Untargeted Metabolomics Reveals the Role of Lipocalin-2 in the Pathological Changes of Lens and Retina in Diabetic Mice

Abstract

Purpose: To identify the role of lipocalin-2 (LCN2) in diabetic cataract (DC) and diabetic retinopathy (DR), diabetes models were established in wild-type (WT) and LCN2 gene knockout (LCN2^−/−) mice by streptozotocin (STZ), this study aimed to investigate the metabolic alterations and underlying pathways in the lens and retina.

Methods: Untargeted metabolomic analysis was performed on the lenses and retinas of WT and LCN2^−/− diabetic mice, and relevant pathways were predicted through bioinformatics analysis.

Results: LCN2 was notably elevated in the anterior capsules of DC and the vitreous humor of DR. Metabolic profiling of the lenses and retinas of diabetic mice indicated that the differential metabolites were mostly amino acids, fatty acids, carbohydrates, and their derivatives. In the lenses of STZ-induced WT mice, the differential abundance score (DA-score) revealed an increase in metabolites associated with the citrate (or TCA) cycle and glucagon signaling pathway, whereas a decrease was observed in metabolites related to cholesterol metabolism. After the knockout of LCN2, the DA-score indicated that the majority of metabolites involved in cholesterol metabolism, cysteine and methionine metabolism, and tryptophan metabolism were diminished. In the STZ-induced retina, there was an increase in metabolites associated with the mTOR signaling pathway, and this increase was inhibited by the knockout of LCN2.

Conclusions: Numerous metabolites exhibited substantial alterations in the lenses and retinas of diabetic mice. Untargeted metabolomics has provided insights into the function of LCN2 in DC and DR. These changes in metabolites, along with their related pathways, could be the mechanisms by which LCN2 modulated DC and DR.

Diabetes mellitus (DM), a chronic metabolic disorder, is known to trigger various ocular complications.¹ Diabetic eye disease remains the leading cause of blindness worldwide.² Notably, diabetic cataracts (DCs) are prone to emerge at an early stage and develop rapidly.³ The formation of DCs involves complex mechanisms, including swelling of lens fibers caused by hyperosmolarity due to the accumulation of polyols,⁴ the acceleration of advanced glycation end product (AGE) accumulation due to hyperglycemia,⁵ and the process of epithelial–mesenchymal transition (EMT).⁶ It was reported that autophagy was impaired in lens epithelium cells (LECs) exposed to hyperglycemia.⁷ Recent studies have also shown that DCs are strongly associated with oxidative stress due to activation of the renin–angiotensin system.⁸ Meanwhile, diabetic retinopathy (DR) is a major eye complication of DM, affecting more than 100 million people worldwide, and it is also an independent risk factor for cataract in diabetic patients.^9–11 Oxidative stress, endoplasmic reticulum stress, apoptosis, and autophagy blockage induced by high glucose play an important role in the formation and development of DR.^12–14 It has been reported that the process of autophagy is affected by oxidative stress and endoplasmic reticulum stress.^15,16 Given that DCs and DR are complications of diabetes with overlapping pathogenic processes, the pursuit of targets that can simultaneously regulate both conditions is advantageous for early detection and therapeutic intervention. 

Lipocalin-2 (LCN2), also known as neutrophil gelatinase-associated lipocalin (NGAL), is a secreted glycoprotein belonging to the lipocalin superfamily,¹⁷ which has been proven to participate in the mechanism of insulin resistance in diabetes. Studies have shown that knockout of the LCN2 gene in mice can prevent insulin resistance caused by a high-fat diet.¹⁸ However, LCN2-deficient mice have also been observed to develop insulin resistance, dyslipidemia, and fatty liver disease.¹⁹ In diabetic retinopathy, downregulation of LCN2 has been found to alleviate retinal vascular dysfunction by inhibiting cellular migration, invasion, angiogenesis, and caspase-1–regulated pyroptosis.²⁰ LCN2 has also been studied as a biomarker for kidney injury, metabolic disorders, non-alcoholic fatty liver disease, and cardiovascular complications.^21–24 LCN2 promotes the process of EMT in cancer cells.²⁵ Recent studies have suggested a potential role of LCN2 in neurological disorders, including Alzheimer's disease, Parkinson's disease, and stroke.²⁶ Current research is progressively uncovering complex interactions of LCN2 within various pathologies and physiological systems, yet its mechanisms are not fully understood. Further rigorous scientific investigation is warranted to clarify the functions of LCN2 and to determine its potential as a therapeutic target or diagnostic biomarker in multiple diseases. 

Several metabolomic studies have identified metabolic alterations in diabetic retinopathy, shedding light on its pathogenic processes.²⁷ In light of this, we opted to utilize streptozotocin (STZ) to induce diabetes in mice, as it exerts its effect by selectively destroying insulin-producing β cells in the pancreas, thereby inducing insulin deficiency and resulting in hyperglycemia.²⁸ Additionally, we selected untargeted metabolomics as our investigative approach, which is not predicated on pre-existing knowledge of the metabolome and has the capacity to uncover novel and unforeseen metabolic alterations²⁹ in order to discern the metabolic alterations in DCs and DR as comprehensively as possible, thereby providing valuable insights for the early prediction and diagnosis of these conditions. 

Materials and Methods

Establishment of Diabetic Mouse Model and Sample Preparation

LCN2 knockout (LCN2^−/−) male mice were obtained from The Jackson Laboratory (024630, strain C57BL/6; The Jackson Laboratory, Bar Harbor, ME, USA), and both male and female mice were homozygote. Genotype identification results of LCN2^−/− are presented in Supplementary Figure S1. Wild-type mice (C57BL/6) were purchased from Slack Animal Co. (Changsha, Hunan Province). We divided all of the 8-week-old male mice into four groups: wild-type control (WT-SC), LCN2^−/− control (LCN2^−/−-SC), wild-type streptozotocin-induced (WT-STZ), and LCN2^−/− streptozotocin-induced (LCN2^−/−-STZ) groups. Mice in the STZ groups were intraperitoneally injected with STZ solution (50 mg/kg for 5 days, MP Biomedicals, Santa Ana, CA, USA). Control groups received an intraperitoneal injection of sodium citrate buffer (ECOTOP SCIENTIFIC Biotechnology Co., Ltd., Guangzhou, China). Blood was collected from the tail vein 1 week after STZ injection using a precision glucose meter (CONTOUR TS; Bayer, Leverkusen, Germany) to confirm diabetes, and mice with blood glucose concentrations >300 g/dL were considered diabetic and used for subsequent experiments. Body weight and blood glucose were measured and recorded every week until the end of the experiment. At 12 weeks after STZ injection, we cut the limbus and removed the lens and retina as carefully as possible to avoid damage, and images of each group of lenses were taken under a stereomicroscope at the same magnification. The lenses and retinas of the same mouse were used as a sample, with eight or nine biological replicates per group. 

Collection of Clinical Specimen

The anterior capsules of the lenses were collected from patients diagnosed with cataract in the ophthalmology department of Xiangya Hospital, Central South University. Phacoemulsification and intraocular lens implantation were performed by the same surgeon, and the anterior capsule was obtained after continuous annular capsulorhexis. Vitreous fluid samples for this study were taken from DR patients and non-DR patients requiring vitrectomy. All specimens were immediately placed into Eppendorf tubes after separation, marked, and placed in a 4°C refrigerator, and stored in a –80°C refrigerator after the operation. Five or six capsule specimens were placed together to form a group. 

Inclusion criteria included being diagnosed as DC by ophthalmology. Exclusion criteria for the DC group were as follows: (1) traumatic cataract, drug-induced and toxic cataract, radiation cataract, and other cataracts with obvious inducements; (2) glaucoma, high myopia, eye trauma, other eye medical history, and eye surgery history; (3) diagnosis of high blood pressure, heart disease, hyperthyroidism, or other diseases having an impact on the metabolic status of the whole body. For the age-related cataract (ARC) group, exclusion criteria were the same as DC as well as no diabetes mellitus. The two groups were matched for age. 

Inclusion criteria for the DR group included a diagnosis of DR being confirmed by fundus examination and fluorescence angiography. Exclusion criteria for the DR group were as follows: (1) previous ocular diseases such as glaucoma, uveitis, retinal vein obstruction, and neovascular glaucoma (not caused by diabetes); (2) previous history of ocular trauma or fundus surgery; (3) endophthalmitis; and (4) systemic diseases such as tumor. For the control group (non-DR), the patients with the following conditions were excluded: rhegmatogenous retinal detachment, idiopathic macular hole, or idiopathic epimacular membrane diagnosed by a fundologist and requiring vitrectomy, as well as the exclusion criteria for the DR group. The two groups were matched for age. 

Ethics Statement

Animal experiments were conducted in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. Informed consent was obtained from the patients for the surgical specimens, and the study adhered to the tenets of the Declaration of Helsinki for research involving human participants. The study protocol was approved by the Ethical Review Committee of Xiangya Hospital, Central South University, China (202112639). 

Genotype Identification

The tail tips of 8-week-old mice were sterilized with 75% alcohol, and 1 to 3 mm of tail tissue was cut. Next, mouse tails were lysed and subjected to a metal bath at 55°C overnight. After centrifugation, proteinase K was inactivated by a metal bath at 99°C. DNA was diluted by adding 130 µL of water to the sample with completed inactivation. Polymerase chain reaction (PCR) of the target gene was then performed according to the reaction system in Supplementary Table S1 and the reaction procedure in Supplementary Table S2. PCR products were electrophoresed in 2% agarose gels and then visualized and photographed under ultraviolet light. The primer sequences used in the experiments are shown in Supplementary Table S3. 

Western Blotting

The anterior capsules were collected during cataract surgery, followed by appropriate additions of radioimmunoprecipitation assay (RIPA) lysate and protease inhibitor (100:1); they were then fully ground. The remaining supernatant was added to 1/4 volume of 5× protein loading buffer (sodium dodecyl sulfate [SDS]), heated and denatured in a metal bath at 95°C for 10 minutes, cooled and centrifuged, then divided into aliquots and stored at –80 °C. The collected vitreous fluid was centrifuged, and the supernatant was added directly to 1/4 volume of 5× protein loading buffer (SDS), followed by denaturation as described above. The 20 µg of protein in each sample was added to each lane for electrophoresis, and then the proteins were transferred to the polyvinylidene fluoride (PVDF) membrane, followed by blocking with 5% skim milk powder. The primary antibody—Anti-Lipocalin-2/NGAL (dilution ratio 1:1000, ab125075; Abcam, Cambridge, UK); Beta Actin Recombinant antibody (dilution ratio 1:10,000, 81115-1-RR; Proteintech, Rosemont, IL, USA); GAPDH Monoclonal Antibody (dilution ratio 1:10000, 60004-1-IG; Proteintech)—was incubated overnight, and the membrane was washed with phosphate-buffered saline with Tween 20 (PBST) to incubate the secondary antibody, HRP-conjugated Goat Anti-Mouse/Rabbit IgG (H+L) (dilution ratio 1:10,000, SA00001-1/SA00001-2; Proteintech) of the corresponding species. Finally, the membrane was washed again using PBST and developed using chemiluminescence solution. 

Hematoxylin and Eosin Staining

The whole eyeball was carefully removed and transferred to eyeball fixation solution. After fixation at room temperature for more than 24 hours, the eyeball was placed in the embedding frame, followed by dehydration with gradient alcohol and repeated xylene immersion and paraffin melting at 65°C. The eyeballs that completed the embedding process were placed in an embedding machine for embedding and sectioning (4 µm), and the slices were baked in an oven at 60°C. For staining, the sections were deparaffinized, stained with hematoxylin and eosin (H&E), and photographed under the microscope after sealing. 

Optical Coherence Tomography

We anesthetized the mice with 1% pentobarbital, and the viscoelastic agent was applied to the eyeballs of the mice after mydriasis to prevent dryness of the ocular surface. Images were taken using Phoenix image-guided optical coherence tomography (OCT; Phoenix Technology Group, Pleasanton, CA, USA). After the gamma value was adjusted, the OCT image of the optic nerve in the center position was obtained, and ImageJ (National Institutes of Health, Bethesda, MD, USA) was used to measure the overall thickness of the retina at the corresponding position in each group. 

Evans Blue Staining

The mouse was secured in a custom-made restrainer, and 2% Evans blue solution (45 mg/kg) was administered via tail vein injection. The immediate blue coloration of the mouse's body, notably on the lips, ears, and paws, confirmed the success of the procedure. The dye was allowed to circulate for 1 hour, after which the eyes were enucleated and fixed in 4% paraformaldehyde (PFA) overnight. The retina was gently dissected for mounting, and an anti-quenching agent was applied. Finally, the retinas were examined under a confocal microscope (Carl Zeiss Microscopy, Jena, Germany) to evaluate the presence and degree of any leakage within the retinal tissue. 

Immunofluorescent Staining

Following the previously described protocol, intact mouse eyes were harvested 3 months post-modeling and fixed overnight in 4% PFA. The eyes were then subjected to a sucrose gradient (10% and 20%) for dehydration and infiltration, followed by embedding and quick-freezing with optimal cutting temperature (OCT) compound. Sections were cut to a thickness of 16 µm and rapidly applied to adhesive slides, then air-dried at room temperature for 24 hours. The slides were then circumscribed with a histology pen around the tissue sections, soaked in PBS for 10 minutes to remove the OCT, and subsequently fixed with pre-chilled 4% PFA for 15 minutes. After the slides were rinsed with PBS for 5 minutes, they were permeabilized with 0.3% Triton X-100 for 15 minutes. The tissue was then blocked with 5% BSA at room temperature for 30 minutes, followed by incubation with the primary antibody (Anti-Lipocalin-2/NGAL, dilution ratio 1:100, ab125075, Abcam; Anti-GFAP, dilution ratio 1:600, 4412, Cell Signaling Technology, Danvers, MA, USA) overnight at 4°C. The slides were allowed to equilibrate to room temperature for 15 minutes, and they were then rinsed with PBS for 10 minutes three times. The fluorescently labeled secondary antibody (Anti-rabbit/mouse IgG (H+L), dilution ratio 1:1000, 4412/8890; Cell Signaling Technology) was applied, followed by incubation in the dark at room temperature for 2 hours. After incubation, the slides were washed with PBS for another three rounds of 10 minutes each. Then, they were incubated with 4′,6-diamidino-2-phenylindole (DAPI) for 5 minutes. Finally, an antifluorescent quencher was applied to the slides and images were captured using a laser confocal microscope (Carl Zeiss Microscopy, Jena, Germany). 

Metabolic Analysis

Sample Preparation

The appropriate sample was weighed into a 2-mL centrifuge tube, and 1000 µL tissue extract (75%, 9:1 methanol:chloroform; 25% H₂O) was added. Then, a steel ball was added for grinding, followed by sonication for 30 minutes and placement on ice for 30 minutes. After centrifugation, all of the supernatant was collected in a centrifuge tube, concentrated, and dried. Finally, 200 µL 50% acetonitrile solution was added to prepare 2-chloro-l-phenylalanine solution (4 ppm) to redissolve the sample, and the filtrate was added to the detection flask for liquid chromatography–mass spectrometry (LC-MS) detection. 

Liquid Chromatography Conditions

The LC analysis was performed on a Vanquish UHPLC System (Thermo Fisher Scientific, Waltham, MA, USA). Chromatography was carried out with an ACQUITY UPLC HSS T3 (150 × 2.1 mm, 1.8 µm; Waters, Milford, MA, USA). The column was maintained at 40°C. The flow rate and injection volume were set at 0.25 mL/min and 2 µL, respectively. A column was used at a flow rate of 0.25 mL/min, temperature of 40°C, and injection volume of 2 µL. In the positive ion mode, the mobile phase was 0.1% formic acid acetonitrile (B2) and 0.1% formic acid water (A2); in the negative ion mode, the mobile phase was acetonitrile (B3) and 5-mM ammonium formate water (A3). 

Mass Spectrum Conditions

Mass spectrometric detection of metabolites was performed on an Orbitrap Exploris 120 (Thermo Fisher Scientific) with an electrospray ionization (ESI) ion source, and data were collected separately in the positive and negative ion modes. The parameters were as follows: sheath gas pressure, 30 arb; auxiliary gas flow, 10 arb; spray voltage, 3.50 kV and –2.50 kV for ESI+ and ESI–, respectively; capillary temperature, 325°C; MS1 range, 100–1000 m/z; MS1 resolving power, 60,000 full width at half maximum (FWHM); number of data-dependent scans per cycle, 4; MS/MS resolving power, 15,000 FWHM; normalized collision energy, 30%; dynamic exclusion time, automatic. 

Quality Control

The R package ropls (R Foundation for Statistical Computing, Vienna, Austria) was used to perform principal component analysis (PCA), partial least squares discriminant analysis (PLS-DA), and orthogonal partial least squares discriminant analysis (OPLS-DA) dimensionality reduction analysis on the sample data. The data were scaled, and the score plot, load plot, and S-plot were drawn to show the differences in metabolite composition between each sample. The model was tested for overfitting using the permutation test. R2X and R2Y represent the interpretation rate of the model for the X and Y matrices, respectively. The P value was calculated according to the statistical test, the variable importance in projection (VIP) was calculated by the OPLS-DA dimensionality reduction method, and the fold change (FC) was calculated. The influence of strength and the interpretation ability of each metabolite component content on sample classification and discrimination were measured to assist the screening of marker metabolites. Metabolite molecules were considered statistically significant for P < 0.05 and VIP > 1. 

Kyoto Encyclopedia of Genes and Genomes Enrichment Analysis

The Kyoto Encyclopedia of Genes and Genomes (KEGG) is a powerful tool for metabolic analysis and metabolic network research in organisms. It helps researchers to study genes and their expression information as a whole network. The vertical axis represents the metabolic pathway impact, and the horizontal axis represents the impact of pathway enrichment. This value can be simply understood as the contribution—that is, the higher the value, the higher the contribution of metabolites detected under this pathway. The color is related to the P value—redder P values are smaller, and bluer P values are larger; smaller the P value represent a more significant impact of the detected differential metabolites on this pathway. 

Statistics

Prism 9.0 (GraphPad, Boston, MA, USA) was used to analyze the statistical correlation of the statistical values. An unpaired t-test was used for comparison between the two groups if the data conformed to a normal distribution. For comparisons among multiple sets of data, two-way ANOVA with Tukey's post hoc was used for bivariate analysis. All values are presented as mean ± SEM, and P < 0.05 was used to indicate statistically significant differences. P < 0.05, P < 0.01, P < 0.001 and P < 0.0001 are indicated by * (#), ** (##), *** (###), and **** (####), respectively. The datasets used and analyzed during the current study are available from the corresponding author on reasonable request. 

Results

Upregulation of LCN2 in Patients With DC and DR

To investigate the expression of LCN2 in DC and DR, ARC and DC groups with similar cataract severity were selected based on slit-lamp observations and preoperative fasting blood glucose tests. At the same time, we collected vitreous humor from patients with DR and non-DR requiring vitrectomy surgery and excluded patients with eye trauma and endophthalmitis. Western blot analysis showed that LCN2 expression was significantly increased in the anterior capsules of DCs and in the vitreous humor of DR compared with ARC and non-DR (Fig. 1), which suggested that LCN2 may play a crucial role in the pathogenesis and progression of DC and DR. 

Figure 1.

 

Upregulation of LCN2 in the anterior capsule of DC and the vitreous humor of DR. (A, B) Western blot analysis and quantitative data for LCN2 expression in LECs from the anterior capsules of ARCs and DCs, with GAPDH used as a loading control. (C, D) Western blot analysis and quantitative data for LCN2 expression in the vitreous humor of DR and non-DR, with β-actin used as a loading control. **P < 0.01, ***P < 0.001 (n = 6).

Pathological Changes in STZ-Induced Diabetic Mice

We used STZ intraperitoneally injected into 8-week-old male mice as a model for inducing diabetes. The weight of mice in the control group increased with age, and the weight of mice in the diabetes model group decreased distinctly compared with the age-matched control group. In diabetic mice, blood glucose increased apparently 1 week after STZ injection and remained at a high level from 2 weeks to 12 weeks (Fig. 2A). It was observed under stereomicroscope that the lenses of diabetic mice in both groups had a slight degree of opacity (Fig. 2B). At the same time, to confirm the occurrence of DR, the Evans blue assay was used to detect retinal vascular leakage (Fig. 2C). In age-matched control mice (WT-SC and LCN2^−/−-SC), the retinal blood vessels exhibited a uniform network structure with equal diameter, regular shape, and even distribution, without fluorescence leakage, whereas the WT diabetic mice demonstrated a marked increase in retinal vascular leakage as indicated by Evans blue staining. Similarly, LCN2^−/− diabetic mice also presented with noticeable angiectatic changes. We further performed OCT imaging and compared the retinal thickness of the age-matched (20 weeks) groups of mice, and retinal thickness was measured equidistant from the center of the optic nerve papilla. We found that overall retinal thickness was reduced in diabetic mice, but no evident thinning was observed in the LCN2^−/−-STZ group (Figs. 2D, 2G). On the other hand, HE retinal staining showed the same change trend as OCT (Fig. 2E), and quantitative analysis revealed that the thinning primarily occurred in the inner plexiform layer (Fig. 2F). These results indicated that LCN2 knockout can improve the retina structure of DR. 

Figure 2.

 

Pathological changes in STZ-induced diabetic mice. (A) Changes in blood glucose and body weight of WT (WT-STZ) and LCN2^−/− (LCN2^−/−-STZ) mice induced by intraperitoneal injections of STZ. The control group received sodium citrate solution (WT-SC and LCN2^−/−-SC). The observation period spanned 12 weeks after STZ injection. ns (blue), compared to the WT-SC; ns (green), compared to the WT-STZ. ****P < 0.0001 compared to the WT-SC (n = 24). (B) Representative images of lenses in four groups of mice. Scale bar: 1 mm. (C) Representative fluorescence signal images of flatmounted retinas after injection of Evans blue dye. Scale bar: 100 µm (n = 4). (D) Representative images of retinal OCT in four groups of mice (n = 3). (E) Representative retinal H&E staining images for the four groups. Scale bar: 50 µm (n = 3).(F, G) Quantitative analysis of total retinal and inner plexiform layer thickness (n = 3). *P < 0.05, **P < 0.01, ***P < 0.001.

Quality Control of Untargeted Metabolomics Data

PCA demonstrated tight clustering of the quality control samples in both cation and anion modes (Figs. 3A, 3B), indicating excellent experimental reproducibility. To determine the differences in metabolic profiles, we performed OPLS-DA score plots to assess the variations between the lens and retina groups. As shown in Figure 3C, significant separation was observed among the LCN2^−/−-STZ group, WT-STZ group, LCN2^−/−-SC group, and WT-SC group in the cation mode. The retinal LCN2^−/−-STZ group, WT-STZ group, LCN2^−/−-SC group, and WT-SC group exhibited a similar separation trend (Fig. 3E). When we compared the four groups of lens samples, the R² and Q² values of the OPLS-DA model were 0.992 and 0.911, respectively. At the same time, for the comparison of the four groups of retinal samples, the R² and Q² values of the OPLS-DA model were 0.997 and 0.845, respectively, which indicated that the models had strong interpretability and predictability. We also employed permutation testing (as depicted in Figs. 3D, 3F) as an external validation approach to conduct permutation analysis on the OPLS-DA model, revealing that the OPLS-DA models were robust and effective. 

Figure 3.

 

Quality control of untargeted metabolomics data. In each image of Fig. 3, A–D represent the LCN2^−/−-STZ group, WT-STZ group, LCN2^−/−-SC group, and WT-SC group of the lenses and E–H represent the LCN2^−/−-STZ group, WT-STZ group, LCN2^−/−-SC group, and WT-SC group of the retinas. (A) Principal component analysis of the lens and retinal groups in cationic mode. (B) Principal component analysis of lens and retinal groups in anion mode. (C, D) Score plots and permutation analysis plot of OPLS-DA among the four lens groups from STZ-induced WT mice and LCN2^−/− mice in cationic mode. (E, F) Score plots and permutation analysis plot of OPLS-DA among the four retina groups from STZ-induced WT mice and LCN2^−/− mice in cationic mode.

Metabolic Changes in Mice Lenses by STZ Induction

In the lenses of WT-SC and WT-STZ mice, 136 differential metabolites were detected, 71 upregulated and 65 downregulated, using the screening criteria of P < 0.05 and VIP > 1 (Figs. 4A, 4B). We selected the top 50 differential metabolites ranked by P value for classification and found that amino acids and fatty acyls were the main categories of differential metabolites (Fig. 4C). As Supplementary Fig. S2A demonstrates, based on P value the top five enriched pathways were ABC transporters, central carbon metabolism in cancer, taste transduction, GABAergic synapse, and phenylalanine metabolism. Additionally, the differential abundance score showed an increase in metabolites related to the TCA cycle and glucagon signaling pathway in the lenses of WT-STZ mice, with a decrease in metabolites related to cholesterol metabolism (Fig. 4D). The chord plot in Figure 4E and the circle plot in Figure 4F provide a comprehensive view of the types of metabolites implicated in the top 10 metabolic pathways with P < 0.05, along with the quantities of these metabolites and their patterns of upregulation or downregulation. Moreover, the z-scores for the metabolic pathways shown in Figure 4F were detailed in Supplementary Table S4A. These results suggest that the STZ-induced diabetes in the mice caused the disorders of glucose metabolism, lipid metabolism, and amino acid metabolism in the lens and verify successful induction of the mouse diabetes model. 

Figure 4.

 

Metabolic changes in mice lenses by STZ induction. In each image of Fig. 4, (B) and (D) represent the WT-STZ group (n = 9) and WT-SC group (n = 8) of the lenses. (A) Volcano plot of 136 differential metabolites. Compared to the WT-SC group, metabolites that were significantly upregulated in the WT-STZ group are marked in red, with VIP > 1, P < 0.05, and fold change (FC) > 1.5. Metabolites that were remarkedly downregulated in the WT-STZ group are marked in blue. Metabolites with VIP > 1, P < 0.05, and FC < 1.5 are indicated in yellow. Non-significant metabolites are represented in gray. The size of the dots corresponds to the magnitude of the VIP values. (B) Heatmap of 136 differential metabolites. Red represents significantly upregulated metabolites in the WT-STZ group compared to the WT-SC group, and blue represents downregulated metabolites. (C) Classification of 136 differential metabolites. (D) The top 20 differential abundance score with P < 0.05 based on KEGG enrichment analysis. DA-score = (number of upregulated metabolites − number of downregulated metabolites)/(total number of differential metabolites in the pathway). (E) Chord plots of the top 10 KEGG enrichment terms with P < 0.05. (F) Circle plots of the top 10 KEGG enrichment terms with P < 0.05. Red represents metabolites with elevated expression, and blue represents metabolites with decreased expression.

Metabolic Alterations in Lenses of STZ-Induced Wild-Type and LCN2^−/− Mice

A total of 54 differential metabolites were detected between the LCN2^−/−-STZ and groups according to the screening criteria as described previously. Twenty-three differential metabolites were upregulated, and 31 differential metabolites were downregulated (Figs. 5A, 5B). Supplementary Figure S2B presents the enriched pathways with statistically significant differences at P < 0.05, with the leading three being glycine, serine, and threonine metabolism; glyoxylate and dicarboxylate metabolism, and cholesterol metabolism. Concurrently, the chord plot illustrates the metabolic products involved in these enriched pathways (Fig. 5C). On the other hand, we observed that after the knockout of LCN2, the DA-score indicated that the majority of the metabolites involved in these pathways were diminished, notably cholesterol metabolism, cysteine and methionine metabolism, and tryptophan metabolism (Fig. 5D). The differential metabolites involved in pathways with the top three P values based on the KEGG enrichment analysis are shown in Figure 5E. Cholesterol, l-homoserine, glyceric acid, glycocholic acid, mesaconate, and l-serine were decreased in the LCN2^−/−-STZ group, whereas the levels of (2R)-2-hydroxy-3-(phosphonatooxy)propanoate and citric acid were increased. Furthermore, we identified 16 metabolites that exhibited significant statistical differences among the WT-SC, WT-STZ, and LCN2^−/−-STZ groups (as shown in Supplementary Fig. S3A). Concurrently, the absence of LCN2 normalized the levels of metabolites such as pyrrole-2-carboxylic acid, hydroxyphenyllactic acid, biotin, cysteinylglycine, and l-methionine S-oxide to the levels observed in the control group. These may represent potential downstream targets regulated by LCN2 and could be potential targets for future interventions. The outcomes hinted that LCN2 made a difference in glucose metabolism, lipid metabolism, and amino acid metabolism in DC. 

Figure 5.

 

Metabolic alterations in lenses of STZ-induced WT and LCN2^−/− mice. In each image of Fig. 5, A and B represent the LCN2^−/−-STZ group (n = 8) and WT-STZ group (n = 9) of the lenses. (A) Volcano plot of 54 differential metabolites. Compared to the WT-STZ group, metabolites that were significantly upregulated in the LCN2^−/−-STZ group are marked in red, with VIP > 1, P < 0.05, and FC > 1.5. Metabolites that were remarkedly downregulated in the LCN2^−/−-STZ group are marked in blue. Metabolites with VIP > 1, P < 0.05, and FC < 1.5 are indicated in yellow. Non-significant metabolites are shown in gray. The size of the dots corresponds to the magnitude of the VIP values. (B) Heatmap of 54 differential metabolites. Red represents significantly upregulated metabolites in the LCN2^−/−-STZ group compared to the WT-STZ group, and blue represents downregulated metabolites. (C) Chord plots of the top 10 KEGG enrichment terms with P < 0.05. (D) The top 20 differential abundance score with P < 0.05 based on KEGG enrichment analysis. (E) The relative expression levels of differential metabolites in the top three KEGG pathways between the two groups. *P < 0.05, **P < 0.01, ****P < 0.0001.

Metabolic Changes in Mice Retinas by STZ Induction

In the retinas of WT-SC and WT-STZ mice, 218 differential metabolites were detected, 145 upregulated and 73 downregulated, using the screening criteria of P < 0.05 and VIP > 1 (Figs. 6A, 6B). We selected the top 50 differential metabolites ranked by P value for classification and found that amino acids and derivatives, carbohydrates, and fatty acyls were the main categories of differential metabolites (Fig. 6C). KEGG pathway enrichment analysis indicated that the top five enriched pathways were ABC transporters, alanine, aspartate and glutamate metabolism, central carbon metabolism in cancer, glycine, serine and threonine metabolism, and vitamin B6 metabolism (Supplementary Fig. S2C). Meanwhile, the DA-score indicated that, in the retinas of WT-STZ mice, the majority of differential metabolites in most of the enriched pathways were elevated. Specifically, there was an increase in metabolites associated with pyruvate metabolism and the mTOR signaling pathway, as well as in ABC transporters, alanine, aspartate and glutamate metabolism, glycine, serine and threonine metabolism, glucagon signaling pathway, glycolysis/gluconeogenesis, and sphingolipid signaling pathway (Fig. 6D). The chord plot in Figure 6E and the circle plot in Figure 6F show the types of metabolites involved in the top 10 metabolic pathways with P < 0.05, as well as the quantities of these metabolites and their trends of upregulation or downregulation. Furthermore, the z-scores for the metabolic pathways illustrated in Figure 6F are detailed in Supplementary Table S4B. These results suggest that diabetes induced by STZ in mouse leads to disruptions of glucose metabolism, lipid metabolism, and amino acid metabolism in the retina. 

Figure 6.

 

Metabolic changes in mice retinas by STZ induction. In each image of Fig. 6, F and H represent the WT-STZ group (n = 8) and WT-SC group (n = 8) of the retinas. (A) Volcano plot of 218 differential metabolites. Compared to the WT-SC group, metabolites that were significantly upregulated in the WT-STZ group are marked in red, with VIP > 1, P < 0.05, and FC > 1.5. Metabolites that were remarkedly downregulated in the WT-STZ group are indicated in blue. Metabolites with VIP > 1, P < 0.05, and FC < 1.5 are indicated in yellow. Non-significant metabolites are represented in gray. The size of the dots corresponds to the magnitude of the VIP values. (B) Heatmap of 218 differential metabolites. Red represents significantly upregulated metabolites in the WT-STZ group compared to the WT-SC group, and blue represents downregulated metabolites. (C) Classification of 218 differential metabolites. (D) The top 20 DA-scores with P < 0.05 based on KEGG enrichment analysis. (E) Chord plots of the top 10 KEGG enrichment terms with P < 0.05. (F) Circle plots of the top 10 KEGG enrichment terms with P < 0.05. Red represents metabolites with elevated expression, and blue represents metabolites with decreased expression.

Metabolic Alterations in Retinas of STZ-Induced WT and LCN2^−/− Mice

A total of 35 differential metabolites were detected in the retinas of LCN2^−/−-STZ and WT-STZ mice, of which 13 were up-regulated and 22 were down-regulated in the E group compared with the F group (Figs. 7A, 7B). Supplementary Figure S2D shows that the top three KEGG pathways were central carbon metabolism in cancer, protein digestion and absorption, and mineral absorption. Upon LCN2 deletion, a decrease in the majority of metabolites associated with these pathways was observed, with notable reductions in phenylalanine, tyrosine, and tryptophan biosynthesis; breast cancer; lysosomes; and the mTOR signaling pathway (Fig. 7D). Specifically, Figure 7C pinpoints estradiol as the differential metabolite in the breast cancer pathway, d-mannose in the lysosome pathway, l-leucine in the mTOR signaling pathway, and l-tryptophan, indole and indoleglycerol phosphate in the phenylalanine, tyrosine, and tryptophan biosynthesis pathway. Relative to the WT-STZ group, the retinal expression of these metabolites was diminished in the LCN2^−/−-STZ group. Metabolites involving multiple enriched metabolic pathways (top three with P < 0.05) are shown in Figure 7E. Except for d-glucose 1-phosphate, which was elevated in the LCN2^−/− STZ group, the levels of all other metabolites were decreased relative to the WT-STZ group. Supplementary Figure S3B shows 17 metabolites that showed significant differences among the WT-SC, WT-STZ, and LCN2^−/−-STZ groups. Additionally, the absence of LCN2 led to the restoration of metabolites such as l-proline, d-mannose, estradiol, miglitol, succinic acid, 5-amino-2-oxopentanoic acid, and l-tryptophan to the levels of the normal control group. These data indicate that glucose metabolism, lipid metabolism, and amino acid metabolism were disrupted in the retina due to STZ induction, which was mitigated by knockout of LCN2. 

Figure 7.

 

Metabolic alterations in retinas of STZ-induced WT and LCN2^−/− mice. In each image of Fig. 7, E and F represent the LCN2^−/−-STZ group (n = 8) and WT-STZ group (n = 9) of the retinas. (A) Volcano plot of 35 differential metabolites. Compared with the WT-STZ group, metabolites that were significantly upregulated in the LCN2^−/−-STZ group are marked in red, with VIP > 1, P < 0.05, and FC > 1.5. Metabolites that were remarkedly downregulated in the LCN2^−/−-STZ group are marked in blue. Metabolites with VIP > 1, P < 0.05, and FC < 1.5 are indicated in yellow. Non-significant metabolites are represented in gray. The size of the dots corresponds to the magnitude of the VIP values. (B) Heatmap of 35 differential metabolites. Red represents significantly upregulated metabolites in the LCN2^−/−-STZ group compared to the WT-STZ group, and blue represents downregulated metabolites. (C) Chord plots of the top 10 KEGG enrichment terms with P < 0.05. (D) The top 20 differential abundance score with P < 0.05 based on KEGG enrichment analysis. (E) Metabolites involving multiple enriched metabolic pathways. *P < 0.05, **P < 0.01, ***P < 0.001.

Lens and Retinas Change Metabolites in STZ-Induced WT Mice

Identifying common biomarkers for DC and DR was key to facilitating early screening and diagnosis. In the lens WT-STZ and WT-SC groups, a total of 136 differential metabolites were detected; similarly, in the retina WT-STZ and WT-SC, 218 differential metabolites were identified according to the screening criteria as described previously. To pinpoint metabolites with more pronounced differences, we sorted them based on the log₂(fold change) and focused on the top 50, which led to the discovery of eight differential metabolites that were consistently altered in both the lens and retina as a result of STZ induction. Four of these were upregulated in both the lens and retina in the model group: L-2-hydroxyglutaric acid, linoleic acid, 2-hydroxyglutarate, and phosphoenolpyruvic acid. In contrast, 8-amino-7-oxononanoate was downregulated in the lens and retina. The expression of the other three metabolites, hydroquinone, β-d-fructose 6-phosphate, and Inosinic acid (IMP), were inconsistent in the lens and retina of the model group. (Figs. 8A, 8B). 

Figure 8.

 

Lens and retinas change metabolites in STZ-induced WT mice. In each image of Fig. 8, B and D represent the WT-STZ (n = 9) group and WT-SC group (n = 8) of the lenses, and F and H represent the WT-STZ group (n = 8) and WT-SC group (n = 8) of the retinas. (A) Overlap of differential metabolites in different groups. (B) Relative expression of differential metabolites in different groups. *P < 0.05, **P < 0.01, ***P < 0.001,****P < 0.0001, #P < 0.05, ##P < 0.01, ###P < 0.001, ####P < 0.0001.

Lens and Retinas Change Metabolites in STZ-Induced WT Mice Compared With LCN2^−/− Mice

LCN2^−/−-STZ and WT-STZ groups collectively exhibited 35 differential metabolites, and the LCN2^−/−-STZ and WT-STZ groups in the retinas revealed 54 differential metabolites. A total of five common differential metabolites were recognized in the lens and retina of both the WT-STZ and LCN2^−/−-STZ groups. Compared with the WT-STZ group, three of these increased in the lens and retina of the LCN2^−/−-STZ, including 5-guanidino-3-methyl-2-oxopentanoate, 12,13-EpOME, and arachidic acid, whereas d-mannose was decreased in both lens and retina. Additionally, the expression of saccharopine varied between the lens and retina (Figs. 9A, 9B). 

Figure 9.

 

Lens and retinas change metabolites in STZ-induced WT mice compared with LCN2^−/− mice. In each image of Fig. 9, A and B represent the LCN2^−/−-STZ group (n = 8) and WT-STZ group (n = 9) of the lenses, and E and F represent the LCN2^−/−-STZ group (n = 8) and WT-STZ group (n = 9) of the retinas. (A) Overlap of differential metabolites in different groups. (B) Relative expression of differential metabolites in different groups. *P < 0.05, **P < 0.01, ##P < 0.01, ###P < 0.001.

Discussion

DC and DR are common ocular complications of diabetes; the incidence of cataracts in diabetic patients is two to five times higher than that in non-diabetic patients.³ At the same time, DR is the main causes of blindness of the working-age population. DR still occurs in diabetic patients despite tight glycemic control.³⁰ The persistent adverse effects of hyperglycemia and the progression of its complications are referred to as “metabolic memory,” with oxidative stress, advanced glycation endproducts, and epigenetic changes being associated with this process. Studies have indicated that this hyperglycemic memory could also impact DR.³¹ 

Before assessing metabolic alterations, we initially confirmed the presence of LCN2 within the retina. Through a process of colocalization with GFAP, a marker for Müller cells, we discovered that the expression of LCN2 in the retina is predominantly found in Müller cells within the ganglion cell layer, inner plexiform layer, and outer plexiform layer (Supplementary Fig. S4), aligning with established research.³² Additionally, LCN2 expression in the retinas of diabetic mice was notably increased, but this increase was significantly reduced in LCN2^−/− mice. Early in diabetic retinopathy, retinal blood vessels tend to leak. Our results for the Evans blue tests showed clear leakage in the retinal vessels of STZ-treated mice, which was lessened with LCN2 gene knockout. This indicates that knocking out LCN2 may provide some relief and protection against the onset and early development of diabetic retinopathy in mice. 

Previous studies have been conducted on targeted metabolomics of serum and aqueous humor in patients with and without diabetes and metabolomics of serum in patients with diabetic retinopathy.^33–36 Studies have shown that LCN2 is involved in the regulation of inflammation, oxidative stress, glucose metabolism, autophagy, and other complex pathological processes,^26,37–39 but whether LCN2 plays a role in regulating the process of DC and DR is still unknown. In the hope of identifying metabolic changes during the development of DC and DR, this study directly detected the changes of metabolites in the lenses and retinas of mice in four groups after the onset of DC and DR. Our study is the first to perform untargeted metabolomics on the lenses and retinas of diabetic mice at the same time to identify targets that can be used to assess the risk of DC and DR, and the findings may be useful in preventing the progression of DC and DR. More importantly, we used LCN2 knockout mice and compared the changes of metabolites in the lenses and retinas of mice after LCN2 knockout to identify the regulatory role of LCN2 in the progression of DC and DR and provide new directions for investigating the prediction and prevention of DC and DR. 

Metabolites identified in the mice lenses and retinas were mainly amino acids and fatty acids, and, among others, the metabolic changes of diabetic mice were mainly concentrated in carboxylic acid and its derivatives, fatty acyl groups. Many studies have found that amino acid metabolism is closely related to diabetes, with amino acids being identified as markers for prediabetes, insulin resistance, and the onset of diabetes; at the same time, amino acid metabolism may show significant differences in the lenses of cataract patients compared to normal individuals.^40,41 A cross-sectional study of diabetic microvascular complications showed that circulating tyrosine levels are an important factor in the detection of DR.⁴² In recent decades, elevated plasma homocysteine levels and low folate status have been observed in many patients with retinal vascular disease, such as retinal vascular occlusion, diabetic retinopathy, and age-related degeneration. Homocysteine-induced toxicity to the endothelium may be involved in the development of retinal vascular disease.⁴³ Specific components of protein branched-chain amino acids, including leucine, l-isoleucine, and valine, are closely associated with insulin resistance and the development of diabetes.⁴⁴ 

Among the differential metabolites detected in the lens and retina of the WT-STZ group compared to the WT-SC group, the differential metabolites shared by both include 2-hydroxyglutarate, β-d-fructose 6-phosphate, 8-amino-7-oxononanoate, linoleic acid, phosphoenolpyruvic acid, hydroquinone, L-2-hydroxyglutaric acid, and IMP. Studies have shown that linoleic acid, an agonist of GPR40 and GPR120, can promote GLP-1 secretion, slow gastric emptying, and improve postprandial hyperglycemia. The mechanism may be related to the GPR120 pathway.⁴⁵ Linoleic acid metabolism was upregulated in high-fat-diet-fed mice.⁴⁶ Clinical interventions consistently showed that dietary supplementation with linoleic acid improved body composition, dyslipidemia, and insulin sensitivity while reducing systemic inflammation and fatty liver disease.⁴⁷ Beta-d-fructose 6-phosphate, a potent activator of d-fructofuranose 6-phosphate-1-kinase and an inhibitor of β-d-fructofuranose 1,6-bisphosphate, plays an important role in the regulation of glucose homeostasis.⁴⁸ Phosphoenolpyruvic acid (PCK-1), a key rate-limiting enzyme of gluconeogenesis, has been shown to improve glucose homeostasis by downregulation of PCK-1.⁴⁹ Hydroquinone O-β-d-glucopyranoside is a glycosylated hydroquinone that has been shown to have antioxidant and antihyperglycemic effects. Arbutin effectively ameliorated impaired glucose tolerance, insulin resistance, dyslipidemia, inflammation, and oxidative stress in diabetic rats, and it can regulate the activity of carbohydrate metabolic enzymes, antioxidants, and fat factors. It is also involved in the regulation of the peroxisome proliferator-activated receptor gamma (PPAR-γ) signaling pathway, which promotes adipocyte differentiation and glucose uptake.^50,51 tert-Butylhydroquinone (TBHQ) alleviates oxidative stress during diabetic retinopathy by upregulating the phosphoinositide 3-kinase (PI3k)/Akt eNOS pathway and partially restoring retinal structure and function, which may play a role in delaying diabetic retinopathy vision loss.⁵² TBHQ inhibits retinal microvascular injury by regulating oxidative stress, inflammation, cell proliferation and death regulation, and vascular system development.⁵³ 

After LCN2 gene knockout, carboxylic acids and their derivatives are also the main metabolites altered in the lenses and retinas of diabetic mice. Metabolites that changed in both the lenses and retinas of the LCN2^−/−-STZ groups included saccharopine, arachidic acid, d-mannose, 5-guanidino-3-methyl-2-oxopentanoate, and 12,13-EpOME. Lys is an intermediate metabolite of lysine that has antiglycation and antioxidation effects and can significantly inhibit the progression of diabetic cataracts in rats.⁵⁴ Mannose can be used in combination with hypoglycemic therapy in the treatment of diabetic cataracts by retaining levels of 31P membrane metabolites in the diabetic lens.⁵⁵ In the LCN2^−/−-STZ group, there was an observed increase in d-mannose levels within the lenses and retinas of mice, indicating a potential protective role of LCN2 knockout. Nitric oxide synthase is an important target for treating diseases such as diabetes, septic shock, and various neurodegeneration diseases; 5-(2-methylisothiourea)-2-amino-3-methylvaleric acid, a closely related synthetic analog of MeArg, strongly inhibits mammalian nitric oxide synthase.⁵⁶ The physiological role of the linoleic acid (LA)-derived epoxide 12(13)-epome is poorly understood,⁵⁷ and it has recently been suggested that diols originating in these species have endocrine functions.⁵⁸ Studies have shown that LA-derived epoxides are lower in patients with type 2 DM.⁵⁹ 

KEGG analysis of the detected metabolites revealed that the metabolites in mice altered by diabetes were mainly enriched in amino acid biosynthesis and metabolism, especially the glycine metabolic pathway, and some of them were also enriched in the related pathways of glucose metabolism. Glycine is a non-essential amino acid with many functions and effects.⁶⁰ A mixture of lysine and other amino acids has been reported to have anti-cataract effects in animal models. Studies have shown that glycine treatment leads to weight gain, decreased blood glucose, and increased plasma insulin levels. Glutathione (GSH) levels were increased and aldose reductase mRNA and protein levels were decreased in glycine-treated rats compared with their respective controls.⁶¹ Studies in human lymphatic endothelial cells (LECs) have shown that glycine can effectively regulate cellular iron homeostasis by synergistically influencing lysosome-dependent ferritin degradation and poly(rC)-binding protein 2 (PCBP2)-mediated transport of ferrous ions to delay the development of cataracts.⁶² In WT diabetic mice, the activation of the glycine degradation pathway in lenses may be involved in the formation of cataracts in mice. 

Within lenses, the activation of the glucagon signaling pathway and TCA cycle pathway in the WT-STZ group, as compared to the control group, confirms the establishment of a diabetic model. Furthermore, we noticed an intriguing phenomenon where the metabolic products of the cholesterol in the WT-STZ group of mice all decreased. The literature suggests that the lens grows throughout life, with epithelial cells differentiating into fiber cells, a process that necessitates a continuous supply of cholesterol to maintain the formation of fiber cell membranes.⁶³ Proliferative lens epithelial cells also require cholesterol to form membranes, which is necessary to complete the cell cycle.⁶⁴ Based on this, we speculate that due to the high demand for cholesterol in the lens, the compensatory downregulation of cholesterol metabolism in the lenses of diabetic mice is required to maintain the need for cholesterol. Following the knockout of the LCN2 gene, the metabolic products of the cholesterol metabolic pathway in the lenses of mice further decreased, which also suggests the protective effect of LCN2 knockout on the lenses of diabetic mice. 

In retinas, all metabolites in the mTOR pathway were elevated in the WT-STZ group of mice compared to the WT-SC group, indicating that this pathway is activated in WT diabetic mice. This activation was reversed after LCN2 gene knockout. The mTOR pathway plays a key role in regulating cell growth, as well as lipid and glucose metabolism.⁶⁵ Studies have indicated that the mTORC1 signaling pathway plays a significant role in glucose homeostasis by regulating the function of pancreatic cells. Initially, high activity of mTORC1 may be beneficial for glucose tolerance, but over time it can lead to a more rapid decline in cellular function.⁶⁶ LCN2 ablation might serve to protect retinal cell functions through suppression of the mTOR signaling pathway, yet it remains to be determined whether this suppressive effect is specifically targeting mTORC1. 

Notably, in both the lens and retina, the metabolism of tryptophan and its related metabolic pathways is downregulated after the knockout of the LCN2. Studies have shown that tryptophan and its metabolites are associated with an increased risk of diabetes.⁶⁷ 

At present, although there are established surgical and other therapeutic methods for addressing DC and DR, the visual loss of some patients is still irreparable. Therefore, it is of great significance to explore the pathogenesis of DC and DR and find the early predictive markers of DC and DR for preserving the visual acuity of diabetic patients. 

The limitations of this study are that the DCs and DR formed in mice maintained in a diabetic state for 3 months were at an early stage, which is not sufficiently convincing for studying the changes in metabolites during the progression of DCs and DR. Most of the metabolites identified so far are related to the pathogenesis of diabetes, and literature identifying metabolites that are directly associated with DCs and DR is limited. 

Conclusions

A large number of metabolites have been discovered to be altered in the lenses and retinas of WT diabetic mice and LCN2^−/− diabetic mice, some of which are involved in the pathogenesis of DC and DR. The pathways involved in the differential metabolites include glycine metabolism, cholesterol metabolism, mTOR pathway, tryptophan metabolism, and some pathways related to glucose metabolism that may have effects on DC and DR. Therefore, LCN2 is expected to be a common target for DC and DR and is meaningful to be studied in the future. 

Acknowledgments

The authors thank Professor Junfeng Mao and Professor Siqi Xiong from the Eye Center of Xiangya Hospital, Central South University, for their significant contributions in collecting vitreous humor samples. 

Supported by grants from the National Natural Science Foundation of China (81974130) and Natural Science Foundation of Hunan Province (2020JJ4882). 

Disclosure: Y. Yang, None; C. Fan, None; Y. Zhang, None; T. Kang, None; J. Jiang, None