Decoding epithelial regeneration in the cornea: multi-omic analysis reveals cellular plasticity as central mechanism

Abstract

Background

Rapid and efficient epithelial regeneration is fundamental for tissue homeostasis and proper function. As the outermost ocular structure, the cornea is transparent, multilayered, and vital for clear vision. Due to its exposed position, the cornea frequently undergoes various forms of injury affecting either the epithelium itself or its surrounding microenvironment, including corneal innervation and the tear film. Corneal abrasion, occurring commonly through trauma or as part of refractive surgical procedures, is typically viewed as a minor event since it usually resolves rapidly. Consequently, the cornea serves as an excellent model for studying epithelial wound healing. However, complications such as persistent epithelial defects or corneal opacity can develop, underscoring critical gaps in understanding the underlying molecular mechanisms.

Methods

Utilizing a unilateral corneal abrasion mouse model, we conducted a comprehensive multi-omics analysis, integrating transcriptomics, proteomics, and epitranscriptomics, to dissect the dynamic molecular responses post-injury in both wounded and contralateral tissues. To elucidate the role of the tear film, we performed additional studies involving lacrimal gland ablation combined with corneal injury. We applied RNA sequencing to profile transcriptomic changes in corneal and lacrimal gland tissues, and mass spectrometry to study tear proteomics and epitranscriptomic modifications.

Results

We revealed a major modulation of the cornea transcriptome after abrasion, suggesting a regulation of pathways including JAK-STAT, Wnt and TGF-β, and a reduction of nucleoside modifications. The lacrimal gland transcriptome and tears proteome were also significantly affected. Plus, we highlighted a bilateralization, both in the cornea transcriptome and tears proteome. In the tear-deficient conditions, the wound closure rate and molecular responses were altered, and the bilateralization was impacted, with an increased matrix remodeling and a modulation of keratins expression.

Conclusions

Our multi-omics analyses revealed extensive epithelial cellular plasticity as a key mechanism driving rapid wound closure, characterized by profound remodeling of transcriptional networks and RNA modifications. Importantly, we uncovered a previously underappreciated role of the lacrimal gland and tear film in mediating bilateral molecular responses following unilateral injury, emphasizing their pivotal roles in tissue regeneration. Additionally, we identified novel regulatory roles for RNA methylation events and critical signaling pathways implicated in epithelial healing.

Graphical abstract

Supplementary Information

The online version contains supplementary material available at 10.1186/s11658-025-00804-9.

Background

Epithelial tissues are endowed with regenerative abilities that are essential for sustaining their critical functions, such as forming protective barriers or regulating absorption and secretion. This regenerative capacity is a key mechanism in maintaining overall tissue homeostasis, whether for physiological renewal or following epithelial damage. Composed of a stratified squamous epithelium, supported by a mesenchymal stromal layer and an inner endothelium, the cornea is a transparent, multilayered structure at the anterior surface of the eye. By virtue of its anatomical position and optical properties, the cornea is essential to focusing and transmitting light to the retina, while simultaneously protecting underlying ocular structures from environmental and internal perturbations[1]. Owing to its exposed location, the cornea is vulnerable to damage from external insults[2], as well as age-related changes[3] and various systemic[4] and local pathologies[5, 6]. When the homeostasis of the corneal microenvironment—which encompasses the corneal epithelium, its innervation, and the tear film—is disrupted, this can result in progressive thickening, leading to opacification, or thinning, leading to its disintegration. Such structural changes lead to corneal blindness, a condition that affects an estimated 28 million people worldwide and is the fourth leading cause of blindness globally.

Among the myriad causes of corneal dysfunction, corneal abrasions are the most frequently encountered ocular injuries in clinical practice[7, 8]. Multiple factors contribute to corneal abrasions, including direct mechanical trauma, chemical injuries, foreign bodies, contact lens wear[9], and inadequate tear film quality[10]. These abrasions involve a nonpenetrating disruption of the epithelial barrier, manifesting as ocular pain, photophobia, and excessive tearing[11]. Clinical confirmation is typically achieved using fluorescein staining under cobalt-blue illumination, which reveals epithelial defects through characteristic dye uptake[12]. In otherwise healthy eyes with proper care, minimal intervention, and normal corneal regenerative capacity[13], these superficial injuries usually heal rapidly (within 24 to 72 h), restoring corneal clarity. Standard treatments include artificial tears to maintain lubrication, topical NSAIDs for pain management, and prophylactic topical antibiotics to prevent secondary infections[11, 14].

However, when not promptly and properly managed, corneal abrasions can lead to persistent epithelial defects[15] and elevated infection risk. Delayed wound closure compromises the corneal protective function, permitting the ingress of various pathogens[16] and increasing the likelihood of ulcers, perforations, severe pain, and vision loss. Corneal abrasions are also a common perioperative complication in patients undergoing general anesthesia for non-ocular surgeries[17], and they are an intentional step in certain refractive procedures. For example, photorefractive keratectomy (PRK) and laser-assisted in situ keratomileusis (LASIK) both involve controlled disruptions of the corneal epithelium to correct refractive errors[18]. Despite their high success rates[19, 20], these surgeries can be complicated by delayed re-epithelialization[21], recurrent erosions[22], or dry eye diseases[23], all of which highlight the delicate balance required for effective corneal wound healing.

The wide range of responses to corneal abrasion—from rapid, complete healing to severe complications culminating in scarring and persistent vision impairment—demonstrates the complexity of the corneal wound healing process. The need to better understand these mechanisms has driven the development of mouse models that replicate human corneal abrasion. Such models utilize controlled epithelial debridement techniques[24–26] and allow precise regulation of wound size. The resulting lesions, evaluated by fluorescein staining, typically heal within 72 h and enable dissection of the cellular and molecular mechanisms underlying corneal epithelial homeostasis and repair. Here, we used a multi-omics approach to understand the impact of corneal abrasions on the epithelial interface and its interplay with the tear film. Our multi-omics analysis reveals that corneal injury triggers rapid molecular remodeling and epithelial dedifferentiation as a core regenerative mechanism. Tear film dynamics critically shape this process, modulating cell identity and extracellular matrix remodeling. We further uncover a tightly regulated bilateral molecular response to unilateral injury, positioning the corneal microenvironment as a central driver of repair. These findings not only suggest new therapeutic avenues for ocular surface diseases but also offer broader insights into epithelial regeneration and microenvironment-driven healing across tissues.

Material and methods

Animals included in this study

Mice were housed in plastic cages, on a standard light cycle (12 h light, 12 h dark), with food and water ad libitum, in a 40–60% relative humidity environment and a 21–22°C ambient temperature. Swiss/CD1 female mice (RjOrl:SWISS, Janvier Labs, France) of 11 to 12 weeks of age were used for all experiments.

Corneal abrasions and wound healing monitoring

Corneal abrasions were performed as previously described[26, 27]. Briefly, mice were anesthetized by intraperitoneal injection with a mix of ketamine (90 mg/kg, Imalgene® 1000, Centravet) and medetomidine (1 mg/kg, Domitor®, Centravet). The epithelium was abraded unilaterally on one eye with an ocular burr (Algerbrush II, reference BR2-5 0.5 mm, Alger company, USA). The abrasion was immediately checked by using a fluorescein solution (1% in PBS, Sigma-Aldrich) under cobalt blue light and a picture was taken. After abrasion, a drop of Ocrygel (TVM) was applied on both eyes, analgesia was given using a buprenorphine (0.1 mg/kg, Burprecare®, Centravet) analgesic solution and mice were woken up with atipamezole (1 mg/kg, Antisedan®, Centravet). The wellbeing of the animals was monitored during the following days. For wound healing monitoring, the eye surface was specifically checked at 1, 2, 3, 4 and 7 days after the abrasion using fluorescein solution on conscious hand-restrained mice and a picture was taken each time until complete wound closure.

Lacrimal gland ablation

Mice were anesthetized by intraperitoneal injection with a mix of ketamine (90 mg/kg) and medetomidine (1 mg/kg). A drop of Ocrygel was applied on both eyes. The mouse was put on its side, the fur under the ear was trimmed carefully, disinfected with 70% ethanol/PBS solution followed by vetedine solution (Vétoquinol). A skin incision was made above the lacrimal gland location, the extraobital lacrimal gland was removed by carefully cutting the duct and the surrounding connective tissues. The incision was sutured with two individual stitches using 6-0 absorbable suture thread (Vicryl® Polyglactin 910, reference W9981, Ethicon). For specific cohorts, a corneal abrasion was performed, as described above, after lacrimal gland ablation. The sutured skin was disinfected with vetedine solution, analgesia was given using a buprenorphine (0.1 mg/kg) analgesic solution and mice were woken up with atipamezole (1 mg/kg). The control mice only received the anaesthesia and the analgesia, but no surgery was performed. The wellbeing of the animals was monitored during the following days.

Sample collection and processing

Mice were euthanazied by cervical disclocation. The eyes were collected by enucleation, using curved scissors to cut the optic nerve, and washed briefly in PBS. For RNA-Seq, qPCR and epitranscriptomics analyses, the corneas were immediately dissected in an RNAse free environment, snap frozen in liquid nitrogen and then stored at -80 °C before the analyses. For RNA-Seq, two corneas were pooled to form each sample, except for three samples at 12H post-abrasion that needed five corneas per sample. For qPCR, five corneas were pooled to form each sample. For epitranscriptomics, three corneas were pooled to form each sample. For each mouse, both eyes were dissected individually, to ensure ipsilateral and contralateral analyses later on. Otherwise, for immunofluorescence labeling, the eyes were fixed in a 4% paraformaldehyde solution (Antigenfix) for 20 min and washed 3 times with PBS for 15 min each. Eyes were then dehydrated for 2H in 50% ethanol/PBS and stored in 70% ethanol/PBS at 4 °C. The extraorbital lacrimal glands were collected by cutting the skin above the lacrimal gland location and carefully cutting the duct and the surrounding connective tissues. The glands were snap frozen in liquid nitrogen and stored at -80 °C before the analyses. Lacrimal gland and cornea samples were taken from different mice for the different time points of the experiments.

Tear samples were collected from conscious hand-restrained mice by using 1µL disposable microcapillary glass tubes (reference 022.7726, CAMAG). Briefly, the mouse was restrained by hand and a microcapillary glass tube was gently tapped on the outer corner of the eye, avoiding contact with the eye itself. Both eyes were sampled for 1 min each (with approximatively 1 tap per second) and stored separately for each mouse in 12µL of ABC (Ammonium Bicarbonate) buffer, provided by the proteomics platform. The samples were temporarily kept on ice while the tear collection was performed and then stored at -80 °C before the analyses. For each experiment, tears were sampled from the same mice for all the time points as a longitudinal follow-up.

RNA-Seq on cornea and lacrimal glands

The same protocol was performed for both cornea and lacrimal gland samples. TRIzol reagent was used to extract total RNA, which was then purified with Qiagen RNeasy kit following the manufacturer’s procedure. Total RNA concentration was quantified with Qubit. RNA ScreenTape Assay (Agilent TapeStation 4200) was used to determine RNA quality. The ribodepletion of 500 ng of good quality total RNA (RIN > 8.8) was achieved using Illumina’s mammalian Ribo-Zero Magnetic Gold Kit HMN/Mouse/Rat (reference MRZG12324, Illumina). Ribodepleted RNA was then used to complete the library preparation with the NEBNext Ultra Directional RNA Library prep kit (reference E7420L, New England Biolabs) using 8 cycles of PCR amplification, followed by single (i7) indexing. Indexed library preparations from each sample were pooled and the NextSeq 500 using a NextSeq High Output 75 cycle flow cell (Illumina) was used for sequencing with 75SE reads at a pool concentration of 1.4 pM.

For data processing, bcl2fastq2 Conversion Software was used to convert BCL files to FASTQ file format and demultiplex samples. Sequenced reads were trimmed for adapter sequence, and masked for low-complexity or low-quality sequence using Trimmomatic. Trimmed reads were mapped to GENCODE mouse Release M30 reference genome (GRCm39) and annotation files using STAR aligner. Counts per gene were calculated using featureCounts software. Differential expression analysis used the DESeq2 software in R environment. Procedures included: normalization of count values between samples using a geometric means, using negative binomial linear model and Wald test to produce p-values, removal of low-expression outliers (using Cook’s distance) to optimize for p-value adjustment and finally, multiple testing adjustment of p-values with Benjamini–Hochberg procedure.

RT-qPCR on cornea

Samples composed of five pooled corneas were used for the extraction of total RNA. CKMix lysing kit and Precellys homogenizer (Bertin Technologies) were used to perform tissue grinding with 1 mL of TRIzol reagent (Life Technologies). RNA purification and DNase treatment were then achieved with the Nucleospin RNA Kit (Macherey Nagel). Quantification of the RNA samples was performed by using the NanodropOne spectrophotometer (ThermoFisher Scientific). The integrity of the RNAs was then checked using the Agilent 2100 bioanalyzer with the RNA 6000 Nano kit (Agilent Technologies). The High-Capacity Reverse Transcription kit (Applied Biosystems) was used following the manufacturer’s instructions to reverse transcribe 500 ng of total RNA in a 20µL final reaction volume, using random primers and the presence of RNase inhibitor. The quantitative PCR experiments were conducted using a TaqMan detection protocol with the QuantStudio 12 K Flex Real-Time PCR System (Life Technologies). The TaqMan assays (ThermoFisher Scientific) selected for each target and reference gene are presented in Table 1. For qPCR reactions, after mixing 3 ng of cDNA with TaqMan FastAdvance Master Mix and TaqMan assay in a final volume of 10µL, samples were loaded on 384-well microplates and 40 cycles of PCR were performed (50 °C/2 min; (95 °C/1 s; 60 °C/20 s) X40). To check the absence of genomic DNA contaminants, negative controls without the reverse transcriptase were implemented. Normalization of the data was achieved by adding three reference genes in the experiments (Hprt1, Ppia, Ubc) and using the GenEx software (MultiD) to select the most stable one (Ppia). Duplicates were made for each measurement, and the determined Ct values were used for analysis. The relative gene expression ratio was determined using the ΔΔCt method.

Table 1.

List of the genes and their identification used for RT-qPCR analysis

Gene	Probe ID
Hprt1	Mm00446968_m1
Ppia	Mm02342430_g1
Ubc	Mm01198158_m1
Il1b	Mm00434228_m1
Wnt2	Mm00470018_m1
Krt14	Mm00516876_m1
Krt16	Mm01306668_g1
Krt17	Mm00495207_m1
Pax6	Mm00443081_m1
Mmp3	Mm00440295_m1
Mmp9	Mm00442991_m1
Timp1	Mm01341361_m1
Vcan	Mm01283063_m1
L1cam	Mm00493049_m1

Proteomics study

The proteomic profile of tear samples was studied. The total protein concentration was quantified using the NanodropOne Microvolume UV–Vis Spectrophotometer (reference ND-ONE-W, ThermoFisher Scientific). Tears proteomic preparation was carried out using the SP3 automated protocol on the LT Bravo Liquid handler (Agilent technologies) as described in Müller et.al[28]. Briefly, 3 µg of total protein from each sample were diluted in SDS to a final percentage of 1% and reduced with 5µL of 80 mM DTT for 30 min at 60 °C. The proteins were then alkylated with 5µL of 200 mM IAA for 30 min at 30 °C. Protein capture was performed by adding 5µL of Cytiva Sera-Mag at 100 mg/mL (1:1 mix E3 and E7) and 35µL of acetonitrile. Beads were washed twice with 80% ethanol and once with acetonitrile. Digestion was performed by adding 250 ng trypsin/lysC mix (Promega) in 35µL of 50 mM ABC overnight at 37 °C and was stopped with 5% of formic acid. Peptide recovery was carried out using a magnetic rack prior to loading onto Evotips according to manufacturer instructions. Peptide separation was performed using a EvoSep One liquid chromatography system (EvoSep)[29] with a PepSep C18 column, 15 cm x 150 µm, 1.5 µm (Bruker Daltonics). Elution was carried out over 34-min gradient corresponding at 30 sample per day (SPD) using mobile phase A (0.1% FA in water; Biosolve) and mobile phase B (0.1% FA in acetontrile; Biosolve). Peptides were analyzed using a trapped ion mobility spectrometry quadrupole time-of-flight mass spectrometer (timsTOF HT, Bruker Daltonics) equipped with a nanoelectrospray ion source (Captive spray, Bruker Daltonics) in positive-ion mode. Data were acquired in a Data-Independent Acquisition (DIA) mode with an isolation window width of 50 m/z and a TIMS accumulation time of 100 ms. The full scan was performed in the mass range 346–1146 m/z and ion mobility range 0.65–1.41 1/Ko. Protein identification was then performed with DIA-NN software (version 1.8.1). The parameters used were the following: the digestion enzyme is trypsin, the number of missed cleavages was 1, the minimum peptide size was 7 amino acids, the precursor charge range was set from 1 to 4, the precursor m/z set from 300 to 1800 and a protein identification FDR was set at 1%. The Uniprot database of the mouse proteome was used as reference (Release_2024_03). Some modifications, induced by the sample preparation protocol, were studied: asparagine deamidation and methionine oxidation as variable modifications and cysteine carbamidomethylation as fixed modification. The maximum number of variable modifications was set to 2. LFQ intensities obtained were processed using the Perseus software (version 1.6.15.0) and DIA-Analyst platform (version 0.8.6). The mean intensity of the three technical replicates was calculated and used for statistical analysis. The LFQ data for each protein were transformed by applying the log2(x) formula and grouped according to condition (CTRL, T6, T12, T18 and T24) for each eye (left and right). For biological replicates, proteins with 100% of valid values were selected. A cutoff of the adjusted p-value of 0.05 (t-statistic correction) along with a log2(fold change) of 1 has been applied to determine significantly regulated proteins in each pairwise comparison for paired samples.

ImmunoHistoChemistry (IHC) studies

The following antibodies were used for IHC studies: rabbit anti-KRT12 (1:200, reference MA5-42701, Invitrogen) and chicken anti-KRT14 (1:500, reference 906004, BioLegend) as primary antibodies, goat anti-Rabbit IgG H&L Alexa Fluor 488 (1:500, reference A11008, Fisher Scientific) and goat anti-chicken IgY H&L Alexa Fluor 568 (1:500, reference ab175711, Abcam) as secondary antibodies respectively. Nuclei were counterstained with Hoechst 33342 (1:5000, reference H3570, Thermofisher Scientific).

Immunofluorescence labeling on whole cornea

To study the expression of KRT12 and KRT14 in the corneal epithelium, eyes were rehydrated for 2H in 50% ethanol/PBS and washed two times in PBS for 10 min each at room temperature. After dissection, corneas were blocked and permeabilized with a 5% Goat Serum (GS) (reference 16210064, Thermo Fisher Scientific) 5% Fish Skin Gelatin (FSG) (reference G7765, Sigma-Aldrich) in 0.5% Triton X-100/PBS solution, with agitation for 1H at room temperature. Corneas were incubated in primary antibody diluted in blocking solution (5% GS 5% FSG in 0,1% Triton X-100/PBS) overnight at 4 °C with agitation and rinsed in 0,1% Triton X-100/PBS at room temperature 3 times for 1H each. Next, samples were incubated in secondary antibody diluted in blocking solution overnight at 4 °C on agitation and rinsed 3 times for 1H each in 0,1% Triton X-100/PBS at room temperature. After the washes, nuclei were stained for 10 min with Hoechst 33342 and washed 5 min in PBS. Corneas were then cut at four cardinal points with a carbon steel surgical blade (15C, reference 0221, Swann-Morton) and flat mounted in Vectashield medium (Vector laboratories, H-1000), with the epithelium facing the coverslip.

Images acquisition and processing

The acquisition of the images was performed as previously described[30]. A Leica Thunder Imager Tissue microscope was used to acquire the whole-cornea images, using the navigator module with the large volume computational clearing (LVCC) process. The LAS X software (version 3.7.4) was used to obtain the images using a 20X/0.55 objective. Imaris Bitplane software (version 9.8.0) was used to process the images. All of the images from a single panel were acquired and processed with the same parameters. For wound healing monitoring, the Fiji measurement plug-in (FIJI (RRID:SCR_002285)) was used to determine the size of the wounded area, highlighted with the fluorescein staining.

Gene Ontology analysis

To analyze the modulation of biological processes (BP) from the Gene Ontology (GO) database and the RNA-Seq data on corneas, statistical analyses were performed using R and RStudio (version 4.4.1). The BiocManager package from the Bioconductor open source software project was used (clusterProfiler and AnnotationDbi packages). Genes with a p-adj < 0.05, baseMean > 50 and log2(FoldChange) > 1 (for upregulated genes) or log2(FoldChange) < 1 (for downregulated genes) were chosen. Genes were then associated with biological processes and an overrepresentation (ORA) analysis was performed using the GO database, with the Mus musculus reference genome as the background dataset (org.Mm.eg.db). Results are presented as dot plots and show the top 15 upregulated or downregulated biological processes.

Statistical analysis

Data were analyzed using GraphPad Prism software (version 10.1.2, Prism, CA, USA) and expressed as the mean ± SD as indicated in the Figure legend. Statistical differences were tested using unpaired two-tailed t-test, or two-way repeated measures ANOVA with Geisser-Greenhouse correction followed by Tukey’s multiple comparisons test, as indicated in the Figure legend. Significant p-values were represented as *p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001.

Results

Corneal abrasion triggers profound molecular changes of the cornea and lacrimal gland

To gain deeper insights into the molecular landscape underlying corneal epithelial wound healing, we performed mechanical debridement[26] and analyzed the transcriptomic signatures of the cornea and lacrimal gland, along with the proteomic profile of the tear film (Fig. 1a, Fig. S1a). Consistent with previous reports[31], the corneal transcriptome exhibited rapid and significant changes within the first 48 h of the wound healing process (Fig. 1b, Fig. S1b). Specifically, at 18 h post-abrasion, 1227 differentially expressed genes (DEGs) showed at least a twofold increase in expression, while 1288 DEGs displayed a reduction of at least 50% (0.5 fold). Notably, among these DEGs, Flna (MAPK signaling), Stat1, Stat2 (JAK-STAT signaling), Tgfb1, Bmp6 (TGF-β signaling), Itga6, Itgb4 (PI3K-Akt and focal adhesion signaling), Vegfa (VEGF signaling), Wnt3a and Wnt5b (Wnt signaling) expression was upregulated, while Wnt2, Wnt11, Senp2 (Wnt signaling), Erbb3, Erbb4 (MAPK signaling) and Lifr (JAK-STAT and stem cells pluripotency signaling) expression was downregulated. The extensive modulation of the corneal transcriptome led to a reduction in RNA and protein processing (Fig. S2) and an upregulation of immune response effectors (Fig. S3), as revealed by the biological processes analysis based on the Gene Ontology (GO) database.

Fig. 1.

Corneal abrasion induces molecular changes in lacrimal gland, tears and cornea. a Schematic overview of the experimental design. Abrasion was performed unilaterally. Samples (i.e. lacrimal glands, tears and corneas) from the abraded side (ipsilateral) were taken before the abrasion as a control and at 6H and 18H post-abrasion. b Volcano plot representations of lacrimal gland RNA-Seq analysis (n = 3 per group), tears proteomics analysis by mass spectrometry (n = 7 per group) and cornea RNA-Seq analysis (n = 5 per group) at 6H and 18H post-abrasion vs control. For RNA-Seq volcano plots, statistically significant genes are represented only. For proteomics volcano plots, all identified proteins are represented and the limit of statistical significance is shown with the horizontal dotted line. Genes and proteins with a fold change lower than 0.5 are colored in blue and their number is indicated in the top left corner of each plot. Genes and proteins with a fold change higher than 2 are colored in red and their number is indicated in the top right corner of each plot. The genes and proteins with an intermediate fold change are colored in grey. c Histograms of RT-qPCR analysis on cornea samples at 18H post-abrasion vs control (n = 7 per group). Data are represented as mean ± SD and normalized to a reference gene (Ppia). Statistical significance was assessed by unpaired two-tailed t-test (*, p < 0.05; **, p < 0.01; ***, p < 0.001, ****, p < 0.0001)

Our transcriptomic analysis of the cornea revealed as well the modulation of enzymes catalyzing the production of modified nucleosides (Table S1). The methyltransferase-like (METTL) proteins (“writers”) promote the formation of RNA modifications, such as N6-methyladenosine (m⁶A) or N3-methyladenosine (m³A). Reversible modifications, such as m⁶A and 3-methylcytosine (m³C), can be removed by demethylases (“erasers”). Interestingly, the expression of m⁶A writers like Mettl7b and Mettl16 was significantly increased up to 24 and 48 h post-abrasion respectively, while the expression of Mettl14 was significantly decreased at 6 and 12 h post-abrasion only. Conversely, the expression of erasers such as Fto was significantly decreased, starting at 12 h post-abrasion. Following these observations, we analyzed the post-transcriptional RNA modifications, known as epitranscriptomics.

Beyond RNA abundance, RNA chemistry represents a novel layer of post-transcriptional gene regulation, playing a key role in cellular adaptation to stress[32]. However, little is known about the post-transcriptional variables shaping the corneal response to stress and healing. To gain deeper insights on this topic, we analyzed total RNA epitranscriptomic modifications in the cornea (Table S2), as these reflect alterations in RNA biology and dynamics[33, 34]. We observed epitranscriptomic modifications affecting various nucleosides at 18 and 72 h post-injury. Notably, these modified nucleosides were significantly reduced following abrasion, with most modifications corresponding to methylation events, including N1-methyladenosine (m¹A), 3-methylcytosine (m³C), 7-methylguanosine (m⁷G) and N6,N6-dimethyladenosine (m^6,6A).

The lacrimal gland exhibited subtle yet significant changes in its transcriptomic profile, displaying only 3 DEGs (2 with at least a twofold increase in expression and 1 with a reduction of at least 50%) at 6 h post-injury but more than 200 DEGs (85 with at least a twofold increase in expression and 144 with a reduction of at least 50%) at 18 h post-injury (Fig. 1b). In contrast, tear proteomic analysis showed a slightly stronger response at 6 h post-injury, with 41 proteins exhibiting at least a twofold change in expression. This intensified by 12 h post-injury, and the expression profile subsequently remained stable at 18 and 24 h post-injury (Fig. 1b, Fig. S1b). Notably, at 18 h post-injury, the expression of growth factors such as EGF, an activator of various signaling pathways (MAPK, PI3K-Akt, JAK-STAT, HIF-1, focal adhesion…), and to a lesser extent GDF5 (activator of the TGF-β pathway), was significantly increased. More specifically, from 6 to 24 h post-abrasion, Egf expression was significantly downregulated in the cornea, but not modified in the lacrimal gland, and the presence of EGF in the tears was significantly increased, reflecting a transfer of the trophic support from intrinsic (autrocrine and/or paracrine) to extrinsic (brought by the lacrimal gland or the conjunctiva). Thus, these transcriptomic and proteomic profiles highlight a dynamic modulation of tear film composition, as well as molecular alterations in the lacrimal gland and cornea throughout the wound healing process.

To further investigate the molecular response to injury, we conducted qPCR analyses on eleven selected DEGs with known relevance to corneal biology at 18 h post-injury (Fig. 1c). Specifically, we examined two secreted ligands: Il-1β, a known marker of corneal inflammation[35], and Wnt2, expressed by limbal stem cells[36]; one transcription factor, Pax6, critical for corneal epithelial cell identity[37]; three intracellular markers, Krt14, Krt16, and Krt17, indicative of epithelial differentiation states and stress[38, 39]; and five extracellular matrix (ECM)-associated factors: Mmp3 (essential for directional epithelial cell migration[40]), Mmp9 (involved in collagen remodeling[41]), Timp1 (an inhibitor of MMP9 activity[42]), Versican (Vcan), which participates in ECM remodeling[43], and L1cam, an ECM component implicated in cell migration and cohesiveness[44].

At 18 h post-abrasion, wound healing is actively underway, and significant molecular changes have already commenced, as indicated by our global molecular data. The qPCR analysis demonstrated substantial modulation in the expression of all examined genes. Notably, Wnt2, Pax6, and L1cam were downregulated, whereas the other genes showed increased expression (Il-1β, Krt14, Krt16, Krt17, Mmp3, Mmp9, Timp1, Vcan). In particular, the elevated expression of Krt14, a marker of undifferentiated corneal epithelial cells[45], and decreased expression of Pax6, crucial for maintaining corneal epithelial cell identity, strongly support a dedifferentiation process during the initial phase of wound healing.

To confirm cellular-level dedifferentiation, immunostainings for KRT14 and KRT12, a marker for terminally differentiated corneal epithelial cells[46], were performed on whole-mounted corneas (Fig. 2a, b). At 18 h post-injury, we observed elevated KRT14 presence primarily localized at the limbus and wound margins, in contrast to its restriction to the limbus in unwounded corneas. Remarkably, Krt14 + cells persisted and extended across the entire corneal epithelium at 72 h post-injury, despite the complete closure of the epithelial wound. KRT12 presence was observed in the whole cornea except for the limbus and some specific regions of the wound border, where KRT14 was also present. This indicates the coexistence of undifferentiated and differentiated epithelial cell populations during wound healing.

Fig. 2.

Corneal abrasion induces a strong expansion of KRT14 expression pattern in the regenerating epithelium. a Schematic overview of the experimental design. Abrasion was performed unilaterally. Samples (i.e. corneas) from the abraded side (ipsilateral) were taken before the abrasion as a control and at 18H and 72H post-abrasion. b Wholemount immunofluorescence staining for differentiated (KRT12) and undifferentiated (KRT14) corneal markers. The abraded area is delimited by the white dotted line. Data are representative of 3 biological replicates. Nuclei were counterstained with Hoechst. Scale bars are 500 µm

Collectively, our findings demonstrate that corneal abrasion triggers significant changes in the corneal and lacrimal gland transcriptomes as well as the tear film proteome, and they point to a dedifferentiation process in the early wound-healing phase. Additionally, the lacrimal gland and tear film also undergo considerable, though less pronounced, molecular remodeling.

Moderate impact of tear film on corneal molecular dynamics during healing

Our next objective was to elucidate the specific role of the tear film by performing extraorbital lacrimal gland ablation alone or combined with corneal abrasion (Fig. 3a). Lacrimal gland excision (LGE) models have been used in previous studies to induce dry eye and evaluate the impact on corneal homeostasis[47, 48], but the consequences during the wound healing process after abrasion have yet to be elucidated. Here, we validated our surgery by analysing the wound closure rate, which was significantly impaired in the tear-deficient condition, as expected from an altered corneal microenvironment (Fig. 3b). To limit confounding effects from dry eye conditions caused by prolonged gland ablation, molecular and cellular analyses were conducted at 18 and 72 h post-ablation. An immunostaining for KRT14 was performed on whole-mounted corneas (Fig. 3c), indicating an expansion of KRT14 expression in peripheral cornea and at the woud margin at 18 h post-surgery, comparable to conditions with intact tear production. Furthermore, the expression of KRT14 expanded across the entire cornea at 72 h post-surgery, paralleling observations made after abrasion with an intact lacrimal gland (Fig. 2).

Fig. 3.

The absence of tears after a corneal abrasion delays the epithelial wound closure. a Schematic overview of the experimental design. Abrasion was performed unilaterally and the extraorbital lacrimal gland was removed on the ipsilateral side at the same time. Samples (i.e. corneas) from the wounded side (ipsilateral) were taken before the surgery as a control and at 18H and 72H post-surgery. b Wound closure quantification of the abraded epithelial surface from the abrasion up to 7 days after the abrasion (DPAb) (n = 5 per group). Data are represented as mean ± SD. Statistical significance was assessed by two-way repeated measures ANOVA with Geisser-Greenhouse correction and Tukey’s multiple comparisons test (**, p < 0.01). c Wholemount immunofluorescence staining for differentiated (KRT12) and undifferentiated (KRT14) corneal markers. The abraded area is delimited by the white dotted line. Data are representative of 3 biological replicates. Nuclei were counterstained with Hoechst. Scale bars are 500 µm

We analyzed the gene expression of the eleven candidates we previously selected in the cornea, in lacrimal gland-ablated conditions alone. Our findings showed that Krt17 and L1cam expression were mildly impacted (up and down, respectively) by lacrimal gland ablation at 18 h post-surgery (Fig. S4a, Fig. S4b). No other candidates from our list were modulated.

When combining lacrimal gland ablation and corneal abrasion, our 11 selected genes were significantly impacted by abrasion, even with an impaired tear production (Fig. 4a, b). Notably, we observed significant changes in the corneal expression of Wnt2, Pax6 and Mmp9. While the downregulation of Pax6 and upregulation of Mmp9 were enhanced in the tear-deficient condition, Wnt2 expression was unexpectedly higher during wound healing with an altered tear film compared to the tear-intact condition (Fig. 4c).

Fig. 4.

The absence of tears after a corneal abrasion mildly alters the molecular pattern of the corneal wound healing process. a Schematic overview of the experimental design. Abrasion was performed unilaterally and the extraorbital lacrimal gland was removed on the ipsilateral side at the same time. Samples (i.e. corneas) from the wounded side (ipsilateral) were taken before the surgery as a control and at 18H post-surgery. b Histograms of RT-qPCR analysis on cornea samples at 18H post-surgery vs control (n = 7 per group). Data are represented as mean ± SD and normalized to a reference gene (Ppia). Statistical significance was assessed by unpaired two-tailed t-test (*, p < 0.05; **, p < 0.01; ***, p < 0.001, ****, p < 0.0001). c Histograms of RT-qPCR analysis on cornea samples at 18H post-surgery vs 18H post-abrasion only (n = 7 per group). Data are represented as mean ± SD and normalized to a reference gene (Ppia). Statistical significance was assessed by unpaired two-tailed t-test (*, p < 0.05; **, p < 0.01; ***, p < 0.001, ****, p < 0.0001)

Next, we examined the epitranscriptomic profile following lacrimal gland ablation, considering that tear deficiency alters the corneal microenvironment and may affect post-transcriptional gene regulation (Table S2). At 18 h post-surgery, the alteration of the tear film alone did not significantly affect the studied nucleoside modifications. However, by 72 h post-surgery, four nucleoside modifications were significantly reduced (m³C, m⁷G, m^2,2G, m^6,6A). Importantly, following abrasion combined with lacrimal gland ablation, significant alterations were observed in the epitranscriptomic profile at 18 and 72 h post-surgery compared to physiological controls, affecting multiple nucleosides. When directly comparing conditions of corneal abrasion with intact and deficient tear prodution (Table S3), only two nucleoside modifications differed significantly between the groups. Noteworthy, m⁶A levels were significantly decreased in the tear-deficient condition at both time points.

Collectively, these findings underscore the critical role of the tear film and a functional corneal microenvironment for efficient wound closure. Our data also reveal distinct molecular changes between wound healing conditions with an intact or an altered tear presence, emphasizing the nuanced role tears play in the healing process.

Bilateral molecular response after unilateral corneal injury

Previous studies have provided evidence of a bilateral response to unilateral corneal injury. For instance, in zebrafish, a second abrasion inflicted on the contralateral eye one hour after injuring the ipsilateral eye led to accelerated wound closure[49]. In the mouse, the relative expression of specific genes in the lacrimal gland was impacted both ipsilaterally and contralaterally 18 h after a unilateral abrasion[50]. However, such a bilateral molecular response has not yet been characterized in mammalian corneal and tear film contexts. To address this gap, we investigated molecular changes occurring in the corneal microenvironment of the contralateral eye following unilateral corneal abrasion.

We first analyzed the transcriptomic profile of the cornea and the proteomic profile of tears from the contralateral eye (Fig. 5a, Fig. S5a). Transcriptomic analysis revealed early, moderate changes, with significant modulations most pronounced at 24 h post-abrasion, later than the ipsilateral response peak. Proteomic data similarly indicated early alterations with intermediate modulation of significant proteins (Fig. 5b, Fig. S5b). Among the DEGs exhibiting at least two-fold increase or 0.5 fold decrease in expression, Stat1, Stat2 (JAK-STAT signaling) Notch4 (Notch signaling), Wnt3a and Wnt6 (Wnt signaling) expression was upregulated, while Ngf (MAPK and PI3K-Akt signaling), Tnf (TGF-β and NF-kB signaling), Fas (MAPK and apoptosis signaling) and Wnt2 (Wnt signaling) expression was downregulated. Biological processes analysis based on the GO database showed broad similarity with ipsilateral changes (Fig. S6, Fig. S7).

Fig. 5.

Corneal abrasion induces contralateral molecular changes in tears and cornea. a Schematic overview of the experimental design. Abrasion was performed unilaterally. Samples (i.e. tears and corneas) from the contralateral side of the abrasion were taken before the abrasion as a control and at 6H and 18H post-abrasion. b Volcano plot representations of contralateral tears proteomics analysis by mass spectrometry (n = 7 per group) and contralateral cornea RNA-Seq analysis (n = 5 per group) at 6H and 18H post-abrasion vs control. For RNA-Seq volcano plots, statistically significant genes are represented only. For proteomics volcano plots, all identified proteins are represented and the limit of statistical significance is shown with the horizontal dotted line. Genes and proteins with a fold change lower than 0.5 are colored in blue and their number is indicated in the top left corner of each plot. Genes and proteins with a fold change higher than 2 are colored in red and their number is indicated in the top right corner of each plot. The genes and proteins with an intermediate fold change are colored in grey. c Histograms of RT-qPCR analysis on contralateral cornea samples at 18H post-abrasion vs control (n = 7 per group). Data are represented as mean ± SD and normalized to a reference gene (Ppia). Statistical significance was assessed by unpaired two-tailed t-test

We expanded these observations through epitranscriptomic analyses of contralateral corneas, identifying no significant modifications at 18 h post-injury (Table S2). At this time point, qPCR analyses demonstrated no significant bilateral alterations in our selected genes compared to controls (Fig. 5c). Cellular-level validation through immunostaining for KRT14 and KRT12 further supported this result, showing no clear bilateral expansion of KRT14 expression, despite a slight enhancement restricted to peripheral cornea, and no change in KRT12 expression domains at 18 and 72 h post-injury (Fig. 6a, b).

Fig. 6.

Corneal abrasion induces a mild expansion of KRT14 expression pattern in the contralateral peripheral epithelium. a Schematic overview of the experimental design. Abrasion was performed unilaterally. Samples (i.e. corneas) from the contralateral side of the abrasion were taken before the abrasion as a control and at 18H and 72H post-abrasion. b Wholemount immunofluorescence staining for differentiated (KRT12) and undifferentiated (KRT14) corneal markers. Data are representative of 3 biological replicates. Nuclei were counterstained with Hoechst. Scale bars are 500 µm

Finally, direct comparisons between ipsilateral and contralateral sides revealed substantial transcriptomic discrepancies, characterized by significant and sustained differences in gene expression persisting up to 48 h post-injury (Fig. S8a, Fig. S8b), as further validated by qPCR analyses (Fig. S8c). At 18 h post-abrasion, 26% of all DEGs showing at least a twofold increase in the wounded eye were also upregulated in the contralateral eye (317 DEGs), whereas 74% were specifically upregulated only in the wounded eye. Conversely, among DEGs exhibiting a reduction of 50% or more, 20% were commonly downregulated in both eyes (259 DEGs), while the remaining 80% were specifically downregulated in the wounded eye (Table S4, Table S5, Table S6). These commonly regulated genes represented 62% of the upregulated DEGs and 53% of the downregulated DEGs detected in the contralateral eye (Fig. S9a, Fig. S9b).

Biological processes commonly downregulated in both eyes differed notably from those exclusively downregulated in the wounded eye. However, the biological processes upregulated in both eyes were consistently associated with immune responses, as indicated by GO analysis (Fig. S10). Despite transcriptomic differences, proteomic profiles of tear fluids remained largely comparable between eyes (Fig. S11a, b). Analysis of significant proteins indicated that 65% of the proteins upregulated (272 proteins) and 63% of the proteins downregulated (278 proteins) in the wounded eye showed significant changes also in the contralateral eye, representing respectively 84% and 79% of the contralaterally regulated proteins (Tables S7, S8, S9). Therefore, 16% of the proteins upregulated and 21% of those downregulated in the contralateral eye lacked corresponding counterparts in the wounded eye (Fig. S9c, d). This suggests that a distinct subset of tear film proteins remains exclusive to each eye, reflecting differential responses or impacts on the cornea. In contrast, immunostaining analyses revealed pronounced KRT14 expansion exclusively on the ipsilateral side, while it was restricted to peripheral cornea on the contralateral side (Fig. S12a, b).

Taken together, these results indicate a significant bilateral molecular response to unilateral corneal injury at transcriptomic and proteomic levels. However, this bilateral effect is not evident at the cellular scale, suggesting a unique and localized molecular signature associated with direct injury exposure.

Tears constrain bilateral injury response

Finally, we sought to clarify the specific contribution of the tear film to the bilateral molecular response observed during corneal wound healing. To address this, we performed unilateral corneal abrasion with or without contralateral extraorbital lacrimal gland ablation and examined the expression of the previously selected genes by qPCR in the contralateral cornea at 18 h post-surgery (Fig. 7a). Notably, Il-1β, Wnt2, Pax6, and L1cam expression levels remained unchanged in the contralateral cornea with deficient tear production after abrasion, compared to physiological controls (Fig. 7b). However, further comparisons between contralateral corneas with intact or impaired tear production revealed significant modulations specifically in genes associated with extracellular matrix remodeling (Mmp3, Mmp9, Timp1, Vcan) and cytoskeletal markers (Krt14 and Krt16), all showing increased expression in the tear-deficient condition (Fig. 7c).

Fig. 7.

The lack of tears on the contralateral side after a corneal abrasion widely alters the molecular state of the unwounded cornea. a Schematic overview of the experimental design. Abrasion was performed unilaterally and the extraorbital lacrimal gland was removed on the contralateral side at the same time. Samples (i.e. corneas) from the contralateral side of the abrasion were taken before the surgery as a control and at 18H post-surgery. b Histograms of RT-qPCR analysis on contralateral cornea samples at 18H post-surgery vs control (n = 7 per group). Data are represented as mean ± SD and normalized to a reference gene (Ppia). Statistical significance was assessed by unpaired two-tailed t-test (*, p < 0.05; **, p < 0.01; ***, p < 0.001). c Histograms of RT-qPCR analysis on contralateral cornea samples at 18H post-surgery vs 18H post-abrasion only (n = 7 per group). Data are represented as mean ± SD and normalized to a reference gene (Ppia). Statistical significance was assessed by unpaired two-tailed t-test (*, p < 0.05; **, p < 0.01)

Expanding our molecular analysis, we evaluated the epitranscriptomic profile of the contralateral corneas (Table S2). At 18 h post-surgery, lacrimal gland ablation alone did not significantly affect the selected nucleoside modifications. A minor but notable reduction was observed in specific nucleosides at 72 h post-ablation (m³C, m⁷G, m^2,2G, m^6,6A). When combining abrasion with contralateral gland ablation, the epitranscriptomic profile at 18 h post-surgery indicated only modest changes, with a significant reduction limited to the m³C nucleoside.

We complemented these molecular findings with immunostainings for KRT14 and KRT12 on whole-mounted contralateral corneas with intact or altered tear production (Fig. 8a, b). Despite qPCR results indicating elevated Krt14 expression, immunostaining showed only a mild expansion localized to peripheral cornea in the tear-deficient condition, without widespread bilateral alterations in KRT12 or KRT14 expression at either 18 or 72 h post-abrasion.

Fig. 8.

The lack of tears on the contralateral side after a corneal abrasion slightly affects KRT14 expression pattern in the peripheral unwounded cornea. a Schematic overview of the experimental design. Abrasion was performed unilaterally and the extraorbital lacrimal gland was removed on the contralateral side at the same time. Samples (i.e. corneas) from the contralateral side of the abrasion were taken before the surgery as a control and at 18H and 72H post-surgery. b Wholemount immunofluorescence staining for differentiated (KRT12) and undifferentiated (KRT14) corneal markers. Data are representative of 3 biological replicates. Nuclei were counterstained with Hoechst. Scale bars are 500 µm

Together, our data suggest that tear film presence plays a moderating role on gene expression changes in the contralateral cornea, regulating extracellular matrix remodeling and cytoskeletal rearrangements during the bilateral response to unilateral corneal injury. This regulatory mechanism likely helps prevent excessive molecular remodeling in the contralateral cornea and maintains tissue homeostasis.

Discussion

Epithelial tissues maintain critical biological functions, from serving as protective barriers to regulating selective absorption and secretion, by continuously undergoing homeostatic renewal. This intrinsic regenerative capacity becomes particularly crucial following injury, activating complex cellular and molecular mechanisms to restore tissue integrity rapidly and efficiently. However, disrupted or impaired healing can lead to severe pathologies, including chronic wounds and fibrosis, highlighting the need for a comprehensive understanding of the underlying regulatory networks. Clarifying these processes at both cellular and molecular levels is essential to developing targeted therapies aimed at enhancing epithelial regeneration and preventing pathological outcomes.

Corneal abrasion, characterized by a transient rupture in the cellular cohesion and the integrity of the epithelial barrier, represents one of the most common threats to corneal homeostasis. Typically, this type of injury resolves rapidly, within a few days, thanks to a robust wound healing process that relies heavily on the corneal microenvironment. During corneal repair, epithelial cells produce critical cellular factors essential for intra-tissue crosstalk, innervation provides neurotrophic factors, and the tear film supplies essential growth factors. Corneal inflammation also plays a central role in wound healing efficiency and is finely regulated by inflammatory cytokines, found both in cornea and tears. Under these conditions, corneal transparency is generally maintained. However, complications can arise, leading to clinical issues such as persistent epithelial defects or poorly-healing ulcerations. This range of outcomes—from efficient healing to severe complications—highlights the need to thoroughly understand the underlying cellular and molecular mechanisms of the wound healing process, including the involvement of the entire microenvironment, an aspect frequently overlooked in previous research work. Here, we provide a dynamic and comprehensive molecular analysis of the cornea and its microenvironment during wound healing following corneal abrasion and extraorbital lacrimal gland ablation.

Our findings demonstrate rapid and extensive changes in both the corneal transcriptome and tear proteome, detectable from 6 h and persisting up to 48 h post-abrasion. These molecular shifts impacted numerous pathways, including TGF-β, HIF-1, and VEGF. Activation of the TGF-β pathway has previously been associated with modulating corneal epithelial cells migration and proliferation, keratocytes proliferation, and their transdifferentiation into myofibroblasts[51]. Several studies have underscored the critical roles of VEGF in nerve regeneration[52] and implicated HIF-1 in corneal pathophysiology[53], though its precise role remains to be clearly defined.

Our data demonstrate the activation of the JAK-STAT signaling pathway, as evidenced by both corneal transcriptomic and tear proteomic analyses. Growth factors typically activate this pathway through their interaction with receptor tyrosine kinases (RTKs), such as epidermal growth factor (EGF) and its receptor EGFR. Given the marked decrease in Egf gene expression within the cornea and the concurrent elevation of EGF protein in tears, our results suggest that JAK-STAT pathway activation in the cornea is mediated primarily by enhanced secretion of EGF into the tear film. This supports the essential role of the tear microenvironment in modulating corneal healing processes. A previous study on corneal epithelial cells also showed that EGF is able to induce Pax6 downregulation, which is in turn needed to control EGF-induced proliferation[54], supporting our hypothesis of a crosstalk between cornea and lacrimal gland, although more investigations should be conducted to address the connections between growth and transcription factors in corneal homeostasis. Moreover, the rapid and efficient modulation of Egf transcript levels may be linked to a distinct epitranscriptomic signature, aligning with the significant RNA stability alterations observed in the cornea. Additionally, although the existence of a unified stress response specifically activating these pathways has not been demonstrated to date, it cannot be excluded. Based on our findings, we hypothesize that during corneal wound healing, the epithelium undergoes transcriptomic reprogramming to promote cellular plasticity, while the lacrimal gland simultaneously enhances its trophic support to facilitate tissue repair.

Additionally, the upregulation of stromal components such as Mmp3, Mmp9, Timp1, and Vcan at 18 h post-abrasion implies significant stromal remodeling, which is essential for keratocyte activation, migration, and transdifferentiation[55]. Also involved in keratinocyte pathophysiology, the stress keratins genes Krt16 and Krt17 were upregulated after abrasion, which is consistent with their known function in adhesion, migration and proliferation during epithelial wounds[39], hinting towards a possible equivalent function in corneal keratocytes. Consistent with previous research work, the healing cornea exhibited increased expression of Krt14, a marker of undifferentiated epithelial cells[45], along with decreased expression of Pax6, which is crucial for maintaining epithelial identity. These observations strongly suggest a dedifferentiation process occurring in the early phases of wound healing.

The lacrimal gland, as part of the corneal microenvironment, produces proteins secreted into the tear film, crucial for maintaining corneal homeostasis. Although tear proteomics studies have expanded recently, particularly in dry eye diseases[56], very few have addressed tear proteomic changes during corneal wound healing in vivo. Our study reveals significant molecular differences between healing processes with intact or deficient tears. Notably, Pax6 downregulation was more pronounced with an altered tear film, suggesting a greater loss of epithelial identity. Additionally, Wnt2 downregulation was less extensive with a deficient tear production, indicating tears might enhance inhibition of the Wnt signaling pathway during healing. Wnt/β-catenin signaling contributes to corneal epithelial homeostasis by regulating proliferation and stratification[57]. Among others, Wnt2 is expressed preferentially in the limbus, where the limbal epithelial stem cells are located, and acts as a regulator of cell proliferation and cell fate[58]. The marked Wnt2 downregulation after abrasion suggests an activation of limbal stem cell niches, complementing epithelial dedifferentiation at wound margins, thus collectively facilitating wound closure. This correlates with our observation of a significantly reduced wound closure rate in the tear-deficient condition, emphasizing tears’ essential role in activating stem cell niches and regulating epithelial identity for efficient healing.

Transcriptomic regulation plays a crucial role in numerous cellular processes underlying tissue development, homeostasis, and function. Post-transcriptional RNA modifications, collectively referred to as the epitranscriptome, occur across diverse RNA species, including messenger RNA (mRNA) and non-coding RNAs (e.g. microRNA). These biochemical nucleotide modifications critically influence RNA stability, splicing, translation efficiency, localization, and degradation, thereby affecting gene expression profiles and cellular physiology. Our findings specifically demonstrated significant alterations within the corneal epitranscriptome following injury, notably a reduction in RNA methylations such as m¹A, m⁶A, m³C, and m⁷G. N1-methyladenosine (m¹A), a reversible modification frequently observed in tRNA, rRNA, and mRNA, impacts RNA stability by altering its secondary structure, and subsequently influences protein translation efficiency. Specifically, m¹A typically promotes translation by stabilizing RNA molecules and facilitating ribosome binding[59, 60]. Our results indicated a notable decrease in m¹A modification at both 18 and 72 h post-abrasion, suggesting diminished RNA stability that may enhance cellular plasticity during the healing process. Importantly, the impairment of tear production post-injury did not alter m¹A levels compared to tear-intact healing conditions, indicating that m¹A modulation occurs independently of tears. Conversely, N6-methyladenosine (m⁶A)—the most abundant reversible RNA modification in eukaryotic cells, pivotal in dynamic transcriptome regulation—generally supports mRNA stability and translation efficiency[61]. Our analysis showed that m⁶A levels remained unaffected by corneal abrasion alone but significantly decreased during healing in tear-deficient conditions, suggesting partial dependence of m⁶A stability on tear presence. These observations underscore the regulatory role of tears in maintaining RNA stability and cellular plasticity after injury by modulating specific RNA modifications, particularly m⁶A. Previous research has implicated m⁶A in processes such as cell differentiation and stem cell self-renewal[62]. Accordingly, our study associates tear film alteration during wound healing with reduced m⁶A levels, hinting at a possible influence of tears on cell differentiation during wound closure. At the transcriptomic level, the expression of writers and erasers of m⁶A in the cornea was significantly affected by the abrasion. The expression of Mettl3 and Mettl14, writers of m⁶A, was significantly reduced at 6 h post-abrasion. Interstingly, Mettl3 expression was then upregulated from 12 to 48 h post-injury, both in the wounded side and the contralateral side. The METTL3 and METTL14 proteins form a heterodimer complex that generates m⁶A modifications in mRNA, with METTL3 holding the catalytic activity and METTL14 supporting the complex stability and facilitating RNA target binding. We also revealed a marked Mettl7b upregulation during corneal wound healing, with at least a ten-fold increase between 6 and 18 h post-abrasion. The METTL3/METTL14 complex[63, 64] and the METTL7B protein[65, 66], through m⁶A methylation, have been involved in cell proliferation and cell migration in various types of tumors and cancers. Consequently, the upregulation of Mettl3 and Mettl7b in cornea after abrasion could facilitate the proliferation and migration of epithelial cells, which is needed for efficient wound closure. Conversely, the expression of Fto, eraser of m⁶A, was significantly reduced during the wound healing process. The downregulation of FTO has been linked to cancer malignancy and tumorigenesis by promoting proliferation and migration[63, 67, 68]. In the context of cornea wound healing, the downregulation of Fto could further support the increased cell dynamics required for efficient woud closure. Nonetheless, more investigations are essential to fully elucidate the functional implications of the epitranscriptome in wound healing and associated cellular differentiation processes.

Evidence of a bilateral ocular response following unilateral injury has previously been documented in a zebrafish model[49, 50], yet such phenomena have not been extensively explored in mammalian systems. To address this gap, we analyzed molecular profiles in the contralateral eye following corneal abrasion. We identified a bilateral molecular response starting as early as 6 h post-abrasion and persisting up to 48 h, detectable through corneal transcriptomic and tear proteomic analyses. However, this bilateral modulation involved distinct molecular candidates that remain to be fully characterized, as none of the specifically selected candidate genes showed contralateral expression variations. Notably, in the contralateral eye, increased expression of genes involved in the Notch signaling pathway (Notch4, Nrarp, Rbpjl) hinted towards an activation of this pathway from 12 to 48 h post-injury. Previous research has associated the Notch pathway with maintaining corneal epithelial cell fate[69], and controlling proliferation and differentiation[70, 71]. Our data support a robust maintenance of epithelial cell identity in the contralateral eye during the wound healing process.

Moreover, significant differences emerged when comparing corneal transcriptomic profiles between wounded and contralateral eyes. At 18 h post-injury, only 26% of differentially expressed genes (DEGs) showing at least a twofold increase in the wounded eye were also upregulated contralaterally. Interestingly, such transcriptomic discrepancies were less evident in tear proteomic profiles, with both eyes maintaining relatively similar molecular profiles—65% of upregulated proteins in the wounded eye were similarly modulated in the contralateral eye. Nonetheless, 35% of the proteins upregulated in the wounded eye and 16% of those upregulated in the contralateral eye were specifically regulated in only one eye. Similarly, 37% of downregulated proteins in the wounded eye and 21% in the contralateral eye showed eye-specific regulation. These findings indicate that bilateralization involves distinct adaptations of the tear film composition in each eye. Consequently, during corneal wound healing, two uniquely modulated tear films are formed, both having different roles, and differing not only from each other but also from the baseline composition observed in an unwounded tear film. Notably, when tear production was deficient, the contralateral side demonstrated pronounced molecular alterations, underscoring the regulatory role of tears in limiting stromal bilateral responses. Specifically, increased expression of Mmp3, Mmp9, Timp1, and Vcan was observed with an impaired tear production. Consistent with our hypothesis regarding the preservation of cell fate in the contralateral eye during healing, tear deficience led to increased molecular expression of Krt14; however, this was not reflected at the cellular level, as KRT14 expression pattern remained unchanged. We only reported a Krt14 change of expression in the contralateral cornea in the tear-deficient condition. In prior research, the increased cell activity in the contralateral side was evidenced when the contralateral eye was also wounded[72, 73]. Here, the molecular modulations taking place in the contralateral eye were studied without any supplemental injury. In the presence of tears, there was no contralateral modulation of Krt14 expression compared to control. In the tear-deficient condition, Krt14 became significantly upregulated compared to control, and remained significant when comparing wounded and contralateral eyes, suggesting that the tears could be preventing its expression. The comparative analysis of these studies and our results could suggest that the regulatory role of the tears on Krt14 expression, which we hypothesized here, is abolished when the contralateral eye is wounded, leading to a stronger Krt14 expression, a faster dedifferentiation and more efficient wound closure.

Collectively, our findings establish a molecular bilateral response occurring early in the wound healing process, sustained for at least 48 h post-injury. Importantly, we provide evidence that bilateralization is not driven primarily by tear film factors, but rather regulated by them. Further investigation is required to clarify the underlying mechanisms. We hypothesize two potential mechanisms: firstly, bilateral responses may predominantly arise via corneal innervation and be regulated by the tear film; alternatively, systemic signals may reach the limbal niche, subsequently signaling to the broader corneal tissue. Future research exploring the specific role of corneal innervation and potential involvement of systemic factors (such as hormones or immune mediators) will be needed to fully elucidate the nature and pathways of this bilateral phenomenon.

Our study underscores the essential role of the microenvironment in corneal wound healing, providing a detailed molecular characterization of changes occurring in both injured and contralateral corneas and tear fluid. This comprehensive approach highlights the necessity of an intact microenvironment for optimal wound repair, offering critical insights into the pathophysiological processes and potential therapeutic targets for improved wound healing outcomes.

Additionally, this research provides valuable insights relevant to wound healing processes occurring in other epithelial tissues, such as skin. The skin microenvironment, characterized by systemic influences, innervation, resident immune cells, and sweat glands, similarly plays a pivotal role in determining wound responses. Exploring how these diverse factors and their interactions influence skin injury responses would significantly enhance our understanding of the microenvironment’s broader role in complex and multifactorial wound healing scenarios.

Such investigations not only deepen our comprehension of fundamental pathophysiological processes but also pave the way for therapeutic strategies specifically designed to target the microenvironment, promoting quicker and more effective wound healing in various pathological conditions where the microenvironment is compromised.

Conclusions

In conclusion, this study emphasizes the essential role of the microenvironment in wound healing, elucidating profound transcriptomic and proteomic changes occurring within the cornea and tear fluid. For the first time, we have extensively characterized molecular events following in vivo corneal abrasion, encompassing both injured and contralateral sides, thereby establishing the crucial requirement of an intact microenvironment for efficient wound closure.

Abbreviations

ABC

Ammonium bicarbonate

BP

Biological processes

DEGs

Differentially expressed genes

DTT

Dithiothréitol

ECM

Extracellular matrix

EGF

Epidermal growth factor

EGFR

Epidermal growth factor receptor

FDR

False discovery rate

FSG

Fish skin gelatin

Fto

Fat ass and obesity-associated protein

GDF5

Growth differentiation factor 5

GO

Gene ontology

GS

Goat serum

HIF-1

Hypoxia-inducible factor-1

IAA

Iodoacetamide

IHC

Immunohistochemistry

Il1b

Interleukin-1 beta

JAK-STAT

Janus kinase/signal transducer and activator of transcription

KRT12

Keratin 12

Krt14

Keratin 14

Krt16

Keratin 16

Krt17

Keratin 17

L1cam

L1 cell adhesion molecule

LASIK

Laser-assisted in situ keratomileusis

LFQ

Label-free quantification

LGE

Lacrimal gland excision

LVCC

Large volume computational clearing

m¹A

N1-methyladenosine

m³A

N3-methyladenosine

m³C

3-Methylcytosine

m^6,6A

N6,N6-dimethyladenosine

m⁶A

N6-methyladenosine

m⁷G

7-Methylguanosine

MAPK

Mitogen-activated protein kinase

Mettl3

Methyltransferase-like 3

Mettl7b

Methyltransferase-like 7b

Mettl14

Methyltransferase-like 14

Mettl16

Methyltransferase-like 16

Mmp3

Matrix metalloproteinase-3

Mmp9

Matrix metalloproteinase-9

mRNA

Messenger ribonucleic acid

NF-kB

Nuclear factor kappa-light-chain-enhancer of activated B cells

NSAID

Non-steroidal anti-inflammatory drugs

ORA

Overrepresentation analysis

Pax6

Paired box 6

PBS

Phosphate-buffered saline

PI3K-Akt

Phosphatidylinositol-3-kinase Protein kinase B

PRK

Photorefractive keratectomy

RNA

Ribonucleic acid

rRNA

Ribosomal RNA

RTKs

Receptor tyrosine kinases

SD

Standard deviation

SDS

Sodium dodecyl sulfate

TGF-β

Transforming growth factor-beta

Timp1

Tissue inhibitor of metalloproteinases 1

tRNA

Transfer ribonucleic acid

Vcan

Versican

VEGF

Vascular endothelial growth factor

Wnt2

Wnt family member 2

Author contributions

Conceptualization: NF, FM; Methodology: NF, ACM, AK, PM, LF, SLG, AA, NN, EJ, JV, VD, AD, CH, KL, ODK, FM; Validation: NF, FM; Formal analysis: NF, LF, SLG, AA, NN, FM; Data analysis: NF, FM; Writing: NF, ACM, AK, PM, LF, SLG, AA, NN, EJ, JV, VD, AD, CH, KL, ODK, FM; Supervision: KL, FM; Project administration: FM; Funding acquisition: FM.

Funding

This research was supported by ATIP-Avenir program, Inserm, the Région Occitanie, ANR (ANR-21-CE17-0061, TeFiCoPa), FRM (REP202110014140), Support for research: I-SITE 2024—program of excellence of the University of Montpellier, CBS2 Doctoral School, and the Fondation Groupama.

Availability of data and materials

The raw data is available upon request and the datasets produced for the current study are publicly available. The RNA-Sequencing data is available in the GEO repository, under the accession numbers GSE292450 (cornea RNA-Seq) and GSE292547 (lacrimal gland RNA-Seq). The mass spectrometry proteomics data is available in the PRIDE repository, via ProteomeXchange, with the identifier PXD061492.

Declarations

Ethics approval and consent to participate

All the experiments conducted on mice were approved by the local ethical committee and the Ministère de la Recherche et de l’enseignement Supérieur (authorization 2016080510211993 version2). All of the procedures were carried out in accordance with the French regulation for the animal procedure (French decree 2013–118) and with specific European Union guidelines for the protection of animal welfare (Directive 2010/63/EU).

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.