KIT ligand produced by limbal niche cells under control of SOX10 maintains limbal epithelial stem cell survival by activating the KIT/AKT signalling pathway

Abstract Homeostasis and function of limbal epithelial stem cells (LESCs) rely on the limbal niche, which, if dysfunctional, leads to limbal epithelial stem cell deficiency (LSCD) and impaired vision. Hence, recovery of niche function is a principal therapeutic goal in LSCD, but the molecular mechanisms of limbal niche homeostasis are still largely unknown. Here, we report that the neural crest transcription factor SOX10, which is expressed in neural crest‐derived limbal niche cells (LNCs), is required for LNCs to promote survival of LESCs both in vivo and in vitro. In fact, using mice with a Sox10 mutation and in vitro coculture experiments, we show that SOX10 in LNCs stimulates the production of KIT ligand (KITL), which in turn activates in LESCs the KIT‐AKT signalling pathway that protects the cells against activated CASPASE 3‐associated cell death. These results suggest that SOX10 and the KITL/KIT‐AKT pathway play key roles in limbal niche homeostasis and LESC survival. These findings provide molecular insights into limbal niche function and may point to rational approaches for therapeutic interventions in LSCD.


SU et al.
LESCs are limbal niche cells (LNCs). [13][14][15][16][17][18][19][20][21] Previous results have shown that LNCs support self-renewal of LESCs, 21 and LNCs transplants prevent LSCD in an alkali burn rabbit model. 22 They likely do this by secreting paracrine factors including KITL, 22 but the detailed mechanisms of KITL regulation in LNCs and how KIT, KITL and its downstream signal pathways are involved are not clearly understood. Here, we use both in vivo and in vitro approaches to provide insights into the regulation of KITL in LNCs and its action in LESCs.
Our studies were guided by the notion that LNCs are developmentally derived from neural crest cells (NCCs), 23,24 unlike LESCs, which are thought to be derived from the ocular ectoderm. 25 In fact, besides LNCs, the neural crest contributes a number of cell types to the development and function of the eye. NCCs give rise, for instance, to the corneal endothelium, the corneal stroma and the stroma of the ciliary body and the iris. Not surprisingly, NCC deficiencies lead to corneal diseases, such as anterior segment dysgenesis (ASD), 26,27 as seen with Axenfeld-Rieger syndrome, 28,29 Peters anomaly, 30 Aniridia 31 and Nail Patella syndrome. 32 The major clinicopathological characteristics of ASD include corneal epithelial dystrophy, disorganized corneal stroma, sclerocornea, corneal opacities and corneal vascularization. 27 Hence, NCCs are key cellular components of corneal development and homeostasis.
A major regulator of the NCC development is the transcription factor SOX10, an SRY-box containing HMG DNA-binding protein. 33,34 In humans, mutations in SOX10 are associated with several developmental NCC defects such as Waardenburg syndrome type IV (also known as Waardenburg-Shah syndrome) and 35 PCWH (peripheral demyelinating neuropathy, central dysmyelinating leukodystrophy, Waardenburg syndrome and Hirschsprung disease). 36 Symptoms include, but are not limited to, deafness, skin pigmentary disorders and neurological defects. Interestingly, in vivo track systems based on SOX10-cre-driven YFP fluorescence have shown that SOX10-positive NCCs are the major contributors to the corneal stroma. 24 Given the above importance of SOX10 in NCCs and the fact that LNCs are essential for supporting self-renewal of LESCs, 21 we here focused on the role of SOX10 in LNCs for LESC maintenance both in vivo and in vitro. Using mice heterozygous for a Sox10 mutation and isolated LNCs as well as their conditioned mediums, we show that a gene dose reduction of Sox10 significantly impairs the ability of LNCs to support LESC maintenance and that SOX10 acts through KITL to activate the KIT-AKT signalling cascade in LESCs. Hence, these findings suggest that the SOX10-KITL/KIT axis is a major component of the supportive function of LNCs for LESCs.

| Animals
Sox10 LacZ/+ (hereafter called Sox10/+) mice, in which one Sox10 allele is rendered non-functional by insertion of a LacZ gene, were originally obtained from Dr Michael Wegner and then transferred to our laboratory from the laboratory of Dr William J. Pavan (NIH).
Genotyping of Sox10/+ mice was carried out as described. 37 All animals were handled according to ethical standards of the Institutional Animal Care and Use Committee of the Wenzhou Medical University (permit number WZMCOPT-090316).

| Isolation and culture of both limbal niche cell and limbal epithelial stem cells
LESCs were isolated from 4-week-old mice by modifying a previously described method. 38 Briefly, eyeballs of mice were washed with DMEM/F12 medium (Sigma-Aldrich) containing 500 IU/mL penicillin (Beyotime Biotechnology) and 500 µg/mL streptomycin (Beyotime Biotechnology). Iris and excessive sclera were carefully removed, and limbal rings were isolated and incubated at 4°C for 16 hours with 1.2 IU/mL dispase II (Sigma-Aldrich) dissolved in Hanks' balanced salt solution (Sigma-Aldrich). Epithelial sheets were then carefully removed under a dissecting microscope, and single cell suspensions were prepared by treatment with 0.25% trypsin-EDTA at 37°C for 5 minutes. Cells were collected by centrifugation at 400 g for 5 minutes and cultured in DMEM/F12 supplemented with 10% FBS (Invitrogen Corporation), 5 ng/mL recombinant mouse EGF (Sigma-Aldrich), 1% ITS liquid media supplement (Sigma-Aldrich), 0.5 µg/mL hydrocortisone (Solarbio), 30 ng/mL cholera toxin (Sigma-Aldrich), 100 IU/mL penicillin and 100 µg/mL streptomycin.
LNCs were also isolated from 4-week-old mice as previously described, 14 except for slight modification as follows: briefly, after limbal rings were isolated and epithelial sheets removed as mentioned above, the remaining limbal rings were cut into 1 mm 3 pieces and incubated overnight at 4°C with DMEM/F12 medium containing 1 mg/mL collagenase A (Sigma-Aldrich). After centrifugation, the pellets were resuspended in E8 medium (Life Technologies) and seeded onto 6-well plates. Two days later, cell debris was carefully removed by aspirating the medium. Adherent LNCs usually grow out to form sphere-shaped colonies 7 days after seeding. Both LNCs and LESCs were characterized by staining for differential expression of marker genes ( Figure S1).

| Colony formation assay
For preparation of conditioned medium, supernatants were collected from LNCs cultured with DMEM/F12 supplemented with 1% FBS for 3 days and then diluted at the ratio of 1:1 with DMEM/ F12 medium containing 1% FBS. Similar procedures were used to prepare conditioned medium derived from LNCs transfected with si-Sox10-1, si-Sox10-2, si-C (non-specific siRNA used as a negative control) or mock-transfected (hereafter called si-Sox10-1-CM and si-Sox10-2-CM, si-C-CMor mock-CM, respectively). LESCs cultured with DMEM/F12 medium supplemented with 1% FBS served as controls.
For colony formation assays, 500 LESCs per well were seeded on the lower chambers of 24-well cell culture inserts and cocultured with LNCs in the upper chambers. Alternatively, 500 LESCs were cultured on 24-well cell culture plates and exposed to conditioned medium as mentioned above. Seven days after cell planting, a Giemsa Staining Kit (Sangon Biotech) was used to visualize colonies of LESCs according to the manufacturer's instructions. Efficiency of colony formation was determined by counting the numbers of colonies with radius greater than 0.2 mm and calculated as a percentage of the number of originally seeded cells.

| Structural analysis of cornea
For structural analyses of cornea, both WT and Sox10/+ eyes were dissected, paraffin embedded, sectioned, and HE stained as described. 39 Areas of the central cornea were photographed, and the thickness of the corneal epithelium was measured.

| Real-time polymerase chain reaction (RT-PCR)
RT-PCR was performed as previously described. 40 Briefly, TRIzol reagent (Life Technologies) was used for total RNA extraction.
Total RNA was then reverse-transcribed into cDNA using a reverse transcriptase kit (Agilent), followed by RT-PCR analysis using SYBR Green technology (Applied Biosystems). All data were normalized with respect to GAPDH expression levels. Sequences of primers used in this study are shown in Table S1.

| siRNA transfection
To knock-down gene expression in LNCs, siRNAs corresponding to sequences of the mouse Sox10 gene were purchased from GenePharma. si-C served as a negative control. Their sequences are as previously described. 40

| Immunostaining
For immunostaining, cells were fixed with 4% paraformaldehyde (PFA) for 20 minutes at room temperature and permeabilized with 0.4% Triton X-100 for 10 minutes, followed by treatment of 1% BSA for 1 hour. Cells were then incubated with specific primary antibodies at 4°C overnight. Staining was revealed by either FITC or Cy3-conjugated secondary antibodies. Photographs were taken using a Zeiss fluorescence microscope.
For immunocytochemical analysis, mice eyeballs were isolated and cryoprotected for preparation of frozen sections. Frozen sections were then immunostained as described above. The primary antibodies used in this study are shown in Table S2.

| Western blot assay
Western blotting was carried out as described. 40 Antibodies used in this study are shown in Table S3.

| Enzyme-linked immunosorbent assay (ELISA) and TUNEL assay
To detect the concentration of SCF in conditioned medium derived from LNCs, ELISA assays were carried out using SCF ELISA Kit (Invitrogen Corporation) according to the manufacturer's instructions.
To reveal apoptotic cells, terminal deoxynucleotidyl transferase dUTP nick end labelling (TUNEL) assay was carried out using In Situ Cell Death Detection Kit (Roche) according to the manufacturer's instructions.

| Statistical analysis
Each experiment was repeated at least three times, and quantitative data are presented as mean ± SD. Statistical significance (P-value) between experimental and control groups was assessed with Student's t test. P < .05 was considered statistically significant.

| SOX10 controls the ability of LNCs to serve as a niche for LESC survival in vitro
To experimentally manipulate and examine the role of LNCs and their ability to support the maintenance of LESCs, both LNCs and

| SOX10 in LNCs is required for survival of LESCs in vivo
Because embryonic or perinatal lethality in homozygous Sox10-null mice precludes an analysis of long-term corneal maintenance, we used Sox10/+ mice to determine whether SOX10 controls LNCs to impact LESCs survival in vivo. Compared with their wild-type (WT) littermates, 2-month-old Sox10/+ mice did not show any obvious abnormalities in the cornea, but at 12 months, they displayed severe corneal dystrophy marked by reduced thickness of corneal epitheliums and disorganized corneal stroma (Figure 2A,B). These results indicate that heterozygosity for a Sox10 loss-of-function allele leads to progressive corneal epithelial dystrophy in mice. Since LESCs, which are self-renewing, are  Figure 1C,D), it hence appears that down-regulation of SOX10 contributes to a disruption of the niche for LESCs, which eventually leads to the reduction in the number of LESCs and to corneal epithelial dystrophy.  22,38,[41][42][43][44][45][46] In addition, as shown above, LNCs act in a paracrine fashion to support LESCs in vitro, and so, we decided to analyse whether SOX10 acts through growth factors to influence the survival of LESCs in LNCs. To identify which factors might be involved, we compared the expression of a series of growth factor genes in si-Sox10 transfected LNCs vs si-C or mock-transfected LNCs. As shown in Figure 3A, knock-down of Sox10 led to downregulation of Egf, Fgf7, Igf1, Kitl, and Wnt5a. Among these genes, Kitl (also known as Stem cell factor, Scf or mast cell growth factor, Mgf) was of particular interest because it plays a pivotal role in LNCs to promote survival of LESC transplants. 22 Hence, we examined

| SOX10 works through activation of the KITL/KIT-AKT signalling cascade to promote survival of LESCs
To determine whether SOX10 indeed acts through KITL produced by LNCs to support survival of LESCs, we first used recombinant KITL to treat cultures of LESCs. As mentioned, LESCs colony formation was diminished, and apoptosis enhanced, when the cells were exposed to si-Sox10-1-CM or si-Sox10-2-CM. Addition of recombinant KITL to such conditioned medium, however, significantly enhanced LESC colony formation and reduced cellular apoptosis ( Figure 4A,D,E,F). Furthermore, Western blotting analysis revealed that the phosphorylation levels of KIT in LESCs were significantly increased provided the cells were exposed to a source of KITL either from LNC-CM or from addition of recombinant KITL to si-Sox10-CM ( Figure 4B,C). These results suggest that SOX10 works through KITL produced by LNCs to support LESCs survival. One of the pathways activated by KITL/KIT signalling is the AKT pathway. 47 To address the question of whether SCF/KIT signalling acts through activation of AKT, Western blotting was carried out to detect the phosphorylation levels of AKT in LESCs. As shown in Figure 5A,B, compared with control cells, the AKT phosphorylation in LESCs was significantly increased when the cells were cultured with LNC-CM, but not when they were cultured with si-Sox10-CM; addition of KITL to the latter conditioned medium, however, increased AKT phosphorylation again. Apparently, the activation pattern of AKT in LESCs parallels that of KIT ( Figure 4B,C), indicating that SOX10-stimulated expression of KITL in LNCs activates the KIT-AKT signalling cascade in LESCs.
As AKT signalling has been shown to control the activation of CASPASE 3 to impact cell apoptosis, 48 we examined whether AKT signalling would protect LESC from apoptosis by inhibiting CASPASE 3 cleavage. Using Western blots, we found, indeed, that the pres-

| D ISCUSS I ON
The interest in LESCs is based on their importance for the development and physiology of the cornea whose structural integrity and translucency has to be maintained for life. In fact, loss or dysfunctions of LESCs can lead to disorganized corneal stroma and corneal epithelial dystrophy and are associated with a number of eye diseases. The replacement of these cells by transplantation, though feasible, has unfortunately not uniformly led to clinical improvements in affected patients. 3 One of the reasons for the therapeutic failures may be the loss or instability of the transplanted cells. It is therefore imperative to delineate conditions under which the therapeutic success may be enhanced.
The development and maintenance of LESCs depends on interactions with neural crest-derived LNCs, which, along with other cell types, form a stem cell niche providing trophic support for LESCs.
Here, we used both genetic in vivo as well as in vitro approaches to examine molecular parameters of LNC-mediated LESC survival. We find that the transcription factor SOX10, known to be instrumental in neural crest cell development, is expressed in LNCs, regulates their development and function 33 and cell-autonomously stimulates a number of growth factor genes, all of which potentially important for LESC survival. Among them we find KITL, known to be critical for the development of a number of neural crest derivatives, to be particularly well stimulated by SOX10. In fact, we find that upon interaction with its single receptor KIT present on LESCs, KITL leads to a paracrine activation of the AKT signalling pathway that in turn prevents apoptosis of LESCs. These results extend previous observations supporting a role for LNCs and KITL signalling in maintaining LESCs. 21,22 The above notion is based on multiple observations. First, both in vivo and in vitro, SOX10 is expressed in LNCs but not LESCs. Second, KITL expression is reduced in LNCs of mice heterozygous for a SOX10 null allele and in cultured LNCs derived from them. Third, LESCs express KIT, the sole receptor for KITL. Fourth, exposure of LESCs to LNCs or LNC-conditioned medium leads to KIT phosphorylation and to activation of the AKT signalling pathway and prevents LESC apoptosis. Fifth, application of conditioned medium from LNCs, whose SOX10 expression has been reduced by application of corresponding siRNAs, results in reduced KIT signalling in LESCs and increased LESC apoptosis, and these effects can be reversed by re-addition of recombinant KITL. Further support of these results comes from the application of inhibitors of KIT and AKT.
That SOX10 is particularly important for LNC development and function is supported by its wider role in the physiology of neural crest cells. 33,49 In fact, loss of SOX10 leads to severe defects in neural crest derivatives. 34 Under appropriate conditions, SOX10 may even reprogramme cells to express neural crest markers 50 and be associated with malignant transformation, for instance of neural crest-derived melanocytes. 51 Nevertheless, it has to be kept in mind that SOX10 is not the only factor critical for LNCs' support of LESCs as there are also other genes, such as Pax3 and Tfap2, that are important in NCC development. TFAP2, for instance, has been shown to directly regulate the expression of Sox10, 52 and PAX3 could synergistically cooperate with SOX10 to regulate fate determination of NCCs. 53 Thus, it will be crucial to extend our current studies to analyse other genes for synergistic interactions with Sox10 in LNCs and their role in providing trophic support for LESCs.
Much as SOX10 is important in development, so is KIT signalling, and this not only in NCCs 34 but also in a variety of other cell types. 47 Hence, the importance of KIT signalling in LESCs is not further surprising. KIT signalling is required for glycogen metabolism of the corneal epithelium, 54 and loss of either KITL or KIT significantly impairs corneal wound healing. 55 Interestingly, KIT is not only expressed in LESCs but also in photoreceptors, lacrimal canaliculus epithelial stem cells and eye wall cells. [56][57][58] These cells are all essential for sustaining the physiological functions of the eye and are located in close proximity to SOX10-positive NCCs or their derivatives. 24 Hence, the SOX10-KIT-AKT axis may have a role in eye development and function beyond its role in maintaining LESCs that we demonstrated here.
Our results suggest a critical role of the SOX10-KITL signalling pathway in LNC for maintaining LESC survival. In addition, other signalling pathways including the BMP and WNT pathway have also been suggested to be essential for LNCs' roles in supporting LESCs. 59 Hence, a detailed knowledge of the molecular parameters of how the stem cell niche supports LESCs may evidently be of chief importance to design rational therapies for LSCD, one of the major causes of reduced visual acuity and blindness. In fact, reconstruction of the niche for LESCs may be an attractive strategy to treat LSCD. Both LNCs and growth factors derived from LNCs, including KITL and PEDF, have been applied towards niche reconstruction, and results from these studies are encouraging. 22,60 Nevertheless, it is currently not clear whether deficiencies in the SOX10-KITL signalling pathway in the niche only affects LESCs, or whether such deficiencies also feedback on the niche cells themselves, in particular on LNCs. Given that mice heterozygous for a loss-of-function mutation in Sox10 develop corneal structural abnormalities relatively late in life and that homozygosity for such Sox10 mutations leads to embryonic or perinatal lethality, it may become necessary to generate conditional niche cell-specific Sox10 knockouts. Such models would also lend themselves to exploration of therapeutic approaches such as adeno-associated virus-based long-term expression of SOX10 for niche reconstruction in LSCD.
In sum, our results provide evidence that SOX10 and KITL enable LNCs in a limbal stem cell niche to provide trophic support for LESCs and that KITL acts through the KIT-AKT signalling cascade in LESCs. These findings contribute to our understanding of the molecular mechanisms underlying limbal niche function. We hope that they also provide hints for potential therapeutic avenues for patients afflicted with the devastating visual dysfunctions associated with limbal stem cell deficiency.

ACK N OWLED G EM ENTS
We thank Drs. Michael Wegner and William J. Pavan for reagents, and thank Dr. Heinz Arnheiter for thoughtful comments and editing of the manuscript. This work was supported by the National Natural

CO N FLI C T O F I NTE R E S T
The authors state no conflict of interest.