A network of Notch-dependent and -independent her genes controls neural stem and progenitor cells in the zebrafish thalamic proliferation zone Open Access

The zebrafish brain continually grows and adds new neurons throughout development and into adult stages, and has a high capacity to regenerate neurons after injury (Kishimoto et al., 2012; Zupanc, 2001). The source of these new neurons are neural stem cells (NSCs) in at least 16 adult neural proliferation zones (NPZs) mostly located at the ventricular wall (Adolf et al., 2006; Grandel et al., 2006; Zupanc et al., 2005). During development, the embryonic neuroepithelium establishes highly active late embryonic and early larval proliferation zones, which develop into neural stem cell niche-like adult proliferation zones. So far, the mechanisms that balance NSC maintenance and neurogenesis in these larval NPZs are not well understood.

Hairy and Enhancer-of-split related (HES/Her) factors in NPZs control neurogenesis by inhibiting proneural gene mediated differentiation (Ishibashi et al., 1994; Kageyama et al., 2007; Sasai et al., 1992). In mice, absence of the basic helix loop helix (bHLH) transcription factors HES1 and HES5 accelerates differentiation, accompanied by loss of radial glia and later-born cell types (Hatakeyama et al., 2004). In Drosophila, the expression of E(spl) genes is Notch-dependent (Jennings et al., 1994), while the expression of hairy, encoding a closely related bHLH factor, is directed by local patterning signals (Riddihough and Ish-Horowicz, 1991). In mouse, Hes1 and Hes5 are downstream targets of Notch signaling, although Hes1 expression and stability are the result of additional Notch-independent input (Kageyama et al., 2007; Ohtsuka et al., 1999).

In zebrafish, the her gene family expressed in NPZs is significantly expanded compared with mammals (Zhou et al., 2012). Nine Hes5 zebrafish homologs are located in two gene clusters, her4.1-4.5 and her12 on chromosome 23, and her15.1, her15.2 and her2 on chromosome 11. There are two Hes1 homologs, her6 and her9, and four Hes6 homologs, hes6, her8a, her8.2 and her13. her4 and her15, her6 and her9, as well as her8a, expression has been characterized in adult NSC zones (Chapouton et al., 2011; Zhou et al., 2012). Zebrafish homologs of Hes1, Hes5 and Hes6 have also been investigated in the neural plate during early embryonic development (Bae et al., 2005; Pasini et al., 2004; Shankaran et al., 2007; Takke et al., 1999; Webb et al., 2011). So far, however, little is known about potential cross-regulatory interactions or redundancies among her genes in NSCs and during neurogenesis in larval NPZs.

Here, we analyze the impact of Notch signaling and her family genes on expression of specific her genes in NSCs and neural progenitor cells (NPCs) to reveal the organization of her gene regulatory networks in neurogenesis. We focus on the highly active early larval thalamic proliferation zone (TPZ) at the ventricular wall of the diencephalon in the thalamus proper (dorsal thalamus), the zona limitans intrathalamica (ZLI) and the prethalamus (Mueller, 2012; Scholpp and Lumsden, 2010), together constituting the thalamic complex. Based on her gene expression, we define distinct NSC and NPC populations in the TPZ. Results from her gene loss-of-function and overexpression experiments reveal cross-regulatory interactions of Notch-dependent and -independent her genes in the TPZ. Our data indicate that Notch-independent her genes are important for NSC maintenance and patterning in the TPZ. Notch-dependent her genes, however, appear to be not strictly required for NSC maintenance, but might rather modulate NPCs.

Based on published information (www.zfin.org; Chapouton et al., 2011; Shankaran et al., 2007; Sieger et al., 2004; Zhou et al., 2012) on zebrafish her family genes (Table S1), we selected those with expression in the diencephalon, namely homologs of mammalian Hes1 (her6, her9), Hes5 (her4;12 cluster and her2;15 cluster) and Hes6 (her8a, her8.2), for analysis of their expression patterns in the TPZ by whole-mount in situ hybridization (WISH; Fig. 1A-L; Fig. S1A-H). Genes within each cluster have high sequence similarity. Our her4 probe was designed to detect expression of all five her4 genes (her4.1-4.5) combined (subsequently referred to as her4). Similarly, our her15 probe detects her15.1 and her15.2 (referred to as her15). We analyzed 48 h post fertilization (hpf) to 96 hpf embryos and early larvae, when the TPZ is a highly active proliferation zone with a diversity of neurogenesis mechanisms involving neurog1, ascl1 and olig2 (Scholpp et al., 2009; Virolainen et al., 2012; see also Fig. 7C and Fig. 11A,B).

In the TPZ, her6, her9, her4, her2 and her15 expression were detected with strong WISH signals, mostly restricted to the wall of the diencephalic ventricle and the adjacent lateral ventricle (Fig. S1A′-H′). her2 has a similar expression domain as her15.1-15.2 in the same cluster. her12 expression appeared to be restricted to the hindbrain and later the midbrain-hindbrain boundary (Figs S1E,E′, S2A,G) and, thus, despite being in a cluster with her4 genes, differs significantly in expression. The Hes6 homologs her8a and her8.2 are broadly expressed in NPZs at 48 hpf, but expression decreases, so that at 96 hpf we detected only low levels of her8.2 and faint her8a expression in the TPZ (Fig. 1E,F,K,L; Fig. S1C-D′). In summary, while expression of the Hes6 homologs her8a and her8.2 decreases during early TPZ formation, Hes1 (her6, her9) and Hes5 (her2, her4.1-4.5, her15.1-15.2) homologs likely provide the major her gene activity in the maturing TPZ.

To assess Notch dependency of expression of Hes1, Hes5 and Hes6 homologs, we first analyzed mind bomb 1 mutants (mib1^ta52b) that lack Notch signaling (Itoh et al., 2003). Second, we globally overexpressed the Notch Intracellular Domain (NICD) by heat shock in 72 hpf larvae (Scheer and Campos-Ortega, 1999). Third, we treated larvae with the Notch inhibitor LY-411575 (Rothenaigner et al., 2011).

The Hes1 homologs her6 and her9 are still expressed at largely normal levels in mib1^ta52b mutant embryos (Fig. 1A-B′; with some ectopic her9 signal in the midbrain), and NICD overexpression only slightly increased her9 and hardly at all her6 expression (Fig. 1G-H′). Upon Notch inhibitor treatment, her6 and her9 expression continued, albeit at slightly reduced levels (Fig. 1M-N′). These experiments demonstrate her6 and her9 expression to be Notch-independent at analyzed larval stages.

In contrast, the expression of Hes5 (her2, her4, her12 and her15) and Hes6 (her8a and her8.2) homologs is lost in mib1^ta52b mutants (Fig. 1C′-F′; Fig.S2A-D′). The ectopic her15 expression in the tectum of mib1 mutants is different from the expression in wild type (WT; Fig. 1D,D′) and, similar to her9, may reflect dysregulation. Expression of Hes5 homologs was strongly induced by NICD, whereas the Hes6 homologs showed only a moderate increase in expression (Fig. 1I-L′). Upon Notch inhibitor treatment, expression of Hes5 and Hes6 homologs was nearly absent (Fig. 1O-R′; Fig.S2G-I′), except her2, which was expressed at reduced levels (Fig. S2J,J′). Thus, her2 may be only partly controlled by Notch.

Expression of her6 and her9 are hereafter referred to as Notch-independent. In contrast, her4 and her15 expression are strictly Notch-dependent. Given that her8a and her8.2 have only low and declining levels of expression in the TPZ on the fourth day of development, we did not include these two genes in our further analysis.

We next investigated, using fluorescent WISH, whether Notch-dependent and Notch-independent her genes are differentially expressed in the TPZ (Fig. 2). High her4 and her6 expression was restricted to the periventricular TPZ wall and co-expressed with the NSC marker sox2 (Fig. S3A-B″). At 24 hpf, her6 is broadly expressed in the presumptive thalamic complex (Scholpp et al., 2009). Then, her6 expression is progressively restricted, and at 72 hpf is largely absent from the ZLI but maintained in the rostral thalamus and prethalamus (Fig. 2A,A′,F; Scholpp and Lumsden, 2010). Absence of high levels of her6 co-expression with irx1b (caudal thalamus) or with shha (ZLI) confirms restriction of the major her6 expression to rostral thalamus and prethalamus (Fig. S3C,D). However, there is some minor co-detection of shha and her6 at the border of the ZLI. At the ventricle along the midline, her4 is strongly expressed where her6 is absent, namely in the ZLI, caudal thalamus and the pretectum (Fig. 2A). In the caudal thalamus, her4 is also expressed laterally adjacent to the ventricular wall (Fig. 2A′). her4 and her15 are mostly co-expressed in the ZLI, caudal thalamus and pretectum (Fig. 2B,B′,G). her9 is expressed only weakly in the ZLI and the caudal thalamus. her6 and her9 are expressed in rather non-overlapping territories (Fig. 2C,C′).

We also directly compared the expression of Notch-dependent her4 and her15 combined with Notch-independent her6 and her9 combined, using pairwise mixed probes, and found that their expression domains together cover the whole ventricular zone of the thalamic complex (Fig. 2D-D″). her4, her15 and, to a lesser extent, her9 are co-expressed in the ZLI and the caudal thalamus, whereas her6 is selectively expressed in prethalamus and rostral thalamus (Fig. 2E,F). her4, her6 and her9 triple fluorescent hybridization chain reaction (HCR) analyzed at cellular resolution (Fig. S4) revealed that domains of high expression of each gene at the ventricle are mostly exclusive at 72 hpf.

Distant from highly sox2-expressing ventricular NSCs, we detected more basally her4 expression at a lower level in cells which may correlate to NPCs. To validate this hypothesis, we analyzed sox2, ascl1b and neurog1 expression (Fig. S5; Movie 1) and found high sox2-expressing cells (Sox2^high) at the ventricular walls largely devoid of ascl1b and neurog1 expression, corresponding to NSCs. In addition, we found a low sox2-expressing domain (Sox2^low) in the region corresponding to low her4 expression, which also expresses ascl1b or neurog1, and thus are NPCs. Based on anti-phospho-Histone H3 (pH3) anti-Sox2 double immunofluorescence, we determined the pH3⁺ versus Sox2⁺ count ratios in WT. The Sox2^high NSCs have significantly higher pH3⁺ versus Sox2⁺ count ratios, and thus mitotic activity, compared with Sox2^low NPCs (P=0.000249; Fig. 7E and Table S7).

In the Sox2^low NPC region (Fig. S6), neurog1 expression largely overlaps with her4, but not with her6 and her9. Similarly, expression of ascl1a/b is largely exclusive from the her6-expressing regions (Fig. S6D). ascl1a/b expression overlaps widely with dla in the Sox2^lowher4 region (Fig. S6E), suggesting that Delta and Notch-dependent her genes control neurog1 and ascl1a/b in these NPC regions. To investigate the relationship of Delta signaling and her genes in NSC and NPC regions, we analyzed expression of dla combined with sox2, her4, her6 and her9 (Fig. S7). Surprisingly, dla expression appears to be higher in the NPC Sox2^low region than in the Sox2^high ventricular zone (Fig. S7A). her4 and dla expression largely overlap in the NPC region (Fig. S7B). In contrast, regions with highest levels of her6 and her9 expression are devoid of dla (Fig. S7C-D‴).

Our data reveal at least three fundamentally distinct her-expressing cell populations in the TPZ (Fig. 11A,B): (1) Sox2^high NSC type 1 at the ventricular wall with predominant her6 but absent dla expression; (2) Sox2^high NSC type 2 at the ventricular wall with high her4 and/or her15, low her9 and, in most regions, also dla expression, but absent her6 expression; (3) Sox2^low NPCs away from the ventricular wall with dla, either ascl1b or neurog1, and low-level expression of her4.

We tested genetically whether her6 and her9 might cross-repress each other. We used CRISPR/Cas9 to delete large parts of the bHLH domains of her6 and her9, and isolated mutant alleles with premature stop codons that leave the remaining aminoterminal peptides non-functional (Fig. 3A,B). Although her6 and her9 single mutant embryos have largely normal morphology, double mutants display an abnormal TPZ phenotype as well as malformations in eye development (Fig. S8A,A′), as reported for Hes1 mutant mice (Nakamura et al., 2008, 2000; Tomita et al., 1996).

We analyzed her6 and her9 single and double mutants (Fig. 3C-N) using WISH probes that detect the shortened mutant transcripts. At 48 hpf, her6 expression levels in several regions appear to be enhanced in her6 mutants, and even stronger in her6,her9 double mutants, whereas her6 expression in her9 mutants appears to be normal in most brain regions (Fig. 3C-F′). Upregulation of her6 in double mutants is even stronger at 96 hpf (Fig. 3G-H′). her9 expression levels in her6 and her9 single and double mutants at 48 hpf appear to not be strongly affected, whereas her9 expression at 96 hpf is slightly increased in double mutants (Fig. 3I-N′). We analyzed her6 and her9 mRNA levels by qPCR on genotyped 96 hpf embryonic heads. We found a significant upregulation of her6 transcripts in her6 mutants and of her9 transcripts in her9 mutants (Fig. 3O; P-values see Table S6), indicating negative autoregulation. Upregulation of her6 and her9 transcripts in her6,her9 double mutants was significantly stronger than in single mutants (Fig. 3O). We conclude that her6 and her9 exert negative cross-regulation on each other, partially redundant with negative autoregulation.

In WT larvae, Sox2⁺ NSC nuclei line the thalamic ventricular walls, including the lateral ventricle away from the midline (Fig. 4A-D″; Movie 2), while Sox2^low nuclei with lower immunoreactivity are detected in subapical positions adjacent to the cell layer that forms the ventricular wall (Fig. 4A′,A₁). her9 and her6 single mutants each have a largely normal distribution of Sox2 positive nuclei (Fig. S8F-H). In her6,her9 double mutants, Sox2^low are largely absent from the subapical region (Fig. 4B′,B₁). In addition, the Sox2^high nuclei along the lateral expansion of the ventricle were strongly reduced and missing at most lateral positions, indicating a loss of ventricular NSCs (Fig. 4A″,B″, white arrowheads; Movie 2). The number of subapical Sox2^low and Sox2^high cell nuclei were both significantly reduced in her6,her9 double mutants (counts of nuclei in Table S7, Fig. 7E′). We used an anti-pH3 antibody to detect cells in mitosis. Along the lateral ventricular wall of her6,her9 double mutants, where Sox2 expression is also missing, pH3⁺ nuclei were largely absent, and ccnd1 expression, which accumulates during G1-to-S phase transition, was reduced (Fig. S8B-E′). Compared with WT, her6,her9 double mutants in the TPZ had a similar distribution of pH3⁺ mitotic nuclei in the Sox2^high ventricular zone, and the count ratios of pH3⁺ among Sox2^high and Sox2^low cells were not significantly different from WT (Fig. 4A′,B′; Movie 2; Fig. 7E′; Table S7). Thus, despite reduction in Sox2⁺ cell number, the mitotic activity of the remaining NSCs and NPCs appears to not be significantly affected.

The strongest co-expression of sox2 and neurog1 or ascl1b is in Sox2^low-expressing NPCs that are not in direct contact with the ventricle (Fig. S5A-F; Movie 1). ascl1b is expressed in a narrow stripe in the rostral thalamus, while neurog1 is expressed in the caudal thalamus (Fig. 4C-C″; ascl1a is also expressed in prethalamus and pretectum, see also Fig. 11A-C). In her6,her9 double mutants, neurog1 and ascl1b cells are intermingled, and the combined domain is smaller and narrower (Fig. 4D-D″, arrowheads), indicating that the NPC domain is reduced. We also investigated neurog1 and ascl1a expression using WISH from 24 hpf to 96 hpf at daily intervals (Fig. S9) and found enhanced neurog1 expression at 24 hpf in her6,her9 double mutants, but in most regions expression was comparable at 48 hpf, and reduced at 72 and 96 hpf compared with WT controls. ascl1a expression is enhanced in double mutants at 24 and 48 hpf, but similar to WT controls at 72 and 96 hpf. Thus, during early neurogenesis, premature differentiation is observed in her6,her9 double mutants, but later the reduced number of NSCs may lead to reduced neurogenesis compared with WT.

In the WT thalamus, shha is expressed in the basal plate and the ZLI, where it extends along the walls of the lateral ventricle (Fig. 5A-D,I). Although at 24 hpf the shha expression in the ZLI appears to be largely normal in her6,her9 double mutants (Fig. 5A,A′,E,E′), at 72 hpf and 96 hpf the ZLI shha expression domain is progressively lost both in its lateral and dorsal domains (Fig. 5). In her6 single mutants shha expression along the ZLI lateral ventricular walls is variable and partially lost (Fig. S10, arrowheads). When we analyzed shha and sox2 expression in her6,her9 double mutants at 24 hpf, we found that shha expression was already reduced at stages when sox2 was still broadly expressed in the neuroepithelium, and a distinct proliferation zone had not yet established (Fig. S11). At this stage, her6 was still expressed in ZLI shha cells. We conclude that loss of Notch-independent her activity affects dorsal expansion of shha expression and establishment of shha in the ZLI temporally before loss of NSCs in the TPZ.

The two her4;12 and her2;15 gene clusters active in the TPZ contain nine Hes5 zebrafish homologs (Zhou et al., 2012) but no other genes (www.ensembl.org). We generated large precise deficiencies of both chromosomal regions using CRISPR/Cas9 (Fig. 6A,B; Fig. S12): the 28 kb deficiency Df(Chr23:her12,her4.1,her4.2,her4.3,her4.4,her4.5)m1364 or m1365, abbreviated as Df(her4;12), and the 30 kb deficiency Df(Chr11:her2,her15.1,her15.2)m1490, abbreviated as Df(her2;15). WISH for her4 and her15 demonstrated that there were no transcripts left in these mutants (Fig. 6C-D′).

We could not observe morphological abnormalities in Df(her4;12) or Df(her2;15) mutant larvae until 30 days post fertilization (dpf) or in fixed 96 hpf larvae (Fig. 6C′,D′). Therefore, we analyzed Df(her4;12),Df(her2;15) double mutants, which also appeared to be morphologically normal and have normal neurog1 expression (Fig. 6E-F′). The Sox2^high NSC nuclei layer covering the ventricular wall of the thalamus proper (arrow in Fig. 7A₁,B₁) and the adjacent subapical Sox2^low NPC nuclei (arrowhead in Fig. 7A₁,B₁) appear to be normal. Both in WT and double mutants, the shape of the nuclei at the ventricular wall is spherical, and the subapical Sox2^low nuclei are elongated. Sox2^high nuclei line the lateral ventricle along its full lateral extent (Fig. 7A″,B″, arrowheads). The counts of Sox2^high and Sox2^low nuclei, the distribution of pH3⁺ nuclei (Fig. 7A-B″,E′; Movie 3), the count ratio of pH3⁺ to Sox2^high and Sox2^low cells (Table S7;Fig. 7E′,E″) and the expression of ccnd1 (Fig. S8) appear to not be significantly different in Df(her4;12),Df(her2;15) compared with WT, suggesting normal proliferation in the TPZ. Thus, NSCs and NPCs appear to develop normally in the TPZ devoid of Notch-dependent her genes.

We also analyzed her gene expression in her6,her9 and in Df(her4;12),Df(her2;15) double mutants. her6 and her9 expression is largely unaffected in Df(her4;12),Df(her2;15) double mutants, in line with their Notch-independent regulation. her4 and her15 expression appears to be decreased in the NPC area of her6 and her9 double mutants (Fig. S13A′,B′). At the ventricular wall her4 and her15 expression do not invade into the her6 domains in the rostral thalamus and prethalamus (Fig. S13A,B), suggesting that mechanisms other than repression by her6 limit her4 and her15 expression in these domains.

We analyzed expression of neurog1, ascl1a/b and olig2 in NPCs. olig2 is expressed in a stripe of cells parallel to the ventricular wall, adjacent to neurog1 (Fig. 7C-C″, arrowheads). Expression domains of all four genes appear to be normal in Df(her4;12),Df(her2;15) double mutants (Fig. 7D-D″). Although more subtle changes cannot be excluded, the NSC and the NPC compartments of the TPZ in the absence of Notch-dependent her gene activity appear to be very similar to WT.

We next investigated potential cross-regulation between Notch-dependent and -independent her genes by heat shock-driven overexpression of Her4 or Her6 using Tg(hsp:her4-FLAG) and Tg(hsp:her6-FLAG) transgenic lines. When Tg(hsp:her4-FLAG) larvae were analyzed using WISH 2.5 h post heat shock, her4 was confirmed to be overexpressed (Fig. 8A,A′), and we detected downregulation of her15 expression (Fig. 8B,B′). Her4 overexpression, however, did not affect the expression levels of her6 and her9 (Fig. 8C-D′). In contrast, overexpression of Her6 strongly downregulated her9 and both her4 and her15 (Fig. 8E-H′). This suggests that Notch-independent her genes negatively repress Notch-dependent her genes, while Notch-dependent her genes repress other Notch-dependent her genes, but do not regulate Notch-independent her genes.

We tested whether Her4 or Her6 might differentially affect proneural gene expression. Upon Her4 heat shock-driven overexpression at 70 hpf and fixation 2.5 h after heat shock, neurog1 expression appeared to be largely unchanged, whereas ascl1b was reduced in 9 out of 14 larvae (Fig. 9A-B′). Late progenitors, evaluated by neurod1 and neurod6b expression, were unchanged 2 h after her4 heat shock (Fig. 9C-D′). sox2 expression was slightly reduced in 9 out of 15 her4 heat-shocked larvae (Fig. 9E-E′). In contrast, overexpression of Her6 strongly downregulated neurog1 and ascl1b expression, and to a lesser extent the late progenitor markers neurod1 and neurod6b (Fig. 9F-I′). In addition, sox2 expression appeared to be slightly reduced in 11 out of 15 Her6 heat-shock larvae (Fig. 9J-J′). We also tested whether overexpression of Her4 or Her6 might have a prolonged effect on proneural gene expression late after heat shock. We analyzed larvae 6 h after the beginning of the 30 min heat shock and found neurog1 and ascl1b expression at normal levels (Fig. S14A-D′), despite their downregulation 2.5 h after heat shock. This suggests that the NSC/NPC regulatory network is robust against perturbations in Her6 expression level, and rapidly re-establishes normal activity. Together, our findings reveal a strong regulatory impact of Her6 on NPCs and NSCs, and indicate that Notch-independent Her6 might be a much more potent inhibitor of neurogenesis compared with Notch-dependent Her4.

We asked whether Notch-dependent and -independent her genes have partially redundant functions. We combined the two deficiencies, eliminating all nine Hes5 homologs together with mutant alleles of her6 and her9, and named embryos homozygous mutant for all 11 her genes ‘her undecimal mutants’ [herUDM; for Df(her4;12),Df(her2;15),her6^−/−,her9^−/−]. The brains of herUDM embryos appeared to be severely malformed, and embryos did not survive beyond 120 hpf. When analyzed at 56 hpf (Fig. 10A-B′), herUDM embryos had the most severe morphological defects in neural tissues that undergo significant expansion by proliferation on the second day of development, such as the cerebellum, the ventral part of the retina and the rhombic lip. Early forming neural tissues, however, appeared to be mostly normal, which might be because of compensation by other her genes expressed during early stages of development, including her3, her8a and her8.2 (Onichtchouk et al., 2010; Webb et al., 2011).

We analyzed Sox2 expression in herUDM embryos and observed a severe but variable phenotype, ranging from strong reduction of both the Sox2^high NSC and Sox2^low NPC compartments (Fig. 10D-D″) to extreme anatomical malformations rendering analysis of the TPZ difficult (not shown). In 72 hpf herUDM embryos, the layer of Sox2^high nuclei along the walls of the lateral ventricle was completely missing (Fig. 10C″,D″). The Sox2^high-expressing nuclei along the medial ventricular wall appeared disorganized, and the ventricular layer of Sox2^high nuclei was interrupted (Fig. 10C-D″; Movie 4). Some Sox2⁺ nuclei were displaced from the ventricular wall, suggesting that the epithelial integrity of the ventricle may be impaired (Fig. 10C₁,D₁, arrows). Similar to her6,her9 double mutants, the Sox2^low NPC nuclei were nearly absent (Fig. 10C₁,D₁, arrowheads). We still detected pH3⁺ nuclei at the ventricular walls of herUDM embryos (Fig. 10D′; Movie 4), suggesting that NSC proliferation is not completely eliminated. The phenotype of herUDM is much more severe than of her6,her9 double mutants with respect to NSC maintenance.

A single her2;her15 cluster WT allele in Df(her4;12) homozygous, Df(her2;15) heterozygous, her6^−/−,her9^−/− mutant larvae partially rescued the herUDM phenotype (Fig. 10E-E″; Movie 5, right). The integrity of the ventricular Sox2^high nuclear layer was restored and resembled the phenotype in her6,her9 double mutants (Fig. 10E₁, arrow, E″; Movie 5). As seen in her6,her9 double mutants, the subapical Sox2^low nuclei were still absent (Fig. 10E₁ arrowhead; Movie 5). In contrast, a single her6 WT allele in Df(her4;12),Df(her2;15),her6^+/−,her9^−/− mutant larvae fully restored the Sox2^high nuclear layer along the lateral ventricular wall (Fig. 10F″ arrowhead; Movie 5, left). The ventricular wall appeared to be intact (Fig. 10F₁, arrow), and the subapical Sox2^low NPC nuclei were present and of normal elongated shape (Fig. 10F′,F₁, arrowhead). Individual effects of her9 or her4,12 were not tested, because they are tightly linked on the her9 ^m1505,Df(her4;12) chromosome 23 we used. When we analyzed herUDM rescue in Df(her4;12) heterozygous, Df(her2;15) homozygous, her6^−/−,her9^+/− by a single chromosome with WT her9 and WT her4;12, we observed a rescue of the dorsolateral Sox2^high nuclei, but only a partial rescue of the Sox2^low NPCs in the thalamus (Fig. S15). Thus, a single her6 allele rescues the herUDM phenotype more efficiently than a single WT chromosome with her9 and the her4;12 cluster, or the her2;15 cluster.

We determined whether combined loss of Notch-dependent her genes and of her9 activity may cause her6 upregulation in anatomical regions in which it may be cross-repressed in WT. We found that in Df(her4;12),Df(her2;15),her9 combined mutants her6 is indeed expressed more broadly along the ventricular wall of the TPZ (Fig. S16). The rescue activity of her6 in herUDM embryos may thus be caused by compensatory upregulation of her6 expression in areas that in WT do not express high her6. In search for the potential origin of these her6 cells, we analyzed the TPZ at higher resolution, and detected few her6 cells intermingled in predominantly her4 and her9 thalamic regions in WT (Fig. S4D′). A small number of dispersed her6 expressing cells remains in the ZLI and caudal thalamus even at 48 hpf, when the early broad her6 expression has recessed from her4 and her9 expression domains (Fig. S4F). Expanded her6 expression in herUDM may thus reflect persistence of early her6 expression, or expansion of small her6 clones. We hypothesize that much of the compensatory expansion of her6 expression in herUDM happens early in development, given that inhibition of Notch signaling from 64-72 hpf (Fig. 1M) did not cause an expansion of her6 expression.

In summary, the severe loss of NSCs and NPCs in herUDM revealed that, first, the 11 her genes analyzed provide most, if not all, Her activity in the TPZ at 72-96 hpf. Second, Notch-dependent and -independent her genes appear to act partially redundant in NSC maintenance. Third, the Notch-independent her6 may substitute for other her activity in NSC and NPC maintenance and expansion in the TPZ.

Delta/Notch signaling and HES/Her transcription factors control NSC maintenance and regulate NPC progression in neurogenesis (Kageyama et al., 2007). Their activities differ during primary neurogenesis in the early neuroepithelium of anamniotes, during embryonic brain growth in neural proliferation zones, and in adult neural stem cell niches (Alunni et al., 2020; Kriegstein and Alvarez-Buylla, 2009). Here, we analyzed Her activities in the zebrafish TPZ.

We find expression of the Hes1 homologs her6 and her9 to be largely independent of Notch signaling, in line with previous reports on earlier stages (Bae et al., 2005; Hans et al., 2004; Latimer et al., 2005). In contrast, Hes1 in mice is predominantly regulated by Notch signaling (Hsieh et al., 1997; Jarriault et al., 1995, 1998; Nishimura et al., 1998), but also by Notch-independent input (Kageyama et al., 2007; Ohtsuka et al., 1999). For the zebrafish Hes5 homologs her4 (Takke et al., 1999; Zhou et al., 2012), her12 (Gajewski et al., 2006), her2 and her15 (Cheng et al., 2015; Shankaran et al., 2007), we find expression to depend on Notch signaling. Similarly, Hes5 expression in mice is Notch-dependent (Ohtsuka et al., 1999). In the TPZ, expression of the zebrafish Hes6 homolog her8a depends on Notch signaling, whereas in the neural plate Notch-independent expression has been reported (Webb et al., 2011). her8.2 appears to be regulated in TPZs by both Notch-dependent and -independent mechanisms.

Differential expression of Notch-dependent and -independent her genes reveals distinct NSC populations in the TPZ (Fig. 11A). We designate Sox2^high NSCs expressing high her6 as NSC type 1, and those expressing high her4 and her15 as NSC type 2. Both NSC type 1 and 2 are proliferative, indicating that quiescent versus active NSC populations may only diversify later in development (Than-Trong et al., 2020). The different NSC types emerge during maturation of the thalamic complex, as early her6 expression comprises the entire presumptive thalamic complex, but becomes gradually restricted (Scholpp et al., 2009). The complementary expression of Hes1 and Hes5 homologs in zebrafish is consistent with previous findings on Hes1 and Hes5 expression in mice (Hatakeyama et al., 2004).

The role of her4 during neural plate stages was reported to be similar to Notch-dependent Drosophila E(spl) genes, as both mediate repression of neurogenesis in the neuroepithelium by lateral inhibition (Nakao and Campos-Ortega, 1996; Takke et al., 1999). We found that, in the larval TPZ, her4 and her15 continue to depend on Notch signaling and are largely co-expressed, but in domains distinct from her6 expression. A similar dichotomy was observed for the expression of Notch-dependent E(spl) genes versus Notch-independent hairy in the fruit fly (Fisher and Caudy, 1998; Nakao and Campos-Ortega, 1996). In the zebrafish neural plate, prepatterning genes mark the inter-proneural domains in a Notch-independent manner (Stigloher et al., 2008). For example, her3 is expressed in two elongated stripes along the anteroposterior axis, flanked by neurog1⁺ progenitor pools (Bae et al., 2000), and her5 inhibits neurogenesis in the intervening zone at the prospective midbrain-hindbrain boundary (Geling et al., 2004; Ninkovic et al., 2005). These homologs of Drosophila hairy appear to act by restricting the proneural domains in the neuroectoderm independent of Notch signaling, similar to Drosophila hairy (Fisher and Caudy, 1998; Geling et al., 2004). Similarly, her6 and her9 may execute prepattern information to maintain NSCs.

Domains of NSC type 1 expressing Notch-independent her6 and NSC type 2 expressing her4, her15 and her9 show a strict spatial organization in the TPZ. A central stripe of NSC type 2 cells coincides with the ZLI. Anterior and posterior directly adjacent to the ZLI are narrow stripes of NSC type 1 cells in the prethalamus and rostral thalamus. Further posterior, there is again a stripe of NSC type2 cells in the caudal thalamus. The NSC type 1 her6 stripes correlate with the her6-expressing domain reported to generate ascl1a gabaergic progenitors, which has been described both for zebrafish (Scholpp et al., 2009; Scholpp and Lumsden, 2010) and mice (Virolainen et al., 2012). In contrast, the NSC type 2 caudal thalamic domain correlates with the domain generating Ngn2 glutamatergic progenitors in mice (Virolainen et al., 2012). Shh signaling organizes these domains of the thalamic complex (Scholpp et al., 2006). In her6 mutants, we find a severe reduction, and in her6,her9 double mutants complete absence of shha expression along the ZLI lateral ventricular walls, while the her9 mutant ZLI appears normal. Despite her6 and her9 being expressed in distinct cell populations, they thus can partially compensate for each other in the ZLI. her6 and her9 may execute patterning information to maintain ZLI signaling activity and NSCs. Our analysis of shha and sox2 expression during early development of her6,her9 double mutants reveals that her6 and her9 activity are required for dorsal expansion of shha expression and establishment of shha in the ZLI, which precede later reduction of sox2 expression in the mutant TPZ. her6 morphant embryos have been shown to lose shha expression in the mid-diencephalic organizer, which was explained by premature differentiation of organizer cells accompanied by the loss of the boundary-characteristic shha expression (Scholpp et al., 2009). A similar Shh phenotype was observed in Hes1,Hes5 double mutant mice (Baek et al., 2006). In mice, Hes1 is expressed at high levels in neural boundaries and organizing centers (Baek et al., 2006), which resulted in a model of two distinct Hes1 functions. In so-called boundaries, high levels of Hes1 constitutively represses neurogenesis, whereas in compartments its expression oscillates and regulates progression of neurogenesis (Kageyama et al., 2007). Along this concept, her6 may execute boundary activities. How both types, NSC type 2 her4- and shha-expressing ZLI organizer cells themselves, and adjacent her6-expressing NSC type 1, work together to maintain the organizer should help to understand the complexity of the ZLI.

We can only speculate on upstream regulation of Notch-independent her6, but note that her6 co-expression in high level Sox2-expressing cells suggests a regulatory link to transcriptional networks maintaining NSCs. Further, although her6 is not expressed in shha ZLI organizer cells itself, based on its position it may be subject to Shh and Wnt signaling from the ZLI (Martinez-Ferre et al., 2013). Thus, we hypothesize that her6 may be controlled by combined NSC regulatory networks and ZLI signaling.

Surprisingly, in Df(her4;12),Df(her2;15) double mutants, Sox2^high and Sox2^low NPCs, as judged by expression of ascl1a/b, neurog1 and olig2 in the TPZ, appear to be largely normal. Thus, in the TPZ, these nine Notch-dependent her genes together appear to be expendable for the initiation of proneural gene expression in NPCs. Our her4 overexpression experiments support this interpretation: in her4 heat-shocked embryos, expression of neurog1 and neurod1/6b are largely normal, and of ascl1b in most embryos only slightly reduced. Together, these findings are surprising, because a strong effect of Notch-dependent her genes on proneural gene expression would be expected (Imayoshi et al., 2013; Takke et al., 1999).

The lack of a phenotype suggests that Notch-dependent her gene activity either may have only a minor contribution to NSC maintenance or be substituted by Notch-independent her gene activity. The rather normal development is also in stark contrast to the severe phenotypes observed previously in studies using other methods for inhibition of canonical Notch signaling. mib1 mutants are a paradigm for loss of Delta signaling and show a severe phenotype (Itoh et al., 2003). However, MIB1-mediated ubiquitylation has been shown to target also other cellular functions, including ciliogenesis (Wang et al., 2016). LY-411575 inhibits gamma-secretase, which has more than 80 targets other than Notch (Haapasalo and Kovacs, 2011). The severe mib1 and LY-411575 phenotypes may thus be enhanced by effects on mechanisms other than Notch signaling. It also needs to be considered that our combined Notch-dependent her mutants target the larval TPZ, but do not affect other Notch-dependent her genes broadly expressed in earlier development, like the Hes6 homologs her8a and her8.2. Finally, inhibition of Notch signaling inhibits all Notch targets, whereas our study selectively investigates those effects mediated by her genes. Thus, our combined analysis of Notch-dependent her genes is likely to be valid for larval TPZ, but does not address earlier embryonic stages or other Notch signaling activities.

In contrast, loss of Notch-independent her6 and her9 results in early premature differentiation and smaller and disorganized domains of ascl1a/b, neurog1 and olig2 in the 72-96 hpf TPZ. Overexpression of her6 eliminates neurog1 and ascl1b expression, and causes a reduction of neurod1/d6b expression, revealing an influence on NPCs entering neuronal differentiation. Similar to our observations, Hes1 single and Hes1,Hes5 double mutants have much stronger neurogenesis phenotypes than Hes5 mutants (Hatakeyama et al., 2004). Our findings so far suggest a strong contribution of Notch-independent her gene activity to control of NPC development, either at the level of generating new NPCs from NSCs or NPC differentiation. As her6 is not expressed in NPCs, we favor the hypothesis that Notch-independent her genes mostly affect generation of new NPCs. This could be by maintaining or increasing the NSC pool, or by controlling NSC-to-NPC transition.

In contrast to combined deficiencies of Notch-dependent her genes, her6,her9 double mutants exhibit a strong NSC phenotype with absent Sox2^high NSC nuclei in the lateral extensions of the ventricle, and a significant overall reduction in the total NSC count in the TPZ (Fig. 11D-F). Even more severely, combined inactivation of Notch-dependent and -independent her genes in herUDMs causes loss of most NSCs and essentially all NPCs in the TPZ, and loss of ventricular wall integrity (Fig. 11G). The herUDM phenotype in severity appears similar to the Hes1,Hes5 double mutant phenotype in mice with premature differentiation and stem cell depletion (Baek et al., 2006; Hatakeyama et al., 2004; Ohtsuka et al., 1999). The severe her6,her9 double mutant phenotype reveals a prominent role in NSC maintenance or proliferation. Our pH3 analysis reveals that loss of Notch-dependent or -independent her genes does not affect NSC proliferation, supporting a prevalent role of Notch-independent her genes in NSC maintenance. The efficient suppression of proneural genes, as observed after her6 overexpression, may be crucial for NSC maintenance.

her6 and her9 show negative feedback auto-regulation and cross-regulation; however, the latter was largely masked in the presence of functional autoregulation, such that the expression of both genes is most strongly upregulated in their double mutants (Fig. 11H). Overexpression of her6 repressed her9, her4 and her15, revealing that Notch-independent her6 repression in a sense is dominant over these her genes. Due to the cluster deletions, we could not investigate autoregulation of Notch-dependent her genes in mutants. her4 overexpression downregulates her15, but did not affect her6 and her9 expression, confirming that these two genes are indeed independent of Notch signaling. Accordingly, her6 and her9 expression is largely unaffected in Df(her4;12),Df(her2;15) double mutants. Redundancy of her genes in zebrafish is a recurring motif that has been shown for her1, her7 and her13.2 in somitogenesis (Henry et al., 2002; Oates and Ho, 2002; Sieger et al., 2006), as well as her5 and her11 (him) at the mid-hindbrain boundary (Ninkovic et al., 2005), but also for Hes1 and Hes3 in mice (Hirata et al., 2001).

When we analyzed individual her genes with respect to their rescue potential in herUDM mutants, a single WT allele of her6 rescued depletion of the activity of 21 other her gene alleles in herUDM embryos. This special activity of her6 may be explained by two mechanisms: First, we observed that in Df(her4;12),Df(her2;15),her9 mutants, her6 expression in the TPZ is not restricted to the rostral thalamus and prethalamus, which may be interpreted as persistent expression of her6 from the broad early thalamic complex her6 domain (Scholpp et al., 2009). Second, although in WT high level expression domains of her6 versus her4/her9-expressing subregions of the TPZ are largely exclusive, we observed few high and some low her6-expressing cells also intermingled in her4 territories. Thus, in the her6-rescued herUDM, her6⁺ clones may expand. The prominent role of her6 may thus be rooted in its early broad expression in the TPZ.

In summary, the expanded number of Hes1/Hes5 homologous genes in zebrafish goes along with a clear separation of Notch-dependency for different her genes. Our genetic analysis demonstrates that the her gene network is largely redundant within the Notch-dependent and -independent her groups, and that the latter may better substitute for the loss of Notch-dependent genes than vice versa. The observed robustness of the network is supported by several regulatory feedback loops and cross-regulation, and makes NSC maintenance robust against severe perturbations of her activity. In the TPZ, Notch-independent her genes predominantly regulate NSC maintenance and contribute to ZLI formation and shha expression. With respect to neurogenesis, we hypothesize that her6 may control lineage transitions, e.g. from NSC to NPC, whereas her4 may rather modulate progression of neurogenesis.

ABTL strain zebrafish (Danio rerio) were kept and bred under standard conditions (Westerfield, 2000). Experiments were performed in accordance with the German Animal Welfare Guidelines. All animal experiments were approved by the ethics committee for animal experiments at the state authority Regierungspräsidium Freiburg (permits 35-9185/G-16/123, 35-9185.81/G-19/19 and 35-9185.81/G-19/54).

Zebrafish embryos were obtained through natural breeding and embryos were raised in E3 medium (5 mM NaCl, 0.17 mM KCl, 0.33 mM CaCl₂, 0.33 mM MgSO₄, supplemented with 2 ppm Methylene Blue). We added 0.2 mM N-Phenylthiourea (PTU, Sigma-Aldrich) to the E3 medium to prevent pigmentation. Embryos were staged as previously described (Kimmel et al., 1995).

Embryos were incubated in E3 (containing 0.2 mM PTU) supplemented with 1× tricaine methanesulfonate (MS-222) and fixed in 4% paraformaldehyde (PFA) for 4 h at room temperature (RT) or overnight at 4°C. Embryos were washed several times with PBST (137 mM NaCl, 2.7 mM KCl, 10 mM Na₂HPO₄, 1.8 mM KH₂PO₄, 1% Tween-20) and dehydrated stepwise with 25%, 50% and 75% MeOH in PBST, and stored at −20°C in 100% MeOH.

DNA was isolated from embryos or tail biopsies of fixed larvae. Samples were heated 5 min to 95°C in 60 µl TE buffer, after addition of 3 µl Proteinase K (20 mg/ml), and digested for at least 4 h at 55°C. After heat-inactivation of Proteinase K for 10 min at 95°C, 3 µl of the DNA sample was used in 30 µl PCR volume. The primer concentration was 0.3 µM unless stated otherwise. PCR programs are supplied in Table S4. For genotyping of embryos used in the qPCR, an alkaline lysis was performed. Tails were cut and transferred to 40 µl 50 mM NaOH and incubated for 45 min at 95°C. The lysis was stopped by addition of 4 µl 1 M Tris-HCl (pH 7.5).

New transgenic lines were generated using the Tol2 kit vector system (Kwan et al., 2007). For heat shock-driven her4 and her6 overexpression, hsp:her4.1-FLAG and hsp:her6-FLAG were cloned by combining a 1.5 kb hsp70 l heat shock promoter fragment (Chi-Bin Chien, Genbank AF158020) with the her4.1 (ENSDART00000079274.4) or her6 (ENSDART00000023613.9) cDNA sequence (from EZRC: kit numbers KIT001605 and KIT000178; www.ezrc.kit.edu) with a FLAG-tag (DYKDDDDK or 5′-GACTACAAAGACGATGACGACAAG-3′) in a vector containing Tol2 sites and a green heart marker (clmc2:EGFP). The her4.1 gene was chosen for overexpression of her4 activity because the her4.1 open reading frame (ORF) represents the consensus amino acid sequence of all five her4 cluster genes (alignment of her4.1 NP_001096598, her4.2 AAH49296, her4.3 NP_001154880, her4.4, NP_001121862, and her4.5 NP_571165). The sequence of the Her4.1 protein we cloned diverges from NP_001096598 only at amino acid 49 valine, which is Her4 consensus and shared with Her4.2 and 4.5. Thus, we assume that the Her4.1 ORF in our transgene has full Her4 activity. The following alleles were used in this study: Tg(hsp70l:her4.1-FLAG)m1541, Tg(hsp70l:her6-FLAG)m1492.

Cas9 mRNA was transcribed from the hspCas9 plasmid (Ansai and Kinoshita, 2014) and Cas9-nanos from the Cas9-nanos-3′-UTR plasmid (Moreno-Mateos et al., 2015) using the SP6 mMessage mMachine kit (Thermo Fisher Scientific), and a polyA tail was added (PolyA tailing kit from Thermo Fisher Scientific). sgRNAs were designed with CRISPRscan (Moreno-Mateos et al., 2015) and ordered as oligonucleotides to generate DNA templates (Gagnon et al., 2014). Annealing of the constant oligonucleotide with the variable oligonucleotide was carried out by cooling from 95°C down to 25°C in 1 h in a PCR machine. Annealed oligonucleotides were filled with T4 DNA Polymerase (New England Biolabs) for 20 min at 11°C. sgRNAs were transcribed for 3-4 h at 37°C using the MEGAscript T7 kit according to protocol (Thermo Fisher Scientific) and dissolved in 15 µl H₂O.

Two sgRNAs (transcribed from p3-Oligo04 and p4-Oligo05) were used to generate the her6 knockout allele m1358. The intragenic deletion generates a frameshift that truncates the protein after amino acid 36 before the bHLH domain. Lower case letters are deleted in the mutant allele m1358: 5′TATAGTGCTTCAAGAagtgattagacagattgactcattttgtttttattttgcattttagtcttctaaacccattatggagaaaagaagaagagcgagaatcaacgaaagcttgggtcagctgaaaacgttaatcttggatgctctgaaaaaagatgtaagtaccgaaagtccgactgagtctttcagttaggctatatctggctggaatattaatgcatttctttatatagccgcatatctaaatcagtctctttcttcactctcagagctccagacactctaaacttgagaaagccgacATCCTGGAGATGACA3′.

Two sgRNAs (transcribed from p186-Oligo26 and p187-Oligo27) were used to generate the her9 knockout allele m1368. The intragenic deletion generates a frameshift that truncates the protein after amino acid 18 before the bHLH domain. Lower case letters are deleted in the mutant allele m1368: 5′ATTGCTGGTGCCCCTgccagtggatctcatactcctgacaagccaaagaatgccagcgagcatagaaagtcttcaaagccaatcatggaaaagcgccgcagagcgagaatcaacgagagccttgggcagctgaagactctcattcttgatgctcttaaaaaagatagctccagacactctaaattggagaaagctgatattctggagatgacagtcaagcacctgcgcaatttacaacgtgttcagatgagcgcagccttgtcagctgacacaaacgtcctcagcaagtaccgcgcaggattcaacgagtgcatgaacgaggtgactcgatttctctctacctgcgagggagtgaatacagaggtcagatcgcgacttcttaaccacctgtCCGGTTGTATGGGAC3′.

Two sgRNAs (transcribed from p211-Oligo38 and p219-Oligo43) were used to generate the 30 kb deficiency Df(Chr11:her2,her15.1,her15.2), which we abbreviate to Df(her2;15) (Fig. S12). We injected 1 nl of the injection mix (6.5 µl total volume), which consisted of 1 µl of each sgRNA (transcribed from Oligo p211 and Oligo p43) plus 4.5 µl Cas9-nanos mRNA (500-800 ng/µl), into one-cell-stage zebrafish embryos. Single embryos at 24 hpf were genotyped using PCR with two primers p213+p222 (0.3 µM each) to test for knockout events (Table S2). The PCR was performed using MyTaq polymerase (Bioline; PCR program in Table S4). If embryos showed the desired deletions, siblings were raised and crossed to ABTL fish. F1 embryos carrying deletions were identified by PCR and siblings were raised to adulthood. Adult F1 fish were fin clipped and genotyped with primers p213+p222 (0.3 µM each)+p220 (0.08 µM) (Tables S2 and S4). Stable heterozygous F2 lines were established and appeared to grow normally.

Two sgRNAs (transcribed from p39-Oligo12 and p47-Oligo16) were used to generate the 28 kb deficiency Df(Chr23:her12,her4.1,her4.2,her4.3,her4.4,her4.5), which we abbreviate to Df(her4;12) (m1364 and m1365 are shown in Fig. S12). We injected 1 nl of the injection mix (8.3 µl total volume), which consisted of 1 µl of each sgRNA (transcribed from Oligo12 and Oligo16) plus 6 µl hspCas9 mRNA (600-1200 ng/µl), into one-cell-stage zebrafish embryos. Then 24 hpf single embryos were genotyped by PCR with p42+p51 (0.3 µM each) to test for knockout events (Table S2). The PCR was performed using MyTaq polymerase (Bioline; PCR program in Table S4). If embryos showed deletions, siblings were raised and crossed to ABTL fish. F1 embryos were identified by PCR and siblings were raised to adulthood. Adult F1 fish were fin clipped and genotyped with primers p42+p43+p51 (0.3 µM each) (Tables S2 and S4). Stable heterozygous F2 lines were established and appear to grow normally.

The her9 gene (chromosome 23: 23,399,520-23,401,305) is only 1.9 Mb away from her4.5 (chromosome 23: 21,469,203-21,471,022), which is its closest neighbor in the her4 gene cluster. To obtain the combined her6,her9, her4;12 cluster and her2;15 cluster mutant strain (herUDM), Df(her4;12) m1364 heterozygous embryos were injected with sgRNAs (transcribed from p186-Oligo26 and p187-Oligo27) to target her9 on the Df(her4;12) chromosome. For this mutant, 1 nl of the injection mix (5 µl total volume), which consisted of 1 µl of each sgRNA and 3 µl Cas9-nanos mRNA (500-800 ng/µl), was injected into one-cell-stage zebrafish embryos. The establishment of stable F2 lines was similar to that described for the large deficiency mutants. The PCRs were performed with primer p197 and p198 and the program can be found in Table S4. The allele was named her9 m1505. Lower case letters are deleted in the mutant: 5′TGCTGGTGCCCCTGCcagtggatctcatactcctgacaagccaaagaatgccagcgagcatagaaaggtaaaaatcacttataatgcgattgcatattgttttagaagaatgacagctgcatcagttattctaaaaaaagagagaatgtcttttaattacgtaaattaattaattttccatttcatctgcagtcttcaaagccaatcatggaaaagcgccgcagagcgagaatcaacgagagccttgggcagctgaagactctcattcttgatgctcttaaaaaagatgtaagttttatctacaactcttgtcatgcttcagtcacccgacgttagtaattctgaaacagtttctaaccgaattctgctcattcacagagctccagacactctaaattggagaaagctgatattctggagatgacagtcaagcacctgcgcaatttacaacgtgttcagatgagcggtaagttgcaagtcagattcctcaagatgataaacttttaacgtgcttttaaaaacgcaatttaatttcctgaatacacaatctaaatacgttttatctttgtttaccagcagccttgtcagctgacacaaacgtcctcagcaagtaccgcgcaggattcaacgagtgcatgaacgaggtgactcgatttctctctacctgcgagggagtgaatacagaggtcagatcgcgacttcttaaccacctGTCCGGTTGTATGGG3′. When breeding the her9 m1505, Df(her4;12) chromosome, we also checked for unwanted crossing over events by PCR for the Df(her4;12) allele.

For pharmacological inhibition of Notch signaling by LY-411575 (Rothenaigner et al., 2011), embryos were incubated in E3 supplemented with 0.2 mM PTU until 64 hpf. Then embryos were incubated in 1× E3 supplemented with 0.2 mM PTU, 2% DMSO, 10 µM LY-411575 for 8 h. Control embryos were treated in 1× E3 supplemented with 0.2 mM PTU and 2% DMSO for 8 h in the same six-well plate as the LY-treated embryos. Embryos were checked for vital functions and fixed at 72 hpf in 4% PFA.

For the overactivation of the Notch signaling pathway, Tg(hsp:Gal4;UAS:RFP) kca4Tg were crossed with Tg(UAS:NICD)kca3Tg (Scheer and Campos-Ortega, 1999).This allowed heat shock-controlled overexpression of NICD. At 64 hpf, embryos were heat shocked for 30 min at 40°C in pre-heated E3 (with 0.2 mM PTU). Heat-shocked embryos were then sorted for the heat shock-induced RFP marker. Non-fluorescent embryos were used as negative controls. Embryos were tested for the UAS:NICD element by PCR (Tables S2 and S4).

Fixed and stored embryos were rehydrated with 75%, 50%, 25% MeOH in PBST and washed several times with PBST. Embryos were digested with Proteinase K (10 µg/ml) in PBST for 45 min at RT. Embryos were re-fixed with 4% PFA for 20 min at RT and then washed several times with PBST. Larvae were blocked in blocking solution [1% bovine serum albumin (BSA), 5% goat serum, 1% DMSO in PBST] for 1 h and incubated with primary antibodies diluted in blocking solution. The anti-Sox2 antibody (20G5, Abcam, ab171380, LOT GR3253929-5; cross-reacts with both zebrafish Sox2 and Sox3 proteins; Westphal et al., 2022) and the anti-phospho-Histone H3 (Ser10) antibody (Merck, 06-570, LOT 3319395) were diluted 1:400. Embryos were incubated in primary antibodies overnight at 4°C and then washed several times for 30 min each with PBST supplemented with 1% DMSO. Embryos were incubated with the secondary antibodies 1:1000 in PBST supplemented with 1% DMSO and 1% blocking reagent (Roche, 1096176) overnight at 4°C. Secondary antibodies were Alexa 555 goat anti-mouse IgG (Thermo Fisher Scientific, A-11001) and Alexa 488 goat anti-rabbit IgG (Thermo Fisher Scientific, A-11070). Embryos were washed several times in PBST, transferred to 80% glycerol in PBST and stored at 4°C protected from light. The embryos were recorded soon after staining.

Whole-mount in situ hybridization by HCR was performed as described in the HCR v3.0 protocol for zebrafish embryo and larvae (Choi et al., 2018) and by Molecular Instruments. The following changes to the protocol were made. In preparation stage 11, 72 hpf larvae were treated with proteinase K (30 µg/ml) for 30 min. In detection stage 1, 15 larvae were used in 2 ml tubes. After amplification stage 6, embryos were washed twice in PBST for 5 min and stored in 80% glycerol in PBS. Probes for neurog1 (ENSDART00000078563.5), asc1a (ENSDART00000056005.5), ascl1b (ENSDART00000183550.1), sox2 (ENSDART00000104493.5) and her6 (ENSDART00000023613.9) were ordered and designed by Molecular Instruments.

We designed probe sets for her4 (ENSDART00000079274.4), her9 (ENSDART00000078936.4), her15 (ENSDART00000055707.6) and olig2 (ENSDART00000060006.5) (Table S3). For example, the olig2 probe set consists of ten probe pairs which are evenly distributed along the mRNA sequence (Table S3). One probe pair consists of two gene-specific 25 bp sequences each with a GC content of 37-85% and a Tm of 55-77°C. The two sequences were separated by a gap of two nucleotides. The B1 specific spacer and initiator sequences were added to the reverse complement of the two gene-specific sequences. The complete probe set was ordered from Sigma-Aldrich. Equal volumes of all 20 single probes were mixed and used as probe mixture containing 5 µM of each probe in detection stage 3.

Specific her probes were designed to avoid cross-hybridization of conserved domains. For the generation of her6, her9, her4.1-her4.5, her15.1-15.2, her2 and her12 whole-mount in situ hybridization probes, sequences spanning the last exon and/or 3′ UTR of the respective genes were amplified by PCR (all primers for probe generation are given in Table S2). The PCRs were performed on WT genomic DNA with PfuUltra II (Agilent). The 50 µl PCR mix contained 6-10 µl dNTP mix (2.5 mM each), 5 µl 10× Pfu buffer, 0.2 mM of each primer, 2 µl DNA and 1 µl PfuUltra II Polymerase. Before cloning, A-overhangs were added by incubating 20 µl of the purified PCR product with 9.7 µl MyTaq buffer, 0.2 µl MyTaq polymerase and 0.1 µl (25 µM) dATPs for 20 min at 72°C. After amplification, the probes were cloned into a TOPO vector according to the manufacturer's protocol (TOPO TA cloning kit, Thermo Fisher Scientific).

For the her4 probe, which recognizes her4.1 (ENSDART00000079274.4), her4.2 (ENSDART00000137573.2), her4.3 (ENSDART00000104209.4), her4.4 ENSDART00000079265.6) and her4.5 (ENSDART00000104206.4), a template was generated with primer p159F and p160R (annealing 57°C, 40 s). For the her15 probe, which recognizes her15.1 (ENSDART00000055707.6) and her15.2 (ENSDART00000055706.6), a template was cloned using primers p157F and p158F (annealing 56°C, 40 s). The her2 (ENSDART00000055709.5) probe template was amplified with p154F and p155R (annealing 56°C, 40 s) and the her12 (ENSDART00000044080.7) probe template was amplified with p152F and p153R (annealing 56°C, 40 s). The her6 (ENSDART00000023613.9) probe template was amplified using p147F and p148R (annealing 57°C, 55 s) and the her9 (ENSDART00000078936.4) probe template was amplified with p149F and p150R (annealing 56°C 40 s). All primers are given in Table S2.

The her8a (ENSDART00000123395.4) probe template was amplified as published in Webb et al. (2011), with p322F and p327R (annealing 56°C, 1 min) and the her8.2 (ENSDART00000101578.4) probe template was amplified with p260F and p259R (annealing 56°C, 1 min). For her8a and her8.2 cDNA was used as a template in the PCR.

sox2 (ENSDART00000104493.5), neurod1 (ENSDART00000011837.6), neurog1 (ENSDART00000078563.5), shha (ENSDART00000149395.3), neurod6b (ENSDART00000185805.1), irx1b (ENSDART00000079114.6; Scholpp et al., 2009), ascl1a (ENSDART00000056005.5; Allende and Weinberg, 1994), ascl1b (ENSDART00000183550.1; Allende and Weinberg, 1994) probes have been previously published (www.zfin.org) and plasmids were validated by sequencing.

Plasmids were linearized and transcribed with T7 or SP6 RNA Polymerases (Thermo Fisher Scientific). Antisense RNA probes were generated using the DIG- or DNP-based RNA labeling kits from Roche (Merck, Germany). The quality of the RNA probe was checked by agarose gel electrophoresis.

Chromogenic whole-mount in situ hybridizations were performed as previously described (Holzschuh et al., 2003). After fixation, storage and rehydration, embryos were washed three times in PBST and treated with 10 µg/ml Proteinase K (AppliChem, 15 min incubation per 24 h of development). Embryos were washed with PBST and fixed in 4% PFA for 20 min at RT. Embryos were washed five times in PBST and incubated in hybridization mix (50% formamide, 5× SSC, 5 mg/ml torula yeast RNA type IV, 50 µg/ml heparin, 0.1% Tween-20) at 65°C for 4 h. Next, embryos were incubated with digoxigenin-labeled antisense RNA probes in hybridization mix (dilution 1:200-1:500) at 65°C overnight. After hybridization, embryos were washed at 65°C for 20 min each with the following solutions: twice with 50% formamide in 2× SSCT, twice with 25% formamide in 2× SSCT for 20 min each, twice in 2× SSCT for 20 min each, three times in 0.2× SSCT for 20 min each. Next, embryos were washed once in 0.1× SSCT in 0.5× PBST for 10 min and twice in PBST for 10 min each at room temperature. Embryos were blocked for 2-3 h with 2% heat inactivated goat serum (Vector Laboratories) and 4 mg/ml BSA (AppliChem) in PBST. Embryos were incubated overnight at 4°C in blocking solution supplemented with the alkaline phosphatase-coupled anti-digoxigenin antibody (Roche, 11093274910, 1:3000). Embryos were washed six times in PBST at room temperature for 20 min each and three times in NTMT [100 mM NaCl, 100 mM Tris-HCl (pH 9.5), 50 mM MgCl₂, 0.1% Tween-20] for 15 min each. For the staining reaction, 0.18 mg/ml BCIP (5-bromo-4-chloro-3-indolyl-phosphate, AppliChem) and 0.45 mg/ml NBT (4-nitroblue tetrazolium chloride, AppliChem) were added to NTMT. Staining was performed in the dark and stopped by three washes in PBST supplemented with 0.1 mM EDTA. For longer storage, embryos were fixed in 4% PFA for 1 h at room temperature. After three more washes in PBST, embryos were transferred to 80% glycerol in PBST supplemented with 0.1 mM EDTA and stored at 4°C protected from light.

Fluorescent in situ hybridizations (FISH) were performed as described above (‘Whole-mount in situ hybridizations’) until addition of probes. The digoxigenin- and dinitrophenol-labeled antisense probes (see ‘Generation of in situ hybridization probes’) together were added to the hybridization mix and the embryos were incubated at 65°C overnight. Washes were performed at 65°C as described above. Embryos were washed in TNT [100 mM Tris-HCl (pH 7.5), 150 mM NaCl, 0.5% Tween-20] and blocked in TNT supplemented with 1% Boehringer Block (Roche) for 2-3 h at RT. The anti-digoxigenin-POD Fab fragment antibody (Roche, 11207733910) was used 1:400 in blocking solution and embryos were incubated overnight at 4°C. Embryos were washed eight times for 15 min each in TNT at room temperature, rinsed in 100 mM borate buffer (pH 8.0) and incubated in staining mix containing 2% dextrane sulfate, 1% tyramide stock solution Alexa Fluor 488 (Thermo Fisher Scientific), 0.0015% H₂O₂ and 112.5 µg/ml 4-iodphenole in 100 mM borate buffer (pH 8.0) for 1 h at room temperature. Embryos were washed three times in TNT, incubated in 0.3% H₂O₂ in TNT for 30 min and then washed five times in TNT for 5 min each at room temperature. Embryos were blocked in 1% blocking reagent (Roche) in 100 mM maleic acid, 150 mM NaCl (pH 7.5) for 2-3 h at RT. The anti-DNP HRP antibody (Perkin Elmer, FP1129) was used at 1:200 in blocking solution and embryos were incubated overnight at 4°C. Embryos were washed eight times for 15 min each in TNT at room temperature, rinsed in 100 mM borate buffer (pH 8.0) and incubated in staining mix containing 2% dextrane sulfate, 1% tyramide stock solution Alexa Fluor 555 (Thermo Fisher Scientific), 0.0015% H₂O₂ and 112.5 µg/ml 4-iodphenole in 100 mM borate buffer (pH 8.0) for 1 h at RT. Embryos were washed three times in TNT and three times in PBST for 5 min each at room temperature. Embryos were transferred to 80% glycerol in PBS and stored at 4°C protected from light.

Imaging of chromogenic whole-mount in situ hybridizations was performed with the Zeiss Axioskop2, Axiovision SE64, Rel.4.9.1 with AxioCam ICc1 with a Plan-Neofluar 20×/0.5 or 10×/0.3 objective and DIC optics. Embryos were mounted in 80% glycerol in PBST supplemented with 0.1 mM EDTA or in 80% glycerol in H₂O supplemented with 1% standard agarose (Bioron). All shown images are single z-planes from an image stack.

Imaging of fluorescent stainings was performed with an inverted microscope with LSM-880 Airyscan (Zeiss). A 40× objective LD LCI Plan Apochromat 40×/1.2 autocorr (Zeiss) was used. Embryos were embedded in 80% glycerol in H₂O supplemented with 1% standard agarose (Bioron). The immersion medium was glycerol. The imaging program ZEN 2.3 SP1 FP3 (black) version 14.0.22.201 was used. If Airyscan was used, image stacks were processed with the ZEN black software using the Airyscan algorithm (v. 2.3 SP1 Zeiss). All images shown are single z-planes from an image stack, except for Fig. 2A,B,C; Fig. 4A,B,C,C″,D,D″; Fig. 7A,B,C,C″,D,D″, Fig. 10C,D,E,F and Fig. S3A,B, which show orthogonal projections of an image stack. Fig. 2D,D′ and D″ show single z-planes from an image stack of laterally mounted embryos.

Counts of cell nuclei (Fig. 7E-E′; Table S7) were performed on single confocal image planes for consecutive optical sections representing similar anatomical regions of the TPZ for each embryo analyzed. The TPZ regions for analysis were chosen on the dorsal-most plane in which Sox2^low cells were visible and Sox2^high cells present at the ventricle. The same regions were analyzed also on the following nine more-ventral z-planes (optical distance of z-planes 2 µm). pH3⁺ nuclei were individually marked and counted using ImageJ2 (v2.9.0/1.53t) with the analyze plugin Cell Counter. pH3⁺ nuclei extending over more than one focal plane were included only once in the count of the central focal plane of an individual nucleus. The Sox2⁺ nuclei are densely packed and assignment of single planes in z to a single nucleus is difficult. The average diameter of a nucleus in this region is 4-5 µm. We counted each nucleus in each plane. At 2 µm optical plane distance, we therefore estimate that each nucleus may be counted ∼2-3 times. Sox2^low versus Sox2^high nuclei were distinguished and counted based on distinctly different relative levels of immunofluorescence intensity in each focal plane.

The assignment of pH3⁺ nuclei to Sox2^low versus Sox2^high was difficult based on the fact that pH3⁺ nuclei are in mitosis, and during mitosis the nuclear envelope resolves, such that the Sox2 immunoreactivity in the cell does not reveal a distinctly bright nucleus. Therefore, if the Sox2 level could not be identified non-ambiguously, we assigned pH3⁺ nuclei in the Sox2⁺ region of the TPZ to Sox2^low or Sox2^high based on the following three criteria: (1) nuclei surrounded on three out of four sides by Sox2^low nuclei were considered Sox2^low; (2) nuclei surrounded on three out of four sides by Sox2^high nuclei were considered Sox2^high; (3) nuclei located at the ventricular surface and on both sides at the ventricular wall between Sox2^high nuclei were considered Sox2^high. The counts of Sox2^high nuclei may in addition be affected by interkinetic nuclear migration. However, as M-phases typically occur at the apical side, this should not affect our evaluation of pH3 nuclei counts in Sox2^high cells. For comparison of mitotic activities, we calculated the fraction of pH3 nuclei among Sox2^high and Sox2^low nuclei counts. Due to multiple counts of individual Sox2⁺ nuclei (but not of pH3 nuclei, see above), we estimate that the mitotic index may be 2-3 times higher than the pH3/Sox2⁺ count ratio. Statistical analysis and graphs were generated using Prism 9 for macOS Version 9.5.0 (525).

Figures were assembled in Photoshop (version 13.0). Image levels were linearly adjusted except for Fig. 5B-C′,F-G′, Fig. S8 and Fig. S10A-F′ to improve visibility in very dark stained areas.

For WISH experiments analyzing WT, compound-treated or heat shock overexpression-treated embryos, shown in Fig. 1G-R′, Fig. 8, Fig. 9, Fig. S1, Fig.S2G-J′ and Fig. S14, for each condition 10-20 larvae were examined for their expression patterns and representative images are shown. When the numbers (n) are provided in images or legends, the number indicates how many larvae were imaged. If staining patterns or intensity varied within one experimental condition (Fig. 9B′,E′,J′), one embryo with a staining pattern representative for the majority of embryos analyzed is shown, and the numbers in the bottom right corner indicate the number of embryos with the representative expression pattern and the total number of embryos analyzed for this condition.

For analysis of genetic mutants or fluorescent WISH shown in Fig. 1A-F′, Fig. 2, Fig. 3C-N′, Fig. 4, Fig. 5, Fig. 6C-F′, Fig. 7, Fig. 10, Fig. S2A-D′, Fig. S3, Fig. S8, Fig. S9, the numbers (n) of embryos imaged for each condition are provided in the figure legends or in the image panels, and in Table S5.

The tail of each 96 hpf larva was cut and transferred to a new tube for genotyping as described above. The rest of each larva was stored individually in 75 µl RNA later solution (Ambion, AM7024) on ice. Following genotype detection by PCR, total RNA was extracted from 2-3 embryos pooled per genotype using the RNeasy Mini Kit (Qiagen, 74106) and QIAshredder columns (Qiagen, 79656). The sample was homogenized in 1% β-mercaptoethanol in 600 µl RLT buffer (Qiagen, 74106). The suspension was transferred to a QIAshredder column and centrifuged for 1 min at 10,000 g. Then 700 µl 70% ethanol was added to the flow through and transferred to RNeasy Mini spin columns. After centrifugation (15 s, 10,000 g), 350 µl RW1 buffer (Qiagen, 74106) was added to the membrane and centrifuged. A digestion step with DNase I was performed by adding 80 µl of the DNase I incubation mix (consisting of 10 µl DNase I stock solution with a concentration of 2.73 Kunitz units/µl and 70 µl RDD buffer; Qiagen, 74106) to the RNeasy spin column membrane for 15 min on the benchtop to remove DNA from the membrane. After two more washing steps according to the manufacturer's protocol, the column was dried by centrifugation for an additional 2 min. Total RNA was eluted by adding 30 µl H₂O to the filter membrane. Tubes were incubated for 1 min at RT and centrifuged for 1 min at 10,000 g. The RNA concentrations were determined by NanoDrop measurements.

We used 100 ng total RNA as a template for reverse transcription into cDNA. cDNA was generated with the SuperScript III RT Kit according to the manufacturer's protocol using oligo(dT)_12-18 (0.5 µg/µl) primers. The qPCR was performed with the SYBR Green Supermix (Sso Advanced Universal, Bio-Rad). The 10 µl qPCR mix contained 5 µl 2× SYBR Green Supermix, 1 µl primer mix (0.5 µM final concentration of each primer) and 2 µl cDNA. The PCR plate was carefully sealed and centrifuged for 10 min at 1077 g. qPCRs were performed with the LightCycler (Roche), and programs are given in Table S4. The data analysis was performed as in Taylor et al. (2019), and actb2 was used as reference gene. We note that for calculations of relative expression, we excluded one of the three biological replicates for the double mutant, as the CT value for actb2 control in this replicate was off. Therefore, the number of biological replicates for her6 and her9 qPCR in her6,her9 double mutants is two. For all others the number of biological replicates is three. Statistical analysis was carried out using Microsoft Excel 2016.

We thank Sylke Lange for technical assistance, Roland Nitschke and Angela Naumann from the Life Imaging Center (LIC) for advice on confocal imaging and image analysis, and Sabine Götter for excellent fish care. We thank Beatrice Weber for initial characterization of her4 and her6 mutant embryos, Masha Miranda Tolentino Voigt for initial characterization of Notch-dependency of her genes, Samuel Wöhrle for initial characterization of Tg(hsp:her6-FLAG) embryos, Leonie Bohnert for establishing the HCR-RNA FISH technique in our lab and Niklas Mayle for performing the WISH procedure for shha expression in her6 and her9 mutants. We thank Christian Altbürger for scientific discussion. The neurod1 and neurog1 plasmids were obtained from Patrick Blader. We thank Chaitanya Dingare for advice on designing the olig2 HCR probe, and Rebecca Peters for help in designing Fig. 11H.