Presenilin enhancer 2 is crucial for the transition of apical progenitors into neurons but into not basal progenitors in the developing hippocampus

The hippocampus is a unique brain structure required for cognitive functions. It is well known that the hippocampus arises from the caudomedial edge of the dorsal telencephalic neuroepithelium adjacent to the cortical hem (CH). The hippocampus can be divided into three subdivisions, including the cornu ammonis (CA), the dentate gyrus (DG) and the fimbria (Altman and Bayer, 1990b,c). The hippocampus is known to play important roles in synaptic plasticity, learning and memory, and spatial navigation (Langston and Wood, 2009; Martin and Clark, 2007; Martin et al., 2000; Wood and Dudchenko, 2021). It is believed that ‘place cells’ in the hippocampus are crucial for spatial navigation in animals. For example, different place cells may fire at different locations and encode information in different tasks (Ainge et al., 2007; O'Keefe and Dostrovsky, 1971; Wood et al., 2000). Conversely, abnormalities in the hippocampus are associated with neurological disorders (Connor et al., 2004; Houser, 1990; Lurton et al., 1997; Tamminga et al., 2010; Walton et al., 2012).

Hippocampal morphogenesis involves a number of complex cellular processes, including formation of the hippocampal primordium, generation of neurons and glial cells in different subregions, and migration of neurons (Altman and Bayer, 1990a,b,c; Angevine, 1965). Whereas neural progenitor cells (NPCs) in the CA give rise to pyramidal neurons (Altman and Bayer, 1990c), those in the primary dentate matrix generate granule neurons, which migrate tangentially and ultimately populate in the DG (Altman and Bayer, 1990a). However, molecular mechanisms underlying hippocampal development are poorly understood.

Presenilin enhancer 2 (Pen2; Psenen) is the smallest subunit of the γ-secretase complex and exhibits multiple cellular functions. First, Pen2 is essential for proteolytic cleavage of presenilin proteins (PSs) (De Strooper, 2003). Second, Pen2 can act as a substrate- binding site for γ-secretase (Fukumori and Steiner, 2016; Shah et al., 2005). Third, Pen2 is essential for the endoproteolytic activity of γ-secretase (Bammens et al., 2011). Fourth, Pen2 can regulate apoptosis. For example, knockdown of Pen2 causes apoptotic cell death in zebrafish (Campbell et al., 2006). Finally, it has been shown that straight knockout (KO) of Pen2 (Psenen) results in embryonic lethality (Bammens et al., 2011), and that conditional KO (cKO) of Pen2 in telencephalic NPCs leads to postnatal lethality (Cheng et al., 2019). Thus, Pen2 is important for the survival of animals.

It has been reported that Pen2 is involved in the 19q13 microdeletion syndrome displaying microcephaly (Forzano et al., 2012; Gana et al., 2012), suggesting that Pen2 is important for the central nervous system (CNS). To study the role of Pen2 in cortical development, we previously took advantage of the Emxl-Cre mouse and generated cortical NPC-specific Pen2 cKO mice (Cheng et al., 2019). The latter exhibit depletion of apical progenitors (APs) and a transiently increased number of basal progenitors (BPs) in the dorsal telencephalon (Cheng et al., 2019), suggesting that Pen2 may control the switch of APs to BPs in the developing cortex. However, it remains unknown what role Pen2 exerts during hippocampal morphogenesis.

To answer the above question, we employed the hGfap-Cre mouse, in which the expression of Cre is controlled by the human Gfap (glial fibrillary acidic protein) promoter (Brenner et al., 1994; Zhuo et al., 2001). It has been shown that Cre is widely expressed in pyramidal neurons in the CA and granule cells in the DG in the hippocampus of the hGfap-Cre mouse (Brenner et al., 1994; Zhuo et al., 2001). The crossing of Pen2^f/f to hGfap-Cre allowed us to generate Pen2 cKO mice, in which Pen2 is inactivated in NPCs in the developing hippocampus. We found that Pen2 cKO mice exhibited hippocampal hypoplasia. We detected massive loss of APs in the Pen2 cKO hippocampus. We observed decreased levels of Hes1 and Hes5, unchanged levels of neurogenin 2 (Ngn2) and increased levels of neurogenic differentiation1 (NeuroD1) in the Pen2 cKO hippocampus. Together, this study highlights the importance of Pen2 in the developing hippocampus.

We aimed to investigate the physiological function of Pen2 in the developing hippocampus. To this end, floxed Pen2 mice were crossed to hGfap-Cre to generate Pen2 cKO (Pen2^f/f;hGfap-Cre) mice. To check the expression of Cre recombinase in the developing hippocampus, the hGfap-Cre mouse was bred with the mTmG reporter (Muzumdar et al., 2007) to obtain hGfap-Cre;mTmG (Fig. 1A). Co-staining experiments revealed abundant GFP⁺/Pax6⁺ cells in the hippocampal neuroepithelium in hGfap-Cre;mTmG mice at E13.5 (Fig. 1B). We found that GFP was distributed mainly in the cytoplasm and the processes of Pax6⁺ cells (Fig. 1B). As Pax6 is a marker for APs, the above immunohistochemistry (IHC) results indicated that Cre is expressed in APs in the hippocampal neuroepithelium. Moreover, we performed co-staining of GFP/Tbr2. We observed numerous GFP⁺/Tbr2⁺ cells in hGfap-Cre;mTmG mice at E13.5 (Fig. 1B). Thus, Cre is also expressed in BPs in the developing hippocampus.

To compare Pen2 levels between control and Pen2 cKO mice (Fig. 1C), western blotting was conducted using hippocampal lysates prepared from mice at postnatal day 0 (P0) (Fig. 1D). We observed a significant reduction of Pen2 in Pen2 cKO mice compared with controls (Fig. 1D). Next, quantitative real-time PCR (qRT-PCR) was performed using RNA samples from hippocampi at E17.5. As expected, levels of Pen2 mRNAs were significantly decreased in Pen2 cKO embryos (Fig. 1E). The above results suggest efficient inactivation of Pen2 in the Pen2 cKO hippocampus.

We next completed Nissl staining using brain sections from mice at P0 and P6 (Fig. 1F). We found that the hippocampus was small in Pen2 cKO mice compared with littermate controls at different ages (Fig. 1F). Quantification results confirmed significant reductions in the average area of the Pen2 cKO hippocampus at P0 and P6 (Fig. 1G). Thus, conditional deletion of Pen2 caused hippocampal malformation.

To explore cellular mechanisms underlying the above phenotype, we examined NPCs by conducting fluorescence immunohistochemistry on Pax6 and Tbr2 using brain sections at embryonic day 16.5 (E16.5) and E17.5. We found that the immunoreactivity of Pax6 in the hippocampal neuroepithelium was comparable between control and Pen2 cKO mice at E16.5. Cell counting results revealed that the average number of Pax6⁺ cells per section in the hippocampal neuroepithelium did not differ between control and Pen2 cKO mice at E16.5 (Fig. 2C). In contrast, the immunoreactivity of Pax6 in the hippocampal neuroepithelium was decreased in Pen2 cKO mice at E17.5 compared with littermate controls (Fig. 2A,B). In line with this, the number of Pax6⁺ cells in the hippocampal neuroepithelium was less in Pen2 cKO mice than in controls (Fig. 2C).

For results on BPs, the immunoreactivity of Tbr2 in the Pen2 cKO hippocampus was comparable with that in controls at E16.5, and it was decreased in Pen2 cKOs at E17.5 (Fig. 2D,E). The average number of Tbr2⁺ cells in the hippocampus per section was unchanged in Pen2 cKO mice at E16.5 compared with controls (Fig. 2F), and it was significantly decreased in Pen2 cKOs at E17.5 (Fig. 2F).

We further studied whether NPCs were affected in the cortex of Pen2 cKO mice. First, to examine the deletion efficiency of Pen2, we performed qRT-PCR analysis on Pen2 using RNA samples from Pen2 cKO cortices at E16.5, and we observed a significant reduction (Fig. S1A), suggesting that Pen2 is inactivated in the cortex. Second, the immunoreactivity of Pax6 in the cortex did not differ between control and Pen2 cKO mice at E17.5 (Fig. S1B,C). We found that the average number of Pax6⁺ cells per section was not significantly different between control and Pen2 cKO mice (Fig. S1D). Third, we found that the immunoreactivity of Tbr2 in the cortex was comparable between control and Pen2 cKO mice (Fig. S2A,B), and that there was no difference on the average number of Tbr2⁺ cells between control and Pen2 cKO mice (Fig. S2C). In addition, we performed immunohistochemistry using a number of markers for different cortical layers (Fig. S3). We observed that the immunoreactivity of Cux1, Ctip2 (Bcl11b) or Tbr1 did not differ between control and Pen2 cKO mice at E17.5, and that the thickness of Cux1⁺, Ctip2⁺ or Tbr1⁺ cell layers was comparable in these two genotypes (Fig. S3A-C). Thus, hGfap-Cre-mediated deletion of Pen2 did not cause significant effect on cortical lamination.

The neuroepithelium is the stem cell niche for APs/BPs in the hippocampal primordium, and it can be divided into the ammonic neuroepithelium (ANE) and the dentate neuroepithelium (DNE) [or the primary (1°) matrix] (Fig. 3A) (Altman and Bayer, 1990b; Galichet et al., 2008). The DNE gives rise to the secondary (2°) matrix and the tertiary (3°) matrix, and neurons in the CA or in the DG are generated from NPCs in the ANE or in the DNE, respectively (Tamminga et al., 2010). We counted Pax6⁺ and Tbr2⁺ cells at E17.5 for each subdivision separately. First, we found that the average number of Pax6⁺ cells per section was significantly decreased in the ANE, the DNE and the 2° matrix in Pen2 cKO mice at E17.5 compared with controls (Fig. 3B). There was no difference in the 3° matrix between control and Pen2 cKO embryos (Fig. 3B). Second, the average number of Tbr2⁺ cells in the ANE, the DNE and the 2° matrix, but not the 3° matrix, in control embryos significantly differed from that in Pen2 cKO mice (Fig. 3C). Overall, these results confirmed the loss of NPCs by deletion of Pen2.

To test the possibility that the depletion of APs in Pen2 cKO mice may be due to deficient proliferation, we performed BrdU pulse-labeling experiments. BrdU was injected intraperitoneally to pregnant dams at E16.5 and embryos were collected 30 min after the injection. Double-staining experiments revealed no significant reduction on the total number of Pax6⁺/BrdU⁺ cells in the hippocampal neuroepithelium in Pen2 cKO mice compared with controls (Fig. 3D,E). Moreover, the ratio of the number of Pax6⁺/BrdU⁺ cells to that of Pax6⁺ cells did not differ between control and Pen2 cKO mice at E16.5 (Fig. 3F). Thus, the ability to enter the synthesis phase (S-phase) of the cell cycle for Pen2 cKO APs was not different from that for controls, suggesting unimpaired proliferation by deletion of Pen2. Hence, the depletion of APs in the Pen2 cKO hippocampus was unlikely due to the proliferative capability of APs.

To test whether deletion of Pen2 affected apoptosis, we conducted immunohistochemistry on cleaved caspase 3 (CC3) using brains sections at E17 and E18.5 (Fig. 3G). We used NPC-specific Ppp2cα cKO mice as the positive control, as these mutants display robust apoptosis in the cortex (Huang et al., 2020). However, CC3⁺ cells were not detected in control and Pen2 cKO mice at E17 and E18.5 compared with Ppp2cα cKO mice (Fig. 3G). Thus, the loss of NPCs in Pen2 cKO mice may not be due to abnormal cell death.

We have recently shown that Emx1-Cre-mediated deletion of Pen2 causes transiently increased population of BPs in the telencephalon (Cheng et al., 2019). As the number of BPs in Pen2 cKO hippocampi was unchanged at E16.5 but decreased at E17.5 (Fig. 2D-F), we then analyzed Tbr2⁺ cells in mice at E17 (Fig. 4A). We found that the immunoreactivity of Tbr2 in the hippocampus was comparable between control and Pen2 cKO mice (Fig. 4A). Moreover, the number of Tbr2⁺ cells in the hippocampus was not significantly different between control and Pen2 cKO mice (Fig. 4B). Overall, the results for Tbr2⁺ cells at E16.5, E17 or E17.5 indicated no transient increase in BPs in the Pen2 cKO hippocampus.

To study whether deletion of Pen2 affected the differentiation of APs into BPs in the hippocampus, we carried out double staining of Sox2/Tbr2 using brain sections at E17 (Fig. 4C). We found that the average number of Sox2⁺/Tbr2⁺ cells in the hippocampus did not differ between control and Pen2 cKO mice (Fig. 4D), suggesting no significant change on BPs being differentiated from APs by deletion of Pen2. To study whether deletion of Pen2 altered neurogenesis, we examined neurons in the hippocampus at E17. Prox1, a marker for granule neurons, was used to perform immunohistochemistry (Fig. 4E). We observed significantly increased number of Prox1⁺ cells in the hippocampus in Pen2 cKO mice compared with littermate controls (Fig. 4F), suggesting transiently increased neurogenesis.

To test whether deletion of Pen2 affected the differentiation of APs into neurons in the hippocampus, we conducted double-staining for Pax6/NeuroD1 and Sox2/Tbr1 using brain sections at E17. We chose NeuroD1 and Tbr1 to label hippocampal neurons for the following reasons. First, hippocampal subregions, including the CA1, the CA3 and the DG were reduced in Pen2 cKO mice at P2 or P6 (Fig. 1F), suggesting that both granule neurons and pyramidal neurons may be affected. Second, it has been shown that NeuroD1 is expressed in immature granule cells in the hippocampus (Nakahira and Yuasa, 2005; Pleasure et al., 2000). Third, it has previously been reported that Tbr1 may label both granule cells (in the CA3 and the DG) and pyramidal cells (in the CA1) in the developing hippocampus at E14, E16 and E18 (Barry et al., 2008).

We found that the immunoreactivity of NeuroD1 (Fig. 5A,B) was increased in the Pen2 cKO hippocampus at E17 compared with controls (control=100%±4.1%, cKO=131%±1.7%; P<0.01). Next, we counted cells doubly positive for Pax6 and NeuroD1. We observed a significantly increased number of Pax6⁺/NeuroD1⁺ cells in the hippocampal neuroepithelium in Pen2 cKO mice compared with controls (Fig. 5C). As NeuroD1 is crucial for the differentiation of granule cells in the hippocampus (Miyata et al., 1999), we further analyzed the immunoreactivity of NeuroD1 in the hippocampal neuroepithelium, and it was significantly increased in the Pen2 cKO group compared with the control (Fig. 5D).

For double-staining of Sox2/Tbr1 (Fig. 5E,F), there was increased immunoreactivity of Tbr1 in the Pen2 cKO hippocampus at E17 compared with littermate controls (control=100%±2%, cKO=126.5%±3.1%; P<0.01). Sox2⁺/Tbr1⁺ cells in the hippocampal neuroepithelium were counted, and the number was highly increased in Pen2 cKO mice (Fig. 5G). Overall, the above results suggest that loss of Pen2 may cause enhanced transition of APs to neurons in the hippocampus.

To explore molecular mechanisms underlying the malformation of the Pen2 cKO hippocampus, we analyzed several downstream targets of γ-secretase. Western blotting was conducted using protein samples from hippocampi at P0. We found that levels of the C-terminal fragment of PS1 (PS1-CTF) were decreased in Pen2 cKO mice compared with controls (Fig. 6A). We observed a significant increase in the levels of the C-terminal fragment of amyloid precursor protein (APP-CTF) in Pen2 cKO mice (Fig. 6A). The above results were in agreement with the notion that Pen2 is an essential component of the γ-secretase complex (De Strooper, 2003).

We next used total RNA samples prepared from the hippocampus at E17.5 to carry out qRT-PCR analyses on Hes1 and Hes5, two key members in the Notch signaling. We found that mRNA levels for Hes1 and Hes5 were significantly decreased in the Pen2 cKO hippocampus compared with controls (Fig. 6B). Moreover, levels for Hey1 and Hey2 mRNAs were also significantly decreased in the Pen2 cKO hippocampus compared with controls (Fig. 6B). These results thus confirmed impaired Notch signaling in Pen2 cKO mice.

We have recently found that Emx1-Cre-mediated deletion of Pen2 leads to increased levels of neurogenin 2 (Ngn2), which may account for enhanced switch of APs to BPs in the dorsal telencephalon (Cheng et al., 2019). Surprisingly, we found that mRNA levels of Ngn2 were unchanged in the hippocampus of Pen2^f/f;hGfap-Cre mice (Fig. 6C). It is likely that Ngn2 may not be involved in Pen2-dependent transition of APs to neurons in the hippocampus. In contrast, our qRT-PCR analysis revealed significantly increased mRNA levels of Neurod1 in the Pen2 cKO hippocampus (Fig. 6C), indicating upregulation of Neurod1 expression. Additionally, we examined Prox1 and Tbr1, which are proneuronal transcriptional factors. However, mRNA levels for Prox1 and Tbr1 in the hippocampus did not differ between control and Pen2 cKO mice at E17.5 (Fig. 6D). Thus, the transcription of Prox1 or Tbr1 was not affected by deletion of Pen2.

Given that Pen2 cKO embryos exhibited loss of NPCs at E17.5 (Fig. 2), the change in APP cleavage products and Notch targets in the E17.5 hippocampus (Fig. 6A-C) could be a consequence of the decreased population of progenitors. To exclude this possibility, we examined major Notch targets in Pen2 cKO mice at E16.5, an age when the loss of NPCs was not seen (Fig. 2). As the mouse hippocampus at E16.5 is very small, the resulting protein lysates are insufficient for us to run western blotting. Therefore, we dissected the E16.5 hippocampus to prepare RNA samples. We then conducted qRT-PCR analysis. First, our results revealed a remarkable reduction in Pen2 mRNA levels in the hippocampus of Pen2 cKO mice at E16.5 compared with controls (Fig. 6E). For Pen2 cKO mice at E16.5, the reduction on Pen2 mRNAs in the hippocampus (Fig. 6E) was greater than in the cortex (Fig. S1A), suggesting higher deletion efficiency of Pen2 in the hippocampus. Second, we observed significantly decreased levels of Hes1 and Hes5 in Pen2 cKO mice at E16.5 compared with controls (Fig. 6E). The above results suggest that hGfap-Cre- mediated deletion of Pen2 is efficient in the hippocampus. Overall, the significant reduction in expression levels of Hes1 and Hes5 in the Pen2 cKO hippocampus at E16.5 strongly suggests that molecular changes in Notch downstream targets precede the occurrence of the NPC loss at E17.5. Third, our qRT-PCR results revealed significantly increased levels of Neurod1 but not Ngn2 in the Pen2 cKO hippocampus at E16.5 (Fig. 6F), suggesting that increased expression of Neurod1 but not Ngn2 may take place before Pen2 cKO mice exhibit massive NPC loss at E17.5. As NeuroD1 is an important proneuronal transcriptional factor (Imayoshi et al., 2008; Miyata et al., 1999), we reasoned that increased Neurod1 might drive the transition of APs to neurons in the Pen2 cKO hippocampus.

The above molecular results suggest that the Notch signaling may be involved in Pen2-dependent transition of APs to neurons. To further test this hypothesis, we used a Tg line expressing the Notch1 intracellular domain (ICD) in a Cre-dependent manner (Cheng et al., 2019). This Tg mouse was bred with Pen2^f/f and Pen2^f/+;hGfap-Cre to generate Pen2^f/f;hGfap-Cre;LSL-N1ICD (Pen2 cKO;N1ICD) mice (Fig. 7A). First, qRT-PCR analysis confirmed significant reductions in Pen2 mRNA levels in the hippocampus of Pen2 cKO mice with and without N1ICD expression compared with littermate controls at E17.5 (Fig. 7B). Second, we found that levels of N1ICD were significantly increased in the hippocampus in Pen2 cKO mice expressing N1ICD compared with those without N1ICD expression (Fig. 7B).

To examine whether Cre-dependent expression of N1ICD affected NPCs in Pen2 cKO mice, brain sections at E17.5 were used for immunohistochemistry on Pax6 and Tbr2 (Fig. 7C-F). Pax6⁺ cells and Tbr2⁺ cells in the hippocampal neuroepithelium were counted and then averaged. First, whereas Pax6⁺ cells were hardly observed in the Pen2 cKO hippocampus, they were abundantly present in control and in Pen2 cKO mice expressing N1ICD (Fig. 7C). Quantification results revealed a significantly increased number of Pax6⁺ cells in the hippocampal neuroepithelium in Pen2 cKO mice expressing N1ICD compared with those without N1ICD expression (Fig. 7D). Second, we observed abundant Tbr2⁺ cells in control and Pen2 cKO mice expressing N1ICD (Fig. 7E). We found that there was a significantly increased number of Tbr2⁺ cells in hippocampal neuroepithelium in Pen2 cKO mice expressing N1ICD compared with those without N1ICD expression (Fig. 7F). Hence, Cre-dependent expression of N1ICD partially rescued populations of APs and BPs in the Pen2 cKO hippocampus.

To examine whether the expression of N1ICD affected the population of neurons in Pen2 cKO mice, we conducted IHC on Prox1 using brain sections at E17.5 (Fig. 8A). We counted Prox1⁺ cells in the dentate migrating stream (DMS) and the DG. We found that there was a significantly decreased number of Prox1⁺ cells in the DMS in Pen2 cKO mice without N1ICD expression compared with those expressing N1ICD (Fig. 8B). Therefore, conditional expression of N1ICD rescued the neuronal population in the Pen2 cKO hippocampus.

To examine whether Cre-dependent expression of N1ICD rescued the Notch signaling in Pen2 cKO mice, we performed in situ hybridization on Hes1 and Hes5 using brain sections at E16.5. We observed very faint Hes1⁺ and Hes5⁺ signals in the hippocampal neuroepithelium of Pen2 cKO mice. We found that Hes1⁺ and Hes5⁺ signals were comparable between control and Pen2 cKO;N1ICD mice (Fig. 8C), suggesting enhanced Notch signaling by the expression of N1ICD in Pen2 cKO mice.

We next performed in vitro experiments to examine the effect of N1ICD on Hes1/Hes5. HEK293T cells were transfected with a vector carrying the N1ICD element and one expressing a luciferase reporter driven by the promoter of Hes1 or Hes5. We found that the expression of N1ICD significantly enhanced the promoter activity of Hes1 or Hes5 (Fig. 8D), suggesting that N1ICD may positively regulate the expression of Hes1 or Hes5. We then constructed plasmids expressing Hes1/HA or Hes5/HA (Fig. 8E). HEK293T cells were then transfected with a vector carrying the Hes1 or Hes5 gene and with one expressing a luciferase reporter driven by the Neurod1 promoter. We found that the expression of Hes1 or Hes5 significantly inhibited the promoter activity of Neurod1 (Fig. 8F), suggesting that Hes1 or Hes5 may negatively regulate the expression of Neurod1. Moreover, we found that co-transfection using vectors expressing Hes1, Hes5 and the luciferase reporter for the Neurod1 promoter also caused significantly decreased promoter activity of Neurod1 in HEK293T cells (Fig. 8F). Together, the above in vitro data suggest that N1ICD may regulate the expression of Neurod1 via Hes1 and Hes5.

Recent evidence has revealed the involvement of Pen2 in microcephaly related to the 19q13 microdeletion (Forzano et al., 2012; Gana et al., 2012). We aim to investigate how Pen2 may regulate hippocampal growth. As straight knockout of Pen2 causes an embryonic lethal effect in mice, conditional KO techniques have been used to produce viable cell type-specific Pen2 cKO mice for postnatal studies (Bi et al., 2021; Cheng et al., 2019; Hou et al., 2021). Here, we have generated a unique mouse model in which Pen2 is inactivated in NPCs in the hippocampal primordium. Our lineage-tracing experiments confirmed the expression of Cre in APs and BPs in the Pen2 cKO hippocampus, which is in line with previous findings (Brenner et al., 1994; Zhuo et al., 2001). Importantly, the observation of abnormal hippocampus in Pen2 cKO mice strongly suggests that Pen2 is required for hippocampal morphogenesis.

It is believed that AP may undergo one of three distinct fates during early stage of the CNS development (Imayoshi and Kageyama, 2014). For example, one AP can generate two daughter APs through symmetric division. Alternatively, one AP may produce one AP and one BP or neuron through asymmetric division (Imayoshi and Kageyama, 2014). One of the most significant findings in this study is the abundance of Sox2⁺/Tbr1⁺ cells and Pax6⁺/NeuroD1⁺ cells in the Pen2 cKO hippocampus. Since these two types of cells represent immature neurons that are being differentiated from APs, the robust increase in them strongly suggests that deletion of Pen2 causes significantly enhanced transition of APs to neurons. In contrast, our BrdU pulse-labeling experiments revealed no significant change in the ratio of the number of Pax6⁺/BrdU⁺ cells to that of Pax6⁺ cells in the hippocampus in Pen2 cKO embryos. Therefore, deletion of Pen2 does not impair the proliferation of APs. In addition, we observed no significant change in the number of CC3⁺ cells in the Pen2 cKO hippocampus. Thus, deletion of Pen2 does not significantly promote cell death. These findings prompt us to exclude the possibility that the loss of APs in the hippocampus of Pen2 cKO mice may be due to abnormal cell cycle or cell death.

In this study, we did not observe any significant increase in the population of Tbr2⁺ cells in the Pen2 cKO hippocampus at any age examined, e.g. E16.5, E17 or E17.5, suggesting that the generation of BPs is not significantly enhanced by deletion of Pen2. Moreover, there was no increase in the number of Sox2⁺/Tbr2⁺ cells in the Pen2 cKO hippocampus, suggesting that the AP-to-BP switch is not promoted by deletion of Pen2. Given that there was an increased number of Prox1⁺ cells and Sox2⁺/Tbr1⁺ cells in the Pen2 cKO hippocampus at E17, we reason that Pen2 might regulate hippocampal growth by inhibiting the transition of APs into neurons but not into BPs. Interestingly, we have recently reported premature generation of BPs in the ventricular zone (VZ) in the Pen2^f/f;Emx1-Cre cortex (Cheng et al., 2019). Different timings for Pen2 deletion may account for distinct phenotypes in Pen2^f/f;Emx1-Cre and Pen2^f/f;hGfap-Cre mice. More specifically, Emx1-Cre-mediated deletion of Pen2 causes increased transition of APs to BPs in the VZ of the telencephalon at E12.5 and E13.5 (Cheng et al., 2019). Whereas hGfap-Cre-mediated deletion of Pen2 results in enhanced transition of APs to neurons in the hippocampal neuroepithelium at E17, it does not affect populations of APs and BPs in the cortex. The above discrepancy may be due to the following reasons. First, the starting time for Cre expression driven by the human Gfap promoter is later than that by the Emx1 promoter (E13.5 versus E10.5). Second, hGfap-Cre-mediated deletion efficiency of Pen2 is higher in the hippocampus than in the cortex. Third, deletion efficiency of Pen2 in the cortex is lower in Pen2^f/f;hGfap-Cre mice than in Pen2^f/f;Emx1-Cre mice.

To identify molecular mechanisms underlying Pen2-dependent hippocampal growth, we have focused on the Notch signaling, a major downstream target of γ-secretase. As expected, mRNA levels of Hes1, Hes5, Hey1 and Hey2 are decreased by deletion of Pen2. Furthermore, the generation of Pen2 cKO mice expressing N1ICD allowed us to observe a significant rescue effect on the population of APs in the hippocampus. Therefore, the Notch signaling may play a crucial role in Pen2-dependent fate determination of APs in the developing hippocampus. Interestingly, roles of Notch1, Rbpjκ or jagged 1 in the postnatal hippocampus (Breunig et al., 2007; Ehm et al., 2010; Lavado and Oliver, 2014) and the adult brain (Ables et al., 2011; Blackwood, 2019; Blackwood et al., 2020) have been studied by several groups. It has been shown that Notch1 may regulate the cell cycle exit of NPCs in the postnatal hippocampus (Breunig et al., 2007), and that jagged 1 regulates the proliferation of NPCs in the postnatal DG (Lavado and Oliver, 2014).

The following evidence indicates upregulation of NeuroD1, but not Ngn2, by deletion of Pen2 in the hippocampus. First, mRNA levels of Neurod1, but not Ngn2, were increased in Pen2 cKO hippocampal lysates. Second, the immunoreactivity of NeuroD1 in the hippocampus was stronger in Pen2 cKO mice than in controls. Third, whereas the promoter activity of Hes1 or Hes5 was enhanced by N1ICD, that of Neurod1 was repressed by Hes1 or Hes5. We reason that NeuroD1, but not Ngn2, may serve as the key molecular mechanism for enhanced transition of APs to neurons in the Pen2 cKO hippocampus. Overall, the above findings are in agreement with the concept that NeuroD1 is important for cell fate specification (Guillemot, 2007). Indeed, it has been shown that straight knockout of NeuroD1 impairs neuronal differentiation (Liu et al., 2000b; Schwab et al., 2000), and that NeuroD1 is required for neuronal differentiation in different brain regions (Liu et al., 2000a; Miyata et al., 1999; Morrow et al., 1999).

As shown in Fig. 9, we propose a cellular model to summarize molecular mechanisms for Pen2-dependent maintenance of NPCs during hippocampal development. First, deletion of Pen2 in hippocampal APs causes loss of γ-secretase activity (Fig. 9). Second, the generation of NICD is prevented, followed by inhibition on the expression of Hes1/5 and other Notch targets (Fig. 9). Third, decreased Hes1/5 causes increased levels of NeuroD1, which may significantly enhance the transition of APs to neurons in the hippocampal neuroepithelium (Fig. 9). Fourth, populations of APs and BPs are subsequently decreased in various hippocampal subregions. It is worth mentioning that the pool of NPCs in the 3° matrix was unchanged in the Pen2 cKO hippocampus at E17.5, suggesting that the number of DG-born neurons may be comparable between two genotypes. As neurons in the DG are generated from NPCs in the 3° matrix, as well as the DNE in the hippocampus (Nakahira and Yuasa, 2005), the reduction in NPC populations in the DNE in Pen2 cKO mice at E17.5 suggests that the number of DNE-born neurons may be significantly decreased in Pen2 cKO mice at and after E17.5. Overall, depletion of NPCs in the Pen2 cKO hippocampal neuroepithelium may result in deficient neurogenesis in hippocampal subregions and consequently lead to hippocampal malformation.

Mouse breeding was conducted under an Animal Protocol approved by the IACUC of the Model Animal Research Center (MARC) at Nanjing University. All mouse experiments were performed in accordance with the Guide for the Care and Use of Laboratory Animals of the MARC at Nanjing University.

Pen2^f/f, LSL-N1ICD and mTmG mice were reported by us recently (Bi et al., 2021; Cheng et al., 2019; Hou et al., 2021; Wang et al., 2021a). Human Gfap-Cre (Brenner et al., 1994) mice were purchased from the Jackson Laboratory. To generate Pen2 cKO mice, we bred Pen2^f/f with hGfap-Cre to obtain Pen2^f/+;hGfap-Cre. The latter were crossed to Pen2^f/f to generate Pen2^f/+;hGfap-Cre (control) and Pen2^f/f;hGfap-Cre (Pen2 cKO) mice. To generate Pen2 cKO mice expressing N1ICD, we bred Pen2^f/f with LSL-N1ICD to produce Pen2^f/+;LSL-N1ICD, which were bred with Pen2^f/f to create Pen2^f/f;LSL-N1ICD. The latter were crossed to Pen2^f/+;hGfap-Cre to generate Pen2^f/+;hGfap-Cre;LSL-N1ICD (control) and Pen2^f/f;hGfap-Cre;LSL-N1ICD (Pen2 cKO;NICD) mice. Genetic background of the mice used in this study was C57BL/6.

The mice were maintained in a SPF room in the core animal facility in the MARC at Nanjing University. Both male and female mice were used in this study. The mice were housed in groups (four to six mice per cage) and had ad libitum access to food and water. The animal room was maintained on a 12 h light/dark cycle. Lights were automatically turned on at 07:00 and switched off at 19:00. The room temperature was kept at 25±1°C. The birth date of the mice was defined as P0, and the day of vaginal plug detection in pregnant mice was E0.5.

After neonatal mice were sacrificed, brains were dissected and then fixed in 4% paraformaldehyde (PFA) at 4°C overnight. Tissues were dehydrated with graded ethanol. After being embedded in paraffin, each block was sectioned sagittally using a microtome (the thickness of each was 10 μm). Each paraffin block contained four brains, including two control and two Pen2 cKO littermates. Sections were deparaffinized with xylene and then treated with 0.1% Cresyl Violet for 2 min, followed by rinsing with distilled water for 1 min. Sections were air-dried and then sealed using neutral resin (Sinopharm Chemical Reagent). For each brain, a total of three sections spaced by 200 μm were used for measurement.

Pregnant mice were sacrificed with CO₂. Embryos were dissected out and then fixed in 4% PFA at 4°C overnight. Tissues were dehydrated using graded ethanol. Brains were embedded with paraffin. Each paraffin block contained one brain and was sectioned coronally using a microtome (10 μm in thickness). Sections were deparaffinized with xylene and then rehydrated with graded ethanol. Sections were boiled in sodium citrate buffer (0.01 M) for 25 min followed by blocking with 5% BSA at room temperature for 30 min. Sections were incubated with the primary antibody at 4°C overnight. For fluorescence immunohistochemistry, sections were incubated with the secondary antibody such as Alexa Fluor 488 goat anti-mouse (Jackson ImmunoResearch, 115-545-003, 1:500), Alexa Fluor 594 goat anti-rabbit (Jackson ImmunoResearch, 111-585-003, 1:500), Alexa Fluor Cy3 donkey anti-rabbit (Jackson ImmunoResearch, 711-165-152, 1:500), Alexa Fluor Cy5 donkey anti-goat (Jackson ImmunoResearch, 715-175-147, 1:500) or Alexa Fluor 488 donkey anti-rat (Abcam, ab150153, 1:500). Primary antibodies used for immunohistochemistry are listed in Table S1. Fluorescence images were captured using a TCS SP5 laser confocal microscope (Leica Microsystems) or a LSM880 laser confocal microscope (Zeiss).

Cortical and hippocampal samples were collected from newborn mice and embryos. Methods for western blotting have been described previously (Liu et al., 2017). Antibodies used for western blotting are listed in Table S1.

To label NPCs in proliferation, BrdU was intraperitoneally injected into pregnant mice (B5002, Sigma Aldrich; 100 mg/kg) and embryos were collected 30 min after the injection.

For quantification studies, three or four embryos in each genotype were used. For each embryo, at least two coronal sections spaced 100 μm apart were used for cell counting purposes. Fluorescence images for Pax6⁺, Tbr2⁺, Pax6⁺/BrdU⁺, Pax6⁺/NeuroD1⁺, Sox2⁺/Tbr1⁺, Sox2⁺/Tbr2⁺ and Prox1⁺ cells were captured under the 20× objective lens of a TCS SP5 microscope. To count cells positive for Pax6 or Tbr2 in the hippocampus, the ANE, the DNE, the 2ry matrix and the 3ry matrix were outlined in images captured from each brain section, and cells were counted for each sub-region separately (Fig. 3B,C). In Figs 2C,F, 3E,F, 4B and 7D,F, the total number of positive cells in the hippocampus was the sum of cells in four different subregions. For results in Figs 4D and 5C,G, positive cells in the hipppocampal neuroepithelium were counted. Adobe Photoshop was used to count different types of cells. The experimenter was blind to genotypes of brain sections.

All the florescence images captured by Leica confocal microscope were converted to 8-bit images and the scale was set in pixels in ImageJ. The neuroepithelium in each section was manually outlined using the freehand selection tool in ImageJ. The immunoreactivity was defined as Raw Signal Intensity divided by the Area. Two sections spaced at 100 µm from each embryo were used for image capture and two images were used for measurement on the immunoreactivity for each embryo.

An ABI StepOne Plus machine was used for qRT-PCR. Total RNA (1μg) of each sample was reverse transcribed using the PrimeScript RT reagent Kit (Takara). RT-PCR was conducted using the 2× RealStar Green Fast Mixture with Rox (GenStar). All the primers used for qRT-PCR analyses were as follows; For Pen2, TGGATTTG CGTTCCTGCCTTTTCT (forward) and ATGAAGTTGTTAGGGAGTGCC (reverse); for Hes1, TCCAAGCTAGAGAAGGCAGACA (forward) and CGCGGTATTTCCCC AACA (reverse); for Hes5, AACACAGCAAAGCCTTCGCC (forward) and AAGCA GCTTCATCTGCGTGTC (reverse); for N1ICD, ACTTGTCAGATGTGGCCTCG (forward) and ATTCAAGTGGCTGATGCCCA (reverse); for Hey1, AGTTAACTCC TCCTTGCCCG (forward) and CGATGATGCCTCTCCGTCTT (reverse); for Hey2, GCTACAGGGGGTAAAGGCTAC (forward) and GAGATGAGAGACAAGGCGCA (reverse); for NeuroD1, CAAAGCCACGGATCAATCTT (forward) and TCCCGGG AATAGTGAAACTG (reverse); for Tbr1, AGAGGCTCTGGAAACACGAA (forward) and ACTCGACTCGCCTAGGAACA (reverse); for Prox1, CAGATGCAT TACCTCGCAGC (forward) and CTGGAACCTCAAAGTCATTTGC (reverse); and for GAPDH, AATGTGTCCGTCGTGGATCT (forward) and CCCTGTTGCTGTAGCCG TAT (reverse).

RNA isolation and reverse transcription are described in the qRT-PCR section. An 848 bp fragment of the Hes1-coding sequence (NM008235.2) was amplified by PCR using the following primers: ATGCCAGCTGATATAATGGA (forward) and TCAGTTCCGCCACGGTCTCC (reverse). A 902 bp fragment from the Hes5 mRNA sequence (NM010419.4) was amplified by PCR using the following primers: AGGACTACAGCGAGGGCTACTC (forward) and TTAGAA GCCTTCAGAACAGCCT (reverse). RNA probe labeled by digoxigenin (DIG) were prepared using the DIG RNA labeling kit (Roche, 11175025910) and purified using a RNA clean-up kit (Bioteck RP1801).

Brains were fixed in 4% PFA/DEPC-H₂O overnight and dehydrated in 30% sucrose/DEPC-PBS at 4°C overnight. Brains were embedded in OCT, then frozen quickly with liquid nitrogen. Coronal sections (12 μm) were prepared using a Leica CM 1950 cryostat at −20°C and stored at −80°C until use. After being incubated with the Hes1 or Hes5 probe at 65°C for 12 h, sections were washed and incubated with anti-DIG-AP Fab fragments (Roche, 11093274910) at room temperature for 3 h. The AP activity was developed using NBT/BCIP substrates (Roche).

Mouse Hes1 and Hes5 were constructed using the pcDNA5-HA vector (Hou et al., 2021; Wang et al., 2021a). The Hes1- and Hes5-coding sequences were amplified by PCR from cDNA libraries prepared from mouse brain. Plasmid expressing N1ICD was purchased from Addgene (Addgene, 20183). For the luciferase reporter assay, the promoter regions of Hes1 (−2000 bp to 46 bp), Hes5 (−1941 bp to 427 bp) or Neurod1 (−2100 bp to 406 bp) were constructed using pGL3-Luc vector.

293 T cells were cultured in DMEM supplemented with 10% FBS and 1% penicillin-streptomycin at 37°C in an incubator, and transfected by Lipofectamine-2000 (Invitrogen). The luciferase assay was performed using Dual-Glo (Promega) according to the manufacturer's protocol 24 h after the transfection. A CMV promoter-driven Renilla luciferase was used as the internal control. Data were obtained from at least three independent experiments.

Data are presented as mean±s.e.m. Owing to highly homogeneous phenotypes in Pen2 cKO mice, the sample size of three or four for each genotype was used for cell counting according to recently published studies (Ahrendsen et al., 2018; Lavado et al., 2010; Liu et al., 2019; Wang et al., 2021b; Zou et al., 2014). GraphPad Prism7 and t-tests (two-tailed) were used for statistical analysis to examine main genotype effects between control and Pen2 cKO groups. One-way ANOVA with multiple comparisons was conducted to evaluate genotype effects in experiments using mice with and without expression of N1ICD. P<0.05 is considered as statistically significant.

We thank Drs Tingting Liu, Jinxing Hou, He Wang and Shanshan Cheng for advice and technical assistance.

The peer review history is available online at https://journals.biologists.com/dev/article-lookup/doi/10.1242/dev.200272