Igf signaling couples retina growth with body growth by modulating progenitor cell division

During embryonic development and postembryonic life of multicellular organisms, the proportions of overall body and organ size are actively controlled, being increased or decreased in a species-specific manner. Scaling of overall body and organ size is known to be regulated by systemic signals, which couple nutritional status to growth (Andersen et al., 2013). Conversely, differential growth of organs can be the result of altered sensitivity to systemic signaling (Tang et al., 2011) as well as of variant intrinsic signaling in the respective organ (Bosch et al., 2017; Twitty and Schwind, 1931).

Different eye dimensions of teleost species have been shown to be of functional relevance, as visual acuity is notably correlated with eye size (Caves et al., 2017). Vision and in particular retinal size and relative cell type composition are highly relevant for habitat-specific function and consequently for speciation. Deep-sea fish retinae show striking structural differences to the retinae of surface-dwelling fish, accommodating their specific need of increased light sensitivity (Caves et al., 2017; Darwish et al., 2015; de Busserolles et al., 2014; Wagner et al., 1998). How retinal growth is regulated in different fish species to achieve differential eye sizes and function is not understood.

Teleost fish such as medaka (Oryzias latipes) and zebrafish (Danio rerio) are particularly appealing model systems for studying organ size scaling. They display life-long postembryonic growth and each organ harbors distinct stem cell populations, which continuously self-renew, generating progenitor and, ultimately, differentiated cells (Aghaallaei et al., 2016; Centanin et al., 2014; Furlan et al., 2017). In the eye, growth of lens, retina and cornea is governed by adjacent stem cell niches (Mochizuki et al., 2014; Nowell and Radtke, 2017; Raymond Johns, 1977), and the precise regulation of continuous growth is of utmost importance to ensure the functional consistency of optical parameters such as photoreceptor density, matching of the focal point, photoreceptor to ganglion cell convergence and, therefore, vision.

Retinal growth in teleosts and amphibians is coordinated by the neuroretinal stem cell niche: the ciliary marginal zone (CMZ) (Hollyfield, 1971; Johns and Easter, 1977; Raymond Johns, 1977; Straznicky and Gaze, 1971). Two proliferating cell populations in the CMZ are responsible for the majority of postembryonic neurogenesis: multipotent, slowly dividing self-renewing stem cells located at the outermost periphery of the CMZ, and rapidly dividing progenitor cells located closer to the differentiated retina (Centanin et al., 2014; Raymond et al., 2006; Tsingos et al., 2019; Wan et al., 2016). Both stem and progenitor cells represent suitable cell type sources for altering retinal size, as they continuously proliferate and likely possess signal transduction machinery necessary for extrinsic signaling modulation. Which signals and which of the two CMZ cell types coordinate retinal growth with overall body size could not be addressed due, in part, to the lack of cell-type specific drivers.

A central integrator of organ size within a growing organism is hormonal signaling, translating extrinsic conditions such as nutrient availability into proportionate and coordinated growth (Andersen et al., 2013; Boulan et al., 2015). The insulin signaling pathway plays a central role in regulating embryonic development and growth, and its key function is underlined by a single nucleotide polymorphism in insulin-like growth factor 1 (igf1) as causative alteration in small dog breeds (Sutter et al., 2007). Similarly, mutations in components of the Igf signaling pathway or their knockdown lead to severe dwarfism phenotypes from fish to humans (Baker et al., 1993; Klammt et al., 2008; Liu et al., 1993; Schlueter et al., 2007). Ultimately, final organ size is specified by systemic signals and their organ-specific interpretation (Shingleton and Frankino, 2018). The expression and localization of Igf pathway components during teleost embryonic and larval development has been assessed in a variety of species (Ayaso et al., 2002; Boucher and Hitchcock, 1998; Radaelli et al., 2003a,b; Zygar et al., 2005). Consistently, Igf ligands and receptors (Igfr) were found in the developing and adult retina. igf1 is expressed in the postembryonic goldfish retina, and its receptor is expressed in the CMZ and inner plexiform layer (IPL) (Boucher and Hitchcock, 1998; Otteson et al., 2002). In zebrafish, Igf1r-mediated signaling is required for proper embryonic development, especially of anterior neural structures, and inhibition results in reduced body size, growth arrest and developmental retardation (Eivers et al., 2004; Schlueter et al., 2007). Despite the expression of its pathway components in the CMZ, the role of Igf1r signaling in the teleost CMZ as well as the consequences of its alteration on size and shape of the continuously growing neuroretina have not been addressed.

In this study, we address the coordination of retinal size with body growth and show that they are coupled by Igf1r signaling in the postembryonic retinal stem cell niche of medaka. We find that global Igf inhibition reduces proliferation in the CMZ. Uncoupling Igf signaling between body and retina by targeted constitutive activation of Igf1r specifically in stem and progenitor cells of the neuroretina increases proliferation, leads to an increase in eye size and uncouples retinal size from body size. Importantly, the resulting oversized retinae are structurally intact and correctly laminated with expanded thickness of the retinal layers, indicating that progenitor differentiation and neurogenesis are self-organized. We demonstrate that Igf1r activation decreases cell cycle length in the CMZ, expands the progenitor population and ultimately leads to increased neuronal cell numbers. By specifically dissecting the individual contribution of stem cells and progenitor cells to the oversized retinae, we uncover that retinal progenitor, but not stem, cells integrate the modulation by Igf1r signaling.

The retinal stem cell niche is located in a continuous ring in the CMZ at the periphery of the teleost eye (Fig. 1A). To address the involvement of the Igf signaling pathway in regulating retinal stem and progenitor cell proliferation in the medaka retina, we first assessed expression of receptors and ligands. We find igf1ra, igf2 and insrb expressed in the CMZ, and in specific layers of the differentiated retina (Fig. S1).

The activity of receptor tyrosine kinases like Igf1r is mediated by ligand-induced conformational changes and subsequent trans-phosphorylation (Laviola et al., 2007). To assess whether Igf1r-mediated signaling is active in the CMZ, we detected the phosphorylated Igf1r (pIgf1r) by immunohistochemistry. Single cells in the progenitor domain of the CMZ (around 2%, estimation based on quantification of Pcna-positive cells in Fig. 4 and Fig. S5) as well as Müller glia (MG) cells in the inner nuclear layer (INL) in the differentiated part of the retina were positive for pIgf1r (Fig. 1B). The sparse distribution of pIgf1r-positive cells likely reflects a snapshot of the highly dynamic process of the activation and phosphorylation of receptor tyrosine kinases (Kiyatkin et al., 2020; Varkaris et al., 2013).

Igf1r expression and activity data in the CMZ are consistent with Igf signaling regulating proliferation in the CMZ to scale retinal size relative to the body. To determine the impact of Igf1r signaling on the proliferation of retinal stem and progenitor cells in the CMZ and on MG cells, we employed the widely used Igf1r inhibitor NVP-AEW541 (Chablais and Jazwinska, 2010; Choi et al., 2013; Huang et al., 2013) and BrdU to detect proliferatively active cells that entered or went through S phase during the incubation time. Although the NVP-AEW541 inhibitor displays a high specificity for Igf1r, an impact on Insr and other related kinases cannot be entirely ruled out (Attias-Geva et al., 2011; García-Echeverría et al., 2004). Fish at hatching stage [Iwamatsu stage 40 (Iwamatsu, 2004)] were incubated in 10 µM NVP-AEW541 or DMSO together with BrdU for 24 h, and analyzed subsequently by immunostaining against BrdU (Fig. 1C,D). Inhibitor incubation resulted in a 30% decrease of cells undergoing S phase in the CMZ (Fig. 1E), indicating that Igf/insulin signaling contributes to regulating proliferation in the CMZ. In the absence of MG cell proliferation in the differentiated retina, an impact of the inhibitor NVP-AEW541 was not detectable. Importantly, the 24 h incubation with the Igf1r inhibitor did not impact on relative eye size (Fig. S2A), nor did average eye diameter or body length differ between inhibitor treated and control groups. Moreover, due to the global nature of the inhibitor assay, a non-cell-autonomous effect of this treatment cannot be excluded. These results suggest that ligands and receptors of the Igf/insulin signaling cascade expressed in the CMZ contribute to CMZ proliferation.

Our results indicate that Igf1r signaling represents a likely target to coordinate and modulate retina-specific growth. To further test this hypothesis, we established and employed tools for uncoupling Igf1r signaling in the neuroretina from the rest of the body by the precise and constitutive activation of Igf1r signaling in the CMZ.

Activation of the Igf signaling pathway was achieved by a constitutively active cd8a:igf1ra chimeric receptor (caigf1r). We targeted its expression to stem and progenitor cells of the CMZ by employing the rx2 promoter. The cd8a:igf1ra variant was generated by an in-frame fusion of the extracellular and transmembrane domain of medaka cd8a and the intracellular domain of medaka igf1ra, as previously described (Carboni et al., 2005). The rx2 promoter drives expression in stem cells and early multipotent progenitors in the CMZ, as well as in MG and photoreceptor (PRC) cells in the differentiated retina (Reinhardt et al., 2015).

The potential of the caIgf1r fusion receptor to induce signaling transduction via the PI3K/Akt axis had been previously validated in several studies by assaying pAkt levels (Carboni et al., 2005; Gusscott et al., 2016; Pappano et al., 2009). Hence, we validated the functionality of the rx2::caigf1r transgenic line and addressed the activation of the signaling cascade downstream of Igf1r in rx2-positive cells by the phosphorylation status of its downstream effector Akt. In retinal sections of wild-type hatchlings, no pAkt staining was present in the peripheral domain in the CMZ (Fig. S2B). In contrast, rx2::caigf1r retinae showed prominent pAkt staining in the CMZ, overlapping with caigf1r expression in the rx2 domain (Fig. S2C), and rx2-expressing MG cells and PRCs were also positive for pAkt. These results show that, in the retina of rx2::caigf1r transgenic fish, Akt is activated downstream of Igf1r in stem and progenitor cells in the CMZ, as well as in MG cells and PRCs in the differentiated retina.

To address the potential of Igf1r signaling to modulate retinal growth, we next examined hatchlings for changes in retinal morphology related to rx2::caigf1r expression. Intriguingly, transgenic rx2::caigf1r hatchlings displayed an oversized retina compared with wild-type siblings (Fig. 2A-C), while other body parts remained unaffected. Relative eye size (anterior-posterior eye diameter normalized to body length) was increased in rx2::caigf1r hatchlings by 15% (Fig. 2D, Table S1), persisting throughout postembryonic growth until adulthood, where relative eye size was enlarged by 9% in rx2::caigf1r fish (Fig. 2E, Table S2). These data indicate that CMZ-targeted activation of Igf1r signaling uncouples retinal growth from overall body growth in medaka, disrupting the coordinated, Igf1r-mediated scaling of retina and body size.

Size increase of a tissue can arise due to different mechanisms of tissue expansion, such as increase in cell size or number, or stretching and increase in fluid or pressure (Ritchey et al., 2012; Stujenske et al., 2011; Veth et al., 2011). To understand how constitutive Igf1r signaling impacts on retina size, we examined cryosections of wild-type and rx2::caigf1r hatchling retinae. Rx2::caigf1r eyes exhibited a noticeable expansion of the neuroretina in all dimensions, in contrast to wild-type controls (Fig. 2F,G). Importantly, nuclear layer morphologies and arrangement indicated that the overall retinal architecture in rx2::caigf1r fish remained intact. The stereotypical structure of the neuroretina with the CMZ at the periphery, and three nuclear and two plexiform layers in the differentiated part was undisturbed by retinal expansion.

We further assessed retinal architecture using the expression of rx2 as a landmark. All cell types expressing rx2 in wild type are present in the oversized rx2::caigf1r retinae, both in the CMZ as well as centrally in MG cells and PRCs (Fig. 2F,G). To characterize the apparent neuroretinal expansion, we compared the neuroretinal thickness in the peripheral, fully laminated region in rx2::caigf1r and wild-type retinae, along a line perpendicular to the IPL. inner plexiform layer (IPL). Retinal column height was increased by 23-24% on average in the dorsal as well as the ventral retina (Fig. 2H). In addition, we measured height of the retinal column in central regions, reflecting the embryonic contribution of rx2 expression in retinal progenitor cells to the oversized retinae. Column height of the central neuroretina in rx2::caigf1r hatchlings was increased by 15% compared with wild type (Fig. 2H). However, the increase was even more pronounced in peripheral, CMZ-derived regions, arguing for a greater impact of the post-developmental contribution via growth mediated by the CMZ.

To further substantiate our hypothesis that the increased size of the neuroretina is predominantly mediated by CMZ-derived cells, we assessed the relative contribution of the differentiated, central retinal region by proliferation analysis. A 2 h BrdU incubation of wild-type and rx2::caigf1r hatchlings yielded very low numbers of BrdU-positive cells per µm² retinal area in both groups, with a slight elevation of proliferative MG cells and PRCs in oversized retinae reflecting a minor to negligible contribution to the apparent neuroretinal expansion (Fig. S3A-C). Moreover, we also excluded the possibility that, reduced apoptosis levels in the retina contributed to the size increase. TUNEL analysis revealed comparable numbers of apoptotic cells in the CMZ of wild-type and rx2::caigf1r hatchlings, and increased levels of TUNEL-positive cells in the central region of rx2::caigf1r retinae (Fig. S3D-F). This indicates that neither increased central proliferation nor decreased apoptosis is a major contributor to cell number expansion in rx2::caigf1r retinae. Taken together, these results show that the activation of Igf1r signaling in the CMZ increases retinal diameter and layer thickness, and the enlargement is achieved by a neuroretinal expansion through the increase in cell number rather than in cell size.

The relative composition of retinal cell types reflects the functionality of the retina and differs considerably between teleost species. To elucidate whether in the oversized retinae cell type numbers were increased proportionally or whether cell type proportions were shifted, we analyzed the neuroretinal cell type composition in rx2::caigf1r and wild-type hatchlings in a 30 µm column spanning the entire neuroretinal column in the fully differentiated region proximal to the CMZ (dashed frame in Fig. 3A-D).

Cell type identity was determined by shape and position or by cell type-specific markers. PRCs are unambiguously characterized by their stereotypic localization in the outermost of the three nuclear layers. Retinal ganglion cells (RGCs) and displaced amacrine cells (ACs) located in the ganglion cell layer (GCL) were classified as one group. In the outer nuclear layer (ONL), cone PRCs were identified as outermost nuclear layer, and rods as the inner layer of the PRC nuclei. Other retinal cell types were distinguished by immunostaining with cell type-specific markers, such as GS, which labels MG cells, and Otx2, which labels bipolar cells (BCs) (Fig. 3A-D). Horizontal (HCs) and ACs in the INL were characterized as being Otx2 negative, together with their specific apicobasal localization and nuclear morphology (Fig. 3C,D). In the oversized retinae, the total cell number within the 30 µm wide column was increased by 15% on average in rx2::caigf1r retinae, with a more prominent increase in the ventral (21%) compared with the dorsal (10%) retina (Fig. 3E). The most prominent shift in the proportion of cell types was detected for ACs (24%) and BCs (31%) (Fig. 3F), while GCL and PRC numbers were rather similar in rx2::caigf1r and wild-type retinae (Fig. 3F). Some differences in cell type number changes were discernible between dorsal and ventral retina, with most distinctive increases in ACs, MG cells and BCs dorsally (Fig. S4A), and in ACs and BCs ventrally (Fig. S4B). These results indicate that Igf1r signaling activation in the CMZ shifts the neuroretinal cell type composition toward the INL and specifically to AC and BC fate.

Igf1r signaling has been shown to influence cell cycle progression in different in vitro and in vivo models (Hodge et al., 2004; Schlueter et al., 2007). To determine the impact of the stem and progenitor cell-specific activation of Igf1r signaling in rx2::caigf1r on cell cycle duration, we next analyzed cell cycle and S-phase length of the proliferating cells in the CMZ of wild-type and rx2::caigf1r hatchlings. To this end, we employed a dual-pulse S-phase-labeling regime using BrdU and EdU (Das et al., 2009; Klimova and Kozmik, 2014). Hatchlings were incubated in BrdU for 2 h, washed and incubated in EdU for 30 min before analysis. BrdU, EdU and Pcna staining (Fig. 4A,B) allowed the quantification of the fractions of cells positive for one or more marker, allowing cell cycle and S-phase length to be determined. In the oversized rx2::caigf1r retinae, the cell cycle length was reduced from 12 h to 10.5 h on average (Fig. 4C), whereas the S-phase length remained constant with an average duration of 4.5 h (Fig. 4D). This had a drastic impact on the number of BrdU-positive cells in the CMZ, which was more than doubled in rx2::caigf1r compared with wild-type fish (Fig. 4E), a fact that is also confirmed by the analysis of the EdU- and Pcna-positive cells (Fig. S5A,B). The ratio of BrdU- to Pcna-positive cells increased from 50% in wild-type to 57% in rx2::caigf1r hatchlings in the dorsal CMZ, and from 55% to 60% in the ventral CMZ, indicating shortened G phases in the rx2::caigf1r CMZ.

Taken together, these data show that, in rx2::caigf1r hatchlings, more cells in the CMZ progress faster through the cell cycle, thereby expanding retinal cell number resulting in oversized retinae with shifted cell type proportions and increased height of the neuroretinal column.

Differential responses of retinal stem and progenitor cells to extrinsic stimuli have been previously described (Centanin et al., 2014; Love et al., 2014). Given that an Igf signaling-mediated proliferative trigger in the combined stem and progenitor cell domain impacts on retinal size, we addressed the individual contribution of the stem and progenitor cell populations to the scaling response. This required to establish a retinal stem cell-specific driver as such a marker with the required specificity was not available for any vertebrate system. Exploiting existing resources for gene expression patterns, we focused on the cytosolic non-specific dipeptidase zgc:114181 (hereafter cndp), which had been identified in a systematic in situ screen of full-length medaka cDNAs (Alonso-Barba et al., 2016) and showed a promising expression pattern at the retinal periphery.

To establish specific tools and address the expression of cndp relative to rx2 we established transgenic reporter lines carrying putative regulatory regions driving a membrane-coupled GFP or nuclear mCherry reporter. The 5 kb genomic region upstream of the cndp-coding region was sufficient to recapitulate the mRNA expression pattern in transgenic reporter lines (cndp::eGFP-caax and cndp::H2A-mCherry) in the retina (Fig. 5A, Fig. S6A-D) as well as in the choroid plexi in the brain (Fig. S6A-D). In the retina, cndp-positive cells were found exclusively in a small, peripheral subset of the rx2-positive domain in the CMZ (Fig. 5A), with position likely representing retinal stem cells. To address the potency of those putative retinal stem cells and validate their nature (stem or progenitor cells), we employed a lineage analysis tool established in the lab that allows the determination of the potency of cells (Centanin et al., 2014). In brief, activation of Cre recombinase specifically expressed in cndp-positive cells was used to trigger a permanent switch in reporter gene expression in those cells and in all of their descendants. Stem cells and progenitor cells can be distinguished by their proliferative capacity. In the assay, stem cells generate continuous stripes (ArCoS) of labeled descending cells always connected to the CMZ. In contrast, progenitor cells, due to their limited proliferative capacity, generate footprints of labeled cells that are separated from the CMZ. We used the cndp promoter (Fig. 5B) characterized above to drive the expression of a tamoxifen-inducible Cre^ERT2 and combined it with the switchable GaudíRSG reporter line (Centanin et al., 2014). A permanent switch of reporter expression from mCherry to GFP was triggered by the tamoxifen-induced activation of nuclear Cre activity. We characterized the distribution, continuity and cell type identity of the clonally labeled progeny to conclude the nature of the cndp-positive mother cell.

GaudíRSG embryos injected with the cndp::Cre^ERT2 plasmid at the one-cell stage were induced with tamoxifen at hatchling stage. After 2-3 weeks, fish were analyzed for GFP expression by whole-mount immunostaining (Fig. 5C). Retinae displayed GFP-positive clones that all originated in the peripheral CMZ and were continuous with the differentiated retina (Fig. 5D,E). This indicates that cndp-positive cells are neuroretinal stem cells. Clones comprised cells of all three nuclear layers (Fig. 5F), demonstrating multipotency of their cells of origin (Centanin et al., 2014). Importantly, we never observed clonal footprints originating from progenitor cells. This establishes cndp as a bona fide marker for neuroretinal stem cells and as an invaluable tool for dissecting stem versus progenitor cell behavior.

To address whether stem or progenitor cell populations are expanded in response to activated Igf1r signaling, we took advantage of the stem cell-specific expression of cndp and used rx2 as a marker for stem and progenitor cells (Fig. 6A). Retinal stem cells were identified by expressing the cndp::H2A-mCherry reporter in a stable transgenic line, and retinal stem and progenitor cells were identified by an antibody specific for Rx2 (Reinhardt et al., 2015). Numbers for retinal stem and progenitor cells were determined in hatchlings of wild type as well as in rx2::caigf1r retinae harboring the cndp::H2A-mCherry reporter (Fig. 6B,C).

Numbers of cndp-positive stem cells were comparable in wild-type and rx2::caigf1r retinae, ranging from 3 to 11 in the dorsal and from 0 to 8 stem cells per section in the ventral CMZ. The number of cndp-positive stem cells was rather stable, with a slight increase in rx2::caigf1r retinae (Fig. 6D). In contrast to that, the Rx2 domain was prominently expanded in rx2::caigf1r hatchlings (Fig. 6E,F), with Rx2-positive stem and progenitor cells in the CMZ more than doubled in rx2::caigf1r versus wild-type retinae (Fig. 6G), arguing that the progenitor, but not the stem cell, population is expanded by Igf1r signaling activation. In line with this observation, pIgf1r staining indicative of the activation of the Igf signaling cascade was detected only in progenitor cells and not in cndp-positive stem cells (Fig. S6E,F).

As retinal stem cell numbers in rx2::caigf1r hatchlings were almost comparable with wild type, we next wanted to understand whether retinal stem cells respond to the modulation of Igf1r signaling. We generated a transgenic line in which activation of Igf1r signaling was specifically targeted to cndp-positive retinal stem cells (cndp::caigf1r). Functionality of the construct was evident through enlarged choroid plexi in cndp::caigf1r fish (Fig. S7A,B). However, the GFP expression domain in the retina was unaltered (Fig. S7C,D). We examined relative eye size in hatchling and adult cndp::caigf1r fish, and neither stage displayed an alteration in relative eye size compared with wild-type siblings (Fig. S7E,F). These results demonstrate that the retinal stem cell population expressing cndp does not expand upon the activation of Igf1r signaling or by accumulation of cndp over time in adult stages. This further confirms and refines our findings that the progenitor, but not stem, cell population is expanded in rx2::caigf1r retinae.

Taken together, we demonstrate that retinal growth can be uncoupled from overall body growth through the activation of Igf1r signaling targeted to the CMZ. This intrinsic modulation elicits differential responses in stem and progenitor cell populations in the medaka retina, leading to a shortened cell cycle and consequential increase of retinal progenitors but not stem cell number (Fig. 6H,I).

In this study, we have investigated how organ growth is coordinated with overall body growth in medaka by uncoupling growth control in the retina versus body. We focused on the retina and dissected the role of Igf1r signaling in regulating proliferation of stem cells versus progenitor cells in the retinal stem cell niche and its impact on retina size and composition at postembryonic stages. We show that proliferation in the retinal stem cell niche is dependent on and triggered by Igf1r-mediated mitogenic signaling. Combining expression analysis, inhibitor assays and gain-of-function approaches, we show that the Igf pathway is functioning in the postembryonic CMZ, where it actively controls proliferation. Our analyses indicate a localized paracrine signaling hub, with Igf1r activity in single progenitors in the CMZ. The inhibition of Igf1r signaling results in a decreased proliferation of CMZ cells. Conversely, the targeted constitutive activation of Igf signaling in the CMZ results in a clear increase of retina size and a shift in relative composition of the retinal layers. Strikingly, this is due to a specific response of the progenitor cells, while the proliferation of stem cells is not affected. Progenitor cells speed up their cell cycle without affecting the subsequent differentiation potential in response to Igf1r signaling activation (Fig. 6I). The specific activation of Igf signaling results in an increased proliferation rate in the progenitor pool that, in turn, causes a shift in retinal layer composition with the potential to impact on functional optical parameters.

Upon activation of Igf1r signaling in the CMZ, the thickness of the neuroretina increases by one-quarter. Intriguingly, the enlarged retinae are structurally intact, displaying correct lamination and differentiation. Although retinal enlargement due to enhanced progenitor proliferation might be an expected phenotype, the resulting intact retinal lamination with shifted cell type composition indicates a high degree of self-organization, depending on the number of progenitor cells. The specific action of the Igf/insulin pathway as signal that integrates nutritional status (Boulan et al., 2015) opens the field for interesting evolutionary hypotheses. In response to an Igf1r-mediated proliferative trigger in the progenitor cell domain, higher retinal columns, and thus an extended light path, are generated. Coordinated changes in relevant optical parameters, such as light path length, lens size and photoreceptor density, might therefore increase visual acuity. Thus, a signal representing nutrient availability can prompt an immediate selective advantage when preferentially acting on the retina and accessory tissues.

In the teleost retina, several populations of progenitor cells exist that differ in their proliferative and lineage capacities. Early progenitor cells in the CMZ are known to undergo self-renewing divisions generating two progenitor cells (PP), while old progenitor cells favor differentiating divisions (Wan et al., 2016). Among these are different lineage-specified progenitors that have been shown to produce different sets of cell type ratios upon modification of their transcriptional signatures (Pérez Saturnino et al., 2018). The activation of Igf1r signaling elicits two major changes in progenitor cells: first, increased numbers of Pcna- and rx2-positive cells indicates an increase in self-renewing PP divisions; second, a pronounced increase in ACs and BCs suggests that Igf1r activation might either convey certain cell fates or favor expansion of a specific progenitor population that is lineage committed to generate INL cells.

We showed that the uncoupling of Igf signaling in the retina by CMZ-specific constitutive activation of Igf1r signaling preferentially expands the rx2-positive progenitor but not the stem cell population. Targeting stem cells by specific Igf1r activation in cndp-expressing cells does not impact on retinal size, indicating that retinal stem cells do not respond to Igf1r signaling with an immediate and long-lasting increase in proliferation, nor could a cumulative effect be observed in adult fish. Conversely, progenitor cell numbers are more than doubled in response to constitutive activation of Igf1r signaling, rendering the rx2-positive progenitor population receptive for this mitogenic stimulus. Differential responses of stem and progenitor cells have been observed in the opposite direction after nutrient deprivation in the Xenopus retina. While stem cells are resistant to nutrient deprivation and to mTOR inhibition, retinal progenitors respond with changes in proliferation and differentiation in a mTOR-mediated manner (Love et al., 2014). These differences between retinal stem and progenitor cells are complementary to our observations whereby stem cells appear resistant to the Igf1r-mediated mitogenic stimulus, while progenitor proliferation is increased. Interestingly, outside the retina, divergence at the level of stem and progenitor cell behavior has been proposed as the basis for varying brain sizes in different amniote species (Nomura et al., 2013). Studies in geckos have proposed that progenitor cells are key for the evolutionary scaling of brain size in amniotes. The lower cell division rates of progenitor cells, their longer cell cycle and the predominance of self-renewing progenitor cell divisions are likely responsible for the comparatively smaller size of the gecko brain (Nomura et al., 2013).

Enlargement of the eye had been reported in few experimental and mutant conditions. Although the overexpression of Yap by mRNA injection could increase eye size in Xenopus embryos (Cabochette et al., 2015), retinal morphology and lamination were disturbed in those Yap-overexpressing tadpoles (Cabochette et al., 2015). Increase in retinal size has also been observed in a patched2 mutant in zebrafish where retinal patterning and morphology are largely intact, but the number of MG cells is reduced (Bibliowicz and Gross, 2009). In the CMZ of patched2 mutant embryos, the number of retinal progenitor cells is increased, but their cell cycle length is unaffected (Bibliowicz and Gross, 2009). We observed that, in response to active Igf1r signaling, the overall cell cycle was shortened without affecting S-phase length, in agreement with results from various systems in which Igf signaling influences G1- or G2-phase length (Hodge et al., 2004; Schlueter et al., 2007; Wang et al., 2015). The observed shortened cell cycle length of retinal progenitor cells of the oversized retinae is likely causal for the expansion of the CMZ and ultimately the entire retina in the rx2::caigf1r fish.

In contrast to eye size expansion resulting from increased retinal cell numbers, zebrafish and goldfish bugeye mutants exhibit enlarged eyes due to increased intraocular pressure (Kon et al., 2020; Stujenske et al., 2011; Veth et al., 2011). The bugeye phenotype is an adult-onset phenotype, and shows retinal cell density decrease and thinning of all nuclear layers. Moreover, bugeye mutants display relative refractive errors concordant with myopia due to vitreous chamber expansion between lens and retina (Veth et al., 2011), which displaces the retina behind the point at which the lens focuses light. Interestingly, the expansion of the eye in adult fish did not impact on the size of the lens in those mutants. We also did not observe an increase in lens size in rx2::caigf1r fish compared with wild-type siblings, indicating that there is no paracrine signal emanating from the neuroretinal niche in response to elevated Igf signaling to coordinate the size of the lens. Those data suggest an independent size control of retina and lens, as opposed to our previous results on neuroretina and retinal pigmented epithelium coordination, where behavior of neuroretinal stem cells in the CMZ drives retinal shape and impacts retinal pigmented epithelium stem cells that passively follow (Tsingos et al., 2019). Uncoupling neuroretinal Igf signaling from the rest of the body, as described here, provides an indication that independent control of retina and lens size exists during development and postembryonic growth. However, expression of the insulin receptor and IGF1 receptor at high levels at the lens periphery in embryonic and P0 mice (Xie et al., 2007), as well as proliferation of lens epithelial cells upon IGF1 treatment in rat lens epithelial explants (Iyengar et al., 2006), indicates that control of lens growth and size could also be modulated by Igf signaling. In the future, it will be important to address whether Igf signaling impacts lens germinal cell proliferation in the medaka retina and whether Igf signaling contributes to the regulation of lens and retina sizes to achieve a continuous functional consistency during the life-long growth of the eye.

The targeted isolated size increase of the retina and the concomitant change in cell type composition opens a wide area for discussing the evolutionary and ecological significance of coordinating retinal size and architecture. Throughout the teleost clade, adaptations to specific habitats and niches are evident in the retinal architecture. Surface-dwelling fish, such as medaka, have two layers of PRCs, one light-sensitive rod and one cone layer responsible for color vision, while zebrafish possess one layer of cones as well as three to four layers of rods for enhanced light perception, as they live in deeper waters (Lust and Wittbrodt, 2018). One particularly interesting example of retinal architecture changes is the ‘four-eyed’ fish Anableps anableps, which displays structural differences within the dorsal and ventral retina to accommodate its specific optic requirements. The ventral retina features an INL that is twice as thick as the dorsal INL, concordant with increased proliferation in the ventral compared with the dorsal CMZ during the development of the larval eye (Perez et al., 2017). As we observe a preferential increase in the INL neuron population in rx2::caigf1r, it is tempting to speculate that an altered proliferation of progenitor cells (initiated by altered Igf signaling) in the ventral CMZ contributes to manifesting the structural differences in the Anableps anableps retina. We speculate that Igf signaling can act as an evolutionary handle through which retinal size, morphology, cell type composition and, consequently, function can be adapted by modifying signaling activity in distinct populations of progenitor cells in the CMZ and beyond.

Medaka (Oryzias latipes) used in this study were kept as closed stocks at Heidelberg University. All experimental procedures and husbandry were performed in accordance with the German animal welfare law and approved by the local government (Tierschutzgesetz §11, Abs. 1, Nr. 1, husbandry permit AZ 35–9185.64/BH and line generation permit AZ 35–9185.81/G-145-15). Fish were maintained in a constant recirculating system at 28°C on a 14 h light/10 h dark cycle. The following stocks and transgenic lines were used: wild-type Cabs, Heino mutants (Loosli et al., 2000), rx2::caigf1r rx2::lifeact-eGFP, cndp::H2A-mCherry, cndp::eGFP-caax, cndp::H2B-eGFP, cndp::caigf1r cndp::H2B-eGFP, cndp::Cre^ERT2 and GaudíRSG (Centanin et al., 2014). All transgenic lines were created by microinjection with Meganuclease (I-SceI) in medaka embryos at the one-cell stage, as previously described (Thermes et al., 2002).

The constitutively active Igf1r variant (Cd8a:Igf1ra, caIgf1r) was generated by an in-frame fusion of the codon-optimized extracellular and transmembrane domain of olCd8a (synthesized by Geneart) and the intracellular domain of the medaka Igf1ra, as previously described (Carboni et al., 2005). To generate the rx2::caigf1r rx2::lifeact-eGFP line, two F0 fish were crossed, and each subsequent generation was derived from one transgenic male outcrossed to wild-type Cabs. All experiments were carried out in the F2 to F4 generations.

The cndp::Cre^ERT2 plasmid (cndp Ensembl ID: ENSORLG00000003701) was generated by cloning the 5 kb cndp regulatory region in a pBS/I-SceI-vector containing a tamoxifen-inducible Cre recombinase. The plasmid contains cmlc2::eCFP as insertional reporter.

For BrdU incorporation, hatchlings (stage 40; Iwamatsu, 2004) were incubated in 2.5 mM BrdU (Sigma-Aldrich) diluted in 1×embryo rearing medium (ERM; 17 mM NaCl, 40 mM KCl, 0.27 mM CaCl₂, 0.66 mM MgSO₄ and 17 mM HEPES) for 2 h. For EdU incorporation, hatchlings were incubated for 30 min in 250 µM EdU (ThermoFisher) diluted in 1×ERM. Quantification of BrdU-positive cells was performed in four retinae from individual hatchlings. Cell counts were performed in z=6 µm of two or three central sections per retina.

For inhibition of Igf1r, hatchling fish were incubated in 10 μM NVP-AEW541 (Selleckchem, 10 mM stock solution solved in DMSO) diluted in 1×ERM at 28°C for 24 h. In a parallel control group, hatchling fish were incubated in 0.1% DMSO/1×ERM at 28°C for 24 h. Directly afterwards, fish were euthanized and fixed for analysis.

For Cre^ERT2 induction, hatchlings were treated with a 5 µM tamoxifen solution (Sigma-Aldrich) in 1×ERM overnight.

Fish were euthanized using 20× Tricaine and fixed overnight in 4% PFA, 1× PTW at 4°C. After fixation, samples were washed with 1× PTW and cryoprotected in 30% sucrose in 1× PTW at 4°C. To improve section quality, the sections were incubated in a half/half mixture of 30% sucrose and Tissue Freezing Medium (Leica) for at least 3 days at 4°C. Serial sections (16 µm) were obtained on a cryostat. Sections were rehydrated in 1× PTW for 30 min at room temperature. Blocking was performed for 1-2 h with 10% NGS (normal goat serum) in 1× PTW at room temperature. The respective primary antibodies were applied diluted in 1% NGS overnight at 4°C. The secondary antibody was applied in 1% NGS together with DAPI (Sigma-Aldrich, D9564; 1:500 dilution in 1× PTW of 5 mg/ml stock) for 2-3 h at 37°C. Slides were mounted with 60% glycerol and kept at 4°C until imaging.

BrdU and Pcna antibody staining was performed with an antigen retrieval step. After all antibody staining and DAPI staining, except for BrdU/Pcna, were complete, a fixation for 30 min was performed with 4% PFA. Slides were incubated for 1.5 h at 37°C in 2 N HCl solution, and pH was recovered by washing with a 40% Borax solution in 1× PTW before incubation with the primary BrdU or Pcna antibody.

The EdU staining reaction was performed after all other antibody staining was completed using the Click-iT EdU Alexa Fluor 647 Flow Cytometry Assay Kit according to the manufacturer's protocol (Thermo Fisher).

TUNEL staining was performed after all other antibody staining was completed using the In Situ Cell Death Detection Kit Fluorescein (Roche). Staining was performed according to the manufacturer's protocol with the following modification: washes were performed with 1× PTW instead of PBS.

Fish were euthanized using 20× Tricaine and fixed overnight in 4% PFA in 1× PTW at 4°C. After fixation, samples were washed with 1× PTW. Fish were bleached with 3% H₂O₂, 0.5% KOH in 1× PTW for 2-3 h in the dark. Retinae were enucleated and permeabilized with acetone for 15 min at −20°C. Blocking was performed in 1% bovine serum albumin (Sigma-Aldrich), 1% DMSO (Roth/Merck), 4% sheep serum (Sigma-Aldrich) in 1× PTW for 2 h. Samples were incubated with primary antibody in blocking buffer overnight at 4°C. The secondary antibody was applied together with DAPI in blocking buffer overnight at 4°C. Primary antibodies were used at 1:200, secondary antibodies at 1:250 and DAPI at 1:500.

The following primary antibodies were used: anti-BrdU (rat; Abcam, ab6326; 1:200), anti-DsRed (rabbit; Clontech, 632496; 1:500), anti-eGFP (chicken; Life Technologies, A10262; 1:500), anti-GS (mouse; Chemicon, MAB302; 1:500), anti-Otx2 (goat; R&D systems, AF1979; 1:200), anti-pAkt (rabbit; Cell Signaling, 4060; 1:200), anti-Pcna (mouse; Millipore, CBL407; 1:100), anti-pIgf1r (rabbit; Abcam, ab39398; 1:100), anti-Rx2 (rabbit; Reinhardt et al., 2015; 1:500) and anti-Zpr-1 (mouse; ZIRC, ZDB-ATB-081002-43; 1:200). The following secondary antibodies were used (at 1:750): anti-chicken Alexa Fluor 488 (donkey; Jackson, 703-485-155), anti-goat Alexa 633 (donkey; Life Technologies, A-21082), anti-mouse Alexa 546 (goat; Life Technologies, A-11030), anti-mouse Alexa 647 (donkey; Jackson, 715-605-151), anti-rabbit Alexa Fluor 488 (goat; Life Technologies, A-11034), anti-rabbit DyLight549 (goat; Jackson, 112-505-144), anti-rabbit Alexa Fluor 647 (goat; Life Technologies, A-21245) and anti-rat DyLight488 (goat; Jackson, 112-485-143). DAPI (Sigma-Aldrich, D9564) nuclear counterstaining was performed as described by Inoue and Wittbrodt (2011).

To determine cell cycle length of retinal progenitor cells, dual-pulse S-phase labeling with BrdU and EdU were used as previously described (Das et al., 2009; Klimova and Kozmik, 2014). This approach allows an estimation of S-phase length by consecutive BrdU and EdU incubation, and subsequent extrapolation of total cell cycle length in the Pcna-positive progenitor population. Hatchlings were incubated for 2 h in BrdU, then 30 min in EdU before fixation. Pcna antibody staining was used to label all cycling retinal progenitor cells. Pcna-, EdU- and BrdU-positive cells, as well as cells positive for only BrdU, were quantified.

Cell-cycle length and S-phase length were determined with the following formulas: T_{cell cycle}=2 h *(Pcna⁺ cells/BrdU⁺ EdU⁻) as calculation of the ratio of all cycling cells (Pcna⁺) to S phase exit (BrdU⁺ EdU⁻); T_{S phase}=2 h *(EdU⁺ cells/BrdU⁺ EdU⁻) as calculation of the ratio of S-phase entry (EdU⁺) to S-phase exit (BrdU⁺ EdU⁻).

Four retinae from individual hatchlings were used for analysis. Cell counts were performed in z=6 µm of two or three central sections per retina. T_{cell cycle} and T_{S phase} were determined for individual sections.

Whole-mount in situ hybridizations using NBT/BCIP detection were carried out as previously described (Loosli et al., 1998). Afterwards, samples were cryoprotected in 30% sucrose in 1× PTW overnight at 4°C and 20 µm thick serial sections were obtained on a Leica cryostat. Sections were rehydrated in 1× PTW for 30 min at room temperature and washed several times with 1× PTW. Slides were mounted with 60% glycerol and kept at 4°C until imaging.

All immunohistochemistry images were acquired by confocal microscopy at a Leica TCS SP8 with 20× or 63× glycerol objective. Sections of whole-mount in situ hybridization were imaged using a Zeiss Axio Imager M1 microscope. Images of whole hatchlings were acquired with a Nikon SMZ18 stereomicroscope equipped with aNikon DS-Ri1 camera.

Images were processed via Fiji image processing software. Statistical analysis and graphical representation of the data were performed using the Prism software package (GraphPad). Box plots show median, and 25th and 75th percentiles, with whiskers from minimum to maximum data points. All data points, represented by black dots, are superimposed on box plots. Normal distribution was tested with the Shapiro-Wilk normality test, and unpaired two-tailed t-tests or Mann–Whitney tests were performed to determine statistical significance. The P<0.05 was considered significant, P values are given in the figure legends. Sample size (n) is mentioned in every figure legend, all sections were treated as independent samples. No statistical methods were used to predetermine sample sizes, but our sample sizes are similar to those generally used in the field. The experimental groups were allocated randomly; no blinding was carried out during allocation.

We thank the Wittbrodt lab for constructive discussions on the project, A. Cornean, T. Tavhelidse, E. Tsingos, V. Weinhardt and L. Zilova for critical reading of the manuscript, and L. Centanin, N.S. Foulkes and S. Lemke for constructive input and feedback on the manuscript. We are grateful to A. Saraceno, E. Leist and M. Majewski for fish husbandry.

The peer review history is available online at https://journals.biologists.com/dev/article-lookup/148/7/dev199133/