Topical NAMPT activation restores mitophagic flux to rescue diabetic keratopathy via the NAD⁺/PINK1 pathway

Abstract

Background

Diabetic keratopathy (DK) is a sight-threatening complication affecting up to 70% of diabetic patients, for which there are currently no effective therapies to halt disease progression. This study seeks to identify a novel DK biomarker and characterize its mechanistic role in disease pathogenesis.

Methods

Proteomic analysis was performed on conjunctival epithelial cells from diabetic patients with and without DK using iTRAQ labeling coupled with liquid chromatography-tandem mass spectrometry (LC-MS/MS). Differential protein expression was validated in diabetic mouse models, human diabetic corneas, and high glucose-induced epithelial cell models. The role of nicotinamide phosphoribosyltransferase (NAMPT) in regulating mitophagy in corneal epithelial cells was further investigated through genetic manipulation. The protective effects of the NAMPT activator P7C3 on corneal nerves and epithelium were evaluated in db/db mice.

Results

NAMPT was identified as a differentially expressed protein. Chronic hyperglycemia selectively depleted mitochondrial NAMPT in corneal epithelial cells, disrupting nicotinamide adenine dinucleotide (NAD) biosynthesis and triggering a self-amplifying cycle of oxidative stress and impaired PINK1-mediated mitophagy, ultimately leading to apoptosis. In diabetic db/db mice, topical application of the NAMPT activator P7C3 enhanced NAMPT activity, restored mitochondrial quality control, attenuated epithelial degeneration, reduced stromal edema, restored tear secretion, and promoted regeneration of corneal nerves and epithelium.

Conclusions

Mitochondrial localization of NAMPT expression loss is a central driver of DK pathogenesis. Topical application of the NAMPT activator P7C3 holds promise for restoring both corneal homeostasis and tear secretion function, making it a novel therapeutic target for oxidative stress-related corneal diseases and Diabetic Dry Eye Syndrome.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12964-026-02784-5.

Introduction

Diabetes mellitus has emerged as a global health crisis, currently affecting more than 537 million adults worldwide, and it is estimated that the total number of cases will increase to 783 million by 2045 [1]. Diabetic keratopathy (DK) remains a clinically underrecognized complication, affecting up to 70% of diabetic patients [2, 3]. whereas systemic complications such as cardiovascular disease have been extensively studied [4, 5], DK involves pathological changes such as tear film instability, corneal nerve degeneration, persistent epithelial defects, and impaired wound healing, potentially leading to ulceration, scarring, and vision loss [6, 7]. Unfortunately, diagnosis is frequently delayed because of the nonspecific nature of its early clinical manifestations [8].

The vulnerability of the cornea to diabetes-related damage is mechanistically linked to its unique metabolic microenvironment. Owing to avascular glucose uptake from the aqueous humor to the cornea via glucose transporter type 1 [9, 10], the cornea is exceptionally sensitive to fluctuations in glucose levels [11]. Chronic hyperglycemia drives glucotoxicity, which synergizes with tear film instability and neurotrophic deficits [12], causing self-reinforcing epithelial dysfunction [13]. Current management strategies for DK include palliative lubrication and prophylactic measures, with transplantation reserved for end-stage cases [8, 14]. Sirtuin 1 targeted antioxidants (e.g., α-lipoic acid) have demonstrated partial efficacy in treating DK [15, 16], but their effects are limited to symptom mitigation [17]. These observations highlight the need to elucidate the metabolic disturbances and compromised neuroprotective mechanisms involved in DK to develop more effective interventions for this multifactorial condition.

Nicotinamide phosphoribosyltransferase (NAMPT), the rate-limiting enzyme in nicotinamide adenine dinucleotide (NAD) biosynthesis, has emerged as a master regulator of cellular metabolism and has demonstrated roles in the oxidative stress response and glucose homeostasis [18]. Emerging evidence has implicated NAMPT dysregulation in type 2 diabetes mellitus (T2DM) and its complications, including nephropathy and endothelial dysfunction [19, 20]. Notably, P7C3-induced NAMPT activation alleviates myopathy in diabetic mice [19], highlighting its therapeutic potential. However, the role of NAMPT in DK remains unexplored, representing a critical knowledge gap.

In this study, we identify mitochondrial NAMPT deficiency as a key driver of DK and show that NAMPT loss occurs earlier and more severely in the lacrimal gland than in the cornea, providing a mechanistic basis for tear secretion deficits that precede corneal lesions. We further demonstrate that topical activation of NAMPT with P7C3 restores corneal homeostasis, supports epithelial repair, and improves tear secretion. Together, these findings link DK to the broader pathology of diabetic dry eye syndrome and highlight NAMPT as a promising therapeutic target for oxidative stress–related corneal and tear secretion disorders.

Materials and methods

Collection of conjunctival epithelial cells

Conjunctival epithelial cells were collected from both eyes of 15 diabetic patients without corneal complications and 15 patients with DK using Supor-200 polyethersulfone membranes (0.2 μm pore size, 13 mm diameter). All participants met the following criteria: (1) had T2DM > 10 years, (2) had an HbA1c < 7.0% for 3 months, and (3) were aged 40–75 years. Patients with organic ocular diseases were excluded. This study was approved by the Ethics Committee of the Second Hospital of Tianjin Medical University and was conducted in accordance with the Declaration of Helsinki. Written informed consent was obtained from all participants.

iTRAQ labeling and liquid chromatography‒tandem mass spectrometry (LC‒MS/MS) analysis

Digested proteins were labeled with 4-plex iTRAQ reagents (AB Sciex, USA) and analyzed via LC‒MS/MS (Eksigent nanoLC 415 (AB Sciex, USA) with a TripleTOF 6600 system (AB Sciex, USA) using the following parameters: flow rate, 10 µL/min; mobile phase A, 2% ACN/0.1% FA; and mobile phase B, 97.9% ACN/0.1% FA. The raw data were processed with MaxQuant software (v1.6.17.0). Protein identification and database searches were conducted using the UniProt human proteome database (March 2020 release, FASTA format), with a false discovery rate (FDR) threshold of < 1%.

Bioinformatics analysis

Median normalization (R preprocessCore) of the log2-transformed data was performed. Unannotated and duplicate proteins were excluded. Differentially expressed proteins (DEPs) were identified via the limma package, with significance thresholds set at a |log2-fold change (FC)|>1.5 and a p value < 0.05. Downstream analyses included Pearson correlation analysis (corrplot), principal component analysis, Gene Ontology (GO) enrichment analysis and Reactome pathway analysis using clusterProfiler.

Human corneal and lacrimal gland samples

Human corneal and lacrimal gland tissues from T2DM patients and nondiabetic controls were acquired from the Tianjin Medical University Eye Hospital Biobank. Participants were enrolled in compliance with the Declaration of Helsinki following approval by the Ethics Committee (approval No. 2024KY-68). The clinical data of the donors are summarized in Table S1.

Animal models and drug treatment

Male db/db mice were used as a diabetes model in this study, and heterozygous db/m mice were used as nondiabetic controls. To evaluate the therapeutic potential of NAMPT activation in diabetic ocular surface disorders, the NAMPT agonist P7C3 (HY-15967; MedChemExpress, USA) was prepared as a 2.5 µg/µl solution in vehicle (0.9% saline containing 40% PEG300 and 5% Tween-80) and administered topically to both eyes of 12-week-old db/db mice. A total volume of 2 µl per eye was instilled twice daily for 4 weeks. A total of 2 µl of P7C3 was administered per eye twice daily for four weeks. The experimental groups included db/m mice (nondiabetic controls); untreated db/db mice (diabetic controls); vehicle-treated db/db mice; and P7C3-treated db/db mice. All procedures were carried out in compliance with the guidelines of the Association for Research in Vision and Ophthalmology (ARVO). The ethics committee of Tianjin Medical University Eye Hospital granted approval for all animal experiments (approval no. TJYY2022122059).

Corneal sensitivity

Corneal mechanosensitivity was assessed using a Cochet-Bonnet esthesiometer (Luneau Ophthalmologie, France) [21, 22]. Briefly, a 60 mm long nylon thread was applied perpendicularly to the corneal apex. Three measurements per eye were taken at 5-minute intervals, and the mean value was used for analysis.

Tear secretion analysis

Tear production analysis was performed with standardized phenol red-impregnated cotton threads (Jingming Biotechnology, China) as previously described [23]. The threads were precisely positioned at the lateral canthal fornix for 30 s using sterile forceps, and the distance that the liquid traveled was measured. Three consecutive measurements were obtained, and the mean wetting length was recorded as the final value.

Corneal fluorescein staining

Corneal epithelial integrity was assessed using standardized 1% fluorescein sodium staining [24]. After the precise administration of 2 µL fluorescein into the inferior conjunctival fornix using a micropipette, the dye was allowed to diffuse for 30 s. The fluorescence signals were then visualized using a slit-lamp biomicroscope with cobalt blue excitation filters. Two independent, blinded corneal specialists quantitatively assessed staining in central and peripheral corneal quadrants using the validated National Eye Institute (NEI)/Industry workshop grading scale (0 = no staining; 1 = 1–5 punctate dots; 2 = > 5 discrete dots or early coalescence; 3 = geographic coalescence). The total corneal fluorescein score was calculated as the sum of all quadrant values [25, 26].

Corneal thickness analysis

Six-micron-thick sections were stained with hematoxylin and eosin (H&E) and imaged using an Olympus BX51 brightfield microscope (Olympus, Tokyo, Japan). Morphometric analysis was performed to determine the epithelial, stromal, and total corneal thickness using CellSens Standard software.

Corneal nerve analysis

Corneal whole mounts were prepared as previously described [21, 27]. Enucleated eyes were fixed in 4% PFA, the limbus was dissected via four radial incisions, and the samples were permeabilized and stained with an Alexa Fluor 488-conjugated anti-βIII-tubulin antibody (1:100, ab195879; Abcam, UK). The samples were flat-mounted with ProLong Diamond Antifade Mountant (Thermo Fisher, USA). Confocal imaging (Zeiss LSM800) was used to quantify the central stromal nerve fiber length via NeuronJ (ImageJ) and peripheral subepithelial nerve density (% βIII-tubulin⁺ area in quadrants; threshold analysis).

Cell culture and treatment

The SV40-transformed human corneal epithelial cell line (STR-authenticated) [28] was routinely cultured in DMEM (5.5 mM glucose, Gibco, USA) supplemented with 10% fetal bovine serum (FBS, Gibco, USA), 1% penicillin‒streptomycin (P/S, Gibco, USA), 5 µg/mL insulin-transferrin-selenium (ITS, Sigma‒Aldrich, USA) and 10 ng/mL recombinant human epidermal growth factor (EGF, Gibco, USA). To model hyperglycemia, 30 mM D-glucose (Solarbio, China; high-glucose (HG) group) was added to the culture medium for 48 h, with 30 mM D-mannitol (Solarbio, China; HM group) used in the osmotic control group.

Human limbal epithelial cells (HLECs) were isolated and cultured as previously described [29]. Briefly, corneoscleral rims obtained after corneal transplantation were incubated in DMEM containing 15 mg/mL Dispase II (Roche Diagnostics, 4942078001, USA) at 4 °C overnight to separate the limbal epithelium. The tissue was then digested with 0.25% trypsin/0.02% ethylenediaminetetraacetic acid (EDTA) at 37 °C for 15 min to dissociate the cells. After resuspension, the cells were seeded onto culture dishes pre-coated with mitomycin C–treated 3T3 feeder cells.

The interventions included the following: Modulation of NAMPT expression via siRNA-mediated knockdown (50 nM, 48 h; sense: 5′-GGUGAAAUAUGAGGAGGAA-3′/antisense: 5′-UUCCUCCUCAUAUUUCACC-3′; GenePha-rma), an FK866 inhibitor (10 nM, 48 h; HY-50876, MedChemExpress), and a P7C3 agonist (10 µM, 48 h; HY-15976, MedChemExpress); Modulation of PINK1 expression via siRNA (50 nM, 48 h; sense: 5′-GAAGCCAUCUUGAACACAA-3′/antisense: 5′-UUGUGUUCAAGAUGGCUUC-3′; GenePharma), a CCCP uncoupler (10 µM, 6 h; HY-100941, MedChemExpress), and nicotinamide mononucleotide (NMN, 500 µM, HY-F0004S, MedChemExpress). Transfection was performed using jetPRIME^® when the cells reached 60% confluency (1.5 × 10⁵ cells/6-well), with nontargeting siRNAs/empty vectors used as controls.

Cell migration assay

Corneal epithelial cells were grown to 90% confluence in 6-well plates. A scratch was made using a 200 µL pipette tip, and wound closure was monitored at 0 and 24 h using an inverted phase-contrast microscope. The residual wound area was quantified with ImageJ software (MRI Wound Healing Tool plugin; NIH). The percentage of wound closure was calculated as [(A₀ - A₂₄)/A₀] × 100%.

Measurement of ROS and mitochondrial ROS levels

Cells were incubated with the ROS-sensitive probe 2’,7’-dichlorodihydrofluorescein diacetate (H₂DCFDA; 25 µM, ab113851; Abcam, UK) at 37 °C for 45 min, washed with HBSS, and imaged via confocal microscopy. TBHP (100 µM) was used as a positive control, and NAC (5 mM) was used as a negative control. The fluorescence intensity was quantified via a microplate reader and normalized to the fluorescence intensity of 5-(and-6)-carboxy-2’,7’-dichlorofluorescein diacetate (carboxy-DCFDA, C369; Invitrogen, USA) to correct for loading variations [30]. The fluorescence intensities were quantified using a microplate reader and are expressed as the H₂DCFDA/carboxy-DCFDA ratios to account for loading variations [31].

Mitochondrial ROS levels were assessed via incubation with MitoSOX Red (500 nM; M36007, Invitrogen, USA) for 15 min at 37 °C and counterstaining with Hoechst 33,342 (H3420, Solarbio, China). Rotenone was used as a positive control, and MitoTEMPO was used as a negative control. The fluorescence intensities were normalized to the intensity of the Hoechst signal and are expressed relative to the baseline mannitol signal.

For confocal imaging, glass-bottom dish-cultured cells were subjected to identical treatments; background-subtracted intensities were analyzed using ImageJ software (NIH) and normalized to the baseline mannitol intensity.

Mitochondrial membrane potential (ΔΨM) assessment

The ΔΨM was assessed using 50 nM tetramethylrhodamine methyl ester (TMRM) (I34361; Thermo Fisher Scientific, USA). The cells were incubated for 30 min at 37 °C, washed with PBS, and immediately imaged via confocal microscopy (10 µM CCCP was used as a control for ΔΨM dissipation).

To assess mitochondrial functional recovery capacity, the cells were first exposed to 4 µM antimycin A (a complex III inhibitor) and 10 µM oligomycin (an ATP synthase inhibitor) for 3 min to induce acute ΔΨM collapse [32]. Following inhibitor washout, the cells were incubated in complete medium supplemented with 20 nM TMRM. Time-lapse imaging was performed at baseline (0 min) and 60 min and 120 min after recovery. Five independent experiments were performed, with five random fields quantitatively analyzed per experiment.

Mitochondrial stress test

The mitochondrial stress test was performed using the Seahorse XF Cell Mito Stress Test Kit (#103015-100, Seahorse Bioscience, USA) according to the manufacturer’s instructions. Mouse corneal epithelial cells were isolated as previously described [33]. Briefly, mouse eyeballs were enucleated and incubated with Dispase II at 4 °C for 18 h to separate the corneal epithelium. The epithelial sheets were then further dissociated with trypsin at 37 °C to obtain a single-cell suspension and seeded into XFe96 microplates. HCECs were seeded into XFe96 microplates and treated as indicated. The sensor cartridge was hydrated overnight with Seahorse XF calibrant at 37 °C in a CO₂-free incubator. On the day of the assay, cells were washed twice with assay medium and incubated in DMEM supplemented with 1 mM pyruvate, 2 mM glutamine, and 10 mM glucose, followed by equilibration at 37 °C for 1 h. Oligomycin (1.5 µM), FCCP (1.0 µM), and rotenone/antimycin A (0.5 µM each) were sequentially injected. Oxygen consumption rate (OCR) values were normalized to cell number per well.

Apoptosis analysis

Apoptotic cells were detected using a TUNEL Assay Kit (C1088; Beyotime Biotechnology, China). The nuclei were counterstained with 4′,6-diamidino-2-phenylindole (DAPI). The apoptotic index was calculated as (TUNEL⁺ cells)/(DAPI⁺ cells) × 100%.

Quantification of NAD+ levels

NAD+ levels in corneal epithelial cells were quantified using a Coenzyme Ⅰ NAD (H) Content Assay Kit (Solarbio, BC0315) in accordance with the manufacturer’s instructions, and the values were normalized to the total protein content.

Immunofluorescence staining

Live cells were stained with MitoTracker™ Deep Red FM (200 nM, M66467; Invitrogen, USA) for 30 min at 37 °C. After fixation (4% PFA) and permeabilization (0.1% Triton X-100), nonspecific binding was blocked with 5% BSA. Following incubation with primary antibodies (Table S2), the cells were incubated with corresponding species-specific secondary antibodies (1:500, ab150113/ab150077, Abcam, UK). The nuclei were counterstained with DAPI (Solarbio, China). Confocal images were analyzed using the Colocalization Finder plugin of ImageJ software (NIH). Pearson’s correlation coefficients were calculated to quantify the localization of the target proteins in mitochondria. Similarly, mouse corneal sections were fixed and permeabilized, blocked, and then incubated sequentially with primary and secondary antibodies for immunofluorescence staining. Fluorescence intensity was quantified by measuring the mean fluorescence intensity from three distinct regions per section (one central and two peripheral fields), and the average of these three measurements was used as the representative value for each sample.

Quantitative reverse transcriptase polymerase chain reaction (qRT‒PCR) analysis

Total RNA was extracted from mouse corneal/lacrimal tissues (EZBioscience, USA), reverse transcribed into cDNA (EZBioscience, USA), and subjected to qRT‒PCR using SYBR Green Master Mix (Roche, Germany) on a LightCycler 480 II system (Roche, Germany). The reactions were run in triplicate; expression levels were normalized to β-actin expression and calculated via the 2 ^{− ΔΔCT} method [21]. The following primers were used: NAMPT forward: 5’-TTGTGTGAATTCCACCGGCT-3’; NAMPT reverse: 5’-GGGGATGGCCAAATACTGCT-3’; β-actin forward: 5’-CATCCGTAAAGACCTCTATGCCAAC-3’; β-actin reverse: 5’-ATGGAGCCACCGATCCACA-3’.

Protein preparation and Western blot analysis

Mitochondrial protein was extracted from human corneal epithelial cells (HCECs) using a Mitochondrial Isolation Kit (C3601; Beyotime Biotechnology, China) via differential centrifugation. Cytoplasmic/nuclear protein was extracted via a Nuclear/Cytoplasmic Extraction Kit (P0028; Beyotime Biotechnology, China). Cells from five 10-cm culture dishes were pooled as one sample for mitochondrial and cytoplasmic/nuclear protein extraction, and the experiment was repeated four times (n = 4). Total protein was extracted using RIPA buffer containing 1 mM PMSF (Solarbio, China) and phosphatase inhibitors (Cell Signaling Technology, USA). For Western blot analysis of total protein, at least five samples were included per group (n ≥ 5).

The proteins were resolved on SDS‒PAGE gels, transferred to PVDF membranes, blocked with 5% milk for 2 h at room temperature, and incubated with primary antibodies (Table S2; 4 °C overnight) followed by HRP-conjugated secondary antibodies (1:5000; Abcam, UK). The signals were detected using Omni-ECL substrate (EpiZyme, China) and quantified via ImageJ software.

Statistical analysis

All the statistical tests were carried out using GraphPad Prism 8. Normality was assessed using the Shapiro‒Wilk test. For comparisons between two groups, Student’s t tests were applied. Comparisons among multiple group (≥ 3 groups) were made by one-way ANOVA. All quantitative data are representative of a minimum of three biological replicates and are reported as the mean ± standard deviation (SD). p < 0.05 was considered to indicate statistical significance.

Results

Progressive downregulation of NAMPT during the course of diabetic keratopathy

Conjunctival impression cytology is a minimally invasive and repeatable sampling method in clinical practice [34]. It enables the collection of superficial conjunctival epithelial cells with minimal tissue damage and has proven useful for evaluating ocular surface diseases, especially those involving barrier dysfunction, inflammation, and epithelial abnormalities [35, 36]. To investigate the molecular mechanisms underlying DK, we first compared the proteomic profiles of conjunctival epithelial cells from diabetic patients with or without clinically diagnosed DK. Using stringent selection criteria, we identified 13 differentially expressed proteins, including 7 upregulated and 6 downregulated proteins. Among these, NAMPT showed a robust and consistent downregulation (Fig. 1A, B). Gene Ontology enrichment analysis further revealed significant perturbations in biological processes related to nonsense-mediated mRNA decay and T-cell proliferation, cellular components such as DNA repair complexes, SNARE complexes, and PP2A phosphatases, and molecular functions including DNA endonuclease activity, oxidized DNA binding, and ribosomal RNA binding (Fig. 1C). These findings suggest that DK is associated with a coordinated dysregulation of DNA damage responses, metabolic homeostasis, and immune regulation.

Fig. 1.

Progressive downregulation of NAMPT during the course of diabetic keratopathy. A Heatmap showing differentially expressed proteins (DEPs) in diabetic patients with or without keratopathy. B Volcano plot displaying the DEPs in diabetic patients with or without keratopathy. C Gene Ontology (GO) enrichment analysis of DEPs, including biological processes (BP), cellular components (CC), and molecular functions (MF). D, E Representative immunofluorescence staining of NAMPT (green) in human tissues. D Corneal tissues from non-diabetic controls (n = 2 donors, 4 samples) and patients with T2DM (n = 1 donor, 2 samples; scale bar, 20 μm). E Lacrimal gland tissues from non-diabetic controls (n = 2 donors, 4 samples) and T2DM patients (n = 1 donor, 2 samples); scale bar, 50 μm. F, G Corneal epithelial integrity in 8-, 12-, and 16-week-old mice assessed by fluorescein staining (n = 7 mice/group). H Tear secretion (n = 15 mice per group). I Corneal sensitivity (n = 8 mice/group). J, K NAMPT mRNA expression in the (J) cornea and (K) lacrimal gland (n = 6 mice/group). L–O Representative Western blot images and corresponding band quantification for NAMPT in the cornea (L, M) and lacrimal gland (N, O) (n = 6 mice/group). All measurements are shown as mean ± SD. Asterisks indicate statistical significance: ^*P < 0.05, ^**P < 0.01, ^***P < 0.001, ^****P < 0.0001; ns, not significant

On the basis of these proteomic findings, we next validated NAMPT expression in human diabetic tissues and in db/db mice to characterize its temporal dynamics during DK progression. Immunofluorescence staining demonstrated a reduction in NAMPT expression and distribution in the corneas and lacrimal glands of patients with diabetes (Fig. 1D, E). In db/db mice, age-dependent hyperglycemia developed (Fig. S1A), accompanied by characteristic ocular surface pathology. Sodium fluorescein staining revealed corneal epithelial lesions that progressed from focal defects in 8-week-old db/db mice to confluent epithelial damage by 16 weeks of age (Fig. 1F, G). Functional impairment followed a sequential pattern, with reduced tear secretion occurring at 8 weeks of age (Fig. 1H), followed by corneal hyposensitivity at 12 weeks of age (Fig. 1I), closely mirroring the clinical course described in diabetic patients [2].

Consistent with these phenotypic changes, quantitative PCR analysis revealed tissue-specific and time-dependent downregulation of NAMPT in db/db mice. NAMPT mRNA expression in the lacrimal gland declined significantly by 8 weeks, whereas corneal NAMPT mRNA levels decreased at 12 weeks (Fig. 1J, K). At the protein level, corneal NAMPT expression remained relatively stable until 16 weeks, when a significant 30.69% reduction was observed (Fig. 1L, M). In contrast, NAMPT protein levels in the lacrimal gland showed a progressive decline (15.65% at 12 weeks and 46.15% at 16 weeks; Fig. 1N, O). Notably, the kinetics of NAMPT depletion closely paralleled the onset and progression of ocular surface dysfunction, supporting a potentially causal role for NAMPT deficiency in diabetic ocular surface degeneration.

NAMPT depletion disrupts mitochondrial biogenesis, dynamics, and membrane homeostasis under HG conditions

To investigate NAMPT-mediated mechanisms under hyperglycemic stress, we first assessed HCECs viability (Fig. S1B) and NAMPT expression (Fig. S1C, D) after supplementing the basal medium (5.5 mM) with increasing concentrations of additional glucose. Based on these results, an extra 30 mM glucose (final concentration 35.5 mM) was selected as the HG model, with an equal concentration of mannitol (HM) used as the osmotic control. Given previous reports of the stress-induced nuclear translocation of NAMPT [32, 33], we performed subcellular fractionation analysis. Interestingly, this analysis revealed a selective depletion of mitochondrial NAMPT by 51.70% (Fig. 2A; Fig. S1E), which was confirmed by a diminished localization of NAMPT in mitochondria (Fig. 2B, C); however, the nuclear/cytoplasmic NAMPT level remained unchanged (Fig. S1F, G). Considering the essential role of mitochondria in cellular homeostasis through coordinated regulation of biogenesis, dynamics, and quality control [34, 35], we examined the effects of NAMPT inhibition. Both pharmacological (FK866, a 46.63% reduction in NAMPT levels) and genetic (siRNA, a 51.94% reduction in NAMPT levels) (Fig. 2D; Fig. S1H) inhibition of NAMPT recapitulated HG-induced mitochondrial dysfunction. This included decreased SOD1 expression (Fig. 2D; Fig. S1I), impaired mitochondrial biogenesis (decreased levels of nuclear respiratory factor 1 (NRF1), mitochondrial transcription factor A (TFAM), and translocase of outer mitochondrial membrane 20 (TOMM20); Fig. 2D; Fig. S1J-L), and disruption of mitochondrial dynamics (reduced levels of phosphorylated dynamin-related protein 1 (DRP1Ser616) and mitochondrial dynamin-like GTPase (OPA1) despite stable total DRP1 levels; Fig. 2E; Fig. S1M-P). H₂DCFDA and MitoSOX staining results also showed increased cellular ROS and mitochondrial superoxide levels, along with the complete loss of mitochondrial membrane potential (ΔΨM) (Fig. 2F-K; Fig. S1Q). Analysis of mitochondrial respiratory function using the Seahorse XF mitochondrial stress test showed that compared to controls, the experimental groups exhibited downregulation in basal respiration, ATP production, maximal respiration, and reserve capacity (Fig. 2L, M). Overall, NAMPT deficiency fully recapitulated the mitochondrial damage caused by hyperglycemia, establishing NAMPT as a central regulator of both redox balance and mitochondrial structural integrity in DK.

Fig. 2.

NAMPT depletion disrupts mitochondrial biogenesis, dynamics, and membrane homeostasis under HG conditions. A Representative Western blots of NAMPT in mitochondrial and cytoplasm fractions. B Fluorescence overlay images showing NAMPT (green), mitochondria (red), and nuclei (blue) distributions. Scale bars: 10 μm (main), 5 μm (zoom). (n = 5/group). (C) Fluorescence intensity profiles along indicated lines (ImageJ line scan analysis). D, E Representative Western blot images. F Representative fluorescence images of H₂DCFDA-stained cells. NAC: N-acetylcysteine negative controls; TBHP: tert-butyl hydroperoxide positive controls. scale bar = 20 μm. G Representative MitoSOX staining images MitoTEMPO: negative control; Rotenone: positive control. Scale bar: 10 μm. H Representative TMRM staining images, scale bar = 10 μm. I The H₂DCFDA/carboxy-DCFDA fluorescence ratio was measured using a microplate reader (n = 5/group). J Quantitative analysis of relative MitoSOX fluorescence intensity (normalized to Hoechst 33342 nuclear staining) measured by microplate reader (n = 5/group). K Quantitative analysis of relative TMRM fluorescence intensity (normalized to Mitotracker) (n = 5/group). L Seahorse mitochondrial stress test in HCECs. M Quantitative statistical analysis of basal respiration, ATP related respiration, maximal respiration and spare respiratory capacity (n = 5/group). All measurements are shown as mean ± SD. Asterisks indicate statistical significance: ^**P < 0.01, ^***P < 0.001, ^****P < 0.0001; ns, not significant

NAMPT-dependent mitophagy failure activates Cyt C -mediated apoptosis in diabetic corneal epithelial cells

The PTEN-induced putative kinase 1 (PINK1)/Parkin RBR E3 ubiquitin protein ligase (Parkin) signaling axis serves as the core regulatory pathway governing mitophagy [37]. To investigate the contribution of NAMPT to this pathway, we quantified mitophagy-related protein expression in HCECs exposed to HG and NAMPT inhibition. Compared with the controls, all experimental groups exhibited significantly decreased PINK1 and Parkin expression and a reduced microtubule-associated protein 1 light chain 3 beta-II/I (LC3B-II/I) ratio, together with sequestosome 1 (p62) accumulation (Fig. 3A–E). Confocal microscopy revealed reduced colocalization of PINK1, Parkin, and LC3B with mitochondria, whereas mitochondrial colocalization of p62 aggregates was increased (Fig. 3J–Q), consistent with a NAMPT-dependent impairment of mitophagic flux.

Fig. 3.

NAMPT-dependent mitophagy failure activates Cyt C-mediated apoptosis in diabetic corneal epithelial cells. A Representative Western blot images. B-I Western blot band quantification (n = 5–7/group). J-Q Representative confocal microscopy images and quantitative Pearson’s correlation coefficient analysis demonstrating colocalization of PINK1, Parkin, LC3B and p62 with mitochondria (MitoTracker, Red). Scale bars: 10 μm (main), 5 μm (zoom). Data represent three independent experimental replicates (n = 3/group) with four randomly selected fields quantified per sample. All measurements are shown as mean ± SD. Asterisks indicate statistical significance: ^*P < 0.05, ^**P < 0.01, ^***P < 0.001, ^****P < 0.0001; ns, not significant

This failure of mitochondrial quality control promoted apoptotic signaling, as evidenced by Western blot analyses. While NAMPT inhibition generally reduced both Bcl-2 and BAX expression (Fig. 3F, G), only FK866-treated cells showed a statistically significant reduction in the Bcl-2/BAX ratio (Fig. 3H). A consistent increase in cytochrome c (Cyt C) levels across NAMPT-deficient groups (Fig. 3I) indicated mitochondrial outer membrane permeabilization, likely mediated by mitochondrial permeability transition pore opening [37, 38]. We further validated these findings in primary HLECs. Specifically, cells were exposed to an additional 20 mM glucose added to the basal medium (final concentration 25.5 mM; HG; Fig. S2A) and, in parallel, treated with the NAMPT inhibitor FK866 or transfected with NAMPT-specific siRNA. Under these conditions, NAMPT was similarly downregulated, PINK1 expression was suppressed, and its recruitment to mitochondria was reduced (Fig. S2B–F). Taken together, these results indicate that hyperglycemia-induced NAMPT depletion promotes Cyt C-mediated apoptosis through impaired mitophagic clearance in diabetic corneal epithelial cells.

Pharmacological NAMPT activation mitigates HG-induced mitochondrial dysfunction in the corneal epithelium

Building upon these findings, we investigated the therapeutic potential of P7C3, a neuroprotective NAMPT activator [38, 39], in HG-treated HCECs. P7C3 treatment significantly accelerated corneal epithelial wound closure, increasing the healing rate from 37.1 ± 8.30% under HG conditions to 58.6 ± 5.90%, comparable to the HM control group (61.33 ± 10.34%) (Fig. 4A, B). In addition, P7C3 reduced intracellular ROS levels (Fig. 4C, D) and decreased mitochondrial superoxide production (Fig. 4D-F). At the molecular level, compared with the HG control, P7C3 significantly upregulated NAMPT expression (Fig. 4G, H) and enhanced antioxidant defenses, as evidenced by increased SOD1 levels (Fig. 4G, I). It also upregulated the mitochondrial biogenesis markers TFAM and TOMM20, whereas the increase in NRF1 expression did not reach statistical significance (Fig. 4G, J–L).

Fig. 4.

Pharmacological NAMPT activation mitigates HG-induced mitochondrial dysfunction in the corneal epithelium. A, B Wound healing progression of HCECs at 0 h and 24 h post-scratching (n = 5/group). Scale bar: 200 μm. C H₂DCFDA/carboxy-DCFDA fluorescence ratio measured by microplate reader (n = 5/group). D Representative fluorescence images of H₂DCFDA (scale bar = 20 μm) and MitoSOX stained cells (scale bar = 10 μm). E MitoSOX relative fluorescence intensity quantification. F Quantitative analysis of relative MitoSOX fluorescence intensity (normalized to Hoechst 33342 nuclear staining) measured by microplate reader (n = 5/group). G Representative western blot images of HCECs. H-L Western blot band quantification (n = 5–6/group). M Representative TMRM fluorescence images at baseline (pre-treatment) and 0, 60, and 120 min after treatment with 4 µM antimycin A plus 10 µM oligomycin (A + O). Scale bar: 10 μm; n = 5/group. N Kinetic profile of TMRM fluorescence intensity changes during membrane potential collapse (n = 5/group). O TMRM signal recovery at 60/120 min post- treatment, expressed as percentage of baseline levels (n = 5/group). All measurements are shown as mean ± SD. Asterisks indicate statistical significance: *P < 0.05, ^**P < 0.01, ^***P < 0.001, ^****P < 0.0001; ns, not significant

Seahorse mitochondrial stress test showed that P7C3 restored basal respiration, ATP production, and both maximal and spare respiratory capacity (Fig. S3A, B). Consistent with these protective effects, we next performed a pharmacological stress test to assess mitochondrial functional recovery. In this assay, antimycin A and oligomycin were used to induce an acute collapse of ΔΨm. This standard metabolic challenge [40] demonstrated that P7C3 restored ΔΨm recovery following injury, with HG+P7C3-treated cells showing 35.36% and 50.52% recovery at 60 and 120 min, respectively, compared with 16.89% and 36.67% for the HG controls (HM control: 38.53% and 58.45%; Fig. 4M–O).

Mechanistically, P7C3 selectively rescued PINK1 expression without altering Parkin levels, increased the LC3B-II/I ratio, reduced p62 accumulation (Fig. 5A-E), and enhanced the recruitment of PINK1, Parkin, and LC3B to mitochondria while decreasing mitochondrial localization of p62 (Fig. 5J–Q). P7C3 exerted antiapoptotic effects, increasin BCL-2 levels and the BCL-2/BAX ratio, limiting Cyt c release (Fig. 5A, F–I), and decreasing apoptosis (TUNEL staining: HG+P7C3, 7.15 ± 0.13% vs. HG control, 26.67 ± 9.43%; HM baseline, 2.34 ± 1.07%; Fig. 5R, S). Consistent with these results, P7C3 similarly restored NAMPT and PINK1 expression in HLECs and promoted the mitochondrial recruitment of PINK1 (Fig. S4A–E). Collectively, these results indicate that P7C3-mediated NAMPT activation protects corneal epithelial mitochondria from hyperglycemia by enhancing mitochondrial biogenesis, mitophagy, and antiapoptotic signaling.

Fig. 5.

Pharmacological NAMPT activation mitigates HG-induced mitochondrial dysfunction in the corneal epithelium. A Representative Western blot images of HCECs. B-I Western blot band quantification (n = 5–6/group). J-Q Representative confocal images and Pearson’s correlation coefficient analyses showing colocalization of PINK1, Parkin, LC3B and p62 with mitochondria (MitoTracker, red). Scale bars: 10 μm (main), 5 μm (zoom). Data represent three independent experiments (n = 3 per group), with four randomly selected fields analyzed per sample. R, S Representative TUNEL staining images and quantification of TUNEL-positive cells (n = 5/group). Scale bar: 100 μm. All measurements are shown as mean ± SD. Asterisks indicate statistical significance: ^*P < 0.05, ^**P < 0.01, ^***P < 0.001, ^****P < 0.0001; ns, not significant

NAMPT regulates PINK1/Parkin-mediated mitophagy in an NAD⁺-dependent manner under HG conditions

To investigate the regulatory role of NAMPT, we treated cells with CCCP (10 µM) to stimulate PINK1 expression to induce mitophagy [41, 42] and knocked down PINK1 with siRNA. Seahorse analysis showed that CCCP partially restored basal respiration, ATP production, maximal respiration, and spare respiratory capacity in si-NAMPT–transfected cells, whereas si-PINK1 largely abolished the P7C3-induced improvements in these parameters (Fig. 6A, B). PINK1 overexpression rescued NAMPT deficiency, increasing Parkin expression 2.4-fold and the LC3B-II/I ratio 3.6-fold and reducing p62 accumulation by 49.2% (nonsignificant) (Fig. 6C; Fig. S5A-D). This enhancement of mitophagic flux conferred antiapoptotic effects, increasing BCL-2 expression and reducing BAX and Cyt C release (Fig. 6C; Fig. S5E-H). Conversely, silencing PINK1 attenuated P7C3-induced mitophagy by suppressing Parkin expression and LC3B-II/I conversion without significantly altering p62 (Fig. 6C; Fig. S5A-D), BCL-2, or BAX expression (Fig. 6C; Fig. S5E -G). However, it increased Cyt c release by 136.7% (Fig. 6C; Fig. S5H). Confocal imaging confirmed that PINK1 overexpression reversed the si-NAMPT-induced decreases in mitochondrial PINK1, Parkin, and LC3B expression, whereas PINK1 knockdown blocked P7C3-mediated autophagosome recruitment (Fig. 6D, E; Fig. S5I-N).

Fig. 6.

NAMPT regulates PINK1/Parkin-mediated mitophagy in an NAD⁺-dependent manner under HG conditions. A Seahorse mitochondrial stress test in HCECs. B Quantitative statistical analysis of basal respiration, ATP related respiration, maximal respiration and spare respiratory capacity (n = 5/group). C Representative Western blot images of HCECs. D, E Representative confocal microscopy images and quantitative Pearson’s correlation coefficient analysis demonstrating colocalization of PINK1 and Parkin with mitochondria (MitoTracker, Red). Scale bars: 10 μm (main), 5 μm (zoom). Data represent three independent experimental replicates (n = 3/group) with four randomly selected fields quantified per sample. F, G Intracellular NAD⁺ levels were measured using an NAD (H) quantification kit (n = 5/group). H Representative Western blot images of HCECs. I Western blot band quantification (n = 6/group). All measurements are shown as mean ± SD. Asterisks indicate statistical significance: ^*P < 0.05, ^**P < 0.01, ^***P < 0.001, ^****P < 0.0001; ns, not significant

Next, we investigated whether NAMPT, as the rate-limiting enzyme for NAD biosynthesis in mammalian cells [43], exerts its effects via NAD. Modulating PINK1 did not alter NAD+ levels (Fig. 6F). Crucially, added NMN (a direct precursor of NAD) to cells in which NAMPT was silenced restored both NAD+ levels (Fig. 6G) and PINK1 expression (Fig. 6H, I). Consistent regulatory relationships were observed in HLECs, where CCCP, si-PINK1and NMN produced similar changes in NAMPT and PINK1 expression (Fig. S6A–G). Collectively, these findings indicate that NAMPT promotes PINK1 expression and its mitochondrial translocation in an NAD⁺-dependent manner to counteract HG-induced mitophagy impairment.

Topical administration of P7C3 ameliorates DK through the restoration of structure and function

To evaluate the therapeutic potential of NAMPT activation in DK, we administered P7C3 topically to the eyes of db/db mice daily for 4 weeks. The experimental groups included db/m mice (nondiabetic controls), untreated db/db mice (diabetic controls), vehicle-treated db/db mice (db/db + V group), and P7C3-treated db/db mice (db/db + P7C3 group). Initial assessments confirmed that P7C3 treatment exerted localized ocular effects without altering systemic metabolic parameters, including body weight and blood glucose levels (Fig. 7A, B). Therapeutic effects emerged progressively, with no significant improvements at the 2-week timepoint (Fig. 7C-F). However, by week 4, compared with the control treatment, P7C3 treatment exerted a significant therapeutic effect, as it resulted in reduced fluorescein staining and increased tear production (Fig. 7C-E). Notably, corneal sensitivity was nearly fully restored to levels comparable to those of db/m controls (Fig. 7F).

Fig. 7.

Topical administration of P7C3 ameliorates DK through the restoration of structure and function. A, B Body weight and blood glucose levels measured before treatment (12 weeks of age) and after 2 weeks (14 weeks of age) and 4 weeks (16 weeks of age) of topical P7C3 administration (body weight: n = 15 mice/group; blood glucose: n = 6 mice/group). C, D Corneal epithelial integrity assessed by fluorescein staining (n = 6 mice/group). E, F Tear secretion and corneal sensitivity analysis (n = 15 mice/group). G Representative H&E-stained corneal sections. Scale bars: 50 μm (overview), 20 μm (detail). H-J Quantification of (H) Full corneal thickness (I) Epithelial layer thickness (J) Stromal thickness (n = 4 mice/group). K Corneal nerve architecture (n = 5 mice/group). Top: Whole-mount βIII-tubulin staining (scale bar: 500 μm). Middle: Stromal nerve trunks (outlined in purple, scale bar: 100 μm) and vortex structures (50 μm). Bottom: subepithelial nerve plexus (scale bar: 20 μm). L Statistics of corneal stromal nerve lengths (n = 5 mice/group). M Statistics of subepithelial nerve density (n = 5 mice/group). All measurements are shown as mean ± SD. Asterisks indicate statistical significance: ^*P < 0.05, ^**P < 0.01, ^***P < 0.001, ^****P < 0.0001; ns, not significant

H&E staining revealed that P7C3 significantly protected against hyperglycemia-induced structural damage. The corneas of diabetic db/db mice presented stromal edema, with a 53.43% increase in stromal thickness (106.00 ± 4.26 μm vs. db/m controls: 69.08 ± 13.56 μm), which is consistent with hyperglycemia-driven dysregulation of hydration [44, 45]. P7C3 treatment reduced the stromal thickness by 20.10% (84.68 ± 7.71 μm), although the cornea was 22.35% thicker than that in the control group (Fig. 7G-J).

Neural regeneration analysis via immunofluorescence staining for βIII-tubulin highlighted the neurorestorative capacity of P7C3. The corneas of diabetic mice exhibited a marked reduction in the stromal nerve length, complete disruption of the central vortex architecture, and a markedly decreased subepithelial nerve density (Fig. 7K). P7C3 treatment significantly increased both the stromal nerve length and subepithelial nerve density, although complete architectural recovery required prolonged intervention (Fig. 7L, M).

P7C3 restores mitophagy and apoptosis in the corneas of diabetic mice

To comprehensively characterize the mechanisms underlying the protective effects of NAMPT activation against diabetic ocular surface pathology, we conducted an integrated analysis of mitophagy and apoptotic regulation. Topical administration of P7C3 for four weeks restored NAMPT expression in the corneas of diabetic mice to the levels in db/m controls, significantly exceeding the levels in untreated and vehicle-treated db/db mice (Fig. 8A, B). This restoration coincided with the upregulation of the mitochondrial outer membrane protein TOMM20 (Fig. 8A, C). P7C3 reversed the hyperglycemia-induced suppression of the key mitophagy regulators PINK1 and Parkin (Fig. 8A, D and E). Quantitative immunofluorescence analysis revealed tissue-specific restoration patterns: NAMPT and PINK1 expression recovered significantly, whereas TOMM20 and Parkin expression remained largely unchanged (Fig. 8F-M).The Seahorse mitochondrial stress test revealed that, compared with db/m controls, mitochondrial respiratory function was markedly impaired in corneal epithelial cells from db/db mice, as evidenced by significant reductions in basal respiration, ATP-linked respiration, maximal respiration, and spare respiratory capacity. In contrast, P7C3 treatment significantly restored mitochondrial function in db/db mice (Fig. S7A, B).

Fig. 8.

P7C3 restores mitophagy and apoptosis in the corneas of diabetic mice. A Representative Western blots of mouse corneal lysates. B-E Western blot band quantification (n = 5 mice/group). F-M Representative immunofluorescence images and fluorescence intensity quantification of NAMPT, TOMM20, PINK1, and Parkin. (n = 3 mice/group; Scale bar = 20 μm). N, O Representative TUNEL staining images and quantification of TUNEL-positive cells (n = 3 mice/group). All measurements are shown as mean ± SD. Asterisks indicate statistical significance: ^*P < 0.05, ^**P < 0.01, ^***P < 0.001; ns, not significant

Notably, the corneas of diabetic mice exhibited concurrent upregulation of proapoptotic markers (BAX/Cyt C) and the antiapoptotic protein BCL-2, whereas the expression of both proapoptotic and antiapoptotic proteins was suppressed in in vitro models of HG conditions. P7C3 selectively reduced BCL-2 overexpression without altering BAX or Cyt C expression (Fig. S7C-G). The percentage of apoptotic corneal epithelial cells in db/db mice (12.69 ± 2.02%) decreased to 5.66 ± 1.09% after P7C3 treatment, remaining elevated above the baseline level in db/m mice (0.71 ± 0.64%; Fig. 8N, O). These findings were further supported by corneal samples from patients with diabetes, which showed reduced TOMM20, PINK1, and Parkin expression compared with healthy controls (Fig. S7H-J). Corneas from patients with diabetes also exhibited a similar increase in apoptosis (Fig. S7K). Collectively, these findings indicate that P7C3 restores mitophagy and reduces apoptosis in the cornea in diabetes.

Discussion

DK poses a significant threat to vision because of its poorly defined pathogenesis and the current reliance on palliative clinical management. To address this, we conducted proteomic analysis of the conjunctival epithelium of diabetic samples to identify key mediators and evaluate therapeutic targets. Here, we identified NAMPT deficiency as a key pathogenic driver of mitochondrial dysfunction in DK, establishing the NAMPT/NAD⁺-mitophagy axis as a therapeutic target for DK.

NAMPT has dual roles as both an intracellular enzyme and a secreted hormone, suggesting its complex regulatory mechanisms [46, 47]. While prior studies have focused on cytoplasmic and nuclear NAMPT dynamics, demonstrating that genotoxic, oxidative, and dicarbonyl stress promote nuclear translocation to activate poly(ADP‒ribose) polymerase (PARP)/sirtuin activation [48, 49], the results of our subcellular fractionation analysis revealed a novel phenomenon: hyperglycemia-induced, cornea-specific depletion of mitochondrial NAMPT. Importantly, compared with the cornea, the lacrimal gland exhibited more pronounced and earlier NAMPT loss—a difference likely attributable to the significantly greater metabolic demand of the lacrimal gland (evidenced by 2–3-fold increase in the maximal OCR, basal ATP production, and the extracellular acidification rate (ECAR)) than corneal epithelial cells [12]. This metabolic disparity may accelerate NAMPT depletion and mitochondrial dysfunction, thereby providing a mechanistic basis for the clinical observation that tear secretion deficits precede corneal lesions in patients with diabetes [50, 51]. These findings suggest that NAMPT depletion and mitochondrial dysfunction are key contributors to both diabetic keratopathy and the broader pathology of Diabetic Dry Eye Syndrome.

NAMPT depletion via siRNA knockdown or FK866 inhibition recapitulated hyperglycemia-induced oxidative stress, underscoring the pivotal role of NAMPT in maintaining mitochondrial homeostasis through the coordinated regulation of biogenesis, fission‒fusion dynamics, and mitophagy. Notably, we identified NAMPT as the master upstream regulator of the PINK1-Parkin axis under hyperglycemic conditions, exerting control over mitochondrial integrity predominantly through a PINK1-dependent pathway. Because NAMPT is the rate-limiting enzyme in the NAD salvage pathway in mammalian cells [43], the regulatory effect of NAMPT on mitophagy is dependent on NAD, as confirmed in our study through supplementation with the NAD precursor NMN. In addition to its canonical role in cellular energy metabolism, NAD is implicated in diverse pathological processes, including diabetes, cancer, obesity, and aging [52]. Recent studies have highlighted the therapeutic importance of NAD: NAD biosynthesis is essential for corneal epithelial wound healing in diabetes [53] and protection against UVB-induced endothelial apoptosis [54]. Preclinical evidence has further demonstrated that subconjunctival administration of NAD or its precursors accelerates corneal wound repair and nerve regeneration in diabetes models [53]. Collectively, these findings establish the NAMPT-NAD axis as a critical therapeutic target for DK, bridging mitochondrial quality control to metabolic restoration.

Notably, the HG in vitro model and the chronic db/db mouse corneal model exhibit seemingly divergent patterns of BCL-2 family protein expression. Under acute HG conditions, BCL-2, BAX, and the BCL-2/BAX ratio are all downregulated in corneal epithelial cells, accompanied by a marked increase in TUNEL-positive cells. In contrast, in db/db corneas exposed to prolonged hyperglycemia and a chronic inflammatory microenvironment, BCL-2 is markedly upregulated, with BAX and the BCL-2/BAX ratio also showing a mild increase, while TUNEL-positive cells continue to accumulate. This pattern suggests concurrent activation of pro- and anti-apoptotic pathways under chronic stress: BAX upregulation reflects sustained ER stress–driven apoptosis [55–57], whereas elevated BCL-2 represents a compensatory survival response that fails to counteract cell death in the presence of strong pro-apoptotic signals [58]. Furthermore, a large body of research has shown that the balance between pro-apoptotic and anti-apoptotic members of the Bcl-2 family is a critical factor in tumorigenesis and diabetic complications [59, 60]. Importantly, by activating NAMPT, P7C3 restored the BCL-2/BAX balance and reduced apoptosis in both models, suggesting that NAMPT activation may suppress apoptosis in diabetic ocular surface disease by improving metabolic and stress status and re-equilibrating BCL-2 family signaling.

Recent research on therapeutic strategies for diabetic complications has increasingly focused on the precise modulation of mitochondrial quality control pathways [61, 62]. Compounds such as notoginsenoside R1 and resveratrol exert beneficial effects through PINK1-mediated mitophagy and AMPK activation, respectively [63, 64]; thus, in our study, we assessed the therapeutic potential of the NAMPT agonist P7C3. Topical P7C3 administration in a DK model exhibited significant therapeutic efficacy, effectively mitigating corneal epithelial damage, increasing tear secretion, and restoring mitotic activity. Notably, the limited reduction in p62 protein levels suggested that P7C3 primarily enhances early-stage autophagosome formation rather than promoting autophagolysosomal degradation, a mechanism of action distinct from that of lysosome-targeting agents such as chloroquine [65]. The cornea is the most densely innervated tissue in the human body, providing crucial protection and nutritional support to the ocular surface [66]. Numerous studies have shown that hyperglycemia impairs corneal nerve innervation [6, 7, 51, 67]. Our findings revealed that P7C3 significantly ameliorates diabetic corneal neuropathy, which aligns with the established neuroprotective effects of P7C3 in central nervous system injury models [38, 39]. This discovery extends the therapeutic potential of P7C3 to diabetic peripheral neuropathy, highlighting its multifaceted regulatory capabilities in neurovascular homeostasis.

Several methodological limitations should be acknowledged in this investigation. First, the 35.5 mM glucose concentration used in our in vitro experiments exceeds the physiological/pathophysiological range relevant to the cornea. To address this, we validated key findings under a more clinically relevant glucose level (25.5 mM) in primary HLECs and observed consistent overall trends. Future studies should evaluate a broader range of glucose concentrations and exposure durations in more physiologically relevant models. Mechanistically, while the data support the NAMPT-mediated regulation of PINK1 mitochondrial translocation via NAD, whether NAMPT synergistically modulates PINK1 through non-NAD-dependent mechanisms (e.g., direct protein interactions or kinase-mediated signaling) remains unvalidated. Notably, the limited availability of ocular specimens from diabetes patients limited our immunofluorescence staining analysis to a single patient-derived corneal/lacrimal gland sample pair, precluding robust statistical validation and potentially compromising generalizability to heterogeneous diabetic populations. In addition, our clinical proteomics screening was performed using conjunctival impression cytology samples because corneal tissue is rarely available in sufficient quantity; future studies incorporating proteomics directly from corneal specimens, when feasible, may provide greater disease relevance and stronger translational value. Nevertheless, addressing these questions will further accelerate clinical translation of NAMPT-targeted therapies for diabetic ocular complications.

Conclusion

These results demonstrate that hyperglycemia-induced mitochondrial NAMPT depletion is a central driver in the pathogenesis of DK. This depletion disrupts mitochondrial fusion-fission dynamics, inhibits PINK1 expression and its recruitment to mitochondria, and consequently impairs mitophagy. The resulting accumulation of dysfunctional mitochondria triggers Cyt C release, initiating an apoptotic cascade. Topical NAMPT activation mitigates hyperglycemia-induced ocular surface damage by restoring mitophagic flux (Fig. 9). Our findings suggest that targeting the NAMPT–NAD⁺–mitophagy axis may offer a promising strategy for restoring mitochondrial bioenergetic homeostasis in the diabetic state, establishing NAMPT as a potential therapeutic target for diabetes-related corneal, lacrimal gland, and nerve injuries.

Fig. 9.

Topical NAMPT activation restores mitophagic flux to rescue diabetic keratopathy via the NAD⁺/PINK1 Pathway. Hyperglycemia induces NAMPT depletion in mitochondrial of corneal epithelial cells, thereby disrupting mitochondrial homeostasis, suppressing mitophagy, and triggering the accumulation of damaged mitochondria. This cascade leads to cytochrome C release and subsequent apoptosis. P7C3 eye drops restore NAMPT activity, inhibit ROS accumulation, promote PINK1 expression and its mitochondrial translocation, and stimulate corneal epithelial and nerve regeneration, thereby alleviating diabetic keratopathy. This figure was created with Adobe Photoshop and BioRender.com

Abbreviations

BAX

Bcl-2 associated X protein

BCL-2

B-cell lymphoma 2

CCCP

Carbonyl cyanide m-chlorophenylhydrazone

Cyt C

Cytochrome c

carboxy-DCFDA

5-(and-6)-carboxy-2’,7’-dichlorofluorescein diacetate

DAPI

4′,6-diamidino-2-phenylindole

DK

Diabetic keratopathy

DRP1

Dynamin-related protein 1

HCECs

Human corneal epithelial cells

HG

High glucose

H₂DCFDA

2’,7’-dichlorodihydrofluorescein diacetate

HLECs

Human limbal epithelial cells

LC3B

Microtubule-associated protein 1 A/1B-light chain 3

NAMPT

Nicotinamide phosphoribosyltransferase

NAD

Nicotinamide adenine dinucleotide

NMN

Nicotinamide mononucleotide

NRF1

Nuclear respiratory factor 1

OCR

Oxygen consumption rate

OPA1

Mitochondrial dynamin-like GTPase

pDRP1

Phosphorylated dynamin-related protein 1

PINK1

PTEN-induced kinase 1

Parkin

Parkin RBR E3 ubiquitin protein ligase

ROS

Reactive oxygen species

SOD1

Superoxide dismutase 1

SQSTM1/p62

Sequestosome 1

TOMM20

Translocase of outer mitochondrial membrane 20

TUNEL

Terminal deoxynucleotidyl transferase dUTP nick-end labeling

TFAM

Mitochondrial transcription factor A

T2DM

Type 2 diabetes mellitus

ΔΨM

Mitochondrial membrane potential

Authors’ contributions

Qian Guo: Conceptualization, Methodology, Investigation, Data curation, Formal analysis, Writing–original draft, Writing – review & editing. Guannan Huang: Conceptualization, Methodology, Investigation, Data curation, Formal analysis. Wanting Li: Methodology, Investigation, Data curation, Validation. Yunjing Ma: Methodology, Validation., Writing – review & editing. Haini Liu: Methodology, Validation. Zihao Liu: Conceptualization, Methodology. Limin Zhu: Methodology, Resources. Lechong Ma: Methodology, Validation, Visualization. Qingjun Zhou: Conceptualization, Supervision, Project administration, Writing – original draft, Writing – review & editing. Shaozhen Zhao: Funding acquisition, Conceptualization, Supervision, Project administration, Resources, Writing – original draft, Writing – review & editing.

Funding

This research received funding from the National Natural Science Foundation of China (grant number 82271062) and the Tianjin Municipal Science and Technology Bureau Major Public Health Special Project (grant number 21ZXGWSY00060).

Data availability

The datasets used and analysed during the current study are available from the corresponding author on reasonable request. In addition, the original uncropped Western blot images used for quantification have been provided as a supplementary file.

Declarations

Ethics approval and consent to participate

The human study was conducted in accordance with the Declaration of Helsinki and was approved by the Independent Ethics Committees of Tianjin Medical University Second Hospital (Approval No. KY2018K001) and Tianjin Medical University Eye Hospital (Approval No. 2024KY-68). Written informed consent was obtained from all participants.

All animal experiments were performed in compliance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research and were approved by the Ethics Committee of Tianjin Medical University Eye Hospital (Approval No. TJYY2022122059).

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.