Hes4 Controls Proliferative Properties of Neural Stem Cells During Retinal Ontogenesis

The retina of fish and amphibian contains genuine neural stem cells located at the most peripheral edge of the ciliary marginal zone (CMZ). However, their cell-of-origin as well as the mechanisms that sustain their maintenance during development are presently unknown. We identified Hes4 (previously named XHairy2), a gene encoding a bHLH-O transcriptional repressor, as a stem cell-specific marker of the Xenopus CMZ that is positively regulated by the canonical Wnt pathway and negatively by Hedgehog signaling. We found that during retinogenesis, Hes4 labels a small territory, located first at the pigmented epithelium (RPE)/neural retina (NR) border and later in the retinal margin, that likely gives rise to adult retinal stem cells. We next addressed whether Hes4 might impart this cell subpopulation with retinal stem cell features: inhibited RPE or NR differentiation programs, continuous proliferation, and slow cell cycle speed. We could indeed show that Hes4 overexpression cell autonomously prevents retinal precursor cells from commitment toward retinal fates and maintains them in a proliferative state. Besides, our data highlight for the first time that Hes4 may also constitute a crucial regulator of cell cycle kinetics. Hes4 gain of function indeed significantly slows down cell division, mainly through the lengthening of G1 phase. As a whole, we propose that Hes4 maintains particular stemness features in a cellular cohort dedicated to constitute the adult retinal stem cell pool, by keeping it in an undifferentiated and slowly proliferative state along embryonic retinogenesis. Stem Cells 2012;30:2784–2795


INTRODUCTION
Contrasting with the mammalian situation, the retina of adult fish and amphibians contains a population of neural stem cells, which allow continuous tissue growth throughout the animal life as well as regeneration following retinal damage [1]. These retinal stem cells (RSCs) reside in a permanently proliferating region located at the margin of the retina, known as the ciliary marginal zone (CMZ; [2]). The spatial organization of the CMZ mirrors the temporal sequence of retinal development with stem cells being found in its most peripheral part, successively followed more centrally by actively dividing progenitors and then by their postmitotic progeny [3,4].
Much progress has been made these past few years in the characterization of CMZ neural stem cells. This includes the formal demonstration of their multipotency and self renewal ability [5,6], advance in the description of their niche [7,8] and the identification of candidate genes and signaling pathways to regulate their postembryonic activity [9][10][11][12][13][14][15][16][17]. How-ever, several questions remain unresolved and in particular, that of their embryonic origin. Whether they arise from a discrete population of cells and how they escape from cell cycle exit and differentiation signals during retinal development is hitherto unknown. Comprehensive analysis of RSC ontogeny and properties necessitates reliable markers to formally identify these immature cells within proliferating heterogeneous cell populations in vivo. To gain insight into the molecular signature of RSCs, we recently performed a large-scale expression screen in the Xenopus CMZ [14]. Among identified RSC markers, we retrieved the Hes4 gene (previously known as XHairy2 in Xenopus and ortholog of zebrafish Her9 and chick cHairy1) that encodes a transcriptional repressor of the bHLH-O family.
Hes family genes are well known as Notch transcriptional targets that can regulate cellular differentiation, cell fate decisions, and embryonic patterning in various developmental systems [18,19]. Notably, several members of this family, including Hes1, which is closely related to Hes4 in terms of sequence similarity [18], have been intensively studied in the Author contributions: W.E.Y.: conception and design, collection and assembly of data, and data analysis and interpretation; C.B.: collection and assembly of data and data analysis and interpretation; J.H. and K.P.: collection and assembly of data; H.T.T. and K.V.: provided Wnt-responsive transgenic animals; M.P.: conception and design, data analysis and interpretation, and manuscript writing; M.L.: conception and design, collection and assembly of data, data analysis and interpretation, and manuscript writing. developing vertebrate brain. In this context, several lines of evidence converge toward a role in boundary formation and maintenance of neural stem/progenitor cells, mainly through prevention of neuronal differentiation [20][21][22][23][24][25][26][27]. Hes1 is expressed as well in the embryonic retina, where it regulates distinct aspects of eye morphogenesis and is required for proper timing of neurogenesis [28][29][30]. Hes1 also emerged as a safeguard of cellular quiescence, through protection against terminal differentiation and permanent cell cycle arrest [31,32]. In contrast to Hes1, the Hes4 gene was largely ignored in mammalian studies, presumably due to the absence of an ortholog in mouse. It is, however, expressed in humans and has been shown to be involved in several aspects of other vertebrate species development. In particular, Hes4 proved to play a significant role in maintaining the undifferentiated state of Xenopus neural crest cells [33][34][35] and zebrafish inner ear progenitors [36]. In addition, a recent publication by Kubo and Nakagawa identified the chick Hes4 ortholog, cHairy1 [37], as highly expressed in the CMZ and required for the maintenance of this structure downstream Wnt signaling [38]. We thus decided to gain further insights into Hes4 expression and function in the developing Xenopus retina.
We found that in contrast to mouse Hes1 [28,30] and chick cHairy1 [39], Hes4 is not expressed in the neural retina (NR) at any stage examined but labels the presumptive retinal pigmented epithelium (RPE) and forming CMZ before being restricted to stem cells of the mature retina. Wnt and Hedgehog signaling pathways contribute to this dynamic expression pattern through positive and negative regulation, respectively. Finally, functional analysis revealed that Hes4 imparts retinal cells with stem-like properties: inhibited commitment toward RPE and neuronal fate, prolonged proliferative capacities, and slow cell cycle kinetics.

Embryo Collection, Transgenic Line
Xenopus laevis embryos were obtained by conventional methods of in vitro fertilization. Xenopus tropicalis transgenic animals carrying the Wnt reporter pbin8LefdGFP construct have previously been described [40] and were obtained by natural fertilization between a wild-type female and a transgenic male carrying a single insertion of the transgene [41]. All experiments were approved by the Direction D epartementale des Services V et erinaires de l'Essonne, Evry, France.

Microinjections and In Vivo DNA Lipofection
Xenopus laevis Hes4 gene exists as two alloalleles, Hes4a and Hes4b, and the later was used in all overexpression experiments. Capped sense RNAs were transcribed using the mMessage mMachine SP6 kit (Ambion, Austin, TX, http:// www.ambion.com). 500 pg of mRNA was injected animally into one or two blastomeres at the two-cell stage, together with GFP or b-galactosidase mRNA (250 pg). Protein activity of GR chimeric constructs was induced by incubating the embryos in 4 lg/ml dexamethasone (Sigma) from stage 12. Treating embryos before this stage lead to severe developmental defects and frequently to developmental arrest. Morpholino oligonucleotides against both alloaleles (''Hes4 Mo''; [34]) were injected into one blastomere at the four-or eightcell stage (8-20 ng). Lipofection experiments were performed at stage 17/18 as previously described [45].

In Situ Hybridization
Digoxigenin-labeled antisense RNA probe synthesis and whole-mount in situ hybridization were performed as previously described [14]. Embryos were then vibratome sectioned (50 lM). Following image capture, labeling area was manually delineated and quantified (in Pixel 2 ) in both the dorsal and ventral CMZ using Adobe Photoshop CS4 software. Shown in graphs are percentage of staining area increase/ decrease compared to the control.

BrdU/EdU Incorporation and Immunohistochemistry
Embryos were injected intra-abdominally with BrdU (5bromo-2 0 -deoxyuridine; 10 mM; Sigma) or EdU (5-ethynyl-2 0deoxyuridine; 1 mM; Invitrogen, Carlsbad, CA, http://www. invitrogen.com) for various durations, depending on the experiment. For birthdating analyses, embryos were injected every 10-12 hours from stage 34 to stage 41. Immunohistochemistry was performed on 12 lm-cryostat or -paraffin sections as previously described [12,41]. Antibodies used are listed in Supporting Information Table 1. Cell nuclei were counterstained with Hoechst (Sigma). Detection of EdU-labeled cells was carried out with the Click-iT EdU Imaging Kit (Molecular Probes, Eugene, OR, http://probes.invitrogen. com). Fluorescent staining was visualized with a M2 Zeiss microscope. Images were captured using a digital Axiocam MRc camera and processed with AxioVision REL 7.8 and Adobe Photoshop CS4 softwares. Retinal area was manually delineated on transverse sections and quantified (in Pixel 2 ) using Adobe Photoshop CS4 software.
appearance of EdU labeling in the mitotic population was taken as an estimation of the average G2 length (T G2 ; [47]). M-phase duration (T M ) was calculated after determining the proportion of cells that were in mitosis (% M), taking into account T C and GF, as determined by cumulative EdU labeling: Finally, G1 duration (T G1 ) was deduced from the above values as

Reverse Transcription and Quantitative Real Time PCR
Reverse transcription of retinal mRNA and quantitative real time PCR (qPCR) were performed as described previously [41]. PCR primer sequences are listed in Supporting Information Table 2.

Quantification and Statistical Analysis
All experiments were performed at least in duplicate. Shown in figures are results from one representative experiment. In each histogram, values are given as mean 6 SEM. In experiments involving two conditions, statistical analysis was performed by Student's t test, while in those involving the comparison of more than two treatments one-way ANOVA followed by Tukey's post hoc tests were performed (NS: not significant; *, p < .05; **, p < .01; ***, p < .001).

Hes4 Expression Likely Reveals the Embryonic Cell-of-Origin of Adult RSCs
The embryonic lineage that gives rise to adult RSCs in Xenopus is currently unknown. Gene expression profile within the CMZ has proved to mirror the temporal genetic sequence of retinal development [3,4,48]. As Hes4 labels the most peripheral stem cell-containing part of the CMZ ( [14] and Figure 1Q, 1R), we hypothesized that establishing its expression profile during retinogenesis might retrospectively give clues about RSC ontogenesis. We thus analyzed the distribution of its transcripts at key stages of eye development and compared it with that of optic field (Pax6), NR (Rx), optic stalk (OS; Pax2), and RPE (Mitf1A) markers ( Fig. 1; Supporting Information Fig. S1). At late neurula stage, Hes4 expression was observed in a restricted proximo-dorsal region of the optic field, clearly contrasting with that of Pax6 (Fig. 1A, 1B). In line with this, analysis on retinal sections at the optic vesicle stage revealed that Hes4 stained the Mitf1A-expressing presumptive RPE and was not detected in the Rx-positive retinal neuroepithelium, except in its most dorsal part at the border with the RPE (Fig. 1K, 1L). Of note, from stage 24-25, Hes4 transcripts were additionally observed in the OS (coinciding with Pax2 expression; Fig. 1L). During optic cup formation, Hes4 expression progressively declined in the differentiating central RPE while remaining high in its peripheral part and in the NR margins ( Fig. 1M-1O). Staining in the margin progressively superimposed with the forming CMZ to finally get restricted to its most peripheral region ( Fig. 1P-1R). As a whole, Hes4 expression delineates a territory clearly distinct from both the differentiating NR and RPE, that is first located in the dorsal part of the optic vesicle, then at the peripheral margin of the optic cup, and that finally coincides with RSC location following completion of embryonic retinogenesis. Importantly, such a dynamic expression pattern strongly resembles that of Gli3 and Smo, two previously described stem-cell specific markers [49]. In addition, we found that the same held true for two other genes that we recently identified as being expressed at the extreme tip of the CMZ, namely Id2 and Wnt8b (Supporting Information Fig. S2 and [41]). Altogether, these data thus suggest that adult RSCs originate from a discrete population of cells located at the RPE/NR border of the optic vesicle.

Hes4 Expression Is Positively Regulated by Wnt Signaling and Negatively by the Hedgehog Pathway
We then aimed at identifying the signaling pathways regulating Hes4 embryonic expression. A potential candidate is the canonical Wnt pathway since it proved to be active in the peripheral retina of various species [10,50,51] and was shown to regulate CMZ expression of cHairy1, the Hes4 ortholog in chick [38]. However, the status of Wnt activity during embryonic retinogenesis of Xenopus has never been investigated in detail. Taking advantage of a Xenopus tropicalis reporter line carrying a destabilized eGFP downstream a synthetic Wnt-responsive promoter [40], we thus analyzed the profile of Wnt activity in the developing retina. Except in the presumptive RPE and OS of the optic vesicle where it was not detected, eGFP staining strikingly superimposed with that of Hes4 (Supporting Information Fig. S3). Consistent with a role in promoting Hes4 expression, we found that pharmacological blockade of Wnt signaling from stage 25 to stage 35 (24-hour IWR1 treatment) significantly lowered Hes4 levels in the CMZ, as inferred from both in situ hybridization and qPCR analyses ( Fig. 2A, 2C, 2F). Unexpectedly, similar inhibition was observed upon a 24-hour treatment with BIO, an activator of the Wnt pathway ( Fig. 2A, 2C). Shortening the treatment to 15 hours, however, led to a consistent increase of Hes4 labeling (Fig. 2B, 2C). As Hes1 is known to repress its own expression [52], these results led us to investigate whether this might also be the case for Hes4. In line with this, we found that misexpression of the Hes4 antimorphic variant Hes4VP16-GR (which transforms the Hes4 repressor into a transcriptional activator) led to ectopic Hes4 staining in the epidermis of injected embryos (Supporting Information Fig.  S4A). Additionally, morpholino-mediated Hes4 knockdown significantly enhanced Hes4 expression in its endogenous domain at the neurula stage (Supporting Information Fig. S4B). As a whole, these results suggest that Hes4 levels in the forming CMZ are positively regulated by the Wnt pathway and modulated by a negative feedback loop.
We next investigated the impact of Hedgehog signaling based on previous data in mouse retinal explants showing that Hes1 expression is stabilized by the Shh effector Gli2 [53]. As a first attempt to address this issue, we compared Hes4 expression pattern during retinogenesis with that of Gli1 as a readout of Hedgehog signaling activity (Supporting Information Fig. S3). As previously described [12,49], Gli1 transcripts were detected at high levels in the presumptive RPE and OS of stage 25 embryos and at low levels in the NR and NR/RPE border. Strikingly, upon optic cup formation, Gli1 and Hes4 clearly exhibited complementary expression patterns, with Gli1 being mainly detectable in the differentiating RPE and Hes4 restricted to the peripheral margin of the retina. Finally, in the mature retina, Gli1 labeling was detected in the central CMZ and was faint in its most peripheral Hes4expressing part. These results suggest that in contrast to mouse Hes1, Hes4 might be downregulated by Hedgehog signaling. In line with this, we found that Hedgehog pathway inhibition with cyclopamine significantly enhanced Hes4 expression in the CMZ (Fig. 2D-2F). Conversely, Hedgehog activation with purmorphamine strongly reduced Hes4 staining (Fig. 2D, 2E). Altogether, it is likely that Hedgehog-mediated inhibition of Hes4 contributes to restrain its expression domain to the peripheral margin of the optic cup and to the stem cell compartment of the CMZ.
Although Hes genes are recognized as canonical targets of Notch signaling, a growing wealth of data indicate that they may function independently in several contexts, including the retina [38,53]. Considering the above data, we investigated whether this might be the case for Hes4 as well. We thus blocked the Notch pathway with DAPT, a chemical compound that inhibits the c-secretase-dependent release of the Notch intracellular domain and was shown to repress Hes1/5 expression in the chick retina [54]. Contrasting with the effects of Wnt and Hedgehog signaling perturbations in the same time window, Hes4 levels were not significantly lowered by a 24hour DAPT exposure (Supporting Information Fig. S5B, S5E, S5F) and were only affected when treatment duration was extended to 48 hours (Supporting Information Fig. S5D, S5E). Such a delay suggests that the Notch pathway does not directly contribute to Hes4 regulation in the forming CMZ.

Hes4 Loss of Function Severely Impairs Eye Formation
We next aimed at addressing Hes4 function during retinal development through a knockdown experiment. Morphological analysis of stage 37 tadpoles injected with Hes4 Morpholinos (Hes4 Mo) revealed severe dose-dependent eye defects ranging from a strong reduction of eye size to a complete absence of eye (Fig. 3A). No discernable eye malformations were observed upon injection of a standard Morpholino (Ctrl Mo; data not shown). In order to exclude a potential toxicity of the Hes4 Mo, we tested whether the eye-absent morphant phenotype could be rescued upon coinjection of Hes4 Mo with an inducible Hes4 construct (Hes4-GR) devoid of Hes4 Mo target sequence. The percentage of eye-absent embryos was scored in the different conditions, and we found that it dropped from 78% in Hes4 Mo-injected embryos to 12.5% in Hes4 Mo plus Hes4-GRinjected ones (Fig. 3B). Such a rescue clearly suggests that the loss of eye is a specific phenotype of Hes4 knockdown. Consistent with this, severe eye defects were observed as well following overexpression in the presumptive neurectoderm of the antimorphic Hes4 variant Hes4VP16-GR (data not shown).
Since two previous studies in Xenopus reported increased apoptosis at neurula stages following Hes4 loss of function [33,35], we wondered whether cell death might account as well for the observed eye malformations. We thus performed anti-active caspase3 immunostaining on stage 25 retinal sections from Hes4 Mo injected embryos. Massive apoptosis was evident on the injected side (Fig. 3C). Thus, as previously proposed in neural crest stem cells [33,35], Hes4 might be required for cell survival in the retina. Of note, cell death was not restricted to Hes4-expressing tissues and spread in particular to the NR, which might be a secondary consequence of the loss of surrounding tissues.

Hes4 Misexpression Prevents RPE Differentiation and Neuronal Commitment of Retinal Precursor Cells
Since the dramatic phenotype of Hes4 knockdown precludes any further functional investigation, we turned to a gain of function strategy using the Hes4-GR inducible construct. Morphological analysis at stage 37 revealed a defective RPE pigmentation (Fig. 3A) and a slight reduction of retinal size in Hes4-GR-injected embryos (23% 6 2.5% decrease of retinal surface compared to controls, as measured on 19 and 37 sections, respectively; p < .001). Effects on retinal cell type differentiation were then monitored using specific markers (Fig.  4A). In line with the aforementioned defects in retinal pigmentation, Hes4 overexpressing retinas displayed a virtual absence of XAR-1 expression, a marker of the differentiated RPE. In addition, substantial perturbations in laminar organization were observed in the central retina (CR). In line with this, both rhodopsin (data not shown) and calbindin stainings were severely reduced suggesting that rods and cones were profoundly missing. Syntaxin expression was strongly decreased in the inner and outer plexiform layers as well as in the optic nerve, consistent with defective interneuron and ganglion cell production. Thus, Hes4 misexpression drastically impairs both RPE and neuronal differentiation.
To get further insights into this phenotype, we checked the status of several markers at earlier stages by in situ hybridization (Fig. 4B). We observed that Rx expression was not affected suggesting that Hes4-overexpressing cells retain a retinal identity. However, Mif1A staining was strongly reduced compared to controls, consistent with an impairment of RPE determination. Besides, expression of the proneural genes Neurog2, Atoh7, and NeuroD4 (previously called XNgnr-1, Xath5, and Xath3, respectively) was also decreased, suggesting that Hes4 inhibits neurogenesis by preventing commitment of precursor cells toward a neuronal fate. Consistent with this, we found that the antimorphic Hes4 variant Hes4VP16-GR was able to induce ectopic neurogenesis, as inferred by N-tubulin staining following mRNA injection in the presumptive epidermis (Supporting Information Fig. S4A).
We next followed the fate of Hes4-overexpressing cells in a clonal analysis using in vivo lipofection. At stage 39, most cells in control clones were already differentiated into neurons as judged by their position and morphology. In contrast, an important proportion of Hes4-overexpressing cells still exhibited a neuroepithelial morphology characteristic of undifferentiated neural precursors (Fig. 4C, 4D). Analysis at stage 41 revealed that this delayed differentiation eventually ended up with an excessive production of Müller glial cells at the expense of neurons (Fig. 4C, 4D). In addition to their typical morphology, Müller cell identity was further confirmed by immunostaining using an anti-CRALBP antibody (data not shown). These data indicate that Hes4 overexpression cell autonomously delays differentiation and reduces neurogenesis. As a whole, these results suggest that Hes4 maintains cells of the retinal margin in an undifferentiated state during embryonic retinogenesis.

Hes4 Misexpression Maintains Retinal Precursors in a Proliferative State
We next examined whether Hes4-dependent blockade of differentiation was accompanied by prolonged proliferative capacities. We first analyzed BrdU incorporation at stage 37 following Hes4-GR mRNA injection. At this point of retinal histogenesis, the majority of cells in the wild-type retina are postmitotic and proliferation is restricted to the CMZ. In contrast, a dramatic increase in BrdU-positive cell number was observed in the CR of Hes4-overexpressing embryos (Fig.  5A, 5B).
We next assayed in lipofection experiments whether these effects on proliferation were cell autonomous (Supporting Information Fig. S6). To challenge the specificity of the Hes4 phenotype, transfection of another Hes family member, Hes2, was performed in a parallel batch of embryos. In contrast to Hes4, Hes2 is not expressed in stem cells but restricted to retinal progenitors where it acts as a gliogenic factor. We previously showed that it does so at least in part by delaying cell cycle exit of late precursor cells, while having no effect on early progenitor cell proliferation [43]. In line with this, we found that Hes2 overexpression increased the percentage of BrdU-positive cells among transfected cells at stage 35 but not at stage 32. In contrast, proliferation rates of Hes4-misexpressing cells were elevated at both stages with respect to the control situation (Supporting Information Fig. S6). This suggests that both early and late retinal precursors are maintained longer in the cell cycle upon Hes4 overexpression.
To examine the timing of cell cycle exit more directly, we performed birthdating experiments (Fig. 5C). BrdU was injected at regular intervals so that it would be constantly available from stage 34 to stage 41, and therefore mark all cells born in that period. We found that the proportion of BrdU-labeled cells among transfected cells was significantly increased following Hes4 transfection, compared to the control situation (Fig. 5D). This clearly indicates that Hes4 delays cell cycle withdrawal of precursor cells. Finally, we wondered whether such prolonged proliferation might account for the aforementioned imbalance between neuronal and glial cell production (Fig. 4D). We addressed this issue by counteracting Hes4-dependent maintenance in the cell cycle through colipofection with Xgadd-45c. This cell cycle inhibitor was previously shown to accelerate retinal cell birthdate in similar assays [44]. We found that Xgadd-45c indeed concomitantly rescued the Hes4-induced delayed cell cycle exit and the deficit in neuronal cells (Supporting Information Fig. S7), strongly suggesting a tight coupling between differentiation and proliferation defects.

Hes4 Misexpression Slows Down Cell Cycle Kinetics of Retinal Precursor Cells
As stated above, Hes4 misexpression leads to a reduction of retinal size, an apparently contradictory phenotype in view of the delayed cell cycle withdrawal of retinal precursor cells. As no increase in apoptosis could be detected following Hes4-GR mRNA injection (data not shown), we thus wondered whether this paradox might be explained by slower cell cycle kinetics of Hes4-overexpressing progenitors compared to controls.
To address this possibility, we first evaluated the mitotic index in the retinal neuroepithelium of stage 25 embryos, using the mitotic marker phospho-Histone H3 (P-H3; [12]). P-H3 positive cell proportion was calculated among all nuclei and further corrected by the GF (percentage of proliferative cells) as determined by EdU cumulative labeling experiments (see below; Fig. 6B; Supporting Information Table 3). We found a significantly lower mitotic index in Hes4-GR-injected optic vesicles compared to the control (Fig. 6A), suggesting that Hes4 overexpression slows down cell cycle speed of retinal precursors.
We next aimed at directly measuring total cell cycle (T C ) and S-phase (T S ) lengths in the NR through a cumulative EdU labeling assay. As shown in Figure 6B and Supporting Information Table 3, the duration of G2 þ M þ G1 (T C À T S ), as inferred from the time point at which the EdU labeling index reached the plateau, was longer in Hes4-misexpressing cells compared to wild-type ones. Calculation of T S and T C using the Nowakowski excel sheet [46] confirmed the hypothesis of extended cell cycle duration following Hes4-GR overexpression (þ39% compared to the control situation; 14.73 vs. 10.62 hours) and showed a slight increase of S-phase length (þ13%; 7.94 vs. 7.00 hours; Supporting Information Table 3).
Finally, in order to determine whether a specific phase of the cell cycle was particularly affected, we individually estimated G2 (T G2 ), M (T M ), and G1 (T G1 ) durations. G2 length was evaluated using the percentage of labeled mitoses (PLM) paradigm [12]. Noticeably, the PLM was consistently lower in Hes4-GR-overexpressing retinas compared to control ones, indicating a delayed S-to M-phase progression (Fig. 6C, 6D). Mean T G2 was estimated at 1.6 hours and 1.2 hours, respectively (þ33%; Fig. 6D; Supporting Information Table 3). Taking into account data gathered from the measurement of the mitotic index shown previously (Fig. 6A), we next determined T M (see Materials and Methods section for calculation details) and finally deduced T G1 as being equal to T C À (T S þ T G2 þ T M ). This revealed that increased time spent in G1 is the main parameter that accounts for the longer cell cycle observed upon Hes4 overexpression (þ148%; 4.94 vs. 1.99 hours), S and G2 phases being lengthened as well although to a lesser extent (Supporting Information Table 3).
The above data suggest that Hes4 transcription factor acts as a regulator of cell cycle speed. In line with this, the stem cell compartment of the CMZ where Hes4 is expressed was recently shown to exhibit slow cell cycle kinetics [6]. As stated above, Hes4 expression at the NR/RPE border of the optic vesicle likely prefigures the subpopulation of cells dedicated to form the adult RSC cohort. We thus wondered whether cells in this territory might be endowed with slow cell cycle kinetics as well. To address this issue, we performed EdU cumulative labeling in stage 22-24 wild-type embryos and analyzed EdU incorporation in two distinct territories: the Hes4-negative NR and the Hes4-expressing NR/ RPE border (Fig. 6E). Calculations of T S and T C at the NR/ RPE border yielded values of 13.57 and 20.26 hours compared to 7.47 and 11.47 hours in the NR (Fig. 6F; Supporting Information Table 3). Altogether these results strongly suggest that Hes4 contributes to the maintenance of slow cell cycle kinetics at the NR/RPE border.

DISCUSSION New Insights into the Embryonic Cell-of-Origin of Adult Xenopus RSCs
Elegant tracing experiments recently allowed to identify the offspring of a single CMZ stem cell during fish postnatal life [5]. However, the embryonic lineage of adult RSCs has never been investigated so far. Our retrospective analysis of Hes4 expression pattern suggests that these cells originate from the RPE/NR border of the optic vesicle. This hypothesis is substantiated by our finding that other stem cell-specific markers of the postembryonic CMZ display similar expression dynamic during retinogenesis: first labeling the presumptive RPE at optic vesicle stage, then the retinal margin of the optic cup, and finally the most peripheral stem-cell containing part of the CMZ in the mature retina. Importantly, it also fits with recent results from four dimensional manual cell tracking in zebrafish, suggesting that part of the optic vesicle medial layer (classically considered as prospective RPE) contributes to the CMZ [55]. Finally, such an early restricted expression of RSC markers challenges the classical schematic representation of retinal histogenesis, which implies that RSCs are initially found throughout the NR of the newly emerged optic vesicle and thereafter persist in its periphery as central cells differentiate. It rather suggests that a presumptive RSC cohort is already segregated from both the NR and RPE lineages as early as the optic vesicle stage (Fig. 7A).

Hes4 Maintains Presumptive RSCs in a Proliferative and Undifferentiated State
Our gain of function analysis revealed that Hes4 misexpression prevents retinal precursors from committing toward retinal fates. Of note, Hes4 ability to block neurogenesis is a function shared by many other Hes family members that are known to transcriptionally repress or functionally antagonize several proneural factors [24]. In particular, cHairy1 has proved to be required to prevent CMZ cells from differentiating into neurons and to be sufficient to endow progenitor cells of the CR with CMZ-like cell properties [38]. We found here that the role of Hes4 during Xenopus retinal development may also extend to an inhibition of the RPE differentiation program. Besides, our results indicate that the Hes4-dependent blockade of neuronal and RPE determination is accompanied by maintenance in a proliferative state. Whether the former effect is a consequence of the latter or vice versa remains to be determined. However, both appear tightly coupled since Hes4-induced defective neurogenesis, as observed in lipofection experiments, can be rescued by forcing cells to exit the cell cycle. We thus conclude that the Hes4 phenotype observed in our overexpression experiments reflects a function dedicated to maintain presumptive adult RSCs in an undifferentiated and proliferative state towards adulthood.

Hes4 Regulates Cell Cycle Kinetics
Our study highlights that Hes4 might also constitute a crucial regulator of cell cycle kinetics. Cell cycle speed is known to vary during the time course of vertebrate retinogenesis [44,56,57], a phenomenon that has been associated with the transition from proliferative to neurogenic divisions in other parts of the central nervous system [58]. However, comparative analysis of proliferation kinetics between different retinal compartments at a given stage has not been performed so far. We found in this study that Hes4-expressing cells at the NR/ RPE border of the optic vesicle exhibit a much longer cell cycle compared to NR precursors. This likely constitutes a hallmark of neural stem cells as observed in adult RSCs of the CMZ ( [6] and our own unpublished data). How cell cycle speed is modulated in these cells is poorly documented. We here demonstrated that Hes4 prolongs cell cycle duration, mainly through a lengthening of G1 phase. In line with this result, sustained overexpression of Hes1 in brain neural pro-genitors was previously shown to downregulate expression of cell cycle regulators such as CyclinD1 and CyclinE2 and to cause G1 phase retardation [26]. Our data thus strongly support the hypothesis that Hes4 regulates some aspects of RSC homeostasis through the maintenance of slow cell cycle kinetics.

Hes4 Is Tightly Regulated by Multiple Inputs
We recently discovered that opposed and counterbalancing functions of Wnt and Hedgehog signaling modulate neural stem/progenitor cell proliferation in the postembryonic CMZ and that the two pathways reciprocally regulate each other's activity [41]. In line with this model, we found that the two pathways exhibit opposite effects on Hes4 expression in the forming CMZ.
The Hes4-dependent maintenance in the cell cycle associated with slow cell cycle kinetics is highly reminiscent of the phenotype resulting from Hedgehog signaling blockade in the retina [12]. Consistent with this, but in sharp contrast with Hes1 regulation in the mouse retina [53], we found that the Hedgehog pathway negatively impacts on Hes4 expression in the forming CMZ. In line with our previously proposed model [12,59], it would thus be interesting to investigate whether Hedgehog-driven downregulation of Hes4 is required for the transition from a slow cycling RSC phenotype to a fast cycling progenitor state (Fig. 7B).
Our results also highlight that the canonical Wnt pathway (a) is active in the presumptive RSC subpopulation and (b) positively modulates Hes4 levels in the forming CMZ. These results suggest that Wnt signaling may function upstream Hes4 in the maintenance of RSCs in a proliferative and undifferentiated state during retinogenesis (Fig. 7B). This hypothesis is substantiated by the fact that the Wnt pathway is required for proliferation in the postembryonic Xenopus CMZ [10], and that Wnt-dependent prevention of chick CMZ cell differentiation is mediated by cHairy1 [38]. A positive link between the Wnt receptors frizzled 5/8 and Hes1 gene has also recently been observed in the context of cell proliferation in the mouse neuroretina [60]. Besides, activation of the Wnt pathway through overexpression of stabilized b-catenin in mouse [61] or in the zebrafish apc mutant [62] results in lower proliferation rates in the peripheral retina. It is therefore tempting to speculate that Wnt signaling might control as well Hes4-driven slow cell cycle speed of neural stem cells in the Xenopus retina.
In addition to these opposite influences of Wnt and Hedgehog signaling, we found that Hes4 is able to repress its own transcription. Such a negative feedback loop has previously been reported for Hes1 and is known to contribute to oscillations of its expression in brain neural precursors [26,63]. However, Hes4-expressing cells in the retina exhibit features that are strikingly reminiscent of boundary cell properties (constitutive repression of proneural genes and reduced proliferation rates). In these boundary cells Hes1, like other Hes-related genes, does not oscillate and is rather persistently expressed at high levels [20,27]. This sustained upregulation is due to the action of Id factors that prevent Hes1 from binding to its own promoter [64]. As shown here, Id2 and Hes4 exhibit a similar expression pattern in the developing retina. It would thus be interesting to determine whether a similar interaction leading to sustained Hes4 expression might be at work in RSCs.

CONCLUSION
Altogether, our study opens new avenue onto the cell-of-origin of adult RSCs and the involvement of the transcriptional repressor Hes4 in their maintenance during embryonic development. Our results indeed suggest that Hes4 expression at the NR/RPE border of the optic vesicle foreshadows the adult RSC pool of the CMZ. Besides, our functional analysis converges toward a model whereby Hes4 acts as a safeguard of neural stemness features in this cell subpopulation by preventing its differentiation and maintaining it in a slowly proliferative state.