The developing central nervous system is partitioned into compartments by boundary cells, which have different properties than compartment cells, such as forming neuron-free zones, proliferating more slowly and acting as organizing centers. We now report that in mice the bHLH factor Hes1 is persistently expressed at high levels by boundary cells but at variable levels by non-boundary cells. Expression levels of Hes1 display an inverse correlation to those of the proneural bHLH factor Mash1, suggesting that downregulation of Hes1 leads to upregulation of Mash1 in non-boundary regions,whereas persistent and high Hes1 expression constitutively represses Mash1 in boundary regions. In agreement with this notion, in the absence of Hes1 and its related genes Hes3 and Hes5, proneural bHLH genes are ectopically expressed in boundaries, resulting in ectopic neurogenesis and disruption of the organizing centers. Conversely, persistent Hes1 expression in neural progenitors prepared from compartment regions blocks neurogenesis and reduces cell proliferation rates. These results indicate that the mode of Hes1 expression is different between boundary and non-boundary cells, and that persistent and high levels of Hes1 expression constitutively repress proneural bHLH gene expression and reduce cell proliferation rates,thereby forming boundaries that act as the organizing centers.
The developing central nervous system (CNS) is partitioned into many compartments, which form units of distinct populations of neurons(Lumsden and Krumlauf, 1996; Kiecker and Lumsden, 2005; Irvine and Rauskolb, 2001; Simeone, 2002). These compartments are demarcated by boundary cells, which regulate neuronal specification in the neighboring compartments. For example, the zona limitans intrathalamica (ZLI), a boundary between the prethalamus and the thalamus,regulates specification of prethalamic and thalamic neurons by secreting Sonic hedgehog (Shh) (Kiecker and Lumsden,2004). The isthmus, a boundary between the midbrain and the hindbrain, regulates their development by secreting Wnt1 and Fgf8(Joyner et al., 2000; Mason et al., 2000; Liu and Joyner, 2001; Rhinn and Brand, 2001; Wurst and Bally-Cuif, 2001). The roof plate and the floor plate, located at the dorsal and ventral regions of the neural tube, respectively, also have similar features. They separate the right and left sides of the neural tube and regulate the dorsoventral patterning of the nervous system by expressing Wnt and BMP family members dorsally and Shh ventrally (Lee and Jessell, 1999; Placzek and Briscoe, 2005). Thus, these structures not only separate the CNS into many units but also regulate neuronal specification of each unit by acting as organizing centers. Genes regulating boundary formation have been intensively analyzed: the isthmus is formed at the interface between Otx2 and Gbx2 expression domains(Broccoll et al., 1999; Millet et al., 1999) whereas the ZLI is formed between Six3 and Irx3 expression domains,depending on Shh (Kobayashi et al.,2002; Braun et al.,2003; Zeltser,2005).
Boundaries are formed by specialized neuroepithelial or radial glial cells,which have properties that are different from those in compartments. These boundary cells are embedded in a specialized extracellular matrix and proliferate more slowly than compartment cells. Furthermore, neurogenesis is delayed or does not occur in boundaries(Lumsden and Keynes, 1989; Guthrie et al., 1991; Kahane and Kalcheim, 1998; Trokovic et al., 2005). Such boundary cell-specific properties are well documented, but little is known about how these properties are regulated. Recent studies revealed that, in zebrafish, activation of Notch signaling regulates rhombomere boundary formation by segregating boundary cells from rhombomeres and inhibiting neurogenesis (Cheng et al.,2004). Although roles of Notch signaling in boundary formation remains to be determined in mouse, we previously found that the repressor-type basic helix-loop-helix (bHLH) gene Hes1, one of the Notch effectors(Jarriault et al., 1995; Kageyama and Nakanishi, 1997; Ohtsuka et al., 1999), is involved in formation of the isthmus(Hirata et al., 2001). In the absence of Hes1 and its related gene Hes3, the isthmic cells are differentiated into neurons and lose their organizer activities prematurely, indicating that Hes1 and Hes3 contribute to maintenance of a neuron-free zone and the organizing center at the isthmus(Hirata et al., 2001). Similarly, the Hes-related genes her5 and him(her11) in zebrafish and XHR1 in Xenopus regulate formation of a neuron-free zone at the isthmus(Geling et al., 2004; Ninkovic et al., 2005; Takada et al., 2005). However, Hes1 is also expressed in compartments, where neurogenesis occurs actively, and it remains to be determined why Hes1-expressing cells in compartments undergo neurogenesis whereas those in boundaries do not. In addition, it is not known whether Hes genes play a role in formation of boundaries other than in the isthmus.
We have found that the mode of Hes1 expression is different in boundary and non-boundary regions in mice. In non-boundary regions, Hes1 is expressed at variable levels: some cells express Hes1 at high or intermediate levels whereas others express no Hes1. Downregulation of Hes1 expression leads to neurogenesis by derepression of proneural bHLH factors such as Mash1 (Ascl1 -Mouse Genome Informatics). By contrast, boundary cells persistently express Hes1 at high levels and are negative for proneural bHLH gene expression. In the absence of Hes genes, however, proneural bHLH genes are ectopically expressed in boundaries, resulting in ectopic neurogenesis and impairment of the organizer activities. Conversely, persistent Hes1 expression in compartmental neural progenitors leads to blockade of neurogenesis and reduction of proliferation rates, properties that are reminiscent of boundary cells. Thus, persistent and high levels of Hes1 expression regulate boundary formation in the developing CNS.
MATERIALS AND METHODS
In situ hybridization
In situ hybridization with digoxigenin-labeled antisense RNA probes was performed as described previously (Tomita et al., 2000).
Pregnant mice were injected intraperitoneally with 120 μg BrdU/g body weight. After 30 minutes, sections of embryos were prepared for immunohistochemistry, as described previously(Ohsawa et al., 2005).
Scanning electron microscopic analysis (SEM)
SEM analysis was done, as described previously(Hatakeyama et al., 2004).
Anti-Hes1 antibody was generated, as previously described(Hatakeyama et al., 2006). Other antibodies used are as follows: anti-BrdU (Becton Dickinson), anti-Shh(Developmental Studies Hybridoma Bank), anti-Ki67 (PharMingen), anti-Mash1(PharMingen), anti-phosphorylated histone H3 (Sigma), anti-GFP (Molecular Probes), anti-nestin (PharMingen), anti-β-tubulin III (TuJ1) (Babco),anti-GFAP (Sigma), and anti-cyclin D1 (Santa Cruz). Biotinylated donkey anti-guinea pig IgG (Jackson), fluorescein avidin D (Vector), Fluorolink Cy3-conjugated goat anti-mouse IgG (Amersham Pharmacia), Fluorolink Cy2-conjugated goat anti-rabbit IgG (Amersham Pharmacia) and FITC-conjugated goat anti-mouse IgG (Vector) were used as secondary antibodies.
Immunohistochemistry and immunocytochemistry
For immunohistochemistry, embryos were fixed in cold 4% paraformaldehyde(PFA) in PBS overnight, rinsed with cold PBS three times and equilibrated in cold 25% sucrose in PBS overnight. Embryos were embedded in OCT compound,frozen at -80°C and sectioned at 15 μm thickness. Sections were washed three times in PBS and preincubated in blocking buffer (1.5% normal goat serum and 0.1% Triton X-100 in PBS) for 1 hour. Sections for Hes1 immunostaining were washed three times in PBS, autoclaved in 0.01 M citrate buffer pH 6.0(105°C, 10 minutes), and washed three times again in PBS before preincubation.
Sections were next incubated in blocking buffer with primary antibodies at 4°C overnight or for 2 days (for Hes1). After being washed in PBS three times, sections were incubated with the secondary antibody in blocking buffer at room temperature for 1 hour and then washed in PBS three times. For Hes1,the tertiary antibody reaction was performed. Samples were then treated with propidium iodide (PI) or DAPI and were mounted with Fluoromount G(SouthernBiotech).
For immunocytochemisty, cells on chamber slides were washed in ice-cold PBS three times, and fixed in 4% PFA at room temperature for 15 minutes. Samples were then washed in PBS three times and preincubated with blocking buffer at room temperature for 20 minutes. Antibody reactions were done as described above. TUNEL was performed using a standard detection kit (Roche).
Measurement of relative Hes1 protein levels
To measure a relative Hes1 protein level on images, each nucleus was circled according to the nuclear staining, and signal intensity of Hes1 immunostaining and nuclear staining in each circle was measured using the program Image-Pro Plus (Ver. 22.214.171.124; MediCybernetics, Silver Spring, MD,USA). The background of a non-Hes1-expressing region was subtracted from each Hes1 signal. This Hes1-specific signal value was divided by the nuclear signal value to normalize the staining efficiency and categorized as (±) for a value of≤ 0.3, (+) for >0.3, (++) for >0.8 and (+++) for >1.3. During M phase, some cells express Hes1 protein in the cytoplasm, but they were excluded from the analysis because cytoplasmic Hes1 does not function in transcriptional repression. As a control experiment, we successfully detected the relative difference in Hes1 protein levels between non-Hes1-expressing cells (±) and transfected Hes1-expressing cells (+ to +++).
Neural progenitor cell (NPC) culture
NPC and retrovirus preparation was done, as described previously(Hirata et al., 2000; Ohtsuka et al., 2001; Hatakeyama and Kageyama,2002). 10 μl of the viral solution was added to each well of poly-l-lysine-coated 8-well plates. After 4 hours, the virus-containing medium was discarded, and the cells were cultured in fresh NPC medium. We usually obtained 20-30 colonies in each well of the eight-well plates.
Persistent and high levels of Hes1 expression and low proliferation rates in boundaries
Hes1 mRNA is expressed at high levels by the neuroepithelial cells of boundaries (Fig. 1A,E,M,P)(Hirata et al., 2001; Hatakeyama et al., 2004; Ohtsuka et al., 2006). We previously found that Hes1 mRNA is increased when Hes1 protein levels are low because Hes1 protein represses its own transcription by negative feedback (Takebayashi et al.,1994; Hirata et al.,2002). To determine the levels of Hes1 protein expression in boundaries, we performed immunohistochemistry. Similar to the mRNA expression,Hes1 protein is also expressed at high levels in the ZLI, the isthmus, the interrhombomeric boundaries, the roof plate and the floor plate (Figs 1, 2 and 3), suggesting that the negative feedback does not work in boundary cells.
In the isthmus, Hes1 protein is expressed at a high level by neuroepithelial cells (Fig. 1B,C), which also express Wnt1(Fig. 1D,E, arrows). These Hes1-expressing cells do not display efficient BrdU uptake(Fig. 1F,G, arrows) (see also Trokovic et al., 2005),compared to the adjacent compartment cells. In the ZLI, Hes1 protein is highly expressed (Fig. 1H,L,asterisks), and the cells that express Hes1 at high levels also express Shh(Fig. 1I-K, brackets). Hes1-expressing cells in the ZLI do not display efficient BrdU uptake(Fig. 1L-O, asterisks),compared to the adjacent compartment cells. Thus, in the isthmus and the ZLI,Hes1-positive cells express Wnt1/Shh and do not actively proliferate. Hes1 protein is also highly expressed in the interrhombomeric boundaries(Fig. 1Q-S). These Hes1-expressing cells are mostly negative for phosphorylated histone H3 (pH3),an M phase-specific marker (Fig. 1T-V, arrows), whereas many of the adjacent rhombomeric cells are positive for pH3 (Fig. 1U,V),indicating that Hes1-expressing cells do not actively proliferate in the interrhombomeric boundaries.
In the spinal cord, Hes1 is expressed in the roof plate and the floor plate as well as in the regions in between along the DV axis(Fig. 2A). In the roof plate and the floor plate, all cells express Hes1 protein at a high level(Fig. 2H,N). The roof plate and the floor plate cells are mostly negative for Ki67, a marker for proliferating cells (Fig. 2A-C, arrowheads),and for BrdU uptake (Fig. 2D-G)whereas many of non-boundary cells are positive for Ki67(Fig. 2A-C, bracket) and BrdU uptake (Fig. 2D,E). In addition, pH3 is expressed by non-boundary cells but not by the roof plate and the floor plate cells (data not shown). Quantification of the Hes1 protein levels relative to the nuclear staining with propidium iodide (PI) revealed that almost all roof plate and floor plate cells express Hes1 protein at high levels (Fig. 2H-J,N-Q). High levels of Hes1 protein expression in these cells were confirmed in all sections of different embryos at E10.5 (data not shown). These results indicate that boundary cells express Hes1 protein at high levels, and that these Hes1-expressing cells do not efficiently proliferate, compared to non-boundary cells. We also examined other stages. As at E10.5, the roof plate and the floor plate cells express Hes1 protein at high levels at E9.5(Fig. 3A-H), E11.5(Fig. 3I-P) and E13.5(Fig. 3Q-X), indicating that boundary cells persistently express Hes1 protein at high levels.
In non-boundary regions, Hes1 protein is also expressed in the nuclei of neuroepithelial and radial glial cells. In Hes1-positive cells non-boundary regions of the spinal cord, some cells express Hes1 protein at a high level whereas others express it at an intermediate level(Fig. 2K-M). Cells that do not express Hes1 protein are also present in the same regions(Fig. 2K-M). Similarly, Hes1 protein levels are variable in non-boundary regions just dorsal to the floor plate (Fig. 2N-P, bracket) and in other compartments (data not shown, see Fig. 1C, in compartments near the isthmus). Quantification analysis showed that Hes1 protein levels are variable in non-boundary cells (Fig. 2Q). These results indicate that the mode of Hes1 expression is different in boundary and non-boundary cells.
Inverse correlation between the Hes1 and Mash1 expression levels
Whereas Hes1, a transcriptional repressor for proneural bHLH gene expression, is expressed in both boundaries and non-boundaries, proneural bHLH genes are expressed only in non-boundary regions. In the spinal cord, the proneural bHLH factor Mash1 is expressed in the same regions as Hes1 at E10.5(Fig. 4A-D). To reveal the relationship between Hes1 and Mash1 expression, we performed double immunohistochemistry. In non-boundary regions, cells expressing Hes1 at a high level do not express Mash1 (Fig. 4E-G, green cells in H) whereas those expressing Hes1 at an intermediate level also express Mash1 at an intermediate level(Fig. 4E-G, striped cells in H). In contrast, those expressing no Hes1 protein express Mash1 at a high level (Fig. 4E-G, a red cell in H). These results indicate that there is an inverse correlation between Hes1 and Mash1 expression in non-boundary cells. Since Hes1 can repress Mash1 expression by binding to the Mash1 promoter(Chen et al., 1997; Castella et al., 1999), it is probable that decreases in Hes1 protein levels lead to de-repression of Mash1 in non-boundary cells.
In boundaries, other proneural bHLH genes such as neurogenin 2(Ngn2) and Math1 (Atoh1 - Mouse Genome Informatics)are not expressed either, although they are highly expressed in non-boundary regions (see Fig. 6). These results raise the possibility that persistent and high levels of Hes1 expression constitutively repress proneural bHLH gene expression in boundaries, thereby making neuron-free zones.
Defects in structures and organizer activities of boundaries in Hes-null mice
To investigate the requirement of persistent Hes1 expression for boundary formation, we next performed loss-of-function analyses. The ZLI normally expresses Hes1 only but, because Hes5 is ectopically upregulated in the ZLI of Hes1-null mice(Hatakeyama et al., 2004), we examined Hes1;Hes5 double-null mice. Although these double-null mice exhibit severe structural defects of the nervous system(Hatakeyama et al., 2004), the forebrain and midbrain are relatively intact at E10.5. However, the ZLI is lacking in the double-null mice (Fig. 5C,D, asterisks), whereas it is clearly observed in the wild type(Fig. 5A,B, arrows). Furthermore, Shh is expressed in the ZLI of the wild-type around E10.5 (Fig. 5I, arrowhead) but it is not detectable in Hes1;Hes5 double-null mice(Fig. 5J, asterisk). In the wild type, the adjacent compartments express Lfng but the ZLI does not (Fig. 5K, arrowhead), and misexpression of Lfng has been shown to repress ZLI formation(Zeltser et al., 2001). Similarly, in the wild type the adjacent compartments express Pax6,although the ZLI does not do so (Fig. 5M, arrowhead). In contrast, in Hes1;Hes5 double-null mice, the Lfng- and Pax6-negative domain is missing(Fig. 5L,N, asterisks). In addition, the bHLH-PAS genes Sim1 and Sim2 are expressed in the ZLI of the wild type (Ema et al.,1996), whereas this expression domain is lacking in Hes1;Hes5 double-null mice (data not shown). These results indicate that the ZLI is not formed in the absence of Hes1 and Hes5.
Hes1 and Hes3 are expressed in the isthmus, and in the absence of Hes1 and Hes3, the isthmus is not properly maintained and prematurely loses Fgf8 expression(Hirata et al., 2001). Similarly, in the Hes1;Hes5 double-null mice, the isthmus is only partially formed (Fig. 5C,D,arrowheads), suggesting that Hes3 may partially compensate for this defect. We further examined the isthmus of Hes3;Hes5 double-null, Hes1;Hes5 double-null and Hes1;Hes3;Hes5 triple-null mice. Hes3;Hes5 double-null mice were found to be normal and were used as a control. In the isthmus of Hes3;Hes5 double-null mice, Wnt1,Fgf8 and En1 are expressed normally(Fig. 5O,R,U, arrowheads). In the isthmus of Hes1;Hes5 double-null mice, however, expression of Wnt1, Fgf8 and En1 is reduced(Fig. 5P,S,V, arrowheads). In the isthmus of Hes1;Hes3;Hes5 triple-null mice, the expression of Wnt1, Fgf8 and En1 is more severely reduced, although they are not completely lost (Fig. 5Q,T,W, asterisks). We observed these defects at both E8.5 (data not shown) and E9.5 (Fig. 5). These results indicate that in the absence of Hes1, Hes3 and Hes5, the isthmus is not properly formed and loses much of the isthmus-specific gene expression. This defect occurs even earlier than in Hes1;Hes3 double-null mice, which display loss of Wnt1 and Fgf8 expression around E10.5(Hirata et al., 2001). In the wild-type and Hes3;Hes5 double-null embryos the expression domain of Otx2 is rostral whereas that of Gbx2 is caudal to the isthmus (Fig. 5X,Aa), but these domains are not clearly separated in Hes1;Hes5 double-null mice at E9.5 (Fig. 5Y,Bb, asterisks)and in Hes1;Hes3;Hes5 triple-null mice at both E8.5(Fig. 5Z,Cc, insets) and E9.5(Fig. 5Z,Cc), indicating that the compartmentalization is impaired in the absence of Hes genes.
In the wild-type hindbrain, the interrhombomeric boundaries are formed at E9.5 and E10.5 (Fig. 5E,F,arrowheads). They are not significantly affected at E9.5(Fig. 5G, bracket) but are somewhat ambiguous at E10.5 (Fig. 5H, bracket) in Hes1;Hes5 double-null hindbrain. Thus,the interrhombomeric boundaries are not properly maintained in the absence of Hes genes.
In Hes1;Hes3;Hes5 triple-mutant mice, Wnt1 expression in the roof plate is still maintained in the midbrain-hindbrain regions(Fig. 5Q) but is mostly missing in the spinal cord (Fig. 6U,asterisk). Similarly, Shh expression in the floor plate is downregulated in Hes1;Hes5 double-null mice (Fig. 6Z,Z′, arrowheads) and is more severely reduced or lost in Hes1;Hes3;Hes5 triple-mutant mice(Fig. 6Aa,Aa′,asterisks). Thus, in the absence of Hes genes, the organizer activities in these structures are not properly maintained.
Ectopic neurogenesis in boundaries of Hes-null mice
We next examined the expression of the proneural bHLH genes Mash1,Ngn2 and Math1 in boundaries of Hes-mutant mice. Mash1and Ngn2 are widely expressed in compartments but not in the ZLI or the isthmus of the wild-type and Hes3;Hes5 double-null embryos(Fig. 6A,D, arrowheads, and data not shown). In Hes1;Hes5 double-null embryos, however, the Mash1-negative domains in the ZLI and the isthmus become narrower or lost in the ventral region (Fig. 6B, asterisk) while the Ngn2 expression domain extends rostrally, resulting in the loss of the Ngn2-negative domain in the ZLI (Fig. 6E, asterisk). Furthermore, in Hes1;Hes3;Hes5 triple-null embryos, the Mash1- and Ngn2-negative domain in the ZLI is missing,resulting in continuous expression of Mash1 and Ngn2 in the ventral part of the forebrain (Fig. 6C,F, asterisks). The Mash1- and Ngn2-negative domain in the isthmus is also missing in Hes1;Hes3;Hes5 triple-null embryos (Fig. 6C,F,arrowheads). Math1 is expressed in the dorsal part of the neural tube caudal to the isthmus in wild-type (Akazawa et al., 1995; Vernay et al.,2005) and Hes3;Hes5 double-null(Fig. 6G) embryos. However, in Hes1;Hes5 double-null mice, Math1 expression is extended rostrally into the midbrain without interruption by the isthmus(Fig. 6H, arrow), and in Hes1;Hes3;Hes5 triple-null mice this ectopic expression is further upregulated (Fig. 6I, arrow). Similarly, in the hindbrain of the wild-type and Hes3;Hes5double-null embryos, Ngn2 and Mash1 are not expressed at the interrhombomeric boundaries (Fig. 6M, arrowheads, and data not shown) whereas these proneural bHLH gene-negative regions at the interrhombomeric boundaries are somewhat obscure in Hes1;Hes5 double-null mice(Fig. 6N, arrowheads) and mostly lacking in Hes1;Hes3;Hes5 triple-null mice(Fig. 6O, bracket). Thus, in Hes-null mutants, the regions negative for proneural bHLH gene expression are missing in the ZLI, isthmus and interrhombomeric boundaries.
In the spinal cord, Math1 is expressed in the dorsal region but not in the roof plate of the wild-type, Hes3;Hes5 double-null and Hes1;Hes5 double-null embryos(Fig. 6P,Q, arrowheads),although it is ectopically expressed in some regions of the roof plate of Hes1;Hes5 double-null embryos (data not shown). In contrast, in Hes1;Hes3;Hes5 triple-null embryos, the Math1-negative region of the roof plate is missing (Fig. 6R, asterisk). Ngn2 is expressed in the ventral part of the neural tube but not in the floor plate of the wild-type, Hes3;Hes5 double-null and Hes1;Hes5 double-null embryos(Fig. 6V,W, arrowheads). However, in some Hes1;Hes3;Hes5 triple-null mice, the Ngn2-negative region of the floor plate is missing(Fig. 6X, asterisk). These results indicate that in the absence of Hes genes, proneural bHLH genes are ectopically expressed in the roof plate and the floor plate.
We next examined expression of delta-like 1 (Dll1), one of the earliest markers for post-mitotic neurons. In the wild-type and Hes3;Hes5 double-null embryos, Dll1 is not expressed in boundaries (Fig. 6J,arrowheads, and data not shown). In contrast, in Hes1;Hes5 double-and Hes1;Hes3;Hes5 triple-null embryos, such Dll1-negative domains are missing (Fig. 6K,L,asterisks). Dll1 expression is also enhanced in all compartments of Hes1;Hes5 double- and Hes1;Hes3;Hes5 triple-null embryos(Fig. 6K,L). In addition, we found that neurons (TuJ1-positive cells) are ectopically differentiated in the roof plate and the floor plate of Hes1;Hes3;Hes5 triple-null embryos(data not shown). Because cell death is not increased in Hes-null mice at these stages (data not shown), it is probable that the cells that should normally form boundaries ectopically express proneural bHLH genes and are differentiated into neurons in the absence of Hes genes. These results support the conclusion that persistent and high levels of Hes1 expression are essential in vivo for the formation of neuron-free zones at boundaries.
Persistent Hes1 expression reduces cell proliferation and maintains undifferentiated cells
We next examined the effects of persistent Hes1 expression on neural progenitors, which were prepared from telencephalic compartments of E11.5 mouse embryos. Neural progenitors were infected with the retrovirus CLIG, which directs EGFP expression (Hojo et al., 2000), and CLIG-Hes1, which directs persistent Hes1 expression in addition to EGFP expression(Fig. 7A). Because retrovirus is infectious only to dividing cells, we were able to monitor the fate of the virus-infected neural progenitors. Each neural progenitor infected with CLIG or CLIG-Hes1 proliferated and formed a clone consisting of EGFP-positive progeny. Cells infected with CLIG proliferated efficiently and formed large clones (Fig. 7B-D,H). The average size of each clone was about 135±12 cells at day 5. In contrast, cells infected with CLIG-Hes1 proliferated less efficiently and formed smaller clones (Fig. 7E-H). The average size of each clone was about 22±2 cells at day 5. We also performed TUNEL assays to determine whether cell death is responsible for the reduction in clonal sizes. Although there was some tendency to increased cell death when infected with CLIG-Hes1 compared to CLIG, the effect was not sufficient for the reduction of the growth rate of cells infected with CLIG-Hes1 (Fig. 7I), suggesting that persistent Hes1 expression reduces the cell proliferation rate without significant cell death. In agreement with this notion, ratios of cells positive for cyclin D1, a G1-specific marker(Momota and Holland, 2005),were increased when CLIG-Hes1 was infected, although the Ki67-positive ratios were not significantly changed (Fig. 7J). Thus, it is likely that persistent Hes1 expression prolongs the G1 phase, thereby reducing cell proliferation.
Cell type composition was also examined. Clones infected with CLIG consisted of progenitors (nestin-positive cells), neurons (TuJ1-positive cells; Fig. 7K-N) and astrocytes (GFAP-positive cells; Fig. 7S-V). In contrast, cells infected with CLIG-Hes1 were mostly negative for TuJ1 (Fig. 7O-R,closed arrowheads) and GFAP (Fig. 7W-Z, arrowheads) even after 5 days in culture. These results indicate that forced Hes1 expression endows compartment cells in culture with properties that are reminiscent of boundary cells: reduction of cell proliferation and blockade of cell differentiation.
Persistent and high levels of Hes1 expression are required for formation of neuron-free zones and organizing centers in boundaries
Here, we found that Hes genes regulate boundary formation in the developing nervous system. Although Hes1 is expressed in both boundary and non-boundary regions, the mode of Hes1 expression is different between the two structures. Wnt1-expressing cells and Shh-expressing cells persistently express Hes1 at high levels in the isthmus and ZLI, respectively. Similarly, all cells in the roof plate and the floor plate persistently express Hes1 protein at high levels. In contrast, levels of Hes1 protein expression are variable in non-boundary regions. Expression levels of the proneural bHLH factor Mash1 display an inverse correlation to levels of Hes1, suggesting that downregulation of Hes1 leads to upregulation of Mash1 in non-boundary regions while persistent and high levels of Hes1 expression constitutively repress Mash1 and generate neuron-free zones in boundaries. Consistent with this notion, inactivation of Hes genes ectopically upregulates proneural bHLH gene expression and leads to ectopic neurogenesis not only in non-boundary but also in boundary regions.
In zebrafish, it has been shown that sustained activation of Notch signaling leads to interrhombomeric boundary formation: an intracellular form of Notch (ICN) and a dominant-active form of Su(H), both of which constitutively activate Notch signaling, segregate cells from non-boundaries to boundaries whereas a dominant-negative form of Su(H), which inhibits Notch signaling, reverses this cell segregation(Cheng et al., 2004). Because the complex of ICN and Su(H) upregulates Hes1 expression(Kageyama and Nakanishi,1997), it is probable that sustained activation of Notch signaling maintains persistent and high levels of Hes1 in boundaries.
Although Hes1 is expressed in most boundaries, there are no significant boundary defects in Hes1-null embryos(Hirata et al., 2001). Hes3 is highly expressed and compensates for Hes1 deficiency in the isthmus. Hes5 is not expressed in boundaries of wild-type embryos, but it is ectopically expressed in boundaries of Hes1-null embryos (Hatakeyama et al.,2004). Thus, all three Hes genes are able to regulate boundary formation, in addition to non-boundary development, although it remains to be determined whether or not the mode of Hes3 and Hes5 expression is different between boundary and non-boundary regions, like Hes1.
In addition to ectopic neurogenesis, the organizer activities are severely impaired in the absence of Hes genes. In Hes1;Hes3;Hes5 triple-null mice, the isthmus is not properly formed, resulting in severe downregulation of Wnt1 and Fgf8 expression. Likewise, in Hes1;Hes5 double-null mice,neither the ZLI structure nor the Shh expression occurs. Furthermore, Wnt1 and Shh expression is reduced or lost in the roof plate and the floor plate, respectively, of Hes1;Hes3;Hes5 triple-null mice. These results indicate that persistent and high levels of Hes1 expression lead to maintenance of neuroepithelial cells of boundaries that act as the organizing centers.
Persistent versus variable Hes1 expression in the developing nervous system
We found that both Hes1 mRNA and Hes1 protein are highly expressed in boundaries. This expression mode is unusual because in most cells, when Hes1 protein is highly expressed, Hes1 mRNA is downregulated by negative feedback (Hirata et al.,2002). There are multiple Hes1-binding sites in the Hes1promoter, and Hes1 protein, a transcriptional repressor(Sasai et al., 1992), inhibits its own transcription by directly binding to its own promoter(Takebayashi et al., 1994; Hirata et al., 2002). It appears that this negative feedback does not work in boundary cells, unlike most other cells. Hes1 protein should be functional, because inactivation of Hes genes leads to ectopic expression of proneural bHLH genes in boundaries. Thus, it remains to be determined why Hes1 represses proneural bHLH gene expression but not its own in boundaries. It has been shown that the Hes1-binding site of the Hes1 promoter is the N box (CACNAG)(Takebayashi et al., 1994)whereas that of the Mash1 promoter is the class C site (CACGCG)(Chen et al., 1997). Thus, one possible mechanism is that the N box-binding activity of Hes1 is modulated by post-translational modification or by heterodimer formation in boundaries. Another possibility is that a boundary-specific factor binds to the N box sites of the Hes1 promoter to prevent Hes1 negative feedback.
In non-boundary regions, Hes1 expression levels are variable, but it is not yet known how Hes1 expression changes during neuronal differentiation. There could be at least two ways of changing the Hes1 expression levels during transition from Hes1-positive undifferentiated cells to Hes1-negative post-mitotic neurons. One is that Hes1 expression might be gradually downregulated and finally disappear while Mash1 expression is gradually upregulated. The other possible way is that Hes1 expression oscillates during transition from undifferentiated cells to neurons, as we previously found that Hes1 expression oscillates with a periodicity of about 2 hours in many cell types (Hirata et al., 2002; Masamizu et al., 2006). Activation of the Hes1 promoter increases Hes1 protein, which in return represses its own transcription by directly binding to its own promoter(Takebayashi et al., 1994). When the transcription is repressed by this negative feedback, Hes1 protein soon disappears because it is rapidly degraded by the ubiquitin-proteasome pathway. Disappearance of Hes1 protein then allows the next round of transcription. In this way, Hes1 autonomously starts oscillatory expression by negative feedback (Hirata et al.,2002). Since this oscillation is observed in many cell types, Hes1 expression could oscillate during transition from undifferentiated dividing cells to post-mitotic neurons. Further studies will be required to determine which way is the case for changes in Hes1 expression. As discussed below, Hes1 oscillation could be required for efficient growth of neural progenitors.
Persistent and high Hes1 expression reduces cell proliferation rate
We found that persistent and high Hes1 expression not only blocks neurogenesis but also reduces cell proliferation rate. It has been reported that persistent Hes1 expression also inhibits both differentiation and proliferation of PC12 pheochromocytoma cells(Castella et al., 2000). In the latter cells, overexpression of Hes1 represses transcription of the cyclin-dependent kinase (CDK) inhibitor p21. Hes1 also has been shown to repress the expression of the CDK inhibitors p21 and p27 in other cells(Castella et al., 2000; Kabos et al., 2002; Murata et al., 2005). However,because p21 and p27 usually retard G1 progression, the relationship between repression of p21 and p27 expression and cell cycle arrest is not clear. It was shown that both p21 and p27 also promote the assembly of cyclin D1-CDK4 complexes and the nuclear import of cyclin D, thereby inducing cell cycle progression (LaBaer et al.,1997; Cheng et al.,1999). Thus, both p21 and p27 are positive and negative regulators of G1-phase progression, and persistent repression of p21 and p27 by Hes1 could lead to reduction in cell proliferation rates.
It has been also reported that persistent Hes1 expression downregulates proliferating cell nuclear antigen (PCNA), an essential DNA replication factor, and leads to cell cycle arrest(Castella et al., 2000; Ström et al., 2000). In addition, Hes1 has been shown to repress expression of E2F, a transcription factor required for transition from G1 to S phase(Hartman et al., 2004). Hes1 forms a heterodimer complex with the Hes-related bHLH factor Hey/Hesr/Herp and represses transcription of E2F by directly binding to its promoter. Thus,persistent Hes1 expression blocks cell proliferation by repressing several genes that are essential for cell cycle progression. However, loss of Hes1 also inhibits proliferation of neural progenitors by upregulation of CDK inhibitors and proneural bHLH genes, inducing premature neurogenesis(Kabos et al., 2002; Hatakeyama et al., 2004). Thus, Hes1 is definitely required for maintenance of neural progenitors,raising the possibility that oscillatory rather than persistent Hes1 expression is required for efficient growth of neural progenitors.
We thank Dr Laure Bally-Cuif for critical reading of the manuscript, and Drs Makio Fujioka, Hiroshi Kiyonari and Kazuki Nakao for discussion and technical assistance. This work was supported by research grants from the Ministry of Education, Culture, Sports, Science and Technology of Japan, and from Japan Society for the Promotion of Science. J.H.B. was supported by the 21st Century COE Program of the Ministry of Education, Culture, Sports,Science and Technology of Japan. J.H. was supported by Research Fellowships of the Japan Society for the Promotion of Science for Young Scientists.