Mouse Rab11-FIP4 regulates proliferation and differentiation of retinal progenitors in a Rab11-independent manner

Abstract

We identified Rab11-family interacting protein 4 (Rab11-FIP4) as a gene strongly expressed in the developing mouse retina. The major transcript encoding a full-length protein, mRab11-FIP4A, was expressed predominantly in neural tissues; whereas an alternative transcript encoding an N-terminally truncated form of the protein, mRab11-FIP4B, was expressed ubiquitously as a minor form. Gain-of-function of mRab11-FIP4A in retina promoted cell cycle exit and increased subpopulations of retinal cells localized in the inner nuclear layer, such as bipolar cells and Müller glia. Reversal of the phenotype was observed in the loss-of-function experiment. Furthermore, Shh signaling was suggested to be involved in these functions. Analysis using truncation mutants revealed the essential role of the N-terminal region containing a conserved EF-hand motif for the retinal phenotypes induced by the expression of mRab11-FIP4A, whereas binding to Rab11 was dispensable, suggesting the involvement of a novel Rab11-independent mechanism for mRab11-FIP4A action in the regulation of retinal development. Developmental Dynamics 236:214–225, 2007. © 2006 Wiley-Liss, Inc.

INTRODUCTION

The vertebrate neural retina contains six major types of neurons and a single type of glial cell; and these cells are organized into three nuclear layers: the outer nuclear layer (ONL), consisting of cone and rod photoreceptors; the inner nuclear layer (INL), consisting of three types of interneurons, i.e., bipolar, amacrine, and horizontal cells, as well as the Müller glial cells; and the ganglion cell layer (GCL), consisting of retinal ganglion cells (RGC) and a small number of displaced amacrine cells. Although all of these types of retinal cells are thought to be derived from a single population of retinal progenitor cells (Perron and Harris, 2000; Ahmad et al., 2004), the generation of each type occurs at distinct stages during eye development and its timing is strictly regulated. To identify genes regulated for expression in a retinal developmental stage-specific manner, we conducted differential display analysis using cDNA prepared from various stages of mouse eyes as templates. As a result, we found that a gene named rab11-family interacting protein 4 (Rab11-FIP4) was expressed with a unique pattern in the developing retina.

Rab11-FIP4 was originally cloned as a gene named KIAA1821 in a human cDNA sequencing project (Nagase et al., 2001) and recently identified as a member of the Rab11-family interacting protein (Rab11-FIPs) family (Wallace et al., 2002a). Rab11 is a member of the Rab family of small GTPases and is known to regulate diverse pathways of vesicle trafficking including protein recycling and intracellular protein transport (Zerial and McBride, 2001) as well as cytokinesis (Riggs et al., 2003). At least six members of Rab11-FIPs, which all share a highly conserved short motif named Rab11-binding domain (RBD) at the C-termini of the proteins, have been so far been identified (Hales et al., 2001; Prekeris et al., 2001; Lindsay et al., 2002; Wallace et al., 2002b), and these molecules are thought to play important roles in the regulation of the vesicle trafficking by Rab11 as downstream effectors (Cullis et al., 2002; Meyers and Prekeris, 2002). Rab11-FIP4 and a closely related protein Rab11-FIP3 are categorized into class II subfamily of Rab11-FIPs by the presence of an N-terminal EF-hand motif as well as by a similarity to the Drosophila protein Nuclear Fallout (Nuf), which was identified as a gene required for cellularization (Rothwell et al., 1998; Riggs et al., 2003). In vitro studies failed to reveal the involvement of Rab11-FIP4 in the Rab11-mediated intracellular transport of transferrin (Wallace et al., 2002b) or its receptor (Hickson et al., 2003), whereas it was demonstrated that Rab11-FIP4 in cooperation with Rab11-FIP3 was required for cytokinesis (Fielding et al., 2005). However, these studies were performed in vitro using human cell lines and the physiological function of Rab11-FIP4 during the vertebrate development is still unclear. We recently identified a zebrafish orthologue of Rab11-FIP4 (zRab11-FIP4) as a gene specifically expressed in the neural tissues, including the developing retina, and found that the amino acid sequence as well as protein motifs were highly conserved between human and zebrafish (Muto et al., 2006). In more detailed analysis, we showed that zRab11-FIP4 plays an important role in regulating the proliferation and differentiation of the retinal progenitors during development. Also, functional interaction between zRab11-FIP4 and shh was indicated. These observations, therefore, have prompted us to examine whether these biological functions of Rab11-FIP4 are conserved in other vertebrates.

Here, we report the identification and characterization of mouse Rab11-FIP4 (mRab11-FIP4). In this study, we found mRab11-FIP4A, a longer form of mRab11-FIP4, to be involved in retinal development. mRab11-FIP4A was predominantly expressed in the developing neural tissues, and gain- and loss-of-function analyses using a retinal explant system indicated that this molecule played roles in regulating the exit of retinal progenitor cells from the cell cycle and their subsequent differentiation into INL cells. We also found that Rab11 could not mimic the phenotype observed by the overexpression of mRab11-FIP4. These results suggest that, although Rab11-FIP4 has been structurally and functionally conserved among vertebrates, the detailed mechanism underlying its action between zebrafish and mouse Rab11-FIP4 appears to be different.

RESULTS

Identification and Cloning of Mouse DD76 cDNA

To identify genes the transcriptional level of which is regulated during eye development, we performed differential display using cDNAs prepared from RNAs obtained from mouse eyes at various stages of development. We sequenced 91 clones that had unique expression patterns during retinal development and focused on a fragment of 221 bp, which we named DD76. Homology search against the Celera and public DNA databases showed that the sequence of this DD76 fragment corresponded to an intron region of a putative mouse gene similar to that of human (h) Rab11-FIP4. We isolated the full-length cDNA by reverse transcriptase-polymerase chain reaction (RT-PCR) in combination with 5′-rapid amplification of cDNA ends (RACE) and found it to encode a putative protein of 635 amino acids with 91% identity to the amino acid sequence of hRab11-FIP4 (Fig. 1A). A motif search identified a single EF-hand motif and the RBD in regions proximal to the N- and C-termini, respectively. A coiled-coil structure and two leucine-zipper like motifs were identified within the C-terminal half. All these characteristic features are conserved in human and zebrafish Rab11-FIP4 (Muto et al., 2006). Moreover, a search of the genomic databases mapped the DD76 gene to the mouse chromosome 11B5 region, which is syntenic to a locus encompassing the hRab11-FIP4 gene on human chromosome 17q11.2 (data not shown). All these results strongly indicate that the DD76 gene is a mouse homologue of hRab11-FIP4; therefore hereafter, we refer to it as mouse Rab11-FIP4 (mRab11-FIP4).

Figure 1. Structure and expression pattern of Rab11-FIP4. A: Schematic representation of protein structures of Rab11-FIP4 homologues. The identity at the amino acid level and the positions of conserved motifs are indicated. The mouse Rab11-FIP4B–specific 5′-region is indicated by the hatched box. B: Northern blot analysis of mRab11-FIP4 in the developing eye and brain was performed by using a probe recognizing both A- and B-forms. Ten micrograms of total RNA prepared from the indicated tissues was applied in each lane. The sizes of RNA markers are shown at the right of the panel. C: Expression of A- and B-forms of mRab11-FIP4 as well as mRab11-FIP3 in various tissues of postnatal day (P) 1 mice (lower) and adult mice (upper) was examined by semiquantitative reverse transcriptase-polymerase chain reaction (RT-PCR). Glyceraldehyde-3-phosphate dehydrogenase (G3PDH) was used as a control in both “B” and “C.”

By the 5′-RACE method, we found an alternative transcript, which was predicted to carry 185 bp of a unique 5′-sequence in place of the N-terminal region containing the EF-hand motif. A putative initiation codon was found in its unique 5′-sequence. We designated this N-terminal–truncated form as Rab11-FIP4B, and the full-length one as Rab11-FIP4A (Fig. 1A).

Rab11-FIP4 Is Expressed Predominantly in the Neural Tissues

By Northern blot analysis using a 1.5-kb cDNA fragment from a region shared by Rab11-FIP4 A- and B-forms as a probe, we found a single 7-kb band in samples derived from the eyes and brain of a postnatal day (P) 1 mouse. The bands were surmised to contain both A- and B-form transcripts because of a slight difference in their mobility (Fig. 1B). As development proceeded, the amount of mRab11-FIP4 mRNA in the eyes gradually increased up to the end of the embryonic period (data not shown); then, after postnatal day 1 (P1), it decreased toward the adult stage (Fig. 1B). In contrast, that in the brain did not change significantly (Fig. 1B). We then examined the expression pattern of mRab11-FIP4 in a variety of P1 and adult mouse tissues by RT-PCR using primers specific for each form of transcript. The A-form in both P1 and adult was predominantly expressed in neural tissues such as eye and brain, whereas the B-form transcript was detected in a broad range of tissues, except the heart and muscle (Fig. 1C).

Next, the spatial and temporal expression pattern of Rab11-FIP4 in developing mouse embryos was examined by whole-mount in situ hybridization. Since we failed to detect A and B transcripts separately by using specific probes, probably due to the short length of the probes (data not shown), we adopted the same probe used for the Northern blot analysis, which hybridized with both A- and B-forms, for the in situ hybridization. At embryonic day (E) 8.5, the mRab11-FIP4 gene was clearly expressed in the region along the midline (Fig. 2A). On the next day (E9.5), the expression level became much higher; and the transcript was detected predominantly in the central nervous system (CNS), including the entire brain, neural tube, and otic vesicles. At this stage, the optic vesicle started to evaginate, and mRab11-FIP4 was strongly expressed in this region (Fig. 2B). In addition to these neural tissues, the branchial arches also expressed Rab11-FIP4 at this stage. At E10.5, the expression pattern was essentially the same as that in E9.5 embryos; weak expression in fore- and hindlimbs was additionally observed (Fig. 2C).

Figure 2. Expression of class II Rab11-FIPs in the developing mouse. A–O: The expression of mRab11-FIP4 in the developing mouse. A–C: Results of whole-mount in situ hybridization of mouse embryos at embryonic day (E) 8.5 (A), E9.5 (B), and E10.5 (C). D–L: Sections subjected to in situ hybridization for detection of mRab11-FIP4 mRNA. Transverse sections of embryonic heads at E10.5 (D), E11.5 (E), E12.5 (F), E13.5 (G), and E16.5 (H), and retinal sections prepared from postnatal mice at postnatal day (P) 1 (I), P5 (J), P10 (K), and adult (L) were examined by in situ hybridization. M–O: Expression of mRab11-FIP4 in areas around the forebrain (M), nose (N), and the inner ear (O) at E16.5 was analyzed by using transverse sections. P–V: Expression of mRab11-FIP3 in the developing retina. Expression pattern of mRab11-FIP3 was assessed using cryosections from embryonic mice (P–R) and postnatal retina (S–V). al, allantois; ba, branchial arch; co, cochlea; cp, cortical plate; di, diencephalon; fl, forelimb bud; gcl, retinal ganglion cell layer; h, heart; hl, hindlimb bud; inl, inner nuclear layer; iz, intermediate zone; le, lens; nbl, neuroblastic layer; nc, nasal cavity; oe, olfactory epithelium; onl, outer nuclear layer; op, optic vesicle; ot, otic vesicle; pe, retinal pigment epithelium; sc, spinal cord; st, striatum; tg, trigeminal ganglion; vg, vagal ganglion; vz, ventricular zone. Scale bars = 400 μm in A–C, 100 μm in D–V.

To analyze in more detail the expression pattern of mRab11-FIP4 in the developing mouse eye, we performed in situ hybridization on tissue sections. At E10.5, expression was found in the retinal pigment epithelium (RPE) but not in the neural retina (Fig. 2D); that in the RPE gradually declined and disappeared by E13.5 (Fig. 2E–G). The expression of mRab11-FIP4 in the neural retina was first detected at E11.5 in a small number of cells in the central area of the retina (Fig. 2E), and a large number of mRab11-FIP4–expressing cells were observed by E12.5 in the GCL (Fig. 2F). The expression in the GCL expanded peripherally at later stages, and finally was detected throughout the retina by E16.5 (Fig. 2H). On the other hand, mRab11-FIP4 started to be expressed in the neuroblastic layer (NBL) around E13.5, and its expression there increased by E16.5 (Fig. 2H). Rab11-FIP4 continued to be expressed in the GCL after birth; however, at later stages, the expression in the NBL gradually disappeared from the outer side (Fig. 2J,K) concomitant with differentiation of photoreceptors, and finally disappeared from the adult retina (Fig. 2L).

In the brain at E16.5, expression of mRab11-FIP4 was observed in the cortical plate (Fig. 2M), as well as in the olfactory epithelium and in some peripheral nerve ganglia (Fig. 2N,O). In addition, otic and nasal capsules expressed mRab11-FIP4 mRNA (Fig. 2O).

In contrast, the closely related gene mRab11-FIP3 (GenBank accession no. AB093257) was expressed ubiquitously in P1 and adult mice when examined by RT-PCR (Fig. 1C). During development, mRab11-FIP3 was expressed throughout the retina in all stages for all the times we examined, although the expression level was relatively higher in RGC (later than E13.5) and INL (later than P5; Fig. 2P–V). It is, therefore, indicated that the predominant expression in neural tissues is a pattern unique to mRab11-FIP4.

mRab11-FIP4 Regulates the Fate Decision of Retinal Cell

To examine the function of mRab11-FIP4 for retinal development, we overexpressed either the A- or B-form of mRab11-FIP4 in retinal explants by using a retrovirus vector containing enhanced green fluorescent protein (EGFP) -driven by the internal ribosomal entry site (IRES; Ouchi et al., 2005). Retinal explants prepared from E17.5 mouse eyes were infected with the retrovirus and cultured for 2 weeks. We first examined the localization of virus-infected cells by examining frozen sections of the explants. When the control virus was used for the infection, more than 80% of the EGFP-positive (virus-infected) cells were localized in the ONL, and the rest of them were found in the INL (Fig. 3A,B). On the other hand, by overexpressing mRab11-FIP4A, we found the proportion of EGFP-positive cells localized in the INL to be significantly increased (Fig. 3A,B). This phenotype was not observed by overexpression of the B-form. We then examined differentiation of the virus-infected cells by immunostaining sections with antibodies against various marker proteins of the subpopulations of retinal cells. Antibodies used were anti-rhodopsin (rod photoreceptors), -HuC/D (RGC and amacrine cells), -PKC (bipolar cells), -glutamine synthetase (GS; Müller glia), and -cyclin D3 (Müller glia) antibodies. The relative proportions of Müller glia, bipolar cells, and amacrine cells were increased by the expression of mRab11-FIP4A (Fig. 3D,E). We observed neither the mislocalization of photoreceptors (Fig. 3D) nor the enhanced cell death in the outer region of NBL or ONL at any of the stages examined (Fig. 3F).

Figure 3. Overexpression of mRab11-FIP4A leads to an increase in the population of inner nuclear layer (INL) cells at the expense of outer nuclear layer (ONL) cells. A: Retinal explants prepared from embryonic day (E) 17.5 mouse embryos were infected with retroviruses encoding mRab11-FIP4A, mRab11-FIP4B, or mRab11-FIP4AΔRBD, and enhanced green fluorescent protein (EGFP) -positive virus-infected cells were examined by immunohistochemistry in sections prepared from 14-day cultures. DAPI (4′,6-diamidine-2-phenylidole-dihydrochloride) was used to visualize nuclei. B: Proportions of EGFP-positive cells in each layer to total EGFP-positive cells in the sections. C: 293T cells were transfected with pMX-mRab11-FIP4 (A, B or ΔRBD) -internal ribosomal entry site (IRES) -EGFP plasmid and the expression of each form of mRab11-FIP4 was examined by Western blotting using an anti–mRab11-FIP4 antibody. EGFP was used as a control. D: Differentiation into each retinal cell type was examined immunohistochemically. Markers used are rhodopsin for rod photoreceptors, PKC for bipolar cells, HuC/D for retinal ganglion cells (RGC) and amacrine cells, and glutamine synthetase (GS) and cyclin D3 for Müller glia. E: Proportions of marker-positive cells among total EGFP-positive cell population. F: Cell death in the virus-infected retina prepared from E17.5 mice examined by TUNEL (terminal deoxynucleotidyl transferase–mediated deoxyuridinetriphosphate nick end-labeling) assay at the indicated days. G: Effect of mRab11-FIP4A expression on subretinal localization of retinal cells examined at different developmental stages. Retinal explants prepared from E17.5 and E15.5 embryos were infected with control or pMXc-mRab11-FIP4A virus and analyzed for the localization of the EGFP-positive cells at the indicated times. Corresponding stages of the in vivo development are indicated between the two panels. Error bars indicate SEM. *P < 0.01; **P < 0.05.

Analysis of the localization of cells at various time points revealed that the increase in the number of cells expressing mRab11-FIP4 in the INL was first recognized at day 5 of the culture period (Fig. 3G). When we used a retinal explant prepared from an E15.5 embryo, which was 2 days younger than the retina used in the experiment just described, an increase in the proportion of cells located in the INL was observed from day 7, indicating that the onset of abnormal sublocalization of the mRab11-FIP4A–expressing cells corresponded to P3 of the developmental stage. The decline in the expression of the endogenous mRab11-FIP4 in the outer region of the NBL between P1 and P5 (Fig. 2I,J) thereby suggests that a persistent expression of mRab11-FIP4A during this period may lead retinal progenitors to differentiate into cells in the INL.

We next examined the effects of down-regulation of the expression of mRab11-FIP4 by using small hairpin RNAs (shRNA) directed against both (ABi) or either (Ai and Bi) of the two forms of mRab11-FIP4 mRNAs (Fig. 4B). For this loss-of-function study, we used the retrovirus vector pSSCG (Yamamichi et al., 2005), which contains the U6 promoter to express shRNA and the EGFP gene driven by the CMV promoter. We examined the efficiencies and specificities of these shRNAs in 293T cells and confirmed that all of shRNAs acted as expected (Fig. 4B). Then, retinal explants prepared from E17.5 mice were infected with the virus encoding these shRNA. When both forms of mRab11-FIP4 were simultaneously knocked down by expressing ABi in the explants, the proportion of EGFP-positive cells in the INL was significantly decreased, and most of them were localized in the ONL (Fig. 4C,D). Subsequent analysis using Ai or Bi revealed that only the Ai had the effect of decreasing the number of INL-located cells. This phenotype is the reverse of that found when mRab11-FIP4A was overexpressed, which suggests that mRab11-FIP4A plays an important role in regulation of differentiation into retinal subpopulations localized in the INL and ONL.

Figure 4. mRab11-FIP4A is required for the differentiation of cell types in the inner nuclear layer (INL). A: Schematic showing regions in mRab11-FIP4 targeted by small hairpin RNAs (shRNA). B: Specificity of shRNAs analyzed in 293T cells. 293T cells were cotransfected with mRab11-FIP4A or mRab11-FIP4B and a vector for the expression of shRNA, and the expression of mRab11-FIP4A and B was examined by Western blotting using total cell lysates (upper panel). The CBB staining pattern of the membrane is shown in the lower panel. C,D: Effect of down-regulation of the endogenous mRab11-FIP4A and/or mRab11-FIP4B on retinal development. C: Retinal explants prepared from embryonic day (E) 17.5 embryos were infected with retroviruses encoding the indicated shRNAs, and subretinal localization was examined in sections prepared from 14-day cultures. D: The proportion of EGFP-positive cells in each layer to total EGFP-positive cells was then determined. Error bars indicate SEM. *P < 0.01.

mRab11-FIP4A Promotes Exit From the Cell Cycle

The determination of cell fate is closely related to the timing of exit from the cell cycle. Therefore, next we analyzed the effect of knockdown or overexpression of mRab11-FIP4A on retinal cell proliferation by examining the expression of Ki67, which is a nuclear antigen expressed in the proliferating cells (Gerdes et al., 1983; Fig. 5B). In the retinal explants infected with either one of the control viruses, approximately 25–27% of the EGFP-positive cells expressed Ki67 at day 4 of the culture period (Fig. 5). We found that the population of proliferating cells was decreased when mRab11-FIP4A was overexpressed, whereas it was increased when mRab11-FIP4A was knocked down, indicating that mRab11-FIP4A promoted the retinal progenitors to exit from the cell cycle. This result may account for the observation that the number of EGFP-positive cells was lower in the Rab11-FIP4A–expressing retina than that in control at day 14, despite equivalent numbers of infected cells at day 3 (Fig. 3A,G). As to the mechanism responsible for this phenomenon, we examined whether down-regulation of mRab11-FIP4 would affect the expression of p27Kip1 or not and found that its expression was partially suppressed in the ABi shRNA-expressing cells (Fig. 5C,D).

Figure 5. mRab11-FIP4A promotes cell cycle exit of the retinal cells. A: Effect of overexpression (left column) and down-regulation (right column) of mRab11-FIP4A on the proliferation of retinal progenitors was analyzed in terms of Ki-67 expression. Retinal explants prepared from embryonic day (E) 17.5 embryos were infected with the indicated retroviruses, and proliferating cells in 4-day cultures were identified immunohistochemically by using anti–Ki-67 antibody. B: Proportions of Ki-67–positive cells to total enhanced green fluorescent protein (EGFP) -positive cell population are shown. C: Effect of mRab11-FIP4 on the expression of p27Kip1 in 4-day cultures was examined immunohistochemically by using anti-p27Kip1 antibody. D: Proportions of p27Kip1-positive cells to total EGFP-positive cell population are shown. Error bars indicate SEM. *P < 0.01, **P < 0.05.

mRab11-FIP4A Regulates the Expression of Genes Involved in Retinal Development

We showed here that mRab11-FIP4A influenced both proliferation and differentiation of developing retinal cells. To gain insight into the mechanism of mRab11-FIP4A–dependent regulation of retinal development, cells ectopically expressing mRab11-FIP4A were sorted as retrovirus-infected (EGFP-positive) cells from retinal explants by flow cytometry and the gene expression patterns in the isolated cells were compared with those in cells sorted from the control virus-infected explants. Retinal explants prepared from E17.5 embryos were infected with either control or mRab11-FIP4A virus and trypsinized to prepare single cell suspension at day5, when the effect of mRab11-FIP4A expression first appeared (Fig. 3G). Proportions of EGFP-positive cells were approximately 24%, 8%, and 22% in control, mRab11-FIP4A–expressing, and mRab11-FIP4B–expressing retinal explants, respectively (Fig. 6A). After staining with propidium iodide (PI), a fluorescent reagent specifically staining dead cells, virus-infected (EGFP-positive) and living (PI-staining negative) cells were enriched by cell sorting using FACSAria (BD Bioscience). In each sample, 4 × 10⁵ cells expressing EGFP at a high level were isolated from 1 × 10⁷ dissociated cells (corresponding to 11 retinal explants) and more than 90% of isolated cells were indeed EGFP-positive (data not shown). cDNA prepared from the isolated cells was subjected to RT-PCR, in which the amount of cDNA was equalized first by the amount of RNA used and then compensated by the expression level of glyceraldehyde-3-phosphate dehydrogenase (G3PDH; Fig. 6B).

Figure 6. Effects of mRab11-FIP4A expression on the expression patterns of genes involved in the retinal development. A: Schematic diagram of cell sorting to isolate retrovirus-infected cells from retinal explants. Retrovirus-infected retinal explants prepared from embryonic day (E) 17.5 embryos were trypsinized at day 5 to prepare single cell suspension and a population of cells expressing enhanced green fluorescent protein (EGFP) at high level was sorted by flow cytometry using FACSAria (BD Bioscience). Proportions of EGFP-positive cells in total retinal cells were indicated. A total of 4 × 10⁵ EGFP-positive cells were sorted from each sample. Total RNA was prepared from the sorted cells, and the gene expression pattern was examined by reverse transcriptase-polymerase chain reaction (RT-PCR). B: Expression of genes involved in the retinal development was analyzed by the semiquantitative RT-PCR. The amount of cDNA used was normalized by the expression of glyceraldehydes-3-phosphate dehydrogenase (G3PDH).

We first examined the growth-related genes, including cyclinD1, p27Kip1, and p57Kip2 (Fig. 6B). As expected, the expression level of cyclin D1 was decreased and those of p27Kip1 and p57Kip2 were increased by ectopic expression of mRab11-FIP4A. The expression of p27Kip1 was apparently, but only slightly, up-regulated, probably because this gene is already expressed in more then 80% of control cells (Fig. 5D). Consistently, in mRab11-FIP4A–expressing cells, the expression of Pax6 and Rx, which are known to be expressed in the retinal progenitors (Ahmad et al., 2004), was decreased. Moreover, as expected from the effects of mRab11-FIP4A on retinal cell differentiation, Müller glial markers, GS, and Vimentin (Hojo et al., 2000), were up-regulated, whereas the expression of rhodopsin was down-regulated (Fig. 6B).

When we assessed the expression of Wnt signaling-related genes, including several ligands (Wnt), receptors (Mfz), and soluble antagonists (Sfrp), the expression of Wnt-5a and -5b, Mfz-3, and Sfrp-3 were significantly enhanced in mRab11-FIP4A–expressing cells, whereas Mfz-4 and Sfrp-2 were detected in the cells at a comparable level to those in control cells. While Wnt signaling is known to regulate proliferation of retinal progenitors during embryonic stages, genes related to Wnt signaling were shown to be expressed even in the perinatal and adult retina, where most cells have ceased proliferating, with patterns unique to each gene (Liu et al., 2003): i.e., Wnt-5a and -5b, Mfz-3, and Sfrp-3 were expressed only in the INL but not in ONL, whereas Mfz-4 was expressed in both INL and ONL. Therefore, given the results obtained from the immunohistochemical analysis, the expression patterns of the Wnt-related genes we observed could be due to an increased proportion of INL cells induced by mRab11-FIP4A expression.

We recently demonstrated that Rab11-FIP4A regulated the retinal development by means of the activation of Sonic Hedgehog (Shh) signaling in zebrafish (Muto et al., 2006). To evaluate whether the expression of mRab11-FIP4A affected on Shh signaling in the mouse retina, we next examined the expression level of Shh as well as its target genes, including Gli family of transcription factors (gli1 to 3) and patched-1 (ptc1). We found that the expression of ptc1 was decreased in cells expressing mRab11-FIP4A (Fig. 6B), thereby indicating that Shh signaling was likely to be impaired in the cells expressing mRab11-FIP4A. In addition, while the expression levels of shh, gli1, and gli2 were not affected, the gli3 was significantly up-regulated when the mRab11-FIP4A was ectopically expressed (Fig. 6B). Gli3 is known to be a negative regulator of Shh signaling (Lipinski et al., 2006), suggesting that mRab11-FIP4A may regulate Shh function negatively by means of the induction of Gli3. The expression patterns of these genes were not affected by expressing mRab11-FIP4B, further indicating the importance of the N-terminal domain on the retinal development (Fig. 6B).

Interaction With Rab11 Is Dispensable for the Regulation of the Retinal Development by mRab11-FIP4

We next examined whether the RBD found in the C-terminal end of Rab11-FIP4 was functional to interact with Rab11 or not. Fusion proteins of the glutathione-S transferase (GST) protein fused to either the N-terminal 375 amino acid (a.a.; GST-4N) or C-terminal 260 a.a. (GST-4C) of mRab11-FIP4 (Fig. 7A) were purified, and the interaction of these proteins with Rab11 was examined by conducting a pull-down assay (Fig. 7B). EGFP-fused human Rab11 protein (EGFP-hRab11), but not control EGFP, was coprecipitated with the GST-4C protein. In contrast, neither the GST-4N protein (Fig. 7B) nor GST (data not shown) was coprecipitated with EGFP-hRab11. The binding of the GST-4C to EGFP-hRab11 was completely abolished by deleting a C-terminal 30-a.a. region encompassing RBD from GST-4C (GST-4CΔRBD), indicating that the RBD was indeed essential for the interaction of Rab11-FIP4 with Rab11. This result is consistent with the previously reported observations on other Rab11-binding proteins (Prekeris et al., 2001).

Figure 7. Neither binding to Rab11 nor Rab11 activity is required for the effect of mRab11-FIP4A on the sublayer localization of retinal cells. A: Schematic representation of the glutathione-S transferase (GST) -Rab11-FIP4–fusion proteins used for pull-down assay. B: Pull-down assay. Whole protein extracts prepared from PLAT-E cells expressing either wild-type enhanced green fluorescent protein (EGFP) -Rab11 or EGFP were incubated with the purified GST-mRab11-FIP4–fusion proteins and then affinity-purified by using glutathione–Sepharose beads. EGFP-Rab11 and EGFP coprecipitated with the GST-fusion proteins were analyzed by Western blotting using an anti-GFP antibody (upper panel). The GST-fusion proteins were visualized by CBB staining after having been transferred onto the membrane (lower panel). The asterisk indicates degradation products of GST-4C and GST-4CΔRBD proteins. C: Myc-tagged mRab11-FIP4A was expressed solely (control, a–c) or coexpressed with wild-type (WT, d–f), constitutively active (Q70L, g–i), or dominant-negative (S25N, j–l) Rab11 fused to EGFP in COS7 cells, and their localization was examined by immunostaining with anti-myc (red) and anti-EGFP (green) antibodies. Nuclei were visualized with DAPI (4′,6-diamidine-2-phenylidole-dihydrochloride). D–G: Rab11 mutants fused to EGFP, i.e., wild-type (WT), dominant-negative (S25N), and constitutively active (Q70L) Rab11, were expressed in retinal explants prepared from E17.5 embryos. D–G: Then, subretinal localization (D,E) and proliferation (F,G) of virus-infected cells were examined immunohistochemically by using frozen sections prepared from 2-week (D) and 4-day (F) cultures. DAPI was used to visualize nuclei. Error bars indicate SEM. *P < 0.01.

To examine whether the binding with the Rab11 plays a role in the biological effects of mRab11-FIP4, we examined the effects of overexpression of a C-terminally truncated mutant of mRab11-FIP4A, mRab11-FIP4AΔRBD, on the retinal explants by means of retrovirus-mediated gene expression. We found that the expression of mRab11-FIP4AΔRBD also increased the proportion of INL-located cells, in fact to a degree rather higher than that achieved by the full-length A-form (Fig. 3A,B), suggesting that the binding with Rab11 is dispensable for the function of mRab11-FIP4 in retinal development. It should be noted that the C-terminally truncated protein (ΔRBD) was able to be expressed stably at the level comparable to those of the A- and B-forms in mammalian cells, including 293T (Fig. 3C) and NIH3T3 cells (data not shown), despite the partial degradation of the GST-fusion protein with the same truncation (GST4C-ΔRBD) in Escherichia coli cells (Fig. 7B).

To further address this notion, we examined the role of Rab11 activity in retinal development by using wild-type (Rab11 WT), dominant-negative (Rab11 S25N), and constitutively active (Rab11 Q70L) Rab11. As already reported (Wallace et al., 2002b), mRab11-FIP4A was localized in the perinuclear region when expressed in a cell line; and its expression there was enhanced by coexpressing it with either Rab11 WT or Rab11 Q70L (Fig. 7C). When coexpressed with Rab11 S25N, mRab11-FIP4A was diffusely distributed in the cytoplasm, suggesting that Rab11-FIP4A forms a complex only with the active form of Rab11. Although we found that the proportion of cells was increased in ONL when wild-type Rab11 was expressed, similar effects were also observed by the expression of either the constitutively active or dominant-negative mutant of Rab11 (Fig. 7D,E), suggesting that the activity of Rab11 had no effect on the sublayer localization of retinal cells. In addition, the expression of either Rab11 or any of its mutants had no effect on proliferation of retinal progenitors or localization of the retinal cells (Fig. 7F,G). All these results, therefore, suggest that mRab11-FIP4 may play a role in retinal development in a Rab11-independent manner.

DISCUSSION

We identified mRab11-FIP4A as a gene predominantly expressed in neural tissues, including the retinal differentiating progenitors and RGC. By gain- and loss-of-function experiments using retinal explants, the involvement of mRab11-FIP4A for cell cycle exit of retinal progenitors through the up-regulation of p27Kip1 as well as down-regulation of cyclinD1 and the fate decision to subpopulations localized in the INL, such as bipolar cells and Müller glia, was revealed. Analysis of the birth date, or the onset of the differentiation, of various retinal subpopulations has indicated that the late-born cells differentiate from progenitors that have undergone longer periods of proliferation (Young, 1985; Livesey and Cepko, 2001). The bipolar and Müller glia cells are the latest-born retinal cells and, therefore, would be expected to be decreased in number when the progenitors exit the cell cycle earlier. However, several lines of evidence suggest that cell cycle regulation and histogenesis of the retinal cells are not simply coordinated. In the developing mouse retinal progenitor cells, the cell cycle is regulated positively by cyclin D1 and negatively by the cdk inhibitors p27Kip1 and p57Kip2 (Dyer, 2003). Cyclin D1-deficient mice showed small eyes with significantly impaired proliferation of retinal progenitors; however, their eye function was normal, suggesting that the histogenesis was unaffected and the neural network of the retinal cells was properly formed in the absence of cyclin D1 (Sicinski et al., 1995). Differentiation of all major types of retinal cells was also observed in mice lacking either one of the above cdk inhibitors (Levine et al., 1995; Dyer and Cepko, 2000, 2001). On the other hand, the misexpression of homeodomain and basic helix–loop–helix transcription factors in the retinal explant promotes differentiation of certain subsets of retinal cells without affecting their cell proliferation or survival (Hatakeyama et al., 2001). On the basis of these observations, we surmised that mRab11-FIP4A regulates the retinal cell fate by the mechanism distinct from that for cell cycle exit.

The single EF-hand motif in the N-terminal domain and the C-terminal RBD are highly similar among orthologues of Rab11-FIP4, suggesting that this gene has been evolutionally conserved. However, our observation revealed that interaction with Rab11 through the C-terminal RBD was dispensable for the retinal phenotypes induced. There are several reports on Rab11-FIPs to analyze the mechanism of the interaction with Rab11 by introducing point mutations in their conserved RBD (Prekeris et al., 2001; Meyers and Prekeris, 2002; Fielding et al., 2005). Although here we used the C-terminal truncation mutant of mRab11-FIP4A mRab11-FIP4AΔRBD, a point mutant of mRab11-FIP4 lacking ability to bind to Rab11 could allow us to examine the role of Rab11 in the retinal development in more detail.

In contrast, the N-terminal domain containing the EF-hand motif was required for mRab11-FIP4A to induce phenotypes when expressed in the retinal explant. The N-terminal EF-hand motif is a Ca2+-binding motif found in many Ca2+-binding proteins; and in many cases, it functions in pairs to induce a conformational change in the protein in a Ca2+-dependent manner. However, some proteins possess an odd number of this motif, and a single unpaired EF-hand is thought to contribute to the interaction with other molecules (Lewit-Bentley and Rety, 2000), suggesting that the function of mRab11-FIP4A in retinal development may be mediated by some molecule(s) interacting with the N-terminal domain.

The EF-hand motif was also found in the N-terminal region of Rab11-FIP3 (Meyers and Prekeris, 2002; Wallace et al., 2002a). Despite mRab11-FIP3 expression in the retina throughout development, we demonstrated that retinal development was perturbed when the expression of mRab11-FIP4A alone was modified. Although we could not exclude the possibility that mRab11-FIP3 plays a role in retinal development, mRab11-FIP4A is likely to have a specific function, which cannot be complemented by mRab11-FIP4B or mRab11-FIP3. There are two possible mechanisms to explain the specificity between Rab11-FIP3 and Rab11-FIP4. One is the interaction with distinct proteins through the EF-hand. Specificity of the protein binding through the EF-hand is in part determined by the primary amino acid sequence within and around this motif (Bhattacharya et al., 2004). The N-terminal sequence surrounding the EF-hand in Rab11-FIP4 is well conserved among species but distinct from that found in Rab11-FIP3, probably mediating the specific protein interaction. The other possible mechanism is differential affinities of Rab11-FIP3 and Rab11-FIP4 for small GTPases. Rab11-FIP3 and Rab11-FIP4 in humans are also known as arfophilin1 and arfophilin2, respectively, and have been shown to interact with Arf5 and Arf6 through a binding site, which was located at the C-terminal regions but distinct from that for Rab11 (Fielding et al., 2005). Biochemical analysis indicated that Rab11-FIP3 bound to Rab11 with an affinity greater than that for Arf6, whereas Rab11-FIP4 was preferentially interacted with Arf6 rather than Rab11 (Fielding et al., 2005), suggesting that Arf6 may be involved in the retinal function of Rab11-FIP4. Although Arf6 was shown to be involved at least in the neurite outgrowth in the chick retina (Albertinazzi et al., 2003), further analysis is required to uncover the role of Arf6 in the retinal development regulated by mRab11-FIP4.

Recent studies indicated that mammalian class II Rab11-FIPs (FIP3 and FIP4) and Drosophila Nuf were shown to regulate the formation of the cleavage furrow in cytokinesis in mammalian cell lines (Fielding et al., 2005; Wilson et al., 2005) and in cellularization in Drosophila embryos (Rothwell et al., 1998; Riggs et al., 2003), respectively. In addition, effects of Rab11-FIP4 expression on the morphology of recycling endosome observed in HeLa cells was reproduced by the expression of Nuf (Hickson et al., 2003). These observations suggested that Nuf was not only structurally but also functionally related to the class II Rab11-FIPs. However, Nuf contains neither N-terminal EF-hand motif nor Arf-binding domain (Hickson et al., 2003; Wilson et al., 2005), and the functional analysis of Rab11-FIP4 in some previous works has been performed using an N-terminally truncated form of the protein (Wallace et al., 2002b; Hickson et al., 2003), suggesting that the role of mRab11-FIP4A we observed in the retinal development was mediated by the mechanism distinct from that previously reported.

In this study, we identified two alternative forms of mRab11-FIP4, A- and B-forms, expressed in developing mouse embryos. We found that zRab11-FIP4 was also expressed as two forms of transcripts with similarities in their structures and expression patterns (Muto et al., 2006). In addition, the loss-of-function experiments using a morpholino antisense oligo demonstrated that zRab11-FIP4A was required for cell cycle exit and differentiation of the retinal progenitors. These results suggest that Rab11-FIP4 has been conserved structurally and functionally between mouse and zebrafish. As in the zebrafish, mouse Rab11-FIP4 may play roles in the differentiation of RGC; however, we could not examine this possibility because of technical difficulty.

zRab11-FIP4A has been implicated to interact functionally with Shh signaling (Muto et al., 2006). Shh was earlier shown to define the timing of the cell cycle exit by regulating the expression of cyclin D1 and p57Kip2 (Shkumatava and Neumann, 2005) and to promote the differentiation of RGC and photoreceptors (Stenkamp et al., 2002; Shkumatava et al., 2004) in the zebrafish retina. We found that all these events were impaired in the zRab11-FIP4A knockdown embryos (morphants) and that the delayed cell cycle exit found in the morphant retina was recovered by activating the Shh signaling (Muto et al., 2006), suggesting that zRab11-FIP4A positively interacts with Shh signaling in the developing zebrafish retina. Shh has also been shown to play important roles in mouse retinal development, but the roles are different from those in the zebrafish retina; i.e., in mouse retina, Shh promotes the proliferation of retinal progenitors (Jensen and Wallace, 1997), represses RGC production (Wang et al., 2005), and accelerates rod photoreceptor differentiation (Levine et al., 1997). Our results indicated that mRab11-FIP4A affected negatively some of these events; and, therefore, a mechanism similar to that of zRab11-FIP4A and Shh cannot be applied for mouse Rab11-FIP4. In an effort to clarify the role of mRab11-FIP4A on the retinal development, we obtained some evidence suggesting the negative regulation of Shh signaling by the expression of mRab11-FIP4A. An analysis of the expression patterns of genes related to Shh signaling showed that the expression of ptc1, which is a major target of Shh signaling, was down-regulated in the retinal cells expressing mRab11-FIP4A. We also found that only gli3 among three gli genes was up-regulated by ectopic expression of mRab11-FIP4A. These expression patterns were similar to those observed in neurosphere derived from neocortex of Shh mutant mice (Palma and Ruiz i Altaba, 2004), suggesting that mRab11-FIP4A regulated the expression of these genes by repressing Shh signaling. Alternatively, the ectopic expression of gli3 might repress Shh signaling in mRab11-FIP4A–expressing cells, as Gli3 was indicated to be a negative regulator of Shh signaling in vivo by means of the analysis using mutant mice lacking Gli3 activity (Meyer and Roelink, 2003; Lipinski et al., 2006). The role of Gli3 in the context of mRab11-FIP4A function remains elusive.

Several lines of evidence have demonstrated that Shh signaling is also regulated by vesicle trafficking at multiple steps, including at the level of the subcellular localization of the Shh receptors Patched and Smoothened (Incardona et al., 2000; Martin et al., 2001) and by the signaling pathway downstream of these receptors (Eggenschwiler et al., 2006). One of the molecules recently identified as a negative regulator of Shh signaling is Rab23, another member of Rab GTPase (Eggenschwiler et al., 2001). Recent studies demonstrated that Rab23 repressed Shh signaling by regulating the activities Gli2 and Gli3 proteins (Eggenschwiler et al., 2006). Rab23 is expressed in eyes (Marcos et al., 2003), and Rab23 mutant mice, known as open brain (opb), were reported to have poorly developed eyes, suggesting that Rab23 is involved in the retinal development by regulating Shh signaling. Rab11-FIP4 was demonstrated to interact with multiple small GTPases (Hickson et al., 2003); it is, therefore, tempting to speculate that Rab11-FIP4 plays a role in the retinal development by regulating membrane trafficking system through interaction with other small GTPases such as Rab23. Further analysis is required to elucidate the exact mechanism of Rab11-FIP4 in retinal development and the interaction of this protein with Shh signaling.

EXPERIMENTAL PROCEDURES

Differential Display and cDNA Cloning of Mouse Rab11-FIP4

For differential display analysis, first-strand cDNA was synthesized from 2.5 μg of total RNA prepared from whole eyes by using GT₁₅V primer (5′-GTTTTTTTTTTTTTTT(A/G/C)-3′) and Superscript II reverse transcriptase (Invitrogen). DD-PCR was performed using a combination of the GT₁₅V primer and an arbitrary 10-mer primer. Then, 30 cycles of PCR reaction (95°C for 0.5 min, 40°C for 2 min, and 72°C for 1 min) were performed, and the products were separated by electrophoresis on a 6% nondenaturing polyacrylamide gel and visualized by staining with SYBR green. Differentially displayed bands were extracted from the gels, amplified under the same PCR conditions, and subcloned into the pGEM-T easy (Promega) or pCR2.1 vector (Invitrogen). Nucleotide sequences of the subcloned fragments were determined, and database searching was performed by using the BLAST algorithm. Based on the partial sequence found in Celera Discovery System, we cloned a cDNA encoding the full length of a putative mouse Rab11-FIP4 protein by RT-PCR in combination with the 5′-RACE method using cDNA prepared from P1 mouse retina as a template.

Plasmid Construction

Plasmids to express GST-fused mRab11-FIP4 in E. coli cells were constructed by subcloning PCR-amplified partial fragments of mRab11-FIP4 into pGEX6P-2 (Amersham). Retrovirus vectors, pMXc-IRES-EGFP (Morita et al., 2000; Tabata et al., 2004; Ouchi et al., 2005) and pSSCG (Yamamichi et al., 2005) were used for the expression of genes and shRNAs, respectively. cDNAs for wild-type, dominant-negative (S25N), and constitutively active (Q70L) Rab11 (Savina et al., 2002), which were kindly donated by Dr. M. I. Colombo of the Universidad Nacional de Cuyo-CONICET, were subcloned into pMX-IRES-EGFP. For shRNA expression, double-stranded oligonucleotides covering shRNA sequences were first subcloned into BbsI and EcoRI sites of a pU6 vector, and then the U6 promoter-shRNA cassettes were further subcloned into pSSCG vector by using BamHI and EcoRI sites.

Animals, Cell Culture, and Transfection

Normal ICR, FVB, and C57BL6 mice were purchased from SLC Japan. All experiments on animals were approved by the Animal Care Committee of the Institute of Medical Science of the University of Tokyo.

COS7, 293T, or PLAT-E cells (Morita et al., 2000), the last being a derivative of HEK293 cells, were grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum. For PLAT-E cells, the medium was supplemented with 1 μg/ml puromycin and 10 μg/ml brastcidine. Exponentially growing cells (4 × 10⁶ cells) were transfected with 2 μg of plasmid by using Fugene6 (Roche Diagnostics GmbH) according to the manufacturer's instructions. The effect of shRNA was examined in 293T cells expressing either the A- or B-form of mRab11-FIP4. The expression of mRab11-FIP4 was examined by Western blotting using rabbit polyclonal anti–mRab11-FIP4 antibody prepared in our laboratory.

Preparation of retinal explants and retrovirus infection was performed as described previously (Ouchi et al., 2005). To quantify the distribution of retrovirus-infected cells, at least 10 sections from two independent retinas were used and EGFP-positive cells on these sections were counted under the fluorescence microscope.

Cell Sorting and RT-PCR

Retinal explants infected with retrovirus were cultured for 5 days and dissociated by incubating in 0.25% trypsin/phosphate buffered saline (PBS) at 37°C for 10 min. After removing aggregated cells by filtration, virus-infected (EGFP-positive) retinal cells in the dissociated single cell suspension were sorted using FACSAria (Becton Dickinson). Gating parameters by forward and side scatters were set to remove remaining dead or aggregated cells, and dead cells were also removed as cells stained with PI. Equal numbers of EGFP-positive cells were isolated from all samples, and total RNA extracted from the sorted cells by a standard AGPC method was used to prepare cDNA. All primers used for the following RT-PCR were designed from cDNA sequences in Genbank.

Pull-Down Assay

GST and GST-fusion proteins were affinity purified from E. coli (BL21DA3 strain) lysates, which were prepared in E. coli lysis buffer (PBS containing 1% Triton X-100 and 1 mM phenylmethyl sulfonyl fluoride [PMSF]), by the binding of the proteins to glutathione–Sepharose 4B beads (Amersham). Mammalian cell lysates containing EGFP-Rab11 or EGFP (control) protein were prepared from PLAT-E cells transfected with either pEGFP or pEGFP-Rab11wt in 0.5 ml of a lysis buffer (50 mM Hepes, pH 7.5, containing 150 mM NaCl, 5 mM MgCl₂, 0.5% Triton X-100, 1 mM PMSF, 2 μg/ml leupeptin, and 1 μg/ml NaVO₃). After removal of debris by ultracentrifugation, supernatants were stored at −80°C until used. Then, the lysate supernatants were thawed and incubated at 4°C for 6 hr with GST or either one of the GST-Rab11-FIP4 proteins bound to the glutathione–Sepharose 4B beads. These samples were extensively washed, after which the bound proteins were subjected to SDS-PAGE. Precipitated EGFP and EGFP-Rab11wt proteins were detected by Western blotting using an anti-EGFP antibody (Clontech).

In Situ Hybridization and Immunohistochemistry

For in situ hybridization, a digoxigenin (DIG)-labeled RNA probe was prepared from cDNA template subcloned into the pGEM-T easy vector (Promega) by using a DIG RNA labeling kit (Roche Diagnostics GmbH). For whole-mount in situ hybridization, mouse embryos (E8.5 to E10.5) were fixed in 4% paraformaldehyde (PFA)/PBS for overnight at 4°C. These embryos were dehydrated and rehydrated by passage through a series of diluted methanol, bleached with 6% hydrogen peroxide, treated with Proteinase K (10 μg/ml) for an appropriate time at room temperature, and then fixed with 4% PFA/0.2% glutaraldehyde solution. Next, the samples were hybridized with DIG-labeled RNA probe for overnight at 70°C. On the next day, the samples were extensively washed; and in situ hybridization and immunohistochemistry for the tissue sections were performed, as described elsewhere (Muto et al., 2006). Antibodies used in this study were anti-rhodopsin (4D2, 1:200; kindly provided gift from Dr. Molday), anti-Hu C/D (1:1,000; Molecular Probes, Inc.), anti-PKC (1:20; Oncogene Research Product), anti-glutamine synthetase (GS, 1:500; Chemicon International), anti-cyclin D3 (1:500, Santa Cruz Biotechnology), anti-p27Kip1 (clone 57, 1:50; BD Transduction Laboratories), anti-Ki67 (1:200, BD Biosciences), and anti-GFP (1:5,000; Clontech).

Acknowledgements

We thank Drs. M.I. Colombo and R.S. Molday for providing Rab11 cDNAs and anti-rhodopsin antibody 4D2, respectively. We thank Drs. H. Sagara and R. Kurita for helpful comments and discussion and also A. Usui for technical support.