Posterior Limbal Mesenchymal Stromal Cells Promote Proliferation and Stemness of Transition Zone Cells: A Novel Insight Into Corneal Endothelial Rejuvenation

Abstract

Purpose: Progenitors for the corneal endothelium have been identified in the transition zone (TZ), but their cellular interactions remain undefined. Posterior limbal mesenchymal stromal cells (P-LMSCs) may support TZ cells in the posterior limbus. This study aims to characterize P-LMSCs and investigate their effects on TZ cells.

Methods: Human P-LMSCs and TZ cells were isolated by explant culture. P-LMSCs were characterized by comparing with anterior limbal mesenchymal stromal cells (A-LMSCs) using immunocytochemistry. TZ cells were cocultured with P-LMSCs in a Transwell, with TZ cell and A-LMSC coculture and TZ cells only as the control groups. The proliferation and wound healing capacity of TZ cells were assessed by EdU assay and scratch wound assay. Colony forming assay, droplet digital PCR, Western blotting, and immunocytochemistry were used to compare the stemness of TZ cells. The effect of P-LMSC conditioned medium on endothelial wound healing was evaluated in organ-cultured mouse corneas. Endothelial regeneration was measured by trypan blue staining.

Results: P-LMSCs expressed similar proteins (vimentin, Nestin, TRA-1-60, Oct3/4) as A-LMSCs. TZ cells cocultured with P-LMSCs had significantly higher proliferation, wound healing speed, and colony-forming efficiency than TZ cells only. TZ cells supported by P-LMSCs expressed higher levels of stem/progenitor markers (Nestin, Sox9, AP-2α, Pitx2) than the control groups. P-LMSC conditioned medium stimulated regeneration of mouse corneal endothelium from the TZ region.

Conclusions: The proliferation and stemness of TZ cells were enhanced by P-LMSCs in both cell and organ culture models. Our study provides an innovative strategy for corneal endothelial rejuvenation.

The corneal endothelium is a monolayer lining the back of the cornea and is vital to maintaining the relatively dehydrated and transparent status of the cornea.¹ Human corneal endothelial cells (CECs) normally possess very limited proliferative and regenerative capacities owing to contact inhibition, cell cycle inhibitors and lack of stimulation of effective growth factors in the anterior chamber.² Loss or dysfunction of CECs, such as in Fuchs’ endothelial corneal dystrophy and bullous keratopathy, leads to endothelial decompensation, corneal edema, and loss of visual acuity eventually.³ Due to the low regenerative capacity in vivo, loss of CECs in the adult human eye is compensated by size expansion of the remaining cells to preserve corneal transparency.⁴ However, for end-stage corneal endothelial disorders, the mainstay of treatment is surgical procedures, mainly including penetrating or endothelial keratoplasty.⁵ However, the availability of corneal transplants is severely restricted by the scarcity of donor tissues globally.⁶ Consequently, alternatives for donor tissues are urgently needed. 

According to previous research, CECs in the peripheral region have significantly higher cell density and proliferation rate compared to those in the center,^7,8 however, their regenerative capacity still seems limited.^9,10 Therefore, it is desirable to find ways to enhance endogenous endothelial regeneration. 

During development, the corneal endothelium is derived from periocular mesenchyme, a subgroup of neural crest, and the latter emerges from the dorsal neural tube.¹¹ Recent evidence has indicated the presence of corneal endothelial progenitors within the posterior limbus in adult humans.¹² These progenitors are thought to exist in the transition zone (TZ), an annular region that starts from the border of peripheral endothelium and ends at the insert non-filtering part of the trabecular meshwork, with Schwalbe's line located centrally.¹³ Human TZ tissues have been proven to express stem/progenitor markers including pluripotency (Sox2, Nanog, Oct3/4), neural crest (Nestin, NGFR, Sox9), and periocular mesenchyme (FoxC1, Pitx2) markers.¹⁴ Our recent study has successfully established primary cultures of TZ cells from adult human TZ explants.¹⁵ These cells demonstrated the ability to proliferate and differentiate into corneal endothelial like-cells and express functional CEC markers (ZO-1, Na⁺/K⁺ ATPase, Col8A1, AQP1, GPC4, N-cadherin, Col8A2, and Slc4a11) at later passages, suggesting that TZ cells are highly likely the stem cell source of the corneal endothelium.¹⁵ Despite the discovery of progenitor cells for the corneal endothelium, it remains uncertain whether dysfunctional TZ cells contribute to corneal endothelial pathologies or accelerated loss of endothelial cells. Moreover, it is also unknown how TZ cells maintain their stem cell properties throughout development to adulthood. 

In the anterior limbus, it has been extensively proven that the anterior limbal mesenchymal stromal cells (A-LMSCs) are critical for the stemness, proliferation, and migration of limbal epithelial stem cells within a unique microenvironment termed the limbal niche, thereby orchestrating the continual renewal of the corneal epithelium.¹⁶ However, whether such cell-cell interactions exist in the posterior limbus has not been studied. Cells residing in the posterior limbal stroma, tentatively named posterior limbal mesenchymal stromal cells (P-LMSCs) might play a comparable role in supporting the proliferation and stemness of TZ cells. 

Here, we aim to characterize P-LMSCs and investigate their effects on TZ cells. We hypothesized that cells isolated from the posterior limbus have similar properties to A-LMSCs, and that these P-LMSCs may compose a stem cell environment conducive to supporting the proliferation and stemness of TZ cells. We characterized P-LMSCs and tested the effect of P-LMSCs on TZ cells in a coculture system and an organ culture model. 

Materials and Methods

Human and Mouse Corneal Tissues

Human donor corneoscleral rims were provided by the New Zealand National Eye Bank, after the central corneal tissues had been utilized for corneal transplantation. A total of 12 donor corneoscleral rims were collected for primary culture. The research was approved by the “Northern A” Health and Disability Ethics Committee of New Zealand (number NTX/07/08/080/AM04) and handled in compliance with the Declaration of Helsinki. 

Mouse corneas were collected from euthanized mice (8-week-old; male; CD1 strain) under SOP836 by staff at the Vernon Jenson Unit of the Faculty of Medical and Health Sciences, University of Auckland. The ethical approval was waived by the Ethics Committee of University of Auckland in view of the routine colony maintenance of excess stock of animals. A total of 24 mouse corneas were used for organ culture. All procedures adhered to the ARVO statement for the Use of Animals in Ophthalmic and Vision Research. 

Cell Culture

The primary culture of human TZ cells was established according to our previously published protocol.¹⁵ Briefly, TZ tissue including the peripheral endothelium, Schwalbe’s line, and the insert portion of the trabecular meshwork was peeled from the posterior side of the human corneoscleral rim. It was further cut into 12 segments and seeded on plates pre-coated with 10 µg/mL fibronectin (341631; Sigma-Aldrich, Allentown, PA, USA) and 35 µg/mL collagen I (A1048301; Thermo Fisher Scientific, Branford, CT, USA). The TZ explants were cultivated in Opti-MEM (31985070; Thermo Fisher Scientific) supplemented with 8% fetal bovine serum (FBS; 16000044, Thermo Fisher Scientific), 5 ng/mL epidermal growth factor (PHG0313; Thermo Fisher Scientific), 40 ng/mL fibroblast growth factor (FGF; 13256-029, Thermo Fisher Scientific), 20 ng/mL nerve growth factor (PHG0126; Thermo Fisher Scientific), 20 µg/mL L-ascorbic acid 2-phosphate (A8960; Sigma-Aldrich), 200 µg/mL calcium chloride (C7902; Sigma-Aldrich), 0.04% chondroitin sulphate (C6737; Sigma-Aldrich), 50 µg/mL gentamicin (15710064; Thermo Fisher Scientific), and 1% antibiotics-antimycotics (15240062; Thermo Fisher Scientific). The culture medium was refreshed every 2 days. Once the cells reached confluency, they were passaged using TrypLE Express (12604013; Thermo Fisher Scientific) at a ratio of 1:4 and seeded according to the downstream experiment. 

Human A-LMSC and P-LMSC cultures were established by modifying previous protocols.^17,18 Specifically, after TZ tissue was peeled from the human corneoscleral rim, the remaining rim was incised at 1 mm inside and outside the anatomic limbus, so that excess cornea, sclera, and conjunctiva were removed. The limbus was digested in 1.2 U/mL Dispase (17105041; Thermo Fisher Scientific) for 40 minutes on a shaker at 37°C to loosen the basement membrane, and the epithelium was scraped off using a scalpel. The remaining limbal stroma was cut into anterior and posterior portions for A-LMSC and P-LMSC explant culture in Dulbecco's Modified Eagle Medium: Nutrient Mixture F-12 (DMEM/F12 1:1, 11320033; Thermo Fisher Scientific) added with 10% FBS and 1% antibiotics-antimycotics. The conditioned medium derived from A-LMSCs and P-LMSCs was obtained from the cultured cells at 60% to 80% confluence every 24 hours. The conditioned medium was then centrifuged at 2000 revolutions per minute (rpm) for 10 minutes, and the supernatant was collected for organ culture of mouse corneas. 

The immortalized human corneal endothelial cell line HCEC-B4G12¹⁹ (from Dr. Li Wen, Sydney Eye Hospital) was used as an experimental control for droplet digital polymerase chain reaction (ddPCR) and Western blotting. B4G12 cells were seeded onto plates pre-coated with 10 µg/mL laminin (L2020; Sigma-Aldrich) and 10 mg/mL chondroitin sulfate, and cultured in Human Endothelial Serum Free Medium (11111044; Thermo Fisher Scientific) supplemented with 2% FBS, 10 ng/mL FGF, and 1% antibiotics-antimycotics. 

Transwell Coculture

TZ cells were cocultured with A-LMSCs or P-LMSCs in a Transwell system, separated by a permeable membrane. The A-LMSCs or P-LMSCs at passage 1 (P1) were first seeded in the 12-well Transwell inserts (3460; Corning, Corning, NY, USA) at a density of 1 × 10⁴/cm². After the A-LMSCs or P-LMSCs reached confluency, P1 TZ cells were seeded into the bottom of the wells, some with coverslips, coated with fibronectin and collagen at a density of 0.2 × 10⁴/cm². The cocultures were maintained in TZ medium for 2 weeks. The 5-ethynyl-2′-deoxyuridine (EdU) assay and scratch wound assay were conducted during the coculture process. The cocultured TZ cells after 2 weeks were further harvested for colony forming assay, ddPCR, Western blotting, and immunocytochemistry. 

EdU Assay

The proliferation of TZ cells was evaluated by the EdU Imaging Kit (C10639; Thermo Fisher Scientific). The P1 TZ cells were seeded on fibronectin and collagen coated coverslips and cocultured with A-LMSCs or P-LMSCs for 3 days. TZ cells without coculture were used as the control group (n = 3 wells in each group). The TZ cells were then cultured in 10 µM EdU labeling solution for 6 hours and fixed with 4% paraformaldehyde (PFA) for 20 minutes. Subsequently, the cells were incubated with 0.5% Triton X-100 for 20 minutes to permeabilize, Click-iT reaction cocktail for 30 minutes to detect EdU, and 1 µg/mL 4′,6-diamidino-2-phenylindole (DAPI; D9542; Sigma-Aldrich) for 10 minutes to label the nuclei. The coverslips were further mounted on glass slides and imaged under a Zeiss Colibri 7 fluorescence microscope. The percentages of proliferating TZ cells in the three groups were assessed using ImageJ software (version 1.45b, National Institutes of Health, Bethesda, MD, USA) and quantified by dividing the number of EdU positive cells by the total cell number as indicated by DAPI staining. 

Scratch Wound Assay

The wound healing ability of TZ cells was measured with the scratch wound assay. Once the TZ cells met confluency after coculturing with A-LMSCs or P-LMSCs in the Transwell system, a horizontal scratch using a sterile 1 mL pipette tip was performed in each well, with non-cocultured TZ cells as the control group (n = 3 wells in each group). Then, the detached cells were rinsed off using phosphate buffered saline (PBS). The remaining TZ cells were continuously cocultured with A-LMSCs or P-LMSCs in TZ medium. The scratching in each well was marked with a marker pen at the bottom of the plate to ensure the same area was imaged at all time points. The TZ cells continued to be cocultured with A-LMSCs or P-LMSCs and imaged at 0, 4, 8, 16, 20, 24, 30, and 44 hours after scratching. The wound area was assessed with ImageJ software. The relative wound area was quantified by dividing the remaining wound area by the initial wound area. 

Colony Forming Assay

After coculturing with A-LMSCs or P-LMSCs for 2 weeks, TZ cells were passaged onto a fibronectin and collagen-coated 6-well at 1000 cells per well and cultured for 12 days to form colonies. TZ cells without coculture were processed in the same procedure (n = 3 wells in each group). The colonies were fixed with 4% PFA for 20 minutes and stained with 1% cresyl violet (C5042; Sigma-Aldrich) for 1 hour. The colony forming efficiency was determined by calculating the ratio of the number of colonies to the seeded number of TZ cells. 

Droplet Digital Polymerase Chain Reaction

TZ cells cocultured with A-LMSCs or P-LMSCs were collected for ddPCR to compare the gene profile of stem cell and differentiated markers, with non-cocultured TZ cells and B4G12 cells as the control groups (n = 3 in B4G12, TZ, and P-TZ and n = 2 in A-TZ). The RNA was isolated by the Purelink RNA mini kit (12183020; Thermo Fisher Scientific). The extracted RNA was tested using the Tape Station (Agilent, Santa Clara, CA, USA) to evaluate the quantity and quality of isolated RNA, followed by the SPUD assay to examine the presence of PCR inhibitors.²⁰ After passing quality control, RNA was reverse-transcribed into cDNA using the SuperScript VILO cDNA Synthesis Kit (11754050; Thermo Fisher Scientific). The successful synthesis of cDNA was verified via PCR amplification of β-actin and agarose gel electrophoresis of the amplified products. Then, the cDNA samples were analyzed using ddPCR. Specifically, each PCR reaction was set up containing 1.1 µL of PrimeTime pre-designed gene expression assay (IDT, Coralville, IA, USA), 11 µL of ddPCR supermix for probes (no dUTP; Bio-Rad, Hercules, CA, USA), and 1.1 µL of cDNA (2 ng/µL) in a total volume of 22 µL. A “no template” control was included alongside the experimental samples for each target/reference gene. Droplet generation involved adding 20 µL of PCR reaction and 70 µL of droplet oil to the DG8 cartridges (Bio-Rad), with droplets formed using QX200 droplet generator (Bio-Rad). Then, 40 µL of the generated droplets were subjected to a C1000 Touch Thermal Cycler (Bio-Rad) for amplification: initial pre-denaturation at 95°C for 10 minutes, followed by 40 cycles of denaturation at 94°C for 30 seconds and annealing at 60°C for 60 seconds, extension at 98°C for 10 minutes, and cooling at 12°C for temporary storage. After amplification, the droplets were individually detected according to the fluorescence signal from each droplet using the QX200 droplet reader (Bio-Rad), and the concentration of positive droplets was analyzed using QuantaSoft analysis software (Bio-Rad). If the concentration numbers were higher than 5000, the cDNA sample was diluted and ddPCR was repeated. In cases where expression was undetectable, increased volumes of cDNA (up to 9.9 µL) were used for repeated reactions. The expression for each gene was normalized to the geometric mean of two most stable reference genes, selected from a set of seven, using the NormFinder algorithm.²¹ B4G12 cells were used as positive controls for CEC gene expression, but this group was not included in the 1-way ANOVA analysis. Detailed information on PrimeTime assays of target and reference genes is provided in Supplementary Table S1. 

Western Blotting

TZ cells in different treatment groups and B4G12 cells (n = 3 in each group) were lysed by radioimmunoprecipitation assay (RIPA) buffer added with protease inhibitor cocktail (04693159001; Roche, Basel, Switzerland) for protein extraction. Protein concentration was measured with the detergent compatible protein assay (5000116; Bio-Rad). A total of 30 µg of protein were then separated via electrophoresis on precast polyacrylamide stain-free gels (Bio-Rad) and subsequently transferred onto polyvinylidene difluoride membranes using a Bio-Rad Trans-Blot Turbo transfer system. The membranes were blocked with 5% trim milk for 1 hour and incubated with primary antibodies overnight at 4°C and secondary antibodies at room temperature for 2 hours. The target protein was detected using Pierce enhanced chemiluminescence (ECL) Plus Substrate (32132; Thermo Fisher Scientific), and the film was scanned by ChemiDoc MP imaging system (Bio-Rad). Band densitometry was analyzed using Image Lab software (Bio-Rad, version 6.1). Relative protein level was calculated as the integrated density of the protein band divided by the integrated density of housekeeping marker α-tubulin on the same blot. The full-length Western blots of target and housekeeping proteins with the ladder are provided in Supplementary Figures S1 to S7. 

Immunofluorescence Staining

Immunofluorescence staining was performed to characterize P-LMSCs at P1 and compare the stem cell protein expression in TZ cells. Specifically, the cells were first fixed with 4% PFA for 15 minutes, permeabilized with 2% Triton X-100 and 10% goat/horse serum in PBS for 60 minutes, and incubated with primary antibodies overnight at 4°C and secondary antibodies for 2 hours at room temperature, with 3 times of washing using PBS between each step. The nuclei were labeled by DAPI, and the slides were mounted with Citifluor antifadent solution (Electron Microscopy Sciences, Hatfield, PA, USA). Images were taken under a Zeiss Colibri 7 fluorescence microscope. The details of antibodies are supplied in Supplementary Table S2. 

Organ Culture of Wounded Mouse Corneas

The mouse corneal tissues, with a small rim of sclera attached, were collected after the removal of connective tissue and conjunctiva from the eyeball. The wounding of mouse corneas was made by carefully scraping the entire endothelium using a silicone tube sheathed on an anterior chamber syringe. The corneas were stained with trypan blue (15250061; Thermo Fisher Scientific) for 2 minutes, rinsed in PBS, and imaged using a Zeiss Discovery V20 stereomicroscope. The wounded corneas were cultured in 3 different media: (1) medium control group: basal medium made of DMEM/F12 added with 10% FBS and 1% antibiotics-antimycotics; (2) A-LMSC CM group: conditioned medium from A-LMSCs mixed with the basal medium at a ratio of 1:1; and (3) P-LMSC CM group: conditioned medium from P-LMSCs mixed with the basal medium at a ratio of 1:1. The corneas were cultured for 2 weeks with media changed every 2 to 3 days. Trypan blue staining was repeated on days 4, 7, 10, and 14. The trypan blue staining area was quantified using ImageJ software. The relative wound area was determined by dividing the trypan blue staining area by the total area of the mouse cornea. 

On day 1 and day 6, the corneas were incubated in the media containing 10 µM EdU for 24 hours and the detection of EdU was performed as described previously. After fixing in 4% PFA for 1 hour, the corneal epithelium was scraped using a scalpel to avoid imaging the EdU positive epithelium in the transparent corneal tissue. All effort was made to remove the epithelium as thoroughly as possible, but some still remained, and they could be identified in images by being out-of-focus. On day 6, the corneal tissues with incorporated EdU were further incubated in ZO-1 and the corresponding secondary antibody, as mentioned above. The nuclei were stained with DAPI and the corneas were flat-mounted in Citifluor antifadent solution. Images were taken under a Zeiss Colibri 7 fluorescence microscope. 

Statistics

All data were shown as mean ± standard error (SE) and analyzed using 1-way ANOVA in SPSS Statistics 29.0 (IBM, Armonk, NY, USA). Homogeneity of variances was first tested between all the groups. When the data met the assumption of homogeneity of variances, Fisher's least significant difference (LSD) test was applied for post hoc pairwise comparisons, otherwise, Games-Howell test was used. P < 0.05 was considered statistically significant. 

Results

Characterization of P-LMSCs

To characterize the phenotype of P-LMSCs, we carried out explant culture, and assessed their morphology and protein expression in comparison to A-LMSCs. Results showed that A-LMSC and P-LMSC cultures could be established from anterior and posterior limbal stromal explants, respectively, and showed similar fibroblastic morphology for several passages (Fig. 1). Immunocytochemistry results demonstrated that P-LMSCs were positive for mesenchymal marker vimentin and stem cell markers Nestin, TRA-1-60, and Oct3/4, all recognized markers for the identification of A-LMSCs (Fig. 2). 

Figure 1.

 

Explant culture of A-LMSCs and P-LMSCs. The A-LMSCs and P-LMSCs showed similar fibroblastic morphology from P0 to P4. Scale bar = 100 µm.

Figure 2.

 

Immunocytochemistry of A-LMSCs and P-LMSCs demonstrating expression of mesenchymal marker vimentin (red) and stem cell markers Nestin (red), TRA-1-60 (red), and Oct3/4 (red) in A-LMSCs (row 1) and P-LMSCs (row 2). Blue = DAPI for nuclei. Inserts show the secondary antibody controls done in the same cell type as the immunohistochemically labeled cells. Scale bar = 50 µm.

Morphology of TZ Cells Cocultured With P-LMSCs

To investigate the effect of P-LMSCs on TZ cells, we performed P-LMSCs and TZ cell coculture (P-TZ) in a Transwell system, with A-LMSCs and TZ cell coculture (A-TZ) as a comparison group and TZ cells alone as a control group. Figure 3A showed TZ tissue dissected from the human donor corneal rim after corneal transplantation. The TZ explant included the peripheral endothelium, Schwalbe’s line, and the insert portion of the trabecular meshwork. Primary cultured TZ cells reached confluency in 20 days (see Fig. 3A). P1 TZ cells demonstrated similar elongated and fibroblastic morphology when coculturing with P-LMSCs, as well as in A-LMSC coculture and without coculture. After coculturing for 7 days, TZ cells reached confluency in P-LMSC and A-LMSC coculture groups, whereas there were some vacant areas in the TZ-only control group (Fig. 3B). 

Figure 3.

 

TZ explant culture and the morphology of TZ cells cocultured with A-LMSCs, P-LMSCs, and TZ cells alone. (A) The dissection of TZ tissue, the structure of the TZ explant, and P0 primary cultured TZ cells at 20 days. (B) The morphology of cells in TZ only control, TZ cocultured with A-LMSCs (A-TZ), and TZ cocultured with P-LMSCs (P-TZ) at days 0, 4, 7, and 14 after Transwell coculture. Scale bars = 500 µm.

Proliferation of TZ Cells Cocultured With P-LMSCs

To compare the proliferation of TZ cells cocultured with P-LMSCs, A-LMSCs, and without coculture, EdU assay was performed in the three groups. Results showed that a significantly higher proportion of nuclei was positive for EdU labeling in TZ cells after coculturing with P-LMSCs (44.38% ± 0.59%, P < 0.001) and A-LMSCs (36.02% ± 2.23%, P = 0.047) compared to TZ only control (24.11% ± 0.72%; Fig. 4). 

Figure 4.

 

EdU proliferation assay of TZ cocultured with P-LMSCs, A-LMSCs, or TZ cells alone. (A–C) EdU (red) incorporation in the TZ, A-TZ, and P-TZ groups. Blue = DAPI for nuclei. Scale bar = 100 µm. (D) The percentage of EdU positive cells in the TZ, A-TZ, and P-TZ groups, n = 3, ^*P < 0.05, ^**P < 0.01.

Wound Healing Capacity of TZ Cells Cocultured With P-LMSCs

A scratch wound assay was performed to compare the wound healing ability of TZ cells. TZ cells cocultured with P-LMSCs demonstrated the highest healing speed after wounding compared to the other two groups (Fig. 5A). Thirty hours after scratching, the average relative wound area was 4.39% ± 0.60% in the P-TZ group (P = 0.002 versus control) and 21.12% ± 3.03% in the A-TZ group (P > 0.05 versus control), compared to 32.36% ± 5.51% in the TZ only control group (Fig. 5B). 

Figure 5.

 

Scratch wound assay of TZ cells cocultured with P-LMSCs, A-LMSCs, and TZ cells alone. (A) Illustrative images of closing wounds in the TZ, A-TZ, and P-TZ groups at 0, 8, 16, and 24 hours after scratching. Scale bar = 100 µm. (B) The percentage of relative wound area in TZ, A-TZ, and P-TZ groups over time, n = 3, ^*P < 0.05, ^**P < 0.01 (versus TZ only control).

Stemness Properties of TZ Cells Cocultured With P-LMSCs

To compare the stemness of TZ cells cocultured with P-LMSCs, A-LMSCs, and without coculture, the colony forming assay was performed. The results of cresyl violet colony staining showed that TZ cells cocultured with P-LMSCs yielded more colonies compared to those without coculture (Fig. 6). Statistical analysis demonstrated a significantly higher colony forming efficiency in the P-TZ group (3.57% ± 0.24%) than the TZ control group (2.20% ± 0.31%, P = 0.007). The colony forming efficiency of the A-TZ group (3.30% ± 0.15%, P = 0.018) was also statistically higher than the control group. 

Figure 6.

 

Colony forming assay of TZ cells cocultured with P-LMSCs, A-LMSCs, and TZ cells alone. (A–C) Cresyl violet staining of TZ cell colonies in the TZ, A-TZ, and P-TZ groups. (D) Colony forming efficiency (CFE) in the TZ, A-TZ, and P-TZ groups, n = 3, ^*P < 0.05, ^**P < 0.01.

Gene Expression of TZ Cells Cocultured With P-LMSCs

Normfinder analysis identified that HPRT1 and POLR2A were the most stable reference gene pair in this dataset, therefore the expressions of target genes were normalized to the geometric mean of HPRT1 and POLR2A. The results of ddPCR demonstrated that TZ cells cocultured with P-LMSCs expressed significantly higher levels of pluripotency gene NANOG (0.0016 ± 0.0003 vs. 0.0007 ± 0.0002, P = 0.034), neural crest genes SOX9 (0.2326 ± 0.0048 vs. 0.1863 ± 0.0080, P = 0.019), and TFAP2A (0.7271 ± 0.0467 vs. 0.5601 ± 0.0509, P = 0.048), and periocular mesenchyme gene PITX2 (1.9485 ± 0.0350 vs. 1.5235 ± 0.1224, P = 0.019) than the TZ only control (Fig. 7). Moreover, the expression of neural crest gene NESTIN in TZ cells cocultured with P-LMSCs was significantly higher than those cocultured with A-LMSCs (1.4644 ± 0.0714 vs. 1.1010 ± 0.0347, P = 0.022) and showed an increasing trend compared to TZ cells without coculture (1.2286 ± 0.0829, P = 0.064). The corneal endothelial gene SLC4A11 in TZ cells was downregulated after coculturing with P-LMSCs (0.0220 ± 0.0003 vs. TZ control 0.0298 ± 0.0009, P = 0.015). However, the corneal endothelial gene AQP1 was upregulated after coculturing with A-LMSCs (1.3046 ± 0.0849 vs. TZ control 0.4168 ± 0.0512, P = 0.004) and P-LMSCs (0.9759 ± 0.1566, P = 0.015). Other corneal endothelial genes, such as COL8A1, TJP1, and ATP1A1, did not show significant differences among TZ cells with and without coculture. Overall, pluripotency genes exhibited very low expressions compared to neural crest, periocular mesenchyme, and CEC genes. 

Figure 7.

 

Relative mRNA expression of TZ cells cocultured with P-LMSCs, A-LMSCs, and TZ cells alone, with samples of B4G12 cells as CEC positive control. Target genes include pluripotency genes SOX2, NANOG, POU5F1, neural crest genes SOX10, NGFR, SOX9, TFAP2A, and NESTIN, periocular mesenchyme genes LMX1B, FOXC1, and PITX2, and corneal endothelial genes SLC4A11, COL8A2, CDH2, GPC4, AQP1, COL8A1, TJP1, and ATP1A1. n = 3 in B4G12, TZ and P-TZ, n = 2 in A-TZ, ^*P < 0.05, ^**P < 0.01, ^#P < 0.1.

Protein Expression of TZ Cells Cocultured With P-LMSCs

Western blot results demonstrated that TZ cells cocultured with P-LMSCs showed significantly higher expressions of neural crest markers Nestin (P < 0.001), Sox9 (P = 0.033), AP-2α (P = 0.035), and periocular mesenchyme markers FoxC1 (P = 0.031), Pitx2 (P = 0.039), and an increasing trend of neural crest marker Sox10 (P = 0.055) than TZ cells alone. Expression levels of Nestin (P < 0.001), Sox9 (P = 0.026), AP-2α (P = 0.032), and Sox10 (P = 0.041) were statistically higher in TZ cells cocultured with P-LMSCs compared to TZ cocultured with A-LMSCs (Figs. 8A, 8B). Moreover, CEC markers Na⁺/K⁺ ATPase, ZO-1, and Col8A1 exhibited similar expressions among TZ cells with different treatments and B4G12 cells (Fig. 8C). Immunocytochemistry confirmed higher levels of Nestin, Sox9, and Pitx2 in TZ cells cocultured with P-LMSCs than TZ cells in the other two groups (Fig. 9). 

Figure 8.

 

Relative protein expression of TZ cells cocultured with P-LMSCs, A-LMSCs, and TZ cells alone, with samples of B4G12 cells as CEC positive control. (A) Neural crest markers Nestin, Sox9, AP-2α, and Sox10. (B) Periocular mesenchyme markers FoxC1 and Pitx2. (C) CEC markers Na⁺/K⁺ ATPase, ZO-1, and Col8A1. The relative protein level of each target marker was normalized to housekeeping marker α-tubulin on the same blot, n = 3, ^*P < 0.05, ^**P < 0.01, ^#P < 0.1.

Figure 9.

 

Immunocytochemistry of neural crest markers Nestin (green), Sox9 (red), and periocular mesenchyme marker Pitx2 (green) in TZ cells alone (column 1), after coculturing with A-LMSCs (A-TZ, column 2) and after coculturing with P-LMSCs (P-TZ, column 3). Blue = DAPI for nuclei. Inserts show the secondary antibody controls done in the same cell type as the immunohistochemically labeled cells. Scale bar = 100 µm.

P-LMSC Conditioned Medium Promotes the Regeneration of TZ Cells in Organ-Cultured Mouse Corneas after Endothelial Scraping

To test whether P-LMSCs could promote the regeneration of TZ cells in wounded cornea tissues, conditioned medium (CM) from cultured P-LMSCs was added onto organ-cultured mouse corneas after entire endothelial scraping. On day 7, the relative wound area was significantly decreased in the P-LMSC CM group (average relative wound area 59.40% ± 2.75%), compared to the A-LMSC CM (75.97% ± 1.19%, P < 0.001) and basal medium control (84.82% ± 1.96%, P < 0.001). Over 14 days of culture, the wounded corneas in P-LMSC CM (relative wound area 31.1% ± 5.49%, P < 0.001, vs. control) and A-LMSC CM (41.79% ± 8.78%, P = 0.005, vs. control) demonstrated superior regeneration of corneal endothelium compared to the basal medium control (72.18% ± 4.66%). Generally, the regeneration of the corneal endothelium originated from the peripheral endothelium - TZ region (Fig. 10). 

Figure 10.

 

P-LMSC conditioned medium (CM) promotes the regeneration of corneal endothelium from the TZ region in organ-cultured mouse corneas compared to A-LMSC CM and basal medium control. (A) Trypan blue imaging of corneas from 0 to 14 days after entire endothelial scraping in the control, A-LMSC CM, and P-LMSC CM groups. (B) The relative wound area of corneas in the control, A-LMSC CM, and P-LMSC CM groups over time, n = 6 each, ^*P < 0.05, ^**P < 0.01 (versus control).

Regeneration of corneal endothelium from the TZ region via proliferation was further confirmed by EdU and ZO-1 labeling. After entire endothelial scraping at 2 days, the EdU incorporation was sparsely detected in the TZ region of the control corneas, but obviously observed in A-LMSC CM and P-LMSC CM groups (Fig. 11), supporting the contribution of proliferation for the corneal endothelial recovery. 

Figure 11.

 

EdU incorporation in organ-cultured mouse corneas at 2 days after endothelial wounding in the medium control, A-LMSC CM, and P-LMSC CM. The bottom row images show the details of the rectangles in the top row images. White dotted lines delineate borders between transition zone (TZ) and peripheral endothelium (PE) in the corneal flat mounts. The stars label out-of-focus EdU labeling from unscraped corneal epithelium. All effort was made to remove the epithelium before imaging, but some remnants still remained. Green = EdU incorporation. Blue = DAPI for nuclei. Scale bars = 100 µm.

On day 7, EdU was substantially incorporated in the TZ region and periphery endothelium of the corneas in P-LMSC CM compared to those in A-LMSC CM and basal medium control (Fig. 12A). Furthermore, ZO-1 staining demonstrated the reformation of a contiguous sheet of CECs with tight junctions in periphery endothelium of the corneas cultured in P-LMSC CM, whereas corneas in A-LMSC CM and basal medium control only showed sparse expression of ZO-1 in TZ region (see Figs. 12A, 12B). 

Figure 12.

 

EdU and ZO-1 staining of organ-cultured mouse corneas at 7 days after endothelial wounding in the basal medium control, A-LMSC CM, and P-LMSC CM. (A) Separate channels of EdU and ZO-1 labeling of corneal flat mounts at low magnification. (B) Composite EdU and ZO-1 labeling of the rectangles in A. The white dotted line delineates the border between transition zone (TZ) and peripheral endothelium (PE). The stars label out-of-focus EdU labeling from remnant epithelium. Green = EdU incorporation. Red = ZO-1 showing tight junctions. Blue = DAPI for nuclei. Scale bars = 100 µm.

Discussion

This study revealed that P-LMSCs expressed similar proteins as A-LMSCs and demonstrated significantly superior stimulation on the proliferation and stemness of TZ cells in both cell and organ culture models. 

P-LMSCs expressed the same recognized markers for A-LMSCs, including mesenchymal marker vimentin and stem cell markers Nestin, TRA-1-60, and Oct3/4.^22,23 Expression of these markers in A-LMSCs has been demonstrated to be essential for them to prevent differentiation and maintain stemness of limbal epithelial stem cells.²³ This suggests that P-LMSCs might have similar supporting potential for the surrounding stem cells. 

TZ cells cocultured with P-LMSCs demonstrated more EdU incorporation and higher wound healing speed. These results show that P-LMSCs can support the proliferation and wound healing ability of TZ cells, both critical properties for endogenous corneal endothelial regeneration under pathologic conditions and in response to injury.²⁴ 

Stemness, another essential property for tissue repair and regeneration,²⁵ was compared in TZ cells with different treatments. Our results demonstrated higher colony forming efficiency in TZ cells supported by P-LMSCs than in TZ cells only. The upregulation of pluripotency gene NANOG, neural crest genes SOX9 and TFAP2A, and periocular mesenchyme gene PITX2 in TZ cells after coculturing with P-LMSCs suggest that the expression of stem cell genes was promoted by P-LMSCs. Western blotting and immunocytochemistry results confirmed increased protein levels of neural crest markers Nestin, Sox9, and AP-2α, and periocular mesenchyme markers FoxC1 and Pitx2 in TZ cells with the support of P-LMSCs. These results collectively suggest that P-LMSCs can stimulate the stemness of TZ cells. 

Previous studies demonstrated the support effect on corneal endothelium by conditioned media from multiple types of MSCs,^26–28 and MSCs from sources closer to the CECs may be superior to those farther away.²⁹ Our findings support that P-LMSCs, rather than A-LMSCs, seemed to be more potent in supporting the stemness of TZ cells, perhaps due to their close anatomical association with TZ cells. The results suggest that P-LMSCs might compose a more favorable microenvironment for the self-renewal and stemness of TZ cells. 

The differentiated properties of TZ cells appeared to be maintained by LMSCs. Herein, we observed similar gene and protein levels of Na⁺/K⁺ ATPase, ZO-1, and Col8A1, suggesting little alteration of corneal endothelial marker expression, and perhaps function, in TZ cells supported by P-LMSCs and A-LMSCs. Interestingly, the gene expression of AQP1 in TZ cells was upregulated by both A-LMSC and P-LMSCs, indicating the support of fluid transport function by LMSCs. Although SLC4A11, encoding solute carrier family 4 member 11, was downregulated in TZ cells after coculturing with P-LMSCs, the expression level of this gene was relatively low compared to other corneal endothelial genes, which might limit its influence on corneal endothelial function. 

By using an immortalized human corneal endothelial cell line, B4G12, as a positive control for corneal endothelial markers,³⁰ we demonstrated great potential applications of TZ cells in treating corneal endothelial diseases. We observed similar protein expressions of Na⁺/K⁺ ATPase, ZO-1, and Col8A1 between B4G12 cells and all TZ cells, suggesting that TZ cells have the capacity to function as normal CECs. We also aimed to use B4G12 cells as a negative control for neural crest and periocular mesenchyme markers. Unexpectedly, B4G12 cells showed protein expressions for AP-2α, Sox10, and Pitx2. This might be attributed to the common problem of genetic or phenotype drift for immortalized cell lines.^31,32 Further experiments, such as trans-endothelial electrical resistance measurement and in vivo transplantation, are needed to test the corneal endothelial function of TZ cells. 

In this study, we provided further evidence supporting TZ cells are the stem cell/progenitor source of corneal endothelium. We observed the regeneration of corneal endothelium from the TZ region after scraping the entire endothelium in mouse corneas, remarkably enhanced by conditioned medium from P-LMSCs. The contribution of proliferation in this process was confirmed by EdU incorporation. Moreover, the reformation of ZO-1 tight junctions within the regenerated endothelium indicated that the intact barrier function was restored by the secretome released by P-LMSCs. These findings were consistent with previous studies which showed that stem cells residing in the TZ region responded to corneal injury to initiate an endothelial repair process.^13,33 Our results further suggest that, with the stimulation of P-LMSC secretome, TZ cells demonstrate stronger regeneration capacity to repair the wounded corneal endothelium and restore natural function. 

The process of corneal endothelial regeneration has not been fully understood. A previous study demonstrated dynamic corneal endothelial–posterior stromal communication traversing Descemet's membrane exclusively in the peripheral area of mature murine and human corneas,³⁴ indicating that CECs might be directly influenced by mesenchymal cells in the posterior limbus. A recent study identified elastic fibers, composed of bundles of microfibrils with or without an elastin core, exclusively within the posterior peripheral corneal stroma of human early embryonic corneas, and absent throughout the rest of the cornea.³⁵ These fibers were found directly anterior to the corneal endothelium, and CEC projections were also found branching toward the mesenchymal cells, supporting the existence of local communication between CECs and adjacent posterior mesenchymal cells. Moreover, a prior study showed increased density and proliferation capacity of human corneal endothelium after stimulating stromal secretion of interleukin-1β.³⁶ Likewise, another study showed that mechanical injury to the corneal endothelium triggered apoptosis of 25% of the posterior stromal cells overlying the site of endothelial injury.³⁷ These studies collectively suggest interactions between the corneal endothelium and the posterior stroma, and potential niche support from P-LMSCs to TZ cells. 

However, to delineate the stem cell niche in the posterior limbus, further studies identifying the secreted factors, cell adhesion, and the precise cellular organization between TZ cells and P-LMSCs are needed.³⁸ Expanding the sample size in future studies could strengthen the robustness of our findings. Nevertheless, our research points to a potential stem cell environment in the posterior limbus, which opens up opportunities for further investigation, including the molecular pathways involved in this crosstalk. 

Conclusions

Our results suggest that P-LMSCs can stimulate the proliferation, wound healing ability, and stemness of TZ cells in both cell and organ culture models. This supports our hypothesis that TZ cells derive critical support from the posterior limbus, where P-LMSCs play an essential role in these cellular interactions. It will help unravel the underlying mechanisms that govern the cell fate of progenitors for corneal endothelium, in order to explore pharmacological interventions towards cell fate manipulation to enhance endogenous regeneration of corneal endothelium. 

Acknowledgments

The authors thank the New Zealand National Eye Bank for supplying donor corneoscleral rims and Professor Trevor Sherwin for obtaining ethical approval for the use of donor corneas in this study. We would also like to acknowledge Li Wen and Associate Professor Con Petsoglou, Sydney Eye Hospital, for the B4G12 cell line, and Monika Valtink, Technical University of Dresden, for agreeing to the cell line transfer. 

Supported by the New Zealand Save Sight Society Grant. 

Disclosure: Y. Xiao, None; C.N.J. McGhee, None; J. Zhang, None