NAMPT Deficiency Promotes Endothelial–Mesenchymal Transition by Inhibiting Energy Metabolism and Lysosomal Acidification in Late-Onset Fuchs Endothelial Corneal Dystrophy

Abstract

Purpose

This study aimed to investigate the role of nicotinamide phosphoribosyltransferase (NAMPT) deficiency in the pathogenesis of Fuchs endothelial corneal dystrophy (FECD) and to explore potential therapeutic strategies.

Methods

NAMPT expression was examined in three GEO datasets (GSE74123, GSE142538, GSE171830) and multiple FECD samples. A UVA-induced late-onset FECD mouse model with NAMPT knockdown was established via intracameral injection of AAV-shNAMPT. Corneal thickness, endothelial morphology, and endothelial-mesenchymal transition (EnMT) status were evaluated. In vitro, a chronic oxidative injury model with NAMPT downregulation was generated using small interfering RNA. RNA sequencing, ATP synthesis, lysosomal acidification, autophagic flux, and V-ATPase subunit expression were assessed to elucidate NAMPT function. Rescue experiments were conducted using NAD⁺ supplementation in vitro and nicotinamide riboside (NR, an NAD⁺ precursor) in vivo.

Results

A significant negative correlation was observed between ATP/energy metabolism and EnMT/extracellular matrix (ECM) remodeling in FECD specimens. NAMPT was consistently downregulated across multiple FECD datasets and the UVA mouse model. NAMPT knockdown induced significant corneal edema, endothelial cell loss, guttae formation, excessive EnMT and thickened Descemet’s membrane in mice. In vitro, NAMPT inhibition altered the expression of energy metabolism-related enzymes, activated ECM remodeling pathways, reduced ATP production, blocked autophagic flux, impaired lysosomal acidification, and reduced V-ATPase expression. Importantly, these functional and molecular defects were ameliorated by supplementation with NAD⁺ in vitro and NR in vivo.

Conclusions

These results indicate that NAMPT preserves corneal endothelial cell density and suppresses EnMT by sustaining energy metabolism, autophagic flux, and lysosomal acidification. Thus, NAMPT may emerge as a novel therapeutic target for delaying FECD progression.

Fuchs corneal endothelial dystrophy (FECD) is a multifactorial degenerative disorder of the cornea. The global incidence among individuals over 50 years old ranges from 4% to 9%, with a notably higher prevalence in females (female-to-male ratio: 2.5–3:1).^1,2 Characteristic pathologic features of FECD include abnormal thickening of Descemet's membrane (DM), forming posterior corneal endothelial excrescences (guttae), along with progressive loss and dysfunction of corneal endothelial cells (CECs), ultimately leading to corneal edema and visual impairment.^3,4 Although corneal transplantation remains the only currently available effective treatment, its application is constrained by the availability of donor tissue, potential surgical complications, and the long-term graft survival rate,⁴ highlighting the urgent need for the exploration of novel therapeutic targets.

Abnormal extracellular matrix (ECM) deposition constitutes the core pathogenic mechanism of FECD, characterized by DM thickening and driven by endothelial-to-mesenchymal transition (EnMT), as evidenced by the upregulation of mesenchymal markers such as α-SMA, vimentin, and fibronectin.^5–7 As a tissue with high oxygen consumption, the corneal endothelium contains abundant mitochondria, and its biological functions are critically dependent on the active ion transport mediated by Na⁺-K⁺-ATPase at the cell membrane surface. Approximately 85% of the energy required for this process is generated through mitochondrial oxidative phosphorylation.^8,9 This study revealed that CECs in patients with FECD exhibit a hypometabolic state, with downregulation of key energy metabolism pathways including glycolysis, the tricarboxylic acid (TCA) cycle, and oxidative phosphorylation. This metabolic suppression may drive aberrant ECM deposition and the progression of EnMT.

Nicotinamide phosphoribosyltransferase (NAMPT), the rate-limiting enzyme in NAD⁺ biosynthesis, plays a critical role in maintaining intracellular NAD⁺ levels and is involved in fundamental biological processes, including cellular energy metabolism and redox homeostasis.^10,11 NAD⁺ levels decline progressively with aging and are closely associated with the aging process.¹² In this study, reduced expression of NAMPT was observed in corneal endothelial specimens from patients with FECD and in corresponding mouse models. Downregulation of NAMPT may disrupt NAD⁺ homeostasis and impair cellular energy metabolism. However, the functional role of this alteration in FECD pathogenesis, particularly its potential involvement in ECM accumulation and the EnMT process, remains to be elucidated.

This study aims to explore the core mechanism by which the NAMPT/NAD⁺ axis regulates energy metabolism in CECs and affects the function of vacuolar-type H⁺-ATPase (V-ATPase) and lysosomal stability, thereby leading to abnormal deposition of ECM and EnMT in the pathogenesis of FECD. The findings will offer new pathogenetic insights for FECD and identify potential therapeutic targets for interventions focusing on the NAMPT/NAD⁺/V-ATPase pathway.

Methods

Data Acquisition and Preprocessing

We queried the GEO database for microarray and RNA sequencing (RNA-seq) datasets related to FECD and identified three datasets: GSE74123, GSE142538, and GSE171830. Among them, GSE74123 and GSE171830 are microarray datasets, including the normal control group and the FECD group; GSE142538 is an RNA-seq dataset, including the normal control group, pre-S FECD group, and late-stage FECD group. We performed probe-to-gene symbol annotation, data standardization, and gene differential analysis on the three datasets and obtained the differentially expressed gene (DEG) profile data.

Gene Set Enrichment Analysis

Gene set enrichment analysis (GSEA) was conducted to evaluate the coordinated expression of predefined biological pathways. Genes were ranked based on the metric of differential expression derived from the initial analysis. Enrichment analysis was performed using the “clusterProfiler” package in R, and the statistical significance of the Normalized Enrichment Score (NES) was assessed through 1000 phenotype-based permutations. A false discovery rate (FDR) q-value of less than 0.05 was considered statistically significant.

Gene Set Variation Analysis and Correlation

To assess pathway activity at an individual sample level, gene set variation analysis (GSVA) was employed using the “GSVA” package, which transforms the gene expression matrix into a matrix of sample-wise enrichment scores for the epithelial-mesenchymal transition (EMT) and ATP synthesis gene sets. Differences in these GSVA enrichment scores across the experimental groups were tested for statistical significance using the “limma” package. Furthermore, to investigate the relationship between EMT and ATP synthesis processes, the Spearman correlation coefficient between their respective GSVA enrichment scores was calculated across all samples, with results visualized using scatter plots.

Human Tissue Samples

This study was approved by the Research Ethics Committee of Chinese PLA General Hospital, and participants or their relatives signed an informed consent form. The human tissue experiments complied with the guidelines of the ARVO Best Practices for Using Human Eye Tissue in Research. The research was conducted in strict compliance with the approved protocols. Three cases of normal corneal endothelial tissues were enrolled as the control group, obtained from residual specimens of corneal grafts provided by the corneal bank. Three cases of pathologic corneal endothelial tissues derived from patients with advanced FECD were included in the experimental group, harvested intraoperatively during endothelial keratoplasty. Following surgical excision, corneal endothelial tissues from patients with FECD were promptly placed in corneal storage medium (Optisol-GS; Bausch & Lomb, Rochester, NY, USA) and briefly stored at 4°C. Normal donors were age-decade matched with patients diagnosed with FECD. Detailed patient information is presented in Supplementary Table S1.

Murine Model of Late-Onset FECD

Female C57BL/6J (8 weeks old) mice were purchased from SPF Biotechnology Co., Ltd. (Beijing, China). The mice were housed under specific pathogen–free conditions at a controlled temperature with a 12-hour light/dark cycle and had free access to food and water. The mice were anesthetized by intraperitoneal injection of 1.25% avertin (0.2 mL/10 g).

A late-onset FECD model was established using a protocol adapted from a previous study.¹³ An ultraviolet A (UVA) LED light source (M365LP1; Thorlabs, Newton, NJ, USA) with an emission peak at 365 nm and a full width at half-maximum bandwidth of 9 nm was employed. The light source was focused onto the mouse cornea to generate a 4-mm -diameter irradiation zone, delivering a UVA fluence of 500 J/cm² over an exposure time of 20 minutes, 57 seconds. The right eye was irradiated as the experimental group, while the left eye was shielded with a sterile dressing as control.

All animal procedures were approved by the Institutional Animal Care and Use Committee of the General Hospital of Chinese PLA and conducted in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research.

Anterior Chamber Injection of Adeno-Associated Virus

Adeno-associated virus (AAV) vectors (serotype 9, U6 promoter) encoding short hairpin RNA targeting the NAMPT sequence (GCTTTGCTACAGAAGTTAACC) or a nontargeting negative control (CGCTGAGTACTTCGAAATGTC) were obtained from GeneChem Biotechnology (Shanghai, China).

Mice were placed under systemic and topical ocular anesthesia, and pupils were dilated using tropicamide. Under microscopic guidance, a 34-gauge beveled needle was introduced into the anterior chamber at a 45° angle approximately 1 mm from the corneal limbus to a depth of 0.5 mm, then gently removed. Subsequently, 2 µL of either AAV-shNAMPT or AAV-shNC (1 × 10¹³ vector genomes/mL) was slowly injected over 1 minute into the anterior chamber using a 10-µL microsyringe (WPI, Sarasota, FL, USA) fitted with a 36-gauge needle. The needle remained in situ for an additional minute before withdrawal to minimize reflux. Postoperative antibiotic ointment was applied to minimize the risk of infection. Transduction efficiency assessed by EGFP fluorescence and NAMPT knockdown efficiency were confirmed at 4 weeks postinjection via immunofluorescence staining of the corneal endothelium.

Pharmacologic NAD⁺ Supplementation

Nicotinamide riboside (NR; Niagen, ChromaDex, Longmont, CO, USA) was administered to mice in drinking water at a concentration of 3.0 g/L. The NR solution was provided ad libitum and refreshed every 3 days to ensure stability and prevent microbial contamination. Based on recorded water consumption, the average daily NR intake was estimated to be approximately 400 mg per kg of body weight, consistent with previously established protocol.¹⁴ No significant difference in water consumption was observed between the NR-treated and control groups (water without NR) throughout the experimental period.

Animal Examination and In Vivo Imaging

Following anesthesia, the mouse corneas were examined using a slit-lamp biomicroscope equipped with a camera (Nikon D100, Tokyo, Japan). Corneal opacity was assessed and scored based on previously established criteria as follows¹⁵: 0 = completely transparent; 1 = mild haze, iris and pupil clearly visible; 2 = moderate opacity, iris and pupil still discernible; 3 = pronounced opacity, iris and pupil barely detectable; and 4 = completely opaque, iris and pupil not visible.

Anterior segment imaging was performed using optical coherence tomography (AS-OCT; Optoprobe ISOCT, Pontypridd, UK), and the central corneal thickness (CCT) was measured using the built-in software.

Finally, corneal endothelium was visualized by a laser scanning in vivo confocal microscopy (IVCM) (Heidelberg Retina Tomograph III with Rostock Corneal Module; Heidelberg Engineering GmbH, Heidelberg, Germany). The system captures two-dimensional coronal sections of 400 × 400 µm (160,000 µm²) at adjustable depths. Throughout IVCM imaging session, mice were securely positioned on a warmed platform to maintain body temperature.

Immunofluorescence Staining of Cornea Buttons and Frozen Sections

Immunofluorescence was performed on frozen sections of the central cornea and cornea buttons. All samples were fixed in 4% paraformaldehyde (PFA) (sections: 10 minutes, room temperature [RT]; tissues: 2 hours, 4°C), permeabilized with 0.3% Triton X-100 (sections: 5 minutes, RT; tissues: 30 minutes, 4°C). Subsequently, nonspecific binding was blocked by incubation with goat serum (ZSGB-BIO, Beijing, China) at RT for 30 minutes. After overnight incubation with primary antibodies at 4°C and subsequent exposure to corresponding Alexa Fluor–conjugated secondary antibodies at RT for 2 hours, the samples were mounted using antifade medium containing DAPI (Beyotime, Shanghai, China). Imaging was performed using a confocal microscope (FV3000-MCPSU; Olympus, Tokyo, Japan) for frozen sections and tissues. Cell numbers and fluorescence intensity of target proteins were quantified using ImageJ software (version 1.52a; National Institutes of Health, Bethesda, MD, USA), with several randomly selected fields evaluated per sample. Cell counting was blindly assessed by two observers. The primary antibodies were used in immunofluorescence are listed in Supplementary Table S2.

Immunofluorescence Staining of Cells

Cells were initially fixed with 4% PFA at RT for 20 minutes, followed by permeabilization with 0.3% Triton X-100 for 10 minutes and blocked with goat serum for 30 minutes. The cells were then incubated overnight at 4°C with primary antibodies diluted in blocking solution. On the following day, the cells were exposed to secondary antibodies and incubated at RT for 2 hours. Nuclear counterstaining was performed using Hoechst 33342 staining solution (Beyotime). Cellular fluorescence was captured and analyzed using a high-content imaging system (Operetta CLS; PerkinElmer, Llantrisant, Wales, UK). The primary antibodies are listed in Supplementary Table S2.

Western Blot Analysis

Protein lysates were extracted from cultured cells using RIPA buffer containing protease and phosphatase inhibitors. Equal amounts of protein were separated by SDS-PAGE (Bio-Rad, Hercules, CA, USA) and transferred to nitrocellulose membranes. After blocking with 5% skimmed milk, the membranes were incubated overnight at 4°C with primary antibodies, followed by horseradish peroxidase (HRP)–conjugated secondary anti-mouse or anti-rabbit antibodies for 1 hour at RT. Protein bands were detected using an enhanced chemiluminescence system (Bio-Rad). Quantification was performed using ImageJ software. The primary antibodies used in Western blot are listed in Supplementary Table S2.

Quantitative RT-PCR Analysis

Following extraction from cultured cells with TRIzol reagent (Invitrogen, Grand Island, NY, USA), total RNA was assessed for purity and concentration. Subsequently, 1 µg RNA served as the template for cDNA synthesis with the iScript cDNA Synthesis Kit (Bio-Rad). Quantitative RT-PCR assays were conducted on a CFX96 system (Bio-Rad) employing iTaq Universal SYBR Green Supermix (Bio-Rad). The mRNA relative expression of target genes was quantified relative to controls via the 2^−ΔΔCT method, with primer sequences detailed in Supplementary Table S3.

Transmission Electron Microscopy

Corneal specimens were trimmed into 1-mm³ tissue and promptly immersed in Electron Microscopy Fixative (Servicebio, Wuhan, China) and incubated at RT for 2 h. Subsequently, the specimens were transferred to 4°C, postfixed with 1% osmium tetroxide, and embedded in Epon 812.¹⁶ After completion of the standard protocol, corneal endothelium was photographed with transmission electron microscopy (TEM) (HITACHI, Tokyo, Japan).

Cell Culture

The human corneal endothelial cell (HCEnC) line B4G12, obtained from the Shandong Eye Institute (Shandong, China), was cultured following established methods.¹⁷ Cells were maintained in Dulbecco's modified Eagle's medium (Gibco, Carlsbad, CA, USA) supplemented with 10% fetal bovine serum (Corning, Corning, NY, USA) and 50 U/mL penicillin–streptomycin (Gibco), and grown at 37°C under a humidified atmosphere of 1% O₂, 5% CO₂, and 94% N₂.

Cell Transfection and Treatment

Cells were transfected with NAMPT-siRNAs (GenePharma, Suzhou, China) using Lipofectamine RNAiMAX (Invitrogen) following the manufacturer's instructions. After 7 hours of transfection, the culture medium was replaced with fresh medium, and the cells were subsequently cultured for a total of 48 hours posttransfection. At the appropriate time points, the cells were subjected to the designated treatments.

For cellular treatment, paraquat (PQ; Merck KGaA, Darmstadt, Germany) was administered at concentrations of 100, 250, 500, 750, 1000, and 1500 µM, with PBS serving as the vehicle control. These working solutions were prepared by serial dilution of a 250-mM stock solution in PBS, which was stored at −20°C. NAD⁺ (Selleck, Houston, TX, USA) was reconstituted in PBS to a 50-mM stock solution and stored at −20°C. For cell treatment, a final concentration of 10 µM was used.

Cell Viability Assay

Briefly, after treatment, HCEnCs were incubated with CCK-8 solution (Beyotime) at 37°C under 5% CO₂ for 1 hour. The absorbance at 450 nm was measured using a Spark multimode microplate reader (Tecan, Grödig, Austria), and cell viability was calculated as a percentage relative to the untreated controls.

RNA Sequencing

Total RNA was extracted from cells with the RNeasy Micro Kit (Qiagen, Hilden, Germany) and subsequently purified using the RNAClean XP Kit (Beckman Coulter, Brea, CA, USA) and the RNase-Free DNase Set (Qiagen). RNA integrity and quality were assessed by measuring the RNA Integrity Number on an Agilent 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA, US) and quantifying concentration with a NanoDrop ND-2000/Qubit 2.0. Sequencing libraries were constructed from the qualified RNA according to standard protocols, which included mRNA isolation, fragmentation, cDNA synthesis (first and second strand), end repair, A-tailing, and adapter ligation. The final library was amplified by PCR, and its quality was verified by assessing concentration (Qubit 2.0 Fluorometer) and size distribution (Agilent 4200). Sequencing was performed on a NovaSeq6000 platform (Illumina, San Diego, CA, USA) in PE150 mode. We have deposited the original RNA-seq data in the GEO database with the accession number (GSE319606). DEGs were identified using the DESeq2 R package with thresholds of |log2 fold-change| ≥ 1 and FDR < 0.05. The resulting DEGs were subsequently subjected to Gene Ontology (GO) enrichment analysis.

ATP and NAD⁺/NADH Assay

The concentration of ATP was determined using a commercial kit (Beyotime) according to the instructions provided with the kit. The intracellular NAD⁺ and NADH levels were quantified using the NAD⁺/NADH Assay Kit (Beyotime), in strict accordance with the manufacturer's protocol.

Mitotraker and LysoSensor Fluorescence Staining

Cells were seeded in a PhenoPlate 96-well TC-treated microplate (Revvity, Waltham, MA, USA) and treated with media containing the respective compounds. MitoTracker Green (Beyotime) was added to cultured cells at a concentration of 600 nM and incubated at 37°C for 30 minutes. After washing with PBS, 1 µM LysoSensor Yellow/Blue DND-160 (Yeasen Biotech, Shanghai, China) was applied to stain the cells at 37°C for 30 minutes. Finally, the cell nuclei were stained with Hoechst 33342 (Beyotime). Imaging of cellular fluorescence was conducted on a high-content imaging system.

Oxygen Consumption Rate

The mitochondrial respiratory function of cells was evaluated by measuring the oxygen consumption rate (OCR) using a Seahorse XFe96 Extracellular Flux Analyzer (Agilent). Before measurement, cells were seeded into XFe96-well microplates (Agilent Technologies) at a density of 5 × 10³ cells/well and treated according to experimental design. Prior to the assay, the growth medium was replaced with Seahorse assay buffer supplemented with 10 mM glucose, 1 mM pyruvate, and 2 mM glutamine (Alicelligent, Beijing, China), followed by incubation for 60 minutes at 37°C in a CO₂-free incubator. OCR was monitored in real time under basal conditions and after the sequential addition of the following compounds: 1.5 µM oligomycin, 0.5 µM FCCP, and 1 µM rotenone/antimycin A (Rot/AA), all provided within the assay kit (Alicelligent). Critical parameters of mitochondrial function, including basal respiration, ATP production, maximal respiration, and spare respiratory capacity, were computed using Agilent Wave Software (version 2.6). All recorded OCR values were normalized to cell number to ensure comparative accuracy.

Measurement of NAD⁺ Concentration in Aqueous Humor

Following euthanasia, aqueous humor (AH) was collected from the mice using a glass micropipette under a stereomicroscope. A perforation was created 1 mm anterior to the corneal limbus. The micropipette tip was then carefully inserted through the cornea and into the anterior chamber, taking care to avoid damaging the iris and the lens. Approximately 5 µL AH was aspirated from each eye. The NAD⁺ levels in the AH were rapidly measured using an automated NAD⁺ analyzer (Nadical, Shenzhen, China).¹⁸

Statistical Analysis

Data are reported as mean ± standard error of the mean and analyzed with GraphPad Prism software (version 10.1.2; GraphPad Software, La Jolla, CA, USA). Differences between groups were assessed using the unpaired two-tailed Student's t-test, one-way analysis of variance (ANOVA), or two-way ANOVA, as dictated by the experimental design. A P value of less than 0.05 was considered statistically significant.

Results

NAMPT May Be Involved in Downregulated Energy Metabolism and Enhanced Extracellular Matrix Remodeling in FECD

We searched all available datasets related to FECD and identified three GEO datasets—GSE74123, GSE142538, and GSE171830—for integrated analysis. The combined analysis revealed that downregulated energy metabolism (e.g., ATP biosynthetic process, oxidative phosphorylation, proton motive force-driven ATP synthesis, proton motive force-driven mitochondrial ATP synthesis) and upregulated ECM remodeling (e.g., extracellular matrix assembly, extracellular matrix organization, positive regulation of epithelial to mesenchymal transition, response to transforming growth factor beta) are two prominent features in FECD samples (Figs. 1A, 1B). Analysis of the GSE142538 dataset revealed that pathways involved in oxidative phosphorylation and ATP generation/metabolism were downregulated as early as the preclinical stage and became more pronounced in late-stage FECD (Supplementary Fig. S1A). Consistently, the GSE171830 dataset exhibited marked downregulation of genes involved in glycolysis, oxidative phosphorylation, and the TCA cycle (Supplementary Fig. S1B). GSVA further indicated a significant negative correlation between these processes, suggesting a potential link between impaired ATP-synthesizing metabolism and EMT pathways in FECD (Fig. 1C).

Figure 1.

Impaired energy metabolism and decreased NAMPT expression coincide with enhanced ECM remodeling in FECD. (A) GSEA plot demonstrating significant negative enrichment of oxidative phosphorylation and ATP synthesis pathways in patients with FECD across the indicated datasets (GSE74123, GSE171830, and GSE142538). (B) GSEA plot demonstrating significant positive enrichment of ECM remodeling pathways in patients with FECD. (C) Scatter plot showing the negative correlation between GSVA enrichment scores for ATP synthesis and EMT pathways in the indicated datasets. (D) Expression levels of NAMPT in corneal endothelium between controls and patients with FECD from the GSE74123 and GSE171830 datasets. (E) Expression levels of NAMPT in corneal endothelium among controls, preclinical patients, and patients with FECD from the GSE142538 datasets. (F) Representative immunofluorescence staining images of NAMPT in human corneal endothelium and quantitative results of mean immunofluorescence intensity (n = 3). Scale bar: 20 µm. (G) GO enrichment analysis results of DEGs following NAMPT knockdown in an in vitro model. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

NAD⁺ levels are closely associated with energy metabolism and participate in various cellular biological processes. As the rate-limiting enzyme for intracellular NAD⁺ synthesis, NAMPT holds significant biological importance. In all three datasets, we observed consistent downregulation of NAMPT expression (Figs. 1D, 1E). Specifically, in GSE142538, which includes both preclinical and late-stage FECD samples, NAMPT expression was moderately decreased in the preclinical stage and significantly reduced in late-stage cases. Consistent with these findings, corneal endothelial whole-mount staining also showed reduced NAMPT expression in patients with FECD compared with normal controls (Fig. 1F).

Interestingly, when we knocked down NAMPT expression in B4G12 cells (a normal CEC line) using small interfering RNA (siRNA) and performed RNA-seq followed by GO analysis, the results similarly highlighted energy metabolism and EMT/ECM remodeling as key biological features (Fig. 1G). These findings suggest that NAMPT may contribute to ECM remodeling processes, potentially through its role in regulating cellular energy metabolism.

NAMPT Expression Was Downregulated During Abnormal ECM Accumulation in UVA-Induced Late-Onset FECD Mouse Models

To model FECD clinically, we established an animal model using UVA irradiation. After irradiation, a significant increase in corneal opacity was assessed by slit-lamp examination with corneal opacity scores rising from 0.00 ± 0.00 to 2.10 ± 0.74 by day 7 (P < 0.05), 1.90 ± 0.88 on day 28 (P < 0.05), and worsening markedly to 3.40 ± 0.70 by day 56 (P < 0.05) (Figs. 2A, 2D). Corneal edema leads to an increase in CCT, which is a hallmark of endothelial cell functional impairment. CCT revealed a significant increase between the sham-irradiated group and the UVA-irradiated group at day 56 (sham: 91.37 ± 6.51 µm; the UVA-irradiated group: 127.27 ± 28.27 µm; P < 0.05) (Figs. 2B, 2E). IVCM revealed that the CECs progressively lost their hexagonal morphology, exhibited a disordered arrangement, and developed “black areas” following UVA irradiation (Fig. 2C). Subsequently, we observed a dynamic change in NAMPT expression in the corneal endothelium after irradiation, which showed an initial transient upregulation on day 7, followed by a sustained decline that became significantly lower than that in the sham group by day 56 (P < 0.05) (Figs. 2F, 2G). Immunologic staining of downregulated ZO-1 and Na⁺-K⁺-ATPase provided further evidence for morphologic and functional impairment of CECs after UVA irradiation (Figs. 2F–K). Concurrently, the EnMT markers, including α-SMA and vimentin, were significantly upregulated after UVA exposure (Figs. 2I–N). These findings suggest that NAMPT expression is downregulated during abnormal ECM accumulation in mice corneal endothelium after UVA irradiation.

Figure 2.

NAMPT expression is downregulated during abnormal ECM accumulation in UVA-induced late-onset FECD mouse models. (A–C) Representative corneal slit-lamp photographs, AS-OCT, and IVCM images of the corneal endothelial morphology in sham controls and UVA-irradiated mice at 7, 28, and 56 days after irradiation. (D) Scoring results of corneal opacity of corneas (n = 10). (E) CCT measurements via AS-OCT (n = 10). (F) Representative immunofluorescence staining images of NAMPT and ZO-1 in mouse corneal endothelium in sham controls and UVA-irradiated mice at 7, 28, and 56 days after irradiation. Scale bar: 20 µm. (G, H) Quantitative analysis of mean immunofluorescence intensity for NAMPT and ZO-1 (n = 5). (I) α-SMA and Na⁺-K⁺-ATPase staining in mouse corneal endothelium in sham controls and UVA-irradiated mice at 7, 28, and 56 days after irradiation. Scale bar: 20 µm. (J) Quantification of CEC density via cell counting (n = 5). (K, L) Quantitative analysis of mean immunofluorescence intensity for Na⁺-K⁺-ATPase and α-SMA (n = 3). (M) Representative immunofluorescence staining images of vimentin in mouse corneal endothelium in sham controls and UVA-irradiated mice at 7, 28, and 56 days after irradiation. Scale bar: 20 µm. (N) Quantitative analysis of mean immunofluorescence intensity for vimentin (n = 3). *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

NAMPT Expression and NAD⁺/NADH Levels Were Altered in an In Vitro Chronic Oxidative Injury Model

Currently, there is still a lack of in vitro cell models that can effectively mimic FECD. Given the important role of chronic oxidative damage in the pathogenesis of FECD, we used PQ treatment to induce cellular oxidative injury. Concentration-gradient experiments showed that 250 µM PQ did not cause significant effects on the viability of B4G12 cells within 24 hours and 48 hours; thus, this concentration was finally selected to establish a chronic oxidative injury model simulating FECD (Supplementary Fig. S2). In cells treated with 250 µM PQ, the expression of NAMPT protein increased, and the mRNA level rose initially but decreased after 48 hours—a temporal pattern paralleling observations in the UVA-irradiated animal model (Figs. 3A–C).

Figure 3.

NAMPT expression and NAD⁺/NADH levels are altered in an in vitro chronic oxidative injury model. (A) Representative blot images of NAMPT at different time points following PQ (which induces cellular oxidative injury). (B) Protein expression of NAMPT (n = 3). (C) The relative mRNA levels of NAMPT (n = 6). (D) Intracellular NAD⁺ levels at indicated time points after PQ treatment (n = 3). (E) Intracellular NADH levels at indicated time points after PQ treatment (n = 3). (F) NAD⁺/NADH ratio at indicated time points after PQ treatment (n = 3). *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

To investigate whether alterations in NAMPT affect cellular NAD⁺/NADH levels, we measured intracellular NAD⁺/NADH concentrations. When cells were treated with PQ for 12 and 24 hours, intracellular NAD⁺ and NADH levels both increased significantly and synchronously, forming a stable NAD⁺/NADH ratio as an initial compensatory response to oxidative stress. By 48 hours, however, both cofactor concentrations decreased while the ratio remained elevated due to an even greater loss of NADH (Figs. 3D–F). This indicates compensatory failure as prolonged stress disrupts synthesis pathways and reduces NAD⁺/NADH production below consumption.

Downregulated NAMPT Promoted CEC Loss and Abnormal ECM Accumulation

To elucidate the role of NAMPT in FECD pathogenesis, we performed intracameral injections of AAV-shNAMPT to achieve corneal endothelium-specific knockdown, whereas control mice received AAV-shNC (experimental design detailed in Fig. 4A). Following adenovirus injection, NAMPT expression in the corneal endothelium was efficiently knocked down to 52.5% of the control level (P < 0.05) (Fig. 4B). Intracameral injection of AAV-shNAMPT and AAV-shNC did not induce corneal endothelial damage, as evidenced by clear corneas and normal CEC morphology. After UVA exposure, AAV-shNAMPT–treated mice exhibited significantly more severe pathology than AAV-shNC controls, characterized by elevated corneal opacity scores (3.00 ± 0.93 vs. 1.63 ± 0.92; P < 0.05), increased CCT (197.15 ± 32.96 µm vs. 123.68 ± 30.46 µm; P < 0.05), and histopathologic changes visible in hematoxylin and eosin staining (Figs. 4C–E). IVCM further revealed profound cellular abnormalities in AAV-shNAMPT–treated corneas, including endothelial enlargement, irregular morphology, and abundant guttae formation, indicative of progressive endothelial decompensation (Fig. 4C). ZO-1 staining revealed a significant reduction in CEC density in AAV-shNAMPT–treated mice (1361.24 ± 108.13 cells/mm²) compared to AAV-shNC controls (1699.06 ± 117.96 cells/mm²; P < 0.05). Concurrently, α-SMA protein expression was markedly upregulated in the AAV-shNAMPT group (P < 0.05) (Figs. 4F–H). These collective findings strongly imply that NAMPT deficiency accelerates corneal endothelial decompensation and abnormal ECM accumulation.

Figure 4.

Knockdown of NAMPT exacerbates the CEC loss and EnMT both in vivo and in vitro. (A) Mice received an anterior chamber injection of AAV-shNC or AAV-shNAMPT, followed by UVA irradiation 4 weeks later. After another 4 weeks, observations and sample collection were performed. (B) Immunofluorescence staining was performed to assess the expression of NAMPT in the corneal endothelium treated with AAV-shNC or AAV-shNAMPT, and the mean fluorescence intensity was quantitatively analyzed (n = 4). Scale bar: 50 µm. (C) Representative corneal slit -lamp images, AS-OCT images, IVCM images, and hematoxylin and eosin staining images of the AAV-shNC, AAV-shNAMPT, UVA+AAV-shNC, and UVA+AAV-shNAMPT groups. (D) Scoring results of corneal opacity in mice (n = 8). (E) Quantitative results of CCT analyzed by AS-OCT (n = 8). (F) Representative images of α-SMA and ZO-1 immunofluorescence staining for corneal endothelium and sections in the UVA+AAV-shNC and UVA+AAV-shNAMPT groups. Scale bar: 20 µm; 50 µm. (G) Results of CEC counting based on ZO-1 staining (n = 5). (H) Quantitative analysis of the mean fluorescence intensity of α-SMA (n = 5). (I) Cells were transfected with sicontrol or siNAMPT the day after plating. The medium was replaced 7 hours posttransfection, and PQ treatment was applied at 24 hours. Cells were collected 48 hours later for subsequent assays. (J) Protein expression and mRNA levels of NAMPT in cells transfected with sicontrol and three different siNAMPTs. (K) Results of cell viability, as assessed by CCK-8 assay, in sicontrol and siNAMPT groups treated with different concentrations of PQ. (L) Representative blot images of NAMPT, α-SMA, and vimentin in the sicontrol, siNAMPT, PQ+sicontrol, and PQ+siNAMPT groups. (M) Protein expression and mRNA levels of α-SMA and vimentin (n = 3). (N) The relative mRNA levels of fibronectin, α-SMA, and vimentin in control, PQ, FK866 (the NAMPT inhibitor), and PQ+FK866 groups (n = 3). *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

To investigate the potential role of NAMPT in this regulatory network, we screened three distinct siNAMPT sequences and selected the most effective one (based on mRNA and protein levels) for subsequent experiments (Fig. 4J). In PQ-treated cells, NAMPT knockdown (via siNAMPT) significantly reduced cell viability and concurrently upregulated the expression of fibrotic markers (α-SMA and vimentin) at both mRNA and protein levels (Figs. 4I, 4K–M). Consistently, treatment with the NAMPT inhibitor FK866 also induced an EnMT phenotype in CECs, as evidenced by significantly elevated mRNA levels of fibronectin, α-SMA, and vimentin (Fig. 4N). These collective results indicate that the downregulation of NAMPT activity promotes fibrotic changes in CECs.

Downregulation of NAMPT Expression Disrupted Autophagic Flux

Under chronic oxidative stress with 250 µM PQ, a time-dependent increase in LC3B-Ⅱ occurred alongside a decrease in P62 (an autophagy receptor and substrate), suggesting enhanced autophagic flux (Fig. 5A). Furthermore, reduced colocalization of Lamp2 (a lysosomal membrane marker) and P62 indicated efficient autophagosome–lysosome fusion and degradation (Fig. 5B). These findings confirm that chronic oxidative stress activates functional autophagic flux in B4G12 cells. Subsequently, we modulated autophagic activity using the inhibitor 3-MA and the activator rapamycin. CCK-8 assays revealed that 3-MA exacerbated PQ-induced cytotoxicity, whereas rapamycin attenuated it, supporting autophagy's role in maintaining cell viability under oxidative stress (Fig. 5C). These results suggest that the activation of autophagy flux promotes cell survival and exerts a cytoprotective effect.

Figure 5.

Under the stimulation of oxidative damage, cells exhibit an elevated autophagy level to sustain cell viability. However, when NAMPT is inhibited, the autophagy flux is disrupted, and apoptosis increases. (A) Representative blot images and quantitative analysis of LC3B and P62 at different time points following PQ treatment. PQ treatment led to an increase in LC3B-II expression and a concurrent decrease in P62 levels (n = 3). (B) Immunofluorescence staining images showing reduced colocalization of P62 and Lamp2 in the PQ-treated group compared with the control group. (C) Cell viability detection results obtained after treating cells with the autophagy inhibitor 3-MA and the activator rapamycin under different concentrations of PQ treatment (n = 3). 3-MA worsened PQ-induced decline in cell viability, while rapamycin elevated it. (D) Representative blot images of NAMPT, LC3B-Ⅱ, P62, and caspase-3 in the sicontrol, siNAMPT, PQ+sicontrol, and PQ+siNAMPT groups. (E) Quantitative analysis of NAMPT, LC3B-Ⅱ, P62, and caspase-3 (n = 3). (F) Immunofluorescence staining images demonstrating the increase in the colocalization of P62 and Lamp2 in the siNAMPT and PQ+siNAMPT groups compared with the sicontrol group. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

We next examined whether NAMPT modulates autophagic flux. Compared with the levels in the cells transfected with sicontrol, the cells with siNAMPT showed an increase in the protein levels of LC3B-II and P62, indicating impaired autophagosome–lysosome fusion and/or lysosomal degradation (Figs. 5D, 5E). Confocal immunofluorescence analysis revealed a significant increase in the colocalization of Lamp2 with P62 in the PQ-treated cells with siNAMPT compared to the sicontrol (Fig. 5F), reflecting autophagic flux blockade. Concurrent with these organellar dysfunctions, the cells transfected with siNAMPT exhibited a significant increase in apoptosis relative to control, as quantified by caspase-3 (Figs. 5D, 5E). These data demonstrate that inhibition of NAMPT could impair autophagic flux in CECs. Similarly, we found that the NAMPT inhibitor FK866 impaired autophagic flux in B4G12 cells under chronic oxidative stress. It increased the levels of LC3B-II and P62 in PQ-treated cells in a dose-dependent manner, thereby enhancing the sensitivity of B4G12 cells to oxidative damage (Supplementary Fig. S3).

Inhibition of NAMPT Disrupted Mitochondrial Respiratory Capacity and ATP Synthesis, Which Reduced the Expression of V-ATPase and Thereby Suppressed Lysosomal Acidification

NAMPT is essential for maintaining intracellular NAD⁺ homeostasis. To functionally validate this paradigm, we performed genetic knockdown of NAMPT in B4G12 cells. Strikingly, NAMPT deficiency not only reduced NAD⁺ levels (P < 0.05) but also concomitantly decreased NADH levels (P < 0.05) (Figs. 6A, 6B). To determine the impact of NAMPT on mitochondrial function, we measured the OCR in response to electron transport chain inhibitors, defining key bioenergetic parameters (Fig. 6C). As shown in Figure 6D, NAMPT knockdown dramatically impaired mitochondrial respiration, as evidenced by decreased maximal respiration, spare respiratory capacity, and ATP production.

Figure 6.

Knockdown of NAMPT in vitro reduces NAD⁺ and NADH levels, suppresses mitochondrial respiration, decreases V-ATPase subunits, and augments the lysosomal pH. (A) Intracellular NAD⁺ levels in sicontrol and siNAMPT-treated cells (n = 3). (B) Intracellular NADH levels in sicontrol and siNAMPT-treated cells (n = 3). (C) Real-time OCR in sicontrol and siNAMPT cells (n = 6). (D) Key parameters of mitochondrial respiration in sicontrol and siNAMPT cells: basal respiration, maximal respiration, spare respiratory capacity, and ATP production (n = 6). (E) The heatmap showing the DEGs in cells of the siNAMPT group compared to the sicontrol group. Red borders mark the downregulated V-ATPase subunit genes in the siNAMPT group. (F) Downregulated V-ATPase subunits in patients with FECD compared with the control group in the GSE171830 dataset. (G) Representative images illustrating Mitotracker and LysoSensor staining (indicating lysosomal pH) in the different groups. Cells transfected with siNAMPT exhibited significantly reduced fluorescence compared with sicontrol, and this effect was markedly enhanced upon PQ exposure. Scale bar: 50 µm. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

To further investigate the pathologic mechanisms underlying NAMPT inhibition, we compared RNA-seq data from sicontrol and siNAMPT-transfected cells. Our analysis revealed that NAMPT inhibition downregulated multiple energy metabolism-related enzymes. Notably, the expression of several subunits of the V-ATPase (ATP6V0/V1), a key proton pump responsible for lysosomal acidification, was also significantly reduced, including ATP6V0A2, ATP6V0D2, ATP6V0E2, ATP6V1D, and ATP6V1E2 (Fig. 6E). This likely constitutes a key mechanism compromising lysosomal acidification and suppressing autophagic flux. In addition, interrogation of the public transcriptomic dataset (GEO: GSE171830) also revealed significant downregulation of multiple V-ATPase subunits in patients with FECD, including ATP6V1G2, ATP6AP1L, ATP6V1C1, ATP6V0E2, ATP6V1G1, ATP6V1E1, ATP6V1A, and ATP6V0D2 (Fig. 6F). Quantitative analysis of live-cell fluorescence imaging revealed a significant reduction in the signal intensity of LysoSensor staining (targeting acidic compartments, predominantly lysosomes) in the cells transfected with siNAMPT compared to the sicontrol. Upon PQ exposure, LysoSensor fluorescence in siNAMPT group cells decreased markedly, indicating that NAMPT deletion elevates lysosomal pH and compromises luminal acidity (Fig. 6G). Collectively, these results indicate that NAMPT suppression triggers a metabolic crisis by depleting NAD⁺/NADH, which disrupts energy metabolism and ATP production. This crisis specifically impairs the V-ATPase, resulting in defective lysosomal acidification and inhibition of autophagic flux—a cascade that may ultimately drive abnormal ECM deposition and EnMT, primarily through the loss of autophagic degradation.

NAD⁺ Supplement Partially Enhanced Mitochondrial Energetic Function, Restored Autophagy Flux, Promoted Lysosomal Degradation, and Increased Expression of V-ATPase

We assessed whether NAD⁺ supplementation could restore metabolic function. First, we confirmed that NAD⁺ supplement did not alter NAMPT levels in siNAMPT-transfected cells (Fig. 7A). Next, we found that PQ treatment markedly reduced ATP production, an effect that was further exacerbated by cotreatment with the NAMPT inhibitor FK866 and siNAMPT. Importantly, NAD⁺ supplementation effectively restored ATP levels in FK866-treated cells. Although it failed to rescue ATP levels following combined PQ and siNAMPT treatment, NAD⁺ supplementation successfully increased ATP in cells subjected to siNAMPT alone (Figs. 7B, 7C). To further clarify the regulatory role of the NAMPT/NAD⁺ axis in autophagic flux, we investigated the effects of NAMPT inhibition and subsequent NAD⁺ supplementation on autophagy-related markers. Inhibition of NAMPT was found to impair autophagic flux, leading to increased LC3-II and P62 expression, as well as enhanced colocalization of P62 and Lamp2, which indicates autophagic flux blockage. We found that after NAD⁺ supplementation, LC3-II expression remained elevated, while P62 expression and the colocalization of P62 and Lamp2 were reduced, suggesting that NAD⁺ could alleviate the autophagic flux blockage induced by NAMPT inhibition (Figs. 7D, 7E and Supplementary Fig. S4). LysoSensor fluorescence staining revealed that siNAMPT-transfected cells exhibited a notably low fluorescence, indicating an elevated lysosomal pH. Conversely, the addition of NAD⁺ to siNAMPT-transfected cells restored the high fluorescence intensity, signifying the recovery of an acidic pH (Figs. 7F, 7G). Investigation into potential mechanisms revealed that the reduction in V-ATPase expression induced by siNAMPT was reversed by NAD⁺ supplementation. Similarly, in the context of PQ treatment, V-ATPase expression, including ATP6V1D, ATP6V1E2, ATP6V0D2, ATP6V0E2, and ATP6V0A2, was lower in siNAMPT-transfected cells than in sicontrol cells; notably, NAD⁺ supplementation elevated V-ATPase expression in the PQ-combined siNAMPT-treated group (Fig. 7H). Collectively, these results demonstrate that NAD⁺ plays an important role in restoring mitochondrial energy function, regulating autophagic flux, enhancing V-ATPase expression, and promoting lysosomal degradation.

Figure 7.

NAD⁺ supplementation induces partial recovery of ATP production, promotes autophagic flux restoration, decreases lysosomal pH, and increases V-ATPase subunit expression. (A) The mRNA expression of NAMPT with or without PQ treatment in sicontrol, siNAMPT, and siNAMPT+NAD⁺ groups (n = 6). (B) Intracellular ATP levels with or without PQ treatment under control, FK866, and NAD⁺ treatment conditions (n = 6). (C) Intracellular ATP levels with or without PQ treatment in the sicontrol, siNAMPT, and siNAMPT+NAD⁺ groups (n = 6). (D) Representative blot images of LC3B and P62 in the sicontrol, siNAMPT, siNAMPT+NAD⁺ groups. (E) Quantitative analysis of LC3B-Ⅱ and P62 (n = 3). (F) Representative images illustrating LysoSensor staining in the different groups. Scale bar: 50 µm. (G) Quantitative analysis of the mean fluorescence intensity of LysoSensor staining (n = 3). (H) The mRNA expression of V-ATPase subunits, including ATP6V1D, ATP6V1E2, ATP6V0D2, ATP6V0E2, and ATP6V0A2, with or without PQ treatment in sicontrol, siNAMPT, and siNAMPT+NAD⁺ groups (n = 6). *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

NR Protected Endothelial Cells and Inhibited Abnormal ECM Accumulation in UVA-Induced Late-Onset FECD Mouse Models

To evaluate the therapeutic potential of NAD⁺ supplementation against UVA-induced FECD progression, mice were administered NR (an NAD⁺ precursor) in their drinking water for 2 months (Fig. 8A). We quantified NAD⁺ levels in mouse aqueous humor and found that UVA irradiation led to a significant decrease in aqueous humor NAD⁺ concentrations, whereas oral NR supplementation partially restored this reduction. These findings suggest that oral NR administration can elevate the ocular NAD⁺ pool (Fig. 8B). Notably, the NR-treated UVA group (UVA+NR group) exhibited significant clinical improvement over the untreated UVA group (UVA group). Specifically, the corneal opacity score was significantly decreased from 2.80 ± 0.92 in the UVA group to 1.60 ± 0.70 in the UVA+NR group (P < 0.05) (Figs. 8C, 8D). AS-OCT measurements indicated that corneal edema was alleviated by NR treatment, as evidenced by a reduction in CCT from 128.30 ± 29.01 µm to 116.09 ± 25.10 µm (P < 0.05); nonetheless, the thickness remained significantly greater than that in the sham group (94.47 ± 3.34 µm, P < 0.05) (Figs. 8C, 8E). IVCM demonstrated a notable improvement in CEC morphology in the NR-treated mice, characterized by a restored hexagonal cellular architecture, increased cell density, and attenuated guttae formation (Fig. 8C). Further analysis demonstrated that NR treatment increased CEC counts, downregulated α-SMA expression, and upregulated Lamp2 expression (Figs. 8F–H). TEM analysis indicated that UVA exposure induced cellular edema and mitochondrial cristae disruption. Conversely, NR treatment alleviated cellular edema, preserved mitochondrial integrity with well-defined cristae, and promoted the formation of autolysosomes. Furthermore, TEM imaging demonstrated that UVA irradiation resulted in a marked increase in DM thickness, which was effectively attenuated by NR treatment (Figs. 8I, 8J). These findings collectively suggest that NR mitigates UVA-induced corneal endothelial damage by autophagic clearance and promoting cellular repair.

Figure 8.

NR treatment alleviates the EnMT level and mitigates UNA-induced DM thickening. (A) After local UVA irradiation of the cornea, mice received NR drinking water treatment for 2 months and were observed, and samples were then collected. (B) NAD⁺ concentration measurements in aqueous humor from the sham, UVA, and UVA+NR experimental groups. (n = 8). (C) Representative corneal slit-lamp photographs, AS-OCT, and IVCM images of the corneal endothelial morphology for each group of sham, UVA, and UVA+NR. (D) Scoring results of corneal opacity of corneas (n = 10). (E) Mouse corneal thickness measurement results (n = 10). (F) Representative immunofluorescence staining images of ZO-1 and α-SMA in corneal endothelium. Scale bar: 20 µm. (G) Representative immunofluorescence staining image of Lamp2 in corneal endothelium. Scale bar: 20 µm. (H) Quantitative analysis results of the average fluorescence intensity of α-SMA and Lamp2 (n = 5). (I) Representative TEM images of CECs and DM in mice for each group of sham, UVA, and UVA+NR. The red triangle indicates autophagic lysosomes. The red line segment indicates DM. Scale bar: 1 µm. (J) Quantitative results of DM thickness (n = 3). *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Discussion

CECs, located at the innermost layer of the cornea, are essential for maintaining corneal transparency by regulating fluid and ion homeostasis as well as metabolism. These cells exhibit high metabolic activity and possess substantial energy demands. In FECD, however, CECs undergo degenerative changes characterized by abnormal ECM deposition that forms guttae, accompanied by a transition of endothelial cells into a mesenchymal phenotype, known as EnMT. This pathologic transformation leads to the loss of both pump and barrier functions of the corneal endothelium, resulting in persistent corneal edema and opacification.^19,20 Currently, there are no effective pharmacologic treatments for FECD, with corneal transplantation remaining the only option for advanced cases. Therefore, the development of therapeutic interventions is of great clinical importance. This study reveals for the first time that the hypometabolic state in FECD is closely associated with EnMT. Specifically, NAMPT, the rate-limiting enzyme in NAD⁺ biosynthesis, was significantly downregulated in CECs of FECD. Through in vitro and in vivo experiments, we demonstrated that NAMPT downregulation reduced NAD⁺ and ATP production, suppressed the expression of multiple V-ATPase subunits, impaired lysosomal acidification, and consequently inhibited autophagy flux, thereby promoting EnMT and aberrant ECM accumulation. Furthermore, exogenous supplementation with NAD⁺ or NR effectively reversed these pathologic alterations, suggesting that the NAMPT/NAD⁺/V-ATPase signaling pathway may serve as a promising therapeutic target for FECD (Fig. 9).

Figure 9.

This schematic diagram elucidates the mechanism by which NAMPT sustains CEC function and inhibits ECM deposition. In FECD, reduced expression of NAMPT in CECs leads to diminished NAD⁺ synthesis, thereby impairing mitochondrial energy metabolism and reducing ATP production. The resulting ATP deficiency compromises the function of V-ATPase, preventing effective proton transport against the concentration gradient. This dysfunction increases lysosomal pH, reduces degradative capacity, and disrupts autophagic flux, ultimately promoting ECM deposition, thickening of DM, and the formation of guttae.

Emerging evidence points to a central role of metabolic failure in FECD. A transcriptomic meta-analysis of FECD demonstrated that impaired energy metabolism, characterized by mitochondrial dysfunction and downregulated oxidative phosphorylation, constituted a core mechanism of the disease, irrespective of the presence of a CTG trinucleotide repeat expansion.²¹ Another study revealed that in the SLC4A11^−/− model of congenital hereditary endothelial dystrophy (CHED), aberrant glutamine metabolism disrupted the TCA cycle and compromised the energy supply in CECs, ultimately impairing endothelial pump function.²² Under physiological conditions, approximately 85% of the energy required by CECs is generated through oxidative phosphorylation—the primary energy production pathway for most cells in aerobic environments. Additionally, under anaerobic or hypoxic conditions, cells rapidly generate energy via anaerobic glycolysis. In contrast, tumor cells exhibit a distinct metabolic preference, relying predominantly on glycolysis even in the presence of oxygen—a phenomenon known as “metabolic reprogramming.”²³ However, degenerative disorders such as FECD display a fundamentally different metabolic profile. Our GSEA analysis of the GSE142538 dataset revealed consistent downregulation of oxidative phosphorylation, the TCA cycle, and ATP metabolism, with this suppression initiating in the preclinical stage and progressively worsening in advanced disease. Furthermore, integrative analysis of three FECD-related transcriptomic datasets (GSE74123, GSE142538, and GSE171830) showed significant downregulation of multiple pathways associated with energy metabolism and ATP synthesis. These findings collectively indicate a generalized “low metabolic state” in CECs of patients with FECD. Additionally, FECD transcriptomes were marked by a significant upregulation of pathways involved in ECM remodeling and EMT. GSVA analysis further identified a significant negative correlation between ATP synthesis and EMT processes. Based on these observations, we hypothesize that the low metabolic state may drive pathologic ECM deposition and EnMT. FECD is a progressive degenerative disorder, and CECs are derived from neural crest cells. Notably, a similar metabolic phenotype has been observed in other neurodegenerative diseases, such as Alzheimer's disease (AD). In patients with AD, global cerebral glucose metabolism is markedly reduced, leading to diminished energy production.²⁴ This pattern of impaired brain glucose and energy metabolism not only affects neuronal function but also promotes Tau protein aggregation through dysregulation of ECM remodeling. Indeed, studies have established a clear association between the degree of cerebral glucose metabolism and the extent of Tau pathology.^25–27

As the rate-limiting enzyme in NAD⁺ biosynthesis, NAMPT plays a critical role in maintaining intracellular NAD⁺ levels and is involved in fundamental biological processes, including cellular energy metabolism and redox homeostasis.²⁸ In mammals, NAD⁺ is synthesized predominantly through the salvage pathway from nicotinamide and nicotinic acid, with only a minor contribution from de novo synthesis starting from tryptophan.¹² Downregulation of NAMPT has been implicated in various aging-related and degenerative disorders, including type 2 diabetes, neurodegenerative diseases, and age-related macular degeneration (AMD).²⁹ Conditional knockout of NAMPT in projection neurons of adult mice resulted in progressive weight loss, motor neuron degeneration, functional deficits, paralysis, and premature death.³⁰ Similarly, photoreceptor-specific deletion of NAMPT resulted in photoreceptor cell death, retinal degeneration, and vision loss, while exogenous administration of nicotinamide mononucleotide conferred retinal protection.³¹ NAMPT/NAD⁺ signaling also contributes significantly to the maintenance of CEC viability and function. Studies revealed that UVB exposure induced corneal edema and apoptosis of CECs in mice, concomitant with reduced NAMPT and NAD⁺ levels; supplementation with nicotinamide (NIC), an NAD⁺ precursor, attenuated apoptosis by activating the Akt signaling pathway.³² Furthermore, NIC inhibited corneal EnMT and preserved endothelial integrity.³³ In this study, we observed decreased NAMPT expression in the corneal endothelium of patients with FECD and in a UVA-induced late-onset FECD mouse model during its decompensation phase. Intracameral injection of AAV-shNAMPT to specifically knock down NAMPT in mouse CECs exacerbated UVA-induced corneal edema, induced morphologic alterations in endothelial cells, and increased α-SMA expression. Similarly, in an in vitro model of chronic oxidative stress using siRNA-mediated NAMPT knockdown or pharmacologic inhibition with FK866, we observed decreased cell viability and elevated expression of EnMT markers such as α-SMA, vimentin, and fibronectin. These findings collectively indicate that under conditions of UVA-induced injury or chronic oxidative stress, suppression of NAMPT promotes EnMT in CECs, implicating NAMPT deficiency in the pathogenesis of EnMT in the corneal endothelium of FECD.

Oxidative stress is a core pathologic mechanism of FECD and an important inducer of EnMT.^6,34 In our study, UVA irradiation and PQ, as oxidative stress sources, significantly induced EnMT in CECs, characterized by loss of hexagonal shape, reduced intercellular junctions, and upregulated α-SMA and vimentin expression. Given the crucial role of oxidative stress in FECD and EnMT, we explored NAMPT expression during FECD progression. In the UVA-induced late-onset FECD mouse model, NAMPT was transiently upregulated in the early stage, likely enhancing NAD⁺ production to maintain energy metabolism and resist EnMT. In contrast, NAMPT expression decreased in the late decompensated phase, impairing NAD⁺ synthesis and causing energy failure, thereby accelerating EnMT progression. This dynamic pattern is consistent with previous observations in macrophages and other cell types.^35–37 Ferroptosis is closely associated with oxidative stress, which is also involved in FECD pathogenesis.³⁸ Chi et al.³⁹ identified miR-23a-3p as a novel antiferroptosis regulator in FECD by targeting PTEN, providing an important protective mechanism against oxidative stress and cell death. Our study, focusing on the NAMPT-mediated energy supply aspect of cell protection under oxidative stress, offers an additional line of defense against oxidative damage.

It is well established that intracellular NAMPT is closely linked to energy metabolism. In our in vitro model with NAMPT knockdown, we observed a decline in NAD⁺/NADH levels, oxygen consumption, and ATP synthesis, which partially recapitulates the “low-metabolic” state associated with FECD. Under this sustained low-energy condition, our study revealed impaired autophagic flux, evidenced by increased LC3B-Ⅱ levels, significant accumulation of P62, and markedly suppressed lysosomal acidification. Although autophagy serves as a protective mechanism during acute energy deficit or stress, this compensatory response becomes chronically compromised as the cellular energy supply dwindles, ultimately leading to organelle dysfunction.^40,41 Our data support this concept, showing that while cells responded to PQ treatment within the first 24 hours with an initial upregulation of NAMPT and enhanced autophagy to maintain viability and phenotype, this was followed by a decline in NAMPT expression by 48 hours. Correspondingly, in chronic oxidative stress models employing either NAMPT-targeting siRNA or the NAMPT inhibitor FK866, we consistently observed autophagy dysfunction and disrupted autophagic flux. This malfunction in the autophagosomal or lysosomal network has also been documented in AMD^42,43 and various neurodegenerative diseases,^44–47 suggesting that such impairment may represent a common pathologic feature of degenerative disorders.

Lysosomes serve as the terminal compartment of the autophagy pathway, responsible for the degradation of proteins and other macromolecules. Lysosomal dysfunction can impair protein degradation capacity and promote the process of EMT. The highly acidic lysosomal environment is maintained by V-ATPase, a multi-subunit proton pump composed of two functional domains: the membrane-embedded V0 domain, which forms the transmembrane channel for H⁺ transport, and the peripheral V1 domain, which hydrolyzes ATP to provide energy for active proton translocation against the concentration gradient.⁴⁸ The activity and function of V-ATPase are highly dependent on cellular energy status. In neurodegenerative diseases, reduced cerebral metabolic activity has been shown to suppress V-ATPase function, leading to impaired lysosomal degradation and accumulation of misfolded proteins both intracellularly and extracellularly.^49,50 In the CHED mouse model, Shyam et al.⁵¹ demonstrated that loss of SLC4A11 increased mitochondrial reactive oxygen species production, which subsequently inhibited nuclear translocation of the transcription factor TFEB, reduced V-ATPase expression, and disrupted lysosomal acidification and autophagic flux, ultimately contributing to corneal endothelial dysfunction. Similarly, in AMD, decreased V-ATPase expression has been shown to cause defective autophagic flux, impaired phagocytosis, accumulation of lipid droplets, and formation of vacuole-like structures.⁵² In renal podocytes, lysosomal inhibition using a V-ATPase inhibitor or its siRNA was found to markedly decrease the epithelial markers (P-cadherin and ZO-1) while increasing mesenchymal markers (FSP-1 and α-SMA). Conversely, inhibition of autophagosome formation with spautin-1 attenuated both the enhancement of EMT and the impairment of autophagic flux.⁵³ In the present study, we found that NAMPT knockdown significantly downregulated V-ATPase expression, compromised endothelial function, and elevated expression of α-SMA and vimentin, indicative of a fibrotic phenotype. Importantly, elevating NAD⁺ levels reversed these effects. These findings suggest that suppression of V-ATPase function and activity underlies the transition of CECs toward a mesenchymal state in low-metabolic conditions, highlighting lysosomal dysfunction as a key mechanism in EnMT pathogenesis. Furthermore, recent studies have found enhanced release of extracellular mitochondria in FECD endothelial cells. When lysosomal function is impaired, the secretion of extracellular vesicles can act as an alternative quality control mechanism to cope with excessive oxidative stress and handle cellular cargo.⁵⁴ Collectively, these observations link NAMPT/NAD⁺ mediated lysosomal homeostasis to both EnMT progression and the alternative extracellular vesicle–dependent quality control pathway, providing new insights into the integrated metabolic and cellular mechanisms underlying FECD pathogenesis.

Our study has certain limitations that should be acknowledged. First, the relatively small size of human samples included in the present study represents a limitation. Future investigations should incorporate a larger collection of human specimens to elucidate the correlation between NAMPT expression and FECD disease stages, as well as to quantify the protein levels and metabolic patterns using proteomics and metabonomics. Second, while we demonstrated that NAMPT and NAD⁺ impact autophagic flux and lysosomal acidification, the specific molecular mechanisms mediating these effects remain undefined. Lastly, FECD is a complex disease involving genetic and environmental factors. Given that our PQ -induced cell model mainly focuses on oxidative stress, it fails to recapitulate the complex pathogenesis of FECD. Thus, future studies should employ genetically modified models or UVA -irradiated cell models for further validation.

In conclusion, this study is the first to establish a causal link between NAMPT, V-ATPase, and corneal endothelial EnMT. We demonstrate that NAMPT, by modulating cellular energy metabolism, influences V-ATPase expression and lysosomal acidification, thereby regulating ECM deposition and the EnMT process. These findings underscore the therapeutic potential of the NAMPT/NAD⁺/V-ATPase axis in FECD.