Introduction

The retina and cornea represent indispensable components of the vertebrate visual system that ensure proper vision (Casey et al., 2023). The retina, a multilayered sensory tissue at the back of the eye, converts light into neural signals, initiating visual perception. It consists of various specialized cells organized into three nuclear layers and two synaptic layers. Photoreceptors, including rods and cones, are located in the outer nuclear layer (ONL) and are responsible for converting light stimuli into electrical signals. These signals are then transmitted to bipolar cells in the inner nuclear layer (INL), which, in turn, establish synapses with retinal ganglion cells (RGCs), whose axons form the optic nerve that send the electrical impulse to the brain. Additionally, horizontal and amacrine cells, also positioned in the INL, provide lateral connections, offering feedback and feedforward signals. Finally, Müller glial (MG) cells, with nuclei located in the INL and bodies extending across the entire retina, provide structural, metabolic, and functional support to the other retinal cells, thus maintaining retinal function and integrity (Baden et al., 2020).

The cornea, positioned at the forefront of the eye, is a transparent and refractive structure responsible for transmitting and focusing light onto the retina, which is devoid of blood vessels. The outermost layer of the cornea is a stratified squamous epithelium that undergoes continuous regeneration. Stem cells responsible for this replenishment reside within a specialized niche located at the periphery, adjacent to the conjunctiva, known as the limbus (Lavker et al., 2004; West et al., 2015). Within this limbal niche, long-lived limbal epithelial stem cells (LESCs) give rise to a more differentiated progeny responsible for the regeneration of the entire corneal epithelium throughout two distinct pathways of cell migration: one within the basal layer, moving from the limbus to the center of the cornea, and another from the basal layer to the terminally differentiated, desquamated cells at the epithelial surface (Yazdanpanah et al., 2017).

SOX9 is a transcription factor involved in the regulation of multiple developmental processes (Jo et al., 2014). In humans, mutations affecting SOX9 cause campomelic dysplasia (CD), a syndrome characterized by skeletal malformations and XY sex reversal (Wagner et al., 1994; Foster et al., 1994). In the adult mice, SOX9 has been shown to play a role in the homeostasis of neural, liver, pancreas, intestine and nail stem cells (Scott et al., 2010; Furuyama et al., 2011; Lao et al., 2022), as well as in the maintenance of Sertoli cells (Barrionuevo et al., 2016), among others.

Several works have addressed the role of Sox9 during the developing eye. In the context of the mouse retina, Sox9 is expressed in multipotent mouse retinal progenitor cells, undergoes downregulation during neuronal differentiation, and then the gene is exclusively expressed in Müller glial cells and retinal pigment epithelia (RPE) until adulthood (Poché et al., 2008; Muto et al, 2009). Deletion of Sox9 in developing retinal cells using a Chx10-Cre mouse line revealed that Sox9 mutant retinae exhibited normal lamination as well as nuclear layer thickness comparable to control samples, indicating that loss of Sox9 alone likely does not affect global retinal fate determination (Poché et al., 2008). shRNA-induced downregulation of Sox9 and Sox8 in retinal explants resulted in a reduced population of Müller glial cells and a moderate increased in the proportion of rod photoreceptors, indicating that both SoxE genes play roles in the specification of Müller glial cells (Muto et al, 2009). Similar results were obtained when Sox9 was conditionally deleted in the developing retina using a Pax6-α-Cre mouse line (Goto et al., 2018). In the mature RPE, Sox9 collaborates with other transcription factors, such as OTX2 and LHX, to regulate the visual cycle, a series of biochemical reactions that regenerate 11-cis-retinal, the chromophore essential for phototransduction in photoreceptor cells (Masuda et al., 2014). Three studies have investigated Sox9 inactivation in RPE cells. While one study reported no obvious morphological abnormalities in the retina and RPE upon conditional deletion of the gene (Masuda et al., 2014), two subsequent studies demonstrated that Sox9 was required for proper choroid differentiation (Goto et al., 2018; Cohen-Tayar et al., 2018).

In the cornea, transcriptome analysis from slow cycling cells identified Sox9 as a putative stem cell marker or determinant, and immunohistochemical analyses confirmed that the gene is expressed in LESCs and in corneal epithelial cells (Sartaj et al., 2017; Parfitt et al., 2015; Menzel-Severing et al., 2018). RNAi silencing of the gene in cultured limbal stem cells resulted in reduced expression of progenitor cell markers and increased expression of differentiation markers. Further analyses showed that Sox9 and Wnt/ß-catenin signaling antagonizes to maintain a balance between quiescence, proliferation, and differentiation limbal stem cells (Menzel-Severing et al., 2018).

Although several studies have addressed the role of Sox9 in retinal and corneal development (Poché et al., 2008; Muto et al., 2009; Masuda et al., 2014; Sartaj et al., 2017; Parfitt et al., 2015; Menzel-Severing et al., 2018), its specific in vivo contributions to the maintenance and regeneration of these tissues in adult mammals remain unknown.

In this work, we used a tamoxifen-inducible Cre/LoxP system to inactivate the Sox9 gene in adult mice and a lineage-tracing mouse strain to study the fate of Sox9-expressing cells. With these tools, in conjunction with scRNAseq analysis, we conducted a comprehensive study of morphological changes, cell-specific marker expression, and lineage tracing to determine the role of Sox9 in maintaining retinal integrity and corneal stem cell homeostasis.

Results

Deletion of Sox9 leads to retinal degeneration in adult mice

To evaluate the role of Sox9 in adult retinal homeostasis, we conditionally deleted the gene in adult mouse cells using a ubiquitous tamoxifen-inducible CAGG-CreER^TM recombinase (Hayashi and McMahon, 2002) and a conditional Sox9^flox/flox allele (Akiyama et al., 2002) (Fig. 1A). Histological examination of retinas from control (Cre-negative Sox9^flox/flox; n = 7) and CAGG-CreER^TM;Sox9^flox/flox (hereafter Sox9^Δ/Δ; n = 24) adult mice (two months old) at different days after TX administration (DATX) revealed that most of the analyzed mutant samples (n = 18/24; Table S1), showed either no relevant morphological defects in the retinal sections or a mild phenotype in which the main retinal nuclear layers (INL and ONL) occupied a smaller area and their edges did not appear as clearly differentiated as in the control retinas. However, we also observed an extreme phenotype in a reduced number of mutant retinas (n = 6/24), characterized by the complete loss of the ONL (Fig. 1B). We categorized Sox9^Δ/Δ retinas into “mild” and “extreme” phenotypes in order to facilitate interpretation of our data. Classification was based on a qualitative assessment of ONL integrity in histological sections. Specifically, samples were classified as “extreme” when the ONL was completely depleted, and as “mild” when the ONL persisted, albeit variably reduced in thickness. This phenotypic classification reflects observable structural differences rather than a fixed quantitative threshold. Some variability exists within the “mild” group, likely due to differences in recombination efficiency and the mosaic nature of tamoxifen-induced Cre-mediated Sox9 deletion. The severity of the phenotype showed no correlation with the time elapsed after tamoxifen administration, as we found mutant retinas with no apparent phenotype at 100 DATX and others with an extreme phenotype at 20 DATX (Table S1). Retinal degeneration was never observed in mice that had not been tamoxifen-treated, nor any other control groups, making the presence of the retinal degeneration allele of photoreceptor cGMP phosphodiesterase 6b (Pde6b^rd1) unlikely in our mice (Bowes et al., 1990). However, we acknowledge that definitive exclusion of this possibility would require PCR-based genotyping.

Figure 1.

During retinal development, Sox9 is expressed in neural RPCs but once the neural retina is differentiated, Sox9 expression is restricted to MG and RPE cells (Poché et al., 2008; Masuda et al., 2014). This is, indeed, the SOX9 expression pattern we observed in retinal sections obtained from adult control mice (Fig. 1C). To explain the phenotypic diversity shown by the TX-treated mice, we assessed the efficiency of Sox9 deletion in mutant mice. For this, we performed SOX9/SOX8 double immunofluorescence, as both transcription factors are normally co-expressed in MG cells (Fig 1C; Muto et al., 2009), and we consistently observed the presence of many SOX8⁺ cells that lacked SOX9 expression in the retinas of mutant TX-treated mice (Fig. 1C). We calculated the percentage of SOX8⁺ cells that also co-expressed SOX9 in control and in Sox9^Δ/Δ retinas with mild and extreme phenotypes (Fig. 1D; Tables S1 and S2). In control MG cells, all SOX8⁺ cells also expressed SOX9. In contrast, in mutants with a mild phenotype, the average percentage was 55%, although with a large standard deviation (55% ± 23%),while in extreme mutants, the percentage was significantly lower (16% ± 12%). Altogether, these results indicate that the severity of the phenotype is directly related to the efficiency of Sox9 inactivation, and that only individuals with Sox9 inactivation occurring in at least 80% of the cells expressing the gene exhibit the extreme phenotype.

Next, we used cell-specific molecular markers to characterize the most affected Sox9-deficient retinas. Initially, we focused on the MG cell layer by means of SOX8 immunofluorescence. We counted the number of SOX8+ cells per 100 μm of INL, and we found a significant reduction in the number of MG cells in mutant retinas (control, 18.5 ± 3.5, n = 7; mutant, 10.5 ± 1.2 n= 6; p = 0.000107, Mann-Whitney U test; Fig 1E; Table S3). In addition to the reduced number of SOX8⁺ cells, we observed areas within some Sox9 mutant INL layers that completely lacked SOX8⁺ cells (Fig. 1C arrowheads), indicating that these regions were entirely devoid of MG cells.

To independently validate the loss of MG cells in Sox9-deficient retinas, we examined the expression of S100, a cytoplasmic marker that labels the processes of adult Müller cells. In control retinas, strong S100 immunoreactivity was observed across the inner retina, outlining the typical radial projections of Müller glia (Fig. 1G). In contrast, Sox9^Δ/Δretinas with an extreme phenotype exhibited a marked reduction in S100 signal (Fig. 1F). Given the diffuse cytoplasmic localization of S100, we quantified its expression by measuring the fluorescence signal within a defined surface area of the retina. This analysis revealed a statistically significant reduction in S100 signal intensity in mutant samples (including both mild and extreme phenotypes) compared to controls (Fig. 1G; Table S4), further supporting the loss of MG cells upon Sox9 deletion.

Since the main observable consequence of Sox9 deletion in the adult mouse retina is the absence of the ONL layer, which harbors both rod and cone photoreceptors, we further examined the status of these cell types by immunofluorescence. Cone photoreceptor cells were assessed by retinal whole-mount staining with two different opsins: OPN1SW (short-wavelength-sensitive opsin, referred to as S opsin) and OPN1LW (medium- and long-wavelength-sensitive opsin, referred to as M opsin). In control retinas, we observed uniform staining on the ventral surface and on the entire surface for S opsin and M opsin, respectively. In contrast, in Sox9-deficient retinas we could only identify small isolated immunofluorescent patches (Fig. 2A. upper). Double immunofluorescence on retinal sections confirmed that mutant retinas lacked expression for both cone photoreceptors (Fig. 2A, bottom). We also performed immunofluorescence for RHO, a marker for rod photoreceptor cells, on retinal whole-mounts and sections. We found a pattern of expression similar to that described for M-opsin in both the control and mutant retinas. These results evidence that Sox9 is required for the maintenance of the two photoreceptor cell types present in the adult mouse retina (Fig. 2B).

Figure 2.

We also studied the status of other retinal cell types. The transcription factor BRN3A was used to identify ganglion cells (Nadal-Nicolás et al., 2009), which were shown to decrease in number in the mutant retinas, compared to control ones (Figs. 2C and 2D and Table S5; n = 5 controls, n = 12 mutants; Mann-Whitney U test, P = 3 × 10^-4). Similarly, double immunodetection of the transcription factors PAX6 and AP2A was used to identify both amacrine and horizontal cells (Fig. 2E), as previously described (Marquardt et al., 2001; Barnstable et al., 1985; Edqvist and Hallböök, 2004), showing a similar reduction in both cell types in degenerated retinas (Figs. 2F and 2G and Table S6; AP2α+ amacrine cells: n = 3 controls, n = 8 mutants; 2-sample T-tests P = 0.029; PAX6+/AP2α− horizontal cells: n = 3 controls, n = 8 mutants; Mann-Whitney U test P = 0.021). These findings indicate that the loss of Sox9 in the adult retina ultimately leads to the degeneration of multiple inner retinal neuronal populations, beyond the previously described effects on photoreceptors and Müller glia.

Photoreceptors undergo apoptosis in the absence of Sox9

Since photoreceptors are absent in severely affected Sox9-mutant retinas, we conducted TUNEL assays to study the role of cell death in the process of retinal degeneration. In control samples (n=5), almost no TUNEL signal was observed in the retina. In contrast, Sox9^Δ/Δ mice (n=15) showed numerous TUNEL+ cells, mainly located in the persisting ONL, indicating that photoreceptor cells were dying (Fig. 3A). Although extensive TUNEL staining in the ONL was clearly observed in two Sox9^Δ/Δ retinas with mild phenotypes, this pattern was not consistently present across the full cohort. In the remaining 13 mutant retinas, we observed a modest but noticeable increase in the number of apoptotic cells compared to controls (Fig. 3B; Table S7). Despite a high frequency of zero counts (particularly among controls), the difference between groups reached statistical significance when analyzed using a zero-inflated Poisson model (P = 0.028; n = 5 controls, 13 mutants). These findings suggest that photoreceptor apoptosis following Sox9 deletion may occur acutely and within a narrow temporal window, making it challenging to capture the full degenerative process at a single time point.

Figure 3.

Müller glial cells undergo reactive gliosis in Sox9-deficient retinas

Finally, we decided to study the expression of the glial fibrillary acidic protein (GFAP), whose expression is upregulated in MG cells undergoing reactive gliosis associated with retinal cell injury (Ekström et al., 1988). As previously reported (Fernández-Sánchez et al., 2015), in control mice GFAP immunostaining was mainly limited to the ganglion cell layer (GCL), in the inner margin of the retina. However, in Sox9-deficient retinas with a mild phenotype, GFAP-positive processes reached the INL layer. In mutant retinas with an extreme phenotype GFAP-positive MG processes were distributed throughout the entire thickness of the degenerated retinas. Thus, in the absence of Sox9, MG cells seem to undergo a progressive process of gliosis (Fig. 3C). To support these observations quantitatively, we measured GFAP fluorescence intensity across defined retinal surface areas in control and Sox9^Δ/Δ mice (Fig. 3D; Table S8). This analysis revealed a statistically significant increase in GFAP signal in mutant retinas compared to controls (Mann-Whitney U test, P = 0.024; n = 4 controls, 10 mutants). These results are consistent with a progressive gliotic response following Sox9 deletion and provide further evidence that MG cells become reactive in the absence of Sox9.

Lineage tracing of Sox9-expressing cells in the limbus and the cornea

We next decided to shed light on the role of Sox9 in the adult cornea. Previous studies have documented the expression of Sox9 in the adult mouse limbal and central cornea epithelium (Menzel-Severing et al., 2018; Sartaj et al., 2017). In line with these findings, we also observed the presence of the SOX9 protein in the basal cells of both structures (Fig. 4A and B), which are the sites housing corneal progenitor and stem cells. To assess the stemness potential of these Sox9-positive cells, we conducted TX-inducible genetic lineage tracing by breeding mice harboring an IRES-CreER^T2 cassette inserted into the endogenous Sox9 locus, Sox9^IRES-CreERT2 (Soeda et al., 2010), to either R26R-EYFP (Srinivas et al., 2001) or R26R-LacZ (Soriano, 1999) reporter mice. We induced labeling of Sox9-expressing cells and their progeny by administering TX to adult mice (two months old) and analyzed reporter expression at different DATX (Fig. 4C). No X-gal-or EYFP-positive cells were detected at any of the stages analyzed in the absence of TX (Fig. S1, control panels). Immunofluorescence for EYFP on limbal Sox9^IRES-CreERT2-R26R-EYFP (Sox9-EYFP) sections at 10 DATX unveiled the presence of discrete clones of EYFP-positive cells (Fig. 4D). By 20 DATX, these clones were larger and extended towards the peripheral cornea. At this latter stage, EYFP-positive clones with cells reaching the outer layers of the epithelium were visible at the limbus-cornea border (Fig. 4E, arrowhead). At 20 DATX, we also observed groups of cells spanning the entire apical-basal axis of the central corneal epithelium (Fig. 4F, arrowhead). These results demonstrate that SOX9⁺ cells residing in the basal layer of the limbal and corneal epithelium are progenitors of the terminally differentiated cells of the corneal epithelium.

Figure 4.

We then investigated the dynamics of stem cell progression through whole mount Xgal staining of Sox9^IRES-CreERT2; R26R-LacZ (Sox9-LacZ) eyes. At first, we conducted a short-term study with a single TX injection and analyzed X-gal staining for up to ten days after treatment (Figs. 4G and S1). We observed the first X-gal-positive cells at 3 DATX, and they were scattered throughout the entire limbal and corneal surface (Fig. 4G, arrows). At 5 DATX, most X-gal⁺ cells were located in the limbal area, and by 10 DATX, the number of stained cells increased notably in this area. These observations suggest that Sox9-expressing descendant cells experience a process of proliferation in the limbus, illustrating their transient cell amplification behavior. To gain insights on the long-term cloning capacity and contribution dynamics of Sox9-progenitor cells, we administered tamoxifen for 10 days in the food and performed X-gal staining from 30 DATX to 365 DATX (Figs. 4H and S1). At 30 DATX, LacZ was mostly expressed in the limbus, although lacZ⁺ clones could also be observed throughout the peripheral and central cornea. Overall, these clones were larger than those observed at 10 DATX. Fifteen days later (45 DATX), the limbal staining began to disperse, allowing the observation of discrete clones. Simultaneously, LacZ⁺ stripes emerging from the limbus and extending into the peripheral cornea became apparent (Fig. 4H, 45 DATX, arrow, and Fig. S1). This situation was evident at 60 DATX, when the thickness and number of the stripes had increased. Additionally, starting at this stage, circumferential clones that remained in the limbus could also be appreciated (Fig. 4H, 60 DATX, arrowheads, and Fig. S1). At 90 DATX, three types of clones were observed: circumferential ones in the limbus (Fig. 4H, 90 DATX, arrowheads, and Fig. S1), stripes originating from the limbus, which often reached the central cornea (Fig. 4H, 90 DATX, arrow, and Fig. S1), and stripes without a base in the limbus (Fig. 4H, 90 DATX, asterisk, and Fig. S1). At 365 DATX, the first two types of clones were mainly observed, albeit in reduced numbers and larger sizes. We also observed some stripes without a base in the limbus (Fig. 4H, 365 DATX, asterisk, and Fig. S1). Thus, Sox9 is expressed in a basal limbal stem cell population with the ability to form two types of long-lived cell clones involved in stem cell maintenance and homeostasis.

Sox9 is involved in limbal stem cell differentiation

To gain insight into the function of Sox9 as a limbal stem cell marker, we used the single-cell RNA-sequencing dataset of isolated epithelial cells from the limbus (with marginal conjunctiva and corneal periphery) recently generated by Altshuler and colleagues (Altshuler et al., 2021). In this study, the authors describe the co-existence of two separated stem populations in the limbus, the quiescent “outer” limbus (OLB) one, adjacent to the conjunctiva, and the active “inner” limbus (ILB) one, adjacent to the peripheral cornea. Unbiased clustering of the sc-RNAseq dataset and subsequent assignament of cell identity revealed the existence of 10 cell populations equivalent to those identified by Altshuler and colleagues (Figs. 5A and S2A). Density plots based on gene-weighted kernel density estimation and violin plots of expression levels revealed high expression levels of Sox9 in OLB cells, and progressive weaker expression in the ILB, corneal basal and corneal suprabasal clusters (Figs. 5B and C). We also detected strong expression for Sox9 in a mitotic cluster (Figs. 5B). Next, we inferred the trajectory of limbal and corneal cells by using the partition-based graph abstraction (PAGA) algorithm, and reconstructing the temporal order of differentiating cells with the diffusion pseudotime software, using the OLB cluster as “root” (Wolf et al., 2019; Haghverdi et al., 2016). The PAGA graph revealed a strong connection between the OLB node and a mitotic cell population with high expression of OLB and stem cell markers (Gpha2, Ifitm3, Cd63, and Krt15), which, in turn, was strongly linked to a second mitotic cell cluster, in which the expression of these markers diminished while the expression of corneal markers (Krt12, Ppp1r3c, and Slurp) increased (Figs. 5D and Fig. S2B). Through the PAGA graph, we also observed that the expression levels of Sox9 correlated with the pseudotime graph (Fig. 5E), indicating that Sox9 expression decreases as transiently amplifying progenitors undergo progressive differentiation from limbal to peripheral corneal cells. To validate these results, we decided to closely examine Sox9 expression in the limbus using immunofluorescence. Previous analyses revealed that the outer limbus is approximately 100 μm wide, while the inner limbus is wider, around 240 μm (Altshuler 2021). We observed that in the region corresponding to the OLB, most cells showed strong Sox9 expression. In the area corresponding to the ILB, this immunoreactivity appeared weaker in the basal layer (corresponding to the ILB proper), and no expression was detected in the suprabasal layers (flattened cells; Fig. 5F left). Double immunofluorescence for SOX9 and P63, which is expressed in basal cells of the limbal epithelium, but not by transient amplifying cells covering the corneal surface (Pellegrini et al., 2001) revealed that Sox9 expression was restricted to P63-positive cells (Fig. 5F right). These observations confirm that Sox9 is expressed in a basal cell population within both the OLB and ILB, and that its expression decreases in differentiated transient amplifying cells. Next, we conducted a comparative analysis of gene expression to examine the differences between cells exhibiting high (>= 0.7) and low expression levels (between 0.7 and 0.15) of Sox9. We identified 2816 deregulated genes, 137 upregulated, and 2679 downregulated (adjusted p-value < 0.05; | avg_log2FC| > 0.1; Fig. 5G; Table S9). Among the upregulated genes we found OLB markers including Gpha2, Ifitm3 and Cd63, in addition to stem cell markers such as Krt14 (Fig. 5G; Table S9).

Figure 5.

Finally, we performed gene ontology analysis using the downregulated genes, and we found terms related to stem cell differentiation, such as “mitotic cell phase transition” (Knoblich, 2008), “autophagy” (Chang, 2020), “cell-cell junction” (Ning et al., 2021), “regulation of epithelial cell migration” (Puri et al., 2020), “regulation of cell morphogenesis” and “stem cell differentiation”, as well as gene pathways involved in these processes including “Wnt-signaling pathway” (Yu et al., 2012), “ERBB signaling” (Hassan and Seno, 2022), “cytokine mediated signaling pathway” (Korkaya et al., 2011) and “regulation of ERK1 and ERK2 cascade” (Lavoie et al., 2020; Fig. 5H; Table S10). Altogether, these results suggest that the population of SOX9-negative cells is more differentiated than that of SOX9-positive cells.

Sox9^Δ/Δ progenitor cells lose their clonogenic capacity

Our results suggest that Sox9 is essential for limbal stem cell differentiation. To test this hypothesis, we used the same tamoxifen-inducible conditionally deleted Sox9 gene in adult mouse cells (two months old) that we used in the retina study. However, histological examination of the limbus and cornea from Sox9Δ/Δ mice at any time up to and including 100 DATX did not reveal any discernible phenotypic effects (Fig. 6A). Immunofluorescence staining for SOX9 (Fig. 6B) failed to reveal any differences in the number of SOX9-expressing cells between mutants and controls (Fig. 6C; Table S11; control, 5.4 ± 0.4 cells/100 µm, n = 7; mutant, 5.9 ± 0.7 cells/100 µm, n= 6; p = 0.147, Welch t test). Similar observations were made for corneal and limbal marker genes such as P63 (Figs. 6B and 6C; Table S11; control, 6.1 ± 0.7 cells/100 µm, n = 6; mutant, 6.1 ± 0.7 cells/100 µm, n= 5; p = 0.147, Welch t test).and Pax6 (Fig. S3A). These observations are consistent with a model whereby tamoxifen induces mosaic patterns of cell Sox9-deleting in the ocular surface, but that the Sox9-null cells cannot survive or proliferate as well as their wild-type neighbors, and are hence outcompeted over time, leading to an essentially wild-type cornea. We tested this model by leveraging this mosaicism to investigate the in vivo clonal capacity of both Sox9-positive and Sox9-negative cells. For this, we generated CAGG-CreERTM;Sox9^flox/flox;R26R-LacZ mice (Sox9^Δ/Δ-LacZ), in which Sox9-deleted cells express the LacZ gene, and therefore, are labelled in blue after X-gal staining (Fig. 6D). As controls, we used CAGG-CreERTM;Sox9^flox/+;R26R-LacZ mice (Sox9^Δ/+-LacZ). Adult mice (two months old) were fed with a tamoxifen-supplemented diet for 10 days to induce LacZ expression followed by a chase period of 14 weeks. This time frame aligns with previous observations reporting the appearance of clear X-gal⁺ stripes in CAGG-CreERTM;R26R-LacZ mice (Dorà et al., 2015). In control Sox9^Δ/+-LacZ corneas, derived from 4 mice we could observe numerous discrete X-gal⁺ patches of different sizes spanning over the entire corneal surface. In all cases, we observed a variable number of elongated stripes originating from the limbus, suggesting that they come from limbal stem cells (Figs. 6E and S3B). In contrast, the majority of Sox9^Δ/Δ-LacZ corneas, derived from 5 mice exhibited minimal blue staining, with only occasional, small X-gal⁺ spots observed. Notably, only one mutant cornea revealed the presence of an elongated stripe originating from the limbus (Fig. 6E, right panels; Fig. S3B). A possible explanation for this clone may be that spontaneous ligand-independent activity of Cre-ER fusion may have occurred in a bona fide limbal stem cell, as previously reported (Vooijs et al., 2001; Kemp et al., 2004; Haldar et al., 2009, Dorà et al., 2015). To quantify this difference, we measured the percentage of blue-stainied surface area, using the animal as the biological replicate. For control animals where both eyes were analysed, the average value was used. This analysis confirmed a significant decrease in the mutant corneas compared to the controls (controls: n = 4 mice Sox9^Δ/+-LacZ, 6.74±2.33; mutants: n = 5 mice Sox9^Δ/Δ-LacZ, 0.85 ± 0.86; Mann-Whitney U-test, p = 0.0159; Fig. 6F; Table S12). These results reveal that during normal in vivo homeostasis, epithelial limbal and corneal stem/progenitor cells require Sox9 to maintain their clonogenic capacity, indicating a key role for this gene in the process.

Figure 6.

Discussion

SOX9 is a master regulatory gene involved in the specification and/or development of embryonic tissues belonging to the three germ layers. In recent years, several studies revealed that its expression is maintained in adult ectoderm and endoderm derived cells indicating its role in the homeostasis of mature organs (Jo, 2014). Here, we analyze the in vivo role of the gene in two structures belonging to the adult mouse eye, the retina and cornea, with completely different anatomy and biological functions. We conditionally deleted Sox9 in the adult mouse retina and observed a phenotype characterized by different degrees of retinal degeneration. In the most affected individuals, this degeneration leads to the disappearance of the ONL layer and almost complete absence of cone and rod photoreceptors. Since Sox9 is expressed in MG and RPE cells (Poché et al., 2008), it is challenging to determine the contribution of each cell type to the final mutant phenotype. Previous studies on Sox9-mutant mice have not described any eye phenotype similar to the one we report here. The main reason is probably that these studies focused on the developing retina and are therefore not useful for understanding how the absence of Sox9 in MG and RPE cells in the adult eye leads to the retinal degeneration observed in our Sox9-deficient adult mutants. In this context, we can obtain better insights by examining the physiological response to RPE and MG cell loss after Cre-mediated activation of diphtheria toxin A (DTA) in these cell types. In the first case, the major retinal abnormality was the formation of regions of photoreceptor rosetting with very mild degeneration of the photoreceptors (Longbottom et al., 2009). In mice with a DTA-induced ablation of MG cells, a wave of cell apoptosis was observed in the ONL but not in the other retinal layers (Shen et al., 2012). In our mutant mice, we did not observe photoreceptor rosetting in the ONL, but a massive apoptosis in this retinal layer. In addition, we observed that some regions within the INL layer were devoid of MG cells. Thus, we favor the hypothesis that continuous expression of Sox9 in MG cells is necessary for normal function and/or survival of this cell type. In the absence of Sox9, MG cells undergo massive retinal gliosis and they cannot provide structural, metabolic, and functional support to the other retinal cells, leading to a massive loss of photoreceptors followed by the depletion of other retinal cell types. In agreement with this, Sox9 expression is upregulated in MG cells after retinal light damage in rats (Wang et al., 2013). However, the retinal phenotype we report here is much more severe than the one observed in both DTA-induced transgenic lines. This could be due to the different efficiency of Cre-induced recombination, but also to the fact that the combined deletion of Sox9 in both MG and RPE may cause a more severe phenotype than the individual cell type-specific mutants.

An important consideration in our model is the potential contribution of non-cell autonomous mechanisms to photoreceptor degeneration. Sox9 is expressed in both MG and RPE cells, and both cell types are known to support photoreceptor viability (Poché et al., 2008; Masuda et al., 2014). Notably, Sox9 and Otx2 cooperate to regulate visual cycle gene expression in the RPE (Masuda et al., 2014), and loss of Otx2 specifically in the adult RPE leads to secondary photoreceptor degeneration through non-cell autonomous mechanisms (Housset et al., 2013). However, RPE-specific deletion of Sox9 does not induce retinal degeneration and in fact results in Otx2 upregulation (Masuda et al., 2014; Goto et al., 2018; Cohen-Tayar et al., 2018), suggesting that Sox9 is not an upstream regulator of Otx2 in this context. Further investigation into the molecular and cellular interactions between MG, RPE, and photoreceptors may help to clarify the indirect pathways contributing to degeneration in the absence of Sox9.

To our knowledge, no retinal abnormality has been described in patients with a mutation in SOX9 (see Hejtmancik and Daiger, 2020). This is probably because CD patients are mostly heterozygous for SOX9 and show early postnatal lethality, so a retinal degeneration phenotype in adulthood was either overlooked or not addressed. However, SOX9/Sox9 has a complex regulatory landscape (Bagheri-Fam et al., 2006; Despang et al., 2019), and mutations in its regulatory region have been associated with tissue-specific human phenotypes (Wunderle et al., 1998; Benko et al., 2009; Kim et al., 2015; Kurth et al., 2009). In this context, transgenic mice carrying a human enhancer located around 500 kb upstream of SOX9 revealed enhancer activity in the eye (Sreenivasan et al., 2017). Thus, screening of the SOX9 regulatory region may provide new molecular mechanisms underlying idiopathic cases of retinal degeneration.

Several expression studies have proposed Sox9 as a marker of epithelial limbal and corneal stem/progenitor cells, with further support from in vitro experiments validating this finding (Sartaj et al., 2017; Parfitt et al., 2015; Menzel-Severing et al., 2018). Through in vivo cell lineage tracing and gene targeting strategies in mice, combined with the analysis of sc-RNAseq data, we shed light on the nature, differentiation dynamics and function of these stem/progenitor Sox9-positive cells. Trajectory analysis of limbal and corneal scRNAseq data revealed that Sox9 is highly expressed in a cell population characterized by the expression of stem cell markers (Gpha2, Ifitm3, Cd63, and Krt14), and that its expression gradually decreases as these cells enter mitosis and differentiate into suprabasal corneal cells. Consistent with this, we observed a substantial proliferation of Sox9-expressing descendant cells, leading to a significant increase in their numbers within the first 10 days after the initiation of the TX treatment. Subsequently, these cells originate small clones with a rapid decrease and finally they give rise to few, large, radially oriented clones. The fact that Sox9 was expressed exclusively in epithelial basal cells, while Sox9-expressing descendant cells were observed throughout the entire corneal epithelium indicates that many Sox9-expressing descendant cells experience a decline in Sox9 expression, initiating their differentiation into cells that proliferate, migrate, and differentiate into suprabasal corneal cells, thus contributing to the maintenance of corneal homeostasis.

The persistence of some small clones (including a radial stripe) of LacZ⁺ Sox9^Δ/Δ cells in the limbal and corneal epithelia of CAGG-CreERTM;Sox9^flox/flox; R26R-LacZ mice nevertheless shows that Sox9 is not absolutely required for the contribution of cells to the limbal and ocular epithelium, but that the lack of Sox9 significantly impairs the ability of mutant cells to compete with wild-type cells. This competition would explain why, in contrast to the whole mount LacZ analysis, Sox9-null cells were not detected in tissue sections of these eyes. A recent preprint (Rice et al, 2024) describes a study that confirms the stem-clonogenic potential of Sox9-positive limbal epithelial cells. That study also reports that conditional deletion of Sox9 leads to abnormal corneal epithelial differentiation and squamous metaplasia in the central cornea. This phenotype was not observed in our study but is explicable because of the mosaic nature of our model, which may allow for rescue of the normal Wnt and Notch signalling by wild-type cells, preventing transdifferentiation (Menzel-Severing, 2018).

We also observed long-lived circumferential clones that never exited the limbus. These latter clones are similar to those produced by OLB cells, as recently shown in lineage-tracing experiments, which predominantly self-renew and minimally contribute, if at all, to the formation of lineages maintaining the corneal epithelium during homeostasis. However, in response to corneal injury, OLB proliferate and play an essential role in wound healing (Altshuler et al., 2021; Farrelly et al., 2021). Consistent with this, we saw that Sox9 is highly expressed in the OLB cluster. Furthermore, previous studies utilizing an H2B-GFP/K5tTA mouse model have shown that Sox9 is expressed in a corneal quiescent/slow-cycling cell population (Parfitt et al., 2015), and wound healing assay using a human corneal organ culture revealed an increase in the number of SOX9⁺ cells in both activated limbal and re-grown corneal epithelial cells (Menzel-Severing et al., 2018). We also observed that the number and extension of the limbal circumferential clones diminished in Sox9^Δ/Δ-LacZ corneas, indicating that Sox9 may also control the proliferation of OLB stem cells.

This study highlights how a single transcription factor can regulate gene expression across neighboring cell types, driving distinct genetic programs and biological functions. In the retina, MG cells play a vital role in providing structural and metabolic support. Here, we demonstrate that Sox9 is essential for the maintenance and survival of MG cells. In this context, previous studies have shown that Notch signaling, which is critical for the maintenance of retinal progenitor cells and the specification of MG cells (Mills and Goldman, 2017), regulates Sox9 expression. Moreover, Sox9 has been shown to regulate the expression of type I collagen (COL1; Wang, 2013) and glial fibrillary acidic protein (GFAP), a key component of the intermediate filaments that form part of the MG cytoskeleton (Kinouchi, 2003). Thus, within the retina, Sox9 appears to fulfill multiple roles, including the regulation of structural protein expression and the promotion of cell survival through anti-apoptotic activity. This situation closely parallels that shown in postnatal and adult Sertoli cells, which die after Sox9 ablation, leading to the complete degeneration of the testicular germinative epithelium. In these cells, Sox9 likewise acts as an anti-apoptotic factor and regulates the expression of extracellular matrix components (Barrionuevo et al, 2009; Barrionuevo et al, 2016). These facts suggest a role for Sox9 in the maintenance of multiple adult tissues. In contrast, in the cornea, Sox9 appears to be involved in the regulation of proliferation and differentiation of corneal epithelial stem/progenitor cells. However, the regulatory mechanisms and downstream targets of Sox9 in this context remain poorly understood. Here, we show that high Sox9 expression is associated with the presence of stem cell markers such as Krt15, P63, and Gpha2, whereas its levels decline as stem cells undergo differentiation, consistent with recent findings (Rice et al, 2024). Furthermore, in vitro studies using limbal epithelial stem/progenitor cells have demonstrated that activation of the BMP, Notch, and Shh pathways leads to upregulation of Sox9, while Wnt signaling antagonizes its activity (Menzel-Severing, 2018). Thus, in the limbus, unlike its role in the retina, Sox9 appears to engage with key molecular pathways that govern stem cell maintenance and differentiation. Future studies will help to elucidate the molecular networks in which Sox9 is involved in both contexts.

In summary, through functional analyses in mice, we have demonstrated multiple roles for Sox9 in the adult eye, acting as an essential maintenance factor preventing retinal degeneration on the one hand, and promoting the differentiation of limbal cells in the cornea on the other.

Material and methods

Mouse lines and crosses

To obtain Sox9 null mutant mice, we crossed Sox9^flox/flox(B6.129S7-Sox9^tm2Crm/J) (Akiyama et al., 2002) to CAGG-CreER^TM(B6.Cg-Tg^{(CAG-cre/Esr1∗)5Amc}) (Hayashi and McMahon, 2002) mice, and the resulting double heterozygous offspring was backcrossed to Sox9^flox/flox mice. Both mouse strains were sourced from The Jackson Laboratory. Mutation occurs after treatment with tamoxifen, as described below. Cre-negative Sox9^flox/flox, tamoxifen-treated mice served as controls. For corneal Sox9 mosaic analyses we crossed CAGG-CreER^TM; Sox9^flox/flox mice to R26R-LacZ (B6.129S4-Gt(ROSA)26Sor^tm1Sor/J) (Soriano, 1999), generating CAGG-CreER^TM; Sox9^flox/flox; R26R-LacZ mice. As control we used CAGG-CreER^TM; Sox9^flox/+; R26R-LacZ mice. For genetic lineage tracing assays, we used Sox9^IRES-CreERT2 (B6.129S7-Sox9^{tm1(cre/ERT2)Haak}) (Soeda et al., 2010) mice obtained from the RIKEN BioResource Research Center (Tsukuba, Japan). These mice were crossed with either R26R-LacZ (B6.129S4-Gt(ROSA)26Sor^tm1Sor/J) (Soriano, 1999) or R26R-EYFP (B6.129X1Gt(ROSA)26Sor^{tm1(EYFP)Cos/J}) (Srinivas et al., 2001) reporter mice, resulting in double heterozygotes. Both of them were provided by The Jackson Laboratory (Bar Harbor, ME). All the experiments started with adult mice (2 months old). Mice were housed under Specific Pathogen-Free (SPF) conditions in the animal facilities of the Center for Biomedical Research (University of Granada, Granada, Spain). The animals had ad libitum access to food and water and were kept in groups. The occupancy density of the cages (microventilated) was in accordance with legal requirements. The cages were kept at a temperature of 22 +/- 2°C, a humidity of 55 +/- 10%, and a 12-hour/12-hour dark light cycle. Mice were provided with activity elements in the form of nest building material and hiding places. The animal procedures and housing conditions were approved by the University of Granada Ethics Committee for Animal Experimentation and the Consejería de Agricultura, Ganadería, Pesca y Desarrollo Sostenible of the Andalussian government, Junta de Andalucía (reference 12/12/2016/177).

Tamoxifen supplementation

For conditional deletion of Sox9, CAGG-CreERTM;Sox9^flox/floxmice were fed a standard diet (Harlan, 2914) supplemented with tamoxifen (TX; Sigma-Aldrich, St. Louis, MO, C8267) at a concentration of 40 mg TX/100 g diet for 10 days. For short-term tracing of Sox9-descendant cells (less than 10 days), TX (Sigma-Aldrich, St. Louis, MO, T5648) was dissolved in corn oil at a concentration of 30 mg/ml and administered at a dose of 0.2 mg per gram of body weight through intraperitoneal injection to Sox9^IRES-CreERT2 mice. For long-term tracing of Sox9-positive cells (more than 10 days), Sox9^IRES-CreERT2 mice were administered TX as detailed above for the conditional deletion analyses.

Estimation of the percentage of tamoxifen-induced, Cre-mediated recombination and GFAP quantification

To estimate the percentage of cells undergoing Cre-mediated recombination in TX-treated CAGG-CreERTM;Sox9^flox/floxmice, the total number of SOX9+ cells was divided by the total number of SOX8+ cells in a 20x microphotograph of a retinal section stained for SOX9 and SOX8.

In parallel, to quantify GFAP expression as a measure of MG reactivity, we analyzed GFAP immunofluorescence intensity across defined retinal surface areas. Given the cytoplasmic distribution of GFAP within glial processes, direct cell counting was not feasible. Instead, fluorescence intensity was measured using ImageJ, within full-thickness retinal regions in 20x microphotographs of a retinal sections stained for GFAP. The total GFAP signal was normalized to the measured area for each section.

Histology and immunofluorescence

Eyes were dissected out and the crystalline lenses removed in order to facilitate histological sectioning. The eyes were fixed in 4% paraformaldehyde (PFA), dehydrated in increasing EtOH/saline solutions (EtOH + 0.9% Nacl), embedded in paraffin, sectioned, and either stained with hematoxylin-eosin or processed for protein immunodetection. For simple and double immunofluorescence, sections were incubated overnight with the primary antibodies, washed, incubated with the appropriate conjugated secondary antibodies for 1 h at room temperature, and counterstained with 40,6-diamidino-2-phenylindole (DAPI). Several sections of eyes from control and mutant mice were consistently mounted on the same slide and processed together. Parallel negative controls were performed in which the primary antibody was omitted. The following antibodies were used: anti-SOX8 (1/500), kindly provided by Dr. Wegner at the Universität Erlangen-Nurnberg (Germany); anti-OPN1SW (1/200, Santa Cruz, sc-14363); anti-OPN1LW (1/500, Chemicon, AB5405); anti-RHO (1/500, Sigma, O4886); anti-BRN3α (1/100, Chemicon, MAB1585); anti-PAX6 (1/40, kindly provided by the Developmental Studies Hybridoma Bank at the University of Iowa), anti-AP2α (1/50, Abcam, ab108311), anti-SOX9 (1/400, Millipore, AB5535), anti-GFP (1/100, Novus Biologicals, NB600-308), anti-p63 (1/500, Master Diagnostica MAD-000479QD), and anti-S100 (prediluted 1/1, Master Diagnostica MAD-000592Q).

The two eyes of each mouse were analyzed in all experiments, and there were no notable differences between them (left-right) either histologically or in the expression pattern of the markers studied. Similarly, we found no differences between the eyes of control mice regardless of whether they were treated with TX or not.

Analysis of retinal and corneal cell death

Retinal and corneal cell death was assessed using the TUNEL Fluorescent In Situ Cell Death Detection Kit (Roche, Mannheim, Germany) according to the manufacturer’s instructions.

Retinal flat whole-mount immunofluorescence

Eyes were fixed in 4% PFA and retinas were dissected in TBS containing 0.05% Tween-20 and 0.1% Triton X-100 for 30 minutes at room temperature (RT). Retinas were then washed in TBS with 0.05% Tween-20 (TBS-T) and exposed to primary antibodies diluted PBS with 4% of BSA overnight at 4°C. After washing with TBS-T, the retinas were incubated with secondary antibodies diluted in PBS with 4% BSA for 3 hours at RT. Finally, the retinas were mounted flat for confocal microscopy.

Whole-mount X-gal staining

The eyes were dissected and cut transversely to facilitate solution penetration and whole-mount β-galactosidase histochemical reaction using X-gal as a substrate (X-gal staining) was performed. Whole-mount X-gal staining was performed overnight as described by Hogan et al. (1995). After staining, eyes were fixed in 4% paraformaldehyde.

Estimation of the percentage of LacZ-positive corneal surface area

The surface area was calculated using the FIJI (V2.1.0) program based on ImageJ. The entire cornea surface was outlined, and the areas of the LacZ stripes were manually selected and marked. Then, the percentage of the blue-stained area relative to the total corneal area was calculated. An automatic threshold could not be applied due to variations in brightness across the cornea pictures.

Imaging

Histological and immunofluorescence images were photographed using a DS-Fi1c camera installed on a Nikon Eclipse Ti microscope (Japan). Retinal flat mount images were obtained using a high-speed spectral confocal microscope Nikon A1 ASHS-1. Subsequently, GIMP2 was used to adjust the color levels and the contrast of the immunofluorescence images. Whole-mount X-gal-stained eyes were photographed using a DP70 digital camera mounted on an Olympus SZX12 stereomicroscope (Japan).

Single-cell RNA-sequencing dataset analyses

We used the single-cell RNA-sequencing datasets of isolated epithelial cells from the limbus (with marginal conjunctiva and corneal periphery) generated by Altshuler and colleagues (Altshuler et al., 2021; Gene Expression Omnibus GSE167992). Bioinformatic analyses were performed with Seurat (version 5.0.3) (Hao et al., 2021) following the author guidelines (https://satijalab.org/seurat/). Cells with unusually high and low numbers of unique feature counts or high mitochondrial counts were filtered out. The data were normalized and subsequently scaled. RunPCA was used for principal component analysis, and cells were clustered at a resolution of 0.6. Uniform manifold approximation and projection was used for nonlinear dimensional reduction. Cluster-specific markers were selected with a minimum average log fold change threshold of 0.25 among genes that were expressed in a minimum of 25% of the cells. Cluster identity was assigned using the following markers: Conjunctiva: Krt17, Krt4, Krt19, Krt6a, Krt13, Krt8 (and lack of Krt12, Slurp1, Ppp1r3c); cornea: Krt12, Ppp1r3c, Slurp; outer limbus: Cd63, Ifitm3, Gpha2; inner limbus: Atf3, Socs3, Mt1; basal cells: Itgb1, Itgb4, Ccnd1; supra-basal cells: Cldn4, Cdkn1a, Dsg1a; and mitosis: Mki67, Top2a, Ccna2. For differential expression of cells expressing high and low levels of Sox9, we selected cells with high Sox9-expression levels (>= 0.7) and cells with low Sox9-expression levels (between 0.7 and 0.15) with the WhichCells and the SetIdent functions of the Seurat package. Differential expression analysis was performed with the FindMarkers function (min.pct = 0.20, logfc.threshold = 0.01). For trajectory analysis we used the Python Scanpy suite (Haghverdi et al., 2016; Wolf et al., 2018) following the general PAGA trajectory inference workflow (https://scanpy.readthedocs.io/en/ stable/tutorials/trajectories/paga-paul15.html) with a Leiden resolution of 0.6. Cluster identity was assigned as described above, and the conjunctiva cluster was removed. For pseudotime analysis (Wolf et al., 2019), the OLB cluster was chosen as the root. Gene Ontology analyses were done with the enrichGO and compareCluster functions of the clusterProfiler R package (Yu et al., 2012).

Statistical analysis

We analyzed datasets derived from distinct experimental conditions using appropriate statistical tests based on preliminary assessments of normality and variance equality. When these conditions were not met, non-parametric approaches were favored for their robustness to data distribution assumptions. Specifically, we employed the Kruskal-Wallis rank-sum test to identify overall differences among group distributions followed by post-hoc Dunn’s multiple comparison tests with Bonferroni adjustment to pinpoint specific pairwise comparisons that contributed to these differences. When no significant departures from normality or equality of variances were detected, parametric methods were deemed suitable. So, we conducted a standard Student t-test to evaluate the hypothesis of equal means between groups. Complete information on these statistical analyses are provided in Supplementary Tables S2-S8, S11 and S12.

Acknowledgements

This work has been funded by the following institutions: Programa Operativo FEDER Andalucía 2014-2020, Consejería de Economía, Conocimiento, Empresas y Universidad, Junta de Andalucía (Ref. A.BIO.106.UGR18) and MICIU/AEI /10.13039/501100011033 and FEDER, UE, Ref PID2022-139302NB-I00 to RJ and FJB; Ministerio de Ciencia e Innovación, Instituto de Salud Carlos III, European Regional Development Fund through the program “Una manera de hacer Europa” (Ref. PI23/00335) to MA.

Additional information

Authors contributions

FJB, FDC, and RJ conceived the project; FJB, FDC, MB, and RJ designed the experiments; FJB, FDC, MB, RJ, AH, VL-S, ML, MAC-B, PL, ACC, YR-C, MA, and JMC performed laboratory experiments; FJB and MB carried out bioinformatic and statistical analyses; RJ, FJB, and MA collected the financial support; FJB, FDC and RJ wrote the first draft of the manuscript and all authors contributed to the final version.

Additional files

Supplementary tables S1-S12

Supplementary figures