Prox1 Protein in Corneal Limbal Lymphatic Vessels Maintains Limbal Stem Cell Stemness and Regulates Corneal Injury Repair

Abstract

Purpose

The purpose of this study was to elucidate the role of the corneal lymphatic Prox1 gene in modulating limbal stem cell stemness and facilitating corneal injury repair.

Methods

The limbal Prox1 gene was knocked down by adeno-associated virus (AAV). The alkali burn model was induced in the naive group, the AAV-sham group, and the AAV-shProx1 group. Anterior segment photography, fluorescein sodium staining, and hematoxylin and eosin (H&E) staining were conducted immediately on days 1, 3, 7, 14, and 21 post-injury. Immunofluorescent (IF) staining was used to assess Ki67, ΔNp63, and K14. Additionally, seq-mRNA technology facilitated a comparative transcriptomic analysis between the AAV-sham and the AAV-shProx1 groups 7 days post-injury. Key regulated genes were verified by protein level. Furthermore, a co-culture model of lymphatic endothelial cells (LECs) and limbal stem cells (LSCs) was used to investigate the proliferation capacity and stemness expression of LSCs.

Results

Fluorescein sodium staining revealed that the epithelial defect area was significantly larger in the AAV-shProx1 group than in the AAV-sham group on days 1 and 3 post-injury (P < 0.05). Ki67, ΔNp63, and K14 expressions were consistently lower in the AAV-shProx1 group than in the AAV-sham group at distinct time points. Additionally, seq-mRNA results demonstrated that genes (Prox1 and Lyve1) were downregulated while inflammatory factors (Ccl2, Ccl7, IL16, IL1R, and TNFsf11) were upregulated in the AAV-shProx1 group compared with the AAV-sham group. When Prox1 was silenced in LECs, the proliferation and stemness of LSCs were markedly downregulated.

Conclusions

The Prox1 and Lyve1 proteins in lymphatic vessels served as pivotal regulated proteins in corneal injury repair. The draining role of lymphatic vessels during corneal injury was indispensable.

The limbal microenvironment is a complex regulatory network where cells, extracellular matrix, and intercellular signaling mutually influence each other to regulate the orderly regeneration and aging of corneal epithelium under physiological conditions.¹ When corneal epithelial injury occurs, the regenerative potential of limbal stem cells (LSCs) is activated. However, if the injury surpasses the regenerative capacity, it leads to persistent corneal epithelial defects and limbal stem cell deficiency (LSCD).² Given the intricate regulatory network within the limbus, identifying and validating effective targets to promote LSCs stemness and proliferative potential offers promising prospects for LSC regeneration. In recent years, single cell sequencing techniques have been utilized to identify several previously unrecognized rare cell populations in mouse and human limbal tissues, including lymphatic endothelial cells (LECs).^3–5 This suggests that the role of LECs in the limbal microenvironment has not yet been fully understood.

Limbal lymphatic vessels play a significant role in post-transplant immune rejection, aqueous humor drainage, and stem cell maintenance.^6,7 Studies have indicated that donor and host antigens are present at the junction between donor grafts and recipient beds following corneal transplantation. These antigens are transferred to lymph nodes, where they activate effector T cells, leading to the production of antibodies that return to the graft-bed junction via venous circulation, thus triggering transplant rejection.⁸ Selective blockage of corneal lymphatic vessels can prolong graft survival. Furthermore, researchers have confirmed the existence of a Prox1⁺Lyve1⁺ connection channel between limbal lymphatic vessels and Schlemm's canal, which plays a role in aqueous humor drainage through a mechanism distinct from traditional pathways.^9–11 Researchers have also found that LSCs adjacent to lymphatic vessels are activated following hair follicle injury, promoting hair follicle stem cell repair.¹² In immunodeficient mice with reduced numbers of lymphatic vessels and lymphocytes, the recovery of corneal epithelium post-injury was slower compared to the control group, suggesting that lymphatic vessels may play a crucial role in LSCs maintenance and injury repair.⁶

However, the interaction between limbal lymphatic vessels and stem cells remains insufficiently explained. Prox1 is a transcription factor critical for the proliferation, differentiation, and sprouting of lymphatic vessels, regulating various essential activities of lymphatic endothelial cells.^13–15 Studies have shown that knocking down Prox1 has downregulated cardiac lymphatic vessels in embryonic mice, leading to delayed heart growth and abnormal myocardial fibers.⁷ Moreover, Prox1 plays an essential role in the differentiation of neural stem/progenitor cells into neurons.¹⁶ These studies suggest that the Prox1 gene is vital for stem cell proliferation and differentiation. The present study aims to investigate whether Prox1 modulates the stemness maintenance of LSCs and the repair of corneal epithelial injuries through its regulatory influence on lymphatic vessel activities.

Methods

Establishment of the Alkali Burn Model

Male C57BL/6 mice aged 6 to 8 weeks old, sourced from SPF Biotechnology (Beijing, China), were used for this study. The animal experiment protocol was reviewed and approved by the Institutional Animal Care and Use Committee (IACUC) of PLA General Hospital (Approval No: 2020-X16-83). All experimental procedures were conducted in strict accordance with the approved guidelines. Additionally, this study adhered to the Association for Research in Vision and Ophthalmology (ARVO) statement for the use of animals in ophthalmic and vision research. The mice were anesthetized with 1% pentobarbital sodium (50 mg/kg IP), ensuring deep anesthesia before operating. The right eye of each mouse was subjected to alkali burn injury, whereas the left eye served as a control. Topical anesthesia was administered using proparacaine eye drops to numb the ocular surface. A 2 mm diameter filter paper disc was soaked in 1 µl of NaOH (0.5 M) and carefully placed on the central cornea for 30 seconds using fine forceps, ensuring firm contact with the corneal surface. Following this, the ocular surface was thoroughly rinsed with physiological saline for 1 minute using a 50 mL syringe, avoiding the mouth and nose of the mouse. The treated eye was then coated with gatifloxacin ophthalmic gel, and the mice were monitored until full recovery to ensure there were no complications from the procedure.

Subconjunctival Injection of Adeno-Associated Virus

The adeno-associated virus (AAV) vectors (serotype 2/9, U6 promoter) were obtained from Hanheng Biotechnology (Shanghai, China), with vector details provided in Supplementary Figure S3 and target/control sequences in Table 1. The vectors were gradually thawed on ice prior to use. Mice underwent pre-procedural screening to exclude ocular abnormalities. Systemic and topical anesthesia were induced. Microliter syringes were flushed with 75% ethanol to ensure sterility, followed by phosphate-buffered saline (PBS). A 5 µL syringe (Hamilton, Switzerland) was loaded with 4.5 µL of either the AAV-shProx1 vector or AAV-sham vector. Under microscopic guidance, the nasal conjunctiva was gently lifted using fine forceps, and 4.5 µL of the solution was injected subconjunctivally, followed by the same procedure for the temporal conjunctiva. The viral titers were 1.6 × 10¹² vg/mL for the target gene and 1.3 × 10¹² vg/mL for the control, with a total injection volume of 9 µL per eye, ensuring precise and reproducible delivery. Post-injection, gatifloxacin ophthalmic gel was applied to prevent infection.

Table 1.

AAV-shProx1 Sequence and siProx1 Sequence

	Top Strand	Bottom Strand
AAV-shProx1	AATTCGCAGGAGGATATATGTAGCAACTCTTCTCGAGAAGAGTTGCTACATATATCCTCCTGTTTTTTG	GATCCAAAAAACAGGAGGATATATGTAGCAACTCTTCTCGAGAAGAGTTGCTACATATATCCTCCTGCG
AAV-sham	GATCCGTTCTCCGAACGTGTCACGTAATTCAAGAGATTACGTGACACGTTCGGAGAATTTTTTC	AATTGAAAAAATTCTCCGAACGTGTCACGTAATCTCTTGAATTACGTGACACGTTCGGAGAACG
siProx1	CGAGAAAGUUACAGAGAAATT	UUUCUCUGUAACUUUCUCGTT
siNC	UUCUCCGAACGUGUCACGUTT	ACGUGACACGUUCGGAGAATT

Anterior Segment Photography and Fluorescein Sodium Staining

Corneal opacity was assessed using a slit-lamp microscope (Optoprobe, England), utilizing a standardized grading system to evaluate the extent of corneal opacity.¹⁷ Corneal opacity was graded on a scale from 0 (completely clear) to 4 (completely opaque), 1 = slightly hazy, and the iris and pupil were easily visible; 2 = slightly opaque, and the iris and pupil were still detectable; 3 = opaque, and the pupils were hardly detectable; and 4 = completely opaque with no view of the pupil. Corneal epithelial defects were assessed using 2% fluorescein sodium staining through cobalt blue light. Image J software (NIH, USA) was used to analyze and calculate the areas of epithelial defects. This assessment was conducted blindly by two experienced ophthalmologists. The severity of corneal epithelial defects was documented accordingly.

Hematoxylin and Eosin Staining

The mice were ethically euthanized at selected time points, and the eyes were fixed in 4% paraformaldehyde (PFA) and embedded in paraffin. Serial sections of 4-µm thickness were cut and stained with hematoxylin and eosin (H&E). The staining process involved several steps: deparaffinization with 2 changes of xylene (10 minutes and 5 minutes), followed by 100% ethanol for 15 minutes, and 95% ethanol for 5 minutes, and then rinsing with running tap water multiple times. Nuclei were stained with hematoxylin for 4 minutes, differentiated in 1% hydrochloric acid ethanol for a few seconds, and blued in ammonia water for a few seconds, followed by a rinse. The sections were then stained with eosin for 5 minutes, rinsed with water, dehydrated through graded ethanol solutions (85% ethanol for a few seconds and 100% ethanol for 5 minutes), cleared in xylene for 5 minutes, and finally mounted with a resinous medium. The slides were scanned using a panoramic microscope (Olympus, Japan) for detailed imaging after the slides were dried.

Corneal Whole Mount Immunofluorescence

Corneas were excised from the euthanized mice, ensuring that a portion of the limbus was retained while avoiding the iris, and then placed in a 48-well plate containing 4% PFA overnight. The PFA was discarded the following day, and the corneas were washed in PBS for 10 minutes, 3 times. In a 96-well plate, the corneas were soaked in 200 µL per well of a mixture containing 0.2% Triton X-100 and blocking goat serum for 2 hours with gentle shaking, followed by incubation with the primary antibody (1:200) at 4°C overnight. On the third day, the corneas were transferred to a 48-well plate and washed again in PBS for 10 minutes, 3 times, and then incubated with a diluted secondary antibody (1:400) at room temperature for 2 hours in the dark. After another set of PBS washes (10 minutes, 3 times), a drop of DAPI was added to a slide, and each cornea was placed concave side down onto the DAPI. The corneas were then cut into four or six equal sections, covered with a coverslip, and sealed with nail polish. Finally, images were captured using a high-content microscopy system (PerkinElmer, Germany) and a spinning disk confocal microscope (Andor, England).

Frozen Section Immunofluorescence

Frozen sections were retrieved from −80°C storage and immediately fixed with 4% PFA for 10 minutes at room temperature. Sections were washed in PBS 5 minutes, 3 times, permeabilized with 0.2% Triton X-100 for 5 minutes, and blocked with goat serum for 30 minutes. Primary antibodies were applied and incubated overnight at 4°C. The next day, following washing, secondary antibodies were incubated for 2 hours at room temperature in the dark. Following washing with PBS for 5 minutes, 3 times, the sections were counterstained with DAPI and mounted with nail polish to secure the cover slip. FV3000 confocal microscopy (Olympus, Japan) was used to observe and document the staining patterns. The pre-configured script file was used to convert each image file to an 8-bit image, perform channel splitting, and process a designated channel with a predefined dark threshold to calculate the mean fluorescence intensity. Key metrics, such as area, mean intensity, minimum, and maximum values were computed and recorded in a file. The script systematically appended results and closed processed images, streamlining quantitative image analysis.

Cell Lines

Two distinct cell lines were used in the present research. The murine limbal epithelium-derived progenitor cell line, TKE2, was generously provided by Dr. Schaffer C.G. Tseng from the Ocular Surface Center in Florida.¹⁸ The cells were cultured in defined keratinocyte serum-free medium (DKSFM; Gibco), supplemented with 1% Human Keratinocyte Growth Supplement (HKGS; Gibco), 5 ng/mL recombinant human epidermal growth factor (EGF; Gibco), and 1% penicillin-streptomycin solution (Gibco). Human LECs were purchased from Shanghai Anwei Cell Company and cultured using the primary endothelial cell culture system (primed-icell-002, icell). Both cell lines were cultured in a humidified incubator with 5% CO₂ and 95% air at 37°C.

Edu Assay

Edu assays were performed using the Beyotime Edu kit (C0075L). To prepare a 2 × Edu working solution (20 µM), 10 mM Edu was diluted at a 1:500 ratio in a cell culture medium. Half of the medium in each well of the 6-well plate (1 mL) was discarded, and 1 mL of the pre-warmed (37°C) 2 × Edu working solution (20 µM) was added. The cells were incubated for 2 hours. After Edu labeling, the culture medium was removed, and 1 mL of fixative was added. The cells were fixed at room temperature for 15 minutes. The fixative was removed, and the cells were washed in each well with 1 mL of wash buffer for 5 minutes, 3 times. The wash buffer was removed, 1 mL of permeabilization solution was added per well, and the cells were incubated at room temperature for 15 minutes. The permeabilization solution was removed, and the cells were rewashed with 1 mL of wash buffer for 5 minutes, 3 times. The Click Reaction Solution (Click Reaction Buffer, CuSO₄, Azide 555, and Click Additive Solution) was prepared according to the kit instructions, and the cells were incubated at room temperature in the dark for 30 minutes. Afterward, the Click Reaction Solution was removed, and the cells were washed with wash buffer for 5 minutes, 3 times. Finally, the cell nuclei were stained with DAPI.

Cell Scratch Assay

In the cell scratch assay, a 6-well plate was pre-marked on the underside with evenly spaced horizontal reference lines, with at least 3 lines per well to guide precise scratch placement. Cells were seeded at a density of 5 × 10⁵ cells per well (or adjusted accordingly based on cell growth rate) to achieve 100% confluence following overnight incubation. The next day, a sterile pipette tip, held perpendicular to the cell monolayer, was used to create linear scratches aligned with the pre-marked reference lines. After scratching, the wells were gently washed twice with sterile PBS to remove detached cells and debris, leaving well-defined and visible wound gaps. The medium was then replaced with serum-free or conditioned medium to promote cell migration. The plate was incubated at 37°C in a 5% CO₂ atmosphere, and at designated time points, the progression of wound closure was monitored via light microscopy (Nikon, Japan).

SiRNA Transfection

The siRNA was procured from Hanheng Biotechnology (Shanghai, China), with the precise sequences detailed in Table 1. The siRNA was centrifuged at 1000 rpm for 1 minute, and then 125 µL of RNase-free water was added to prepare a 20 µM solution. A cell confluence of approximately 50% was ensured for transfection. For a hole of 6-well plate, 250 µL of Opti-MEM plus 5.0 µL of Lipofectamine RNAiMAX (Invitrogen, USA) was prepared and was allowed to stand for 5 minutes. A 100 nM concentration was used for the transfection, specifically by preparing 250 µL of Opti-MEM plus 10.0 µL of siRNA. The Lipofectamine RNAiMAX and siRNA mixture was allowed to stand at room temperature for 20 minutes. The original medium was replaced with fresh medium, and the above mixture was added. SiRNA knockdown efficiency was verified by RT-qPCR and Western blot 48 hours after transfection.

RNA Sequencing

Total RNA was isolated using the SMART-Seq HT Kit. Paired-end libraries were prepared with the TruSeq RNA Sample Preparation Kit (Illumina, USA). Briefly, poly-A mRNA was purified with poly-T magnetic beads, fragmented at 94°C for 8 minutes, and reverse transcribed into cDNA. Second strand cDNA was synthesized, end-repaired, A-tailed, and ligated with adapters. The cDNA library was purified and enriched by PCR. The libraries were quantified using a Qubit 2.0 Fluorometer (Life Technologies, USA) and validated with an Agilent 2100 Bioanalyzer (Agilent Technologies, USA). Cluster generation was done with cBot at 10 pM, and sequencing was performed on the Illumina HiSeq Xten (Illumina, USA). Differentially expressed genes were identified using the DESeq2 R package, with criteria set at a log2 fold-change of ≥ 1 and a false discovery rate of < 0.05. Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) biological process enrichment analysis were performed using the cluster Profiler R package, applying a P value cutoff of 0.01 and a q value cutoff of 0.05 to identify significant pathways.

Western Blot

The tissue was minced with scissors, and 30 µL of RIPA buffer (Beyotime Biotechnology, China) was added. The mixture was homogenized using an ultrasonic processor (Sonics, USA) at 50% power (5 seconds on and then 5 seconds off) until entirely disrupted. The homogenate was incubated on ice for 30 minutes. It was then centrifuged at 12,000 rpm for 30 minutes at 4°C, and the supernatant was transferred to a clean 1.5 mL tube. Protein concentration was measured using the BCA Kit (ThermoFisher, USA). Protein concentrations were adjusted to 1.5 to 3 µg/µL with RIPA buffer and 6 × SDS loading buffer (Beyotime Biotechnology, China). The samples were heated at 95°C for 15 minutes and were cooled on ice for 10 minutes. They were then loaded onto a 10% polyacrylamide gel (Bio-Rad, USA) and were run at 200 V until the proteins were separated. The proteins were transferred to a PVDF membrane and were blocked with 5% non-fat milk for 2 hours. The membrane was incubated with primary antibody in 2 × casein at 4°C for 12 to 36 hours. It was washed with TBST for 10 minutes, 3 times. The membrane was incubated with a secondary antibody in 2 × casein at room temperature for 2 hours. It was rewashed with TBST for 10 minutes, 3 times. Finally, protein bands were detected using a chemiluminescence imaging system (Bio-Rad, USA). The information of the primary antibodies used in the present research are provided in Table 2.

Table 2.

Antibodies Used in the Present Research

Antibody	Source	Identifier
Prox1	Abcam	Ab199359
Lyve1	Thermo Fisher Scientific	14-0443-82
K12	Abcam	Ab185627
ΔNp63	Proteintech	12143-1-AP
K14	Proteintech	10143-1-AP
Ki67	Abcam	Ab16667
IL-1β	Abcam	Ab283818
TNF-α	Abcam	Ab183218
Ccl2	Thermo Fisher Scientific	MA5-17040
CD68	Abcam	Ab303565
CD11c	Cell Signaling Technology	97585
β-actin	Proteintech	66009-1-lg

Quantitative Real-Time PCR

To isolate RNA, 1 mL of Trizol was added into 2 corneas and incubated at room temperature for 5 minutes, an ultrasonic processor (Sonics, USA) was used to disrupt the tissue entirely. After adding 200 µL of chloroform and mixing, the mixture was centrifuged at 12,000 rpm for 15 minutes at 4°C. The aqueous phase was transferred to a new tube, mixed with isopropanol, and centrifuged again. The resulting pellet was washed twice with 75% ethanol, air-dried, and resuspended in 10 µL of DEPC water for RNA concentration measurement and storage. RNA quality was assessed by measuring the absorbance at OD260 nm and OD280 nm, ensuring A260/A280 ratios between 1.9 and 2.1. The cDNA was synthesized using the PrimeScript RT reagent kit (Bio-Rad, USA). RT-qPCR was performed using SYBR Green PCR Master Mix (Bio-Rad, USA) on an ABI Prism 7000 instrument. Data were presented as mean ± SEM from three independent experiments. Primer sequences are detailed in Table 3.

Table 3.

Primers Used for Reverse Transcriptase qPCR

Gene	Forward (5′-3′)	Reverse (3′-5′)
Ki67	ATCATTGACCGCTCCTTTAGGT	GCTCGCCTTGATGGTTCCT
K14	GAAGAACCGCAAGGATGCTGAG	TGCAGCTCGATCTCCAGGTTCT
K12	CGAGAGTGGTATGAAACACGGAG	CGTTGCTGACTCTGGCAGAAAC
ΔNp63	GTATCGGACAGCGCAAAGAACG	CTGGTAGGTACAGCAGCTCATC
IL-1β	TGGACCTTCCAGGATGAGGACA	GTTCATCTCGGAGCCTGTAGTG
TNF-α	GGTGCCTATGTCTCAGCCTCTT	GCCATAGAACTGATGAGAGGGAG
Prox1	AGAAGGGTTGACATTGGAGTGA	TGCGTGTTGCACCACAGAATA
GAPDH	AGGTCGGTGTGAACGGATTTG	GGGGTCGTTGATGGCAACA

Statistical Analysis

Data were expressed as mean ± SEM. Statistical analyses were performed using 1-way ANOVA followed by Tukey's multiple-comparison test for comparisons among more than two groups, and two-tailed Student's t-test for comparisons between the two groups. GraphPad Prism (Dotmatics, USA) was used for statistical analysis and graphical representation of data. Significance levels were indicated as follows: *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001.

Results

The Limbal Lymphatic Vessels Were not Adjacent to the Mature Corneal Epithelium but Were Spatially Adjacent to the Limbal Stem Cells

Figure 1A illustrated the co-localization of Lyve1 and Prox1 within the limbal lymphatic endothelial cells. ΔNp63, which marks the LSCs, was detected in the basal epithelial layer of the limbus, demonstrating spatial proximity to Lyve1- and Prox1-marked lymphatic vessels (Fig. 1B). K14, another marker of LSCs, was also present in the basal epithelial layer and adjacent to Lyve1- and Prox1-positive lymphatic vessels (Fig. 1C). This spatial correlation between LSCs and lymphatic vessels suggested that lymphatic vessels might influence LSC activities through specific proteins and signaling molecules. Figure 1D depicted the immunofluorescence of the mature corneal epithelial cell marker K12 and the lymphatic vessel marker Lyve1. K12 expression was confined to the corneal epithelial layer, whereas Lyve1- and Prox1-positive lymphatic vessels were identified in the superficial stroma of the limbus, indicating a lack of spatial overlap between these markers.

Figure 1.

The spatial position of corneal lymphatic vessels and their proximity to limbal stem cells. (A) Immunostaining of Prox1 and Lyve1 in the corneal whole mounts from wild-type (WT) mice delineated the co-localization of Prox1 and Lyve1 and spatial positioning of the limbal lymphatic vessels (N = 4 eyes). (B) Immunostaining of ΔNp63, Prox1, and Lyve1 in the corneal whole mounts from WT mice revealed that ΔNp63-labeled limbal stem cells were spatially juxtaposed to the limbal lymphatic vessels (N = 4 eyes). (C) Immunostaining of Prox1, Lyve1, and K14 in the corneal whole mounts from WT mice indicated that K14-labeled limbal stem cells were spatially juxtaposed to the limbal lymphatic vessels (N = 4 eyes). (D) Immunostaining of K12, Prox1, and Lyve1 in the corneal whole mounts from WT mice demonstrated that the localization of the corneal epithelium was not adjacent to the limbal lymphatic vessels (N = 4 eyes).

AAV-shProx1 Significantly Reduced Prox1 Protein and mRNA Levels in the Limbal Region

Mice were divided into AAV-sham and AAV-shProx1 groups based on the type of subconjunctival AAV injection, and corneal samples were collected from both groups 4 weeks later (Fig. 2A). Immunofluorescence staining of corneal tissues revealed a significant downregulation of Prox1 protein in the AAV-shProx1 group compared to the AAV-sham group (Figs. 2B, 2C). The mRNA analysis corroborated these findings, showing a significant reduction in Prox1 mRNA levels in the AAV-shProx1 group (Fig. 2D). In homeostasis, anterior segment photography of the two groups of mice demonstrated comparable corneal clarity with no discernible differences in transparency (Fig. 2D). Fluorescein sodium staining revealed no significant variations in epithelial defect areas (Figs. 2E, 2F). Histological analysis indicated no differences in the limbal epithelial thickness between the groups (Fig. 2G). Furthermore, there were no statistically significant differences in fluorescence intensity in limbal epithelial cell proliferation (Ki67) or stemness biomarkers (ΔNp63 and K14) between the AAV-sham and AAV-shProx1 groups (Figs. 2H–K).

Figure 2.

The Prox1 gene in the mouse corneal limbus was successfully knocked down by using AAV-shProx1. No differences in corneal epithelial thickness or limbal stemness markers were observed between Prox1 knockdown mice and controls in homeostasis. (A) Wild-type mice were given subconjunctival injections of AAV-sham or AAV-shProx1, and corneal samples were collected 4 weeks later. (B) Immunostaining of Prox1 and Lyve1 in the corneas of mice treated with AAV-sham or AAV-shProx1 (N = 4 animals/group). (C) Quantitative analysis of the average fluorescence intensity of Prox1 in the AAV-sham group and the AAV-shProx1 group (N = 4 animals/group). (D) Quantitative analysis of Prox1 mRNA levels in the AAV-sham group and the AAV-shProx1 group (N = 3 animals/group). (E) Anterior segment photographs and sodium fluorescein staining images, which were taken 4 weeks after subconjunctival AAV injection in both groups, revealed no abnormalities in the corneal epithelium (N = 6 animals/group). (F) Hematoxylin and eosin (H&E) staining of the corneal limbus in the AAV-sham and the AAV-shProx1 groups showed no difference in epithelial thickness between the two groups (N = 4 animals/group). (G) Statistical analysis of corneal limbal epithelial thickness in the AAV-sham and AAV-shProx1 groups indicated no significant difference (N = 4 animals/group). (H) Immunofluorescent staining of Ki67, ΔNp63, and K14 at 4 weeks after subconjunctival AAV injection in both groups showed no significant differences in the proliferative or stemness markers in homeostasis (N = 4 animals/group). (I–K) Quantitative analyses of the average fluorescence intensities of Ki67, ΔNp63, and K14 in the AAV-sham and AAV-shProx1 groups confirmed no significant differences in proliferation or stemness in homeostasis (N = 4 animals/group).

The Corneal Epithelial Wound Healing Rate and the Recovery of Corneal Limbal Stemness Following Alkali Burn Were Delayed Through the Knockdown of Prox1 Expression in the Mouse Corneal Limbus

Mice were categorized into the AAV-sham and AAV-shProx1 groups based on AAV injection status, 4 weeks later, an alkali burn was induced on the cornea, and corneas were collected 7 days post-injury (Fig. 3A). In the AAV-shProx1 group, the Prox1 protein expression was significantly downregulated compared to the AAV-sham group at 7 days post-injury, with the difference being statistically significant (Figs. 3B, 3C). The mRNA analysis further validated these results, revealing a marked downregulation of Prox1 mRNA expression in the AAV-shProx1 group at 7 days post-injury (Fig. 3D). Fluorescein sodium staining demonstrated that the area of corneal epithelial defects was significantly larger in the AAV-shProx1 group compared to the AAV-sham groups on days 1 and 3 post-injury, indicating a slower healing rate in the Prox1 knockdown group (Figs. 3E, 3F). H&E staining demonstrated that beginning on day 1, the corneal limbal epithelium in the knockdown group was notably thinner than that in the control group, suggesting a decelerated rate of epithelial repair (Figs. 3G, 3H). Immunofluorescence staining of Ki67 in corneal sections showed a significant decrease in Ki67-positive cells in the AAV-shProx1 group compared to the AAV-sham group at day 3 post-injury (Figs. 3I, 3J). The AAV-Sham group peaked in Ki67-positive cells on day 3, whereas the AAV-shProx1 group peaked on day 14, indicating a delay in LSCs proliferation due to Prox1 knockdown. The ΔNp63 staining revealed a decrease in ΔNp63 expression in the AAV-sham group post-injury, with recovery by day 7. In contrast, the AAV-shProx1 group showed prolonged downregulation of ΔNp63 until day 14, indicating damage and delayed recovery of LSCs stemness (Figs. 3K, 3L). Similar trends were observed with other stemness markers, such as K14, which showed significant downregulation in the AAV-shProx1 group compared to the AAV-sham group at days 7, 14, and 21 post-injury (Figs. 3M, 3N). Slit-lamp microscopy revealed that the corneal transparency scores in the AAV-shProx1 group were significantly higher (indicating greater opacity) than those in the naive and AAV-sham groups on days 7, 14, and 21 post-injury (Supplementary Figs. S1A, S1B). There was no significant difference in corneal transparency between the naive and AAV-sham groups at any time point.

Figure 3.

The corneal epithelial wound healing rate and the recovery of corneal limbal stemness following alkali burn were delayed through the knockdown of Prox1 expression in the mouse corneal limbus. (A) WT mice were given subconjunctival injections of AAV-sham or AAV-shProx1. Four weeks later, an alkali burn was induced on the cornea, and the corneas were collected 7 days post-injury. (B) Four weeks after the subconjunctival injection of AAV-sham or AAV-shProx1 in the WT mice, an alkali burn was induced, and the corneas were harvested 7 days post-injury for Prox1 and Lyve1 immunofluorescent staining (N = 6 animals/group). (C) Immunofluorescence quantitative analysis at 7 days post-burn of the AAV-sham and AAV-shProx1 groups (N = 6 animals/group). (D) RT-qPCR results revealed that Prox1 mRNA expression at 7 days post-burn in the AAV-sham and AAV-shProx1 groups (N = 3 animals/group). (E) Fluorescein sodium staining of the ocular surface was performed in both groups at the time of injury and on days 1, 3, 7, 14, and 21 post-injury (N = 6 animals/group). (F) Line-graph analysis of corneal epithelial defect areas at each time point of the AAV-sham and AAV-shProx1 groups (N = 6 animals/group). (G) H&E staining of the cornea on days 1, 3, 7, 14, and 21 post-burn in the AAV-sham and the AAV-shProx1 groups (N = 4 animals/group). (H) Statistical analysis of the limbal epithelial thickness at different time points post-burn of the AAV-sham and the AAV-shProx1 groups (N = 4 animals/group). (I) Ki67 immunofluorescent staining at the corneal limbus on days 1, 3, 7, 14, and 21 post-burn revealed an initial increase and subsequent decrease in Ki67 expression in both groups (N = 4 animals/group). (J) The temporal line graph of Ki67-positive cells in both groups showed that Ki67 levels first increased and then decreased. In the AAV-sham group, the peak occurred on day 3, whereas in the AAV-shProx1 group, the peak was delayed until day 14 (N = 4 animals/group). (K) ΔNp63 immunofluorescent staining at the corneal limbus on days 1, 3, 7, 14, and 21 post-burn of the AAV-sham and the AAV-shProx1 groups (N = 4 animals/group). (L) The temporal line graph of ΔNp63-positive cells in both groups showed that from day 7, ΔNp63 fluorescence intensity was consistently lower in the AAV-shProx1 group (N = 4 animals/group). (M) K14 immunofluorescent staining of the corneal limbus on days 1, 3, 7, 14, and 21 post-burn of the AAV-sham and the AAV-shProx1 groups (N = 4 animals/group). (N) The temporal line-graph of the K14-positive cells in both groups indicated that from day 7, K14 fluorescence intensity remained lower in the AAV-shProx1 group compared with the AAV-sham group (N = 4 animals/group).

Lymphatic Endothelial Cells Enhanced the Proliferative Capacity and Stemness Maintenance of Limbal Stem Cells In Vitro

To elucidate the interplay between LSCs and LECs, we first used a conditioned medium, results demonstrated a marked increase in Edu-positive and Ki67-positive cells in the conditioned medium group versus the serum-free medium group after 24 hours of incubation (Supplementary Figs. S2A–E). Furthermore, the wound healing assay revealed a significantly enhanced migratory capacity in the conditioned medium group at 48 hours compared to the serum-free medium group (Supplementary Figs. S2F, S2G). Then, a Transwell-based co-culture system was used to investigate the influence of LECs on TKE2 (Fig. 4A). Results demonstrated a marked increase in Edu-positive and Ki67-positive cells in the co-culture group relative to the control group after 24 hours of incubation (Figs. 4B–E). These findings indicated that LECs potentiated the proliferation of limbal stem cells. To gain further insights into the effects of LECs on the preservation of stemness in LSCs, we assessed the expression levels of key stemness-associated proteins following 6, 12, and 24 hours of co-culture. The data revealed an upregulation of ΔNp63 at 6 hours post-co-culture, whereas K14 exhibited significant upregulation at 24 hours compared to the control group (Figs. 4F, 4G, and 4I). The RNA results exhibited a consistent trend with protein expression. At 6 hours post-co-culture, ΔNp63 was upregulated in the co-culture group compared to the control group (Fig. 4J). Additionally, K14 expression was elevated in the co-culture group at 6, 12, and 24 hours post-co-culture (Fig. 4H). Collectively, these results suggested that LECs contribute to the maintenance and enhancement of stemness in LSCs.

Figure 4.

Co-culture with lymphatic endothelial cells (LECs) enhances the stemness of corneal limbal stem cells. (A) The schematic diagram illustrated the experimental design for the co-culture and control groups. (B) Edu immunofluorescence images of the co-culture and control groups at 24 hours. (C) Quantitative analysis of Edu-positive cells at 24 hours showed that the co-culture group had significantly more Edu-positive cells than the control group (N = 3 experiments/group). (D) Ki67 immunofluorescence images of the co-culture and control groups at 24 hours. (E) Quantitative analysis of Ki67-positive cells at 24 hours showed that the co-culture group had significantly more Ki67-positive cells than the control group (N = 3 experiments/group). (F) Western blot analysis of stemness markers K14 and ΔNp63 in TKE2 cells at 6, 12, and 24 hours of co-culture and the control group. (G) The relative protein expression of K14 was significantly higher in the co-culture group at 24 hours than in the control group (N = 3 experiments/group). (H) The mRNA level of K14 was significantly higher in the co-culture group at 24 hours than in the control group (N = 3 experiments/group). (I) The relative protein expression of ΔNp63 was significantly higher in the co-culture group at 6 hours than in the control group (N = 6 experiments/group). (J) The mRNA level of ΔNp63 was significantly higher in the co-culture group at 6 hours than in the control group (N = 3 experiments/group).

Silencing the Prox1 Gene of LECs Affected the Stemness and Proliferative Capacity of Limbal Stem Cells In Vitro

LECs were divided into the siProx1 group and the si-negative control (siNC) group based on the siRNA added to the LECs (Fig. 5A). Figure 5D indicated that 100 nM siProx1 significantly reduced Prox1 mRNA expression compared to the siNC group. This knockdown effect was further confirmed at the protein level, the protein expression of Prox1 in the LECs of the siProx1 group was significantly lower than that of the control group (Figs. 5B, 5C). When co-culturing Prox1-knockdown LECs with LSCs, the proportion of Edu-positive and Ki67-positive cells was significantly lower in the siProx1 group compared to the siNC group (Figs. 5E–H). Additionally, stemness marker analysis showed that K14 expression levels were decreased in the siProx1 group after 24 hours of co-culture, whereas ΔNp63 expression was downregulated as early as 12 hours into co-culture (Figs. 5I, 5J, and 5L). The RNA results were consistent with the protein expression data. At 6, 12, and 24 hours post-co-culture, K14 expression was downregulated in the siProx1 group compared to the siNC group (Fig. 5K). Similarly, at 12 and 24 hours post-co-culture, ΔNp63 expression was downregulated in the siProx1 group relative to the siNC group (Fig. 5M). These findings suggested that Prox1 knockdown significantly affected the proliferation and stemness characteristics of LSCs.

Figure 5.

The proliferative capacity and stemness of corneal limbal stem cells were reduced by knocking down the Prox1 gene in LECs. (A) The schematic diagram of the experimental design for the siNC and siProx1 groups. (B) Western blot validation of Prox1 expression of LECs in siNC and siProx1 groups (N = 3 experiments/group). (C) Relative protein expression of Prox1 in LECs was significantly lower in the siProx1 group compared to the siNC group (N = 3 experiments/group). (D) The mRNA levels of Prox1 in LECs were significantly lower in the siProx1 group compared to the siNC group (N = 3 experiments/group). (E) Edu immunofluorescence images of TKE2 cells in the siNC and siProx1 groups after 24 hours of co-culture. (F) Quantitative analysis of Edu-positive TKE2 cells at 24 hours of co-culture showed that the siProx1 group was significantly lower than the siNC group (N = 6 experiments/group). (G) Ki67 immunofluorescence images of TKE2 cells in the siNC and siProx1 groups after 24 hours of co-culture. (H) Quantitative analysis of Ki67-positive TKE2 cells at 24 hours of co-culture showed that the siProx1 group was significantly lower than the siNC group (N = 6 experiments/group). (I) Protein expression levels of the stemness markers K14 and ΔNp63 in TKE2 cells from the siNC and siProx1 groups after 6, 12, and 24 hours of co-culture. (J) The relative protein expression of K14 in TKE2 cells was significantly lower in the siProx1 group compared to the siNC group at 24 hours of co-culture (N = 4 experiments/group). (K) The mRNA level of K14 in TKE2 cells was significantly lower in the siProx1 group compared to the siNC group at 24 hours of co-culture (N = 5 experiments/group). (L) The relative protein expression of ΔNp63 in TKE2 cells was significantly lower in the siProx1 group compared to the siNC group at 12 hours of co-culture (N = 3 experiments/group). (M) The mRNA levels of ΔNp63 in TKE2 cells were significantly lower in the siProx1 group compared to the siNC group at 12 and 24 hours of co-culture (N = 3 experiments/group).

The Enrichment of Differential Genes Following Prox1 Knockdown was Predominantly Associated With Immune Modulation and Alterations in Inflammatory Cytokines

Figure 6A presented the transcriptomic sequencing of corneas 7 days post-alkali burn injury in the AAV-shProx1 and the AAV-sham groups, highlighting the differential gene expression between the two groups. The results demonstrated that genes such as Prox1, Lyve1, Abcg8, Trp53, and Mapk8 were downregulated in the AAV-shProx1 group, suggesting that the knockdown of Prox1 led to the downregulation of genes related to regeneration and differentiation of lymphatic vessels. Additionally, compared to the AAV-sham group, the AAV-shProx1 group exhibited upregulation of lymphocyte cytosolic proteins (Lcp1 and Lcp2), integrin family members (Itga5 and Itgb2), and inflammatory factors (Ccl2, Ccl7, IL16, IL1R, and TNFsf11). Figure 6B displayed a volcano plot of the differential gene expression, further corroborating that the upregulated genes were predominantly associated with enhanced inflammation, whereas the downregulated genes were enriched in pathways related to cell proliferation and development. GO and KEGG enrichment analyses of the differential genes indicated significant enrichment in pathways such as the negative regulation of lymphocyte migration, negative regulation of leukocyte chemotaxis, primary immunodeficiency, the intestinal immune network for IgA production, and positive regulation of acute inflammatory response to antigen (Figs. 6C, 6D). KEGG classification showed that the differentially expressed genes were enriched in the immune system, developmental system, and signal transduction processes (Fig. 6E).

Figure 6.

Differentially expressed genes following Prox1 knockdown are primarily associated with immune regulation and changes in inflammatory factors. (A) The heatmap showed the differentially expressed genes in the corneas of mice from the AAV-shProx1 group compared with the AAV-sham group at 7 days post-injury (N = 3 animals/group). (B) The volcano plot displayed the upregulated and downregulated genes in the corneas of mice from the AAV-shProx1 group compared with the AAV-sham group at 7 days post-injury. (C) GO enrichment analysis showed that genes affected by Prox1 knockdown were enriched in pathways such as negative regulation of lymphatic vessels and positive regulation of antigen-induced acute inflammatory responses. (D) KEGG enrichment analysis illustrated that top signaling pathways were associated with primary immunodeficiency and intestinal immune network for IgA production. (E) KEGG classification showed that the differentially expressed genes were enriched in the immune system, developmental system, and signal transduction processes.

The Formation of Newly Generated Lymphatic Vessels and the Expression of Lyve1 After Corneal Alkali Burns Were Reduced by Knocking Down Prox1. Reduced Lymphatic Drainage Capacity led to the Accumulation of Inflammatory Factors and Immune Cells Around the LSCs, Impairing the Proliferation and Stemness of LSCs

Immunofluorescence staining of corneal whole mounts post-alkali burn showed increased Lyve1-positive lymphatic vessels over time in all groups (Fig. 7A). Comparative analysis revealed significant downregulation of Lyve1 expression in the AAV-shProx1 group compared to the AAV-sham groups on days 7, 14, and 21 post-injury (see Figs. 7A, 7B), indicating that the modulation of Lyve1 expression manifested subsequent to the alterations in Prox1. Thus, Prox1 and Lyve1 may play a crucial role in regulating LSCs stemness and corneal repair processes. As elucidated by the preceding RNA sequencing analysis, inflammatory cytokines, such as Ccl2, Ccl7, Lcp1, Ly6c1, TNFsf11, IL16, and CD8, were upregulated in the AAV-shProx1 group. Western blot was utilized to corroborate the differential expression of IL-1β, TNF-α, and Ccl2 in the corneas of the 2 groups of mice 7 days post-injury. The findings revealed a statistically significant upregulation of IL-1β and TNF-α in the knockdown group, whereas there was no significant difference in Ccl2 expression between the two groups (Figs. 7C–F). The mRNA results were consistent with the protein expression data. In the AAV-shProx1 group, the mRNA levels of IL-1β and TNF-α were significantly upregulated compared to the AAV-sham group at 7 days post-injury, with statistically significant differences (Figs. 7G, 7H). Furthermore, immunofluorescence analysis provided additional validation of the hypothesis, demonstrating increased accumulation of IL-1β and TNF-α in the limbal region of the AAV-shProx1 group compared to the AAV-sham group (Figs. 7I–L). A significantly higher accumulation of CD68⁺ macrophages (Figs. 7M, 7N) and CD11c⁺ dendritic cells (Figs. 7O, 7P) was observed in the limbus of the AAV-shProx1 group compared to the AAV-sham group, highlighting enhanced immune cell infiltration. This evidence indicated that the reduction in lymphatic vessels impeded the drainage of inflammatory cytokines and immune cells, resulting in their accumulation within the cornea. Consequently, this accumulation inflicted damage upon limbal stem cells, attenuated their stemness, and delayed epithelial repair processes.

Figure 7.

The formation of newly generated lymphatic vessels and the expression of Lyve1 after corneal alkali burns were reduced by knocking down Prox1. The resulting decrease in new lymphatic vessels leads to reduced lymphatic drainage, causing injury-induced inflammatory factors and immune cells to accumulate in the cornea. This accumulation exerted a negative regulatory effect on the proliferative function and stemness maintenance of limbal epithelial stem cells. (A) Immunofluorescent staining for Lyve1 in the AAV-Sham and AAV-shProx1 groups at pre-burn and days 1, 7, 14, and 21 post-burn (N = 4 animals/group). (B) Temporal quantitative analysis of Lyve1 expression in the AAV-sham and AAV-shProx1 groups (N = 4 animals/group). (C) Protein expression of inflammatory factors IL-1β, TNF-α, and Ccl2 of corneas in AAV-sham and AAV-shProx1 groups at 7 days post-burn. (D–F) Statistical comparison of the relative protein levels of IL-1β, TNF-α, and Ccl2 in corneal tissues of the 2 groups at 7 days post-burn (N = 10 animals/group). (G, H) Statistical analysis of the relative mRNA levels of IL-1β and TNF-α in the corneas of the two groups (N = 6 animals/group). (I) IL-1β immunofluorescent staining at the corneal limbus of AAV-sham and AAV-shProx1 on days 7 post-burn. (J) Statistical analysis of the mean fluorescence intensity of IL-1β in the two groups (N = 6 animals/group). (K) TNF-α immunofluorescent staining at the corneal limbus of AAV-sham and AAV-shProx1 on days 7 post-burn. (L) Statistical analysis of the mean fluorescence intensity of TNF-α in the two groups (N = 6 animals/group). (M) CD68 immunofluorescent staining at the corneal limbus of AAV-sham and AAV-shProx1 on days 7 post-burn. (N) Statistical analysis of the mean fluorescence intensity of CD68 in the two groups. N = 4 animals/group. (O) CD11c immunofluorescent staining at the corneal limbus of AAV-sham and AAV-shProx1 on days 7 post-burn. (P) Statistical analysis of the mean fluorescence intensity of CD11c in the two groups (N = 4 animals/group).

Discussion

LSCD poses a formidable clinical challenge, with its multifaceted etiology remaining incompletely elucidated. Although current therapeutic approaches are efficacious for mild LSCD, their effectiveness diminishes considerably for moderate to severe cases.^19–22 Contemporary researches that focus on LSCs and the signaling pathways which sustain LSCs stemness have predominantly concentrated on the Wnt, Notch, TGF-β/BMP, Sonic Hedgehog, and YAP/TAZ pathways.²³ Wnt ligands typically activate three Wnt pathways to regulate stem cell proliferation, differentiation, and migration, including the canonical Wnt/β-catenin pathway, the non-canonical Wnt/planar cell polarity (PCP) pathway, and the non-canonical Wnt/Ca²⁺ pathway. Studies have shown that the distribution of Wnt signaling regulators differ between the human limbus and the central cornea.²⁴ Downregulation of Fzd7 in the Wnt signaling pathway has been demonstrated to promote the downregulation of LSC markers K14, Abcg2, and ΔNp63, and the upregulation of K12.²⁵ Notch1 receptor expression has been observed in the basal epithelium of the mouse limbus, and blocking the Notch pathway with small molecule inhibitors DAPT and SAHM1 inhibits LSCs proliferation and stemness expression.²⁶

The cornea is traditionally considered an immune-privileged structure and devoid of blood and lymphatic vessels under normal conditions. Over the past 2 decades, the studies of lymphatic markers (Lyve1, Prox1, Vegfr3, and podoplanin) have advanced our understanding of lymphatic involvement in the cornea.²⁷ Lymphatic vessels serve as biological conduits for draining interstitial fluid from tissues to the bloodstream and function as immune organs, facilitating antigen presentation.²⁸ Research on lymphatic vessels in the cornea primarily centers on transplantation rejection. Studies have indicated that the use of an integrin α5β1 inhibitor can effectively suppress limbal lymphangiogenesis in the graft bed, leading to longer survival time of corneal grafts compared to the control group.²⁹ Additionally, lymphatic vessels have been implicated in the formation of pterygium,³⁰ aqueous humor outflow pathways,¹⁰ and choroidal melanoma.³¹

Comparative studies have revealed that LSCs secrete more pro-lymphangiogenic factors, such as IL-1β, Cxcl10, TGF-β1, and VEGF-A, compared with mature corneal epithelial cells.³² Other studies suggest that LSCs inhibit lymphocyte proliferation, thereby reducing inflammation and neovascularization.³³ However, the role of lymphatic vessels and LSCs remains inadequately elucidated. Research on the interaction between lymphatic vessels and LSCs is sparse, and the concept of limbal lymphatics as a component of the limbal microenvironment is novel. This study aims to investigate the role of lymphatic vessels in the homeostasis and injury repair of LSCs.

Recent studies have increasingly highlighted the role of lymphatic vessels in tissue regeneration. Limbal cells are classified into two populations: an inner population that maintains epithelial homeostasis and an outer population that mediates epithelial repair during injury, with the outer population closely associated with blood and lymphatic vessels.⁶ Given the histological homology between skin and corneal epithelium, Sox9 and Sox18-marked stem cells in the skin’s dermis are closely adjacent to Lyve1⁺Vegfr3⁺Prox1⁺ lymphatic vessels. During skin injury, increased secretion of angiopoietin-like protein 4 by stem cells drives lymphatic remodeling, regulating hair follicle regeneration.¹² Similarly, in the regenerating adult zebrafish heart, extensive lymphangiogenesis occurs following cryoinjury. Zebrafish with disrupted Vegfc/Vegfr3 signaling exhibit significant defects in scar size reduction post-injury, failing to develop complete cardiac lymphatics.³⁴ These findings suggested that the lymphatic system can influence myocardial regeneration potential, implying that lymphatic vessels may be integral to the LSC microenvironment. The present research revealed that knocking down Prox1, a key gene for lymphangiogenesis, affects both corneal epithelial wound healing and LSC stemness.

In Prox1 cardiac knockout mice, the absence of Prox1 resulted in smaller hearts with reduced weight and impaired myocardial cell proliferation, along with increased myocardial cell apoptosis.¹⁰ In murine models of radiation therapy and bone marrow ablation, knocking down the Prox1 gene in bone marrow lymphatic vessels impaired hematopoietic capacity and the stemness of osteogenic stem cells. This effect was regulated through the secretion of Cxcl12 by LECs, which binds to the Cxcr4 receptor on bone-derived Myh11-positive cells.³⁵ Moreover, extensive studies have demonstrated that Prox1 plays a pivotal role in the regeneration and differentiation of glial cells, serving as a key regulator within the “go and stop” signaling pathways of glial cell dynamics.³⁶ Additionally, Prox1 has been implicated in the formation of late-stage crystallin proteins in newts, with its function closely associated with the Wnt signaling pathway.³⁷ Deletion of Lyve1 in mice, which inhibited the docking and transit of leukocytes through the lymphatic endothelium, led to exacerbating chronic inflammation and long-term deterioration of cardiac function.³⁸ Furthermore, studies have confirmed that Lyve1-positive lymphatic vessels are closely adjacent to intestinal stem cells³⁹ and skin stem cells.¹² Studies have demonstrated the mutual regulatory interaction between Lyve1 and Prox1.⁴⁰ Consistent with these findings, our study showed that Prox1 knockdown led to decreased Lyve1 expression of lymphatic vessels in the cornea post-alkali burn, leading to fewer Ki67-positive cells at the limbus, reduced stemness expression, and delayed repair.

In addition to their role in antigen presentation, lymphatic vessels also function in draining metabolic waste products. The results indicated that the pro-inflammatory cytokine IL-1β and TNF-α levels were markedly elevated in the knockdown group compared to the control group (Fig. 8). Studies have shown that the accumulation of inflammatory cytokines such as IL-1 is observed in the cornea post-alkali burn.⁴¹ Further research indicated that CCR2-marked monocytes, through TNF-α, damage LSCs and promote apoptosis, whereas CX3CR1-marked resident macrophages protect LSCs.⁴² Our study suggested that the reduction in lymphatic vessels due to Prox1 knockdown diminished the drainage of inflammatory cytokines such as IL-1β and TNF-α from the cornea to the systemic circulation, leading to their accumulation in the superficial stroma. This accumulation damaged LSCs and slowed epithelial recovery.

Figure 8.

The schematic diagram illustrated the role of limbal lymphatic vessels in regulating the stemness of limbal stem cells and corneal epithelial wound repair. Limbal lymphatic vessels reside in the superficial stromal layer of the limbus and are adjacent to limbal stem cells. Prox1 influenced the stemness and proliferation of limbal stem cells primarily via paracrine signaling by regulating the activity of lymphatic endothelial cells and proteins such as Lyve1. On the other hand, the drainage function of lymphatic vessels is widely recognized. Prox1 knockdown reduced lymphangiogenesis, thereby weakening the drainage pathway for inflammatory factors (IL-1β and TNF-α) and immune cells (macrophages and dendritic cells) following corneal injury. This led to an accumulation of inflammatory factors and immune cells in the corneal stroma, exerting detrimental effects on limbal stem cells.

Following Prox1 knockdown and corneal injury, CD68⁺ macrophages and CD11c⁺ dendritic cells accumulated at the limbus due to impaired lymphatic drainage (see Fig. 8). The role of macrophages in regulating LSCs was not fully understood, but they were generally classified into M1 pro-inflammatory and M2 anti-inflammatory phenotypes. M1 macrophages inhibited LSCs function, whereas M2 macrophages promoted it.^42–44 Our findings suggested a predominance of M1 macrophages, contributing to LSC dysfunction and impaired corneal repair. Additionally, in dry eye models, netrin-1 knockdown has been shown to suppress dendritic cell-mediated Th17 activation, alleviating inflammation.^45,46 Consistently, our results indicated that reduced lymphangiogenesis increases dendritic cell accumulation, which enhances Th17 activation. Th17 cells secrete pro-inflammatory cytokines, such as IL-17, IFN-γ, and IL-1β, impairing LSC stemness and proliferation, thereby hindering corneal regeneration and homeostasis.

The present research suggested that co-culturing LECs with TKE2 cells enhances the proliferative capacity and stemness of TKE2 cells. However, when Prox1 gene expression in LECs was knocked down and these cells were co-cultured with TKE2 cells, a significant reduction in the proliferative capacity and stemness of TKE2 cells was observed compared to the normal co-culture group. These results indicated that Prox1 in LECs may regulate the proliferative capacity and stemness of LSCs by modulating related signaling pathways, such as the expression of Lyve1, and the secretion of other proteins.

The present study had certain limitations. Although TKE2 cells are widely used in limbal stem cell researches,^47–49 the use of human primary LSCs would provide a more accurate representation of the biological activities of human LSCs. The application of human primary LSCs in transplantation showed promising prospects. For instance, ABCB5-positive cells have been shown to enhance epithelial proliferation and migration rates post-transplantation.⁵⁰ Additionally, transplantation of corneal epithelial sheets derived from human primary limbal stem cells⁵¹ or pluripotent stem cells⁵² into the ocular surface of patients with LSCD has demonstrated significant improvements in visual acuity and reduction in corneal neovascularization. However, human primary LSCs face challenges, such as limited availability, restricted passage numbers, and loss of stemness during passaging.²¹

In conclusion, our study highlights the pivotal role of the Prox1 and lyve1 in limbal lymphatic vessels for maintaining LSCs stemness and facilitating corneal injury repair. The Prox1 knockdown group revealed significant impairments in lymphangiogenesis, delayed epithelial wound healing, and reduced stemness marker expression in LSCs. These findings underscore the integral role of the lymphatic system within the limbal microenvironment, offering valuable insights into the molecular interactions that sustain LSCs function and corneal homeostasis. The elucidation of this signaling pathway provides a promising foundation for the development of targeted therapeutic strategies to enhance LSCs viability and promote effective corneal regeneration, addressing critical unmet needs in ophthalmology. By establishing the foundational role of lymphatic vessels in the limbal niche, this research enriches our understanding of LSCs biology and paves the way for future investigations into the complex interplay within the limbal microenvironment, with significant implications for advancing regenerative medicine and improving clinical outcomes for patients suffering from LSCD and other corneal pathologies.