This study obtained mice, including wild-type (WT) C57BL/6J, Cd274[-/-], and Pdcd1lg2[-/-] mice, from Cyagen Biosciences Inc. (Suzhou, China). Programmed death ligand 1 (PD-L1) is encoded by Cd274 gene, and programmed death ligand 2 (PD-L2) is encoded by Pdcd1lg2 gene.The mouse Cd274 gene (GenBank accession number: NM_021893.3; Ensembl: ENSMUSG00000016496) is located on mouse chromosome 19. 7 exons have been identified, with the ATG start codon in exon 2 and TAA stop codon in exon 7. The mouse Pdcd1lg2 gene (GenBank accession number: NM_021396.2; Ensembl: ENSMUSG00000016498) is located on mouse chromosome 19. 6 exons have been identified, with the ATG start codon in exon 2 and TAG stop codon in exon 5. Exon 3 of mouse Cd274 to exon 5 of mouse Pdcd1lg2 were selected as target region. Cas9 and gRNA were co-injected into fertilized eggs for knockout (KO) mouse production. The pups were genotyped by PCR followed by sequence analysis. All experiments were carried out following guides from the Animal Care and Use Committee of Fudan University (Shanghai, China). Our study protocols were approved by the Association for Research in Vision and Ophthalmology Statement for the use of animals in ophthalmic and vision research. To avoid errors caused by sex differences, the same number of male and female mice were included in each experimental group.

Mice were anaesthetised using an intraperitoneal injection of chloral hydrate (500 mg/kg). An increase in IOP was triggered unilaterally by the injection of 2 μL superparamagnetic iron oxide (Bang Laboratories, Inc.) into the anterior chamber of the right eye of each animal based on a surgical microscope according to previous report (20, 21). Mice that developed signs of inflammation (including clouding of the cornea, oedematous cornea, iris prolapse, lens injury, and hyphema) were excluded from the study.

IOP of mice was measured on the 3rd, 7th, 10th, 14th, 20th and 30th days after superparamagnetic iron oxide injection. The measurement of IOP was made in conscious mice with a hand-held rebound tonometer (Icare TonoLab; Colonial Medical Supply, Franconia, NH). In brief, the tonometer was positioned in front of the eye, and thus the probe hit the cornea perpendicularly. After eliminating the highest and lowest values, the average IOP was presented automatically when six measurements were made. In addition, machine-generated mean was regarded as one reading. Three measurements were made at a given time point. At the same time, the average of these three values was recorded as the overall IOP. Because the IOP varies throughout the day, all IOPs were made at the same time of day (between 14:00 and 17:00 h).

Eyeball specimens were fixed in formalin the night before and were embedded and sectioned into 10-μm slices. Before staining, deparaffinisation was performed by applying xylene. Rehydration of sections was performed in an ethanol gradient before washing in a phosphate-buffered solution (PBS). Subsequently, the sections were stained with HE. A microscope (Leica DM4000, Germany) was employed to analyse retinal tissue morphology.

After lysis of mouse retinas in the RIPA buffer (Beyotime, Shanghai, China), the BCA approach (Beyotime) was utilised to determine protein content. Proteins (around 20 μg) were separated through 10% or 12% SDS-PAGE, followed by transfer on PVDF membranes (0.45 μm; Millipore, Shanghai, China). Later, 5% Bovine Serum Albumin was utilised to block membranes under ambient air for a 1-h period, followed by overnight incubation under 4°C using anti-Iba1 (1:1000, #ab178846, Abcam, Cambridge, MA, USA), anti-iNOS (1:1000, #ab178945, Abcam, Cambridge, MA, USA), anti-CD206 (1:200, AF2535, R&D Systems), anti-STAT1 (1:1000, #ab239360, Abcam, Cambridge, MA, USA), anti-STAT1 (phospho S727, 1:1000, #ab109461, Abcam, Cambridge, MA, USA), anti-STAT6 (1:1000, #ab32520, Abcam, Cambridge, MA, USA), and anti-STAT6 (phospho Y641, 1:1000, #ab263947, Abcam, Cambridge, MA, USA), with β-Tubulin being the endogenous reference. Membranes were rinsed sufficiently through PBST three times. Subsequently, the membrane was exposed to 1-h HRP-labelled secondary antibody incubation. At last, chemiluminescence was used to analyse the expression of each protein.

After the eye specimens were rapidly obtained from anaesthetised mice with no perfusion fixation, they were subject to 1-h fixation using 4% paraformaldehyde (PFA) under 4°C, followed by the removal of the anterior segment of the eyes. Later, the specimens were dehydrated using gradient sucrose solutions (10% for 30 min and 20% for overnight dehydration) under 4°C, followed by embedding of eyecups within the optical coherence tomography (OCT) compound (Tissue Tek, Torrance, CA, USA). Thereafter, this assay sliced the retina into 10-μm vertical sections with a freezing microtome (Leica, Nussloch, Germany). The retinal slices were rewarmed at room temperature for 30 min, followed by 15-min 4% PFA fixation, 1-h blocking using the solution that contained 0.1% Triton X-100 and 5% goat serum albumin supplemented within PBS under ambient temperature for a 1-h period, and overnight probing using primary antibodies under 4°C. Primary antibodies listed below were used in this study, rabbit-anti-Iba1 (1:100, #ab178846, Abcam, Cambridge, MA, USA), goat-anti-CD206 (1:8, #AF2535, R&D Systems), and rat-anti-CD16/CD32 (1:50, #BD553141, BD Biosciences). Membranes were then sufficiently rinsed using PBS, followed by an additional 1-h fluorescence secondary antibody incubation (1:1000), anti-rat IgG Alexa-Fluor-555 (Thermo Fisher Scientific, Inc., USA), anti-rabbit IgG Alexa-Fluor-488 (Thermo Fisher Scientific, Inc., USA), and anti-goat IgG Alexa-Fluor-633 (Thermo Fisher Scientific, Inc., USA) under ambient temperature. Sections and coverslips were washed and sealed using the anti-fade mounting agent and stained with 4’,6-diamidino-2-phenylindole (DAPI) (Abcam, Cambridge, MA, USA). This work employed Olympus FV1000 confocal laser scanning microscope for image acquisition. The value of Iba1, CD16/CD32, and CD206 positive cells were calculated. Image-Pro Plus software (version 6.0; Media Cybernetics Inc., Rockville, MD, USA) was used to analyse immunofluorescence images.

TRIzol reagent was used to extract total RNA with the manufacturer’s instructions. The NanoDrop 2000 spectrophotometer (Thermo Fisher Scientific, USA) was used for evaluating RNA content and quality. Agilent 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA, USA) was used for assessing RNA integrity. TruSeq Stranded mRNA LT Sample Prep Kit (Illumina, San Diego, CA, USA) was used in library construction following specific protocols. OE Biotech Co. Ltd. (Shanghai, China) is credited for transcriptome sequencing and analyses.

This study applied Illumina HiSeq X Ten platform for library sequencing to produce paired-end reads of 150 bp. First, Trimmomatic was used to process raw reads in the FASTQ format (22), while clean reads were obtained by removing low-quality ones. Later, HISAT2 was utilised to map clean reads against human genome (GRCh38) (23). Cufflinks was utilised to determine gene FPKM (24, 25), whereas HTSeqcount (26) was utilised to obtain the gene read numbers. R package DESeq (2012) was used to analyse DEGs (27) selected upon the thresholds of Foldchange (FC)>2 or <0.5 and p<0.05. DEGs expression profiles within diverse samples and groups were analysed by hierarchical clustering. According to hypergeometric distribution, DEGs were then subject to gene ontology (GO) as well as Kyoto Encyclopaedia of Genes and Genomes (KEGG) (28) analysis with R package. StringTie was subsequently applied in read reassembly (29), and reference genome was compared against known genes annotated to extend gene structure and identify new transcripts with Cuffcompare. The number of gene counts in each sample was standardised using the DESeq2 software, and the expression was assessed as the base mean value. The difference multiple was calculated, and the significance level of the difference was evaluated using a negative binomial distribution test. Finally, protein-coding DEG were screened in accordance with the results of the difference multiple and different significance test. After identification, DEGs were enriched and analysed by GO. DEG functions were described and combined with GO annotation results. In the GO function enrichment analysis, the numbers of DEGs of each GO entry were counted, and the significance of DEG enrichment of each GO entry was computed with the use of a hypergeometric distribution algorithm. The calculation returned a P-value describing enrichment significance (each biological process, cellular component, and molecular function was computed using Fisher’s exact test). The lower the P-value, the greater the statistical significance. Genes for follow-up studies were selected according to GO analysis results and biological significance. KEGG is the main public database used for pathway analysis of differentially expressed protein-coding genes, and the significance of DEG enrichment in each pathway was computed by adopting a hypergeometric distribution test.

Data were analysed by GraphPad Prism (ver. 9.0, GraphPad Software, USA). Quantitative results were presented as mean ± SD. Two groups were compared using t-test (two-tailed) and Fisher’s exact test. P<0.05 was considered statistically significant.

Injection of superparamagnetic iron oxide superparamagnetic Iiron oxide into the anterior chamber of mice to block aqueous humour outflow and simulate angle-closure glaucoma (20, 30) increased the IOP in both WT and PD-L1/PD-L2 dKO mice ( Figure 1A ). Each mouse received a superparamagnetic iron oxide injection into the right eye, the experimental eye, whereas the left eye, the control eye, did not receive such injections. IOP monitoring began 3 days before the administration of the superparamagnetic iron oxide injection. The IOP values of WT or PD-L1/PD-L2 dKO mice that did not receive superparamagnetic iron oxide injections were not significantly different at any time point, whereas the IOP values of experimental eyes significantly increased from day 3 to day 30 after the superparamagnetic iron oxide injection was administered. The IOP values were not significantly different between WT and PD-L1/PD-L2 dKO mice ( Figure 1A ). Haematoxylin and eosin (HE) staining of the experimental eyes 30 days after the administration of the superparamagnetic iron oxide injection showed no significant difference in retinal morphology between WT ( Figure 1B ) and PD-L1/PD-L2 dKO mice ( Figure 1C ).

Data suggest that the injection of superparamagnetic iron oxide into the anterior chamber comparably increases the IOP values of WT and PD-L1/PD-L2 dKO mice.

BRN3A antibody can specifically tag RGCs. The effects of PD-L1/PD-L2 dKO on RGC apoptosis in the established COHT model were assessed using WB and immunofluorescence staining (31). Immunofluorescence staining of retinal sections from normal mice not exposed to superparamagnetic iron oxide injections showed large numbers of BRN3A-positive cells, indicating high numbers of RGCs in the mouse retina under normal conditions. The number of RGCs in PD-L1/PD-L2 dKO mice (19.7 ± 2.3) was higher than that in WT mice (16.1 ± 3.9), which is statistically significant. After the administration of the superparamagnetic iron oxide injection into the anterior chamber, the number of RGCs decreased with time. The number of surviving RGCs in PD-L1/PD-L2 dKO mice was greater than in WT mice on day 10 (16.8 ± 3.3) and day 30 (13.2 ± 1.5) after injection ( Figures 2A, B ). This finding was verified using WB, and substantial RGC apoptosis was detected in eyes exposed to COHT ( Figure 2C ). However, the decrease in BRN3A expression in dKO mice was lesser than that in WT mice. The BRN3A expression levels in WT mice on days 10, 20, and 30 after superparamagnetic iron oxide injection were 0.63 ± 0.11, 0.49 ± 0.38, and 0.38 ± 0.26, respectively. The corresponding values in dKO mice were 0.89 ± 0.03, 0.80 ± 0.05, and 0.68 ± 0.16, respectively ( Figure 2D ). Therefore, we calculate and compared the ratio of RGCs reduction in WT and dKO mice for each time point after superparamagnetic iron oxide injection ( Figure 2E ). On days 10, 20, and 30, RGCs reduction in WT mice are higher than those in dKO mice.

These results indicate that in this mouse COHT model, PD-L1/PD-L2 dKO prevents damage to RGCs caused by increased IOP.

Immunofluorescence staining of microglia show that a small number of positive cell markers, such as Iba1, can be found in both WT and PD-L1/PD-L2 dKO mice without superparamagnetic iron oxide injection, and activated microglia were rarely detected. After injecting superparamagnetic iron oxide into the anterior chamber of the eye, retinal staining revealed a large number of positive cells that had migrated to the ganglion cell layer, indicating that after RGC injury, microglia were overactivated and aggregated (5–7, 32–34), accompanied by a severe inflammatory reaction. However, the retinas of WT and PD-L1/PD-L2 dKO mice showed different trends in microglial polarisation after the increase of IOP induced by the superparamagnetic iron oxide injection. Microglia activated in WT mice mainly associated with the M1 phenotype ( Figure 3A ), whereas microglia in PD-L1/PD-L2 dKO mice were mainly associated with the M2 phenotype ( Figure 3B ). The analysis of immunofluorescence-positive cells using ImageJ software showed that in animals without superparamagnetic iron oxide injection, the numbers of M1 microglial cells were almost the same in the eyes of WT mice (1.31 ± 0.64) and PD-L1/PD-L2 dKO mice (1.48 ± 0.60). After superparamagnetic iron oxide injection, the number of M1 microglia in the retina increased with time. The greatest difference between WT mice (2.32 ± 1.24) and PD-L1/PD-L2 dKO mice (1.62 ± 0.59) was observed after 10 days. After 30 days, M1-polarised microglia in WT and PD-L1/PD-L2 dKO mice were at 2.56 ± 1.12 and 2.47 ± 0.54, respectively ( Figure 3C ). M2 microglial cells in WT and PD-L1/PD-L2 dKO mice without superparamagnetic iron oxide injection were at 0.66 ± 0.61 and 1.18 ± 0.32, respectively. Retinal microglia with M2 phenotype also increased with time after superparamagnetic iron oxide injection. M2 microglia in WT and PD-L1/PD-L2 dKO mice were at 1.24 ± 0.47 and 1.67 ± 0.64, respectively, on day 10. The difference between WT (1.62 ± 0.62) and PD-L1/PD-L2 dKO (2.71 ± 0.73) mice in M2 microglia was greatest after 30 days ( Figure 3D ). WB confirmed these observations ( Figure 3G–J ) and revealed that the increase in Iba1 expression in dKO mice was greater than that in WT mice ( Figure 3E ). The Iba1 expression levels of WT and dKO mice 10 days after superparamagnetic iron oxide injection were 1.34 ± 0.09 and 1.61 ± 0.23 of those of the non-injected groups, respectively, 1.47 ± 0.05 and 1.77 ± 0.12 on day 20, and 1.84 ± 0.36 and 2.36 ± 0.30 on day 30 ( Figure 3F ).

This suggests that PD-L1/PD-L2 dKO can polarise mouse retinal microglia into the M2 phenotype.

In general, STAT1 phosphorylation is increased in M1-activated microglia, whereas STAT6 phosphorylation is upregulated in M2-activated microglia (35). Figure 4 shows that following superparamagnetic iron oxide injection, STAT1 phosphorylation levels in the retinas of WT mice and dKO mice increased, representing an increased activation of M1 microglia. The degree of STAT1 phosphorylation in dKO mice was greater than that in WT mice on days 10 (2.95 ± 0.32 and 2.16 ± 0.14), 20 (4.14 ± 2.76 and 2.97 ± 1.71), and 30 (9.54 ± 3.55 and 4.54 ± 0.38) after superparamagnetic iron oxide injection ( Figure 4I ). For STAT6, the phosphorylation levels of dKO mice were greater than that of WT mice on days 10, 20, and 30 after injection. The values were 4.95 ± 4.49 vs. 1.87 ± 1.43, 2.19 ± 1.13 vs. 1.38 ± 0.35, and 1.40 ± 0.69 vs. 1.01 ± 0.17, respectively ( Figure 4J ). Our study found that the phosphorylation degree of STAT1 and STAT6 in both WT and dKO mice increased after receiving superparamagnetic iron oxide injection, especially in dKO mice. The difference was that the phosphorylation degree of STAT1 increased over time, while the phosphorylation degree of STAT6 decreased over time. Therefore, receiving superparamagnetic iron oxide injection will enhance the polarization of M1 and M2 of mouse retinal microglia, especially in dKO mice.

So we infered that dKO mice can promote the polarization of M1 and M2 microglia in COHT model.

To explore changes of mRNA expression in PD-L1/PD-L2 dKO mice in COHT model, we investigated differential gene expression analysis between WT and dKO mice at different time poionts (10, 20 and 30 days) in the COHT model. A total of 289 DEGs were found in 10 days group including 244 up-regulated DEGs and 45 down-regulated DEGs ( Figure 5A ), the data in 20 days group were 399 DEGs in total including 327 up-regulated DEGs and 72 down-regulated DEGs ( Figure 5B ). Moreover, 1038 DEGs were obtained in 30 days group including 461 up-regulated DEGs and 577 down-regulated DEGs ( Figure 5C ).

In order to understand the gene function and the pathway of valid DEGs involved, we performed GO and KEGG analyses of 1726 DEGs. The top ten terms of biological processes (BPs), cellular component (CCs) and molecular functions (MFs) were listed separately ( Figures 5D–F ). In BPs, GO functions were mainly enriched in immune system process, innate immune response and immune response. CCs regarding extracellular region, extracellula space and collagen-containing extracellula matrix. MFs including cytokine activity and chemokine activity were found.

Similarly, the pathways related to immune processes and regulations were also found in KEGG enrichment analysis, such as the NF-κB, TNF, PI3K/Akt and toll-like receptor signalling pathways etc. ( Figures 5G–I ). These results suggested that immune factors were critical in COHT of the blocking PD-1 pathway.