We performed a prospective nonrandomized study at the Department of Ophthalmology, Shanghai General Hospital, Shanghai Jiao Tong University School of Medicine from January to December 2021. DR patients (12 patients) included in this study were diagnosed with diabetic retinopathy requiring surgery, with or without vitreous hemorrhage. Among the 7 control patients, five had macular epiretinal membranes and two had a macular hole, and they need surgery at the same time without hypertension and diabetes mellitus (DM).

At our clinic, patients who were clinically assessed as eligible were asked if they were interested in participating in the study. Before applying, they must decide on whether they would like to participate or not, and if so, they should provide written informed consent.

The minimum age of participants was 18 years old. Inclusion criteria included the diagnosis of diabetic retinopathy and either diabetes mellitus type 1 or type 2. The control group included patients diagnosed with macular epiretinal membranes or macular holes who needed pars plana vitrectomy.

Exclusion criteria included a history of subtotal or complete vitrectomy (3 port pars plana vitrectomy), treatment with an antiangiogenic agent, partial vitrectomy with drug administration, or laser coagulation within the 90 days before screening or treatment with a long-acting corticosteroid within the last 90 days before screening. In addition, patients who had undergone a cataract operation or posterior capsule opacification treatment within the past 90 days or had any other eye disease or clinically significant glaucoma evident at the time of screening were excluded. Patients with uncontrolled intraocular pressure (≥30 mmHg) were excluded.

Additional exclusion criteria included patients with uncontrolled hypertension (>160/90 mmHg (systole/diastole), poorly controlled diabetes (HbA1c >10%), or those diagnosed with an autoimmune disease. Patients were excluded if pregnant or breastfeeding during the study, if they had participated in a clinical trial within 30 days before screening, and if they had any other condition, that at the discretion of the investigator, was deemed to be inconsistent with participation in the trial. Finally, patients who reported drug or alcohol abuse within the 180 days before screening were excluded.

A blood sample (5–10 ml blood) was taken before the surgery to determine the SPARC level. A minimally invasive 3-port partial pars plana vitrectomy was performed in all cases by one vitreoretinal surgeon in the operating room. The extracted vitreous sample was placed directly in the freezer at a temperature of −70°C. Once completed, the removal of the probe leaves a self-sealing wound, which reduces the risk of leakage from the eye and limits the penetration of pathogens from the outer ocular surface.

Serum and vitreous samples were sent for laboratory analysis following surgery. The laboratory received no information about whether each sample was part of the DR or control group. This was to allow for laboratory analysis to be conducted in a blinded fashion. The SPARC level was detected using an enzyme-linked immunosorbent assay (ELISA) (DY941-05, R&D, Minneapolis, USA).

The sample of human proliferative membranes and epiretinal membranes were fixed in 2% PFA for 2 h and then frozen in isopentane (−55°C). Next, the eyes were embedded in paraffin and cut into 5-μm-thick sections, which were then mounted on glass slides. Paraffin-embedded sections were dewaxed with xylene, washed by gradient alcohol or distilled water, and stained with hematoxylin for 1–3 min. Sections were then differentiated by 1% hydrochloric alcohol, turned blue by PBS, stained by eosin, and dehydrated using gradient alcohol. After permeabilization with xylene, the tissues were photographed under microscopy (BX53, Olympus, Japan).

In addition, the tissues were processed for immunohistochemistry (IHC) analysis using an anti-SPARC antibody (ab225716, Abcam, Cambridge, the UK). Before the IHC procedure, sections were deparaffinized and rehydrated by immersing the slides in xylene and alcohol at descending concentrations. Colocalization of individual SPARC, TRIB3, and BRN3A, or GFAP and vimentin proteins in the retinal sections were detected using fluorescent confocal microscopy. Fluorescence intensity was measured using ImageJ software. Absorption control was performed using the recombinant human SPARC protein.

The RNA-sequencing data of DR patients and corresponding clinical information were obtained from the GEO dataset (GSE102485) on December 31, 2021. mRNA expression data of 11 DR patients and 3 healthy people were collected.

The differential expression of mRNA in DR tissue and normal control samples was evaluated using the R software (Version 3.8; http://www.bioconductor.org/packages/release/bioc/html). The heatmap and volcano of these genes were plotted using the R software. Pearson correlation analyses were performed to identify gene-to-gene correlation. The p-value of 0.05 was considered the significant threshold in all tests.

Gene ontology (GO), including the biological process (BP), cellular component (CC), and molecular function (MF) categories, was conducted with the “ggplot2” package in the R software. Similarly, this package was also utilized to perform the Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis.

The STRING website (https://string-db.org/) uses the protein–protein interaction network to study the interaction between protein names (SPARC) and organisms (Homo sapiens). The most important parameters are defined as follows: the minimum required interaction rating [low confidence (0,150)], the effectiveness of the network edge, the maximum number of interactions to be displayed (no more than 50 first shells), and the active interaction sources.

To further the function of SPARC, we used GeneMANIA tools (http://genemania.org/) to understand the coexpression, colocation, genetic interaction, pathway, physical interaction, and predicted and shared protein domains of SPARC.

Human retinal capillary endothelial cells (HRCECs; PriCells Biotechnological Co. Ltd., Wuhan, China) were cultured in DMEM (Gibco, Grand Island, NY, USA), 10% fetal bovine serum (FBS, Gibco), 100 IU/ml penicillin, and 100 μg/ml streptomycin (Gibco). All cells were incubated in a humidified atmosphere at 37°C under 5% CO₂ air. Subculturing was performed every 2 days, and cells were seeded into 96-well plates with the density of 4 × 105 cells/ml, cultured for 1 h, and detached with 0.25% trypsin. d-Glucose was added to the medium at a final concentration of 30 mmol/L to generate high-glucose (HG) conditions, while cells treated with DMEM (low glucose) with 5 mmol/L served as the low-glucose group. In addition, cells cultured in basal DMEM supplemented with 25 mmol/L glucose served as the normal medium conditions (control group).

Overexpression plasmids for SPARC and ITGB1 as well as the silencing plasmids sh-NC, sh-SPARC, and sh-ITGB1 were synthesized by Shanghai GenePharma Biological Co. Ltd. (Shanghai, China). CDS sequences of SPARC and ITGB1 were obtained from NCBI, followed by polymerase chain reaction (PCR) amplification. Plasmids were established by cloning the corresponding sequences of SPARC to pcDNA3.1 (+) vector through restriction enzyme sites. Confluent cells were then transfected with 1 μg sh-SPARC plasmid according to the manufacturer’s protocol in the Lipofectamine 2000 kit (Invitrogen, Carlsbad, CA, USA). Meanwhile, cells received treatment with vector + dimethyl sulfoxide (DMSO), sh-SPARC + DMSO, sh-ITGB1 + DMSO, and sh-SPARC + sh-ITGB1, followed by treatment with DMSO and 0.5 or 1.0 μM transmethylase inhibitor 5-Aza (Sigma) solution.

HRCECs were cultured at a density of 4 × 105 cells per 6-cm well, and they were 80%–90% confluent at the time of transfection. Two micrograms of each plasmid was transfected with 4 μl lipofectamine 2000 reagent (Invitrogen, Karlsruhe, Germany) according to the manufacturer’s recommendations, and the empty vector was transfected into HRCECs as a control for comparison.

A Cell-Counting Kit-8 (CCK-8) assay was used to evaluate the cell viability. HRCECs were treated with sh-SPARC for 24 h. The cells were plated into 96-well plates for 48 h in the incubator. At the end of the incubation period, 10 μl CCK-8 solution was added to each well, and the plates were returned to the incubator for an additional 2 h at 37°C. Absorbance was measured at 450 nm using a microplate reader.

Tube formation assay was performed as we previously described (22). HRCECs were placed on the Matrigel and treated with sh-SPARC for 12 h. Tube formation was quantified by counting the number of connected cells in randomly selected fields and dividing by the total number of cells in the same field.

HRCECs were harvested, and total proteins were obtained using RIPA Lysis Buffer (Beyotime, Shanghai, China). The protein concentration was determined using the BCA Protein Quantitation kit (Beyotime, Shanghai, China). The proteins were then separated by SDS-PAGE and transferred to PVDF membranes. The membranes were blocked with 5% nonfat milk and 0.1% Tween-20 for 1 h at room temperature. Samples were then incubated with the primary antibodies overnight at 4°C, which were the primary antibodies against SPARC (66426-1-Ig, Proteintech), ITGB1 (ab199056 Abcam), vascular endothelial growth factor receptor (VEGFR) (ab134191, Abcam), integrin-linked kinase (ILK) (ab134179, Abcam), and fibronectin 1 (FN1) (ab52480, Abcam). After being washed three times with TBST buffer, the membranes were incubated with secondary antibodies (ab150077, Abcam) at 4°C for 3 h. Enhanced chemiluminescence (ECL) reaction reagents were used for visualization, and ImageJ software was employed to analyze the results. The protein bands were quantified and normalized to the expression of glyceraldehyde 3-phosphate dehydrogenase (GAPDH).

The total RNA was extracted using Trizol reagent (TaKaRa, Dalian, Liaoning, China), and the cDNA was synthesized with RNA Transcription Kit (TaKaRa). RT-qPCR was then performed using the SYBR Green RT-qPCR system. SPARC, ITGB1, VEGFR, ILK, and FN1 mRNA expression were tested in our study mRNA expression were tested in our study. The relative expressions of mRNAs were evaluated by the 2−ΔΔCq method, and the internal normalization control is GAPDH ( Supplementary Table S1 ).

HRCECs were seeded in 6-well plates, then collected by trypsinization, washed in 4°C PBS, and trypsinized into cell suspensions. Phosphatidyl-serine (PS) exposure due to the flipping of the plasma membrane, a concomitant feature during apoptosis, was evaluated by PE Annexin V Apoptosis Detection Kit (559763, BD Pharmingen) by flow cytometry analysis containing PE Annexin V and 7-AAD. The experiments were independently repeated three times.

Recent studies found that dysregulation of SPARC is associated with a variety of obesity-related diseases, including type 2 diabetes and its complications related to obesity, kidney and liver disease, cardiovascular disease, and cancer (23). Therefore, we investigated the SPARC expression in human serum and vitreous samples and histological results in proliferative membranes. The demographic data of the participants are shown in Table 1 . HE staining showed that the epiretinal membrane is mainly composed of extracellular matrix (ECM) such as collagen fiber, and the nucleus is round and spindle shaped in the control group. DR group proliferation membrane is mainly manifested as vascular fiber membrane, which is characterized by more luminal structures surrounded by endothelioid cells ( Figure 1A ). To this end, we analyzed and quantitated SPARC level and immunoreactivity. Overall, the serum and vitreum SPARC levels in the DR group were highly increased significantly than in the control group ( Figures 1B, C ). Representative control and DR of a 67-year-old and 56-year-old man are shown in Figure 1D . Strong evidence of DR immune responses in retinal endothelial cells, ganglia, and photoreceptor cells was observed in a 67-year-old nondiabetic man compared with the control ( Figure 1E ).

Further analysis showed the differential gene expression of DR patients in the GEO dataset. The heatmap and volcano map are shown in Figures 2A , B . SPARC and ITGB1 expressions are significant differences between DR patients and healthy patients. GO disease analysis also found that SPARC was correlated with electroretinogram abnormality, photophobia, rod-cone dystrophy, and disorder of the eye ( Supplementary Figure S1B ). GO target gene analysis investigated ZNF513, AUTS2, ZBTB44, and SUPT16H ( Supplementary Figure S1C ), and GO-regulated gene analysis also found that NRL, ETV7, PAX6, and CRX can regulate the expression of SPARC ( Supplementary Figure S1C ). GO enrichment and GO network found that SPARC plays an important role in retinal development in the camera-type eye ( Figure 2C ; Supplementary Figure S1A ). The GeneMANIA network showed that SPARC significantly correlated with STAB1, FN1, VEGF, and ILK ( Figure 2C ). Protein–protein network also investigated the correlation of SPARC with ERBB4, ALB, COL1A1, and MMP2 ( Figures 2D, E ).

To verify these results, we test the mRNA of SPARC, ITGB1, ILK, and FN1 levels in DR and control tissues and found that the DR group has higher SPARC, ITGB1, ILK, and FN1 mRNA levels than the control group ( Figure 3A ). It is indicated that a high SPARC level is correlated with the development of DR, and the SPARC expression level also links to ITGB1 and ILK.

HG-induced retinal pigment epithelium can stimulate the development of DR. Therefore, we also analyzed the SPARC level in a different dose of HG-induced cell models. We found that the SPARC level can be increased significantly in HG medium (30 mmol/L) in than low glucose medium (5.5 mmol/L) after 7 days of incubation, and the effect of HG treatment on SPARC showed a dose-dependent manner. qPCR assay showed that the HG environment can also increase the SPARC, VEGFR, ITGB1, IL-6, TGF-β, and ILK mRNA levels than the low-glucose group ( Figure 3 ). In this part, we further demonstrated that the expression of SPARC and related genes and proteins in the HRCEC cells were significantly increased under HG condition.

In addition, we also analyzed the protein level in the HG-induced HRCEC cell model. The HG environment increased the expression of ITGB1, ILK, FN1, and VEGFR compared with the low-glucose-induced HRCEC cells ( Figure 4A, B ).

Subsequently, the study aimed to understand the role of SPARC and SPARC-related genes such as ITGB1 in the development of DR. We first transfected HRCECs with the sh-SPARC-interfered lentiviruses carrying red fluorescent protein. A representative figure of immunofluorescence has been shown in Figure 5A . The SPARC protein level has also been tested by Western blotting and was found to be significantly lower in the sh-SPARC group than in the control group, showing that the SPARC silence model is successful ( Figure 5 ).

After the establishment of the cell silencing model, we detected the changes in SPARC-related gene expression. Firstly, the SPARC mRNA level was significantly decreased in the sh-SPARC group than in the sh-NC group, and then, the ITGB1, ILK, VEGFR, FN1, IL6, and TGF-β mRNA levels also markedly decreased in the sh-SPARC than in the sh-NC group.

Consisting with the mRNA expression of SPARC-related genes, SPARC knockdown could decrease the protein level of ITGB1, ILK, FN1, and VEGFR significantly than the control group ( Figure 4A, B ).

ITGB1 overexpression and ablation demonstrate that the former can significantly increase the expression of VEGFR, FN1, IL6, and TGF-β in mRNA levels, whereas the latter could decrease the expression of VEGFR, FN1, IL6, and TGF-β in mRNA levels.

In addition, the ITGB1 overexpression or ablation could increase or significantly decrease the protein expression levels of ILK, FN1, and VEGFR. The sh-SPARC/ITGB1-OE group could also increase the ITGB1 protein level while the sh-SPARC/ITGB1-sh group decreased the ITGB1 protein level significantly. However, sh-SPARC/ITGB1-OE could only increase the protein level of FN1 and VEGFR compared with the sh-SPARC/ITGB1-sh group ( Figure 6 ).

The cell apoptosis assay was conducted, and HG-induced HRCECs had a high apoptosis ratio when compared with the control group ( Figure 7A ). Firstly, HG-induced HRCECs can increase apoptosis levels in the control group more than normal glucose, and sh-SPACR can significantly decrease the apoptosis ratio in HG levels more than normal glucose ( Figure 7B ). In normal and HG ( Figure 7D ), the sh-ITGB1 group increased the apoptosis ratio significantly more than the sh-NC. In addition, sh-SPARC-ITGB1 also increased apoptosis significantly more than in the sh-SPARC-NC in HG ( Figure 7E ).

We analyze the cell viability by CCK-8 tests in Figure 7A . After 1 or 2 days postincubation ( Figures 8A, B ), sh-SPARC significantly decreased cell viability more than the sh-NC group ( Figures 8C–E ). Apoptosis analysis showed that sh-SPARC cells exhibit higher apoptosis levels than the sh-NC group in Figure 7C . Furthermore, ablation of ITGB1 decreased cell viability significantly more than the NC group, and overexpression of ITGB1 increased cell viability more than in the NC group ( Figures 9A –C ).

In Figure 9A , the tube formation test has been investigated among the NC, sh-SPARC, sh-ITGB1, and sh-SPARC-ITGB1 groups. The corresponding image was shown, and the results revealed that the sh-SPARC, sh-ITGB1, and sh-SPARC-ITGB1 groups could reduce tube formation rate and tube area covered than the NC group ( Figures 9B, C ).