RaxL regulates chick ganglion cell development

Mechanisms of Development 120 (2003) 881–895 www.elsevier.com/locate/modo

RaxL regulates chick ganglion cell development Kiyo Sakagami, Akiko Ishii, Naoko Shimada, Kunio Yasuda* Graduate School of Biological Sciences, Nara Institute of Science and Technology, 8916-5 Takayama, Ikoma 630-0101, Japan Received 14 March 2003; received in revised form 27 May 2003; accepted 20 June 2003

Abstract RaxL is a paired-like homeobox gene involved in vertebrate eye morphogenesis. We examined RaxL protein expression patterns during chick retinal development in combination with ganglion cell markers including the RA4 antigen, cBrn-3, Islet-1 and neuronal type III btubulin. Double-immunostaining demonstrated that downregulation of RaxL protein correlates with upregulation of ganglion cell markers in the ganglion cell layer (GCL). To explore this correlation in vivo, we performed gain- and loss-of-function experiments by electroporating retroviral vectors encoding wild-type and dominant-negative-RaxL into the optic vesicles of stage 10 chick embryos. Infection with virus expressing RaxL led to a 35% decrease in Islet-1-positive ganglion cells at E5.0 and a complete loss of ganglion cells at E15, with no effect on displaced amacrine cells in the GCL. When dominant-negative RaxL was expressed, the total number of cells in the GCL increased by , 40% at E5.0 but was reduced to 40% at E15, due to ectopic apoptosis in the GCL from E9 to E15. These results suggest that RaxL gives an inhibitory effect on ganglion cell development and that the loss of RaxL expression is required for maintenance of ganglion cells. q 2003 Elsevier Ireland Ltd. All rights reserved. Keywords: RaxL; Retinal ganglion cells; Islet-1; RA4; Sox2; Displaced amacrine cells; Electroporation

1. Introduction The neural retina provides a good model system for studying mechanisms of cell fate determination in the vertebrate central nervous system, as it is a simple neuronal tissue, composed of seven types of cells organized into three cell layers (Dowling, 1987). The neural retina originates from the inner layer of the optic cup, which evaginates from the anterior forebrain (Jean et al., 1998; Mey and Thanos, 2000). The neural retina primordium is a population of proliferating progenitor cells that gives rise to six types of neuronal cells and one type of glial cell to form an intricate neural network (Dowling, 1987). These seven types of cells differentiate at precise but overlapping times (Cepko et al., 1996; Marquardt and Gruss, 2002). During retinal development, these cells constitute three nuclear layers: the outer nuclear layer (ONL) consists of rod and cone photoreceptor cell bodies, the inner nuclear layer (INL) is composed of bipolar, horizontal and amacrine cells, as well as the Muller glia cells, and the ganglion cell layer (GCL) contains retinal ganglion cells (RGCs) and the so-called displaced amacrine cells (Galvez et al., 1977; Masland, 2001). * Corresponding author. Tel.: þ 81-743-72-5550; fax: þ81-743-72-5559. E-mail address: [email protected] (K. Yasuda). 0925-4773/$ - see front matter q 2003 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/S0925-4773(03)00163-1

RGCs are the sole output neurons in the retina. They relay light information from the eyes to the brain and are the first retinal cells to be generated during vertebrate retinogenesis (Young, 1985). During chicken retinal development, RGCs arise from embryonic day 2 (E2) to 9 (E9) (Prada et al., 1991). All RGCs are born in the ventricular zone (VZ), after which they are immediately differentiated, begin expressing the RA4 antigen and migrate to the vitreal surface to form the future GCL (Waid and McLoon, 1995). RGC specification and differentiation begins initially in the central region of the developing retina, gradually spreading to the peripheral region in a wave-like progression (McCabe et al., 1999) that is thought to involve a complex set of gene and protein interactions (Jean et al., 1998). Recent advances have begun to unravel the molecular events that control the specification and differentiation of RGCs. For example, the neurogenic gene, Notch, is expressed in retinal progenitors and the activation of the Notch signaling pathway in the retina has been shown to suppress neuronal differentiation, while promoting the formation of Mu¨ller glia cells (Austin et al., 1995; Furukawa et al., 2000). In contrast, upregulation of proneural basic helix –loop –helix (bHLH) transcription factors such as Xath5 and Xath3 has been shown to promote RGC formation in Xenopus (Kanekar et al., 1997; Perron et al.,

882

K. Sakagami et al. / Mechanisms of Development 120 (2003) 881–895

1999). It has also been noted that the POU domain transcription factor Brn-3c is first expressed in migrating RGC precursors (Liu et al., 2000), and its expression is initially activated by Ath5 factors in newly generated RGCs (Liu et al., 2001). Even given the above information, however, the molecular mechanisms underlying determination of RGC fates remain poorly understood. It is known that paired-like homeobox transcription factors play important roles in retinal cell fate determination and differentiation across diverse systems (reviewed by Jean et al., 1998; Mey and Thanos, 2000). Recently, the vertebrate paired-like homeobox gene, Rx/rax, was identified and shown to be expressed in the developing neural retina (Casarosa et al., 1997; Chuang and Raymond, 2001; Furukawa et al., 1997; Mathers et al., 1997). Overexpression experiments with the Xenopus Rx gene resulted in the formation of ectopic retinal tissues (Andreazzoli et al., 1999; Mathers et al., 1997), and mouse embryos carrying a null allele of the Rx gene do not form an optic cup and do not develop eyes (Mathers et al., 1997). This phenotype argues for the importance of the Rx/rax gene in establishment and maintenance of retinal cells (Mathers and Jamrich, 2000). Rx promotes Mu¨ller glial cell differentiation in the rat and chick RaxL plays a role in photoreceptor differentiation (Chen and Cepko, 2002; Furukawa et al., 2000). However, the various roles of Rx/rax in differentiation of other retinal cell types are still not clear. In this paper, we report the functional characterization of the Rx/rax chick homologue, RaxL, in RGC differentiation. Although RaxL expression is detected in all neural retinal cells on E3, in RGCs the expression fades completely by E7 and the protein is then restricted to photoreceptor cells by E15. We found that RaxL expression is correlated with RGC differentiation. Overexpression of RaxL using the RCAS retrovirus resulted in a decrease of RGCs at E5.0, and complete loss of RGCs at E15. In contrast, expression of a dominant-negative RaxL led to a 40% increase of RGCs at E5.0. These results suggest that RaxL negatively regulates RGC development and that the loss of RaxL expression is required for maintenance of ganglion cells during chick retinal development.

were subjected to Western blot analysis using the RaxL antiserum. The results showed that the RaxL rabbit antiserum specifically recognized RaxL and did not crossreact with Rax (data not shown). This antiserum was used to detect RaxL protein by immunohistochemical analysis of E4.0-P2 central retina sections. RaxL immunostaining patterns were correlated well with those of RaxL in situ hybridization (data not shown; Chen and Cepko, 2002), also confirming the antibody specificity. At E4.0, immunoreactivity for RaxL was observed in all cells in the VZ and, weakly in some cells in the GCL (Fig. 1A). As CNotch-1 is expressed in progenitor cells of the developing retina

2. Results 2.1. RaxL expression patterns during chick retinal development To examine the role of Rx/rax in RGC differentiation, we characterized the expression of RaxL protein during chick retinal differentiation. For this, we generated a rabbit polyclonal anti-RaxL antiserum and examined its specificity for RaxL, since two distinct but closely related genes, Rax and RaxL, are present (Ohuchi et al., 1999; Chen and Cepko, 2002). Plasmids expressing Rax or RaxL were transfected into NIH3T3 cell cultures, and the cell lysates

Fig. 1. Immunohistochemical analysis of RaxL expression during chick retinal development. Retinal sections at the indicated developmental stages were immunostained with anti-RaxL antibodies (A,D,G,J,L), stained with Hoechst for nuclear staining (B,E,H,K,M) and processed for in situ hybridization with CNotch-1 (C,F,I) and RaxL (L,O). Developmental stages according to Hamburger and Hamilton are shown on the left. Scale bars: in A is 50 mm for A-O. GCL, ganglion cell layer; VZ, ventricular zone; INL, inner nuclear layer; ONL, outer nuclear layer.

K. Sakagami et al. / Mechanisms of Development 120 (2003) 881–895

883

(Austin et al., 1995; Bao and Cepko, 1997; Henrique et al., 1997), we examined CNotch-1 mRNA expression. In situ hybridization using the CNotch-1 probe showed that CNotch-1 was also expressed in RaxL-positive cells of the VZ (Fig. 1C), demonstrating that RaxL-positive cells in E4 retina are retinal progenitor cells. At E7, RaxL protein was restricted to the outer half of the INL, (Fig. 1D), correspondingly CNotch-1 was expressed in the same region (Fig. 1F). RaxL was undetectable in the GCL at E7, in which most of RGCs are differentiated. In contrast, high levels of RaxL were detected in the ONL (Fig. 1D), being coincident with the initiation of photoreceptor cell differentiation. At E11, when almost all RGC progenitors had become postmitotic and RGC differentiation is complete (Snow and Robson, 1994; Waid and McLoon, 1995), RaxL expression was undetectable in the GCL, though it was seen in the middle of the INL and at high levels in the ONL (Fig. 1G). At this time, CNotch-1 expression was restricted to the middle of the INL (Fig. 1I). At E15, RaxL expression was downregulated in the INL (Fig. 1J,L) and RaxL expression became restricted to the ONL at P2 (Fig. 1M), whereas in situ hybridization showed no RaxL mRNA expression (Fig. 10). 2.2. RaxL expression becomes lost in Islet-1-positive RGCs in the GCL RaxL expression patterns suggest that downregulation of RaxL is correlated with formation of RGCs during early retinal development. To test this in more detail at the protein level, we first performed double-immunostaining for RaxL and Islet-1, an RGC marker (Austin et al., 1995; Thor et al., 1991), on sections of central retinas from E3.0 to E7.0 chick embryos. At E3.0, all cells were positive for RaxL and no reactivity for Islet-1 was observed. At E3.5, Islet-1-positive cells became visible at regular intervals, three- to four-cells apart, lined up at the innermost layer of the VZ (Fig. 2A(a –e)) and the majority of these cells were positive for RaxL. Closer scrutiny revealed that the Islet-1-staining in RaxL-expressing cells first appeared at E3.25 (stage 20; Fig. 2B(a –c)). As development proceeded from E3.5 to E6.0, the number of Islet-1-positive cells in the retina progressively increased while RaxL expression was gradually downregulated until no RaxL expression was detected in the majority of Islet-1-positive cells in the GCL at E6.0 (Fig. 2A(f –j)). Quantitative analysis of RaxL-positive and Islet-1positive cells from E3.0 to E7.0 demonstrated that the number of Islet-1-positive cells increased gradually from E3.5 to E7.0 (Fig. 2B(a)). As cells expressing RaxL in the GCL gradually decreased during this period, the number of Islet-1-positive/RaxL-negative cells showed a sharp increase (Fig. 2B(c)), whereas the number of doublepositive cells remained constant between E3.5 and E6.0 (Fig. 2B(b)). These results indicate that a small and constant population of cells among the cells expressing RaxL in

Fig. 2. Double-immunostaining for RaxL and Islet-1 during early stages of RGC development. (A) Spatiotemporal expression patterns of RaxL and Islet-1, a marker of RGCs, at E3.5 and E7.0. Retinal sections at the indicated developmental stages were immunostained for RaxL (a,f), and Islet-1 (b,g), and stained with Hoechst for nuclear visualization (d,i). Merged images for RaxL and Islet-1 (c,h) and for RaxL and Hoechst (e,j) are shown and each photograph at high magnification of marked region (c,h) are shown below. Developmental stages according to Hamburger and Hamilton are shown on the left. Double-positive cells are indicated by white arrow and single-positive cells by white arrowhead. (B) Developmental changes of Islet-1-positive cells in the GCL of E3.0-E7.0 central retinas. Histograms show Islet-1-positive cells (a), Islet-1-positive/RaxLpositive cells (b), and Islet-1-positive/RaxL-negative cells (c). The ‘Y’ axis represents the number of cells per 70 mm length retina. Islet-1 expression is first expressed in RaxL-positive cells in the vitreal surface at stage 20 (E3.25), and downregulation of RaxL expression is likely correlated with generation of Islet-1-positive RGCs. Scale bars: in a is 50 mm for a– j and in each photograph at high magnification 10 mm.

the vitreal surface of the retina begins to express Islet-1 from E3.25, and that subsequently RaxL becomes downregulated in these cells and is absent at E7.0. 2.3. Loss of RaxL expression correlates with terminal differentiation of RGCs The above RaxL expression data show a correlation between downregulation of RaxL and RGC differentiation. To examine when RaxL becomes lost during

884

K. Sakagami et al. / Mechanisms of Development 120 (2003) 881–895

Fig. 3. Double-immunostaining for either RaxL or cBrn-3 with RGC markers on sections of central retinas. Sections of E5.0 central retinas were double-immunostained for RaxL (A,E,I) and RA4 (B), cBrn-3 (F) or TUJ1 (J), and sections of E4.0 central retinas for cBrn-3 (M,M0 ) and RA4 (N,N0 ) and sections of E6.0 for cBrn-3 (R,V) and either Islet-1 (Q) or TUJ1 (U). Merged images are shown (C,G,K,O,O0 ,S,W) along with nuclear staining with Hoechst (D,H,L,P,P0 ,T,X). RaxL-positive and -negative cells with RA4- or cBrn-3 or TUJ1-postive cells are shown by arrow and arrowhead, respectively, in A –L. The rectangle part in M,N,O,P are shown at high magnification in M0 ,N0 ,O,0 P0 , where cBrn-3-positive/RA4-positive cell (arrow) and cBrn-3-positive/RA4-negative cell (arrowhead) are shown. cBrn-3-positive and -negative cells with either Islet-1 or TUJ1 are shown by arrow and arrowheads, respectively, in Q –X. RaxL is expressed in newly generated and migrating presumptive RA4- and cBrn-3-positive RGCs.

gangliogenesis, we performed double-immunostaining for RaxL and several RGC markers on sections of E5.0 retinas. RGCs become committed either during or immediately after the terminal cell division of RGC progenitors in the mitotic layer of the retina. RGCs are generated as the presumptive RGCs at the mitotic layer (Waid and McLoon, 1995) and then migrate to the vitreal surface to finally differentiate into RGCs by extending nerve fibers to the brain (Mey and Thanos, 2000; Snow and Robson, 1995). At E5, RGCs at various developmental stages are present in the central retina (Snow and Robson, 1994; Waid and McLoon, 1995). As they represent cells just after the last mitosis, migrating cells and migrated RGCs to the vitreal surface, sections of E5.0 central retinas were used for immunohistochemical analysis. We used RA4, anti-cBrn-3 antibody, and antiIslet-1 antibody and TUJ1 to classify RGCs into different stages of specification, migration and differentiation, respectively. The RA4 antigen is first expressed in RGCs immediately after their terminal division (McLoon and Barnes, 1989; Silva et al., 2002; Waid and McLoon, 1995), making it the earliest RGC marker. cBrn-3 is a POU domain transcription factor expressed in RGCs during and after their migration to the GCL, making it a middle-stage marker (Lindeberg et al., 1997; Liu et al., 2000). Islet-1 is used as a marker for migrated RGCs to the GCL, although Islet-1 is expressed in cholinergic amacrine cells at later stages (data not shown; Galli-Resta et al., 1997). TUJ1 stains neuronal type III b-tubulin, which is expressed in terminally differentiated RGCs, making it a late RGC marker (Snow and Robson, 1994). In E5 neural retina, the RA4 antigen and RaxL colocalized in cells located in both the outermost mitotic layer and in the thickness of the VZ (Fig. 3A – D), representing RGCs immediately after the terminal cell division and migrating RGCs, respectively. cBrn-3 signals were observed in the inner half of the INL and in the GCL, where RaxL was expressed (Fig. 3E –H). cBrn-3 signals were overlapped with RaxL in the INL, but a subpopulation of cBrn-3-positive cells in the GCL were not stained for RaxL (Fig. 3E– H). TUJ1-positive cells were just present in the GCL and only a small population of these cells was stained for RaxL (Fig. 3I – L), indicating that RaxL expression became lost immediately after RGCs terminally differentiated into TUJ1-positive RGCs. Recent studies have suggested that cBrn-3 is likely expressed in the migrating RGCs (Lindeberg et al., 1997; Liu et al., 2000). We assessed this by double-immunostaining for the RA4 antigen and cBrn-3 (Fig. 3M – P0 ), since the RA4 antigen is known to be a marker for the migrating

Islet-1 becomes detectable in the late migrating and migrated RGCs. RaxL expression then fades in migrated RGCs, which in turn become TUJ1positive in the GCL. Scale bars: in D is 50 mm for A–D, M–X, and in H 50 mm for E– L and in P0 10 mm for M0 –P0 . GCL, ganglion cell layer; VZ, ventricular zone; ML, mitotic layer.

K. Sakagami et al. / Mechanisms of Development 120 (2003) 881–895

RGCs (McLoon and Barnes, 1989; Waid and McLoon, 1995). Double-immunostaining demonstrated that a subpopulation of the RA4-positive migrating RGCs was positive for cBrn-3 (Fig. 3M0 – P0 ). A subpopulation of the cBrn-3-positive migrating RGCs began to express Islet-1, and all of them became Islet-1-positive when they migrated to the GCL (Fig. 3Q – T). These results show that Islet-1 became detectable in all cBrn-3-positive migrated RGCs. This makes Islet-1 later marker than cBrn-3 during RGC development. TUJ1 immunoreactivity was then detected in the cBrn-3-positive cells in the GCL (Fig. 3U – X). Thus, the RA4-positive migrating RGCs are stained for RaxL, cBrn-3 and Islet-1 while migrated and differentiated RGCs-expressing cBrn-3, the RA4 antigen, Islet-1 and neuronal type III b-tubulin are negative for RaxL. The present double-immunostaining identified several steps in RGC formation. Taken together, these data indicate that loss of RaxL expression correlates with RGC formation in the GCL. 2.4. RaxL transactivates a rhodopsin promoter To assess biological functions of RaxL in chick RGC formation, we first analyzed RaxL and RaxL-EnR transcriptional activity in a heterologous system using transient transfection assays. We constructed a RaxL expression plasmid and a dominant-negative RaxL (RaxLEnR) expression plasmid. Rax-EnRc is carrying the repressor domain (EnR) of Drosophila Engrailed protein (Badiani et al., 1994; Han and Manley, 1993) in place of the RaxL C-terminal region containing the oar/paired-tail domain (OAR) (Mathers et al., 1997; Norris and Kern, 2001). As it has been reported that a construct comprising a fusion of the EnR and RaxL with deletion of the OAR functions as a dominant-negative form of RaxL (Andreazzoli et al., 1999; Chen and Cepko, 2002). We examined whether RaxL-EnR also functions as a dominant-negative form of RaxL using a reporter plasmid containing a chick rhodopsin 0.2 kb promoter upstream of a luciferase reporter gene, Rhod-luc (Fig. 4A). This rhodopsin promoter contains a putative RaxL and Rx/rax binding site, RET-1/PCE-1 (TAATTG) (Chen and Cepko, 2002; Kimura et al., 2000). The RaxL expression plasmid and the Rhod-luc reporter plasmid were cotransfected into NIH3T3 cells. Increased amounts of transfected RaxLexpressing plasmid led to a dose-dependent increase of luciferase activity up to a 13-fold increase relative to a control with mock vector (Fig. 4B). To assess whether RaxL-EnR could inhibit RaxL transcriptional activity, the plasmid expressing RaxL-EnR was cotransfected together with both the RaxL plasmid and the reporter plasmid. A 3.3-fold amount of RaxL-EnR-expressing plasmid relative to a fixed amount of RaxL plasmid caused a complete reduction of reporter activity back to the level observed in the control with mock vector (Fig. 4B). These results

885

Fig. 4. Transcriptional activity of RaxL and RaxL-EnR in a luciferase reporter assay driven by chick rhodopsin promoter. (A) Schematic representation of RaxL and RaxL-EnR effector constructs and rhodopsin promoter reporter construct. (B) Either RaxL or RaxL-EnR expression vectors or both RaxL and RaxL-EnR at different ratios were transfected into NIH3T3 cell cultures together with the luciferase reporter construct and pEFX3-b-gal as an internal control. RaxL transactivated the reporter in a dose-dependent manner. RaxL-EnR repressed transcriptional activation of RaxL in a dose-dependent manner, showing that it acts as a dominantnegative RaxL.

show that RaxL has the ability to transactivate promoters of genes containing RaxL target sequences and that RaxLEnR is capable of inhibiting these endogenous promoters in vivo. 2.5. Expression of exogenous RaxL results in a decrease in the number of RGCs To examine whether RaxL plays a key role in RGC formation, we ectopically expressed wild-type and dominant-negative RaxL in optic vesicles using retroviral expression vectors (Hughes et al., 1987). We electroporated retroviral vectors bearing RaxL (RCAS-RaxL) or RaxL-EnR (RCAS-RaxL-EnR) into the optic vesicles of stage 10 chick embryos. After electroporation, transfected retinal cells should generate virus particles to subsequently infect neighboring retinal cells, creating RCAS-RaxL- or RCASRaxL-EnR-producing domains in the retina. For analysis of RaxL function, we used virus-transfected domains at or near the center of the retina, where RGC formation initiates (McLoon and Barnes, 1989; Snow and Robson, 1994). Viral infection was confirmed in frontal cross sections of E5 virus-infected chick eyes by in situ hybridization using an RCAS envelope probe (Fig. 5B) and by immunostaining using anti-gag antibody (Fig. 5I,J). RCAS envelope signals were clearly detected in the infected retina and in a small area of the brain, but it was completely absent from the uninfected control sections (Fig. 5B) and spanned the entire thickness of the E15 retinas (Fig. 5I,J). These infected domains of the retinas were used for further analysis.

886

K. Sakagami et al. / Mechanisms of Development 120 (2003) 881–895

To examine effects of RaxL and RaxL-EnR in RGC formation of chick retinas infected at stage 10, we performed histological and immunohistochemical staining for sections of RaxL or RaxL-EnR virus-infected retinas at E15, when RGC formation is completed. Haematoxilin – Eosin (HE) staining showed that control and RaxL-EnRinfected retinas displayed the developmental morphology including well-defined GCL, INL and ONL (Fig. 5C– E). However, we found that RaxL-infected retina sections contained very few cells in the GCL and showed an almost complete lack of the optic nerve fiber layer (OFL). The GCL in the RaxL virus-infected retinas consisted of a single layer of cells rather than the two rows seen in control retinas (Fig. 5C,D) and lacked the OFL (Fig. 5D,G). Cells in the GCL were interspersed at two rows in the RaxL-EnR virusinfected retinas (Fig. 5E,H). To quantify the numbers of cells in each layer per section, we quantified the number of cells in HE stained sections (Fig. 5K,L). RaxL and RaxLEnR infection resulted in a 75 and 60% decrease, respectively, in the number of cells in the GCL (Fig. 5K), and a slight decrease (, 8.5%) was observed in the number of cells in the RaxL-EnR infected INL, leading to a slight decrease in the total number of cells per section in the RaxLEnR virus-infected retinas (Fig. 5L). Our results suggest that overexpression of RaxL and RaxL-EnR interfered RGC formation. 2.6. Overexpression of RaxL does not affect displaced amacrine cell differentiation

Fig. 5. Morphological abnormalities in E15 RaxL- and RaxL-EnR-virus infected retinas. Frontal sections of RaxL virus-infected E5.0 heads were stained with Haematoxilin–Eosin (HE) (A) and processed for in situ hybridization using DIG-labeled probe for the RCAS-envelope protein (B). Sections of central retinas of E15 chick embryos infected with RaxL or RaxL-EnR viruses at stage 18 were stained with HE (C–E), Hoechst dye (F– H), or were immunostained for RCAS-gag (I,J). Histograms show the cell number per section in the GCL (K), the INL (L) and the ONL (L). The ‘Y’ axis represents the number of cells per 180 mm length retina. Total represents the total cell number in the three layers. Scale bars: in B is 1 mm for A,B and in J 50 mm for C –J. The all retina sample number is five. P values are based on the Student’ t-test of two-tailed distribution as follows: *, p , 0:001: OFL, optic nerve fiber layer; GCL, ganglion cell layer; INL, inner nuclear layer; ONL, outer nuclear layer.

To determine whether virus infected retinal cells present in the GCL are characteristic of RGCs at E15 when RGC development is completed, we analyzed the expression of RGC markers cBrn-3 and neuronal type III b-tubulin by immunohistochemistry using anti-cBrn-3 (Fig. 6A – C) and TUJ1 antibodies (Fig. 6D – F), respectively. In this stage, anti-cBrn-3 and TUJ1 antibodies specifically recognize RGCs. In the control retina, about 70% of cells in the GCL were stained for both neuronal type-III b-tubulin and cBrn-3. However, surprisingly, we found no RGC marker expression in retinas infected with RaxL virus (Fig. 6B,E), although Hoechst staining had shown a single cell-thick GCL layer in these retinas (Fig. 6K,N). On the other hand, the single cell-thick GCL layer observed in RaxL-EnR infected retinas did stain with TUJ1 and cBrn-3 antibodies (Fig. 6C,F), indicating the presence of RGCs. We hypothesized that the single cell-thick GCL in RaxLinfected retinas might consist of displaced amacrine cells, which are known to be present in the GCL (Galvez et al., 1977). Accordingly, we examined whether the TUJ1negative cells expressed the displaced amacrine cell marker Sox2 (Le et al., 2002) by immunostaining retinal sections with anti-chicken Sox2 antibody (Fig. 6G – O). In control retinas, the majority of cells in the cell row facing the INL were positive for Sox2 (, 30% of total cells) (Fig. 6M), which was not found in the other row (Fig. 6M). In contrast,

K. Sakagami et al. / Mechanisms of Development 120 (2003) 881–895

887

the GCL was reduced by , 60%, but both RGC and displaced amacrine cells were present with their ratio similar to the control (Fig. 6L,O). These results suggest that regulation of RaxL expression is indispensable for RGCs development and that an immediate loss of RaxL expression is necessary for RGCs migrated to the GCL, otherwise RGCs would be killed. 2.7. Overexpression of RaxL affects Islet-1-positive RGC formation To examine which stages of RGC development is affected with RaxL, RaxL or RaxL-EnR transfected central retina on E3.5 and E5.0 were performed for section immunostaining using anti-Islet-1 antibody to examine a mid RGC marker expression. Islet-1-positive cells were observed in the GCL of control retinas at regular three- to four-cell intervals at E3.5 (Fig. 7A). When development proceeded to E5.0, Islet-1-positive cells increased to a sideby-side density in the GCL (Fig. 7G). In RaxL-transfected retinas, Islet-1-positive cells were decreased at E3.5 and at E5.0 (Fig. 7B,H), while RaxL-EnR production led to an increase of the Islet-1-positive cells in the GCL (Fig. 7C,I). These observations were confirmed by quantitative analysis (Fig. 7P,Q) showing that expression of RaxL resulted in a 35% decrease of Islet-1-positive cells in the GCL at E5.0. Conversely, RaxL-EnR expression led to a 40% increase of Islet-1-positive cells in the GCL at E5.0 (Fig. 7P,Q), suggesting that RaxL-EnR promotes RGC formation at early stages. These results suggest that levels of RaxL protein play an important role in Islet-1-positive RGC formation. 2.8. Overexpression of RaxL affects RGC development

Fig. 6. Immunohistochemical analyses of E15 RaxL- and RaxL-EnR-virus infected retinas for ganglion and displaced amacrine cell markers. Control or retinas infected with RaxL or RaxL-EnR virus at stage 18 were harvested at E15 and immunostained for cBrn-3 (A–C), TUJ1 (D– F), or Sox2 (G –O) and double-stained for Hoechst and Sox2 (J –O). In control retinas, some of cell in the outer cell layer in the GCL are positive for Sox-2 (arrow) but all cells in the inner cell layer are negative (arrowhead). In RaxL overexpressing retinas, no TUJ1- and cBrn-3-double-positive RGCs were observed (B,E) and all cells were Sox2-positive (H,K,N; arrows), indicating that all cells of the single cell layer in the GCL are characteristic of displaced amacrine cells (H,K,N). In RaxL-EnR expressing retinas, both type of cells were present (I,L,O) although the total number of cells in the GCL was reduced. Scale bars: in A is 50 mm for A –L and in M 10 mm for M–O. GCL, ganglion cell layer; INL, inner nuclear layer; ONL, outer nuclear layer.

all cells of the single cell-thick GCL of the RaxL infected retinas were positive for Sox2 but not for either TUJ1 nor cBrn-3 (Fig. 6N), indicating that only displaced amacrine cells are present in the GCL of RaxL-infected retinas. Overexpression of RaxL resulted in inhibition of RGC differentiation or maintenance during retinal development. In the RaxL-EnR infected retinas, the total cell number in

It was apparent that RaxL interfered with RGC formation. RGC formation consists of several steps, including proliferation, specification, migration, differentiation and maintenance. Thus, we examined which of these steps might be blocked by overexpression of RaxL. We first used a BrdU incorporation assay to test whether RaxL or RaxL-EnR affects cell proliferation. For BrdU labeling, we injected BrdU solution into virus-infected or control eye cavities of E5.0 chick embryos. After 3 h incubation, the retinas were harvested and fixed, followed by immunohistochemical staining for BrdU. BrdU signals in RaxL- and RaxL-EnR-infected retinas were comparable to those of the control retina (Fig. 8A – C). We quantified the numbers of BrdU- and Hoechst-positive cells in these control and virusinfected retinal sections. As the total number of cells per section of each retina was almost the same, the ratio of BrdU-positive cells to total cells was quantified to examine the effect of RaxL on proliferation (Fig. 8G). The histogram indicates that expression of wild-type and dominantnegative RaxL does not significantly affect the population of proliferating cells in the retinas. Thus, we concluded

888

K. Sakagami et al. / Mechanisms of Development 120 (2003) 881–895

that RaxL has no significant effect on proliferation in the retinas. Next we examined whether RaxL functions in either specification or migration of presumptive RGCs. The RA4 antigen is the earliest marker expressed just after terminal division of RGC progenitor cells and is also expressed in the migrating RGCs. cBrn-3 is expressed in the migrating and migrated RGCs. Overexpression of RaxL reduced RA4 and cBrn-3 signals in the VZ and GCL (Fig. 8H,I,K,L), while the dominant-negative form increased the number of RA4positive cells in the GCL but decreased that of RA4-positive cells in the VZ (Fig. 8H,J) and increased cBrn-3 cells (Fig. 8K,M). These results indicate that RaxL affects each step of specification, migration and differentiation of RGCs to some extent. As a whole, overexpression of RaxL resulted in a considerable decrease in the number of TUJ1-positive cells in the GCL, while an increase was observed in retinas expressing RaxL-EnR (Fig. 8N –P). As CNotch-1 is reported to inhibit RGC formation (Austin et al., 1995; Henrique et al., 1997), the expression of endogenous CNotch-1 in RaxL or RaxL-EnR infected E5.0 retina sections was examined (Fig. 8T – V). CNotch-1 signal intensities were likely enhanced in sections of the RaxL-expressed retinas but reduced in those of RaxL-EnR expressed retinas. Taken together, these results demonstrate that RaxL plays a key role in RGC development but has no significant role in the proliferation of presumptive RGCs. 2.9. Perturbation of RaxL levels induces apoptosis in both the INL and GCL

Fig. 7. Immunostaining for Islet-1 in sections of E3.5– E5.0 virus-infected retinas. Expression of exogenous RaxL or RaxL-EnR affects early specification of presumptive RGCs. Immunohistochemistry for the mid RGC marker Islet-1 was performed for sections of E3.5 and E5.0 control (A,D,G,J,M) or RaxL (B,E,H,K,N) or RaxL-EnR (C,F,I,L,O) virus-infected central retinas. RCAS virus infection was confirmed by immunostaining for RCAS-gag (M–O). Nuclei were visualized with Hoechst (D –F,J –L). Histograms indicate the number of Islet-1-positive cells in the GCL at E3.5 (P) and E5.0 (Q). The ‘Y’ axis represents the number of cells per 70 mm length retina. In RaxL overexpressing retina, the number of Islet-1-positive cells decreased. On the other hand, in RaxL-EnR expressing retina, the number increased compared to the control. The all retina sample number is five. P values are based on the Student’ t-test of two-tailed distribution as follows: *, p ¼ 0:09; **, p ¼ 0:22; ***, p , 0:001: Scale bars: in A is 50 mm for A –O. GCL, ganglion cell layer; VZ, ventricular zone.

In early stages of retinogenesis at E3.5– 5.0, the total cell number in the retina and GCL was not significantly affected by infection with either RaxL or RaxL-EnR virus compared to the control. This shows that perturbation of RaxL levels in the retinal progenitor cells did not significantly influence retinal cell proliferation, although RaxL affected Islet-1positive cell formation in the GCL at E5.0. However, by E15 the total cell number in the GCL of virus infected retinas was greatly reduced, suggesting that a substantial loss or apoptosis of RGCs must occur between E5.0 and E15. There are two phases of retinal cell death, the first peaking at E6 and the second at E12 (Frade et al., 1997), so we used TUNEL (Terminal dUTP nick-end labeling) assay to assess apoptosis during retinal development on RaxL and RaxL-EnR-producing retinas (Fig. 9). Apoptotic nuclei were not detected at E6.0 (data not shown), indicating that overexpression of RaxL protein did not affect the first phase of retinal cell death. When we examined the second phase of cell death, we found dispersed TUNEL-positive nuclei in the INL at E9-E15, and very few apoptotic cells in the GCL (Fig. 9A,D,G). RaxL overexpressing retinas displayed a slight increase of apoptotic cells in the INL and GCL from E9 to E15 (Fig. 9B,E,H). Cell death in the RaxL virusinfected retina was observed in both the GCL and the inner half of the INL close to the GCL at E12-15,

K. Sakagami et al. / Mechanisms of Development 120 (2003) 881–895

889

while TUNEL-labeled dying cells spanned the whole INL in the control retina. In contrast, the RaxL-EnR expressing retinas showed evidence of increased apoptotic cells in both the GCL and INL both at E9 and E12 (Fig. 9C,F), consistent with the finding of Chen and Cepko (2002). No TUNEL labeled cells could be found at E15 in the RaxL-EnR expressing retinas, though positive cells could still be found at this time in control and RaxL overexpressing retinal cells (Fig. 9G – I). We confirmed by double-immunostaining that TUNEL labeled cells were positive for virus envelope marker (Fig. 9H,K,I,L, insert). These results suggest that enhanced cell loss in the INL and GCL of retinas infected with either RaxL or RaxL-EnR virus between E5.0 and E15 is due to apoptosis.

3. Discussion The aim of this study is to reveal the molecular mechanisms underlying RGC formation and the involvement of RaxL during chick retinal development. We performed immunohistochemical analysis for expression of RaxL and RGC stage-specific markers including RA4, cBrn-3, RaxL, Islet-1 and TUJ1 during chick retinal development. Double-immunofluorescence studies made RGCs at different stages distinguishable, suggesting a correlation between loss of RaxL and RGC formation. Next, we used gain- and loss-of-function experiments in ovo to examine whether RaxL is involved in RGC formation. Overexpression of RaxL reduced the presence of RGCs at E5 and led eventually to a complete loss of RGCs due to apoptosis but had no effect on the displaced amacrine cells in the GCL at E15. In contrast, although overexpression of RaxL-EnR increased RGC number at E5, almost of both RGCs and displaced amacrine cells in the GCL were killed through apoptosis by E15. Our results demonstrate that RaxL is involved in development of RGCs in chick retinas.

Fig. 8. Immunohistochemical examination for BrdU incorporation and RGC markers on sections of virus-infected E5.0 retinas. Expression patterns of early RGC markers on the control, RaxL and RaxL-EnR virus infected central retinas were examined at E5.0 by immunohistochemistry and in situ hybridization. Sections were performed for BrdU labeling (A – C) and for Hoechst (D–F). The ratio of BrdU-positive cells to Hoechstpositive cells is presented by histogram (G). Sections were stained for RA4 (H –J) and cBrn-3 (K –M) and processed for in situ hybridization with CNotch-1 (T– V). Consecutive sections were double-stained for TUJ1 (N – P) and RCAS-gag (Q –S). In RaxL-virus infected retina, RGC marker expression was slightly reduced, and that of CNotch-1 was increased. On the other hand, in RaxL-EnR-virus infected retina, RGC marker expression increased slightly and CNotch-1 decreased. Scale bars: in A is 50 mm for A–F and in H for H–V. GCL, ganglion cell layer; VZ, ventricular zone.

890

K. Sakagami et al. / Mechanisms of Development 120 (2003) 881–895

Fig. 9. Apoptotic cell detection by TUNEL assay of E9-E15 virus-infected central retinas. Retinas were harvested at E9 (A –C), E12 (D–F) and E15 (G –O) from the control (A,D,G), RaxL- (B,E,H) or RaxL-EnR (C,F,I) virusinfected retinas, followed by TUNEL labeling (A –F). E15 sections were triple-stained for TUNEL (G–I), RCAS-gag (J –L) and Hoechst (M–O). In the control retinas, very dispersed TUNEL labeled cells (arrow) were found only in the INL from E9 to E15. In RaxL overexpressing retinas, TUNELpositive cells were detectable in the GCL at both E12 and E15, although INL patterns were similar to those found the control (H). In contrast, in RaxL-EnR expressing retinas, many TUNEL-positive cells (arrow) were detected in the GCL and the innermost area of the INL at E9 and E12. At E15 no TUNEL-positive cells were detected. Scale bars: in A is 50 mm for A –O. GCL, ganglion cell layer; INL, inner nuclear layer; ONL, outer nuclear layer.

3.1. Co-expression of RaxL and RGC markers during chick retinal development RGCs are the first neurons produced during vertebrate retinal development, and pass through several stages

including the proliferation of retinal progenitors, specification and migration of presumptive RGCs, terminal differentiation and maintenance of RGCs. We used immunohistochemical analysis to examine the coexpression of RaxL with ganglion cell markers RA4, cBrn-3, Islet-1 and TUJ1. Proliferating retinal progenitor cells have been shown to express Pax6, CNotch-1 and RaxL (Bao and Cepko, 1997; Chen and Cepko, 2002; Marquardt et al., 2001; Ohuchi et al., 1999). Pax6 resides at or near the top of the hierarchy of genes controlling eye morphogenesis (Callaerts et al., 1997; Halder et al., 1995) and the neurogenic gene CNotch-1 inhibits the specification of RGCs in progenitor cells (Austin et al., 1995; Bao and Cepko, 1997; Henrique et al., 1997). The RA4 antigen is expressed in presumptive RGCs immediately after terminal division of the retinal progenitor cells in the retinal proliferative zone and specifically in this cell type throughout retinal histogenesis (McLoon and Barnes, 1989; Waid and McLoon, 1995). RaxL is expressed in retinal progenitor cells in the VZ (Chen and Cepko, 2002). The RA4-positive presumptive RGCs initiate the migration to the vitreal surface of the NR (McLoon and Barnes, 1989; Waid and McLoon, 1995) and then express cBrn-3 when they reach the intermediate region of the VZ. The members of the Brn-3 POU domain transcription factors, Brn-3a, Brn3b, and Brn-3c, are crucial for retinal development and maintenance (Erkman et al., 1996; Gan et al., 1999; Liu et al., 2000; Xiang, 1998). In the chick retina, cBrn-3c expression appears to be spatiotemporally coincident with RGC formation (Liu et al., 2000). Immunohistochemical analyses using anti-human Brn-3a and Brn-3b antibodies crossreactive to cBrn-3a and cBrn-3b showed that cBrn-3a and cBrn-3b are expressed only in RGCs after they have arrived in the GCL (Liu et al., 2000). Misexpression of these factors is capable of promoting RGC development or differentiation. Islet-1 belongs to the family of LIM homeodomain transcription factors involved in cell fate determination in many different systems. Islet-1 is a marker for migrated RGCs to the vitreal surface and cholinergic amacrine cells (Galli-Resta et al., 1997; Thor et al., 1991). The TUJ1 antibody recognizes neuronal type III b-tubulin, which is expressed in RGCs only after they have migrated to the vitreal surface (Snow and Robson, 1995). The expression patterns of these markers in our present results, combined with previous studies, showed that they are differentially expressed during the stages of RGC development, as summarized in Fig. 10A. (1) Proliferating retinal progenitor cells express Notch-1, Pax6 and RaxL. (2) Presumptive RGCs express the RA4 antigen, immediately after the terminal division of retinal progenitor cells located in the ventricular surface. (3) The RA4-positive presumptive RGCs then migrate to the vitreal surface. When they reach the intermediate region of the VZ, the migrating RGCs begin to express cBrn-3 and maintain its expression throughout the developmental process. (4) The migrating RGCs begin to express Islet-1 when they

K. Sakagami et al. / Mechanisms of Development 120 (2003) 881–895

891

1989; Waid and McLoon, 1995) and demonstrated that RA4 marks RGCs early in the specification and migration stage, RA4 and cBrn-3 mark RGCs in the middle of migration stage, Islet-1 marks RGCs later in the migration stage, and loss of RaxL marks RGCs differentiated stage. 3.2. RaxL plays a role in negative regulation of RGC development

Fig. 10. (A) Summary of gene expression during chick RGC generation, migration and differentiation. RaxL expression is maintained during the newly generated and migrating presumptive RGCs and early steps of migrated RGCs. (B) Schematic representation of effects of the expression of exogenous RaxL or dominant-negative RaxL-EnR. In RaxL infected retina, there are no RGCs due to apoptosis; all cells present in the single layer GCL are displaced amacrine cells. In RaxL-EnR infected retina, although the total number of GCL cells was decreased through apoptosis, both cells are present. VZ, ventricular zone; GCL, ganglion cell layer; INL, inner nuclear layer; ONL, outer nuclear layer; OFL, optic fiber layer.

migrate close to the vitreal surface of the retina. (5) When they have reached the innermost surface of the VZ, neuronal type III b-tubulin can be detected in the migrated RGCs still immunoreactive for RaxL. (6) The terminally differentiated RGCs is characterized by a complete loss of RaxL. Our work confirmed previous studies (McLoon and Barnes,

RaxL infection reduced the number of RGCs at E5.0 without affecting cell proliferation and cell death (Fig. 8A – G). Conversely, inhibition of RaxL by RaxLEnR infection increased RGC production by 1.4-fold. These results suggest that the number of RGCs was inversely related to the level of RaxL activity. The previous study on the molecular mechanisms of RGC specification showed that the Notch-1 transmembrane receptor and its cellsurface ligand Delta are involved in RGC fate specification of retinal progenitor cells without affecting proliferation (Austin et al., 1995; Henrique et al., 1997). Inhibition of the Notch-1 signaling pathway by antisense treatment promoted RGC production (Austin et al., 1995), demonstrating that Notch-1 negatively regulates RGC specification. Our data suggested that RaxL misexpression increased CNotch-1 mRNA levels (Fig. 8M,N). These results suggest that RaxL might, directly or indirectly, regulate RGC specification via Notch signaling. It is also possible that RaxL might inhibit the function of factors such as NeuroD and Ath-5, which are involved in promoting RGC specification. Although RaxL overexpression does not significantly affect proliferation of progenitor cells at E5.0, it has been reported that misexpression of Rx mRNA in Xenopus embryos induces hyperproliferation and ectopic neural retina tissue in the neuroectoderm (Andreazzoli et al., 1999; Mathers et al., 1997). This might be explained by the fact that Xenopus Rx mRNA was introduced into one cell of two-cell stage embryos where Rx was absent. This led to expanding Rx expression domains, resulting in an ectopic eye structure. In contrast, the failure of RaxL to induce proliferation is probably due to the timing of RaxL expression in our experiments. As we introduced RaxL into retinal progenitor cells, which were already expressing RaxL, exogenous RaxL failed to induce proliferation. However, how RaxL negatively regulates RGC development without affecting cell cycle remains to be elucidated. Recently, Chen and Cepko (2002) reported that RaxL is not involved in non-photoreceptor differentiation. When a dominant-negative form of RaxL (EnRaxLDC) was introduced in vivo into the early chick eye using retrovirus, EnRaxLDC caused no significant reduction in expression of markers of non-photoreceptor cells such as RGC, bipolar, horizontal or amacrine cells. Their finding that RaxL is not involved in RGC differentiation is inconsistent with our present results showing the involvement of RaxL in RGC development. This discrepancy may come from observations based on different assay systems. Chen and Cepko

892

K. Sakagami et al. / Mechanisms of Development 120 (2003) 881–895

(2002) examined Brn3a expression as an RGC marker at E9.0 by in situ hybridization, whereas we examined Islet-1 and TUJ1 expression as an RGC marker at E3.5 –E5.0 by immunohistochemical analysis. However, when we examined cBrn-3 expression for E5-9 RaxL or RaxL-EnRinfected retinas by in situ hybridization, there were no apparent changes on cBrn-3 expression in either (data not shown). Based on these observations alone, we might also conclude that RaxL is not involved in RGC differentiation, consistent with their observation. However, when Islet-1 and TUJ1 were used to identify RGCs at E3.5– E5.0 by immunohistochemical analysis, it was obvious that RGC development was significantly affected. Chen and Cepko (2002) have also reported that RaxL is involved in the initiation of photoreceptor differentiation. Since EnRaxL DC caused a significant reduction in expression of photoreceptor cell markers. We also found that RaxL is involved in photoreceptor development, as overexpression of RaxL-EnR affected photoreceptor differentiation. When rhodopsin and iodopsin expression was examined in E18 RaxL-EnR-infected retinas by immunostaining, RaxL-EnR caused an increase in expression of rhodopsin but a reduction in expression of iodopsin (data not shown). These observations demonstrate that RaxL is involved in photoreceptor cell differentiation and probably in cone photoreceptor cell differentiation, being consistent with the finding that RaxL is specifically expressed in cone photoreceptors (data not shown). Taken together, these results demonstrate that RaxL plays stage-specific multiple roles in chick retinal development. However, further studies are needed to clarify these issues. 3.3. RaxL functions in maintenance of RGCs In neural retinas expressing exogenous RaxL, the number of RGCs was reduced to about 65% at E5.0 and 40% at E15 of control numbers, while neural retinas expressing the RaxL-EnR showed an initial increase in the number of RGCs to 40% at E5.0 followed by a significant decrease to 40% at E15 (Fig. 5K). Thus, this significant reduction in retinal cell numbers should occur between E5 and E15. It has been known that two phases of apoptosis in chick retinas occur, peaking at E6 and E12 (Frade et al., 1997; Hughes and McLoon, 1979). Our present TUNEL showed that the first phase of apoptosis was not affected by exogenous or dominant-negative RaxL expression (data not shown). However, we did observe changes in the second phase of apoptosis in experimental retinas. Retinas with exogenous RaxL expression showed increased cell death in the GCL and INL by E12 (Fig. 9D,E,G,H). In dominant-negative RaxL-EnR expressing retinas, apoptosis was evident at E9 (Fig. 9C and Chen and Cepko; 2002) but no longer evident at E15 (Fig. 9I), though it was still observed in the RaxL overexpressing and control retinas (Fig. 9H). These results suggest that RGCs require complete loss of RaxL protein for their maintenance, and that prolonged RaxL expression

triggers cell death by apoptosis. High levels of RaxL expression in retinal progenitors may delay the expression of factors required for the later differentiation of RGCs without affecting proliferation, leading to the accumulation of undifferentiated RGCs in the GCL. These RGCs which had failed to be terminally differentiated in RaxL overexpressing retinas might be killed by apoptosis. In contrast, a reduction in levels of RaxL in retinal progenitors by expression of RaxL-EnR advanced the timing of RGC and displaced amacrine cell differentiation, presumably resulting in enhanced early cell death in the INL that resulted in a balanced, but reduced cell population in the GCL.

4. Experimental procedures 4.1. Chick embryos Fertilized eggs from White Leghorn hens were purchased from local suppliers and incubated at 38.5 8C in a humidified incubator. Chick embryos were staged according to Hamburger and Hamilton (1951). 4.2. Plasmids To generate pEFX3-Fg-RaxL, the EcoRV/BamH I DNA fragment encoding the chicken RaxL was amplified using total RNA extracted from E8 retinas and the appropriate primers (GGGATATCAGATGTTCCTCAATAAGTGTGAGG and GCGGATCCTCAAATGGGCTGCCAGGTC). The PCR product was subcloned into the EcoRV/BamH I site of pEFX3-Fg. To generate pEFX3-Fg-RaxL-EnR, the Nco I DNA fragment encoding the Drosophila Engrailed repressor domain EnR was isolated from pCAGGS-Tlx-EnR (Yu et al., 2000), blunt-ended and subcloned into the bluntended Sal I site of pEFX3-Fg-RaxL. To ensure expression of the fusion protein GST-RaxL, the Asp718/BamH I fragment encoding RaxL was blunt-ended and subcloned into the blunt-ended BamHI/SalI site of pGEX6p-1 vector (Amersham). To generate pGEX-6p-1-cBrn-3a, the DNA fragment encoding 124 –183 amino acids of cBrn-3a (Lindeberg et al., 1997) was amplified by RT-PCR using GGAGGATCCCTGGACCACCTCAACTCTGC and CAGGAATTCTGGGTCCGTCTCTGTCTCG as primers and inserted into the BamH I/EcoR I site of pGEX-6p-1. The DNA fragment containing the chicken Rhodopsin promoter (2 203 , þ 67 bp) was amplified from chick genomic DNA using primers TAGGATCCAGAGGATGTGTGAGGAGG and CTGATATCTCAGTCGCACTCAGTGTCG. The amplified fragment was subcloned into the BamH I/ EcoRV site of pCRII. To obtain the reporter plasmid pGVRhod (2 203 , þ 67), the EcoRV DNA fragment of the promoter region containing PCE-1, an Rx/rax-binding sequence (TAATTG), was subcloned into the BamH I/ EcoRV site of the pGV-B vector. To generate pRCASRaxL, the Cla I DNA fragment encoding full-length RaxL

K. Sakagami et al. / Mechanisms of Development 120 (2003) 881–895

was isolated from pEFX3-Fg-RaxL and subcloned into the Cla I site of the pRCAS(A) vector (Hughes et al., 1987). To generate pRCAS-RaxL-EnR, the Cla I DNA fragment encoding RaxL-EnR was isolated from pEFX3-Fg-RaxLEnR and subcloned into the Cla I site of pRCAS vector. All constructs were verified by DNA sequencing.

893

control eyes and three virus-infected eyes of a given stage were analyzed. Fluorescent images were observed under a fluorescent microscope (Axiophot, Zeiss) or a confocal microscope (Fluoview FV500, Olympus). Two-color images (40 or 60 £ ) were captured digitally from random fields. To count the total and Islet-1-positive cells, 4– 7 images were analyzed.

4.3. Preparation of GST-fusion proteins 4.6. In situ hybridization Fusion proteins of GST-RaxL and GST-cBrn-3a were expressed in Escherichia coli strain BL21(DE3) induced by the addition of 10 mM isopropyl thio-b-galactoside at 37 8C for 1.5 and 3 h, respectively. The proteins were purified on a glutathione Sepharose 4B affinity column and treated with PreScission Protease according to the manufacturer’s instructions (Pharmacia Biotech). The proteins were concentrated by centrifugation at 2500 rpm for 30 min with a Centricon 30 (Amicon), then quantified by SDS-PAGE using albumin as a standard and adjusted to the concentration of 0.1 mg/ml. 4.4. Production of antibodies against RaxL and cBrn-3 Eleven week-old female Japanese white rabbits and 7week old SD male rats were used for generation of antiRaxL and anti-cBrn-3a antiserum, respectively. These animals were immunized by subcutaneous injections of 50 –200 mg of the purified recombinant protein in Freund’s adjuvant (Sigma) thrice at about 3 weeks intervals. Seven days after the fourth injection, sera of the rabbits and rats were collected and tested for antibody specificity by Western blot analysis. As the anti-cBrn-3a antiserum failed to distinguish cBrn-3a from cBrn-3c, we termed cBrn-3 for the antigens recognized by this antibody. 4.5. Histology and immunohistochemistry Control and virus infected embryos or eyeballs at various developmental stages were collected and fixed in 95% methanol/5% acetic acid overnight for paraffin embedding. Eyeballs were sectioned at 10 mm and immunohistochemistry was performed as previously described (Reza et al., 2002). Polyclonal and monoclonal primary antibodies and dilutions used to detect RGCs were as follows: rabbit polyclonal anti-RaxL antiserum at 1:100; rat polyclonal anti-cBrn-3 antiserum at 1:100; mouse monoclonal antiIslet-1 antibody at 1:100 (Hybridoma Bank, 40.2D6); TUJ1, mouse monoclonal anti-neuronal type III b-tubulin antibody at 1:500 (BAbCO); RA4 mouse anti-unknown protein at 1:100 (McLoon and Barnes, 1989); rabbit polyclonal antiSox2 antiserum at 1:200 (Shimada et al., 2003); mouse monoclonal anti-RCAS-gag antibody at 1:100 (Hybridoma Bank, AMV-3C2). Secondary antibodies used were fluorolink Alexa488 and Alexa594 (Molecular Probe) conjugated goat anti-rabbit, mouse or rat IgG at the dilution of 1:500. For section immunostaining experiments, at least three

Section in situ hybridization was performed as previously described (Yu et al., 2000). Digoxigenin (DIG)labeled RNA probes (Promega) were generated according to the manufacturer’s instructions. To generate probes for CNotch-1 and RCAS-envelope mRNA detection, pBSNotch1 and pBS-RCAS/env (generously provided by Dr Ogura) was digested by Xba I and EcoR I, respectively, then transcribed with T7 RNA polymerase for antisense probe. 4.7. Cell cultures, transfection and luciferase assay NIH3T3 cells were cultured at 37 8C in Dulbecco’s modified Eagle’s medium (DMEM) supplemented 10% fetal calf serum (BRL). Three hundred nanograms of the luciferase reporter plasmid (Rhod-Luc), 250 ng b-galactosidase expression plasmid (pEFX3-b-gal), with the appropriate amounts of pEFX3-RaxL and/or pEFX3-RaxL-EnR for 500 ng of total plasmid DNA were transfected onto NIH3T3 cell cultures in 12-well dishes using DEMRI-C according to the manufacturer’s protocol (Life Technologies). After 6 h of transfection, the medium was replaced with 10% FBS/1% chick serum/DEME. Cells were incubated for a further 48 h, then whole cell lysates were prepared in extraction buffer (0.1 ml of 25 mM Trisphosphate, pH7.8, 15% glycerol, 2% CHAPS, 1% L -aphosphatidylcholine, 1% BSA, 4 mM EGTA, 8 mM MgCl2, 1 mM dithiothreitol, 0.4 mM APMSF). The luciferase activities of the lysates were measured with a Microtiter Plate Luminometer (DYNEX). Transfection efficiency in each dish was normalized based on b-galactosidase activity. All luciferase activity values represent mean ^ standard deviation of results obtained from duplicates of three independent transfection experiments. 4.8. Electroporation Electroporation was performed according to the procedure described (Momose et al., 1999). Plasmid DNA solution containing the RCAS retrovirus vector (1.0 mg/ml) was injected into the left optic vesicles of Hamburger and Hamilton stage 10 chick embryos in ovo. Immediately after injection, the embryos were subjected to electroporation by applying two 5 V pulses for 25 ms each using a pulsegenerator (T-820, BTX). Following electroporation, the embryos were incubated until E3 – E15. For each experiment, 20 embryos were electroporated. We checked for

894

K. Sakagami et al. / Mechanisms of Development 120 (2003) 881–895

successful electroporation by performing immunostaining or in situ hybridization. About 70 and 20% of electroporated embryos were alive at E5 and at E15, respectively. We performed at least three independent experiments for each electroporation. 4.9. BrdU pulse-labeling in vivo To label dividing cells in the retinas, BrdU solution (3 ml of 10 mM BrdU) was injected into the optic cavity of E5.0 chick embryos, which were incubated for an additional 3 h before embryo collection. Sections of labeled retinas were processed for immunostaining by detecting BrdU incorporation as follows. Sections were treated with 4 M HCl for 10 min, blocked in 10% goat serum/PBST and then reacted with the primary antibody BrdU Ab-1 (diluted at 1:20; 85208, Neo Marker, Co.), followed by reaction with the secondary antibody Alexa594 (Molecular Probe). Fluorescent images were observed under a confocal microscope. 4.10. TUNEL labeling TUNEL analysis was performed according to the manufacture’s instructions (in situ Apoptosis Detection Kit; Takara). Fluorescent images were observed under a confocal microscope.

Acknowledgements We are grateful to Dr S.C. McLoon for the RA4 antibody, to Dr C.L. Cepko for pCScRax, to Dr T. Ogura for pBS-Notch1 and pBS-RCAS/env, and to Drs I. Smith, J.E. Dowling and H.M. Reza for valuable comments and critical reading of the manuscript. This work was supported in part by Grants-in-Aid for The 21st Century COE Program from the Ministry of Education, Science, Sports and Culture of Japan. K.S. is the recipient of a fellowship from The 21st Century COE Program.

References Andreazzoli, M., Gestri, G., Angeloni, D., Menna, E., Barsacchi, G., 1999. Role of Xrx1 in Xenopus eye and anterior brain development. Development 126, 2451–2460. Austin, C.P., Feldman, D.E., Ida, J.A., Cepko, C.L., 1995. Vertebrate retinal ganglion cells are selected from competent progenitors by the action of Notch. Development 121, 3637– 3650. Badiani, P., Corbella, P., Kioussis, D., Marvel, J., Weston, K., 1994. Dominant interfering alleles define a role for c-Myb in T-cell development. Genes Dev. 8, 770 –782. Bao, Z.Z., Cepko, C.L., 1997. The expression and function of Notch pathway genes in the developing rat eye. J. Neurosci. 17, 1425–1434. Callaerts, P., Halder, G., Gehring, W.J., 1997. Pax-6 in development and evolution. Annu. Rev. Neurosci. 20, 483–532.

Casarosa, S., Andreazzoli, M., Simeone, A., Barsacchi, G., 1997. Xrx1, a novel Xenopus homeobox gene expressed during eye and pineal gland development. Mech. Dev. 61, 187–198. Cepko, C.L., Austin, C.P., Yang, X., Alexiades, M., Ezzeddine, D., 1996. Cell fate determination in the vertebrate retina. Proc. Natl. Acad. Sci. USA 93, 589–595. Chen, C.M., Cepko, C.L., 2002. The chicken RaxL gene plays a role in the initiation of photoreceptor differentiation. Development 129, 5363–5375. Chuang, J.C., Raymond, P.A., 2001. Zebrafish genes rx1 and rx2 help define the region of forebrain that gives rise to retina. Dev. Biol. 231, 13 –30. Dowling, J.E., 1987. The retina—An Approachable part of the Brain, Harvard University Press, Cambridge, MA. Erkman, L., McEvilly, R.J., Luo, L., Ryan, A.K., Hooshmand, F., O’Connell, S.M., Keithley, E.M., Rapaport, D.H., Ryan, A.F., Rosenfeld, M.G., 1996. Role of transcription factors Brn-3.1 and Brn-3.2 in auditory and visual system development. Nature 381, 603 –606. Frade, J.M., Bovolenta, P., Martinez-Morales, J.R., Arribas, A., Barbas, J.A., Rodriguez-Tebar, A., 1997. Control of early cell death by BDNF in the chick retina. Development 124, 3313–3320. Furukawa, T., Kozak, C.A., Cepko, C.L., 1997. rax, a novel paired-type homeobox gene, shows expression in the anterior neural fold and developing retina. Proc. Natl. Acad. Sci. USA 94, 3088–3093. Furukawa, T., Mukherjee, S., Bao, Z.Z., Morrow, E.M., Cepko, C.L., 2000. rax, Hes1, and notch1 promote the formation of Mu¨ller glia by postnatal retinal progenitor cells. Neuron 26, 383–394. Galli-Resta, L., Resta, G., Tan, S.S., Reese, B.E., 1997. Mosaics of islet-1expressing amacrine cells assembled by short-range cellular interactions. J. Neurosci. 17, 7831–7838. Galvez, J.M., Puelles, L., Prada, C., 1977. Inverted (displaced) retinal amacrine cells and their embryonic development in the chick. Exp. Neurol. 56, 151–157. Gan, L., Wang, S.W., Huang, Z., Klein, W.H., 1999. POU domain factor Brn-3b is essential for retinal ganglion cell differentiation and survival but not for initial cell fate specification. Dev. Biol. 210, 469 –480. Halder, G., Callaerts, P., Gehring, W.J., 1995. Induction of ectopic eyes by targeted expression of the eyeless gene in Drosophila. Science 267, 1788–1792. Hamburger, V., Hamilton, H.L., 1951. A series of normal stages in the development of the chick embryo. J. Morphol. 88, 49–92. Han, K., Manley, J.L., 1993. Functional domains of the Drosophila engrailed protein. Embo J. 12, 2723–2733. Henrique, D., Hirsinger, E., Adam, J., Le Roux, I., Pourquie, O., IshHorowicz, D., Lewis, J., 1997. Maintenance of neuroepithelial progenitor cells by Delta– Notch signalling in the embryonic chick retina. Curr. Biol. 7, 661–670. Hughes, W.F., McLoon, S.C., 1979. Ganglion cell death during normal retinal development in the chick: comparisons with cell death induced by early target field destruction. Exp. Neurol. 66, 587 –601. Hughes, S.H., Greenhouse, J.J., Petropoulos, C.J., Sutrave, P., 1987. Adoptor plasmid simplify the insertion of foreign DNA into HelperIndependent retroviral vectors. J. Virol. 61, 3004–3012. Jean, D., Ewan, K., Gruss, P., 1998. Molecular regulators involved in vertebrate eye development. Mech. Dev. 76, 3–18. Kanekar, S., Perron, M., Dorsky, R., Harris, W., Jan, L., Jan, Y., Vetter, M., 1997. Xath5 participates in a network of bHLH genes in the developing Xenopus retina. Neuron 19, 981 –994. Kimura, A., Singh, D., Wawrousek, E.F., Kikuchi, M., Nakamura, M., Shinohara, T., 2000. Both PCE-1/RX and OTX/CRX interactions are necessary for photoreceptor-specific gene expression. J. Biol. Chem. 275, 1152–1160. Le, R.D., Rayner, K., Rex, M., Wigmore, P.M., Scotting, P.J., 2002. The transcription factor cSox2 and Neuropeptide Y define a novel subgroup of amacrine cells in the retina. J. Anat. 200, 51–56.

K. Sakagami et al. / Mechanisms of Development 120 (2003) 881–895 Lindeberg, J., Klint, P., Williams, R., Ebendal, T., 1997. Identification of a chicken homologue in the Brn-3 subfamily of POU-transcription factors. Brain Res. Dev. Brain Res. 100, 169 –182. Liu, W., Khare, S.L., Liang, X., Peters, M.A., Liu, X., Cepko, C.L., Xiang, M., 2000. All Brn3 genes can promote retinal ganglion cell differentiation in the chick. Development 127, 3237–3247. Liu, W., Mo, Z., Xiang, M., 2001. The Ath5 proneural genes function upstream of Brn3 POU domain transcription factor genes to promote retinal ganglion cell development. Proc. Natl. Acad. Sci. USA 98, 1649–1654. Marquardt, T., Gruss, P., 2002. Generating neuronal diversity in the retina: one for nearly all. Trends Neurosci. 25, 32–38. Marquardt, T., Ashery-Padan, R., Andrejewski, N., Scardigli, R., Guillemot, F., Gruss, P., 2001. Pax6 is required for the multipotent state of retinal progenitor cells. Cell 105, 43–55. Masland, R.H., 2001. The fundamental plan of the retina. Nat. Neurosci. 4, 877–886. Mathers, P.H., Jamrich, M., 2000. Regulation of eye formation by the Rx and pax6 homeobox genes. Cell Mol. Life Sci. 57, 186–194. Mathers, P.H., Grinberg, A., Mahon, K.A., Jamrich, M., 1997. The Rx homeobox gene is essential for vertebrate eye development. Nature 387, 603–607. McCabe, K.L., Gunther, E.C., Reh, T.A., 1999. The development of the pattern of retinal ganglion cells in the chick retina: mechanisms that control differentiation. Development 126, 5713–5724. McLoon, S.C., Barnes, R.B., 1989. Early differentiation of retinal ganglion cells: an axonal protein expressed by premigratory and migrating retinal ganglion cells. J. Neurosci. 9, 1424–1432. Mey, J., Thanos, S., 2000. Development of the visual system of the chick. I. Cell differentiation and histogenesis. Brain Res. Brain Res. Rev. 32, 343–379. Momose, T., Tonegawa, A., Takeuchi, J., Ogawa, H., Umesono, K., Yasuda, K., 1999. Efficient targeting of gene expression in chick embryos by microelectroporation. Dev. Growth Differ. 41, 335–344. Norris, R.A., Kern, M.J., 2001. Identification of domains mediating transcription activation, repression, and inhibition in the paired-related homeobox protein, Prx2 (S8). DNA Cell Biol. 20, 89–99.

895

Ohuchi, H., Tomonari, S., Itoh, H., Mikawa, T., Noji, S., 1999. Identification of chick rax/rx genes with overlapping patterns of expression during early eye and brain development. Mech. Dev. 85, 193– 195. Perron, M., Opdecamp, K., Butler, K., Harris, W.A., Bellefroid, E.J., 1999. X-ngnr-1 and Xath3 promote ectopic expression of sensory neuron markers in the neurula ectoderm and have distinct inducing properties in the retina. Proc. Natl. Acad. Sci. USA 96, 14996–15001. Prada, C., Puga, J., Perez-Mendez, L., Lopez, R., Ramirez, G., 1991. Spatial and temporal patterns of neurogenesis in the chick retina. Eur. J. Neurosci. 3, 559–569. Reza, H.M., Ogino, H., Yasuda, K., 2002. L-Maf, a downstream target of Pax6, is essential for chick lens development. Mech. Dev. 116, 61– 73. Shimada, N., Aya-Murata, T., Reza, M.H., Yasuda, K., 2003. Cooperative action between L-Maf and Sox2 on d-crystallin gene expression during chick lens development. Mech. Dev. 120, 455–465. Silva, A.O., Ercole, C.E., McLoon, S.C., 2002. Plane of cell cleavage and numb distribution during cell division relative to cell differentiation in the developing retina. J. Neurosci. 22, 7518–7525. Snow, R.L., Robson, J.A., 1994. Ganglion cell neurogenesis, migration and early differentiation in the chick retina. Neuroscience 58, 399–409. Snow, R.L., Robson, J.A., 1995. Migration and differentiation of neurons in the retina and optic tectum of the chick. Exp. Neurol. 134, 13 –24. Thor, S., Ericson, J., Brannstrom, T., Edlund, T., 1991. The homeodomain LIM protein Isl-1 is expressed in subsets of neurons and endocrine cells in the adult rat. Neuron 7, 881 –889. Waid, D.K., McLoon, S.C., 1995. Immediate differentiation of ganglion cells following mitosis in the developing retina. Neuron 14, 117– 124. Xiang, M., 1998. Requirement for Brn-3b in early differentiation of postmitotic retinal ganglion cell precursors. Dev. Biol. 197, 155–169. Young, R.W., 1985. Cell differentiation in the retina of the mouse. Anat. Rec. 212, 199–205. Yu, R.T., Chiang, M.Y., Tanabe, T., Kobayashi, M., Yasuda, K., Evans, R.M., Umesono, K., 2000. The orphan nuclear receptor Tlx regulates Pax2 and is essential for vision. Proc. Natl. Acad. Sci. USA 97, 2621–2625.