Abstract Purpose To develop a method for enriching keratinocyte progenitor cells (KPCs) and establish a limbal niche (LN)-mediated transdifferentiation protocol of KPCs into corneal epithelial cells. Methods Limbal niche cells (LNCs) were isolated from limbal tissues through enzymatic digestion and characterized. Conditioned medium from LNCs cultures was collected. KPCs were enriched by rapid adhesion of Matrigel and subsequently cultured in either an LNCs-conditioned medium supplemented with KSFM (LN-KS) or SHEM (LN-SH) for 14 days. Corneal-specific marker expression was assessed to evaluate transdifferentiation efficiency. Key transcription factors and signaling pathways involved in the transdifferentiation process were identified through single-cell and RNA sequencing, and were validated by western blot and quantitative real-time PCR. Results Both LN-KS and LN-SH protocols successfully induced corneal epithelial cell transdifferentiation from KPCs, with LN-KS demonstrating higher efficiency in generating CK12 + and p63 + cells (p < 0.001). RNA sequencing analysis and western blot have revealed significant activation of STAT3 and PI3K/AKT signaling pathways. Inhibition of STAT3 blocked the activation of PI3K/AKT signaling pathway and impaired corneal epithelial cell transdifferentiation. Conclusions This study demonstrates the ability of LN to promote KPCs transdifferentiation into corneal epithelial cells in vitro, and this process is partially mediated by the STAT3/PI3K/AKT signaling pathway. Graphical abstract [46]graphic file with name 13287_2024_4129_Figa_HTML.jpg Supplementary Information The online version contains supplementary material available at 10.1186/s13287-024-04129-w. Keywords: Corneal epithelial cells, Transdifferentiation, Keratinocyte progenitors, Limbal niche, STAT3, PI3K/AKT signaling pathway Introduction The cornea, a transparent, avascular structure at the front of the eye, consists of multiple layers. The outermost layer, the corneal epithelium (CE), is continually exposed to external factors and requires rapid regeneration. This process is facilitated by limbal epithelial progenitor cells (LEPCs) residing in a specialized niche, the limbus, at the corneoscleral junction. As adult stem cells with robust proliferative and self-renewal capacity, LEPCs are essential for maintaining CE integrity and transparency [[47]1, [48]2]. Destruction of LEPCs due to ocular burns or severe ocular diseases results in limbal stem cell deficiency (LSCD), a primary cause of most blinding keratopathies [[49]3]. In these cases, invasion of the cornea by conjunctival epithelial cells leads to chronic inflammation, persistent corneal defects, neovascularization, and vision loss [[50]4]. Various surgical techniques, including autologous limbal stem cell transplantation, allogeneic corneal transplantation, and cultured corneal limbal epithelial transplantation (CLET), have been employed to treat LSCD [[51]5]. However, the limited availability of healthy limbal tissue and the need for prolonged systemic immunosuppression restrict the efficacy of these approaches [[52]6]. In the pursuit of innovative therapies for corneal diseases, alternative cell sources have been highlighted [[53]7, [54]8], including hair follicle stem cells [[55]9], mesenchymal stem cells [[56]10, [57]11], pluripotent stem cells [[58]12, [59]13] and oral mucosal epithelial cells (OMECs) [[60]14, [61]15]. Notably, OMECs have emerged as a promising autologous source of corneal epithelial stem cells. Transplantation of cultured OMECs has yielded encouraging preliminary results in patients with bilateral LSCD [[62]16, [63]17]. Nevertheless, this approach currently faces challenges related to inconsistent outcomes and limited oral mucosa availability. Stemming from ectoderm similar to CE, skin keratinocytes (SKCs) exhibit a broader transdifferentiation potential than OMECs, offering an unlimited cell source [[64]18–[65]20]. While existing studies have described protocols for generating corneal epithelial cells from SKCs, these methods often suffer from prolonged timelines, suboptimal efficiency, and a lack of understanding regarding underlying developmental mechanisms. Notably, Gopakumar et al. produced CK12-positive cells from SKCs in 31 days [[66]21], while Sakurai et al. generated CK14-positive cells within 5 weeks, falling short of mature CK12-positive corneal epithelial cells [[67]22]. Consequently, there is a clear need to refine SKCs culture systems to enhance transdifferentiation efficiency. Given that corneal epithelium can differentiate into epidermis under the influence of dermal signals [[68]23], SKCs possess the intrinsic capacity to acquire a corneal epithelial phenotype when exposed to appropriate corneal limbal cues. Our previous research demonstrated that the limbal niche (LN) effectively promotes corneal epithelial cell transdifferentiation from OMECs within 14 days [[69]14]. Therefore, we employed LN in this study to provide essential corneal developmental signals. To further optimize this process, we focused on differentiating keratinocyte progenitor cells (KPCs) enriched from primary SKCs into corneal epithelial cells rather than utilizing the entire SKCs population. Herein, we established a two-step transdifferentiation protocol for KPCs. Initially, KPCs were enriched from SKCs through Matrigel-based adhesion culture. To induce corneal epithelial transdifferentiation, we recapitulated the microenvironment of the corneal limbus in vitro by utilizing conditioned media from LNCs. This approach enabled the generation of multiple populations of corneal epithelial cells and provided preliminary evidence for the involvement of the STAT3/PI3K/AKT signaling pathway in this process. Methods Study animals The reporting of animal experiments in this study adhered to the ARRIVE guidelines. All procedures were approved by the Institutional Animal Care and Use Committee of Tongji Medical College, Huazhong University of Science and Technology (IACUC number 3869, approved on October 10, 2023). Sprague–Dawley rats (aged 6–8 weeks, in the resting phase of hair growth) were obtained from the Experimental Animal Center of Huazhong University of Science and Technology (Wuhan, China). Rats exhibiting no visible abnormalities in appearance or behavior were randomly assigned to control and treatment groups using a random number generator. The rats were housed in a pathogen-free environment under controlled conditions (24 °C, 12-h light/dark cycle, 50% humidity) with ad libitum access to food and water. All animals were anesthetized via inhalation of isoflurane (95% oxygen and 5% isoflurane, provided by RWD Life science Co.,Shenzhen, China; catalog number 550933) and were subsequently euthanized by cervical dislocation when necessary. Following euthanasia, all rats were used for skin keratinocytes and corneal epithelial cells isolation. Corneal epithelial transdifferentiation The experimental design of this study is schematically illustrated in Additional file [70]1: Fig. S1. Following isolation from the limbal region, LNCs were seeded onto Matrigel-coated well plates (5 mg/cm^2; catalog number C356234; Corning) at a density of 1 × 10^4/cm^2. LNCs were maintained in adherent culture for an additional 21 days (passaged to P3) in MESCM, with medium replacement three times weekly to achieve LNC purification. Subsequently, the culture medium from P3 LNC cultures was collected and centrifuged at 12,000 rpm for 5 min, and the supernatant was designated as LNCs conditioned medium. To initiate corneal epithelial transdifferentiation, obtained KPCs were transferred to Matrigel-coated six-well plates (5 mg/cm^2; catalog number C356234; Corning) for 14 days, with daily medium changes. This induction phase was conducted under one of three culture conditions: (1) KSFM condition, employing commercial keratinocyte serum-free medium (catalog number 17005042; Thermo Fisher Scientific) supplemented with recombinant human epidermal growth factor (rhEGF) (Thermo Fisher Scientific) and bovine pituitary extract (BPE) (Thermo Fisher Scientific); (2) LNCs-KS condition, utilizing KSFM supplemented with LNCs conditioned medium at a 1:1 ratio; (3) LNC-SH condition, employing supplemented hormonal epithelial medium(SHEM) supplemented with LNCs conditioned medium at a 1:1 ratio. SHEM was composed of DMEM/F12 (Gibco) supplemented with 5% fetal bovine serum (Gibco), 2.5 μg/mL insulin, 2.5 μg/mL transferrin, 2.5 ng/mL sodium selenite, 0.45 μg/mL hydrocortisone, 20 ng/mL human epidermal growth factor, 10 ng/mL hLIF, 50 μg/mL gentamicin, and 1.25 μg/mL amphotericin B. Primary LNCs and LECs culture LNCs isolation and culture were performed as follows. Following anesthesia and spinal cord dislocation in rats, eyeballs were enucleated and separated into limbus and central cornea using a 3.5 mm diameter trephine. Excessive sclera, conjunctiva, iris, trabecular meshwork, and endothelium were meticulously removed. Corneoscleral rims were then clipped 1 mm inside and outside the anatomic limbus and divided into six sections. LNCs clusters were obtained by removing limbal epithelial sheets using neutral protease II (Sigma-Aldrich) followed by digestion in 1 mg/mL collagenase A (Sigma-Aldrich) for 3 h. LNCs clusters were further digested with 0.25% trypsin–EDTA at 37 °C for 15 min to yield single cells. Cells were then suspended at a density of 1 × 10^4 cells/cm^2 in six-well plates coated with 5% Matrigel in modified embryonic stem cell medium (MESCM). MESCM consisted of DMEM/F12 (Gibco), supplemented with 10% knockout serum, insulin (5 μg/mL), transferrin (5 μg/mL), sodium selenite (5 ng/mL), basic fibroblast growth factor (4 ng/mL), human leukemia inhibitory factor (hLIF, 10 ng/mL), gentamicin (50 μg/mL), and amphotericin B (1.25 μg/mL). Upon reaching confluence, cells were passaged at a ratio of 1:4 up to passage 3 (P3). The expression of mesenchymal stem cell markers Sox2 and Oct4 in P3 LNCs was analyzed by quantitative real-time PCR and Western blotting, while N-cadherin expression was assessed by immunofluorescence. Additionally, double immunofluorescence staining of vimentin (Vim) with CK12, ΔNp63α, or PAX6 was performed to confirm the purity of isolated LNCs. In the single-cell culture system for limbal epithelial cells(LECs), corneoscleral rims were dissected into small segments and incubated with 10 mg/mL Dispase II (Sigma-Aldrich) at 37 °C for 30 min. The limbal epithelium was carefully detached under a dissecting microscope and subsequently treated with 0.25% trypsin–EDTA at 37 °C for 5 min to obtain a single-cell suspension. The cell suspension was seeded at a density of 200 cells/cm^2 onto six-well plates coated with 5% Matrigel in KSFM. Primary SKCs culture Rat dorsal skin was prepared in advance and subsequently excised. Tissues were washed thrice in cold phosphate-buffered saline (PBS) containing penicillin (100 U/mL), streptomycin (100 U/mL), and amphotericin B (1.25 mg/mL). Skin samples were placed in a 100 mm Petri dish, and excess subcutaneous fat, blood vessels, hair, and connective tissue were meticulously removed. The remaining tissue was cut into 2 × 2 cm pieces and digested with Dispase II (10 mg/mL) at 37 °C for 30 min. The epidermis was peeled from the dermis using forceps under a dissecting microscope and divided into smaller segments. These segments were then treated with 0.25% trypsin–EDTA at 37 °C for 15 min. The resulting suspension was filtered through a 70 µm nylon strainer and centrifuged at 1400 rpm for 10 min. The supernatant was discarded, and the pellet was resuspended in KSFM to obtain a single-cell suspension. Cell counts were determined using a hemocytometer, and cells were seeded into a six-well coated dish. The culture plate was incubated at 37 °C in a humidified atmosphere containing 95% air and 5% CO[2]. The medium was replaced after 24 h. Preparation of matrigel Matrigel™ Basement Membrane Matrix (BD Biosciences, San Jose, CA, USA) was prepared for coating as described previously (Ghaffarinia et al., 2023). Briefly, Matrigel was diluted 1:20 in Dulbecco’s Modified Eagle’s Medium/F12 (DMEM/F12, 1:1; Hyclone). The diluted Matrigel (50 μL/cm^2) was applied to culture plates and incubated at 37 °C with 5% CO₂ for 1 h. Prior to cell seeding, the Matrigel coating was removed. All procedures were conducted on ice. Quantitative real-time PCR Total RNA was extracted using TRIzol Reagent (Thermo Fisher Scientific) and subsequently converted to cDNA using a reverse transcription kit (GeneCopoeia, Rockville, MD, USA). Quantitative real-time PCR amplification was initiated with a preliminary pre-denaturation step at 50 °C for 2 min, followed by 40 cycles of denaturation at 95 °C for 10 s, annealing at 95 °C for 30 s, and extension at 60 °C for 30 s. Gene expression levels were quantified using the comparative cycle threshold (Ct) method, with β-actin serving as an endogenous control for normalization. Each experiment was performed in triplicate, and primer sequences are listed in Additional file [71]2: Table S2.1. Western blotting Cells were washed with PBS and subsequently lysed using radioimmunoprecipitation assay (RIPA) lysis buffer (Solarbio) supplemented with a protease and phosphatase inhibitor cocktail (Solarbio). The lysate was centrifuged to collect the supernatant. Protein concentration was determined using the bicinchoninic acid (BCA) protein assay (Beyotime Biotechnology). Proteins were separated by sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE) and transferred to a polyvinylidene fluoride (PVDF) membrane according to standard protocols. Membranes were blocked with 5% non-fat milk in Tris-buffered saline with Tween-20 (TBST) and incubated with primary antibodies overnight at 4 °C. Following three TBST washes, membranes were incubated with secondary antibodies for 1 h at room temperature. β-actin or glyceraldehyde-3-phosphate dehydrogenase (GAPDH) served as loading controls. Immunoreactive bands were visualized using chemiluminescence and quantified by densitometry using BandScan software. Antibody information is provided in Additional file [72]2: Table S2.2. Immunofluorescence staining Tissue sections and single cells were fixed in 4% paraformaldehyde for 15 min, permeabilized in PBS containing 0.5% Triton X-100 (Sigma-Aldrich) for 60 min, and blocked with 2% bovine serum albumin (BSA) for 1 h at room temperature. Subsequently, samples were incubated with primary antibodies diluted in 1% BSA and 0.1% Triton X-100 overnight at 4 °C. Following three PBS washes, sections and cells were incubated with secondary antibodies for 1 h at room temperature and washed an additional three times with PBS. Nuclei were stained with 4',6-diamidino-2-phenylindole (DAPI; Invitrogen) for 5 min at room temperature, washed thrice in PBS, and mounted in 50% glycerol (Sigma-Aldrich). Images were captured using a laser scanning confocal microscope (LSM700; Carl Zeiss Microscopy, White Plains, NY, USA). Fluorescent signal quantification in three randomly selected image areas was performed using ImageJ software (National Institutes of Health, USA). Antibody information is provided in Additional file [73]2: Table S2.2. RNA sequencing analysis For library preparation, 1 µg of total RNA per sample was subjected to poly(A) selection using oligo(dT) beads, followed by reverse transcription into cDNA using the SuperScript Double-Stranded cDNA Synthesis Kit (Invitrogen) with random hexamer primers. The resulting cDNA underwent end repair, phosphorylation, and adenylation according to Illumina's library preparation protocol. Size selection of 300 bp fragments was performed using 2% Low Range Ultra Agarose gel electrophoresis, followed by PCR amplification with Phusion DNA Polymerase (NEB) for 15 cycles. Quantification of libraries was achieved using a TBS380 fluorometer (Turner Biosystems). Paired-end sequencing (150 bp reads) was conducted on an Illumina HiSeq X Ten/NovaSeq 6000 platform. Raw sequencing data underwent quality control assessment using FastQC, and adapter sequences were trimmed with Trimoraic. Differential expression analysis was performed using DEseq2, identifying genes with |log2FC|> 1 and a Q value ≤ 0.05 as significantly differentially expressed. Data acquisition and processing For this study, single-cell RNA sequencing (scRNA-seq) data were procured from the GEO database, specifically datasets PRJNA891516 and [74]GSE122043. Preprocessing of scRNA-seq data included filtering for genes expressed in at least three cells and cells expressing a minimum of 200 genes. Mitochondrial content was assessed using the PercentageFeatureSet function within the Seurat R package. Subsequently, differential expression analysis was performed to identify genes significantly expressed across experimental conditions. Pathway enrichment analysis In contrast to single-gene studies, Gene Set Enrichment Analysis (GSEA) leverages gene lists to identify significant differences in biological functions or pathways between two groups (Subramanian et al., 2005). For our analysis, we utilized gene sets curated from the Molecular Signatures Database (MSigDB) ([75]https://www.gsea-msigdb.org/gsea/msigdb/index.jsp), encompassing KEGG, Gene Ontology (GO), Reactome, and Hallmark pathways. From this comprehensive collection, we selected pathways relevant to our specific research question for input into GSEA. Transcription factor analysis To analyze transcription factor (TF) activity, we employed the "SCENIC" algorithm (Aibar et al., 2017) ([76]https://github.com/aertslab/SCENIC). This algorithm infers co-expression regulatory networks between TFs and candidate target genes from scRNA-seq datasets. The analysis entails three key steps: (1) identification of candidate target genes based on co-expression networks, (2) recognition of directly accessible genes through motif analysis to construct regulons (groups of co-regulated genes), and (3) scoring of all regulons. These steps are implemented through the "GENIE3", "RcisTarget", and "AUCell" functions within the SCENIC framework. Statistical analysis Statistical analyses were performed using SPSS Statistics. Data are presented as mean ± standard deviation (SD) and were compared using two-tailed unpaired t-tests and one-way analysis of variance. A p-value < 0.05 was considered statistically significant. Results Localization of KPCs and LEPCs in rat skin epidermis and limbus We initially found that in normal rat cornea and limbus, LEPCs expressing ΔNP63α were located in the basal layer of the corneal limbal epithelium. In contrast, ΔNP63α expression was very faint in the central corneal epithelium. Similarly, KPCs exhibited ΔNP63α expression at the epidermal basal membranes. Importantly, immunofluorescence staining confirmed the previously reported expression patterns of CK12 and PAX6 in corneal epithelial cells, as well as CK10 and CK1 in keratinocytes (Fig. [77]1A). Fig. 1. [78]Fig. 1 [79]Open in a new tab The molecular expression patterns in the skin and cornea. Immunofluorescence of the epidermis, central cornea, and limbus showed that CK12 and PAX6 were identified as corneal epithelial cell markers, whereas CK1 and CK10 as the keratinocyte markers. Besides, limbal epithelial progenitor cells and keratinocyte progenitor cells all expressed ΔNP63α, which had been detected only on the basal part of the epithelium. (scale bar = 50 μm) Characterization of LNC markers To isolate LNCs, limbal segments were subjected to enzymatic digestion using dispase II and collagenase. The resulting LNCs were plated on Matrigel-coated well plates and further purified using MESCM. In primary cultures, abundant rounded epithelial cells were observed in suspension on day 2, followed by the formation of typical epithelial colonies by day 5, with fibroblast-like LNCs present in the surrounding area. LNCs were passaged at a ratio of 1:4 up to passage 3 (P3). Morphological observations revealed a significant decline in rounded epithelial cells and the emergence of predominantly fibroblast-like cell outgrowths by P3 (Fig. [80]2A). To characterize the epithelial and mesenchymal components during LNC culture, CK12 and vimentin expression was assessed by quantitative PCR. A gradual decrease in CK12 + epithelial cells and an increase in vimentin + mesenchymal cells were observed with each passage (Fig. [81]2B) (p < 0.001). Immunofluorescence staining of P3 LNCs confirmed the predominance of vimentin-positive cells and the absence of CK12 expression, indicating successful LNC purification (Fig. [82]2C). Given the critical role of embryonic stem cell (ESC) markers in LNC niche function, we compared ESC marker expression in LNCs expanded to P3. The mRNA levels of SOX2 and Oct4 were significantly upregulated followed by generations(p < 0.001 and p < 0.01, respectively) (Fig. [83]2D). Immunofluorescence staining corroborated these findings, demonstrating increased SOX2 and Oct4 expression in P3 LNCs (Fig. [84]2E). Fig. 2. [85]Fig. 2 [86]Open in a new tab Molecular phenotype characterization of LNCs. A Morphologically, representative images of LNCs cultured in MESCM at different stages of purification have been captured. Images showed that during the purification process, circular cells gradually disappeared, and dormant fibroblasts mainly aggregated in P2 LNCs (scale bar = 50 μm). B-C Quantitative real-time PCR showed that followed by generations, the expression of Vimentin has been upregulated, and CK12 expression was no longer observed. The purified P3 LNCs were double-stained for Vim (red fluorescence) and CK12 (green fluorescence), ΔNP63α (green fluorescence) (scale bar = 50 μm). D-E Quantitative real-time PCR showed Oct4 and SOX2 have been enriched in P3 LNCs. Furthermore, P3 LNCs were single-stained for Oct4 and SOX2 (scale bar = 50 μm). Data are from three independent experiments. *p < 0.05, **p < 0.01, ***p < 0.001 Utilizing KSFM and Matrigel, KPCs were enriched from the primary SKCs culture As illustrated in additional file [87]1: Fig. S2A, the epidermis was meticulously separated from epidermal tissue under microscopic observation following dispase II digestion and subsequently dissociated into individual keratinocytes using trypsin. SKCs were initially cultured in suspension using one of three induction media: KSFM, defined keratinocyte serum-free medium (dKSFM), or SHEM. Morphological analysis revealed pronounced fibroblast contamination in cultures using dKSFM and SHEM, while SKCs maintained a relatively uniform keratinocyte morphology and predominant growth in KSFM (Additional file [88]1: Fig. S2B). Quantitative real-time PCR demonstrated that KSFM significantly enhanced SKCs proliferation, as evidenced by elevated CK10 mRNA expression (all p < 0.001) and reduced vimentin expression (all p < 0.001) (Additional file [89]1: Fig. S2C). These findings indicate that KSFM supports SKCs expansion. Previous studies have established Matrigel as an in vitro platform that mimics the basement membrane, facilitating epithelial cell adhesion and expansion of epithelial progenitor cells. To assess the impact of Matrigel on stemness maintenance and proliferative capacity of cultured SKCs, the expression levels of the stemness marker ΔNP63α and the proliferation marker Ki67 were evaluated. Quantitative real-time PCR results demonstrated a consistently slower decline in Ki67 and ΔNP63α expression in SKCs cultured on Matrigel compared to control cells (Ki67: all p < 0.01; ΔNp63α: p < 0.05 and p < 0.001) (Additional file [90]1: Fig. S2 D, E). Consequently, subsequent experiments favored adherent culture on a Matrigel-coated substrate using KSFM. Cell outgrowths emerged from attached aggregates as early as day 2, forming a distinct confluence by day 5. Cells matured in KSFM exhibited a polygonal epithelial morphology, characterized by small, cuboidal cells arranged in a regular, cobblestone-like pattern (Additional file [91]1: Fig. S2F). To investigate the enrichment of KPCs and the optimal timing for transdifferentiation, cultured SKCs were maintained in KSFM and subcultured at four-day intervals. Immunofluorescence staining revealed that both P0 and P1 SKCs predominantly consisted of progenitor cells with a relatively low proportion of mature CK10-positive cells (Fig. [92]3A). Quantitative real-time PCR analysis demonstrated that while P0 KPCs exhibited the highest ΔNP63α expression, there was no significant difference in ΔNP63α levels between P0 and P1 KPCs (ΔNP63α: all p < 0.001) (Fig. [93]3B). Additionally, CK10 and vimentin mRNA expression was markedly lower in P1 SKCs compared to P0 SKCs (Fig. [94]3C, D) (CK10: p < 0.001; vimentin: p < 0.001). Western blot analysis corroborated these findings (Fig. [95]3E, F). These results indicate that although both P0 and P1 SKCs contain a higher proportion of progenitor cells, P0 SKCs exhibit significant fibroblast contamination. Therefore, P1 SKCs were selected for subsequent transdifferentiation experiments. Fig. 3. [96]Fig. 3 [97]Open in a new tab Utilization of KSFM and Matrigel improved the adhesion and stemness maintenance of keratinocyte progenitor cells. A IF staining showed that P0 and P1 SKCs contained dominant p63α positive cells and faint CK10 positive cells. However, P0 SKCs contained more Vimentin-positive cells than P1 SKCs (scale bar = 50 μm). B-F Quantitative real-time PCR (B-D) and western blotting (E–F) of ΔNP63α, CK10, Vimentin showed that ΔNP63α expression of P0 and P1 SKCs was significantly higher than P2-P3 SKCs.However, the expression of ΔNP63α did not significantly differ between P0 SKCs and P1 SKCs. In addition, CK10 expression showed the lowest level at P0-P3 SKCs.Vimentin expression showed the highest level at P0-P3 SKCs and was very faint from P1 SKCs. Data are from four independent experiments. *p < 0.05, **p < 0.01, ***p < 0.001 LN facilitated transdifferentiation of KPCs to corneal epithelial cells To evaluate the influence of LN on KPC transdifferentiation into corneal epithelial cells and optimize this process, KPCs were subjected to transdifferentiation in two induction media: LNCs conditioned medium supplemented with KSFM (LN-KS) and LNCs conditioned medium supplemented with SHEM (LN-SH). KSFM alone served as a control to assess spontaneous transdifferentiation efficiency. Cells cultured in KSFM exhibited a highly keratinized morphology with evidence of stratification at the end of the study. Conversely, KPCs treated with LN displayed a relatively undifferentiated phenotype compared to the control (Fig. [98]4A). To investigate the effects of induction medium on transdifferentiation, corneal epithelial cells marker expression was analyzed. Immunofluorescence staining revealed a significantly higher abundance of CK12 + cells in LN-KS-induced KPCs compared to LN-SH and KSFM groups (all p < 0.001) (Fig. [99]4B,C). Furthermore, CK12 protein level was markedly elevated in LN-KS KPCs compared to LN-SH and KSFM groups (all p < 0.001) (Fig. [100]4D), indicating a greater extent of transdifferentiation. Additionally, a higher proportion of p63α-positive cells was observed in the LN-KS culture system (all p < 0.001) (Fig. [101]4E), suggesting that KSFM contributed to maintaining KPC stemness. These findings collectively demonstrate that LN effectively promotes KPC transdifferentiation into corneal epithelial cells in vitro, and using KSFM enhances the efficiency of this transdifferentiation process. Fig. 4. [102]Fig. 4 [103]Open in a new tab Limbal niche facilitated the transdifferentiation of keratinocyte progenitor cells into corneal epithelial cells. A Representative images of KPCs cultured in KSFM, LN-KS medium, and LN-SH medium. The image of LECs has also been captured and added (scale bar = 50 μm). B–F Immunofluorescence staining (B and C) of LN-KS and LN-SH cells and quantitative analysis (D) of CK12 + cells percentages showed that LN treatment contributed to the generation of CK12 + cells. Western blotting findings (E) confirmed that LN-KS treatment produced more significant CK12 expression than LN-SH induction. Quantitative analysis (D) of p63α + cells percentages revealed that the LN-KS group showed more remarkable p63α + cells populations than the LN-SH group, which indicated KSFM is more sufficient in maintaining the stemness of KPCs(scale bar = 50 μm). Data are from four independent experiments. *p < 0.05, **p < 0.01, ***p < 0.001 Transdifferentiation of KPCs into corneal epithelial cells mediated by the STAT3/PI3K /AKT signaling pathway By integrating and analyzing single-cell data from the normal rat cornea and skin obtained from the Gene Expression Omnibus (GEO) database, we identified significant differences in gene expression between these two tissue types. Following batch effect correction using Harmony, dimensionality reduction, and clustering analysis, we generated UMAP plots representing distinct cell clusters. The results revealed substantial heterogeneity between cells derived from the two tissue types (Additional file [104]1: Fig. S3A). Furthermore, using classical epithelial markers, we isolated epithelial cell populations from both the skin and cornea (Additional file [105]1: Fig. S3B). The analysis showed that, in addition to the typical expression of Krt10, skin epithelial cells also expressed high levels of stemness-associated genes, whereas corneal epithelial cells exhibited elevated expression of Krt12 (Additional file [106]1: Fig. S3C). In addition, Differential analysis also has revealed that corneal epithelial cells predominantly upregulated genes (Additional File [107]2: Table S1), including mature corneal epithelial cell specific markers krt12 [[108]46], ALDH3A1 [[109]47], transcription factors STAT3 [[110]48]; regulators of corneal epithelium cell fate determination Dkk2 [[111]49]and PAX6 [[112]50]. Among these studies, Dickkopf-2 (Dkk2), is expressed in the periocular mesenchyme at E11.5 and in the mesenchyme at E14.5, and it contributes in specifying the phenotype of corneal epithelium via bidirectional signaling between stromal and epithelial cells [[113]49]. Paired box 6 (PAX6) is a crucial transcription factor for developmental regulation of cornea [[114]51], which governs corneal epithelial cell fate determination [[115]50]. STAT3 plays a pivotal role in cell signaling and transcriptional activation, regulating cell proliferation, transdifferentiation, and apoptosis. Conversely, skin cells exhibited high expression of genes such as mature keratinocytes markers krt10, krt1 [[116]52], keratinocytes development regulator SOX4 [[117]53, [118]54], CXCL14 [[119]55], H19 [[120]56]. Among these, epidermal specific knockout of SOX4 reportedly promoted keratinocyte differentiation, and loss of SOX4 affected the expression of genes involved in cytoskeleton/ECM organization and Skin development [[121]53, [122]54]. CXCL14, a chemokine secreted by epidermal keratinocytes that rhythmically activates the innate immune system, suppresses skin bacterial proliferation [[123]55]. Long noncoding RNA H19 mediates the proliferation and migration of keratinocytes [[124]56] (Fig. [125]5A).To preliminarily validate the genes altered during transdifferentiation, we used qPCR to detect the expression of these differentially expressed genes in normal KPCs and LN-KS KPCs. Our results showed that in the early stages of transdifferentiation, the markers CK1 and CK10, Cxcl14 of keratinocytes were significantly reduced, and more importantly, the expression of PAX6, CK12, and ALDH3A1 was significantly increased. This evidence suggests that induction of LN promotes the corneal transdifferentiation of keratinocytes. However, SOX4, DKK2, and H19 did not show significant changes. In addition, due to the possibility of early transdifferentiation, a portion of the progenitor cell population may be transforming into CK12 positive cells (Additional file [126]1: Fig. S3D).Further enrichment analysis indicated that corneal cell clusters were primarily enriched in pathways, including IL6/JAK/STAT3, PI3K/AKT/mTOR, and Notch signaling (Fig. [127]5B). Employing the SCENIC transcription factor algorithm to analyze transcriptional regulatory activity changes in skin and corneal cells, we observed significant activation of the STAT family and related inflammatory transcription factors in corneal cells (Fig. [128]5C). Subsequent analysis of STAT family gene expression revealed a substantial upregulation of STAT3 in corneal limbal cells compared to SKCs, suggesting a potential link between STAT3 and the transformation of SKCs into corneal cells (Fig. [129]5D). To further investigate these findings, we collected rat LN-KS-KPCs and control KPCs for bulk sequencing analysis, followed by differential gene analysis and enrichment analysis (Fig. [130]5E, F, Additional File [131]2: Table S1). The results indicated activation of signaling pathways such as PI3K/AKT and MAPK in LN-KS KPCs, suggesting a potential role for the PI3K/AKT pathway in KPC transdifferentiation into corneal epithelial cells (Fig. [132]5G). Fig. 5. [133]Fig. 5 [134]Open in a new tab RNA sequencing analysis and determination. A Volcano plot displaying differentially expressed genes between corneal and skin cells. B GSEA enrichment analysis of corneal versus skin cells. C Heatmap of transcription factor activity based on the SCENIC algorithm for corneal and skin cell populations, showing the top 15 pathways with the highest transcriptional activity. D Expression levels of STAT family characteristic genes in corneal and skin cells, with boxes representing the median and interquartile ranges. E–F Heatmap and volcano plot of differentially expressed genes in skin and corneal samples from bulk sequencing. G Enrichment analysis of differentially expressed genes in corneal versus skin sequencing samples Previous studies have established a strong association between STAT3 and the JAK signaling pathway. In this pathway, extracellular signaling molecules, such as cytokines, activate JAK, which subsequently phosphorylates STAT3, leading to its activation and translocation to the nucleus for gene regulation. Beyond the JAK-STAT3 pathway, STAT3 interacts with other signaling cascades, including PI3K/Akt and MAPK, forming a complex network that governs diverse cellular functions [[135]24–[136]26]. Our sequencing results also highlighted the importance of the PI3K/Akt and MAPK pathways. To investigate the protein levels of STAT3 and key components within these signaling cascades, we performed western blotting. Results indicated a 2.5-fold increase in p-STAT3 expression in LN-KS KPCs compared to control KPCs (p < 0.01). Furthermore, phosphorylation levels of PI3K and Akt were elevated approximately twofold (p < 0.001) and 2.5-fold (p < 0.001), respectively, in LN-KS KPCs compared to controls (Fig. [137]6A and 6B). These findings align with the upregulated STAT3 expression (p < 0.001). However, phosphorylation levels of ERK and P38 MAPK did not exhibit significant differences between LN-KS KPCs and control KPCs (Additional file [138]1: Fig. S1G and S1H). Fig. 6. [139]Fig. 6 [140]Open in a new tab LNCs facilitate the transdifferentiation of KPCs towards corneal epithelial cells via the STAT3 signaling pathway. A and B WB confirmed that the protein level of STAT3 was significantly higher in LNCs-treated KPCs than in control SKCs (p = 0.045). In addition to an increased level of STAT3, the phosphorylation levels of PI3K and AKT were significantly increased by LNCs treatment (all p < 0.001). C and D WB showed that Stattic treatment inhibited the phosphorylation of STAT3 of LN-KS treated KPCs (p < 0.001). The phosphorylation levels of PI3K and AKT were significantly lower in KPCs co-treated with Stattic + LNCs than in KPCs treated with LNCs only (all p < 0.001). Data are from three independent experiments. *p < 0.05, **p < 0.01, ***p < 0.001 To determine the involvement of STAT3-activated PI3K/Akt signaling in the transdifferentiation process, a specific STAT3 antagonist, Stattic (20 mg/mL), was administered from day 0. Western blot analysis revealed a significant downregulation of p-STAT3 expression in LN-KS KPCs following Stattic treatment (p < 0.001). In addition, the significantly reduced phosphorylation levels of PI3K (p < 0.001) and AKT (p < 0.001) have been confirmed in Stattic-treated LN-KS KPCs compared to untreated controls (Fig. [141]6C and 6D). Moreover, 54.9% ± 1.7% of cells were CK12-positive in the LN-KS culture system. However, this percentage decreased to 25.3% ± 1.8% after Stattic treatment (Fig. [142]7A, [143]B). Western blot analysis further confirmed suppressed CK12 protein levels in LN-KS KPCs treated with the STAT3 inhibitor (p < 0.001). Importantly, STAT3 inhibitor treatment alone failed to induce CK12 expression in KPCs (Fig. [144]7C, [145]D). Additionally, quantitative real-time PCR demonstrated PAX6 activation in LN-KS KPCs, which was impaired by Stattic treatment (all p < 0.001) (Fig. [146]7E). These findings collectively indicate that activation of the PI3K/AKT signaling pathway is essential for LN-mediated corneal epithelial cell transdifferentiation. Fig. 7. [147]Fig. 7 [148]Open in a new tab Partial deprivation of transdifferentiation towards corneal epithelial cells after STAT3 inhibitor treatment. A and B IF staining (A) and quantification of CK12 + cells (B) showed that Stattic treatment significantly decreased the population of CK12 + cells (CK12: p < 0.001) (scale bar = 50 μm). C and D Studies of quantitative real-time PCR showed that the corneal epithelial cell marker CK12 and PAX6 of KPCs co-treated with LN-KS + Stattic was significantly lower than that of KPCs treated with LN-KS only (all p < 0.001). Stattic alone did not affect the transdifferentiation of KPCs towards corneal epithelial cells. E and F Western blot findings showed that Stattic treatment suppressed the protein levels of CK12 in KPCs treated in LN-KS (p < 0.001). Data are from three independent experiments. *p < 0.05, **p < 0.01, ***p < 0.001 Discussion In this study, we investigated the effects of the limbal niche on the transdifferentiation of KPCs into corneal epithelial cells and preliminarily assessed the role of the STAT3/PI3K/AKT signaling pathway in this process. Initially, we validated biomarker phenotypes for the epidermis, central cornea, and limbus, confirming CK12 and PAX6 as corneal-specific markers, consistent with previous findings [[149]27, [150]28]. Moreover, our results demonstrated the presence of epidermal progenitor cells in the basal layer of epidermal epithelium, which expressed ΔNP63α similarly to LEPCs in the basal layers of limbal epithelium [[151]29, [152]30]. Analogous to the LEPCs extraction protocols, we employed dispase enzyme to remove the dermis from skin tissues, thereby obtaining the epidermis keratinocytes (Calenic et al., 2015; Li et al., 2017). Furthermore, morphological and PCR findings indicated that KSFM is more conducive to SKCs expansion. The capacity to adhere to type IV collagen is recognized as a crucial characteristic of epithelial cells. Previous differentiation studies have leveraged this property to isolate epithelial progenitor cells from the epithelial cell population [[153]11, [154]22]. Additionally, p63α has been established as a marker associated with the stemness and cell migration of various epithelial cell lineages, including corneal epithelial cells [[155]18, [156]31]. Similarly, in our study, we opted for adherent culture on a Matrigel-coated substrate for KPCs expansion due to the higher adhesion rate of p63α-positive cells compared to suspension culture. To further validate the KPCs enrichment process in P0-P3 SKCs cultured in KSFM, we examined the expression of p63α, CK10, and vimentin. Results indicated that both P0 and P1 SKCs predominantly contained progenitor cells, but P0 SKCs also included a significant population of fibroblasts. Therefore, we favored P1 SKCs for subsequent transdifferentiation experiments due to their enhanced KPCs enrichment. Previous studies have confirmed the capacity of LNCs to effectively maintain LEPCs in an undifferentiated state in vitro [[157]32, [158]33], making them widely used as feeder cells for in vitro LEPC cultivation [[159]34, [160]35]. Moreover, LNCs have been shown to facilitate the transdifferentiation of corneal epithelial cells from OMECs and hair follicle stem cells by providing essential corneal development signals [[161]9, [162]14]. Notably, our research has pioneered and optimized the culture protocol for LNCs, capable of delivering corneal development signals to other stem cells, and identified SOX2, Oct4, and N-cadherin as LNC markers. Consistent with previous reports [[163]33, [164]34], we employed a dispase-enzyme and collagenase extraction method to obtain LNCs. Dual immunofluorescence staining of P3 LNCs confirmed their purification. Furthermore, RT-PCR and immunofluorescence staining identified the expression of ESC markers Oct4 and Sox2 in P3 LNCs. To further assess the transdifferentiation state of KPCs following induction with LN-KS and LN-SH, we examined the expression of CK12, PAX6, and ΔNP63α. Compared to control groups (LECs and KPCs), KPCs cultured in KSFM exhibited a highly mature, keratinized state at the endpoint, characterized by an abundance of corneal cells. Conversely, LN-treated groups formed regularly arranged, cobblestone-shaped structures. Immunofluorescence revealed that LN significantly improved the overall yield of CK12-positive cells in both LN-KS and LN-SH cultures. Notably, approximately 59.1% of cells in the LN-KS group exhibited CK12 positivity, while the LN-SH culture system showed 21.2% CK12-positive cells by the study endpoint. Importantly, LNCs conditional medium upregulated protein expression of CK12. Additionally, the activation of PAX6, a key transcription factor during early eye development, further supported the generation of corneal epithelial cells. Transdifferentiation under spontaneous conditions (maintained in KSFM throughout) was insufficient for generating CK12-positive cells. Interestingly, while both culture methods demonstrated the ability of LN to induce KPCs transdifferentiation towards a corneal epithelial cell fate, the LN-KS group generated a higher proportion of CK12 + corneal epithelial cells. This may be associated with the higher number of p63α + cells observed in the LN-KS culture system. These findings suggest that KSFM may be more favorable for maintaining KPCs stemness than SHEM. Overall, these results confirm the capacity of both culture systems to generate corneal epithelial cell populations from a limited number of KPCs. Further optimization has the potential to improve yields across both conditions. While a definitive marker specific to LEPCs remains elusive, the transcription factor p63, particularly its ΔNP63α isoform, is strongly associated with stemness and is highly expressed in the basal layer of the corneal limbus [[165]12, [166]18]. Mutations in the p63 gene result in severe abnormalities, leading to syndromes of ectodermal dysplasia that affect the cornea and other ectodermal tissues [[167]36, [168]37], highlighting the critical role of this transcription factor in normal corneal development. Moreover, the clinical significance of p63 expression is underscored by findings demonstrating a 78% success rate for limbal stem cell transplants with over 3% p63-positive cells, compared to only 11% success in LSCD patients receiving grafts with 3% or fewer p63-positive cells [[169]38]. In this context, our protocols achieved 29.05% ± 1.1% and 15.1 ± 1.3% p63-positive cells, respectively, which may potentially translate to favorable clinical outcomes. While many of the underlying developmental mechanisms and signaling pathways remain unclear, our RNA bulk and single-cell sequencing data underscore the critical roles of the transcription factor STAT3 and pathways such as PI3K in this process. STAT3, a widely expressed transcription factor, was found to be activated during transdifferentiation. It is well-established that STAT3 participates in cell proliferation, stem cell activation, and tissue regeneration through ligand binding and subsequent activation of various signaling pathways, including MEK/ERK, PI3K/Akt, and mTOR [[170]39]. STAT3 has also been implicated in regulating corneal endothelial barrier function by influencing ZO-1 expression. Additionally, STAT3-activated PI3K-AKT-NF-κB signaling plays a pivotal role in corneal epithelial cell proliferation [[171]40, [172]41]. Our study demonstrated that STAT3 inhibition significantly decreased the expression of p-PI3K, p-AKT, and CK12 in LN-KS KPCs without affecting p-ERK1/2 and p-P38 MAPK levels. These findings suggest that LNCs induce corneal development in KPCs by activating the STAT3/PI3K/Akt signaling cascade. Transdifferentiation of corneal epithelial cells from keratinocytes has proven challenging, with most previous studies relying on undefined factors such as conditioned medium, feeder cells, or amniotic membrane [[173]42–[174]45]. Moreover, the transdifferentiation of p63-positive corneal epithelial progenitor cells has been inadequately examined and quantified, primarily focusing on qualitative analyses. These studies have generally failed to generate mature corneal epithelial cells expressing terminal transdifferentiation markers CK3 and CK12. For example, Sakurai et al. reported colonies containing p63- and CK14-positive cells after five weeks of transdifferentiation [[175]22]. Additionally, Gopakumar et al. described a time-consuming method to generate CK12-positive cells, requiring 31 days of transdifferentiation [[176]21]. In contrast, our transdifferentiation method is highly efficient and consistent, producing corneal epithelial cells within three weeks. Moreover, previous studies primarily focused on differentiating SKCs extracted using collagenase, which often leads to dermal fibroblast contamination. Our method emphasizes KPCs enrichment rather than relying solely on SKCs. Several techniques have been proposed for isolating progenitor cells, including flow cytometry, magnetic separation, serial filtration based on cell size, rapid adherence to collagen IV, or collecting floating keratinocyte cells in monoculture to avoid trypsinization. In the present study, we opted for rapid adherence to Matrigel to generate a specific population of KPCs from primary SKCs cultures. While our findings provide valuable insights, there are a few limitations existing in the current culture conditions, which need to be further refined for future clinical applications. Firstly, to enhance the production of mature corneal epithelial cells and achieve a stratified cell sheet resembling the native corneal epithelium, some consistent and efficient stratification methods should be implemented, such as air–liquid interface culture [[177]44]. Additionally, the current experiments are focused on cellular-level validation. To further assess the efficacy of our approach in reconstructing the cornea, future corneal wound healing experiments will need to be conducted to evaluate the therapeutic outcomes on the ocular surface. Besides, single-cell sequencing, RNA sequencing, and western blotting have revealed the crucial role of the PI3K/AKT pathway in triggering this process. Studies have suggested that niche cells can secrete various chemotactic factors to activate receptors on the cell surface, thereby altering cell developmental pathways. Therefore, further exploration of underlying mechanisms could involve conducting proteomic analyses of limbal niche cells conditional medium (LN), followed by assessing significant secreted proteins and relevant receptors of transdifferentiated cells. This would deepen our understanding of the corneal signals provided by LN and potentially allow us to introduce supplements during in vitro transdifferentiation to enhance corneal yield. Last, but not least, our method has utilized the animal-derived cells and some animal products, such as the knockout serum and bovine pituitary growth factor(BPE), limiting its application in human LSCD therapy. Thus, the transdifferentiation method described here is in urgent need to be refined to be fully defined and xeno-free by using human autologous cells and substituting the animal-asociated conditions present in KSFM and MESCM. These modifications would align the protocol with good manufacturing practice standards, thereby increasing its potential for clinical translation. Conclusion In conclusion, the present study describes an efficient method for transdifferentiating rat skin keratinocyte progenitor cells toward CK12-positive corneal epithelial cells by recapitulating corneal development signals. Single-cell sequencing and western blotting have verified this transdifferentiation has been triggered by STAT3/PI3K/AKT signaling pathway, which can be partially abolished by inhibition of STAT3. In addition, with minor modifications, the current method can be carried out in fully defined and xeno-free conditions, improving its reliability and the safety of resulting cell populations. Moreover, our research has provided valuable insights for developing novel cell-based therapies that enable accurate, personalized, and reliable drug discovery for treating corneal diseases. Supplementary Information [178]13287_2024_4129_MOESM1_ESM.zip^ (9.1MB, zip) Additional file 1: Fig. S1. Schematic Outline of the Study. The culture conditions, duration for each stage, and main analyses are shown. Endpoint analyses included immunofluorescence, sequencing, western blot and qPCR.. Fig. S2. A Images of peeling off the epidermis after digesting skin via dispersing enzymes. B-C Morphological imagesand quantitative real-time PCRshowed that compared with SHEM and dKSFM, the use of KSFM suppressed the expansion of Vimentin-positive cells and promoted CK10-positive cells proliferation. D-F Quantitative real-time PCR manifested that Matrigel slowed the downregulation of NP63α and KI67 of SKCs after serial passaging. Typical images of KPC expansionat different time points. G and H Western blot findings showed that phosphorylation levels of ERK and P38 MAPK did not significantly differ between LN-KS KPCs and control KPCs. *p < 0.05, **p < 0.01, ***p < 0.001. Fig. S3. A Uniform Manifold Approximation and Projectionplot showing the main cell types from all samples. B UMAP plots indicating the expression of selected marker genes to determine epithelial subtypes. C Violin plot showing expression levels of selected genes in Corneal and skin epithelial cells. D QPCR showing the differential gene expression between KPCs and LN-KS KPCs. Fig. S4. A Unprocessed scan of Western bloting for Vimentin, CK10 and p63α in Fig. 3E. B Unprocesed scan of Western bloting for CK12 and β-actin in Fig. 4D. C Unprocesed scan of Wersten blotting for STAT3, P-STAT3 and β-actin in Fig. 6A. Fig. S5. A Unprocesed scan of Western bloting for PI3K, P-PI3K, AKT, p-AKT and β-actin in Filg. 6A. B Unprocesed scan of Westem blotting for ERK, p-ERK, P38MAPK,p-P38MAPk and β-actin in Fig. S1H. Fig. S6. A Unprocesed scan of Western bloting for STAT3, P-STAT3,PI3K, P-PI3K, AKT, p-AKT, and β-actin in Fig. 6B. B Unprocesed scan of Westem blotting for CK12 and β-actin in Fig. 7C. [179]13287_2024_4129_MOESM2_ESM.zip^ (286.6KB, zip) Additional file 2: Table S1. Differentially expressed genes between Cornea and Skin epithelial cells in single RNA sequencing. Table S2.. Acknowledgements