Development 127, 2515-2522 (2000) 2515 Printed in Great Britain © The Company of Biologists Limited 2000 DEV4317

Glial cell fate specification modulated by the bHLH gene Hes5 in mouse retina

Masato Hojo1,2, Toshiyuki Ohtsuka1, Nobuo Hashimoto2, Gérald Gradwohl3, François Guillemot3 and Ryoichiro Kageyama1,* 1Institute for Virus Research, Kyoto University and 2Department of Neurosurgery, Kyoto University Graduate School of Medicine, Sakyo-ku, Kyoto 606-8507, Japan 3Institut de Génétique et de Biologie Moléculaire et Cellulaire, CNRS/INSERM/Université Louis Pasteur/Collège de France, 67404 Illkirch, CU de Strasbourg, France *Author for correspondence (e-mail: [email protected])

Accepted 29 March; published on WWW 23 May 2000

SUMMARY Neurons and glial cells differentiate from common retrovirus significantly increased the population of glial precursors. Whereas the gene glial cells missing (gcm) cells at the expense of neurons. Conversely, Hes5-deficient determines the glial fate in Drosophila, current data about retina showed 30-40% decrease of Müller glial cell number the expression patterns suggest that, in mammals, gcm without affecting cell survival. These results indicate that homologues are unlikely to regulate gliogenesis. Here, we Hes5 modulates glial cell fate specification in mouse retina. found that, in mouse retina, the bHLH gene Hes5 was specifically expressed by differentiating Müller glial cells Key words: bHLH, Glial specification, Hes5, Müller glia, Retina, and that misexpression of Hes5 with recombinant Mouse, Cell fate

INTRODUCTION homologues have been characterized (Gcm1/Gcma and Gcm2/Gcmb) (Akiyama et al., 1996; Kim et al., 1998; Retina provides a powerful model system to investigate the Kanemura et al., 1999; Reifegerste et al., 1999) and it was mechanisms of cell fate decision in mammalian central nervous shown that one of them, Gcm1, can substitute functionally for system. From common retinal precursors, six types of neurons Drosophila gcm by transforming neurons into glia in and one type of glial cells (Müller glia) differentiate in the Drosophila embryos (Kim et al., 1998; Reifegerste et al., 1999), overlapping but precise time course (Cepko et al., 1996). These suggesting that mammalian gcm homologues may have a seven types of cells constitute three nuclear layers in the retina: conserved gliogenic function, at least, in Drosophila embryos. the outer nuclear layer (ONL), which contains rod and cone However, neither of the mammalian gcm homologues show the photoreceptors, the inner nuclear layer (INL), which contains expression pattern that may account for generation of glial cells bipolar, horizontal and amacrine interneurons as well as Müller in mammalian nervous system and therefore these genes are glia, and the ganglion cell layer, which contains ganglion cells unlikely to regulate glial fate decision in mammals (Kim et al., (projection neurons). Although the kinetic pattern of generation 1998). of these cell types has been well characterized, the molecular Here, we found that the bHLH gene Hes5 (Akazawa et al., mechanism for glial cell fate determination is not well 1992), one of Notch effectors (Kageyama and Nakanishi, 1997; understood. It was previously shown that activation of EGF Ohtsuka et al., 1999), is specifically expressed by differentiat- receptor increases the population of Müller glial cells (Lillien, ing glial cells in mouse retina and that misexpression of Hes5 1995), but the intracellular mechanism for gliogenesis still directs neural precursor cells toward the glial fate at the expense remains to be determined. Recent data demonstrated that of the neuronal fate. Furthermore, Müller glial cell number was transcription factors of the bHLH and homeobox classes reduced in Hes5-deficient retina. These results demonstrate positively or negatively regulate differentiation of many retinal that, in mouse retina, Hes5 modulates the glial cell fate cells, but so far no genes that determine the glial fate are known specification. in the retina (Tomita et al., 1996a,b; Burmeister et al., 1996; Chen et al., 1997; Freund et al., 1997; Furukawa et al., 1997a,b; Kanekar et al., 1997; Mathers et al., 1997; Brown et al., 1998; MATERIALS AND METHODS Jean et al., 1998; Morrow et al., 1999). In situ hybridization In Drosophila, the gene glial cells missing (gcm) is known to In situ hybridization analysis of Hes5 was performed as previously determine the glial fate: in loss-of-function gcm mutants, nearly described (Ishibashi et al., 1995; Ishii et al., 1997). Digoxigenin- all glial cells fail to differentiate while ectopic gcm expression labeled antisense RNA probe was synthesized in vitro, as previously converts presumptive neurons into glia (Hosoya et al., 1995; described (Akazawa et al., 1992). The probe was hybridized to 12 µm Jones et al., 1995; Vincent et al., 1996). Two mammalian gcm cryostat sections of mouse retinae. 2516 M. Hojo and others

Dissociation of retinal cells Amersham), mouse anti-HPC1 (1:200; Sigma), mouse anti-calbindin To count the cell numbers, retinae were dissociated, as previously (1:200; Sigma) and rabbit rhodopsin antibodies (1:2000; LSL). described (Morrow et al., 1998). Briefly, neural retinae were incubated Affinity-purified rabbit anti-Hes5 antibody was raised against the with 0.1% trypsin in HBSS lacking Ca2+/Mg2+ for 10 minutes at room peptide, AGPKSLHQDYSEGYSW, which is common to mouse and temperature and then soybean trypsin inhibitor was added (a final rat Hes5. To detect these antibodies, FITC-conjugated goat anti-mouse concentration of 2 mg/ml). The cells were pelleted, resuspended in IgG (1:200; Vector), FITC-conjugated goat anti-rabbit IgG (1:200; HBSS containing 100 µg/ml DNaseI, triturated and plated on poly-D- Vector), biotinylated anti-rabbit IgG (1:200; Vector), biotinylated anti- lysine-coated eight-well glass slides (Nalge Nunc). After incubation mouse IgG (1:200; Vector), FITC-avidin (1:2000; Vector) and Texas for 90 minutes at 37°C in 5% CO2, the cells were fixed with 4% red avidin (1:2000; Vector) were used. For BrdU staining, BrdU was paraformaldehyde for 10 minutes at room temperature and subjected to applied (2 µM) to the retinal explants 2 days after viral infection and immunostaining (see below). Cells were counted manually with the explants were cultured in the continued presence of BrdU. Sections masking the genotypes. were then made and stained with anti-BrdU antibody (1:5; Becton Dickinson). TUNEL assay was performed with a detection kit Retinal explant culture (Boehringer-Mannheim). Pictures were taken by confocal microscopy Retinal explant culture was performed, as previously described (Caffé (Carl Zeiss). et al., 1989; Sparrow et al., 1990; Stoppini et al., 1991; Tomita et al., 1996a). Eyes were isolated from mouse embryos and neonates and transferred to PBS solution. The neural retina without pigment RESULTS epithelium was placed onto a Millicell chamber filter (Millipore: diameter 30 mm, pore size 0.4 µm) with the ganglion cell layer upward. The chamber was placed in a 6-well culture plate, which Hes5 is specifically expressed by differentiating contained 0.8 ml of 50% MEM with Hepes, 25% Hank’s solution, 25% Müller glial cells heat-inactivated horse serum, 200 µM L-glutamine, and 6.75 mg/ml In situ hybridization analysis showed that, at postnatal day 1 glucose (Stoppini et al., 1991). Explants were incubated at 34°C in 5% (P1), Hes5 was expressed in the ventricular zone of the CO2. developing mouse retina, which contains common precursors Construction and infection of retrovirus for neurons and glial cells (Fig. 1A). During the postnatal period Hes5 expression was gradually restricted to the inner CLIG was constructed as follows. The SphI-SacI fragment of pIRES- EGFP (Clontech) was isolated, and the SphI and SacI sites were nuclear layer (INL) and, by P6, Hes5 was expressed only in the converted to EcoRI and the blunt end, respectively. This fragment was INL (Fig. 1B). After P7, Hes5 expression in the INL decreased ligated to EcoRI- and HpaI-digested pCLNCX v.2 (kindly provided by and became undetectable by P14 (Fig. 1C-E). At P6-8, Müller I. Verma) to replace the neo and the CMV promoter with the IRES- glial cells (vimentin+) were differentiating in the INL, where EGFP. The resultant vector was pCLIG. Hes5 cDNA was cloned into Hes5 expression was observed (Fig. 1G-I). Expression of the EcoRI site of pCLIG. The retroviral DNAs were transfected into the ecotropic packaging cells ψ2mp34 (Yoshimatsu et al., 1998), and 3 days later the medium was recovered and concentrated, as previously described (Tsuda et al., 1998). Virus solution was applied to the retinal explants, and after 2 weeks of culture, the explants were fixed with 4% paraformaldehyde, dehydrated in 25% sucrose, embedded in OCT compound (Miles) and sectioned. Immunostaining and TUNEL Sections and dissociated cells on slides were preincubated in PBS containing 5% goat serum and 0.1% Triton X-100 for 30 minutes and incubated in 1% goat serum and 0.1% Triton X-100 with the following primary antibodies (dilution rate): mouse anti-myc (1:500; Invitrogen), mouse anti- Fig. 1. Hes5 expression and Müller glial differentiation in the developing retina. (A-E) In situ hybridization vimentin (1:1; Histofine), of Hes5. Hes5 was initially expressed in the ventricular zone at P1 and later the expression was restricted to mouse anti-glutamine the INL. By P14, the expression disappears. (F-J) Expression of the Müller glial marker vimentin (Vim). synthetase (1:200; Chemicon), Müller glial cells differentiated in the INL, where Hes5 expression is observed. (K-O) Expression of the anti-GFP (1:500; Clontech), Müller glial marker glutamine synthetase (GS). By the time when GS was well expressed (P14), Hes5 mouse anti-PKC (1: 200; expression became undetectable. VZ, the ventricular zone. Scale bar, 50 µm. Glial cell fate specification by Hes5 2517

2D). In contrast, none of Hes5+ cells (green in Fig. 2F,H) expressed the retinal neuronal markers such as protein kinase C (PKC) (red in Fig. 2G,H) and MAP2 (data not shown). These results indicated that Hes5 is specifically expressed by differen- tiating glial cells in the retina at P7. Thus, Hes5 expression was initially observed in common precursors but later became restricted to differentiating Müller glial cells. Misexpression of Hes5 promotes gliogenesis To determine whether Hes5 is functionally involved in gliogenesis, we attempted to misexpress Hes5 in the developing retina. Two replication-incompetent retroviruses, CLIG and CLIG-Hes5, were generated (Fig. 3A). CLIG directed the expression of green fluorescent protein (GFP) as a marker, and CLIG-Hes5 directed coexpression of Hes5 and GFP. To verify the coexpression of Hes5 and GFP, Hes5 was fused with the myc-epitope tag and immunohistochemistry with anti-myc and anti-GFP antibodies was performed. We found that almost all GFP+ cells successfully coexpressed Hes5 (Fig. 3C-E). However, since in transient transfection assay the transcrip- tional repression activity of Hes5 seemed to be somewhat inhibited by the myc tag (data not shown), Hes5 without myc tag was expressed for most experiments. The viruses were applied to retinal explants prepared from mouse embryos at day 17 (E17) and, after two weeks, the fate of the virus-infected cells was determined by monitoring GFP+ cells. When the retina was infected with the control virus CLIG, the majority (>80%) of the virus-infected cells became rod photoreceptors in the outer nuclear layer (ONL) (Figs 3B, 4A) and only about 8% became Müller glial cells (Fig. 4A), as previously described (Turner and Cepko, 1987). In contrast, when the retina was infected with CLIG-Hes5, rods significantly decreased (Fig. 4A) and more than 50% of the virus-infected cells became Müller glial cells, which had the cell body in the INL and long processes extending to the ONL and to the ganglion cell layer (Figs 3C-E,F,I, 4A). These CLIG-Hes5-infected cells expressed the Müller glial markers vimentin (Fig. 3F-H) and glutamine synthetase (Fig. 3I-K). The other cell types were not significantly affected by Hes5 (Fig. 4A). Thus, Hes5 promoted Fig. 2. Hes5 expression in dissociated retinal cells. P7 retinae were gliogenesis and inhibited generation of neurons, particularly dissociated and subjected to immunocytochemistry. The rate of + + rods. Hes5 cells is less than 10% at P7 and the area abundant in Hes5 The gliogenic activity of Hes5 could be the result of cells is shown here. (A-D) Hes5 and vimentin expression was examined. Hes5+ cells (green in B and yellow in D) expressed proliferation of glial cells and apoptosis of neurons rather than vimentin (red in C and yellow in D). Hes5− cells (indicated by conversion of precursors to the glial fate at the expense of the arrowheads in A) did not express vimentin. (E-H) Hes5 and PKC neuronal fate. To distinguish between these possibilities, expression was examined. None of Hes5+ cells (green in F,H) proliferation and death of virus-infected cells were analyzed. To expressed PKC (red in G,H). Thus, Hes5 is specifically expressed by detect cell proliferation, BrdU was applied to the retinal explant differentiating Müller glial cells at P7. Scale bar, 20 µm. cultures. Retinal explants prepared at E17 were infected with virus on the same day and BrdU was applied to the cultures 2 days after viral infection (corresponding to E19-P0). After another Müller glial marker, glutamine synthetase, was also culture in the continued presence of BrdU for 12 days, observed but weakly at P7 (Fig. 1M). The glutamine synthetase approximately 15% each of CLIG- and CLIG-Hes5-infected expression became very strong at P14 (Fig. 1O), at which time cells incorporated BrdU. Therefore, there was no significant Hes5 expression was undetectable (Fig. 1E), suggesting that difference in BrdU incorporation between the two infections Hes5 is expressed by differentiating, but not mature, Müller (Fig. 5A,B), indicating that Hes5 did not promote cellular glial cells. proliferation during this stage. To determine the extent of cell To show decisively that Hes5 is specifically expressed by dif- death, the retinal explants were subjected to TUNEL assay at ferentiating glial cells, P7 retinae were dissociated and various time points: 4, 8 and 14 days after viral infection. The subjected to immunocytochemistry. Approximately 80% of majority of both CLIG- and CLIG-Hes5-infected cells were Hes5+ cells (green in Fig. 2B, and yellow in Fig. 2D) expressed negative for the TUNEL assay at every time point (Fig. 5C,D; the glial marker vimentin (red in Fig. 2C, and yellow in Fig. data not shown) and therefore Hes5 did not induce apoptosis 2518 M. Hojo and others either. These results indicated that the gliogenic activity of Hes5 difficult to show a statistical increase of some cell types such as was not the result of glial proliferation or neuronal death but rods. most likely of conversion of precursor cells toward the glial fate at the expense of the neuronal fate. Interestingly, activation of EGF receptor also increases Müller glial cells (Lillien, 1995) DISCUSSION and thus Hes5 could function in the EGF signalling although it remains to be determined whether activation of EGF receptor Hes5 modulates glial fate specification in mouse upregulates Hes5 expression. retina Previous studies demonstrated that Notch activation by Delta We provide evidence that the bHLH gene Hes5 promotes glial at different stages during development generated different cell fate specification in mouse retina. Hes5 is initially expressed by types in Xenopus (Dorsky et al., 1997). Since Hes5 is one of the common precursors for neurons and glial cells, but later Hes5 Notch effectors (Ohtsuka et al., 1999), we next examined expression is restricted to differentiating Müller glial cells. whether infection of CLIG-Hes5 at different stages generated Furthermore, misexpression of Hes5 significantly increased the different cell types. CLIG-Hes5 was thus infected at the earlier (E15) and later stages (P1) of the developing retina. In both infections, Müller glial cells significantly increased (Fig. 4B) and rods decreased (Fig. 4C), like infection at E17. The other types of neurons were not affected in all infections (data not shown). Thus, Hes5 always directed gliogenesis, irrespective of the time of misexpression. The clonal size of CLIG-Hes5-infected cells was very similar to that of CLIG-infected cells in all infections (Fig. 4D), thus agreeing with the above notion that Hes5 is not involved in cell growth or survival during this period. Decrease of Müller glial cells in Hes5-null retina To further clarify the gliogenic function of Hes5, the retinae of Hes5-mutant mice (Ohtsuka et al., 1999; Cau et al., 2000) were analyzed. Vimentin-positive Müller glial cells appeared around P6 and P7 in both wild-type and Hes5-mutant retina (Fig. 6A), indicating that the onset of glial differentiation is not affected by Hes5 mutation. However, the density of Müller glial cells was somewhat reduced by Hes5 mutation at P7 (Fig. 6A). The number of Müller glial cells (Vim+, GS+) was also lower at P14 and adult in Hes5-mutant retina than the wild type (Fig. 6B,C). TUNEL assay indicated that cell death was not increased by Hes5 mutation (Fig. 6A-C), suggesting that generation, but not survival, of Müller glial cells was affected in the absence of Hes5. To precisely determine the glial cell numbers, P14 and adult retinae were dissociated and Müller glial cells were counted (Fig. 7E,F). In the wild-type retina, the number of Müller glial cells was approximately 4.0-5.0% (GS+, Vim+) of the total retinal cells at both P14 and adult (Fig. 7A,B,E). In contrast, this number was reduced to 2.8-3.2% (GS+) or 2.3-2.8% (Vim+) in the Hes5- mutant retina (Fig. 7A,B,F). Thus, Hes5 mutation showed 30-40% reduction of Müller glial cell Fig. 3. Infection of CLIG and CLIG-Hes5 in the retina. (A) Schematic structure population. To determine whether this decrease of of recombinant retroviruses, CLIG and CLIG-Hes5. CLIG directs GFP Müller glial cells correlated with increase of particular expression only and CLIG-Hes5 directs Hes5 and GFP expression from the + upstream LTR promoter. IRES, internal ribosomal entry site. (B) Infection of cell types, the number of other cells such as rods (Rh ) + and bipolar cells (PKC+) was counted. However, CLIG. The majority of the virus-infected cells (GFP ) became rods in the ONL. (C-E) Infection of CLIG-Hes5. Here, Hes5 fused with the myc tag was difference between wild type and Hes5 mutant retina expressed. This virus directed coexpression of GFP (C,E) and Hes5 (D,E). This was not statistically significant (Fig. 7C,D). Thus, no virus-infected cell became a Müller glial cell. (F-H) Many CLIG-Hes5-infected particular cell types were favored by Hes5 mutation cells (green) expressed the Müller glial marker vimentin (red). (I-K) Many although, since the decrease of glial cells CLIG-Hes5-infected cells (green) expressed the Müller glial marker glutamine corresponded to 2% of the total retinal cells, it was synthetase (red). Scale bar, 20 µm. Glial cell fate specification by Hes5 2519 population of Müller glial cells while decreasing that of rods Campos-Ortega, 1996; Fisher and Caudy, 1998; Ligoxygakis et without causing cell proliferation or death. Thus, it is most al., 1999). Instead, gcm determines the glial fate in Drosophila likely that Hes5 converts precursor cells toward the glial fate at (Hosoya et al., 1995; Jones et al., 1995; Vincent et al., 1996). the expense of the neuronal fate. In Drosophila, Hes5 Currently, two mammalian gcm homologues are known homologues, namely hairy and the bHLH genes in the (Akiyama et al., 1996; Kim et al., 1998; Kanemura et al., 1999; Enhancer of split complex, inhibit neural development and Reifegerste et al., 1999) but neither shows the expression none of them are known to promote gliogenesis (Nakao and pattern that may account for gliogenesis (Kim et al., 1998; Kanemura et al., 1999; Reifegerste et al., 1999). Since mammalian Gcm1/Gcma has been shown to rescue the gcm mutation in Drosophila (Kim et al., 1998), we misexpressed Gcma and Gcmb with retrovirus in mouse retina to examine whether they have a gliogenic activity in mammalian tissues. However, neither promoted gliogenesis in mouse retina (unpublished data). Thus, in mammals Hes5, but not gcm homologues, promotes gliogenesis and the mechanism of gliogenesis seems to have diverged during evolution between mammals and Drosophila. It has been suggested that mammalian and Drosophila glial cells are not functionally conserved since terminal differentiation genes expressed in

Fig. 4. Effects of misexpression of Hes5 on retinal cells. (A) Ratios of cell types. CLIG and CLIG-Hes5 were applied to retinal explants at E17, and after two weeks of culture the cell types were determined according to the position and morphology with the aid of the following markers: neurofilament (ganglion and horizontal cells), HPC1 (amacrine cells), calbindin (amacrine and horizontal cells), PKC (bipolar cells), vimentin (Müller glia), glutamine synthetase Fig. 5. Cell proliferation and death in the virus-infected retina. The (Müller glia) and rhodopsin (rods). Hes5 increased the population of retinal explant cultures were prepared from E17 embryos and virus Müller glial cells and decreased that of rods. (B) Comparison of the was applied to the cultures on the same day. (A,B) BrdU was applied ratios of Müller glial cells in infections at E15, E17, and P1. The to the retinal explants two days after viral infection (corresponding to open and closed bars represent CLIG and CLIG-Hes5 infection, E19-P0) and the explants were cultured in the continued presence of respectively. Hes5 always promoted gliogenesis, irrespective of the BrdU. Sections were examined two weeks after viral infection. The time of infection. (C) Comparison of the ratios of rods in infections majority (>80%) of cells infected with CLIG (A, green) and CLIG- at E15, E17 and P1. Hes5 always inhibited rod differentiation. Hes5 (B, green) did not incorporate BrdU (red). Thus, these virus- (D) Comparison of clonal sizes in infections at E15, E17 and P1. The infected cells were not proliferating. Due to the presence of BrdU in clonal sizes of CLIG- and CLIG-Hes5-infected cells were almost the the cultures, the morphology of retinal cells was somewhat affected. same. In each experiment, virus-infected cells were classified based (C,D) TUNEL assay was performed four days after viral infection. upon position, morphology and immunohistochemistry and those Cells infected with CLIG (C, green) and CLIG-Hes5 (D, green) were that were unable to be classified were excluded. More than 200 negative for TUNEL (red). Thus, these virus-infected cells were not clones in at least two independent retinal explants were examined dying. The labeled cells in D are probably migrating immature and the standard error is shown with each bar. Müller glial cells. Scale bar, 20 µm. 2520 M. Hojo and others

Fig. 6. Müller glial cell development in Hes5-mutant retina. Retinal sections of wild-type (+/+) and Hes5-deficient mice (−/−) at P7 (A), P14 (B), and adult (C). At P7, Müller glial cells (Vim+, weakly GS+) were differentiating in both wild-type and Hes5-deficient retina, indicating that the onset of glial differentiation was not affected by Hes5 mutation. In addition, cell death (TUNEL) was not affected either by Hes5 mutation. However, the density of Müller glial cells was always lower in Hes5-deficient retina than in the wild type. Thus, generation, but not survival, of Müller glial cells was affected by Hes5 mutation. Scale bar, 50 µm.

Fig. 7. The ratios of retinal cells of Hes5-mutant mice. (A-D) The ratios of retinal cells. P14 and adult retinae were dissociated and subjected to immunocytochemistry to determine the cell number. The ratio of Müller glial cells (GS+, Vim+) was reduced in Hes5-mutant retina at both P14 and adult (A,B). *Changes compared to the wild type are statistically significant (t-test, P<0.05). In contrast, there was no statistically significant difference in the ratios of bipolar cells (PKC+) and rods (rhodopsin (Rh)+) between wild-type and Hes5- mutant retina (C,D). At least two independent retinae were examined and more than 1,000 cells in each retina were counted. The standard error is shown with each bar. (E,F) Representative figures of dissociated P14 retinal cells. The wild-type (E) and Hes5(−/−) retinae (F) were dissociated and subjected to immunocytochemistry with anti-GS Ab. In these figures, 14 out of 259 wild-type cells (E) and 9 out of 282 Hes5(−/−) cells (F) were positive, indicating that the number of Müller glial cells was reduced in the absence of Hes5.

glial cells are not molecularly conserved (Kim et al., 1998). Although the retina is a good model system to investigate the Therefore, this lack of conservation of glial features may relieve mechanism of the nervous system, it remains to be determined the gliogenic genes from pressure to be maintained. whether the gliogenic function of Hes5 is universal in the whole Glial cell fate specification by Hes5 2521 nervous system. Since Hes5 is widely expressed in the demonstrated that Hes1 keeps cells at the undifferentiated developing nervous system (Akazawa et al., 1992), it is likely precursor stage while Hes5 promotes the glial fate. In contrast, that Hes5 is indeed involved in gliogenesis in other regions than Mash1 regulates the neuronal fate (Brown et al., 1998). Thus, in the retina. However, recent data showed that Notch activation mouse retina all the cell fate may be regulated by bHLH genes. induced Hes5 expression and inhibited oligodendrocyte differ- entiation (Wang et al., 1998), suggesting that Hes5 could be Notch-Hes pathway inhibitory to differentiation of some glial cells. However, in this Activation of Notch leads to two different things, keeping cells system, it is possible that the Notch pathway may activate other undifferentiated and promoting different cell type specification genes besides Hes5, such as Deltex, which can inhibit differen- (Artavanis-Tsakonas et al., 1999). How these apparently tiation genes independently of Hes genes (Matsuno et al., 1998; different things happen after Notch activation is not well Ordentlich et al., 1998). Thus, it needs further study to clarify understood, particularly in mammals. We previously showed the exact function of Hes5 in gliogenesis in other nervous that both Hes1 and Hes5 expression is induced by activation of system. Notch (Nishimura et al., 1998; Ohtsuka et al., 1999). Our present study raises the possibility that, if Notch can Possible mechanism for gliogenesis by Hes5 differentially activate Hes1 and Hes5, then it is directly It remains to be determined how Hes5 specifies the glial fate in involved in cell fate specification in addition to inhibition of cell the retina. One possible mechanism is that Hes5, a DNA- differentiation. If Notch activates both Hes1 and Hes5 in the binding repressor (Akazawa et al., 1992), may repress retina, then the cells should remain undifferentiated since Hes1 expression of neuronal bHLH genes such as Mash1, NeuroD inhibits both neuronal and glial differentiation. In contrast, if and Math3 (Tomita et al., 1996b; Takebayashi et al., 1997; Notch activates only Hes5 in a certain situation, then Notch Roztocil et al., 1997; Morrow et al., 1999) and lead to the glial activation may choose the glial fate instead of keeping cells fate. Supporting this idea, we and others reported that, in the undifferentiated. This situation could happen if Hes1 kinase retina of Mash1- or NeuroD-deficient mice, Müller glial cells phosphorylates the basic region of Hes1 since this phosphoryla- increase in number (Tomita et al., 1996b; Morrow et al., 1999). tion inactivates the Hes1 activity (Ström et al., 1997) while such The notion that Hes5 may repress Mash1 expression is also phosphorylation sites are not present in the basic region of supported by the finding that Mash1 expression is prematurely Hes5. Although we do not have definitive evidence for upregulated in the regions where Hes5 expression disappears in differential activation of Hes1 and Hes5 by Notch in the retina, RBP-J or Notch1 mutant mice (de la Pompa et al., 1997). Thus, our data indicate the first evidence for a nuclear factor in the antagonistic regulation between Hes5 and the neuronal mammalian Notch pathway, which may directly regulate the bHLH genes such as Mash1 may determine the ratio of cell fate specification. neuronal to glial cell numbers. However, in the retina of Hes5- deficient mice, many Müller glial cells still differentiated, We thank Hiroyuki Miyoshi and Inder Verma for the retroviral suggesting that Hes5 may be redundant in gliogenesis. Hes1 is vector pCLNCX v.2, Yoshiki Hotta and Masato Nakafuku for cDNA ψ unlikely to compensate for the gliogenic activity of Hes5 since constructs, and Kazuhiro Ikenaka for 2mp34. This work was Hes1 has no such activity, raising the possibility that an as yet supported by Special Coordination Funds for Promoting Science and Technology and research grants from the Ministry of Education, unidentified bHLH gene may compensate for the gliogenic Science, Sports and Culture of Japan and Japan Society for the activity of Hes5. Promotion of Science. Even though Hes5 may repress neuronal bHLH genes, it still remains to be determined which transcription factors function downstream of or in collaboration with Hes5 for gliogenesis. REFERENCES Our preliminary data suggest that some homeobox genes can promote differentiation of INL-specific cell types, such as Akazawa, C., Sasai, Y., Nakanishi, S. and Kageyama, R. (1992). Molecular bipolar, amacrine and Müller glial cells, but cannot choose the characterization of a rat negative regulator with a basic helix-loop-helix neuronal or glial fate. Thus, it is possible that some homeobox structure predominantly expressed in the developing nervous system. J. Biol. genes may collaborate with Hes5 to determine the Müller glial Chem. 267, 21879-21885. Akiyama, Y., Hosoya, T., Poole, A. M. and Hotta, Y. (1996). The gcm-motif: a cell fate in the retina while a combination of the same novel DNA-binding motif conserved in Drosophila and mammals. Proc. Natl. homeobox genes and the neuronal bHLH gene such as Mash1 Acad. Sci. USA 93, 14912-14916. may determine the fate of INL-specific neurons. Artavanis-Tsakonas, S., Rand, M. D. and Lake, R. J. (1999). Notch signaling: Whereas Hes1 has been shown to repress Mash1 expression cell fate control and signal integration in development. Science 284, 770-776. Brown, N. L., Kanekar, S., Vetter, M. L., Tucker, P. K., Gemza, D. L. and (Ishibashi et al., 1995; Chen et al., 1997), Hes1 does not Glaser, T. (1998). Math5 encodes a murine basic helix-loop-helix promote gliogenesis but inhibits both neuronal and glial differ- transcription factor expressed during early stages of retinal neurogenesis. entiation unlike Hes5 (Sasai et al., 1992; Ishibashi et al., 1994; Development 125, 4821-4833. Tomita et al., 1996a). We do not know yet why the two related Burmeister, M., Novak, J., Liang, M.-Y., Basu, S., Ploder, L., Hawes, N. L., genes, Hes1 and Hes5, have such different functions. Since the Vidgen, D., Hoover, F., Goldman, D., Kalnins, V. I., Roderick, T. H., Taylor, B. A., Hankin, M. H. and McInnes, R. R. (1996). Ocular retardation basic regions of Hes1 and Hes5 are different, it is likely that mouse caused by Chx10 homeobox null allele: impaired retinal progenitor their exact target genes are also different, and therefore it is proliferation and bipolar cell differentiation. Nature Genet. 12, 376-383. possible that, while Hes1 represses other genes such as Caffé, A. R., Visser, H., Jansen, H. G. and Sanyal, S. (1989). Histotypic differ- homeobox genes in addition to Mash1, Hes5 does not. entiation of neonatal mouse retina in organ culture. Curr. Eye Res. 8, 1083- 1092. During neural development, cells either remain as Cau, E., Gradwohl, G., Casarosa, S., Kageyama, R. and Guillemot, F. precursors, differentiate as neurons, or differentiate as glial (2000). Hes genes regulate sequential stages of neurogenesis in the olfactory cells. Our present as well as other previous studies epithelium. Development 127, 2323-2332. 2522 M. Hojo and others

Cepko, C. L., Austin, C. P., Yang, X., Alexiades, M. and Ezzeddine, D. Lillien, L. (1995). Changes in retinal cell fate induced by overexpression of EGF (1996). Cell fate determination in the vertebrate retina. Proc. Natl. Acad. Sci. receptor. Nature 377, 158-162. USA 93, 589-595. Mathers, P. H., Grinberg, A., Mahon, K. A. and Jamrich, M. (1997). The Rx Chen, H., Thiagalingam, A., Chopra, H., Borges, M. W., Feder, J. N., Nelkin, homeobox gene is essential for vertebrate eye development. Nature 387, 603- B. D., Baylin, S. B. and Ball, D. W. (1997). Conservation of the Drosophila 607. lateral inhibition pathway in human lung cancer: a hairy-related protein (HES- Matsuno, K., Eastman, D., Mitsiades, T., Quinn, A. M., Carcanciu, M. L., 1) directly represses achaete-scute homolog-1 expression. Proc. Natl. Acad. Ordentlich, P., Kadesch, T. and Artavanis-Tsakonas, S. (1998). Human Sci. USA 94, 5355-5360. deltex is a conserved regulator of Notch signalling. Nature genet. 19, 74- Chen, S., Wang, Q.-L., Nie, Z., Sun, H., Lennon, G., Copeland, N. G., 78. Gilbert, D. J., Jenkins, N. A. and Zack, D. J. (1997). Crx, a novel Otx-like Morrow, E. M., Belliveau, M. J. and Cepko, C. L. (1998). Two phases of rod paired-homeodomain protein, binds to and transactivates photoreceptor cell- photoreceptor differentiation during rat retinal development. J. Neurosci. 18, specific genes. Neuron 19, 1017-1030. 3738-3748. de la Pompa, J. L., Wakeham, A., Correia, K. M., Samper E., Brown, S., Morrow, E. M., Furukawa, T., Lee, J. E. and Cepko, C. L. (1999). NeuroD Aguilera, R. J., Nakano, T., Honjo, T., Mak, T. W., Rossant, J. and Conlon regulates multiple functions in the developing neural retina in rodent. R. A. (1997). Conservation of the Notch signalling pathway in mammalian Development 126, 23-36. neurogenesis. Development 124, 1139-1148. Nakao, K. and Campos-Ortega, J. A. (1996). Persistent expression of genes of Dorsky, R. I., Chang, W. S., Rapaport, D. H. and Harris, W. A. (1997). the Enhancer of split Complex suppresses neural development in Drosophila. Regulation of neuronal diversity in the Xenopus retina by Delta signalling. Neuron 16, 275-286. Nature 385, 67-70. Nishimura, M., Isaka, F., Ishibashi, M., Tomita, K., Tsuda, H., Nakanishi, S. Fisher, A. and Caudy, M. (1998). The function of hairy-related bHLH repressor and Kageyama, R. (1998). Structure, chromosomal locus, and promoter of proteins in cell fate decisions. BioEssays 20, 298-306. mouse Hes2 gene, a homologue of Drosophila hairy and Enhancer of split. Freund, C. L., Gregory-Evans, C. Y., Furukawa, T., Papaioannou, M., Genomics 49, 69-75. Looser, J., Ploder, L., Bellingham, J., Ng, D., Herbrick, J.-A. S., Duncan, Ohtsuka, T., Ishibashi, M., Gradwohl, G., Nakanishi, S., Guillemot, F. and A., Scherer, S. W., Tsui, L.-C., Loutradois-Anagnostou, A., Jacobson, S. Kageyama, R. (1999). Hes1 and Hes5 as Notch effectors in mammalian G., Cepko, C. L., Bhattacharya, S. S. and McInnes, R. R. (1997). Cone-rod neuronal differentiation. EMBO J. 18, 2196-2207. dystrophy due to mutations in a novel photoreceptor-specific homeobox gene Ordentlich, P., Lin, A., Shen, C.-P., Blaumueller, C., Matsuno, K., (CRX) essential for maintenance of the photoreceptor. Cell 91, 543-553. Artavanis-Tsakonas, S. and Kadesch, T. (1998). Notch inhibition of E47 Furukawa, T., Morrow, E. M. and Cepko, C. L. (1997a). Crx, a novel otx-like supports the existence of a novel signaling pathway. Mol. Cell. Biol. 18, 2230- homeobox gene, shows photoreceptor-specific expression and regulates 2239. photoreceptor differentiation. Cell 91, 531-541. Reifegerste, R., Schreiber, J., Gülland, S., Lüdemann, A. and Wegner, M. Furukawa, T., Kozak, C. A. and Cepko, C. L. (1997b). rax, a novel paired- (1999). mGCMa is a murine transcription factor that overrides cell fate type homeobox gene, shows expression in the anterior neural fold and decision in Drosophila. Mech. Dev. 82, 141-150. developing retina. Proc. Natl. Acad. Sci. USA 94, 3088-3093. Roztocil, T., Matter-Sadzinski, L., Alliod, C., Ballivet, M., Matter, J.-M. Hosoya, T., Takizawa, K., Nitta, K. and Hotta, Y. (1995). glial cells missing: a (1997). NeuroM, a neural helix-loop-helix transcription factor, defines a new binary switch between neuronal and glial determination in Drosophila. Cell transition stage in neurogenesis. Development 124, 3263-3272. 82, 1025-1036. Sasai, Y., Kageyama, R., Tagawa, Y., Shigemoto, R. and Nakanishi, S. Ishibashi, M., Moriyoshi, K., Sasai, Y., Shiota, K., Nakanishi, S. and (1992). Two mammalian helix-loop-helix factors structurally related to Kageyama, R. (1994). Persistent expression of helix-loop-helix factor HES-1 Drosophila hairy and Enhancer of split. Genes Dev. 6, 2620-2634. prevents mammalian neural differentiation in the central nervous system. Sparrow, J. R., Hicks, D. and Barnstable, C. J. (1990). Cell commitment EMBO J. 13, 1799-1805. and differentiation in explants of embryonic rat neural retina: comparison Ishibashi, M., Ang, S.-L., Shiota, K., Nakanishi, S., Kageyama, R. and with developmental potential of dissociated retina. Dev. Brain Res. 51, 69- Guillemot, F. (1995). Targeted disruption of mammalian hairy and Enhancer 84. of split homolog-1 (HES-1) leads to up-regulation of neural helix-loop-helix Stoppini, L., Buchs, P.-A. and Muller, D. (1991). A simple method for factors, premature neurogenesis, and severe neural tube defects. Genes Dev. 9, organotypic cultures of nervous tissue. J. Neurosci. Meth. 37, 173-182. 3136-3148. Ström, A., Castella, P., Rockwood, J., Wagner, J. and Caudy, M. (1997). Ishii, Y., Fukuda, K., Saiga, H., Matsushita, S. and Yasugi, S. (1997). Early Mediation of NGF signaling by post-translational inhibition of HES-1, a basic specification of intestinal epithelium in the chick embryo: a study on the helix-loop-helix repressor of neuronal differentiation. Genes Dev. 11, 3168- localization and regulation of CdxA expression. Dev. Growth Diff. 39, 643- 3181. 653. Takebayashi, K., Takahashi, S., Yokota, C., Tsuda, H., Nakanishi, S., Jean, D., Ewan, K. and Gruss, P. (1998). Molecular regulators involved in Asashima, M. and Kageyama, R. (1997). Conversion of ectoderm into a vertebrate eye development. Mech. Dev. 76, 3-18. neural fate by ATH-3, a vertebrate basic helix-loop-helix gene homologous to Jones, B. W., Fetter, R. D., Tear, G. and Goodman, C. S. (1995). glial cells Drosophila proneural gene atonal. EMBO J. 16, 384-395. missing: a genetic switch that controls glial versus neuronal fate. Cell 82, Tomita, K., Ishibashi, M., Nakahara, K., Ang, S.-L., Nakanishi, S., 1013-1023. Guillemot, F. and Kageyama, R. (1996a). Mammalian hairy and Enhancer Kageyama, R. and Nakanishi, S. (1997). Helix-loop-helix factors in growth of split homolog 1 regulates differentiation of retinal neurons and is essential and differentiation of the vertebrate nervous system. Curr. Opin. Genet. Dev. for eye morphogenesis. Neuron 16, 723-734. 7, 659-665. Tomita, K., Nakanishi, S., Guillemot, F. and Kageyama, R. (1996b). Mash1 Kanekar, S., Perron, M., Dorsky, R., Harris, W. A., Jan, L. Y., Jan, Y. N. and promotes neuronal differentiation in the retina. Genes Cells 1, 765-774. Vetter, M. L. (1997). Xath5 participates in a network of bHLH genes in the Tsuda, H., Takebayashi, K., Nakanishi, S. and Kageyama, R. (1998). developing Xenopus retina. Neuron 19, 981-994. Structure and promoter analysis of Math3 gene, a mouse homolog of Kanemura, Y., Hiraga, S., Arita, N., Ohnishi, T., Izumoto, S., Mori, K., Drosophila proneural gene atonal. J. Biol. Chem. 273, 6327-6333. Matsumura, H., Yamasaki, M., Fushiki, S. and Yoshimine, T. (1999). Turner, D. L. and Cepko, C. L. (1987). A common progenitor for neurons and Isolation and expression analysis of a novel human homologue of the glia persists in rat retina late in development. Nature 328, 131-136. Drosophila glial cells missing (gcm) gene. FEBS Lett. 442, 151-156. Vincent, S., Vonesch, J.-L. and Giangrande, A. (1996). glide directs glial fate Kim, J., Jones, B. W., Zock, C., Chen, Z., Wang, H., Goodman, C. S. and commitment and cell fate switch between neurones and glia. Development Anderson, D. J. (1998). Isolation and characterization of mammalian 122, 131-139. homologs of the Drosophila gene glial cells missing. Proc. Natl. Acad. Sci. Wang, S., Sdrulla, A. D., diSibio, G., Bush, G., Nofziger, D., Hicks, C., USA 95, 12364-12369. Weinmaster, G. and Barres, B. A. (1998). Notch receptor activation inhibits Ligoxygakis, P., Bray, S. J., Apidianakis, Y. and Delidakis, C. (1999). Ectopic oligodendrocyte differentiation. Neuron 21, 63-75. expression of individual E(spl) genes has differential effects on different cell Yoshimatsu, T., Tamura, M., Kuriyama, S. and Ikenaka, K. (1998). fate decisions and underscores the biphasic requirement for Notch activity in Improvement of retroviral packaging cell lines by introducing the wing margin establishment in Drosophila. Development 126, 2205-2214. polyomavirus early region. Hum. Gene Ther. 20, 161-172.