Early Neurogenesis Was Affected in the Forebrain of Dll1 Functional Ortholog dlc Mutant Zebrafish

To explore whether loss of Dll1 expression affects brain development in other vertebrates, zebrafish were utilized, as the genome is well studied and genetically manipulated zebrafish are available. In zebrafish, the delta family, containing dla, dlb, dlc, and dld, has been identified as an ortholog of the Dll family in mammals (Fig. S1). While the sequence of Dll1 is highly similar to sequences of dla and dld (Fig. S1), the dynamic expression pattern of Dll1 is different from either of them during embryogenesis. For example, Dll1 is expressed in both somite and brain development in an oscillatory expression pattern [12], while dla is not expressed in the presomitic mesoderm (PSM), and dld is constantly expressed in PSM [36]. Nonetheless, dlc was found to be the only gene of Delta/Serrate/lag-2 (DSL) ligands with oscillating expression in developing somite and serves as the segmentation oscillation modulator during somitogenesis of somite formation in zebrafish [36, 37]. As an oscillatory expression pattern is a critical characteristic of Dll1 in progenitors during cortical development [12], the allied periodic expression patterns of Dll1 and dlc suggest that dlc might be a functional homolog of mammalian Dll1 in zebrafish. We, therefore, examined the effects of the loss of dlc in the brains of zebrafish using beatm98 zebrafish, named bea−/−, compared to the bea+/- controls [23].

We found that the zebrafish carrying the homozygous dlc mutation survived until the adult stage with normal reproductive ability (Fig. 1a). Nonetheless, the adult dlc mutant fish showed smaller body size and lighter body weight (Fig. 1b) as well as lighter forebrain weight compared to the bea+/- controls (Fig. 1c). To inspect whether the forebrain composition was affected in dlc mutants, we prepared 12 µm cryosections for immunostaining and in situ hybridization analysis. The results showed no significant difference in the proportion of HuC/D-positive neurons between adult dlc mutants and controls (Fig. 1d), although somite formation was found to be affected at early embryogenesis, as previously reported (Fig. S2a) [38]. To account for the robust regenerative capacity of zebrafish [39], which may obscure the effects of dlc expression loss on early pallium neurogenesis, we analyzed the dlc mutant zebrafish at 2 days post-fertilization (dpf) (Fig. S2b) and 5 dpf (Fig. 1e). We found that the number of nuclei in the hemisphere was unaffected based on DAPI staining (Fig. 1f). Nonetheless, the results of in situ hybridization analysis showed that the area where cells expressed the neural progenitor gene sox2, intermediate neural progenitor gene eomesa, or neuronal gene HuC/elavl3 was significantly increased (Fig. 1g, h and S3).

Next, we asked whether the minor phenotype was due to compensation for the loss of dlc expression by other delta genes by examining the expression of all delta genes, including dla, dlb, and dld, as well as dlc (Fig. S4). As reported previously [23], at 5 dpf, while dlc expression was mildly downregulated in the forebrain of dlc mutant (Fig. S4), a severe somite defect was easily observed compared to the control (Fig. S2a), indicating the functional loss of dlc in bea−/− zebrafish. The expression of dla (p = 0.086) and dlb (p < 0.001) in the forebrain of dlc mutant fish was increased, and the expression of dld was not changed compared to bea+/- controls. While the upregulated dla and dlb expression implies a possible compensation for losing functional dlc expression to activate Notch signaling, the expression of the Notch effector her6 was significantly decreased (Fig. S4). The downregulated her6 expression supports the phenotype of prematuration (Fig. 1g, h, and S3) and indicates that the upregulated dla and dlb expression may not fully compensate for the dlc functions in dlc mutant fish. This finding underscores the significant role of dlc in regulating zebrafish forebrain development. Notably, these results are partially in accord with the Dll1 conditional knocking out phenotype, showing that neuronal gene expression was increased and progenitor gene expression was downregulated in Nestin-cre; Dll1fl/fl cortices, suggesting a phenotype of massive prematuration in the expanse of the progenitor pool [40]. Taken together, our results indicate that the loss of dlc would affect neurogenesis in the developing forebrain in a manner similar to the Dll1-deficient mammalian neocortex despite no gross abnormality being observed in the adult forebrain.

Notch Signaling in Forebrain Development Is Not Solely Controlled by Lateral Inhibition Among Vertebrates

Phenotypic differences in Notch signaling deficiency between mice and zebrafish may be due to inconsistent expression patterns of delta gene Dll1 and its orthologs among species, as well as delta transcriptional effector her gene, during forebrain development across species. To this, we collected forebrain samples from developing zebrafish and mammalian mice as well as turtles (Pelodiscus sinensis) and avians (Gallus gallus). In the developing forebrain of larval zebrafish, progenitors were aligned in a T-shaped region [41], contrasting with the ventricular zone observed in tetrapods, including amphibians, reptiles, and mammals (Fig. 2) [42]. In the developing mouse neocortex, Dll1 was stochastically enriched in the ventricular zone and decreased during later stages, along with Dll1 transcriptional effector Hes1 (Fig. 2a) [12]. A similar stochastic expression pattern of Dll1 could be observed in the developing forebrains of stage 17 turtles and chickens at E5 and 7 (Fig. 2b, c). Furthermore, unlike in mice, Dll1 and Hes1 expressions were not downregulated at later stages in chicken brains (Fig. 2c). In larval zebrafish forebrains, dlc signals were enriched in the T-shaped germinal region of the pallium and co-expressed with the known neural progenitor gene sox2, similar to mouse cortical progenitors [43] (Fig. 2d, e and S4), while dld, the sequence-wise ortholog of Dll1, was generally expressed in the forebrain, including subdivisions such as pallium and thalamus (Fig. S4 and S5). This conserved specific enriched expression pattern of dlc in the germinal zone of zebrafish forebrain, and Dll1 in the germinal zone of mouse neocortex supports our hypothesis that dlc is the functional ortholog of mammalian Dll1 in zebrafish. Taken together, these results from multiple species show that Dll1 and its orthologs, dlc, are expressed in a similar stochastic pattern in the developing forebrain of vertebrates.

Fig. 2

Expression pattern of Dll1, Hes1, and orthologs in developing vertebral forebrain. a In situ hybridization of Dll1 and Hes1 in developing mouse dorsal telencephalon at E12.5 and E15.5 using probes specifically targeting mouse Dll1 and Hes1 mRNA. b In situ hybridization of Dll1 and Hes1 in developing dorsal telencephalon of stage 17 turtles using probes specifically targeting turtle Dll1 and Hes1 mRNA. c In situ hybridization of Dll1 and Hes1 in developing chicken dorsal telencephalon at E5 and E7 using probes specifically targeting chicken Dll1 and Hes1 mRNA. d In situ hybridization of dlc and her6 in developing forebrain of 5 dpf larval zebrafish using probes specifically targeting zebrafish dlc and her6 mRNA. e Two-color in situ hybridization of dlc and sox2 in developing forebrain of 5 dpf larval zebrafish. Colocalized signals are indicated by arrowheads. P, pallium; DT, dorsal thalamus (thalamus); VT, ventral thalamus (prethalamus); Po, preoptic region; lfb, lateral forebrain bundle [13]. Scale bars are indicated

Previous reports have indicated that oscillated Notch signaling in mammalian neurogenesis is controlled by the homeostasis of oscillatory gene Hes1, which has an irregular salt-and-pepper pattern that differs from the simple conventional lateral inhibition with an ordered salt-and-pepper pattern [44, 45]. Therefore, in an oscillatory pattern where the distance between Hes1-positive cells varies, it is unlikely to fit the simple lateral inhibition situation where there is a great central tendency of distance between two signal-positive cells. To test this, we measured the minimal distance from dlc/Dll1-positive cells to the nearest dlc/Dll1-positive cells and the nearest cell (Fig. 3). Quantitative analysis showed that the pattern of minimal distance of dlc-positive cells in larval zebrafish was significantly different from Dll1-positive cells in both chickens and mice (Fig. 3a). If Dll1 expression was controlled solely by lateral inhibition, the coefficient of variation (c.v.) of the relative distance of the two nearest Dll1-positive cells would theoretically be limited. However, in line with the previous results for the developing mouse neocortex, the standard deviation of the relative distance was high and close to the average value, with c.v. = 0.8879 [12]. Similar to the results in mice, a moderate but not limited coefficient of variation for the relative distance of the two nearest dlc- or Dll1-positive cells was also observed in zebrafish forebrain at 5 dpf, with c.v. = 0.5803, and a high value for chicken cortex at embryonic day 5, with c.v. = 0.8121 (Fig. 3a). Notably, the relative distance between the two nearest Dll1-positive cells in zebrafish was significantly different compared to either chickens or mice, and no significance could be detected between chickens and mice.

Fig. 3

dlc-positive cells were stochastically distributed in the forebrain of larval zebrafish. a Quantitative analysis of the relative minimal distance between a signal-positive cell to the nearest signal-positive cell in the developing forebrain of 5 dpf larval zebrafishes, E5 chicken, and E12.5 mouse. * represent significance with p value < 0.01 by Student’s t-test. b–d L() analysis of the spatial distribution of Dll1-positive cells in the developing forebrain of E12.5 mouse (d), Dll1-positive cells in the developing forebrain of E5 chicken (c), and dlc-positive cells in the developing forebrain of 5 dpf larval zebrafish (d)

If Notch signaling is not controlled solely by lateral inhibition, it may be activated periodically, specifically in an oscillatory pattern. To determine whether the distribution of dlc/Dll1-positive cells in the developing forebrain of zebrafish and chickens was similar to that in mice in which Dll1 was expressed in the known oscillatory pattern, we performed L function analysis, L(), to investigate the distribution of dlc/Dll1-positive cells in the germinal zone (Fig. 3b–d). Because the distribution of dlc/Dll1-positive cells was inherited from the distribution of all cells on the slices, the spatial patterns of dlc/Dll1-positive cells should be compared to the L() of all cells. L() analysis results showed that the distribution of dlc/Dll1-positive cells in the brains of all species analyzed in the present study was not random when compared to the simulated results but rather shared a similar distribution pattern of local regularity and global clustering (Fig. 3b–d and S6). This implies that the neural progenitors may be selected to express the Delta gene, dlc/Dll1, using a conserved mechanism across mice, chickens, and zebrafish.

dlc Oscillates in the Developing Zebrafish Forebrain

As dlc oscillates during somite formation [23] and dlc-positive cells could be stochastically observed in the zebrafish forebrain (Figs. 2 and 3), we asked whether dlc was expressed in an oscillatory pattern during forebrain development, as has been observed in the developing mouse forebrain. To address this, we monitored the expression of dlc in the developing zebrafish forebrain using a transgenic fish, tg(dlc::dlc-mCherry), in which normal somitogenesis was reported [34]. This transgenic strain contained 3 kbp 5′ upstream promoter region and 8 kbp full dlc sequence fused with mCherry/red fluorescent protein. To evaluate whether dlc-mCherry could serve as an appropriate reporter to monitor oscillatory expression, we examined the dynamic expression of dlc-mCherry in the PSM, where oscillatory dlc expression can be observed and is critical to somite formation. The live imaging results showed periodic mCherry expression and clear somite boundaries during 140 min of monitoring (Fig. S7a-c). Additional findings from crossing dlc-mCherry with dlc mutant zebrafish showed that somite defects could be partially rescued by dlc-mCherry, as shown by the observation of nine somites formed in dlc-mCherry crossed with dlc mutant zebrafish, compared to only four somites in dlc mutant zebrafish (Fig. S2a and S7d).

The results of live imaging of mCherry signals and the rescue of somite defects by dlc-mCherry in dlc mutants suggested that the molecular behavior of dlc-mCherry is similar to innate dlc expression in controls. We then used this transgenic fish with dlc-mCherry to monitor dynamic dlc expression during forebrain development in zebrafish. A GFP construct was introduced into 4-cell-stage embryos via microinjection to label the shape of cells, and samples were collected at 2 dpf for 5 h of live-imaging recording (Fig. 4). To monitor dlc-mCherry signals within cells, we delineated cellular boundaries based on GFP expression and monitored the dynamic mCherry signals over a 300-min period (Fig. 4a–b”). The raw signals of mCherry were then processed, and peaks were determined using the imaging processing procedure described in “Materials and Methods,” and the resulting waves displaying the corrected mean mCherry signal intensity within individual cells were presented (Fig. 4a–b”, S8 and S9). In order to examine whether the waves matched the periodic pattern with a consistent amplitude and duration of period, we aligned the first peak of all analyzed cells as the zero of pseudo-timelines and applied autocorrelation analysis, a function to identify periodic patterns on MATLAB software (Fig. 4a”, b”, S8 and S9). Based on the results of the autocorrelation analysis (Fig. S9), we found that the dynamic expression patterns of dlc-mCherry were various and not apparently synchronized across all the cells tracked. Further clustering analysis based on the k-means machine learning strategy [46], three groups of oscillating behaviors could be clustered using 5 oscillating properties of robustness, including mean peak-to-peak interval, mean amplitude, mean square of autocorrelation coefficient, range of autocorrelation coefficient, and the number of cycles (Fig. 4c).

Fig. 4

Fluctuating dlc expression was not synchronized in the developing forebrain of larval zebrafish. a, b Time-lapse imaging of mCherry signals and the heatmap of corrected mCherry signals in GFP cells (8*8 μm). Dashed lines delineate the cellular boundary of GFP cells. Time (minutes) was indicated. a’, b’ Line charts show the amplitude of mCherry signals in cell 10 (a’) and cell 35 (b’). a”, b” Curve diagrams show the autocorrelation results of cell 10 (a”) and cell 35 (b”). a, a’, a” Time-lapse imaging and the analyses of cell 10 in group 1. b, b’, b” Time-lapse imaging and the analyses of cell 35 in group 3. c Box and dot plots show the mean peak-to-peak interval, mean amplitude, mean square of autocorrelation coefficient, and range of autocorrelation coefficient in three groups. p values by Student’s t-test are indicated. d Line charts show the mean amplitude in each group. e The pie chart shows the composition of analyzed GFP cells

Among these three groups, group 1 exhibited a significant difference from groups 2 and 3 in terms of oscillating properties and the wave of mean amplitude (Fig. 4c, d). As shown in the montage of sequential images, the intensity of dlc-mCherry signals of cell 10 in group 1 displayed a clear up-and-down dynamic expression pattern, in contrast to that of cell 35 in group 3 (Fig. 4a–a’ and b–b’). Also, the wave of mean amplitude in each group showed a similar pattern as that of the individual cells (Fig. 4d and S8). The autocorrelation analysis of cells in group 1 showed distinct cycles between peaks and troughs, while cells in group 2 or 3 showed some connected peaks or troughs (Fig. 4a”, b” and S9). Collectively, our findings suggested that group 1 exhibited the most robust periodic behavior, whereas some cells in the other two groups also displayed periodic yet noisy waves (Fig. S8 and S9), indicating a distinct oscillatory pattern from the oscillatory Dll1 pattern in mice [12]. These results suggested the dlc oscillations of zebrafish forebrain progenitors occurred in a spectrum manner during determination (Fig. 4e), implying a prototypical pattern of dlc-involved Notch signaling in forebrain progenitors of zebrafish.

Different Effects of Attenuating Notch Signaling on Neuronal Differentiation in Developing Forebrain of Zebrafish Compared to Chickens and Mice

If some of the dlc expression in the developing zebrafish forebrain exhibited an oscillatory pattern similar to that of Dll1 in the developing mouse forebrain, we then investigated the effects of attenuating this dlc-involved Notch signaling on neural differentiation during forebrain development in mice, chickens, and zebrafish. To this, we applied a γ-secretase inhibitor, MK-0752, to block Notch signaling during the critical period of neural differentiation. Several previous studies have demonstrated that blocking Notch signaling promotes neural differentiation in the cortical progenitors in the developing mouse cerebral cortex [6, 47]. Therefore, we first examined the effect of blocking Notch signaling using MK-0752 in mice.

Brain slice culture was utilized to prolong the exposure time to MK-0752 in the developing forebrain (Fig. 5a, left panel). As the Tbr2-positive intermediate progenitor population was primarily affected by the Notch signaling dysfunction condition [6], we measured Tbr2-positive intermediate progenitors after MK-0752 treatment. After 20 h of treatment, Tbr2-positive intermediate progenitors in the ventricular zone (VZ) were significantly increased in the MK-0752 treatment group compared to the DMSO control group, which is consistent with previous studies (Fig. 5a) [6, 47]. We then employed a similar slice culture procedure using chicken brain at E5 (Hamburger and Hamilton stages 24/25), administering MK-0752 or DMSO as the control in the culture medium for 24 h, during which the time chicken cortex experiences massive neurogenesis (Fig. 5b, left panel) [32]. As cortical progenitors in chicken and zebrafish undergo neural differentiation to produce neurons directly without producing Tbr2-positive intermediate progenitors [48], we measured the changes in neuron population in chicken and zebrafish brains after MK-0752 treatment. The immunostaining results showed that the number of Tuj1-positive neurons that expressed young neuronal gene Tuj1 was significantly increased in the neuronal zone (NZ) outside of the ventricular zone on chicken pallium (Fig. 5b), indicating that inhibiting Notch signaling with MK-0752 promotes neural differentiation in both chicken pallium and mouse cerebral cortex.

Fig. 5

Blocking Notch signaling increased neural precursor numbers in developing mouse and chicken dorsal telencephalon but not in the larval zebrafish forebrain. a Schematic diagram of experimental design of MK-0752 treatment in developing mouse cerebral cortex, immunostaining and quantitative analysis of Tbr2-positive cells (nuclei) in VZ. VZ, ventricular zone; IZ, intermediate zone; CP, cortical plate. Solid lines indicate ventricular surface. Nuclei: DAPI staining. b Schematic diagram of experimental design of MK-0752 treatment in developing chicken pallium, immunostaining, and quantitative analysis of Tuj1-positive cells (nuclei) in NZ. Tuj1-positive cells are defined as nuclei surrounded by Tuj1-positive signals (arrowhead). Dashed lines indicate the pia surface and solid lines indicate the ventricular surface. Nuclei: DAPI staining. c Schematic diagram of experimental design of MK-0752 treatment in developing larval zebrafish forebrain, in situ hybridization, and quantitative analysis of area with HuC/elavl3 signals. Close arrowheads indicate positive signals and open arrowheads indicate negative signals. Scale bars are indicated. Error bars represent standard deviation, and data points are shown in dots. *Significance with p value < 0.01 by Student’s t-test

Finally, we investigated the effects of blocking Notch signaling in developing zebrafish forebrain. To avoid potential impact on somitogenesis, treating the zebrafish with MK-0752 from the fertilized egg stage is not applicable. Also, due to the challenge of specifically introducing MK-0752 to the zebrafish brain, we could not apply the method we used in mice and chickens. Thus, we treated larval zebrafish with MK-0752 at 4 dpf at 28.5 °C and assessed the effects on neuronal differentiation using in situ hybridization of the neural gene HuC/elavl3 in the forebrain 24 h after treatment (Fig. 5c, left panel). Surprisingly, we found a downregulation of neural gene HuC/elavl3 in zebrafish forebrain after MK-0752 treatment, which contradicts our findings in mice and chickens (Fig. 5a, b) and in dlc mutant fish (Fig. 1g, h). This unexpected result, in conjunction with some oscillatory expression patterns of dlc in zebrafish, suggests that Notch signaling is rudimentary in the developing zebrafish forebrain when compared to the developing mouse neocortex.