Mouse model and cell lines

All animal-related experiments were approved by the Animal Care and Use Committee of Sichuan Provincial People’s Hospital and were performed in accordance with the Declaration of Helsinki. All experimental procedures were carried out strictly in accordance with the approved study protocols. The mice were raised under regular light conditions with a 12-h light/12-h dark cycle and unrestricted access to food and water. Euthanasia was performed by dislocating the cervical vertebrae of mice after anaesthesia with 1% sodium pentobarbital anaesthetic (at a dose of 10 μL/g body weight).

Pcyt1aloxP/+ (Pcyt1afl/+) transgenic mice on a C57BL/6 J background were generated by Cyagen Biosciences (Suzhou, China). By employing the CRISPR/Cas9 approach, we generated conditional Pcyt1a knockout mice by replacing exons 4–5 of Pcyt1a with a homologous region containing two loxP sites. This modification resulted in a frameshift in the PCYT1A transcript and caused the loss of function of the Pcyt1a gene. The primer pairs F1 (5′-CTAACTACAAGACGCTTTCCCCA-3′) and R1 (5′-GCTTCAGTCCCTTGGCTTTAGAA-3′) were used to identify the upstream loxP sequence.

To prevent off-target effects caused by CRISPR/Cas9, we crossed Pcyt1afl/+ mice with C57BL/6 J mice for six generations before mating them with Six3-Cre mice. Genotyping of the Six-Cre transgenic mice was carried out by PCR using Six-F (5′-GAACGCACTGATTTCGACCA-3′) and Six-R (5′-GCTAACCAGCGTTTTCGTTC-3′). The HEK293T and ARPE-19 cell lines were purchased from ATCC and cultured in DMEM (Gibco, USA) at 37 °C in a 5% CO2 incubator. For knockdown of the PCYT1A gene, ARPE-19 cells were transduced with a lentivirus carrying shRNA targeting PCYT1A (5′-CCCTTTCTGTCCCATTACCTT-3′, Genechem, Shanghai, China) or negative control shRNA (5′-TTCTCCGAACGTGTCACGT-3′) according to the manufacturer’s protocol. The use of lentiviruses in this study was approved by the Institutional Biosafety Committee of Sichuan Provincial People’s Hospital.

RNA isolation, reverse transcription–polymerase chain reaction (RT-PCR), and RT-qPCR

Mouse retinal tissue RNA was isolated using TransZol Up reagent, and reverse transcription was subsequently performed using EasyScript All-in-One First-Strand cDNA Synthesis SuperMix for qPCR. The primer pairs used for different genes are listed in Additional file 2: Table S2. Target gene expression was normalized to GAPDH mRNA expression.

Single-cell RNA-seqSample preparation

Experiments were conducted using 3-month-old wild-type C57BL/6 J mice. For isolation of viable single cells from retina/retinal pigment epithelium (RPE)/choroid tissue, the mice were euthanized, and the eyes were enucleated for further processing. Each eye was dissected under a stereomicroscope to remove the anterior segment, including the cornea, lens, iris, and ciliary body. The neural retina and RPE/choroid tissue were gently scraped from the sclera. After PBS rinses, the tissue was transferred to tissue storage solution and sent to Genergy Bioscience Co., Ltd. (Shanghai), for further processing.

Normalization of low-mass cell filtration and gene expression data

Low-quality cells were filtered using Seurat software. By default, cells with gene counts greater than or equal to 200, mitochondrial genes with unique molecular identifier (UMI) sequence counts less than or equal to 10%, erythrocyte marker genes with UMI sequence ratios less than or equal to 10%, or bicellular cells identified by the Scrublet software were considered low-quality cells and were filtered and removed. For the remaining high-quality cells, we used Seurat’s NormalizeData function to normalize the gene expression counts of each cell under the default parameters.

Cluster analysis of samples

For the number of UMI sequences of high-quality single cells and genes within the sample, the total number of UMIs per cell and the scaling factor ratio of 10,000 were calculated to correct for the depth of cellular sequencing to perform a standardized normalization process. Based on the diffusion coefficient measure of the high degree of variability of genes, the similarity between the cells was determined using principal component analysis (PCA) downscaling, and the closer the distance of the samples was, the closer the expression trend of the genes was. The top 30 principal components with the largest variance in the PCA results were visualized by the reduction method t-distributed stochastic neighbour embedding (TSNE). Seurat was used to separately cluster high-quality populations of cells. The PCA space was used to construct a nearest neighbour KNN map based on Euclidean distances, and then, the Louvain modularity optimization algorithm was used to cluster the cell populations.

Cell type discrimination analysis

The FindAllMarkers function was used to identify specific marker genes for a cell subpopulation. This process was performed by sequentially comparing cells of a specific cell subpopulation with all other cells by the Wilcoxon rank sum test to identify differentially expressed genes in that cell subpopulation. A Bonferroni-corrected P value less than 0.05 was used as a threshold to define statistically significant differentially expressed genes. Genes whose average expression in a specific cell subpopulation was more than twofold greater than the average expression in other cell subpopulations were selected as marker genes. The cell type to which each cell subpopulation belonged was annotated using previously reported passive cell type-related marker genes and the top-ranked differentially expressed genes. Detailed information on the retinal cell clusters is provided in Additional file 2: Table S3.

ERG recordings

As outlined in our previous study, ERG recordings were obtained [17, 60]. Over the course of the night, the mice underwent dark adaptation, and subsequent procedures were conducted in a dim-red light environment. Mice were subjected to deep anaesthesia via inhalation of a mixture of xylazine (80 mg/kg) and ketamine (16 mg/kg). To prepare the mice for the ERG test, we administered tropicamide, phenylephrine, and tetracaine (0.5%) to the mouse eyes before the procedure. Subsequently, the ERG responses of the mice were recorded and analysed using the Espion Visual Electrophysiology System from Diagnosis, LCC (Littleton, MA, USA). Dark-adapted ERGs were measured in response to varying intensities of light flashes.

Optical coherence tomography (OCT)

Mice were anaesthetized with 1% sodium pentobarbital anaesthetic (at a dose of 10μL/g body weight) prepared in 0.9% saline, and tobramycin eye drops (Runzheng, China) were added to the cornea. After the pupils of the mice were allowed to dilate, an OCT scan was started, and images were collected according to standard OCT test procedures. The images were also assessed and processed using the accompanying image processing software.

Immunohistochemical analysis

For retinal staining, the eyes of Pcyt1a-RKO (RKO) mice and control (Ctl) mice were collected, fixed in 4% paraformaldehyde for 2 h, and then soaked in 30% sucrose in PBS for another 2 h at 4 °C. Following lens removal, the eyes were inserted into an optimal cutting temperature (OCT) solution and sliced at a thickness of 12 μm. The sections were blocked and permeabilized with blocking buffer (5% FBS, 0.5% Triton X-100, 0.01% NaN3 in PBS) and then incubated with the primary antibody overnight at 4 °C. After being rinsed in PBS three times, the sections were incubated with Alexa Fluor 488/594-conjugated goat anti-rabbit/mouse secondary antibodies, and the nuclei were co-labelled with DAPI.

Western blotting (WB)

For WB, samples were homogenized in standard RIPA lysis buffer containing Complete Protease Inhibitor Cocktail (Roche, Switzerland). For the determination of the protein concentration, an enhanced BCA protein assay kit was used. Standard SDS–PAGE procedures were followed. The samples were blocked with 8% skim milk in TBST buffer for 2 h at room temperature before being incubated with primary antibodies in blocking solution overnight at 4 °C. Signals were developed using SuperSignal West Pico Chemiluminescent Substrate (Thermo Fisher, USA). The relevant antibody information is listed in Additional file 2: Table S4. Uncropped blots are shown in Additional file 3. ImageJ was used to determine the relative protein density.

LD formation analysis

Retinal primary Müller cells and ARPE-19 cells were collected after induction with 50 μM oleic acid (OA) (fatty acid-free bovine serum albumin: OA dissolved in EBSS at a ratio of 6:1) for 3 h. LDs were labelled with BODIPY 493/503 (catalogue number: D3922, Sigma, USA). Each labelled LD was automatically counted using Image-Pro Plus software. The statistical analysis was then carried out using independent samples or multiple Student’s t-test with GraphPad Prism 8.0 software.

Primary Müller glial (MG) cell isolation

A 24-well plate was lined with round coverslips and washed three times with PBS for 5 min. Then, 500 μL of poly-L-lysine (catalogue number: P4707, Sigma, USA) was added, and the plate was incubated overnight in a cell incubator. Poly-L-lysine was recovered, and 200 μL of laminin (catalogue number: L2020, Sigma, USA) was added. PBS diluted to a working concentration of 2 μg/cm2 was added to the well plates. Retinas from 12-day-old mice were rinsed with HBSS and then placed in 1 mL of preprepared papain (catalogue number: LS003119, Worthington, USA; working concentration: 20 U/mL). The tissue was processed into small pieces by slow aspiration 5 times with a 2.5-mL syringe. After treatment in the previous step, the tissue suspension was digested in an incubator at 37 °C for 30 min. One millilitre of DMEM (containing 10% FBS and 1% PS) was added to terminate the digestion, and the cells were filtered through a 40-μm nylon membrane. The filtered cell suspension was collected into a 15-mL centrifuge tube at 800 × g for 4 min. The cells were resuspended in MACS buffer (with serum in PBS) and washed again by centrifugation at 800 × g for 4 min. The supernatant was removed, and 1 mL of astrocyte medium was added for resuspension. The single-cell suspensions were added to a prewrapped 24-well plate and incubated at 37 °C in a 5% CO2 cell incubator. After 3 days, the cells were observed under a microscope for well attachment, and the medium was replaced with fresh astrocyte medium. After 7 days of culture, the cells were induced with OA (catalogue number: O1008, Sigma, USA) for 3 h. Samples were collected, and slices were prepared.

Free fatty acid (FFA) detection

For quantification of FFAs in retinas or cells, an FFA detection kit (catalogue number: ml092765, Mlbio, Shanghai, China) was used. Retinas from 2-month-old mice were homogenized and centrifuged at 5000 × g and 4 °C for 5 min, and the organic phase was collected for measurement. For cells, an extraction solution was added, and the mixture was sonicated, shaken for 15 min, and centrifuged at 5000 × g for 5 min at 4 °C. The organic phase was collected for analysis. The sample was mixed with working solution and centrifuged at 4 °C and 5000 × g for 5 min. The upper organic phase was separated, reagent IV was added, and the absorbance at 550 nm was measured after 10 min. The content of FFAs was calculated using the standard curve.

Reactive oxygen species (ROS) assay, fatty acid oxidation (FAO) assay, PC analysis, and α-ketoglutarate (α-KG) analysis

Two-month-old RKO mice and their littermate controls were chosen for this study. Retinas were isolated and homogenized, followed by centrifugation at 3000 × g for 10 min at 4 °C to obtain the supernatant for analysis. After treatment with 0.25% trypsin, the cells were harvested by centrifugation and lysed via cell sonication. The supernatant was collected for measurement after centrifugation at 3000 × g and 4 °C for 10 min. The experimental procedures were performed according to the manufacturer’s instructions for the kit used. Relevant information is included in Additional file 2: Table S4.

Measurement of mitochondrial Fe2+

Retinal tissue was processed into a single-cell suspension using the same method used for primary MG cell culture. The level of mitochondrial Fe2+ was determined using Mito-FerroGreen (catalogue number: M489, Dojindo, Japan). After the culture medium was removed, the cells were washed three times with serum-free medium, stained with Mito-FerroGreen working solution at 37 °C and 5% CO2 for 30 min in a cell culture incubator, collected by centrifugation after removal of the supernatant, washed with PBS and resuspended. The fluorescence intensity was measured using a flow cytometer with an excitation wavelength of 488 nm and an emission wavelength of 520 nm.

Analysis of lipid peroxides in cells

The cells were seeded into a 6-well plate and incubated overnight at 37 °C in a 5% CO2 incubator. After removal of the culture medium, the cells were washed twice with PBS and then treated with OA (50 μM) for 3 h before collection. For retinal tissue, the retinas of 2-month-old mice were dissected and prepared as single-cell suspensions. Working solution [prepared according to the instructions (catalogue number: L267, Dojindo, Japan)] was added to the prepared samples and incubated at 37 °C in a 5% CO2 incubator for 30 min. After the working solution was removed, the cells were washed twice with PBS. The cells were then digested, centrifuged, washed with PBS, and resuspended in 1 mL of PBS, after which the FITC fluorescence intensity was assessed using a flow cytometer (NovoSamplerQ, Agilent, USA).

GSSG/glutathione (GSH) analysis

A GSSG/GSH kit (catalogue number: G263, Dojindo, Japan) was used to determine the GSSG/GSH ratio. Retinas from 2-month-old mice were homogenized in 600 μL of 5% SSA solution and centrifuged at 8000 × g for 10 min, after which the supernatants were collected. The SSA concentration was adjusted to 0.5% with ddH2O. The cells were collected at 200 × g for 10 min at 4 °C, lysed with 80 μL of 10 mM HCl, and subjected to two cycles of freezing/thawing. Then, 20 μL of 5% SSA was added, and the mixture was centrifuged at 8000 × g for 10 min. The reaction mixture was prepared according to the kit instructions, and the absorbance was measured at 405 nm.

Proteomic analysisSample processing and collection

Three 2-month-old RKO mice and three littermate controls were selected, and retinal tissues were isolated after sacrifice. SDT buffer (4% SDS, 100 mM Tris–HCl, pH 7.6) was added to the sample. The lysate was homogenized, sonicated, and then boiled for 10 min. After centrifugation at 14,000 × g for 15 min, the supernatant was filtered through 0.22-μm filters. The proteins were quantified with a BCA protein assay kit (P0012, Beyotime). A total of 50–200 μg of protein/sample was reduced with 100 mM DTT for 5 min at 100 °C. The detergent, DTT, and low-molecular-weight components were removed using UA buffer (8 M urea, 150 mM Tris–HCl, pH 8.5) through repeated ultrafiltration (Sartorius, 30 KD). Iodoacetamide (100 mM IAA in UA buffer) was added to block reduced cysteine residues, and the samples were incubated for 30 min in darkness. The filters were washed three times with 100 μL of UA buffer and two times with 100 μL of 50 mM NH4HCO3 buffer. The protein suspensions were digested with 4 μg of trypsin (Promega) in 40 μL of 50 mM NH4HCO3 buffer overnight at 37 °C. The resulting peptides were collected as a filtrate and desalted on a C18 column.

Proteomic data analysis

Missing proteomic data were processed using the R package DEP (v1.16.0, https://bioconductor.org/packages/release/bioc/html/DEP.html) and then compared between groups by limma (v3.50.3, https://bioconductor.org/packages/release/bioc/html/limma.html) for determination of intergroup differences. The R package clusterProfiler was used for Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis. KEGG pathways with adjusted P values less than 0.05, as calculated by hypergeometric tests and the Benjamini–Hochberg method, were defined as significantly enriched pathways. The top 15 enriched pathways were selected for visualization.

Transmission electron microscopy

Mice were sacrificed by cervical dislocation after anaesthesia, and the eye was quickly removed after the removal of the periocular tissue. The samples were briefly rinsed with 1 × PBS, and electron microscopy fixative (3% glutaraldehyde) was quickly injected into the eye to prefix the samples; the samples were then fixed with 1% osmium tetroxide. The tissue was then infiltrated sequentially with low to high acetone concentrations (30%, 50%, 70%, 80%, 90%, 95%, 100%) and dehydrated step by step. The samples were sequentially infiltrated with dehydrating agent and embedding agent at ratios of 3:1, 1:1, and 1:3, followed by embedding with embedding agent. Sections were sliced using a Leica (EM-UC7) ultrathin sectioning machine at 60–90-nm thickness, spread, and fixed to a copper mesh. The samples were stained by sequential immersion in uranyl acetate (10–15 min) and lead citrate (1–2 min). The copper mesh was first observed at 6000 × , and then, the target area was selected for image capture.

Statistical analysis

All data are presented as the mean ± standard deviation (SD), and statistical analysis was performed using the GraphPad Prism version 8.0 software. Statistical significance was calculated using unpaired Student’s t-test or multiple Student’s t-tests. Values with p-value lower than 0.05 (P < 0.05) were considered statistically significant.