Rat retinal Müller cells (CP-R117, Procell, Wuhan, China) were cultured in specific DMEM (CM-R117, Procell, Wuhan, China). For high glucose (HG) treatment, the Müller cells were cultured (24 h) in a complete culture medium with 25 mM glucose (G8150, Solarbio, Beijing, China) [24]. After HG/normal glucose (NG) induction, the rat retinal Müller cells were collected and used for transcriptome sequencing. All cell lines were validated by Glutamine Synthetase (GS) immunofluorescence and tested negative for mycoplasma. Cells were all cultured in a humidified incubator (37 °C, 5% CO₂).

Transcriptome sequencing was performed by the Novogene Company (Novogene, Beijing, China). Sequencing libraries were built using NEBNext® Ultra™ Directional RNA Library Prep Kit (NEB, Ipswich, MA, USA). Briefly, the mRNA was fragmented and purified with poly-T oligo-attached magnetic beads (Beckman Coulter, Brea, CA, USA). Then, first-strand and second-strand cDNAs were synthesized. Polymerase chain reaction (PCR) was performed using universal PCR primers and Index (x) Primer. Finally, the Library preparations were sequenced on an Illumina Hiseq 2000 platform (https://support.illumina.com/sequencing/sequencing_instruments/hiseq_2000/documentation.html). Differential expression analysis, Gene Ontology (GO) enrichment analysis (https://geneontology.org/) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis (https://www.kegg.jp/) were conducted.

Short hairpin RNA (shRNA) for GAS1 (shGAS1; target sequence: 5^′-ACGACTACTACGACGAAGAATAT-3^′)/negative control (shNC; target sequence: 5^′-AATTCTCCGAACGTGTCACGT-3^′), small interference RNA (siRNA) for Nrf2 (siNrf2; si1 target sequence: 5^′-CAGCATGATGGACTTGGAATTGC-3^′; si2 target sequence: 5^′-GGGGTAAGTCGAGAAGTGTTTGA-3^′; si3 target sequence: 5^′-GGCCTTTTTCGCTCAGTTACAAC-3^′), and siNC (target sequence: 5^′-TTCTCCGAACGTGTCACGT-3^′) were obtained from VectorBuilder (Guangzhou, China). Overexpression plasmids for GAS1 and NC plasmids were synthesized by Sino Biological, Inc. Supplementary Materia exhibits coding sequence (CDS) of GAS1. The shRNA, siRNA, or plasmids were transfected (48 h) into Müller cells with 80% density in 6-well plates with transfection reagent (L3000008, Bluefbio, Shanghai, China), followed by cell collection.

Total RNA from Müller cells (a RNA isolation kit; R0017M, Beyotime, Shanghai, China) was detected using the UV spectrophotometer (NanoDrop, Thermo, Waltham, MA, USA) and synthesized into cDNA using First-strand cDNA synthesis kit (D7168M, Beyotime, Shanghai, China), followed by mixing with SYBR green (ALH185, Baiaolaibo, Beijing, China) and primers of a relative gene. The gene expressions were examined by a QuantStudio 6 System (Applied Biosystems, Waltham, MA, USA). Primers’ information was as follows: GAS1 forward (F): 5^′-GACCCCGTTTGCCTGTTTTC-3^′, reverse (R): 5^′-CAGTGCTCCCGATCATCTCC-3^′; Nrf2 F: 5^′-TCACACGAGATGAGCTTAGGGCAA-3^′, R: 5^′-TACAGTTCTGGGCGGCGACTTTAT-3^′; $\beta$-actin F: 5^′-CATGTACGTTGCTATCCAGGC-3^′, R: 5^′-CTCCTTAATGTCACGCACGAT-3^′. The relative mRNA expression levels were calculated by 2^{-Δ⁢Δ⁢Ct} method [25].

Transfected/HG-triggered Müller cells (5 $\times$ 10³) were seeded in 96-well plates (24 h), and then reacted with Cell Counting Kit-8 (CCK-8) reagent (10 µL; 96992, Merk, St. Louis, MO, USA) (2.5 h). Cell viability was gauged through a microplate reader (450 nm; Varioskan LUX, Thermo, Waltham, MA, USA). Cell viability (%) = (optical density (OD)_Experimental – OD_blank)/(OD_Control – OD_blank) $\times$ 100.

The Annexin V-FITC/propidium iodide (PI) kit (40302ES60, YEASEN, Shanghai, China) was used in this assay. Briefly, transfected/HG-triggered Müller cells were suspended in 100 µL binding buffer, and then incubated with 5 µL Annexin V-FITC and 10 µL PI staining buffer (40302ES60, YEASEN, Shanghai, China) (20 min). Following cell incubation with 400 µL binding buffer, cell apoptosis signal was examined using a flow cytometer (Attune NxT, Thermo, Waltham, MA, USA).

Post culture of transfected/HG-exposed Müller cells with BODIPY 581/591 C11 (S0043S, Beyotime, Shanghai, China) (20 min, 37 °C) and phosphate buffered saline (PBS) (C0221A, Beyotime, Shanghai, China) washing, lipid ROS levels were assessed using fluorescence microplate reader (GloMax® Discover Microplate Reader, Promega, Madison, WI, USA) (excitation: 581 nm; emission: 591 nm).

GSH and GSSG levels in transfected/HG-exposed Müller cells were evaluated using a total glutathione/oxidized glutathione assay kit (A061-1-2, Nanjing Jiancheng Bioengineering Institute, Nanjing, China). Concretely, the collected Müller cells were ground to pieces, and 10 µL cell samples were mixed with 100 µL reagent 1 and 10 µL reagent 2 (2 min), as well as with 50 µL reagent 3. Then, the OD value was evaluated using a microplate reader (405 nm wavelength) for the calculation of total GSH. Meanwhile, 100 µL cell samples were mixed with 2 µL reagent 5 and 5 µL reagent 6, after which 10 µL mixed solution was further blended with 100 µL reagent 1 and 10 µL reagent 2 (2 min), followed by addition of 50 µL reagent 3. Next, the OD value was evaluated using a microplate reader (405 nm wavelength) for the calculation of GSSG level. Finally, the level of GSH was further calculated as the formula provided by the manufacturer’s manual.

MDA levels in transfected/HG-exposed Müller cells were evaluated employing an MDA assay kit (ab118970, Abcam, Cambridge, UK). In brief, the collected Müller cells were mixed with 300 µL MDA lysis buffer and 3 µL BHT stock/BHT, and then underwent centrifugation (13000 $\times$g, 10 min) to obtain the supernatant. Then, 200 µL supernatant and 600 µL developer VII/TBA reagent were cultivated (1 h, 95 °C). Finally, the OD value was detected using a microplate reader (532 nm) for the calculation of the MDA level as the formula provided by manufacturer’s instructions.

Fe²⁺ levels in transfected/HG-exposed Müller cells were evaluated using an Iron assay kit (ab83366, Abcam Cambridge, UK). Briefly, the homogenate of Müller cells (50 µL) experienced incubation with 5 µL iron reducer and 5 µL assay buffer (30 min, 37 °C), followed by culture with 100 µL iron probe (1 h, 37 °C). The OD value was read by a microplate reader (593 nm) for Fe²⁺ level calculation.

Quantification of GST, HO-1, and NQO1 in transfected/HG-exposed Müller cells was performed employing GST (ab65326, Abcam, Cambridge, UK), HO-1 (ab279414, Abcam, Cambridge, UK), and NQO1 (ab184867, Abcam, Cambridge, UK) assay kits as per the manual.

Total protein in transfected/HG-exposed Müller cells was extracted by a RIPA buffer (R0020, Solarbio, Beijing, China) containing a protease inhibitor cocktail (GCPI0015, Vokai Biotechnology, Beijing, China) and PMSF (D16793, Vokai Biotechnology, Beijing, China). The cell nuclear proteins were isolated using a kit (P0027, Beyotime, Shanghai, China) and underwent quantification using a BCA detection kit (KGP903, KeyGEN BioTECH, Nanjing, China). The proteins were incubated with loading buffer (D16796, Vokai Biotechnology, Beijing, China) (5 min, 100 °C), further separated by SDS-PAGE gel (GCPE0017, Vokai Biotechnology, Beijing, China), and transferred to PVDF membrane (YA1701, Solarbio, Beijing, China). The membrane was then incubated with blocking buffer (P0023B, Beyotime, Shanghai, China) overnight, primary antibody overnight (4 °C), and secondary antibody (2 h). By means of ECL luminescent liquid (GCPW0121, Vokai Biotechnology, Beijing, China), the protein bands were detected by a ChemiDoc MP-specific system (Bio-Rad, Hercules, California, USA). The primary and secondary (Beyotime, Shanghai, China) antibodies included Keap1 (60027-1-Ig, 1:4000, Proteintech, Wuhan, China), Nrf2 (20733, 1:1000, CST, Boston, MA, USA), LaminB (66095-1-Ig, 1:20000, Proteintech, Wuhan, China), GAS1 (PA5-101615, 1:1000, Invitrogen, Waltham, MA, USA), $\beta$-actin (A2-F6, 1:10000, Huabio, Hangzhou, China), goat anti-rabbit IgG (A0277, 1:3000, Beyotime, Shanghai, China), and goat anti-mouse IgG (A0286, 1:3000, Beyotime, Shanghai, China).

Data from two groups experienced dissection using independent-sample t-test, and those from multiple groups were analyzed using one-way ANOVA by Tukey’s post-hoc test using GraphPad 8.0 software (GraphPad Software, San Diego, CA, USA). Measurement data were expressed as mean $\pm$ SD. p 0.05 meant statistical significance.

HG treatment resulted in decreased viability of Müller cells (p 0.01, Fig. 1A) and increased apoptosis (p 0.001, Fig. 1B,C). Subsequently, the lipid peroxidation in Müller cells was evaluated. As shown in Fig. 1D–G, the lipid ROS, GSSG, and MDA in HG-treated Müller cells were elevated (p 0.01), and the GSH was reduced (p 0.001). Furthermore, the Fe²⁺ level in Müller cells was raised by HG treatment (p 0.001, Fig. 1H). These results revealed that HG treatment decreased the viability and increased apoptosis, lipid peroxidation, and Fe²⁺ level in Müller cells.

Fig. 1.

HG treatment decreased the viability and increased apoptosis, lipid peroxidation, and Fe²⁺levels in Müller cells. (A–H) The viability (CCK-8) assay (A), the apoptosis (flow cytometry) (B,C), and the levels of lipid ROS, GSH, GSSG, MDA, and Fe²⁺ (commercial reagent kit) (D–H) in HG-exposed (24 h) Müller cells. **p 0.01, ***p 0.001 vs. NG. CCK-8, Cell Counting Kit-8; NG, normal glucose; HG, high glucose; ROS, reactive oxygen species; GSH, glutathione; GSSG, glutathione disulfide; MDA, malondialdehyde. n = 3 in each group.

In HG-exposed Müller cells, transcriptome sequencing data unveiled many differentially expressed genes (DEGs) (Fig. 2A,B), with 34 downregulated genes and 339 upregulated genes in line with the volcano plot. GAS1 level was lower in NG-treated Müller cells than HG-treated Müller cells (Fig. 2B). GO analysis provided insights into the biological process (BP), molecular function (MF), and cell component (CC) of DEG’s enrichment (Fig. 3A). Meanwhile, KEGG’s enrichment results show that Glycosphingolipid Biosynthesis-globo and isoglobo Series, Motor proteins and selenocompound metabolism is the top three metabolic channels of the enrichment (Fig. 3B). Furthermore, GAS1 expression in HG-treated Müller cells was discovered to be upregulated (p 0.001, Fig. 3C).

Fig. 2.

GAS1 was differentially expressed in HG-treated Müller cells. (A,B) Heat map (A) and volcano map (B) of NG/HG-treated (24 h) DEGs in Müller cells (transcriptome sequencing), showing decreased GAS1 expression in the NG group relative to HG group. NG, normal glucose; HG, high glucose; GAS1, growth arrest-specific 1; DEGs, differentially expressed genes.

Fig. 3.

GAS1 was upregulated in HG-treated Müller cells. (A,B) BP, MF, and CC of enrichment between GO and KEGG of DEGs (A) and the pathways enrichment between GO and KEGG of DEGs (B) (GO and KEGG analysis). (C) GAS1 expression in HG-exposed (24 h) Müller cells (quantitative real-time PCR). ***p 0.001 vs. NG. NG, normal glucose; HG, high glucose; GAS1, growth arrest-specific 1; BP, biological process; MF, molecular function; CC, cellular componenta. n = 3 in each group.

To evaluate how GAS1 impacted HG-treated Müller cells, shGAS1 was transfected into Müller cells. Fig. 4A–C revealed increased GAS1 expression by HG treatment (p 0.001), while shGAS1 reversed the role of HG (p 0.001). Through CCK-8 and flow cytometry assays, it was discovered that HG treatment attenuated viability (p 0.001, Fig. 4D) and enhanced apoptosis (p 0.001, Fig. 4E,F) of Müller cells, but shGAS1 offset these effects of HG (p 0.001). Besides, detection of oxidative stress (Fig. 4G–I) proved HG induction augmented concentrations of GST, HO-1 and NQO1 in Müller cells (p 0.001), while GAS1 downregulation reduced GST, HO-1 and NQO1 levels following HG treatment (p 0.001). The above discovers revealed that GAS1 downregulation promoted the viability and inhibited the apoptosis and oxidative stress of HG-treated Müller cells.

Fig. 4.

GAS1 downregulation promoted the viability and inhibited the apoptosis and oxidative stress of HG-treated Müller cells. (A–I) GAS1 expression (western blotting and quantitative real-time PCR) (A–C), viability (CCK-8 assay) (D), apoptosis (flow cytometry) (E,F), and the concentrations of GST, HO-1, and NQO1 (commercial reagent kit) (G–I) in shGAS1-transfected and HG-treated Müller cells. ***p 0.001. NG, normal glucose; HG, high glucose; GAS1, growth arrest-specific 1; GST, glutathione-s-transferase; HO-1, heme oxygenase-1; NQO1, quinone oxidoreductase 1; shNC; negative control for shGAS1. n = 3 in each group.

Then, examination of lipid peroxidation (Fig. 5A–D) indicated the relative levels of lipid ROS, GSSG, and MDA were increased (p 0.01), while the GSH level was reduced (p 0.001) in Müller cells due to HG treatment; however, the role of HG was offset by shGAS1 (p 0.001). Meanwhile, the Fe²⁺ level in HG-treated Müller cells was increased (p 0.001, Fig. 5E), which was counteracted by shGAS1 (p 0.001, Fig. 5E).

Fig. 5.

GAS1 downregulation reduced the lipid peroxidation and Fe²⁺ levels of HG-treated Müller cells. (A–E) Levels of lipid ROS, GSH, GSSG, MDA, and Fe²⁺ in shGAS1-transfected and HG-treated Müller cells (commercial reagent kit). **p 0.01, ***p 0.001. NG, normal glucose; HG, high glucose; GAS1, growth arrest-specific 1; ROS, reactive oxygen species; GSH, glutathione; GSSG, glutathione disulfide; MDA, malondialdehyde; shNC; negative control for shGAS1. n = 3 in each group.

Mechanistically, Keap1/Nrf2 signaling transduction was evaluated (Fig. 6A–D), revealing upregulated Keap1 and total Nrf2 as well as downregulated nuclear Nrf2 in HG-treated Müller cells (p 0.001), which however were all reversed after shGAS1 transfection (p 0.001). These findings suggested that GAS1 downregulation promoted Keap1/Nrf2 signal in HG-treated Müller cells.

Fig. 6.

GAS1 downregulation promoted the Keap1/Nrf2 signal in HG-treated Müller cells. (A–D) Expressions of Keap1, total Nrf2, and nuclear Nrf2 in shGAS1-transfected and HG-treated Müller cells (Western blot). ***p 0.001. NG, normal glucose; HG, high glucose; GAS1, growth arrest-specific 1; Keap1, Kelch-like ECH-associated protein 1; Nrf2, nuclear factor erythroid 2-related factor; shNC, negative control for shGAS1. n = 3 in each group.

To confirm that GAS1 impacted HG-treated Müller cells through regulating the Keap1/Nrf2 signal, Nrf2 was silenced in Müller cells (p 0.05, Fig. 7A–C). Due to the highest transfection efficiency, si2 was chosen for later use. Meanwhile, GAS1 was also overexpressed in Müller cells (p 0.001, Fig. 7D–F). In HG-treated Müller cells, GAS1 overexpression weakened viability (p 0.001, Fig. 7G) and induced apoptosis (p 0.001, Fig. 7H,I), which were reversed by Nrf2 silencing (p 0.01, Fig. 7G–I). Moreover, detection of oxidative stress revealed that GAS1 overexpression raised the concentrations of GST, HO-1, and NQO1 in HG-treated Müller cells (p 0.05, Fig. 7J–L), which were further offset by Nrf2 silencing (p 0.05, Fig. 7J–L). Accordingly, it was conjectured that GAS1 may mediate Keap1/Nrf2 signal to impact viability, apoptosis, and oxidative stress of HG-treated Müller cells.

Fig. 7.

Nrf2 downregulation reversed the role of GAS1 in the viability, apoptosis, and oxidative stress of HG-treated Müller cells. (A–C) The transfection efficiency of siRNA for Nrf2 and (D–F) GAS1 overexpression plasmids (western blotting and quantitative real-time PCR). (G–L) Viability (CCK-8 assay) (G), apoptosis (flow cytometry) (H–I), and the concentrations of GST, HO-1, and NQO1 (commercial reagent kit) (J–L) in transfected and HG-exposed Müller cells. *p 0.05, **p 0.01, ***p 0.001. HG, high glucose; GAS1, growth arrest-specific 1; HO-1, heme oxygenase-1; NQO1, quinone oxidoreductase 1; Nrf2, nuclear factor erythroid 2-related factor; NC, negative control for GAS1 overexpression plasmid; siNC, negative control for siNrf2. n = 3 in each group.

Changes in lipid peroxidation and Fe²⁺ levels in HG-treated Müller cells were examined (Fig. 8A–D). GAS1 overexpression increased lipid ROS production, GSSG level and MDA level (p 0.01), and reduced GSH level in HG-treated Müller cells (p 0.01); however, Nrf2 silencing counteracted these impacts of GAS1 (p 0.01). Meanwhile, the Fe²⁺ level of HG-treated Müller cells was upregulated by GAS1 overexpression (p 0.001, Fig. 8E), which was then offset by Nrf2 downregulation (p 0.001, Fig. 8E). We thus inferred that GAS1 impacted lipid peroxidation and Fe²⁺ levels of HG-treated Müller cells by regulating Keap1/Nrf2 signal.

Fig. 8.

Nrf2 downregulation reversed the role of GAS1 in the lipid peroxidation and Fe²⁺levels of HG-treated Müller cells. (A–E) Levels of lipid ROS, GSH, GSSG, MDA, and Fe²⁺ in transfected and HG-exposed Müller cells (commercial reagent kit). **p 0.01, ***p 0.001. HG, high glucose; GAS1, growth arrest-specific 1; ROS, reactive oxygen species; GSH, glutathione; GSSG, glutathione disulfide; MDA, malondialdehyde; Nrf2, nuclear factor erythroid 2-related factor; NC, negative control for GAS1 overexpression plasmid; siNC, negative control for siNrf2. n = 3 in each group.

Finally, Keap1/Nrf2 signaling transduction was evaluated (Fig. 9A–D). The expressions of Keap1 and total Nrf2 were upregulated while nuclear Nrf2 expression was downregulation in HG-treated Müller cells after GAS1 overexpression (p 0.001); however, the effects of GAS1 overexpression were all neutralized by Nrf2 deficiency (p 0.001). These outcomes further confirmed that GAS1 can repress Keap1/Nrf2 signaling transduction.

Fig. 9.

Nrf2 downregulation reversed the role of GAS1 in Keap1/Nrf2 signaling transduction in HG-treated Müller cells. (A–D) Expressions of Keap1, total Nrf2, and nuclear Nrf2 in transfected and HG-exposed Müller cells (Western blot). ***p 0.001. HG, high glucose; GAS1, growth arrest-specific 1; Keap1, Kelch-like ECH-associated protein 1; Nrf2, nuclear factor erythroid 2-related factor; NC, negative control for GAS1 overexpression plasmid; siNC, negative control for siNrf2. n = 3 in each group.