Hes genes and neurogenin regulate non-neural versus neural fate specification in the dorsal telencephalic midline

The telencephalic hemispheres are formed by bilateral evagination of the anterior end of the neural tube. The dorsal telencephalon is further subdivided along the medial-lateral axis into three regions. The most lateral region becomes cortical neuroepithelium, which later gives rise to neurons and glial cells of the cerebral cortex. The medial region (the dorsal telencephalic midline region) is divided into the most medial part, the choroid plexus epithelium, and an intermediate part, the cortical hem, which is a major source of Cajal-Retzius cells of the neocortex(Grove et al., 1998; Meyer et al., 2002; Takiguchi-Hayashi et al.,2004; Yoshida et al.,2005). Cajal-Retzius cells are distributed in the neocortex and guide neuronal migration. It has been shown that this medial-lateral patterning of the dorsal telencephalon is regulated by a combination of transcription factors and secreted signaling factors. For example, the homeodomain transcription factors Msx1/2 and Lhx2/Foxg1 (Bf1) are involved in the development of the choroid plexus and the cortical neuroepithelium,respectively, whereas secreted factors such as Bmps regulate specification of the choroid plexus epithelium by inducing Msx1 and repressing Lhx2/Foxg1 expression(Bach et al., 2003; Xuan et al., 1995; Furuta et al., 1997; Porter et al., 1997; Monuki et al., 2001; Panchision et al., 2001; Hébert et al., 2002; Fernandes et al., 2007).

The choroid plexus is unique in the brain, because it is a non-neural secretory tissue. It produces the cerebrospinal fluid and functions as a blood-brain barrier. The choroid plexus derives from both epithelial and mesenchymal components, with the epithelium facing the ventricular lumen. The choroid plexus epithelial cells are generated from neuroepithelial cells like other cell types of the central nervous system, such as neurons, astrocytes and oligodendrocytes (Sturrock,1979; Thomas and Dziadek,1993; Awatramani et al.,2003; Currle et al.,2005; Hunter and Dymecki,2007). The role of Bmp signaling in the development of the choroid plexus has been intensively analyzed. It has been shown that misexpression of the constitutively active form of Bmp receptors results in an expansion of the choroid plexus at the expense of the cortical neuroepithelium(Panchision et al., 2001),whereas inactivation of the Bmp receptor results in defects of specification of choroid plexus epithelial cells(Hébert et al., 2002; Fernandes et al., 2007). Bmp signaling induces expression of the homeodomain factors Msx1/2, which are involved in the development of the dorsal midline region(Bach et al., 2003; Hébert et al., 2002). However, the precise mechanism underlying generation of this non-neural tissue during the development of the nervous system is, as yet, undetermined.

It is well known that in Drosophila, the basic helix-loop-helix(bHLH) repressor genes of hairy and Enhancer of split[E(spl)] regulate non-neural versus neural fate specification in the ectoderm(Campos-Ortega and Jan, 1991). Cells expressing proneural bHLH genes at higher levels, such as the achaete-scute complex, adopt the neural fate and express Delta, which then activates Notch signaling of neighboring cells. Activation of Notch signaling leads to upregulation of E(spl) genes, which promote non-neural fate specification by repressing proneural genes (lateral inhibition). Thus, proneural and E(spl) genes antagonistically regulate neural versus non-neural cell fate specification. These results raise the possibility that the bHLH repressor genes such as Hes genes, mammalian hairy and E(spl) homologues(Kageyama et al., 2007), are likewise involved in the formation of non-neural tissues in the developing mammalian brain. Although it has been shown that Hes genes repress proneural gene expression(Ishibashi et al., 1995; Chen et al., 1997; Hatakeyama et al., 2004), no previous analyses have shown that Hes genes regulate non-neural versus neural fate specification in the mammalian brain. Hes genes have been shown to maintain neural progenitors or to promote gliogenesis(Ross et al., 2003; Kageyama et al., 2007; Miller and Gauthier,2007).

In this study, we found that the prospective choroid plexus epithelium of the telencephalon expresses both the proneural bHLH gene neurogenin 2(Ngn2) and the repressor genes Hes1 and Hes5, and gives rise to two cell lineages: choroid plexus epithelial cells and Cajal-Retzius cells. Furthermore, Hes1-, Hes3- and Hes5-null mutations lead to the upregulation of Ngn2, to a lack of choroid plexus epithelial cells and to the promotion of Cajal-Retzius cell differentiation. Overexpression of Ngn2 had similar effects. These results suggest that Hes and Ngn2 genes antagonistically regulate the specification of non-neural (choroid plexus)versus neural (Cajal-Retzius cell) fate in the mouse brain.

Mouse Ngn2 cDNA in pEF-BOS was introduced into the developing brain by microelectroporation, as described previously(Fukuchi-Shimogori and Grove,2001; Fukuchi-Shimogori and Grove, 2003). Transfected cells were monitored by co-electroporation of pEF-EGFP.

The floxed Hes1 targeting vector (see Fig. 4A) was linearized with NotI and transfected into TT2 cells, and G418-resistant clones were selected. Genomic DNA was digested with HindIII and analyzed by Southern blot using a 0.6 kb HindIII-BamHI fragment as a 5′-probe. Neomycin selection cassette was removed by transient Cre expression in the targeted TT2 cells. Genotypes were determined by PCR using the following primers: 5′-CAGCCAGTGTCAACACGACACCGGACAAAC-3′ and 5′-TGCCCTTCGCCTCTTCTCCATGATA-3′. The sizes of PCR products for floxed and wild-type alleles are 272 bp and 224 bp, respectively.

Hes1 conditional knockout (cKO) mice were obtained by crossing homozygous Hes1 floxed mice with Emx1-Cre(Iwasato et al.,2000);Hes1^+/-(Ishibashi et al., 1995) mice. Hes1;Hes3;Hes5 cKO mice were acquired by crossing Hes1(floxed/floxed);Hes3^-/-;Hes5^-/-with Emx1-Cre;Hes1^+/-;Hes3^-/-;Hes5^-/-. Emx1-Cre;Hes1(floxed/+);Hes3^-/-;Hes5^-/-embryos were normal and used as control. Nes-CreER^T2;Hes1;Hes3;Hes5 cKO mice were acquired by crossing Hes1(floxed/floxed);Hes3^-/-;Hes5^-/-with Nes-CreER^T2line5-1;Hes1^+/-;Hes3^-/-;Hes5^-/-. Nes-CreER^T2line5-1;Hes1(floxed/+);Hes3^-/-;Hes5^-/-embryos were normal and were used as control. Tamoxifen (6 mg/35 g body weight) was administered by oral gavage to the pregnant mice at E9.5. Rbpj cKO mice were obtained by crossing Emx1-Cre;Rbp-j^+/- with homozygous Rbp-j floxed mice. These mice were maintained on ICR or C57BL/6;ICR mixed background.

A transgene containing 5 kb upstream fragment of Msx1 gene(MacKenzie et al., 1997; Takahashi et al., 1997), EGFP cDNA and SV40 poly-adenylation sequence was used to generate pMsx1-EGFP mice. Mice were analyzed at E10.0 (n=5), E10.5 (n=4) and E11.5(n=4).

X-gal staining was performed as described previously(Imayoshi et al., 2006). Immunohistochemistry was performed as described previously(Imayoshi et al., 2006) with primary antibodies against β-tubulin III/Tuj1 (Covance), GFP (Molecular Probes), reelin (Calbiochem), Msx1/2 (DSHB, The University of Iowa, Department of Biological Sciences), Ngn2 (Santacruz), doublecortin (DCX) (Santacruz), p73(Neomarkers), Calretinin (Swant) and Hes1 (aa86-278), which was produced as previously described (Baek et al.,2006). Goat or donkey anti-species IgG conjugated with Alexa 488 or Alexa 594 (Molecular Probes) were used as secondary antibodies. Sections were analyzed with LSM510 confocal microscopy. In situ hybridization was carried out as described previously(Ohsawa et al., 2005) using mouse reelin, p73, Math2, Slit1, Robo1, Er81 (GenBank Accession Number, BI901885; IMAGE, 5663418), Cux2, Rorb (GenBank Accession Number; IMAGE, 6490704), Prox1, Steel, Ka1, Nt3, Big1(Shinozaki et al., 2004), Hey1, Hey2, Bmpr1a, Lmx1a, Ttr, Wnt3a, Mash1, Egfp, Msx1, Wnt3a,Wnt2b (GenBank Accession Number, AI893147; IMAGE, 353765), Bmp4(GenBank Accession Number, AA473799; IMAGE, 873328), Hes1, Hes5, Msx2,Foxg1 (GenBank Accession Number, AI893944; IMAGE, 388688), Lhx2,Lhx5, Ngn2 and Ngn1 probes.

To determine the cell lineage of the prospective choroid plexus epithelium of the telencephalon, we first examined whether or not neurogenesis occurs in this region. The homeodomain factor Msx1 is expressed in this region of mouse embryos from E10.0 to E11.5 and in the choroid plexus epithelium at E12.5(Fig. 1B,E,F; see Fig. 3I,Q). At E10.5, subsets of Msx1⁺ cells expressed the proneual factor Ngn2(Fig. 1C-C″, arrowheads),suggesting that neurogenesis occurs in the prospective choroid plexus region. In agreement with this notion, some differentiating neurons (DCX⁺)were found in this region (Fig. 1D-D″), although they did not express Msx1, suggesting that Msx1 is downregulated when Ngn2⁺ cells start neuronal differentiation. To trace the lineage of Msx1⁺ cells, we generated transgenic mice carrying the Msx1 promoter-driven EGFP reporter. Because EGFP is relatively stable, it can be used as a short-term lineage tracer that detects cells expressing Msx1 both currently and previously. As expected, EGFP was specifically expressed in the prospective choroid plexus region of transgenic mice from E10.0 to E11.5(Fig. 1E-M″). At E10.0,96.6±3.8% of EGFP⁺ cells expressed Msx1, indicating that the EGFP expression occurred specifically in Msx1⁺ cells(Fig. 1E-E″). At E10.5 and E11.5, subsets of EGFP⁺ cells expressed Ngn2(Fig. 1H-H″,K-K″,arrowheads) and the neuronal marker Tuj1(Fig. 1I-I″,L-L″,arrowheads), indicating that some Msx1⁺ cells indeed differentiated into neurons. Interestingly, p73, a marker of Cajal-Retzius cells, was also expressed (Fig. 1J-J″,M-M″, arrowheads), suggesting that neurons formed in the Msx1⁺ region are Cajal-Retzius cells.

To further analyze the cell lineage of the prospective choroid plexus region, we introduced pEF-EGFP, which directs EGFP expression under the control of the elongation factor 1α promoter, into the prospective choroid plexus region at E9.5 by using an in utero microelectroporation method(Fukuchi-Shimogori and Grove,2001; Fukuchi-Shimogori and Grove, 2003). At E10.5, all EGFP⁺ cells resided in the Msx1⁺ prospective choroid plexus region of the dorsal telencephalic midline (n=6) (Fig. 2A-C). However, at E11.5, many EGFP⁺ cells migrated tangentially into the cortical neuroepitheium, and some of them had already reached the marginal zone of the piriform cortex (n=6)(Fig. 2D-F). At E12.5, most of the cells that migrated laterally seemed to have reached the piriform cortex(Fig. 2G, arrow), and only two regions, the choroid plexus epithelium (the origin, Fig. 2G, asterisk) and the piriform cortex (the destination, Fig. 2G, arrow), were labeled with EGFP (n=5). This finding suggests that the electroporated region gives rise to migrating cells around E11.5 and ceases the formation by E12.5. It has been reported that Cajal-Retzius cells in the piriform cortex expressed Lhx5(Yamazaki et al., 2004), and the EGFP⁺ cells in this region seemed to express this marker(Fig. 2J,K). Furthermore, many of the EGFP⁺ cells expressed Tuj1(Fig. 2L,L′) and reelin(63.5%, n=148) (Fig. 2M,M′, arrowheads) but not calretinin (11.4%, n=210) (Fig. 2N,N′). These results suggest that the electroporated region gives rise to Cajal-Retzius cells destined for the piriform cortex around E11.5. In this experiment, EGFP was not expressed in the hem(Wnt2b⁺, Fig. 2H,I), a known source of Cajal-Retzius cells. Furthermore, cell migration from the electroporated region ceased by E12.5, although the hem is known to generate migrating Cajal-Retzius cells even after E13.5(Takiguchi-Hayashi et al.,2004). These results support the notion that these Cajal-Retzius cells do not derive from the cortical hem but from the prospective choroid plexus region. At E12.5, the cells remaining at the origin expressed the choroid plexus-specific marker transthyretin (Ttr)(Fig. 2O,P) but not reelin(Fig. 2Q). These results suggest that the prospective choroid plexus region(Msx1⁺Wnt2b^-) gives rise to two distinct cell types: Cajal-Retzius cells (neural) and the choroid plexus epithelium (non-neural) around E10.5 to E11.5.

To reveal the molecular mechanism of the fate choice in the dorsal midline region, we examined expression of Hes1 and Hes5 from E10.5 to E12.5. The telencephalic choroid plexus forms bilaterally at the dorsomedial edge of the telencephalon (Fig. 3A). At E10.5, the epithelium of the dorsal telencephalic midline region expressed Bmp4 and the homeodomain gene Lmx1a, which regulates development of the choroid plexus and the cortical hem(Fig. 3C,D)(Millonig et al., 2000; Kuwamura et al., 2005). Likewise, Hes1 was expressed in this region, as well as in the neighboring diencephalic and telencephalic neuroepithelium, while Hes5 was expressed at a lower level in this region than in the telencephalic neuroepithelium at E10.5(Fig. 3E,F). At E11.5, the Bmp4 and Lmx1a expression domain was elongated(Fig. 3G,H) and gradually divided into two regions, the choroid plexus epithelium(Msx1⁺,Ttr⁺) and the cortical hem(Wnt2b⁺) (Fig. 3I-K,N). At this stage, Hes1 and Hes5 expression continued but was gradually downregulated in Ttr⁺ cells of the prospective choroid plexus region (Fig. 3L,M). At E12.5, the choroid plexus epithelial cells became thin and cuboidal (Ttr⁺), whereas cells in the cortical hem were still pseudostratified (Fig. 3R). Bmp4 and Lmx1a were expressed in both regions, whereas Msx1 and Wnt2b expression occurred in the choroid plexus epithelium and in the cortical hem, respectively(Fig. 3O-R). At this stage, Hes1 and Hes5 expression occurred at a high level in the cortical hem but was almost completely repressed in the differentiated choroid plexus epithelium (Fig. 3S,T). These results show that Hes1 and Hes5 expression occurs in the prospective choroid plexus region at E10.5 to E11.5, when fate choice between choroid plexus epithelial cells and Cajal-Retzius cells takes place,and is downregulated at E12.5, when the cell fate is completely specified.

The above results suggest that both Hes1 and Hes5 are expressed in the prospective choroid plexus epithelium when neural versus non-neural cell fate specification occurs. To reveal the role of Hes genes in this region, we decided to examine Hes-null mice. However, Hes1;Hes5 double-null embryos die by E11 before the establishment of the telencephalon(Hatakeyama et al., 2004), and thus they were not suitable for analysis. To overcome this problem, we generated Hes1 floxed mice, in which the region containing exons 2 to 4 was deleted by Cre recombinase (Fig. 4A,B). These mice were crossed with Emx1-Cre mice, which had previously been shown to efficiently result in the recombination of floxed alleles in the dorsal telencephalon(Iwasato et al., 2000). It has been reported that expression of Emx1 starts at E9.5(Yoshida et al., 1997). To monitor the Cre-mediated recombination, we crossed Emx1-Cre mice with R26R reporter mice (Soriano, 1999). Recombination occurred efficiently in the dorsal telencephalic neuroepithelium, including progenitors to the choroid plexus epithelium at E10.5 to E12.5 (Fig. 4C-E). We generated the Hes1 conditional knock-out (cKO) mice by crossing Hes1 floxed mice and Emx1-Cre mice. In Hes1 cKO mice, Hes1 expression in the dorsal telencephalon was downregulated around E10.5 (Fig. 4J, asterisk) and was lost by E11.5 (Fig. 4O,asterisk). Thus, compared with the control, where Hes1 expression was lost by E12.5, downregulation of Hes1 occurred 1-2 days earlier in Hes1 cKO mice. No apparent defect was observed in the developing telencephalon of Hes1 cKO mice (data not shown), probably owing to compensation by other members of the Hes family such as Hes5, which was upregulated in Hes1 cKO mice(Fig. 4K, compare with 4H). Additionally, Hes3 could be upregulated in the absence of Hes1 and Hes5, and we decided to make mice lacking Hes1, Hes3 and Hes5. Because Hes3;Hes5 double-null mice are apparently normal (Hatakeyama et al.,2004), we generated Hes1 cKO mice on a Hes3;Hes5 double-null background(Hes1;Hes3;Hes5 cKO).

Although it has previously been shown that the midbrain, the hindbrain and the spinal cord can develop severe defects such as premature depletion of neural progenitors and disruption of the neural tube structures in conventional Hes1;Hes3;Hes5 KO mice(Hatakeyama et al., 2004; Baek et al., 2006), it was surprising that the cortical hem and the cortical neuroepithelium were only mildly affected in Hes1;Hes3;Hes5 cKO mice. Hes1;Hes3;Hes5 cKO mice showed accelerated differentiation of neurons, including Cajal-Retzius cells in the dorsal telencephalon (see Fig. S1 in the supplementary material), but there were many neural progenitors, and the laminar structures of the neocortex and the hippocampus were not affected (see Fig. S2 in the supplementary material). Hes-related genes Hey1 and Hey2 were found to be expressed in the cortical hem and the cortical neuroepithelium (see Fig. S3 in the supplementary material), and Hey1 expression was upregulated in Hes1;Hes3;Hes5 cKO mice (see Fig. S3A,B in the supplementary material). It has previously been shown that both Hey1and Hey2 inhibit neuronal differentiation and promote maintenance of neural progenitors, such as Hes genes(Sakamoto et al., 2003). Therefore, such mild phenotypes of the dorsal telencephalon of Hes1;Hes3;Hes5 cKO mice were probably due to compensation by Hey1 and Hey2. However, Hey1 and Hey2 were not expressed in the prospective choroid plexus epithelium(see Fig. S3 in the supplementary material and data not shown), and Hes1;Hes3;Hes5 cKO mice displayed a severe defect of the choroid plexus, as described below.

In the control mice, the neuroepithelial cells at the midline became flattened from E11.5 to E12.5 (Fig. 5A,A′,B,B′), and the thin cuboidal epithelium protruded into the lateral ventricles around E12.5 to E15.5(Fig. 5C). By contrast, in Hes1;Hes3;Hes5 cKO mice, the dorsal midline cells were not flattened but remained pseudostratified at E11.5 and E12.5(Fig. 5D,D′,E,E′,asterisks). This region was morphologically very similar to the neighboring cortical and diencephalic neuroepithelium. Even at E15.5, no choroid plexus was formed in Hes1;Hes3;Hes5 cKO mice(Fig. 5F, asterisk). Furthermore, expression of the choroid plexus epithelium-specific gene Ttr was not detectable in the mutant mice at E11.5 or E12.5(Fig. 5H,J), although it had already occurred in the control (Fig. 5G,I). These results indicate that the choroid plexus is completely missing in Hes1;Hes3;Hes5 cKO mice. Similarly, the choroid plexus in the fourth ventricle was severely affected in Hes1;Hes5 conventional KO mice at E10.5 (see Fig. S4 in the supplementary material), although it was not yet formed in the telencephalon of both the wild-type and Hes1;Hes5 conventional KO mice at this stage.

We then examined the lineage of Cajal-Retzius cells which originate from the prospective choroid plexus region and migrate into the piriform cortex. We found that there were more Cajal-Retzius cells(reelin⁺,Lhx5⁺) in the marginal zone of the piriform cortex of Hes1;Hes3;Hes5 cKO mice than in the control mice at both E11.5 (Fig. 5K,L,R,S, arrows) and E12.5(Fig. 5M-Q,T,U, arrows). This finding suggests that Cajal-Retzius cell development is enhanced in the absence of Hes genes. Although significant defects were not observed in the cortical development (see Fig. S2 in the supplementary material), it is possible that overall acceleration of cortical neurogenesis is involved in enhancement of Cajal-Retzius cell formation in the piriform cortex of Hes1;Hes3;Hes5 cKO mice. However, Cajal-Retzius cell formation was not significantly affected in the pallial-subpallial boundary region of the mutant mice (see Fig. S5 in the supplementary material). Furthermore, we generated Hes1;Hes3;Hes5cKO mice by using Nes-CreER^T2 mice(Imayoshi et al., 2006), in which Hes1 was knocked out in the cortical neuroepithelium and the hem but not in the choroid plexus region (see Fig. S6A-F′ in the supplementary material), which developed normally (see Fig. S6I,L in the supplementary material). In these mice, Cajal-Retzius cell formation was not significantly affected in the piriform cortex (see Fig. S6M-P′ in the supplementary material). These results suggest that inactivation of Hes genes in the prospective choroid plexus region mainly contributes to enhancement of Cajal-Retzius cell formation in the piriform cortex, although the possibility of contribution by overall accelerated neurogenesis is not totally excluded.

It was previously shown that the telencephalic choroid plexus is missing in the absence of the Bmp receptor gene Bmpr1a(Hébert et al., 2002). We therefore examined expression of Bmp signaling and related molecules in Hes1;Hes3;Hes5 cKO mice. At E11.5, in these mutant mice, the expression domain of Bmp4 and Lmx1a was reduced in size (Fig. 6A.A′,B,B′), and the expression of the downstream homeodomain genes Msx1 and Msx2 was severely downregulated compared with the control (Fig. 6C,C′,D,D′). Thus, Bmp signaling was attenuated in the absence of Hes genes. However, expression of the Bmp receptor Bmpr1a(see Fig. S7A,B in the supplementary material) and of Noggin, an antagonist of Bmp (data not shown), as well as its responsiveness to Bmp (see Fig. S7C-F in the supplementary material) were not affected in Hes1;Hes3;Hes5 cKO mice. The expression domain of Wnt3a was also reduced in size at this stage(Fig. 6E,E′), although expression of Foxg1 and Lhx2, which are required for cortical development (Xuan et al.,1995; Porter et al.,1997; Monuki et al.,2001), was not significantly affected(Fig. 6F,F′,G,G′). In Hes1;Hes3;Hes5 cKO mice, the dorsal telencephalic midline was reduced in size, but the cortical neuroepithelium did not expand. Cell death and proliferation were not responsible for the reduction in size of the dorsal telencephalic midline (see Fig. S7G-L in the supplementary material).

At E12.5, in the control mice, the telencephalic midline region was clearly separated into the choroid plexus epithelium and the cortical hem, while Bmp4 and Lmx1a were expressed in both regions(Fig. 6H,I). In Hes1;Hes3;Hes5 cKO mice, the prospective choroid plexus region remained pseudostratified, and the Bmp4 and Lmx1a expression domain became smaller(Fig. 6H′,I′). In the control, Msx1 was expressed at a high level in the choroid plexus epithelium and at a low level in the ventral part of the cortical hem(Fig. 6J), while Msx2was expressed in both the choroid plexus epithelium and the cortical hem(Fig. 6K). In Hes1;Hes3;Hes5 cKO mice, both Msx1 and Msx2 were expressed at very low levels(Fig. 6J′,K′,asterisks). However, the Wnt3a expression domain was not significantly changed between control and Hes1;Hes3;Hes5 cKO mice at this stage(Fig. 6L,L′). Expression of the homeodomain gene Lhx5 in the eminentia thalami, which physically links the telencephalic choroid plexus to the diencephalon(Hébert et al., 2002),was not significantly affected in Hes1;Hes3;Hes5cKO mice, indicating that the diencephalon is not expanded in the absence of Hes genes (Fig. 6M,M′). This finding reveals that inactivation of Hes genes leads to attenuation of Bmp signaling and lack of the choroid plexus epithelium with no expansion of the cortical and diencephalic neuroepithelium.

In Hes1;Hes3;Hes5 cKO mice, Cajal-Retzius cells increased in number in the piriform cortex at E12.5. Furthermore, neurogenesis was accelerated in the dorsal telencephalic midline region of Hes1;Hes3;Hes5 cKO mice at E10.5 and E11.5(Fig. 7B,D, asterisks) compared with the control mice (Fig. 7A,C). We then sought to determine the mechanism for this enhanced Cajal-Retzius cell formation in Hes1;Hes3;Hes5 cKO mice. In the dorsal telencephalic midline (Lmx1a⁺) of wild-type embryos, Ngn1 and Ngn2 were expressed at E10.5 and E11.5 (Fig. 7G-K). Interestingly, at E10.5, Hes1 and Ngn2 were co-expressed by many cells(Fig. 7E, arrowheads), but the expression became mostly segregated at E11.5(Fig. 7F, arrows), suggesting that Hes1⁺Ngn2⁺ cells gradually become either Hes1⁺ or Ngn2⁺ cells during this period. In Hes1;Hes3;Hes5 cKO mice, Ngn1 and Ngn2 expression were highly upregulated at both E10.5 and E11.5(Fig. 7L-P, asterisks,7R,R′,S) compared with the control(Fig. 7G-K,Q,Q′,S). These results suggest that inactivation of Hes genes leads to upregulation of Ngn1 and Ngn2 expression, which contributes to enhanced Cajal-Retzius cell formation.

To further clarify the role of Ngn2 in Cajal-Retzius cell formation, we next examined Ngn2-null mice(Fode et al., 2000). The number of Cajal-Retzius cells (reelin⁺, p73⁺),which are derived from the dorsal telencephalic midline, was reduced in Ngn2-null mice compared with the control mice (see Fig. S8 in the supplementary material), indicating that Ngn2 indeed contributes to Cajal-Retzius cell formation. Nevertheless, there was no significant difference in the number of Cajal-Retzius cells in the piriform cortex (data not shown). It is partly because Cajal-Retzius cells in this region come from other regions in addition to the choroid plexus region(Bielle et al., 2005). Furthermore, the dorsal telencephalic midline region developed normally in Ngn2-null mice (see Fig. S9 in the supplementary material),suggesting that Ngn1 compensates Ngn2 to some extent.

We found that in the absence of Hes genes, Ngn2 expression was upregulated and that Cajal-Retzius cells in the piriform cortex increased in number at the expense of the choroid plexus cell fate. We then examined whether misexpression of Ngn2 in the dorsal telencephalic midline promotes formation of Cajal-Retzius cells at the expense of the choroid plexus. Misexpression of Ngn2 in the dorsal telencephalic midline at E9.5 inhibited the development of the choroid plexus(Fig. 8D′,E′)(n=4). Furthermore, this misexpression of Ngn2 generated more Cajal-Retzius cells (reelin⁺) in the piriform cortex(Fig. 8F,G). These results suggest that misexpression of Ngn2 in the dorsal telencephalic midline at E9.5 promotes formation of Cajal-Retzius cells at the expense of the choroid plexus.

To examine the plasticity of the differentiation competency at a later stage, we electroporated the Ngn2 vector at E11.5 (this procedure should induce the ectopic expression around E12). Misexpression of Ngn2 increased Cajal-Retzius cell formation (reelin⁺) from the cortical hem (Wnt2b⁺, Fig. 8H,I) but did not affect choroid plexus formation (Ttr⁺) at E12.5(Fig. 8J,K). These results suggest that the choroid plexus epithelial region loses competency to produce Cajal-Retzius cells by E12.5.

The above results indicate that Hes-expressing cells and Ngn2-expressing cells are segregated in the dorsal telencephalic midline region around E10.5 to E11.5, and that Hes-expressing cells adopt the choroid plexus fate, whereas Ngn2-expressing cells adopt Cajal-Retzius cell fate. We then sought to determine the mechanism responsible for this segregation. The most likely mechanism is Notch-mediated lateral inhibition: proneural genes such as Ngn2 induce expression of the Notch ligand, leading to activation of the Notch pathway and to the induction of Hes1/Hes5 expression in neighboring cells (Kageyama et al., 2007). We thus examined mice mutant for Rbpj, an essential effector of Notch signaling (Tanigaki and Honjo, 2007). However, because conventional Rbpj-null mice die very early (Oka et al.,1995), we generated Rbpj cKO mice by crossing floxed Rbpj mice (Han et al.,2002) with Emx1-Cre mice. In these cKO mice, although Hes5 expression was downregulated, Hes1 was still expressed(see Fig. S10C,D,K,L in the supplementary material), and Ttrexpression occurred normally (see Fig. S10E,M in the supplementary material). These results indicate that the Notch-Rbpj pathway is not involved in segregation of Hes- and Ngn2-expressing cells.

It has been shown that Cajal-Retzius cells are born at multiple places in the developing telencephalon, such as the cortical hem, the septum and the pallial-subpallial boundary(Takiguchi-Hayashi et al.,2004; Yoshida et al.,2005; Bielle et al.,2005). By lineage-tracing analysis, we found that cells in the prospective choroid plexus region have potential to give rise to Cajal-Retzius cells first, and later differentiate into choroid plexus epithelial cells. Thus, neural (Cajal-Retzius) and non-neural (choroid plexus epithelial) cells are sequentially born in the dorsal telencephalic midline region. We further showed that Hes1⁺Ngn2⁺ cells gradually become either Hes1⁺ or Ngn2⁺ cells, and that inactivation of Hes1,Hes3 and Hes5 upregulated expression of Ngn2,accelerating Cajal-Retzius cell formation at the expense of the choroid plexus(Fig. 9A). In these mutant mice, it is likely that almost all cells in the prospective choroid plexus epithelium adopted Cajal-Retzius cell fate and migrated into the piriform cortex, because the choroid plexus epithelium is completely lacking. However,it is also possible that some cells remain as pseudostratified epithelial cells, and further experiments will be required to resolve this issue. Similarly, overexpression of Ngn2 enhanced formation of Cajal-Retzius cells and inhibited differentiation of the choroid plexus. These results suggest that Hes1⁺Ngn2⁺ cells are bi-potent and become segregated into Hes1-expressing cells that adopt the choroid plexus fate and Ngn2-expressing cells that adopt Cajal-Retzius cell fate, and that Hes and Ngn2 antagonistically regulate the non-neural versus neural fate decision in the dorsal telencephalic midline region(Fig. 9B). However, it is also possible that these two cell types derive from two different types of progenitor cells rather than from bi-potent cells. We were not able to resolve this issue decisively, because it is technically difficult to perform a clonal dissociation culture of the E10.5 dorsal telencephalic midline.

It is surprising that the prospective choroid plexus region initially give rise to Cajal-Retzius cells before differentiating into the choroid plexus epithelium. Because the hem, a well known source of Cajal-Retzius cells, is located next to the prospective choroid plexus region, it is possible that these two regions are not clearly separated at early stages and thus some cells in the boundary region could contribute to Cajal-Retzius cell formation. However, around E10.5 to E11.5, neurogenesis occurs widely in the prospective choroid plexus region and is not restricted to the boundary to the prospective hem region (Fig. 1). Furthermore, Cajal-Retzius cell migration from the prospective choroid plexus region ceases by E12.5 (Fig. 2G), although that from the hem continues even after E13.5(Takiguchi-Hayashi et al.,2004). These data support the notion that these Cajal-Retzius cells derive from the prospective choroid plexus region.

In the dorsal telencephalic midline of wild-type mice, Hes1 and Hes5 expression occurred at high levels until E11.5 but was then downregulated in the choroid plexus epithelium at E12.5. In our Hes1;Hes3;Hes5 cKO mice, Hes1 expression occurred at a lower level at E10.5 and was lost around E11.5. Thus, in Hes1;Hes3;Hes5 cKO mice, Hes1 expression was repressed only 1 or 2 days earlier than in the control. Nevertheless, we found profound defects (loss of the choroid plexus), suggesting that Hes expression around E10.5 to E11.5 has a crucial role in the specification of the choroid plexus. At this stage, Hes-expressing cells and Ngn2-expressing cells seem to be segregated in the dorsal telencephalic midline region, but after this stage, the cell fates seem to be determined and are unchangeable. In accordance with this notion, development of the choroid plexus was severely affected by electroporation of the Ngn2 vector at E9.5 (expression occurs around E10) but not at E11.5(expression occurs around E12). These results suggest that the differentiation competency becomes unchangeable soon after E11.5 in the dorsal telencephalic midline. The mechanism underlying how segregation of choroid plexus epithelial cells and Cajal-Retzius cells is regulated is not known. Lateral inhibition mediated by Notch signaling is a possible mechanism. However, inactivation of Rbpj, an essential mediator of Notch signaling, neither abolishes Hes1 expression nor significantly affects the choroid plexus development, thus suggesting that Notch signaling is not involved in this process.

Although the choroid plexus was completely missing in Hes1;Hes3;Hes5 cKO mice, the boundary between the prospective choroid plexus epithelium and the diencephalon was not affected. Thus, it is likely that none of the cells in the prospective choroid plexus epithelium adopted the diencephalic cell fate. This finding suggests that these prospective choroid plexus epithelial cells do not have the competency to become cell types other than choroid plexus and Cajal-Retzius cells.

Our finding that Hes1 expression is not regulated by the Notch-Rbpj pathway raised another important question: which factors regulate Hes1 expression in the dorsal telencephalic midline? One of the candidates is Bmp signaling, because previous studies have shown that activation of Bmp signaling induces Hes1 expression in cultured cells(Dahlqvist et al., 2003). Additionally, our preliminary study also showed that treatment with Bmp leads to increased Hes1 expression in neural progenitor cultures (I.I.,T.S., T.O. and R.K., unpublished). Furthermore, Bmp genes are expressed at high levels in the dorsal telencephalic midline. Thus, Bmp signaling seems to be important for Hes1 expression in this region. Conversely, Hes genes are required for maintenance of Bmp signaling, because expression of Bmp and of its downstream genes is severely downregulated in Hes1;Hes3;Hes5 cKO mice. Apparently, Cajal-Retzius cells do not express Bmp, so premature differentiation of these cells may lead to loss of Bmp expression. We speculate that Hes genes maintain Bmp-expressing cells by inhibiting Cajal-Retzius cell formation rather than directly activating Bmp expression.

It has been shown that Bmp signaling is required locally for the development of the dorsal telencephalic midline but not for the medial-lateral patterning of the dorsal telencephalon(Hébert et al., 2002). Regions where Bmp signaling is inactive seem to become the neural cells(cortical hem and cortical neuroepithelium), whereas regions with a high Bmp activity become the non-neural cells (the choroid plexus). This effect of Bmp signaling is reminiscent of the epidermal versus neural fate specification of Xenopus. In early Xenopus embryos, Bmp signaling induces naïve ectoderm to adopt the epidermal fate, whereas anti-Bmp factors such as noggin and chordin inhibit Bmp signaling and promote the neural fate specification (Sasai and De Robertis,1997). It is likely that the Bmp-Hes pathway regulates the choroid plexus fate, whereas the Bmp antagonist-Ngn pathway regulates the Cajal-Retzius cell fate (Fig. 9B). A full understanding of this process, however, will require further analysis, including the functional interaction between bHLH and homeodomain factors that are required for the choroid plexus formation.

We thank Shigeyoshi Itohara, François Guillemot, Tasuku Honjo,Phillip Soriano and Yoko Suda for reagents. We also thank Hitoshi Miyachi for help in generating pMsx1-EGFP mice. This work was supported by the Grants-in-aid from the Ministry of Education, Culture, Sports, Science and Technology of Japan, and by the Uehara Memorial Foundation. I.I. was supported by the 21st Century COE Program of the Ministry of Education, Culture, Sports,Science and Technology of Japan and by Research Fellowships of the Japanese Society for the Promotion of Science for Young Scientists.