Notch1 and Notch2 collaboratively maintain radial glial cells in mouse neurogenesis

During mammalian corticogenesis, Notch signaling is essential to maintain neural stem cells called radial glial cells (RGCs) and the cortical architecture. Because the conventional knockout of either Notch1 or Notch2 causes a neuroepithelial loss prior to neurogenesis, their functional relationship in RGCs remain elusive. Here, we investigated the impacts of single knockout of Notch1 and Notch2 genes, and their conditional double knockout (DKO) on mouse corticogenesis. We demonstrated that Notch1 single knockout affected RGC maintenance in early to mid-neurogenesis whereas Notch2 knockout caused no apparent defect. In contrast, Notch2 plays a role in the RGC maintenance as Notch1 does at the late stage. Notch1 and Notch2 DKO resulted in the complete loss of RGCs, suggesting their cooperative function. We found that Notch activity in RGCs depends on the Notch gene dosage irrespective of Notch1 or Notch2 at late neurogenic stage, and that Notch1 and Notch2 have a similar activity, most likely due to a drastic increase in Notch2 transcription. Our results revealed that Notch1 has an essential role in establishing the RGC pool during the early stage, whereas Notch1 and Notch2 subsequently exhibit a comparable function for RGC maintenance and neurogenesis in the late neurogenic period in the mouse telencephalon.


Introduction
The evolutionarily conserved Notch signaling pathway is one of the most essential signaling pathways for regulating the proliferation of stem cells in various developmental processes (Andersson et al., 2011;Siebel and Lendahl, 2017). Mammals have four Notch homologues, Notch1, 2, 3, and 4. Experiments on mice have shown that mutant mice show systemic developmental defects and embryonic lethality at earlier embryonic stages; this shows that in mammalian development, Notch1 and Notch2 largely contribute to embryogenesis (Conlon et al., 1995;Hamada et al., 1999;Krebs et al., 2003bKrebs et al., , 2000McCright et al., 2006McCright et al., , 2001Radtke et al., 1999;Swiatek et al., 1994). Notch1 and Notch2 are thought to be redundant (Krebs et al., 2003a), and these two genes are expressed in radial glial cells (RGCs) of the dorsal cortex (Higuchi et al., 1995;Irvin et al., 2001). While previous studies have identified that Notch1 largely contributes to the maintenance of neural stem cells (NSCs) (Yang et al., 2004;Yoon et al., 2004), the early lethality of Notch1 and Notch2 knockout mice has limited the phenotypic analysis of telencephalon development from the mid-neurogenic stage (around E13.5) onward. The functional differences between Notch1 and Notch2 for the maintenance of NSCs throughout telencephalon development remains elusive. In the present study, we address how Notch signaling regulates NSC maintenance in the dorsal cortex throughout neurogenesis by combinations of the conditional knockout for Notch1 and Notch2 genes.

Animals
All animal experiments were performed in compliance with the guidelines for animal experiments at the RIKEN Center for Biosyshttps://doi.org/10.1016/j.neures.2020.11.007 0168-0102/© 2020 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).

Table 1
Knockout mice used in this study.

Microscopic observations
We observed antibody-stained cryostat sections in a confocal microscope (LSM880, inverted laser scanning confocal microscope, Carl Zeiss). Whole brains shown in Fig. S1 were photographed by a stereo fluorescence microscope (Macro Zoom Fluorescence Microscope System, MVX10 with DP71 digital microscope camera, Olympus).

Measurement of cell cycle exit rate
The cell-cycle exit rate is defined as the rate of cells that exit cell cycle during approximately one cell cycle length among mitotic cells (Chenn and Walsh, 2002). To label proliferating cells, EdU (1 mg/25 g body weight, A10044, Thermo Fisher Scientific) was administrated intraperitoneally into pregnant mice at E11. We fixed the embryos 12 h after the EdU administration to detect EdU incorporation in a single cell cycle, given that the cell cycle length at E11.5 was about 8.1 h (Takahashi et al., 1995). The brains were collected, fixed, followed by double staining for EdU and Ki67, which is a marker expressed in mitotic cells. No less than three sections from three different brains were counted for the fractions of EdU+/Ki67+ cells and all EdU incorporating cells. We calculated the cell cycle exit rate as the fraction of (EdU + Ki67-cells) relative to all EdUincorporated cells (EdU + Ki67+ plus EdU + Ki67-cells) in the unit area of each dorsal cortex at E11.5. No less than three sections from three different brains were counted for the fractions of EdU+/Ki67+ cells and all EdU incorporating cells.

Single cell isolation, library construction and sequencing
Details for the single cell dataset analysis from E14 mice have been described in the method section of a separately submitted manuscript (Wu et al., 2020). Single cell isolation and library construction from E11 and E16 mice have been described there.

Bioinformatics analysis of single cell data
Sequenced data was mapped to 10 mm using cellranger 2.0.2 (10X GENOMICS). All data were further analyzed by a bioinformatics pipeline Seurat (2.3.4) (Butler et al., 2018).

The loss of Notch1 but not Notch2 enhances neurogenesis during early neurogenesis
We first investigated the effects of Notch1 / and Notch2 / on the differentiation of RGCs during the early to mid-neurogenic period, by crossing Emx1-Cre KI driver line (Guo et al., 2000) with Notch1 flox/flox or Notch flox/flox mouse lines (Table 1). The numerical ratio of RGCs and differentiating cells was measured at embryonic day E11.5 (a late proliferative stage) and E13.5 (the mid-neurogenic stage when layer 4 neurons are generated), by counting cells expressing Sox2, a marker expressed in NSCs and those expressing Tbr2, a marker for committed neural progenitors in the dorsal cortex. We found that at both E11.5 and E13.5, the elimination of Notch1 in the dorsal cortex greatly reduced cell number per unit width (100 m) of cryosections (hereafter referred to as cell number/unit area) of RGCs, whereas Notch2 / showed no significant change in RGC cell number/unit area ( Fig. 1A-C).

The simultaneous loss of Notch1/2 functions causes synergetic defects
We also examined Notch1 / ; Notch2 / DKO phenotype in corticogenesis. We observed that Notch1 / ; Notch2 / brain showed more drastic defects than those of Notch1 single knockout. There was a remarkable increase in Tbr2 (+) differentiating cells even at E11.5 ( Fig. 1A and C). At E13.5, only a small number of proliferating RGCs remained, and epithelial surface was compromised in the dorsal cortex of DKO brains ( Fig. 1B and D). These DKO phenotypes suggest that the simultaneous loss of Notch1 and Notch2 synergistically affects the maintenance of RGCs and neurogenesis (see Discussion).

3.3.
Notch2 is dispensable in the presence of Notch1 from the early to mid-neurogenic period To evaluate the rate of neurogenesis, we measured cell-cycle exit rate for Notch1 / and Notch2 / single KO and Notch1 / ; Notch2 / DKO brains. Notch1 / single KO and Notch1 / ; Notch2 / DKO showed severe increases in the cell cycle exit compared to double heterozygotes ( Fig. 2A and B); however, Notch2 / single KO brains did not show such results.
Next, we investigated whether proliferating cells, including both apically dividing RGCs and basally dividing intermediate progenitors (IP), changed their cell number/unit area in those mutant brains. We measured the cell numbers of RGCs and IPs undergoing mitosis in the unit area of the developing lateral cortex by staining for phospho-histone H3. In Notch1 / ; Notch2 / DKO brains, non-apical dividing cells, which were supposed to be IPs, were increased ( Fig. 3A and C), suggesting precocious neurogenesis at E11.5. At E13.5, Notch1 / brains also showed a significant increase in the IP cells, suggesting an enhanced differentiation of RGCs into IPs by Notch1 KO alone. Conversely, in Notch1 / ; Notch2 / DKO brains at E13.5, RGCs and total dividing cells were largely decreased in cell number/unit area, suggesting that a large part of proliferating cells (RGCs and IPs) had been differentiated ( Fig. 3B and D). We noted cell death in the ventricular zone (VZ) of Notch1 / brain (Fig. 3B) as previously described (Mason et al., 2006). Interestingly, Notch2 knockout did not significantly alter the cell cycle exit rate nor affected the proliferation of progenitor cells, both at E11.5 and E13.5. Taken together, these data suggest that Notch2 is dispensable as far as the Notch1 locus is normal in the developing telencephalon during the early (E11.5)-mid (E13.5) neurogenic period. However, in the absence of Notch1 activity, Notch2 can partly take over Notch1 for RGCs self-renewal (see Discussion).
3.4. Notch1 and Notch2 might have a comparable level of signaling activity at late neurogenic stage Next, we examined phenotypes of various Notch mutant brains at a much later stage such as E18.5, where neurogenesis had been nearly completed. To visualize the developing dorsal cortex where a various combination in the dosage of Notch1 and Notch2 genes was depleted, we crossed Notch1 / and/or Notch2 / mice with CAG-CAT-GFP mice (Kawamoto et al., 2000), where the Emx1-Cre active cells were visualized with GFP (Fig. 4A). Since we found an apparent variety in size of the brains depending on genotypes (Fig. S1), we counted the total number of Sox2(+) cells in the VZ of the telencephalon in the genotypes shown in Fig. 4B. Interestingly, Notch1 +/+ ; Notch2 / brains showed a similar reduction of Sox2 (+) cell number to that in Notch1 / ; Notch2 +/+ brains, suggesting an increase in Notch signaling activity by Notch2 to a comparable level with that by Notch1, in contrast to the early stages. We also found that RGCs were completely depleted in the Notch1 / ; Notch2 / DKO mice. Furthermore, brains bearing a single Notch gene dosage such as Notch1 +/ ; Notch2 / and Notch1 / ; Notch2 +/ also exhibited a severe depletion of RGCs, compared to brains of double heterozygous Notch1 +/ ; Notch2 +/ or that of a single KO of Notch1 / or Notch2 / , while it was less severe than that of the Notch1 / ; Notch2 / DKO brains (Fig. 4B); when the dosage effects of Notch genes are examined, the wild type with the genotype of Notch1 +/+ ; Notch2 +/+ would be the appropriate control. We obtained only two wild type (Notch1 +/+ ; Notch2 +/+ ; Emx1-Cre +/-) animals from the crosses for these experiments, which gave 1482, and 1386 for Sox2 (+) cell number. Whereas these values are larger than the average value for (Notch1 +/ ; Notch2 +/ ), these values cannot be statistically compared with numerical values of the other genotypes, because of too few sample numbers for the wild type. However, it is highly likely that phenotypic severity depends on the Notch gene dosage, being increased according to a decrease in the dosage of the four genes of Notch1 and Notch2 at the late embryonic stage, as far as we compare the Sox2 (+) cell numbers among the above genotypes available for statistical comparisons. Those mice that had only one dose of the Notch1 or Notch2 gene (Notch1 +/ ; Notch2 / and Notch1 / ; Notch2 +/ ) showed a severe phenotype in Sox2 (+) cell number to a similar degree. This again implies that Notch1 and Notch2 have comparative levels of signaling activity (or expression) for RGC maintenance at the late embryonic stage, in contrast to their large difference at the earlier stage ( Fig.1) Tbr2 = 105.6 ± 15.5 cells (n = 9 embryos). 10 pregnant females were used. (D) Histograms of Sox2(+) cells (left) and Tbr2(+) cells (right) in E13.5 dorsal cortices for the same genotype series as in C. Notch1 +/ ; Notch2 +/ : Sox2 = 236.8 ± 29.3 cells, Tbr2 = 126.7 ± 18.7 cells (n = 7 embryos); Notch1 / : Sox2 = 181.5 ± 17.1 cells, Tbr2 = 136.4 ± 8.6 cells (n = 3 embryos); Notch2 / : Sox2 = 245.7 ± 28.9 cells, Tbr2 = 107.7 ± 16.8 cells (n = 5 embryos); Notch1 / ; Notch2 / : Sox2 = 115.8 ± 18.2 cells, Tbr2 = 154.6 ± 9.5 cells (n = 3 embryos). Five pregnant females were used. Values are the mean ± S.D. Statistical significance calculated using one-way ANOVA with Tukey's correction for multiple comparisons; *P < 0.05, **P < 0.01, ***P < 0.001 and ****P < 0.0001 (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article).
It might not be straightforward to compare the Notch1 and Notch2 activities from these data because a previous study showed a differential transcriptional activity between Notch1 and Notch2 (Shimizu et al., 2002). However, the simplest interpretation is that the difference between Notch1 and Notch2 in the slope of mRNA increases over time; that is, an almost zero level at E11.5 to a certain degree of Notch2 expression is a major reason why the relative activity between Notch1 and Notch2 changes temporarily. Notch1 plays a major role at early stage, but the overall contribution of Notch1 and Notch2 is comparable to each other in RGC maintenance represented by Sox2 (+) cell abundance.

Notch1 is the major Notch signaling receptor at early neural development
In this study, we have obtained multiple lines of evidence that Notch1 and Notch2 act collaboratively in RGC maintenance. Notch1 plays a more substantial role than Notch2 in the early developmental stage because we do not detect a significant defect in the progenitor pool size or neurogenesis in Notch2 / brains during E11.5∼E13.5. This is consistent with the observation of Notch1 / brains by a previous study (Yang et al., 2004). We observed that combinatorial knockout of Notch1 and Notch2 exhibits a severe RGC depletion in the telencephalon than that of defects by Notch1 single knockout, indicating that Notch2 has some activity as a Notch signaling receptor. Alternatively, Notch1 knockout may possibly induce an upregulation of the Notch2 gene, and consequently compensate the decrease in Notch signaling activity by the loss of Notch1 to some degree, compared with the simple situation of the depletion of Notch1 activity plus the normal level of Notch2 activity. In such a case, Notch1 / ; Notch2 / DKO brains show the phenotypic sum due to the loss of both wild type Notch1 activity and an overactivated Notch2. Whether the lack of one Notch gene affects the expression of the other Notch genes will be an interesting issue to be tested. Especially for Notch2 expression level in cortical progenitor cells of the Notch1 mutant at E13.5, needs to be examined. Taken together, Notch1 and Notch2 are major receptors of Notch signaling in the embryonic telencephalon development, while Notch1 accounts for the major signaling receptor activity for NSC self-renewal during the proliferative and early neurogenic stages.

Role of Notch3 in NSC maintenance and neurogenesis at the embryonic development
Notch1/Notch3 double mutant has no significant difference from Notch1 single knockout in the brain phenotype (Krebs et al., 2003b), suggesting that Notch3 does not appear to have a role in the embryonic brain development, whereas Notch3 is essential for the maintenance of quiescent neural stem cells (Basak et al., 2012;Kawai et al., 2017). We note the possibility that the embryonic Notch3 expression (Fig. 5H) may contribute to NSC proliferation, given that the depletion of Delta-like1 (Dll1), a major ligand of Notch signaling, shows a much severe depletion of NSC at an earlier stage, compared with the Notch1/Notch2 double mutant (data not shown).

Interaction between Notch and various signaling pathway
At the downstream of Notch, combinatorial Hes1, Hes3 and Hes5 signaling play an important role in the maintenance of NSCs (Hatakeyama and Kageyama, 2006). These Hes genes play a pivotal role by interacting with not only Notch but also other signaling proteins assuming regional and context dependent expression such as BMP/SMAD, SHH and JAK/STAT to converge various signaling inputs into the NSC property (Kriegstein and Alvarez-Buylla, 2009;Mehler, 2002). Thus, it is intriguing issue to address the question of whether the differential temporal pattern of Notch1 and Notch2, in cooperation with spatially and temporally varying signaling pathways, contributes to the control of the NSC fate determination such as active cycling, gliogenesis, and quiescence.
In this study, we demonstrate that Notch1 is a major player during stem cell proliferation and neurogenesis while Notch2 shows up in the later stages of neurogenesis. In this sense, the situation might resemble the ventral side of the telencephalon, whose subventricular zone maintains adult NSCs, when they are quiescent. Notch2 relatively works for their maintenance, whereas Notch1 mainly functions for the activation of adult NSCs (Engler et al., 2018). It is intriguing to assume a higher dependency on Notch2 to maintain adult neural stem cells gradually appears from embryonic stage, whereas we need to examine the temporally differential expression pattern of Notch1 and Notch2 in the dorsal cortex is also preserved in the dentate gyrus primordium, as well as in the ventral side of the embryonic cerebral cortex.

Declaration of Competing Interest
The authors report no declarations of interest.