Single-Cell Transcriptome Analysis of the Retina During the Development of Normal Tension Glaucoma in Mice Authors: Xinna Liu, Di Zhang, Mengxian Du, Zhengbo Shao, Xikun Xu, Wulian Song, Sisi Chen, Xianghui Li, Leyi Qiu, Fengyi Guo, Huiping Yuan Published: 2025 ABSTRACT AbstractGlaucoma is characterized by the progressive degeneration of retinal ganglion cells (RGCs), the pathogenesis of which is still unknown. Many studies have reported that retinal glial cells play an important role in neuron degeneration, but the underlying mechanisms are not well defined. Systematic analysis at the single-cell level is crucial for better understanding the molecular alterations in major retinal cell types and the interactions between glial cells and RGCs during disease progression. Here, we performed single-cell RNA sequencing (scRNA-seq) on OPTN E50K mutant mice, an in vivo model of normal tension glaucoma (NTG), of different ages and analyzed them by bioinformatics methods to obtain a complete gene expression profile of retinal cells in NTG. We identified the transcriptional signatures of RGCs and changes in their cellular interactions with glial cells during NTG development. Microglia were initially reactive in the early stage of NTG and progressively increased TNF-α expression with age, contributing to retinal degeneration, in which the CD74–MIF pathway played an important role. In addition, Müller cells interacted with microglia and became reactive in NTG. Our study provides a detailed analysis of RGCs and retinal glial cells and enhances the understanding of the mechanism of optic nerve damage during NTG progression, suggesting promising new targets for diagnostic and future therapeutic strategies.Supplementary InformationThe online version contains supplementary material available at 10.1007/s10753-025-02422-1.Keywords: OPTN E50K mutation, Normal tension glaucoma, Aging, Single-cell RNA sequencing, Glial cells Introduction Glaucoma is the leading cause of blindness worldwide, and its prevalence continues to increase with the ageing of the global population [1]. It is a neurodegenerative disease characterized by the progressive degeneration of retinal ganglion cells (RGCs), resulting in vision loss and eventual blindness, but the underlying mechanisms are still unknown. The only proven clinically treatable therapy is lowering intraocular pressure (IOP), which is unable to restore damaged RGCs and protect against blindness and is not an ideal option for normal-tension glaucoma (NTG) patients with no increased in IOP [2]. The limitations of current treatments have encouraged in-depth research on the mechanism of optic nerve injury in NTG. Previous studies on glaucoma have focused mostly on the pathological mechanism through which elevated IOP induces damage to RGCs [3], and an integrative and elaborate strategy to elucidate the time-course alternations of cellular interactions and molecular changes in each type of retinal cell during NTG progression is lacking. The E50K mutation of the optineurin (OPTN) gene is a leading cause of NTG, which directly affects RGCs and leads to particularly severe glaucomatous symptoms but with no increased in IOP [4, 5]. In previous studies, we successfully established OPTN E50K mutant (E50K) mice as an NTG model by CRISPR/Cas9 gene-editing technology and verified their age-related glaucomatous manifestation [6–8]. The single-cell RNA sequencing (scRNA-seq) technique, which allows deep sequencing and identification at the single-cell level, has become an option for studying different cell types and cellular interactions. In this context, we performed scRNA-seq and bioinformatics analysis on the retinas of E50K mice from 3 to 18 months to profile the gene expression profiles of RGCs and neuroglia, and characterized the disease-associated molecular changes and cellular mechanisms underlying retinal degeneration, offering new insights for early diagnosis and new therapies for glaucoma. Method Animal Models All experimental protocols were approved by the Institutional Animal Care and Use Committee at Harbin Medical University and conformed to the Guide for the Care and Use of Laboratory Animals (NIH, 8th Edition, 2011). And this study was approved by Ethics Committee of the Second Affiliated Hospital of Harbin Medical University (Permit Number: YJSDW2024-049). Based on previously described protocols and results in our studies, mice with OPTN E50K mutation was established as a NTG mouse model and wild type C57BL/6 J mice (WT) served as a control [9]. Mice with OPTN E50K mutation at 3-, 12- and 18-months were designated as E3, E12 and E18, respectively, while the age-matched WT control groups were correspondingly defined as W3, W12 and W18. Mice were housed in the special pathogen-free (SPF) animal facility at the Second Affiliated Hospital of Harbin Medical University. Preparation of Retina Cell Suspensions and Libraries for Single-Cell RNA Sequencing Before scRNA-seq was performed, several measuring procedures were conducted to verify the age-points for analysis. Briefly, immunostaining was used to assess RGCs loss and visual function was assessed by a light/dark test and optomotor response tests during NTG development [8, 10], compared to their age-matched WT counterparts (Supplemental Fig. 1). Both WT and E50K mice demonstrated age-related visual impairment, with the E18 mice displaying more severe retinal degeneration similar to clinical NTG and E12 mice showing an initial change similar to E18. Thus, E50K mice at 3 age-points (3-, 12- and 18-months) were sacrificed for the scRNA-seq with the age-matched WT mice as control groups. Retina tissue obtained from each group was placed in a sterile RNase-free culture dish containing an appropriate volume of calcium-free and magnesium-free PBS on ice. After cutting and washing, tissues were then dissociated into single cells using a dissociation solution at 37 ℃ with gentle shaking for 20 min and terminated by adding PBS containing 10% fetal bovine serum (FBS). The resulting cell suspension was filtered, centrifuged at 300 g for 5 min and then resuspended in PBS. After red blood cell lysis, the suspension was added into Dead Cell Removal MicroBeads (MACS 130–090–101) and Miltenyi® Dead Cell Removal Kit (MACS 130–090–101). The cell viability of single-cell suspensions obtained should exceed 85% and the concentration need to be 700–1200 cells/μl. Then the suspensions were processed using 10 × Chromium system (10 × Genomics), following cDNA amplification and library construction, which sequenced on the Illumina NovaSeq 6000 sequencing system by LC-Bio Technology Co., Ltd., (HangZhou,China) at a minimum depth of 20,000 reads per cell. Bioinformatic Analyses Sequencing results were converted to FASTQ format using Illumina bcl2fastq software (version 2.20) and imported into R package Seurat (version 3.0) for filtration, normalization, cell clustering, and differential gene expression analysis based on the calculated single-cell expression matrix by CellRanger. Cells, in which genes were fewer than 500 and mitochondrial gene ratio exceeded 25%, were excluded. A total of 210,754 cells were used to further analysis after quality control. To visualize the data, we employed t-Distributed Stochastic Neighbor Embedding (tSNE) for cell projection. Differential expression analysis of scRNA-seq datasets was conducted by the FindAllMarkers function and filtering parameters were gene expression in cells above 10%, p-value < 0.01 and log2(Fold change) threshold of 0.26. Differential gene enrichment analysis was performed to gain further insights into the biological functions of differentially expressed genes (DEGs) across enriched Gene Ontology (GO) terms pertaining to biological processes, cellular components, and molecular functions. Additionally, metabolic pathways associated with scRNA-seq data were obtained from the publicly available Kyoto Encyclopedia of Genes and Genomes (KEGG) database. CellPhoneDB (www.cellphonedb.org) was utilized for the prediction of ligand-receptor expression and interaction, wherein the strength of the interaction between ligands and receptors is associated with their respective gene expressions. For each gene within a cell group, the percentage of cells expressing that particular gene and its mean expression were calculated only when both ligand and receptor genes were expressed in at least 10% of cells, with p-value < 0.05. Cell Culture Lentiviruses have been utilized to successfully deliver E50K mutation in rat retinal cells (R28), murine microglia (BV2) and rat Müller cells (RMC-1) after depleting endogenous OPTN to establish stable in vitro cell models of NTG, which designed as E50K group (E50K) and cells infected with WT OPTN was control group (Con) in these mentioned cells respectively. After infection, cells were performed antibiotic selection and assessed efficiency by western blotting, then for further experiments. Quantitative Real-Time PCR (qPCR) Total RNA was extracted from cultured cells with Trizol and converted into cDNA using Transcription First Strand cDNA Synthesis kit (Roche, UK). Real-time qPCR was carried out according to the manufacturer's protocols with LightCycler® 480 SYBR Green I Master (Roche, UK). mRNA expression for genes of interest was normalized with β-actin using the 2ΔΔct method. Immunofluorescent Staining Each group contained three mice. Eye tissues obtained from each group were fixed with 4% PFA overnight at 4 °C, placed in a 25% sucrose gradient for dehydration, and then frozen in liquid nitrogen, as we previously report [8]. Frozen eyes were embedded in Optimal Cutting Temperature media and cut into 6 µm sections. After washing with PBS, eye tissue sections were blocked in 0.5% goat serum for 1 h, then subsequently incubated in primary antibodies overnight at 4 °C. To examine RGCs, microglia and Müller cells, anti-Neun (1:50, Abcam), anti-Iba-1 (1:50, Abcam) and anti-CRALBP (1:50, Abcam) antibodies were used, respectively. Tissue sections were incubated with secondary antibody for 1 h at room temperature, followed by DAPI for 5 min to counterstain nuclei. Images were taken using a confocal fluorescence microscope and quantified with Image-J software. Statistical Analyses All values are presented as mean ± standard error of the mean (SEM). Statistical analysis was performed using GraphPad Prism version 10.0 (GraphPad Software Inc.). Data was analyzed by unpaired Student’s t-test for comparisons between 2 groups, or one-way analysis of variance (ANOVA) for comparisons among 3 or more groups. P < 0.05 was considered statistically significant. Result Single-Cell Retinal Transcriptomic Atlas of the NTG Mouse Model We performed scRNA-seq transcriptome analysis using the 10 × Genomics platform on E50K mice at 3, 12 and 18 months of age, timepoints consistent with the increasing severity of the glaucomatous phenomenon, as shown in Supplemental Fig. 1, and designated as E3, E12 and E18, respectively. WT mice of the same age served as matched control groups (W3, W12 and W18). The entire procedure is illustrated in Fig. 1A. Following data processing and quality control filtering, 110,641 and 100,113 retinal cells derived from WT and E50K mice, respectively, were retained for further analysis. The cells were unbiasedly classified into 36 transcriptionally distinct clusters by Seurat analysis (Fig. 1B). On the basis of the expression of well-established markers in different retinal cell types (Fig. 1C), the 36 clusters were identified as Cone, Rod, Amacrine Cell, Bipolar cell, Müller cell, Horizontal cell, Astrocyte, Endothelial cell, Microglia and RGC in each group of both E50K and WT mice via a tSNE plot (Fig. 1D). There was no notable difference in cell types between WT and E50K mice with age, but the number of retinal neurons and glias were changed (Fig. 1E). We found that the number of neurons decreased while the number of glias increased with age in both E50K and WT mice, especially in the E50K group, and that the reduction in the number of neurons occurred earlier, which was in accordance with the RGC loss mentioned in Supplemental Fig. 1. Single-cell transcriptomic analysis of OPTN E50K mutant mice with aging. (A) Experimental work flow. Retinas from both OPTN E50K mutant (E50K) and wild-type (WT) mice at 3-, 12- and 18 months were isolated (designated as E3, E12, E18, W3, W12 and W18) for single cell suspension, then sequenced and analyzed by single-cell RNA sequencing (scRNA-seq) platform. (six samples from three mice assigned into one biological replicate, n = 3 replicates per group). This figure is created by biorender. (B) t-Distributed Stochastic Neighbor Embedding (tSNE) plot revealing 36 specific clusters representing the major retinal cell types. (C) The expression of known cell-type specific marker genes in cell groups showing by violin plots. (D) tSNE plot displaying the known cell types of E50K and WT retina at different age. (E) The distribution of retinal neurons and glial cells in each group Transcriptomic Analysis of the E50K Mutation-Dependent Effect on RGCs During NTG Development Progressive RGC loss is indicative of glaucoma pathogenesis. Thus, we first surveyed the transcriptomic changes between E50K RGCs and WT RGCs. We observed that the expression of genes, including Rps18, Rplp2, Rps25 and Mrpl18, significantly increased and that of Rps15a, Mrpl12, Mrps36 and Mrpl41 decreased in E50K RGCs (Fig. 2A). KEGG analysis of these genes was then conducted, and pathways enriched in clusters related to the cellular process, environmental information processing, genetic information processing, human diseases, metabolism and organismal systems, where subcluster signal transduction had the largest number of pathways (Fig. 2B), were identified. Considering that E12 is an early stage of NTG (Supplemental Fig. 1), we compared the KEGG pathways of all significantly differentially expressed transcripts in E12/E3 and W12/W3 and identified 26 E50K mutation-specific pathways at this time point (Fig. 2C), including the PI3K-Akt, Rap1, mTOR, FoxO, HIF-1 and JAK–STAT signaling pathways (Fig. 2D). The effect of OPTN E50K mutation on RGCs. (A) Volcano plot demonstrating the dual thresholds differentially expressed genes (DEGs) of RGCs in E50K mice versus WT mice. Each dot representing a single quantified gene and colored dots representing DEGs with blue indicating relative decrease and red increase. (B) DEGs were analyzed by Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways. (C) Venn diagrams representing the significantly enriched KEGG pathways in E12 versus E3 and W12 versus W3. (D) Bubble chart of the top E50K-specific quantified pathways in E12/E3. (E) Heatmap exhibiting the interaction profiles between every two cell types in retina. Bar charts showing the alterations on cellular interactions between RGCs and microglia (F) as well as RGCs and macroglia (G) in E50K mice The retina consists of several well-defined cellular layers, including different types of neurons and glia, and these cells play important roles in supporting RGC heath. To determine their underlying relationships, we analyzed the overall cellular communication in retinal cells overall and detected strong relationships among RGCs, macroglia and microglia (Fig. 2E). Interestingly, the interactions of RGCs with both microglia and macroglia were strongly decreased in E12 mice (Fig. 2F and G), and the underlying mechanism is worthy of exploration. Alterations in the Reactive State and Transcriptomic Features of Microglia in NTG Emerging evidence suggests that microglial activation is an important pathological mechanism, even perhaps the initial changes in neurodegenerative diseases, including glaucoma [11, 12]. However, the dynamic changes in the morphology and function of microglia during NTG progression are still not clear. Thus, we performed immunofluorescence staining with the specific microglia marker Iba-1 on E50K retinas of different ages to observe the quantity and morphological differences in NTG (Fig. 3A and B). We found microglia was small in number at young age with tiny cell bodies in both the WT and E50K. The number of microglia increased and their morphology changed to an amoeboid shape with aging, especially under the E50K mutation, and even advanced at 12 months of age compared with the WT control. For further understanding, we analyzed transcriptomic changes in these E50K microglia, and compared DEGs in the context of aging to elucidate the underlying mechanism of NTG. Notably, inflammatory factors such as TNF, CCL4, IL-1α, IL-1β, CCL2 and CCL3 were dramatically upregulated in the microglia of E50K mice with increasing age (Fig. 3C). These increased mRNA levels of TNF-α, CCL2, CCL3, CCL4, IL-1α and IL-1β were also supported by the in vitro E50K model of microglia (Fig. 3D). Owing to the E12 and E18 groups exhibiting similar changes, we conducted KEGG pathway enrichment for the two groups and investigated the 8 overlapping KEGG pathways (Fig. 3E). After ranking them on the basis of their rich factor and p value, we found that neurodegenerative Parkinson’s diseases (PD), Huntington’s diseases and Alzheimer’s diseases (AD), and amyotrophic lateral sclerosis were commonly enriched, and the most significant P value was associated with the TNF signaling pathway in signal transduction (Fig. 3E). TNF-α is one of the most classical cytokines that regulates the inflammatory response and induces neuroinflammation in degenerative disease. Our retinal sections stained for TNF-α, confirmed that the TNF-α expression started to change in E12 microglia and was strongly increased in E18 microglia, paralleling the alteration in the reactive state of microglia during NTG development (Fig. 3F and G). We also explored the pathological effects in an in vitro microglial model, and the results verified that TNF-α was greatly increased in the E50K group (Fig. 3H and I). These results suggested that microglia transformed into a reactive state in an early stage of NTG and started to secrete inflammatory factors, contributing to RGC death. Alterations to reactive state of microglia in NTG. (A-B) Retinal flat mount from WT and E50K mice were stained against Iba-1 (red), captured by confocal fluorescence microscope and calculated to gain the microglia density in the ganglion cell layer (GCL) by image J. Scale bars, 50 μm. (C) The relative expression of the inflammatory factors in the microglia of three E50K groups. (D) The mRNA levels of TNF-α, CCL2, CCL3, CCL4, IL-1α and IL-1β were determined by real-time qPCR in BV2 cells, in which E50K mutation were delivered by lentivirus, designed as E50K group (E50K) and matched control group as Con. (E) The overlapped KEGG pathways enriched in microglia from both E12 and E18 group. (F-G) Retinal sections were immunostained for TNF-α (green) in all three E50K groups. Cell nuclei were visualized with DAPI (blue). Scale bar, 50 µm (n = 3/group). (H-I) BV2 cells of E50K and Con group were immunostained for TNF-α (green). Data shown as mean ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001. Data were analyzed using Two-Way ANOVA (B), Student’s t-test (D and I) and One-Way ANOVA (G) Since communication between microglia and RGCs also began to change in the early stage of NTG (Fig. 2F), we applied CellphoneDB analysis to investigate the potential correlations underlying the interactions of these two types of cells during NTG development. The number of ligand‒receptor pairs in each E50K group is shown in Fig. 4A, and considering our previous results (Figs. 2F and 3), we identified three interactions specially associated with older E50K microglia and RGCs. Notably, the expression of CD74 and TNF receptors was significantly increased in microglia, and the expression of their corresponding ligands, macrophage migration inhibitory factor (MIF) and PTPRS, was upregulated in RGCs, especially the expression of the MIF–CD74 ligand–receptor pair with the top mean value (Fig. 4B and C). CD74, the dominant cell-surface receptor for MIF, leads to the activation of a plethora of pathways that are essential for cell survival, differentiation and proliferation [13] and is a hallmark of microglia in neurodegenerative disease [14–16]. Taken together, these results suggest that MIF–CD74 signaling may play an important role in retinal neuroinflammation during NTG. Thus, retinal sections from all E50K groups were stained with CD74 and Iba-1 (Fig. 4D and E) and MIF and Neun (Fig. 4F and G) to confirm NTG microglia-RGC crosstalk. The number of both CD74- and Iba-1- positive cells increased with age, and the percentage of CD74+ microglia (Iba-1+) reached approximately 80% of all microglia in the E18 retina. MIF expression, which is located in the ganglion cell layer (GCL), was significantly greater at E12 and E18 than at E3, especially at E12. E50K mutation affected cellular communication between microglia and RGC during NTG progress. (A) Venn diagrams representing the number of ligand-receptor pairs between microglia and RGCs in three E50K groups. (B-C) Dot plot representing the expression levels of selected ligand-receptor genes in each cell type. The color of each dot represents the gene’s expression level across the cell (average expression) and the size of each dot corresponds to the percentage of cells in the cell expressing the gene (percent expressed). (D-E) Retinal sections were stained against CD74 (green) and Iba-1 (red) in all E50K groups. Cell nuclei were visualized with DAPI (blue). The number of CD74+ and Iba-1.+ merged cells was calculated by image J. (F-G) Retinal sections were stained against MIF (red) and Neun (green) in all E50K groups. Cell nuclei were visualized with DAPI (blue). Data shown as mean ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001. Data were analyzed using One-Way ANOVA (E and G) Moreover, to verify the role of CD74 in microglia, we reclustered the microglia and identified five distinct clusters: C0, C1, C2, C3 and C4 (Fig. 5A). We found that CD74 was specifically expressed in cluster 3 (Fig. 5B), which also highly expressed Apoe, H2-Aa, H2-Ab1, H2-Eb1 and TNF (Fig. 5C). In addition, biological process (BP) gene sets enriched in cluster 3 were specifically related to inflammatory and immune responses, apoptotic processes, and inflammatory factor production (Fig. 5D). In accordance with previous observations, these results suggest that the mechanism by which microglia undergo inflammatory activation and immune deregulation in response to the E50K mutation acting on RGCs may be associated with the MIF–CD74 signaling axis, which shares a common pathological pathway with AD and aging [17, 18]. The subclassification and DEGs analysis of microglia. (A) Uniform Manifold Approximation and Projection (UMAP) diagram for classification of microglia. (B) The DEGs expressed in each cluster of microglia. Log2(fold change) ≥ 0.26. (C) Violin diagram of the marker genes specifically expressed in cluster 3 of microglia. (D) The biological process (BP) enrichment map of cluster 3 Alterations in Transcripts and Cellular Communication in Müller Cells During NTG Progression Müller cells constitute 80–90% of retinal macroglia and are involved in the maintenance of retinal homeostasis [19]. Many studies have shown that the interactions among different types of glial cells embedded in the retina contribute to neuronal injury, including glaucoma, but the underlying signaling cascades are mostly unknown [20, 21]. Therefore, we explorated scRNA-seq data to clarify the role of Müller cells in NTG progression. First, we observed the proportions of Müller cells and astrocytes among macroglia, which was consistent with recent findings (Fig. 6A). We compared marker genes in these two groups and detected high expression of Gfap, Pdgfra, Pax2 and lgfbp5 in astrocytes and high expression of Rlbp1, Glul, Gpr37 and Slc1a3 in Müller cells (Fig. 6B). Moreover, the expression of these marker genes was not affected by the E50K mutation (Fig. 6C). Therefore, we chose Rlbp1 (CRALBP) as a marker of Müller cells and GFAP as a marker of astrocytes for subsequent retinal immunostaining to clarify the changes in these cells during NTG development. Our results revealed that the cell bodies of Müller cells are located in the inner nuclear layer (INL) and that their cell processes spread across the whole retina, which is in accordance with the reported characteristics of these cells (Fig. 6D). The number of Müller cells increased with increasing age in both the E18 and W18 groups (Fig. 6E). In addition, the astrocytes in the E18 group were more reactive, with thicker somas, numerous processes and high GFAP immunoreactivity, in line with recent findings of the astrocyte phenomenon in glaucoma (Fig. 6F and G) [22]. The classification and distribution of macroglia. (A) UMAP diagram for classification of macroglia. (B-C) The marker genes expression of Müller cells and astrocyte in E50K and WT retina. (D-E) Retinal sections of all E50K and WT groups were stained against CRALBP (red). Cell nuclei were visualized with DAPI (blue). (F-G) Retinal flat mounts from E50K mice were stained against GFAP (green) and captured by confocal fluorescence microscope. Scale bars, 50 μm. Data shown as mean ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001. Data were analyzed using Two-Way ANOVA (E) and One-Way ANOVA (G) To further investigate the transcriptomic alterations of Müller cells, we performed unsupervised clustering analysis and classified them into five subpopulations, which were named C0 to C4 (Fig. 7A and B). Next, we compared these five clusters of Müller cells in all groups and found that the cluster proportions differed with age (Fig. 7C). Surprisingly, we found that C0 was greatly increased in both the E18 and W18 groups compared with the other clusters, suggesting that C0 might be the main reason for the increase in Müller cells at old age. GO analysis revealed that the DEGs in C0 were enriched mainly in biological processes related to cell fate, including cell differentiation, maturation, migration and proliferation; negative regulation of neuron projection development and neuron differentiation; positive regulation of apoptotic process; and the molecular function of glutathione transferase, oxidoreductase and DNA-binding transcription factor activity (Fig. 7D). To explain the senescence-related influence on both old E50K and WT C0 Müller cells, we compared the DEGs of C0 in W18 versus W12 (Fig. 7E) as well as E18 versus E12 (Fig. 7F), finding 68 DEGs in WT and 79 DEGs in E50K. Among them, 27 genes were differentially expressed in both the W18 and E18 groups (Fig. 7G), and the majority of them (25 genes) were upregulated, especially in the E18 group (Fig. 7H), indicating that age-related Müller cell proliferation might share similar pathological mechanisms in different genotypes. Indeed, KEGG enrichment was largely concentrated in the TNF signaling pathway, oxidative phosphorylation and neurodegenerative diseases (Fig. 7I). The subclassification and DEGs analysis of Müller cells. (A) UMAP diagram for classification of Müller cells. (B) Heatmap depicted DEGs in each cluster of Müller cells. (C) Relative proportion of each Müller cell subcluster in WT and E50K retina. (D) Bioinformatics analysis of DEGs in C0 by enriched Gene Ontology (GO) function in the Database for Annotation, Visualization, and Integrated Discovery (DAVID) database. Volcano plot demonstrating DEGs in W18 compared to W12 (E) and E18 compared to E12 (F). Log2(fold change) ≥ 0.26. Dots with blue indicating relative decrease and red indicating increase. (G) Venn diagrams representing the overlap of DEGs quantified in W18/W12 and E18/E12 groups. (H) Heatmap depicting the relative expression of the overlapped genes in four groups. (I) KEGG pathway analysis of the overlapped genes in (G) Müller cells and microglia are reportedly activated in glaucoma and work cooperatively in the context of RGCs glaucomatous damage [20, 22]. To investigate how the two types of cells interact in NTG, we performed CellPhoneDB between Müller cells and microglia, and the number of cellular interactions enriched in the E50K groups is illustrated in Venn diagrams (Fig. 8A). Interestingly, 14 pairs of ligand-receptor pairs were specifically associated with E12 and E18, among which the TNF-NOTCH1 ligand-receptor pair was significantly enriched (Fig. 8B). Furthermore, the expression of TNF and the corresponding ligand NOTCH1 was also high in disease-associated clusters of microglia (C3) and Müller cells (C0) (Fig. 8C). Therefore, we tested whether the TNF-NOTCH1 pathway mediated the crosstalk of Müller cells and microglia under the E50K mutation. Lentiviruses have been utilized to successfully deliver the E50K mutation in microglia and Müller cells to establish an in vitro model of NTG; thus, we confirmed our results via cell experiments in BV2 and RMC-1 cells with a Transwell coculture system. The results revealed that the mRNA levels of TNF-α, CCL2, CCL3, CCL4, IL-1α and IL-1β in E50K BV2 cells and the protein level of NOTCH1 in E50K RMC-1 cells were increased after coculture compared with those in WT BV2 cells and RMC-1 cells after coculture (Fig. 8D-F). Similarly, E50K BV2 cells transformed into a reactive state, and the morphology of E50K RMC-1 cells also changed (Fig. 8G and H). These findings suggest that the interaction between Müller cells and microglia tends to augment retinal inflammation damage via positive feedback. E50K mutation affected cellular communication between microglia and Müller cells during NTG development. (A) Venn diagram representing the number of ligand-receptor pairs between microglia and Müller cells in three E50K groups. Dot plot representing the overlapped ligand-receptor interactions between microglia and Müller cells (B) as well as C3 of microglia and C0 of Müller cells (C) in E12 and E18 expect E3 group respectively. (D) The mRNA levels of TNF-α, CCL2, CCL3, CCL4, IL-1α and IL-1β were determined by real-time qPCR in E50K BV2 cells (E50K), which were co-cultured with E50K RMC-1 cells, and WT BV2 cells (Con) co-cultured with WT RMC-1 cells. (E and F) Representative Western blot images and quantification of Notch1 in E50K RMC-1 cells (E50K) and WT RMC-1 cells (Con), which were both co-cultured with BV2 cells under matched condition. (G) The Con and E50K BV2 cell groups in (D) were stained against Iba-1 (red). Cell nuclei were visualized with DAPI (blue). (H) The Con and E50K RMC-1 cell groups in (E) were stained against CRALBP (red). Cell nuclei were visualized with DAPI (blue). Data shown as mean ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001. Data were analyzed using Student’s t-test (D and F) Discussion Among the gene mutations associated with glaucoma, the OPTN E50K mutation is well known to cause NTG, which directly affects RGCs under normal IOP and results in a particularly severe degenerative phenotype in patients; however, the pathological mechanism of this condition remains unclear [4, 23]. The retina is a well-organized structure that includes a heterogeneous mix of well-defined cell types. Thus, previous studies on glaucomatous optic nerve injury lacked a comprehensive analysis pertaining to each specific type of retinal cell. To address this research gap, in the present study, scRNA-seq was performed to elucidate the transcriptomic profiles, biochemical pathway alterations and cellular interactions in RGCs and glial cells during NTG development, providing evidence for molecular signatures as well as promising early treatments and diagnostic targets in NTG. Aging is an important risk factor for glaucoma, which has also been confirmed by degenerative damage in the retinas of old E50K mice retina [7, 8]. However, recent studies have been limited in elucidating the molecular changes associated with progressive visual impairment in NTG. Therefore, we conducted scRNA-seq on the retinas of E50K mice at different ages to map the molecular alterations of each type of cell in a time-dependent manner, and we added 12-month-old retinas as a check point for their tendency toward glaucomatous phenomena in terms of RGC density and visual function (Supplemental Fig. 1). The results revealed that DEGs in RGCs in the early stage of NTG were enriched mainly in pathways involved in metabolism, autophagy and axon regeneration, which have been verified to play important roles in glaucoma pathogenesis [24–26], suggesting that energy metabolism deficiency and autophagy dysfunction in RGCs might be early events in the mechanism of NTG. These findings are in agreement with the neuroprotective effects reported in recent studies that focused on mTOR, PI3K-Akt and HIF-1 signaling to attenuate RGC death [24–26]. Unfortunately, the number of RGCs is small in the retina under physical conditions, and owing to the characteristics of NTG, the number of RGCs is severely decreased in 18-month-old E50K retinas, which limits scRNA-seq analysis, and a better strategy is needed. Glial cells, which outnumber neurons by at least 10 to 1 in the retina, closely interact with RGCs [20]; thus, morphological and functional changes in retinal glial cells crucially affect RGC health. Microglia are sensitive to various insults in the retina, act as antigen-presenting cells, participate in the immunologic response and phagocytose apoptotic cells to maintain tissue homeostasis. In the normal retina, microglia are in a resting state, reside in the inner plexiform layer (IPL) and outer plexiform layer (OPL), and are characterized by tiny cell bodies with elongated ramified processes. Under pathological conditions, microglia transform into a reactive phenotype and are able to proliferate and move to injury sites, displaying rounder and larger cell bodies and pseudopodia [10, 11], which are also found in the old E50K retina (Fig. 3A and B). Accumulating evidence suggests that inflammation is a vital player in neurodegenerative disease and that microglial activation may even be the first event in glaucoma [27, 28]. Our results revealed that cellular communication between microglia and RGCs in NTG was altered at the 12-month early stage, possibly through CD74-MIF pathway. The cluster of microglia with specific expression of CD74 (C3) was confirmed to have functional implications in inflammatory and immune responses. The expression of CD74 in microglia and MIF in RGCs was significantly increased in 12-month-old E50K mice. Similarly, microglia in NTG, similar to those in healthy controls at a young age, initially exhibited enlarged cell bodies with radial processes and oriented toward the GCL, with increased cell numbers and TNF-α expression at 12 months (Fig. 3), when RGC density and visual function slightly began to change, indicating the transition of microglia to a reactive state at the early stage of the disease. A 18 months, microglia with high CD74 expression in NTG exhibited a typical amoeboid shape with increased proliferation and inflammatory factor release, corresponding with severe RGC loss and visual dysfunction. In contrast, during the physiological aging process, microglia were activated later and only slightly increased in number and TNF-α expression in the WT retina. These results indicate that reactive microglia are positively correlated with neurodegenerative severity during NTG progression. This disease-associated activation of microglia has also been observed in other glaucoma models [29, 30]. MIF, a cytokine expressed in neurons, reportedly acts through the CD74 receptor in highly activated microglia and triggers inflammatory responses in the brain and retina [17]. The CD74-MIF pathway targets various downstream effectors, such as ERK, MAPK, and PI3K-AKT signaling and apoptotic processes, and contributes to the regulation of cellular survival, proliferation and differentiation [13, 31], which are also enriched in our E50K-mutant RGCs and C3 microglia. Furthermore, C3 microglia also expressed high levels of APOE and TNF, which are specifically expressed in glaucoma and AD-associated microglia [32, 33]. These findings suggest that CD74 could be an important marker for reactive microglia in NTG. We postulate that aging-driven dysfunction of E50K mutant RGCs causes increased expression of MIF, which leads to microglial recruitment and activation via the CD74 receptor, triggering an inflammatory cascade through increased levels of TNF-α and other inflammatory factors and resulting in RGC death in a positive feedback manner. Although the possible pathological process underlying RGC damage was determined by scRNA-seq analysis, further experiments are necessary in our future studies. Macroglia play important roles in maintaining homeostatic conditions for neurons in the central nervous system. Diseased macroglia alter interactions with neurons and cause neuron degeneration through neurotoxic effects and inadequate survival support [20, 34]. Müller cells constitute the major component of macroglia in the retina of most mammals and account for 90% of total glial cells. This cell, in which the soma is localized in the INL and its processes extend to the inner and external limiting membrane, is the only cell type spanning all retinal layers and widely contacts neighboring neurons and vessels. Therefore, Müller cells contribute to monitoring the homeostasis of the retinal microenvironment and supply trophic factors to RGCs. This monitoring and support function is often compromised in disease states such as glaucoma. However, the precise mechanisms underlying these changes remain unclear, which encouraged us to investigate them in greater depth. Several previous studies did not clearly emphasize the difference between Müller cells and astrocytes in experiments in which the same marker was chosen to label both types of cells [35, 36]. Our scRNA-seq analysis and retinal immunostaining results revealed that GFAP and CRALBP are suitable indicators of astrocytes and Müller cells, respectively, and demonstrated the transition to a reactive state by both kinds of macroglia in NTG optic nerve injury (Fig. 6), which is in agreement with recent findings on glaucoma [3, 22]. This discrepancy may partly explain some conflicting reports about Müller cells and astrocytes in glaucoma [37]. Müller cells and microglia are two important types of neuroglia, and their responses to pathological stimulation in glaucoma have been well studied; however, the intricate relationships between these two kinds of cells are still not clear and have recently been of great concern recently. Müller cell activation has been reported to initiate microglial reactivity in glaucoma and trigger the release of excessive cytotoxic factors in return, whereas in some glaucoma models, microglial activation is initiated earlier than that of Müller cell [11, 27, 38, 39]. These results indicate that whether Müller cells or microglial are activated earlier depends on the type of retinal injuries. In NTG, microglia were activated and induced neuroinflammation at 12 months of age, earlier than Müller cell gliosis, which was obviously reactive at 18 months of age. In addition, the CellphoneDB analysis and coculture experiments revealed the potential mechanism of the TNF-NOTCH1 signaling axis in microglia and Müller cells, including their functional clusters. Notch1 signaling plays an important role in controlling activated astrocyte proliferation after brain injury; therefore, inhibition of Notch1 signaling has been reported to promote astrocyte-derived neurogenesis [40–42]. Müller cells, specialized astroglia in the retina that share some characteristics with retinal stem cells, have been reported to regenerate retinal neurons after injury in some species but limit the spontaneous regenerative response in the mammalian retina [43, 44]. Therefore, recent studies have revealed the underlying mechanism of the reprogramming of mature Müller cells, tipping the balance from the gliotic response to the reparative state after injury. Inducing the glial cell redifferentiation into neuron-like cells could be one of the most promising endogenous repair mechanisms in neurodegenerative diseases. Our findings indicated that in NTG, microglia initially react as they age, subsequently increasing TNF-α expression and leading to Müller cell gliosis via Notch1 signaling, which affects microglial activation in turn. The regulatory role of Notch1 signaling in the Müller cell neurogenic program is worthy of in-depth investigations in our future studies. Although our findings provide a detailed analysis of the pathological mechanism of RGCs and retinal glial cells during NTG progression, there are still some limitations in this study. First, this study focused primarily on RGCs, microglia and Müller cells, while the transcriptomic changes and cellular interactions of other retinal cell types, such as photoreceptors and endothelial cells, during NTG progression require further investigation. Second, although we confirmed that CD74 is an important marker for reactive microglia in NTG, further experiments are necessary in the future. Finally, the regulatory role of Notch1 signaling in the Müller cell response to NTG has not been fully explored. In vitro and in vivo studies are needed to further elucidate the precise molecular mechanisms involved and validate the promising reparative effect on glaucoma. Therefore, future studies should focus on these limitations and provide new viable strategies for NTG treatment. Conclusion In summary, this study applied scRNA-seq analysis of retinal cells from E50K mice at different ages and characterized the molecular changes in and interactions between RGCs and glial cells during NTG progression. Furthermore, microglia interact with RGCs via CD74-MIF signaling and enter a reactive state at an early stage, triggering an inflammatory cascade. In addition, the TNF-NOTCH1 pathway- mediated crosstalk between Müller cells and microglia in NTG tended to augment retinal inflammation damage. These findings improve the understanding of the mechanism of glaucomatous optic nerve injury and provide promising therapeutic targets for retinal degeneration. Supplementary Information Below is the link to the electronic supplementary material. (PDF 275 KB) (PDF 144 KB) (DOCX 93 KB) Acknowledgements This work was supported by the National Natural Science Foundation of China (82070956 to H-PY, 81970799 to Z-BS, 82000889 to W-LS); Heilongjiang Postdoctoral Scientific Research Developmental Fund (LBH-Z21184 to X-NL); China Postdoctoral Science Foundation (2021M701022 to X-NL); Longjiang Excellent Master's and Doctoral Theses of New Era (LJYXL2022-014 to X-NL); Key Scientific Research Projects of Heilongjiang Provincial Science and Technology Department (2022ZX06C14 to W-LS). Author Contributions X-NL, DZ and M-XD contributed to the design of the study, analysis of the data, completion of experiments and wrote the manuscript; Z-BS contributed to the design and writing; X-KX and W-LS assisted to sample collection and immunostaining; S-SC contributed with immunostaining and behavior tests; X-HL, L-YQ and F-YG assisted to the literature search, analysis and manuscript preparation; H-PY contributed to conception and design, final approval of manuscript and funding acquisition. Data Availability The data that support the findings of this study are available within the article and its supplementary materials. Raw data are available from the corresponding author upon reasonable request. Declarations Competing Interest The authors declare no competing interests. Footnotes Xinna Liu, Di Zhang and Mengxian Du contributed equally to this work. Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Supplementary Materials (PDF 275 KB) (PDF 144 KB) (DOCX 93 KB) Data Availability Statement The data that support the findings of this study are available within the article and its supplementary materials. Raw data are available from the corresponding author upon reasonable request.