fig 2A, 12, 18, and 24 hae, apical surface values for Neurog2 were inserted in the ΔMaml1 graph for comparison). Moreover, unlike in the control and Neurog2 situations, low N-Cadherin levels were no longer restricted to cells with a small apical surface (Fig 2B) and Tuj1-positive nascent neurons with abnormally large apical domains were observed (Fig 2C). Taken together, these data suggest that upon precocious blockade of Notch signaling, N-Cadherin reduction and neuronal differentiation occur before apical constriction is complete.

We then investigated whether the effects observed at the single cell level would have a global impact on the integrity of the NT. Very strikingly, in contrast to control and Neurog2 situations, ΔMaml1 overexpression led to a noticeable decrease of all apical markers analyzed on transverse views at 24 hae (Fig 3A and S4B Fig). This resulted one day later in a severe disruption of the ventricular wall associated with the presence of ectopic neuronal masses protruding into the spinal cord lumen (Fig 3B). Remarkably, only a fraction of these ectopic neurons corresponded to transfected cells (Fig 3B, see arrowheads), suggesting that the down-regulation of apical markers in the transfected population was sufficient to induce a massive disorganization at the tissue scale. While blocking Notch activity results in a decrease of apical markers all along the dorsal–ventral axis at 24 hae (Fig 3A and S4B Fig), breaches in the ventricular wall were observed one day later almost exclusively in the ventral region of the NT. Motor neurons are the first neurons to be detected in the spinal cord (at E2) and are already extensively differentiated at E3 in the ventral NT. This suggests that a large population of nascent motor neurons has collectively delaminated between E2 and E3, which may render the ventral NT more sensitive to a weakening of the apical network.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Effects of Notch signaling and apical constriction modulators on apical markers and tissue integrity.

(A) Transverse sections of the NT transfected at E2 with the indicated constructs, harvested at E3 and immunostained for N-Cadherin (red). (B) Transverse sections of the NT transfected at E2 with the indicated constructs, harvested at E4 and immunostained for N-Cadherin (red); and for Sox2 (red) and HuCD (blue) to label progenitors and neurons, respectively. Transfection is reported by GFP expression. Summary: Schematic of the effects observed on tissue integrity. Gray cells correspond to electroporated cells. Scale bar represents 50 μm. See also S4 Fig. ΔMaml1, dominant-negative Mastermind-like 1; E, embryonic day; EP, electroporation; GFP, green fluorescent protein; HuCD, neuron-specific RNA-binding proteins HuC and HuD; N, neuron; Neurog2, Neurogenin 2; NT, neural tube; P, progenitor; RII-C1, Shroom3 binding site on ROCK2; Shroom3, Shroom family member 3; Sox2, SRY (sex determining region Y) box 2.

https://doi.org/10.1371/journal.pbio.2004162.g003

The down-regulation of apical markers following Notch blockade is correlated with the presence of differentiating cells displaying large apical domains (Fig 2C). This suggests that in the control situation (or in Neurog2 expressing cells), apical constriction may help to confine low N-Cadherin levels to only small fractions of the apical junction network and contribute to preserving epithelial integrity when neurons delaminate. To functionally test this hypothesis, we sought to alter the size of the apical area in differentiating cells. Apical constriction was shown to rely on actomyosin contraction and is regulated by Rho-GTPases family members [39]. A typical example of apical constriction is observed in the neurulation process, during which the actin-binding protein Shroom family member 3 (Shroom3) induces apical constriction by recruiting Rho kinases (ROCKs) to adherens junctions [40]. We found that overexpression of Shroom3 forced apical constriction. Conversely, a fragment of ROCK2 designated as RII-C1 (Shroom3 binding site on ROCK2) shown to dominantly interfere with the interaction between endogenous full-length ROCK2 and Shroom3 led to an increase in apical areas, suggesting that Shroom family members are active at neurogenic stages and regulate the size of the apical footprint of neuroepithelial cells (S4C and S4D Fig). Co-transfection of Shroom3 with ΔMaml1 strongly reduced the apical area, bringing it down to the value measured in the Neurog2 situation, and increased N-Cadherin apical level (S4C and S4E Fig). Strikingly, this rescued the ΔMaml1 phenotypes: apical markers distribution was restored at 24 hae (Fig 3A, S4B Fig) and tissue integrity was no longer affected at 48 hae (Fig 3B). By contrast, inhibiting Shroom-ROCK2 interaction in Neurog2 transfected cells through overexpression of the RII-C1 fragment led to an increase of the apical area (S4C and S4D Fig), which correlated with a decrease of apical markers on transverse (S4B Fig) and apical views (S4C and S4E Fig), mimicking the ΔMaml1 overexpression phenotype. Consistently, this was followed by a disruption of the ventricular wall at 48 hae (Fig 3B), while Shroom3 and RII-C1 alone had no effect on apical marker localization (S4B Fig).

Taken together, these results suggest that sustained Notch activity is necessary in prospective neurons to allow reduction of the apical size to take place before apical junction markers are down-regulated and neurons delaminate, therefore preserving the integrity of the ventricular wall.

Dll1 levels control differentiation through the regulation of Notch activity

Having shown that maintenance of Notch signaling is critical during the last steps leading to differentiation, we next investigated the mechanisms regulating the level of Notch activity during this transition. Increase in proneural gene expression is known to be required for differentiation and is correlated with a reduction of Notch activity. However, the connection between these two events remains to be clarified. In the chick spinal cord, the Notch ligand Dll1 is an early target of Neurog2 [35], and functional approaches in the mouse cortex suggested that Dll1 expression was necessary and sufficient for neural differentiation [14]. We first investigated the role of Dll1 on neurogenesis in the chick spinal cord. Consistent with published results, we observed a strong increase in the differentiation rate of Dll1 transfected cells 48 hae (S5A Fig). By contrast, down-regulation of Dll1 following mosaic electroporation of a short hairpin RNA (shRNA) against chick Dll1 [41] reduced differentiation (S5B Fig). We then used the Hes5-VNP transgenic line to investigate the level of Notch activity following gain and loss of Dll1 function, focusing on HuCD⁻ undifferentiated cells. Consistent with their impact on differentiation, gain and loss of Dll1 function led to a decrease and an increase of Notch activity, respectively (S5C and S5D Fig). It should be noted that Dll1 is widely expressed in the spinal cord except for the dorsal dI6 and intermediate V1 interneuron domains. As Notch activity and differentiation rate following Dll1 misexpression were analyzed in the dorsal and intermediate regions of the NT irrespective of the endogenous expression of Dll1, our results may be slightly underestimated.

Mib1 blocks the ability of Dll1 to Cis-inhibit Notch signaling

Dll1 expression in a differentiating cell could lead to reduced Notch activity either indirectly by Trans-activation of Notch signaling in the neighbors that would therefore not Trans-activate in return (according to the lateral inhibition with feedback model) or directly through Cis-inhibition of Notch receptors in the same cell [20]. However, answering this question in vivo cannot be obtained solely by Dll1 misexpression, which would impact both Trans and Cis phenomena. Thus, we decided to take advantage of the ability of the ubiquitin ligase Mindbomb1 (Mib1) to promote Notch Trans-activation. We first tested the ability of Dll1 alone or with Mib1 to induce Trans-activation of Notch signaling in undifferentiated cells (HuCD-negative) 24 h after transfection. To this end, the intensity of the Hes5-VNP reporter was measured in non-transfected “neighbor cells” contacted by a minimum of four transfected cells. Dll1 alone was unable to Trans-activate signaling in neighbors (Fig 4B). In contrast, our measures upon Mib1 co-transfection indicated a trend towards increased Notch activity, although it failed to reach statistical significance (Fig 4B). We recently reported that in normal conditions, Mib1 is strongly enriched at the centrosome and barely detectable at the membrane in the NT, suggesting that only a fraction of it interacts with Dll1 [27]. To potentiate this interaction, we engineered a version of Mib1 constitutively tethered to the plasma membrane (mbMib1) by addition of an N-terminal myristoylation sequence. Remarkably, co-transfection of Dll1 with mbMib1 resulted in a significant increase of Notch activity in neighbor cells (Fig 4B). We reasoned that this higher Notch activity in neighbors should hinder their ability to differentiate. Indeed, while Dll1 alone had no impact on the differentiation rate of neighboring cells, the latter was consistently reduced following co-transfection of Mib1 and mbMib1 (Fig 4A and 4C). These data suggest that endogenous Mib1 is limiting and that Dll1 can Trans-activate the Notch pathway only when co-transfected with Mib1.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Mib1 blocks the ability of Dll1 to Cis-inhibit Notch signaling.

(A) Top: Transverse sections of the NT transfected at E2, with the indicated constructs and harvested at E4. Immunostaining for Sox2 (blue) and HuCD (green) labels progenitors and neurons, respectively. Transfection is reported by H2B-Cherry expression. Arrowheads indicate ectopic Sox2⁺ progenitors adjacent to HuCD⁺ transfected neurons. Bottom: Summaries of the effects of Dll1 and Mib1 on neurogenesis. Red and gray cells correspond to electroporated (ep) and non-electroporated cells, respectively. Round and star-shaped cells correspond to progenitors and neurons, respectively. Blue outlines indicate cells changing fate, autonomously or non-autonomously, in each condition. (B, D) Quantification of the Hes5-VNP signal intensity in HuCD⁻ cells either (B) non-transfected (surrounded by at least four transfected cells) or (D) transfected 24 hae with the indicated constructs. Data represent fold change compared to control. (B) n = 54, 35, 35, 54 cells were analyzed for control, Dll1, Dll1+Mib1, and Dll1+mbMib1, respectively. (D) n = 58, 59, 59, 66 cells were analyzed for control, Dll1, Dll1+Mib1, and Dll1+mbMib1, respectively. Data were collected from four to six embryos for each experimental group. ns, p > 0.05; **p < 0.01; ***p < 0.001 (Kruskal-Wallis test). (C, E) Quantification of the differentiation rate in (C) non-transfected neighbors (number of non-transfected HuCD⁺ cells adjacent to a HuCD⁺ transfected cell on the total number of adjacent cells) or (E) transfected cells (number of HuCD⁺ cells on total transfected cells) 48 hae with the indicated constructs. Data represent mean + SEM. n = 14 (6 embryos), 10 (8 embryos), 14 (6 embryos), 18 (6 embryos) sections were analyzed for control, Dll1, Dll1+Mib1, and Dll1+mbMib1, respectively. ns, p > 0.05; *p < 0.05; **p < 0.01; ***p < 0.001 (one-way ANOVA). Analyses were performed on the same sections for (B) and (D), and for (C) and (E). Underlying data are provided in S1 Data. Scale bar represents 50 μm. See also S5 Fig. Ct, control; Dll1, Delta-like 1; E, embryonic day; ep, electroporated; hae, hour after electroporation; Hes5, Hairy and Enhancer of Split 5; HuCD, neuron-specific RNA-binding proteins HuC and HuD; H2B-Cherry, Histone 2B fused to Cherry; mbMib1, Mib1 constitutively tethered to the plasma membrane; Mib1, Mindbomb1; ns, nonsignificant; NT, neural tube; Sox2, SRY (sex determining region Y) box 2; VNP, Venus-NLS-PEST.

https://doi.org/10.1371/journal.pbio.2004162.g004

We then analyzed the same parameters in transfected cells. Dll1 alone led to a noticeable decrease of Notch activity 24 hae in HuCD⁻ cells (Fig 4D), accompanied by an increased differentiation rate 48 hae (Fig 4A and 4E). If this effect was relying on a feedback-based lateral inhibition mechanism, as it was previously proposed, one would expect Mib1 to enhance the phenotype observed with Dll1 by promoting Trans-activation in neighbors. On the contrary, we observed that Mib1 and mbMib1 induced an increase of Notch signaling (Fig 4D) and a reduction of the differentiation rate compared to Dll1 alone (Fig 4A and 4E).

Taken together, these results indicate that Dll1 overexpression promotes differentiation of neural progenitors cell autonomously through Cis-inhibition of Notch signaling and that Mib1 is able to block this effect by converting Dll1 from a Cis-inhibiting to a Trans-activating ligand.

Mib1 blocks Notch Cis-inhibition to defer differentiation and preserve neuroepithelial integrity

We then sought to address whether Cis-inhibition of the Notch pathway by endogenous ligands occurs in the neuroepithelium. The above results suggest that Mib1 may promote Notch response not only in signal-receiving neighbors through Trans-activation but also in the signal-sending cell by blocking the Cis-inhibition process. To test this, we interfered with Mib1 function using a dominant negative version lacking its ring finger domain (ΔMib1) [42], which retains the interaction with Delta ligands but is unable to promote their maturation and endocytosis. Blocking Mib1 activity should therefore enhance Cis-inhibition and reduce Notch signaling cell autonomously. Indeed, overexpression of ΔMib1 reduced Notch activity (Fig 5A) and increased differentiation (Fig 5B), thus mimicking the effects of Dll1 alone, while co-electroporation of Dll1 and ΔMib1 did not significantly enhance the effect of either construct. However, blocking Mib1 function in a massive manner is also likely to alter Notch Trans-activation among contacting neighbor and sister cells. Thus, to restrict our analysis to isolated cells, embryos were electroporated under clonal conditions at E3 in order to target cells during the neurogenic peak and harvested shortly after (8 h) to minimize the probability of cell division. Clonal inhibition of Mib1 resulted in a significant decrease of Notch activity in electroporated cells as early as 8 hae (Fig 5C), providing strong evidence that Cis-inhibition takes place endogenously in the vertebrate nervous system.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. Mib1 blocks Notch Cis-inhibition to defer differentiation and preserve neuroepithelial integrity.

(A) Quantification of Hes5-VNP intensity in HuCD⁻ cells transfected with the indicated constructs at E2 and harvested 24 hae. Data represent fold change compared to control. A minimum of 108 cells were analyzed for each group. ns, p > 0.05; ***p < 0.001 (Kruskal-Wallis test). (B) Quantification of the differentiation rate (number of HuCD⁺ cells on total transfected cells). Data represent mean + SEM. n = 12 (3 embryos), 13 (6 embryos), 15 (6 embryos), 12 (4 embryos) sections were analyzed for control, Dll1, Dll1+ΔMib1, and ΔMib1, respectively. ns, p > 0.05; ***p < 0.001 (one-way ANOVA). (C) Quantification of the Hes5-VNP signal intensity in HuCD⁻ and isolated cells transfected at E3 with the indicated constructs at low voltage (15 V) and harvested 8 h later. Data represent fold change compared to control. n = 37 (3 embryos), 31 (7 embryos), 15 (3 embryos), and 25 (4 embryos) cells were analyzed for control, ΔMib1, Neurog2, and Neurog2+ΔMib1, respectively. **p < 0.01 (Mann-Whitney U test). (D) Left: Transverse sections of the NT transfected at E2 with the indicated constructs, harvested at E3 and immunostained for HuCD (green) to label neurons. Transfection is reported by H2B-Cherry expression. Right: Quantification of the differentiation rate (number of HuCD⁺ cells on total transfected cells). Data represent mean + SEM. For 24 hae, n = 14, 12, 19, 7 sections collected from six embryos for each experimental group were analyzed for control, Neurog2, Neurog2+ΔMib1, and ΔMib1, respectively. ***p < 0.001 (one-way ANOVA). Underlying data are provided in S1 Data. (E, F) Transverse sections of the NT transfected at E2 with the indicated constructs and immunostained for N-Cadherin (red); and for Sox2 (red) and HuCD (blue) to label progenitors and neurons, respectively. Transfection is reported by H2B-Cherry expression (green). Scale bar represents 50 μm. ΔMib1, dominant-negative Mib1; Ct, control; Dll1, Delta-like 1; E, embryonic day; EP, electroporation; hae, hour after electroporation; Hes5, Hairy and Enchancer of Split 5; HuCD, neuron-specific RNA-binding proteins HuC and HuD; H2B-Cherry, Histone 2B fused to Cherry; Mib1, Mindbomb1; Neurog2, Neurogenin 2; ns, nonsignificant; NT, neural tube; Shroom3, shroom family member 3; Sox2, Sox2, SRY (sex determining region Y) box 2; VNP, Venus-NLS-PEST.

https://doi.org/10.1371/journal.pbio.2004162.g005

While Mib1 blockade reduces Notch activity (Fig 5A and 5C), unlike Neurog2 it is not sufficient to rapidly force cells to exit the cell cycle and differentiate. However, it is essential for the process of asymmetric division and Mib1 loss-of-function will increase neurogenesis on a longer term [27]. Consistent with this, differentiation was only mildly increased at 24 hae compared to Neurog2 overexpression (Fig 5D), with no effect on N-Cadherin levels (Fig 5E). However, longer incubation times resulted in more neurons induced (Fig 5B) and large breaches in the ventricle (Fig 5E), suggesting that Mib1 regulates both the differentiation rate and the delamination process. Importantly, Shroom3 co-expression rescued NT morphology at 48 and 72 hae (Fig 5E). These results suggest that Mib1-dependent Notch maintenance is required to regulate the pace of differentiation and to allow proper neuronal delamination.

To bypass the effects of Mib1 in binary fate decisions and further characterize its function in the delamination and differentiation of prospective neurons, we performed similar experiments in cells also expressing Neurog2. ΔMib1 and Neurog2 co-expression led to a sharp decrease of Notch activity in prospective neurons at 8 hae (Fig 5C) and to a dramatic increase in differentiated HuCD⁺ cells at 24 hae compared to ΔMib1, Neurog2, or even ΔMaml1 alone (compare Fig 5D with S3F Fig). We then assessed the localization of apical markers at different times following transfection of Neurog2 and/or ΔMib1 (Fig 5E and 5F). Whereas neither ΔMib1 nor Neurog2 alone had any effect, N-Cadherin level was reduced upon co-expression at 24 hae, and breaches along the ventricular wall could be observed one day later and occasionally as early as 24 hae (Fig 5F). Moreover, co-transfection of Shroom3 rescued the morphology of the NT at 48 hae (Fig 5F).

Mib1 was previously shown to control the rate of neurogenesis in vertebrates by promoting Notch Trans-activation [43]. Our results suggest that Mib1 promotes Notch activity not only through Trans-activation in signal-receiving neighbors but also in the signal-sending cell by blocking the Cis-inhibition process. Overall, our data indicate that Mib1 actively sustains Notch signaling in prospective neurons to regulate the pace of differentiation and to allow proper neuronal delamination.

Ethics statement

All animal experiments, breeding, and care was compliant with the UK Animals (Scientific Procedures) Act 1986 and was authorized under a project license approved by the Roslin Institute Animal Welfare and Ethical Review Body and the UK Home Office.

Experiments performed with non-hatched avian embryos in the first two thirds of embryonic development time are not considered animal experiments according to the Directive 2010/63/EU.

Embryos

JA57 chicken fertilized eggs were provided by EARL Morizeau (8 rue du Moulin, 28190 Dangers, France). They were incubated at 38 °C in a Sanyo MIR-253 incubator for the appropriate time.

Production of the Hes5-VNP transgenic chicken line

The Hes5-VNP-NLS-PEST (Hes5-VNP) reporter transgene [28] was cloned into a lentiviral vector in reverse orientation to prevent the polyA sequence of the transgene from negatively affecting lentiviral packaging efficiency. Transgenic chicken production was carried out by injection of packaged pseudovirus generated from the Hes5-VNP lentiviral vector into blastoderm-stage chicken embryos in new laid eggs. Injected embryos were cultured to hatch and of six chicks, one male was shown to have the transgene present in blood DNA and, at sexual maturity, in semen DNA. The chimeric male (NOR4-21) was bred with stock hens and two transgenic G₁ male offspring were identified at hatch (NOR4-21:92 and:108). The position of the transgene insert sites in the chicken genome was determined by nested primer amplification of the insert site followed by sequencing, for both G₁ cockerels. Both carried a single transgene insert site in noncoding regions of the genome. A homozygous transgenic line was established from NOR4-21:92 to provide embryos homozygous for the Hes5-VNP transgene.

In ovo electroporation and plasmids

Electroporation in the chick NT was performed at E2 or E3 by applying 5 pulses of 50 ms at 25 V with 100 ms in between. For mosaic transfection analysis (Fig 5C, S5B and S5D Fig), lower voltage (3 pulses of 50 ms at 15 V with 950 ms in between) were applied to obtain isolated cells.

The following constructs have been previously described: pCX-EGFP-ZO1 [54], a gift from F. Matsuzaki; pCIG [55]; pCAGGS-ΔMaml1-EGFP [41], a gift from C. Marcelle; pRFP-RNAiC-cDll1-A and pRFP-RNAi-cDll1-B [41], a gift from C. Marcelle, were electroporated together and pRFP-RNAiC [56] was used as a control; pCA-Flag-Shroom3-Full and pCA-EGFP-HA-RII-C1 [40], a gift from M. Takeichi; pCAGGS-cMib1 [27]; pCAGGS-NICD was purchased from Addgene [57].

The following constructs were generated for this study: pCAGGS-Ngn2 was obtained by removing the IRES-GFP fragment from pCIG-Ngn2 [58], a gift from K. Storey. The chick version of Dll1 (cDll1) was cloned and inserted into pCAGGS and pCAGGS-IRES-H2B-Cherry. To generate a membrane-tethered version of Mib1 (pCAGGS-mbMib1), a myristoylation membrane localization sequence (MGCIKSKEDKGPAM from c-Yes kinase [59]) was inserted N-terminally upstream of Mib1, to not interfere with the C-terminal ring finger enzymatic domain of Mib1. For the dominant negative Mib1 (ΔMib1) [60], a version lacking the ring finger domain (aa 1–767) was amplified from the cMib1 and inserted into pCAGGS and pCAGGS-IRES-H2B-Cherry [27]. Other plasmids used are: pCX-EGFP (0.5 μg/μL), pCX-H2B-EGFP (0.5 μg/μL), pCX-iRFP-ZO1 (0.2 μg/μL), and pCAGGS-TetOn-IRES-H2B-iRFP (0.2 μg/μL). All plasmids were used at 1 μg/μL except where otherwise mentioned.

FlashTagging

FlashTagging procedures were adapted from [31]. CellTrace CFSE (Life Technologies, #C34554) was injected at 0.5 mM concentration into E2(HH12) or E2.75 chick NT. Embryos were incubated at 38 °C for the appropriate time until dissection. EdU was deposited at 4 h intervals as described below and schematized in Fig 1C and S2B Fig.

EdU labeling

Proliferating cells in the NT were labeled by in ovo incorporation of 5-ethynyl-2′-deoxyuridine (EdU). One hundred microliters of a 100 μM solution of EdU diluted in PBS was deposited on the embryo. Embryos were incubated for 1 h (S3B Fig) or more for cumulative EdU labeling (Fig 1C, S2C–S2E Fig), then dissected and fixed as described above. Immunodetection of EdU incorporated cells was carried out on cryostat sections using the Click-iT EdU imaging kit (Invitrogen).

Immunohistochemistry

Chick embryos were fixed for 1 h in ice-cold 4% formaldehyde/PBS and rinsed 3 times in PBS. For cryosections, they were equilibrated at 4 °C in PB/15% sucrose and embedded in PB/15% sucrose/7.5% gelatin before sectioning. Before immunostaining, cryosections were equilibrated at room temperature, degelatinized in PBS at 37 °C 3 times 5 min, before a 30-min blocking step in PBS-0.1%Triton/10% fetal calf serum (FCS). Slides were then incubated with the primary antibodies diluted in the blocking solution at 4 °C overnight. The following day, slides were washed 3 times 5 min in PBS-0.1%Triton, incubated 2 h with the adequate secondary antibodies at room temperature, washed again 3 times, and mounted with DAPI containing Vectashield (Vector Labs).

For en face views, fixed embryos were cut along their midline and bathed 1 h in blocking solution (PBS-0.3%Triton/10%FCS), followed by overnight incubation at 4 °C with the primary antibodies diluted in the blocking solution. The next day, embryos were washed 4–5 times with PBS-0.3%Triton, incubated overnight at 4 °C with the secondary antibodies, washed again 3 times 10 min in PBS-0.3%Triton and flat-mounted (apical side facing the coverslip) with DAPI containing Vectashield.

Primary antibodies used are: chicken anti-GFP (Aves Lab, 1:800); mouse anti-HuCD (clone 16A11, Life Technologies, 1:50); guinea-pig anti-Neurog2 (a gift from B. Novitch [61] 1:32,000); rabbit anti-phospho-Histone H3 (Millipore, 1:250); rabbit anti-Pax6 (Millipore, 1:500); mouse anti-N-Cadherin (clone GC-4, Sigma Aldrich, 1:100) (BD Biosciences, 1:250); mouse anti-βIII-tubulin (clone Tuj1; Covance, 1:500); rabbit anti-Par3 (Millipore, 1:1,000); mouse anti-ZO1 (clone 1A12, ThermoFischer, 1:100); goat anti-Sox2 (clone Y-17, Santa Cruz, 1:100). Secondary antibodies coupled to Alexa Fluor 488, Cy3, or Alexa Fluor 649 were obtained from Jackson laboratories.

In situ hybridization

In situ hybridization on gelatin mounted cryosections was performed as previously described [62]. All of the probes were synthesized using a DIG RNA labeling kit (Roche) as specified by the manufacturer. Antisense probes were prepared from the following linearized plasmids: cHes5.1 (a gift from D. Henrique), cHes1 (a gift from S. Bel-Vialar), and cDll1 (a gift from Olivier Pourquié). To generate hΔMaml1, cMib1, mShroom3, and mRII-C1 antisense probes, primers containing T3 and T7 overhangs were used to PCR amplify a region from the corresponding expression plasmids. The purified amplicon was then used as the template for antisense probe synthesis using T3 or T7 RNA polymerase.

Gelatin-mounted cryosections from overnight-fixed tissue were equilibrated at room temperature and degelatinized in PBS at 37 °C 3 times 5 min. Slides were treated 20 min in RIPA buffer (150 mM NaCl, 1% NP-40, 0.5% Na deoxycholate, 0.1% SDS, 1 mM EDTA, 50 mM Tris pH 8.0), postfixed in 4% paraformaldehyde/PBS for 10 min, and washed 3 times 5 min with PBS. The slides were then transferred in Triethanolamine buffer (100 mM triethanolamine, acetic acid 0.25% pH 8.0) for 15 min and washed 3 times 5 min in PBS. Slides were prehybridized during 1 h with 500 μL of hybridization solution (50% formamide, 5X SSC, 5X Denhardt’s, 500 μg/mL herring sperm DNA, 250 μg/mL yeast RNA) and hybridized overnight at 70 °C with the same solution in the presence of the heat-denatured DIG-labeled RNA probes. The following day, slides were placed in post-hybridization solution (50% Formamid; 2X SSC; 0.1% Tween20) at 70 °C, then washed in 0.2X SSC for 30 min at 70 °C and finally in 0.2X SSC at RT for 5 min. Slides were washed with buffer 1 (100 mM maleic acid, pH 7.5, 150 mM NaCl, 0.05% Tween 20) during 20 min at room temperature, blocked for 30 min in buffer 2 (buffer 1/10% FCS), followed by overnight incubation at 4 °C with the anti-DIG antibody (Roche) diluted 1:2,000 in buffer 2. The following day, slides were washed 3 times 5 min with buffer 1 and equilibrated for 30 min in buffer 3 (100 mM Tris pH 9.5, 100 mM NaCl, 50 mM MgCl₂). The signal was visualized by a color reaction using 500 μL of BM-Purple (Roche). The color reaction was allowed to develop in the dark at room temperature during 30 min–4 h and was stopped with PBS-0.1% Tween20.

Image acquisition and processing

Optical sections of fixed samples (en face views from half embryos or transverse views from cryosections) after immunofluorescence were obtained on a confocal microscope (model SP5; Leica) using 20× and 63× (Plan Neofluar NA 1.3 oil immersion) objectives and Leica LAS software. For image processing and data analysis, we used the ImageJ and FIJI software [63, 64]. Images were finally subjected to brightness and contrast adjustments to equilibrate channel intensities and background using ImageJ and FIJI software.

Image quantifications

Statistical analyses

The number of embryos and analyzed cells or sections are indicated in the figure legends. All data processing and statistical analyses were performed using Excel and GraphPad Prism softwares. For data following a normal distribution, significance was assessed using either a Student t test (S1B-Right, S2E, S2F, S3A, S3B, S3D, S3F, S5A, S5B and S6C Figs) to compare the mean of two groups or one-way ANOVA (Figs 2A–2C, 4C, 4E, 5B and 5D, S2D, S4D, S4E, S6A, S6B, S6D and S6F Figs) with Tukey correction to compare the mean of three or more groups. Data represent mean + SEM, ns, p > 0.05; *p < 0.05; **p < 0.01, ***p < 0.001. For the analysis of Hes5-VNP intensity distributions, significance was assessed using a Mann-Whitney U test (Fig 5C, S1B-Left, S3C, S3E, S5C and S5D Figs) to compare the median of two groups or a Kruskal-Wallis test (Figs 1C, 1D, 4B, 4D and 5A, S1C Fig) with Dunn’s correction to compare the median of three or more groups. ns, p > 0.05; *p < 0.05; **p < 0.01, ***p < 0.001.

DAPT NT culture

A trunk explant spanning the brachial to lumbar region was dissected from E3 Hes5-VNP embryos and grown in culture medium (F12/Penicillin Streptomycin/Sodium pyruvate 1 mM) for 8 h at 38.5 °C. We added to the culture medium either DAPT (N-(3,5-difluorophenylacetyl-L-alanyl)-S-phenylglycine t-ButylEster [InSolution γ-Secretase Inhibitor IX; Calbiochem] at a final concentration of 10 μM dissolved in DMSO) or DMSO alone at the indicated time (see schematic in S1C Fig). At the end of the culture period, embryos were fixed as described above and processed for immunohistochemistry.