Proteomic Comparison of Aged and OPTN E50K Retina in The Development of Normal Tension Glaucoma

Background: Progressive degeneration of retinal ganglion cells (RGCs) is a major characteristic of glaucoma, whose underlying mechanisms are still largely unknown. An E50K mutation in the Optineurin (OPTN) gene is a leading cause of normal tension glaucoma (NTG), directly affecting RGCs without high intraocular pressure and causing severe glaucomatous symptoms in clinical settings. A systematic analysis of the NTG mouse model is crucial for better understanding of the underlying pathological mechanisms for glaucoma. Methods: To elucidate proteomic and biochemical pathway alterations during NTG development, we established an OPTN E50K mutant mouse model through CRISPR/Cas9. Retinal proteins from resulting mice exhibiting glaucomatous phenotypes were subject to tandem mass tag-labelled quantitative proteomics. Retinal proteomes were analyzed through bioinformatics methods to characterize the molecular and functional signatures of NTG. Results: We identied 6364 quantitative proteins in our proteomic analysis. Bioinformatics analysis revealed that OPTN E50K mice experienced protein synthesis dysregulation, age-dependent energy defects and autophagy-lysosome pathway dysfunction. Certain biological features, including amyloid deposition, RNA splicing, microglia activation and reduction of crystallin production, were similar to Alzheimer’s Disease (AD). Conclusions: Our study is the rst to describe proteomic and biochemical pathway alterations in NTG pathogenesis during disease advancement. Several proteomic signatures overlapped with retinal changes found in the AD mice model, suggesting the presence of common mechanisms between age-related degenerative disorders, as well as prospective new targets for diagnostic and therapeutic strategies.


Introduction
Glaucoma is the leading worldwide cause of irreversible vision loss, with increasing prevalence [1]. It is characterized by the progressive degeneration of retinal ganglion cells (RGCs), resulting in visual impairment and eventual blindness. Reducing intraocular pressure (IOP) is the most practiced therapy for glaucoma in the clinic, but it is unable to reverse visual impairment and prevent progression to blindness, as well as being insu cient for normal tension glaucoma (NTG) patients, whose IOP is always within the normal range. This observation suggests that glaucoma may develop from a complex interaction of multiple factors, aside from high IOP levels alone. Thus, it is critical to investigate molecular changes and speci c biomarkers for early diagnosis and new treatment targets. The E50K mutation of the optineurin (OPTN) gene is known to cause NTG, along with yielding particularly severe clinical glaucomatous symptoms [2][3][4]. Several E50K-mutant cell models have been applied [5][6][7], but in vivo studies are still rare. Therefore, we used CRISPR/Cas9 gene-editing technology to introduce the E50K mutation into mice to fully mimic NTG pathologies in vivo, followed by tandem mass tag (TMT) quantitative proteomics to perform systematic analysis of the biochemical process within both young-and old-age retinas.
Bioinformatics analysis were used to understand molecular neuropathology and potential hallmarks of NTG.
Increasing numbers of studies have indicated a possible relationship between glaucoma and Alzheimer's Disease (AD) [8], both of which are age-related neurodegenerative diseases with some common biological features. For instance, aggregates of OPTN was also observed within neuro brillary tangles and dystrophic neurites in AD [9], though the exact mechanism behind their occurrence is still obscure. Thus, we also compared the proteomic results for those two diseases to better understand the mechanism underlying retinal degeneration and provide new insights for neurodegenerative diseases.

Experimental design
The aim of this study was to investigate the molecular and functional characteristics of the NTG mouse model with the OPTN E50K mutation. Protein expression were assessed by quantitative proteomics, combined with liquid chromatography-tandem mass spectrometry (LC-MS/MS) and TMT labeling.
Seventy-two retinal samples in total were used, comprising biological triplicates of 2 genotypes (OPTN E50K and wild type) at 2 age-points (3-and 18-months old). They were randomly divided into three 4-plex TMT experimental sets and then analyzed via MS (Fig. 1a). Among 7019 identi ed proteins, 6364 proteins in all retinal samples were considered as satisfactory for subsequent quantitative and bioinformatics analyses (Fig. 1b).  Fig. 1). Mice with OPTN E50K mutation at 3-and 18-months were designated as EY and EO, respectively, while the age-matched WT control groups were correspondingly de ned as WY and WO.

Animal models and their visual conditions
Before mice were sacri ced for the proteomics analysis, several measuring procedures were conducted to verify their visual conditions [10]. Brie y, IOP was monitored using an Icare Tonometer (Finland) (Supplemental Fig. 2a), and visual function was assessed by a light/dark test (Supplemental Fig. 2b-c). Hematoxylin & eosin staining was used to assess retinal morphology (Supplemental Fig. 2d-f). Western blot analysis and immunostaining both con rmed that OPTN protein dramatically decreased in mutant mice, compared to their age-matched WT counterparts (Supplemental Fig. 2g-

Bioinformatics analyses
Gene set enrichment analysis (GSEA), based on quanti able proteins, was performed to elucidate the integrated rankings and differences of pathways between groups [11]. Values were derived by permuting the proteins sets 1000 times within a core pathway extracted from Gene Ontology (GO), Kyoto Encyclopedia of Genes and Genomes (KEGG), and cell type signatures. Normalized enrichment score (NES) was the primary statistic for examining gene set enrichment results, and its statistical signi cance was estimated by the nominal P-value. Signi cantly-enriched pathways were selected at a nominal Pvalue level of 0.05 (NOM P-val) and FDR q-value of 0.25 (FDR q-val).
Differentially expressed proteins (DEPs), selected by meeting the signi cance criteria on a two-sample ttest P-value (P < 0.05) and fold change threshold (≥ 1.20 for up-regulation or ≤ 0.833 for downregulation), were used for further analysis between two comparative groups. DAVID Bioinformatics (version 6.8) [12] were used to classify the GO, especially the molecular function (MF) of the proteins. MF was evaluated by Fisher's exact test to obtain P-values, and their values also ranked from low to high. Canonical pathways, diseases and biological functions, as well as upstream regulators and networks were enriched using Ingenuity Pathway Analysis (IPA, QIAGEN) [13]. Canonical pathway analysis allowed identi cation of function-speci c genes through assessing the P-values obtained from Fisher's exact test.
Analytical algorithms of IPA were carried out on DEPs to predict putative upstream regulators, such as kinases and transcription regulators, as well as downstream effects on known biological pathways. They also assigned the magnitude of activation states as a Z-score (z > 2 indicates activation, while z<-2 represents inhibition) for pathways or regulators, based on the quantitative values of involved proteins. Interactions between proteins were illustrated by networks based on Ingenuity Pathway Knowledge Base, with a cut-off score of 20 (Supplemental Fig. 3).

Quantitative real-time PCR (qPCR) analysis
Total RNA was extracted from retinas with Trizol, then 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 GAPDH using the 2 ΔΔct method.

Western blot analysis
Protein samples were diluted in loading buffer and heated at 95°C for 10 min. Equal amounts of protein were electrophoresed on 12% or 15% SDS-polyacrylamide gels, then transferred onto polyvinylidene uoride membranes. Membranes were then blocked with 5% skim milk solution for 1 hour at room temperature, followed by incubation with primary antibodies against OPTN (Proteintech, 1:1000), LC3B (Cell Signalling Technology, 1:1000), and P62 (Proteintech, 1:1000) overnight at 4°C. After three washes, the membranes were incubated with horseradish peroxidase-conjugated secondary antibody (1:10000, Zhongshan) for 1 hour at room temperature. Protein bands were visualized by enhanced chemiluminescence (ECL, Biosharp), and quanti ed with ImageJ.

Immuno uorescent staining
Mice were anesthetized and perfused with 4% paraformaldehyde (PFA). Eye tissues were xed with 4% PFA overnight at 4°C, placed in a 25% sucrose gradient for dehydration, and then frozen. 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 hour, then subsequently incubated in primary antibodies overnight at 4°C. To examine microglia and astrocytes, anti-Iba-1 (1:50, Abcam) and anti-GFAP (1:50, Proteintech) antibodies were used, respectively. Tissue sections were incubated with Alexa488 goat anti-rabbit secondary antibody for 1 hour at room temperature, followed by DAPI for 5 minutes to counterstain nuclei. Images were taken using a uorescence microscope and quanti ed with Image-Pro software.

Statistics
All values were presented as mean ± SEM. Statistical analysis was conducted using Prism version 7.0 (GraphPad Software). Data were analyzed by Student's t-test for two-group comparisons, or one-way analysis of variance (ANOVA) followed by Tukey's post-hoc test for comparisons among 3 or more groups. P < 0.05 was considered statistically signi cant.

Results
The role of the aging process in NTG pathogenesis Aging is the primary risk factor for neurodegenerative diseases, including glaucoma [14,15]. To investigate the role of senescence in NTG development, we performed TMT-labelled LC-MS/MS proteomics analysis on OPTN E50K and WT control mice retinas, at both young and old ages. This analysis identi ed 7019 proteins, of which 6364 were quanti able from all the retinal tissues used (Fig. 1b). GSEA was performed based on those quanti ed proteins from all the experimental sets. Agerelated biological pathways were identi ed based on whether the protein expression for a prior-de ned set of genes was statistically enriched in old retinas or depleted in young controls under GSEA, and ltered by NOM P-val < 0.05 and FDR < 25%. The number of KEGG and biological process (BP) gene sets enriched in EO and WO groups were illustrated by the Venn diagram plots in Figs. 1c and 1d, respectively. For KEGG, 7 sets were EO-speci c, 25 WO-speci c, and 24 was found for both groups, while 11 was EO-speci c, 128 WO-speci c, and 31 was present in both categories for BP ( Fig. 1c-d). These ndings overall suggested similar aging effects being present in both pathological and healthy conditions for aged mice. The distribution pattern of these enriched proteins within EO and WO groups was assessed through volcano plots with log-transformed protein abundance ratios. The results showed that 134 (77-up, 57-down) DEPs (P < 0.05, FC ≥ 1.20 or ≤ 0.833) were found in EO (Fig. 1e), and 853 DEPs (519-up, 334-down) in WO mice (Fig. 1f). Among those DEPs, 84 were found within both EO and WO groups (Fig. 1g).
Molecular functions for these proteins were then analyzed by the DAVID database (Fig. 2a), where the structural constituent of eye lens category had the most signi cant P-value and largest number of DEPs, mostly composed of crystallin proteins. It is also worth noting that the crystallin family is also involved in binding unfolded proteins as molecular chaperones. The distribution pattern of different crystallin family members within triplicates from each of the 4 groups was assessed through a heat map (Fig. 2b), which suggested that crystallins were dramatically down-regulated with age, especially for the EO group. This downregulation was also supported by decreased mRNA levels for Crybb3, Cryba2, Cryab, Crybb2, and Crybb1 in accordance with the decreased protein abundance (Figs. 2c-g). As it has been documented that decreased crystallin levels can reduce its neuroprotective effects, rendering retinal neurons more vulnerable to cytotoxic injury and cell death [16], it is therefore worthy to note that a signi cant crystallin reduction also occurs in EY mice. Owing to this crystallin downregulation among both EO and EY groups, we investigated the 7 KEGG and 11 BP pathways found to be EO-speci c and ranked them based on their NES values (Supplemental Table 1). The most signi cantly upregulated pathways oxidative phosphorylation (OXPHOS), peroxisome and regulation of actin cytoskeleton, as found among the KEGG gene sets, while the most signi cantly downregulated pathways, as determined by BP, were sensory perception, cell cycle DNA replication and regulation of proteasomal ubiquitin dependent protein catabolic process in EO retinas. These results indicated that aside from the aging process, speci c responses from NTG mice are present for the development of the pathological phenotype.
Comparative analysis of the E50K mutation-dependent effect on the retina We performed GSEA for protein enrichment analysis on biological processes in EO and EY mice, compared with their respective WO and WY controls to gure out the E50K mutation-speci c effects.
Twenty-nine BP gene sets were found to be speci cally associated with the E50K mutation, no matter the age of the mice (Fig. 3a), which were mainly divided into 2 categories, comprising mRNA splicing (Fig. 3b) and transmembrane transport (Fig. 3c). With respect to mRNA splicing, the E50K mutation is suggested to speci cally affect RNA processing via spliceosome, while for transmembrane transport, organic acid, drug, and anion transport are affected. Strikingly, 8 of those 29 BP gene sets, all associated with genetic information processing, demonstrated opposing trends between EY and EO groups, possibly indicating the occurrence of cross-talk between the aging process and the E50K mutation (Fig. 3d). This possibility is supported by the nding that much more BP gene sets are affected in EO versus EY groups (155 in EO versus 10 in EY), suggesting a synergistic effect between the E50K mutation and aging.
Young OPTN-E50K mutants have similar protein expression pro les as old WT mice To determine whether EY mice have a similar pattern of protein expression as WO mice, we analyzed BP alterations in EY versus WY, as well as WO versus WY retinas. Indeed, 16 processes were found to demonstrate the same expression trend in both EY and WO groups (Supplemental Table 2). Since we also found similar mRNA and protein expression levels between EY and WO retinas for the crystallin family ( Fig. 2b-g), we tried to investigate the underlying relationship between these two conditions. EY retina had 228 up-regulated and 442 down-regulated DEPs present (Fig. 4a). Interestingly, 185 proteins were also identi ed in WO retina (Fig. 4b) and the majority of them (137 proteins) was downregulated under both conditions. Closer examination of these proteins found that they were classi ed as ribosomal, crystallin family, or histone proteins. The distribution pattern for triplicates belonging to WY, EY, or WO groups was displayed in a heat map (Fig. 4c). To better understand their underlying relationships, IPA analysis were performed based on DEPs between EY and WO retinas, where the top 6 canonical pathways for EY were EIF2 signaling (p = 2.51E-35), regulation of eIF4 and p70S6K signaling (p = 6.31E-16), mammalian target of rapamycin (mTOR) signaling (p = 3.47E-10), coronavirus pathogenesis pathway (p = 1.58E-12), sirtuin signaling pathway (p = 9.33E-07), and glutathione-mediated detoxi cation (p = 1.41E-06). EIF2 signaling (p = 3.16E-33), regulation of eIF4 and p70S6K signaling (p = 3.98E-14), and mTOR signaling (p = 1.26E-12) were also among the top 6 canonical pathways for WO retinas, but the other 3 pathways were G beta gamma signaling (p = 3.16E-14), cardiac hypertrophy signaling (p = 6.31E-12), and CXCR4 signaling (p = 1.58E-11) (Fig. 4d). Notably, both EY and WO had the same top three pathways, which also extended to their associated proteins. For instance, both EY and WO shared 45 ribosomal subunit proteins whose expression all decreased under their associated reduced EIF2 signaling comparisons. Such inhibition was predicted by their z-scores (-6 in EY/WY, -3.042 in WO/WY). These similarities in protein abundance and pathway operation between EY and WO mice suggest that E50K mutation may advance age-related degenerative changes in retina, thereby yielding a cumulative impact on visual impairment in NTG.
Canonical pathways and protein networks involved in accelerating NTG pathogenesis in aged E50K mutant mice As the E50K mutation causes severe clinical symptoms at advanced ages, we tried to gain a more detailed understanding of NTG pathological changes in the context of aging. As 155 biological processes were found to be speci cally enriched in EO versus WO mice (Fig. 3a), we analyzed the presence of signi cantly modulated proteins between those 2 groups and identi ed 1564 DEPs (952 down-regulated, 612 up-regulated; Fig. 5a). A hierarchical clustered heat map of the DEPs re ected opposing protein abundance distributions between EO and WO groups among the three triplicates (Fig. 5b). Diseases and biological functions analysis (Fig. 5c), as well as canonical pathway (Fig. 5d) enrichment was analyzed by IPA on DEPs. The bubble chart (Fig. 5c) showed the top diseases and biological functions present in EO mice, where notably, degeneration of retinal cells was found to be the most enriched. The top canonical pathways induced by E50K mutation were mitochondrial dysfunction (p = 3.98E-18), OXPHOS (p = 1.26E-17), synaptogenesis signaling pathway (p = 2.04E-09), EIF2 Signaling (p = 6.03E-10), mTOR signaling (p = 2.04E-09), phototransduction pathway (p = 2.19E-08), phagosome maturation (p = 2.24E-07), protein ubiquitination pathway (p = 8.13E-05), AMP-activated protein kinase (AMPK) signaling (p = 8.51E-05), and apoptosis signaling (p = 3.02E-04) (Fig. 5d). Mitochondria provides energy for many intracellular biochemical processes via OXPHOS and comprises ve key enzyme complexes. In our dataset, 49 DEPs were associated with mitochondrial complexes, including 20 subunits of complex I (NADH dehydrogenase), 1 subunit each for complex II (succinate dehydrogenase) and III (cytochrome b-c1 complex), 9 subunits of complex IV (cytochrome-c oxidase) and 18 subunits of complex V (ATP synthase), respectively ( Fig. 5e-f).
Upstream pathway analysis of IPA is capable to predict the presence of molecules governing the regulation of a speci c cluster of DEPs. One such molecule, mTOR, was identi ed and predicted to be activated in EO mice (Supplemental Fig. 3a), based on its high z-score (3.622), statistically signi cant Pvalue (p = 4.11E-06), and mTOR signaling being enriched as one of the top canonical pathways. IPA analysis is also able to classify DEPs into different functional units found within networks. Interestingly, mTOR was integrated with mitochondrial complexes in the same interaction network associated with cellular development (Score = 23; Supplemental Fig. 3b). As a protein kinase, mTOR regulates important biological processes, including metabolism, protein synthesis and particularly autophagy, where it serves as a central regulator by modulating multiple aspects of this process, such as initiation, processing, and termination. Active mTOR inhibits autophagy induction, while OPTN serves as an autophagy receptor involved in autophagy induction. E50K mutation negatively affects OPTN expression, leading to the autophagy pathway being disrupted, and resulting in injured RGCs. Even though autophagy was not enriched as one of the top canonical pathways, we observed a number of autophagy-associated proteins with decreased expression in both EY and EO retinas comparing with age-matched controls, including Atg3 (0.72-fold) and Atg4b (0.82-fold) (Supplemental Table 3).
To verify mitochondria-related proteomic results and the predicted autophagy inhibition, we performed qPCR to determine the mRNA levels for selected mitochondrial gene candidates whose expression greatly changed in the EO retina, as well as Western blotting to measure expression of the well-known autophagy markers LC3 and P62 in all four groups. The downregulated mRNA levels for Ndufb8, Ndufb6, Ndufa13, Ndufb5, Cox4i2, Cox4i1 and Atp6v0c, along with upregulated levels for Ndufv3, Cox17 and Atp6v1f in the EO group were consistent with the abundances of their corresponding proteins (Fig. 6a-b). Western blot results also showed increased P62 levels, as well as LC3 II/I ratios, under E50K mutation, which was accentuated with increased age (Fig. 6c-f). Increases in those values are hallmarks of E50K-induced autophagy blockage [7,17]. Other key autophagy-associated proteins are SNAREs and RABs, both of which are involved in facilitating autophagosome and lysosome fusion. We found that the functions for Page 9/18 22 SNARE-associated proteins were signi cantly inhibited for lysosomes (NOM P-val = 2.94E-03, FDR qval = 3.37E-02) and SNARE interactions in vesicular transport (NOM P-val = 0.00E + 00, FDR q-val = 7.55E-04) under GSEA in EO mice (Supplemental Fig. 4 and Table 4). As for RABs, 27 proteins in the family were found to be downregulated in the EO group (Supplemental Table 5), while 7 proteins were also found to have reduced expression in the EY group, compared to their age-matched WT counterparts. Together, these results suggested that autophagy and its associated pathways were signi cantly disrupted in NTG retinas, even at younger ages.

NTG shares pathological phenotypes with Alzheimer's disease
We performed GSEA analysis to determine the speci c pathological pathways in EO mice involved with NTG development. Among these signi cantly enriched pathways, those associated with metabolism, as well as the neurodegenerative Alzheimer's (AD) and Parkinson's diseases (PD) were upregulated, while most genetic information processing and degradation-associated pathways were downregulated (Fig. 7a). Recent studies have reported amyloid-beta (Aβ) deposition, the hallmark of AD, also occurring in glaucoma retina, as well as evidence supporting an association between AD and glaucoma through sharing common pathological mechanisms, such as mitochondrial dysfunction [18][19][20], abnormal accumulation of misfolded proteins [21], glial activation, and in ammation [22][23][24]. In addition, recent proteomic research have investigated dysregulated mRNA splicing and decreased crystallin in both AD and glaucoma retinas [25][26][27] (Fig. 7b). OPTN has been observed within neuro brillary tangles and dystrophic neurites in AD, implying it may also be a contributory gene there [9]. Our proteomic results con rmed changes in mitochondria and autophagy of OPTN E50K retinas, as well as the spliceosomeassociated gene set being more enriched under GSEA for EO mice, especially for serine/arginine-rich (SR) proteins including Srsf6 (1.40-fold), Srsf4 (1.35-fold) and Sf3a1 (1.32-fold). The upregulation of these proteins in NTG were veri ed by qPCR results for their corresponding mRNAs (Figs. 7c-d).
Retinal sections were then stained with Iba1 and GFAP to con rm the NTG microglia-astrocyte-neuron crosstalk, in which both microglia and astrocytes were mostly located in the ganglion cell layer (GCL) (Fig. 8a). The number of both Iba1 + and GFAP + cells in the GCL increased with age, especially in EO mice (Fig. 8b-c). From this nding, we postulated that the microglia and astrocytes may also play major roles in neuroin ammatory conditions, such as AD, as both diseases demonstrated similar retinal signatures. This postulation was supported by GSEA showing enrichment of proteins associated with both cell types in EO compared to WO mice (Supplemental Fig. 5). We tried to determine whether NTG increases the likelihood of AD susceptibility by examining Aβ deposition through retinal immunostaining, where it was increased in E50K mutant mice compared to their age-matched WT counterparts (Fig. 8d-e). Another common risk factor between NTG and AD was the crystallin family, which was demonstrated under immunostaining for both Crybb3 and Cryab proteins (Fig. 8e, f-g). Lower expression levels were found for the EO group, compared to the WO group, for both proteins, while downregulation was only present for Cryab in EY mice compared to WY ones (Fig. 8f-g) [26,27]. These data revealed that NTG with OPTN E50K mutation is a widespread age-related neurodegenerative condition, which shares common pathological phenotypes with AD in terms of glial cell activation, reduced crystallin production, and βamyloid aggregation.

Discussion
OPTN E50K mutation is the most common mutation for NTG [2,3], causing severe symptoms in patients [4]. Its pathological mechanism is still obscure, and systematic analysis is rare. In the present study, CRISPR/Cas9 was used to produce OPTN E50K mice. Proteomic analysis was performed to elucidate proteomic and biochemical pathway alterations during NTG development with respect to mitochondrial function, autophagy, and protein synthesis. Comparative analysis between NTG and AD also provided evidences for shared retinal pathological manifestations, suggesting common pathogenesis between those disorders. CNS neurons are highly energy-demanding cells, dependent on mitochondrial OXPHOS [19]. Thus, mitochondrial dysfunction is a major cause of aging and neurodegenerative disease, including glaucoma [28]. In old E50K mice, 180 DEPs are identi ed as mitochondrial matrix and membrane components. Complex I, the largest multi-subunit complex, coupling NADH-ubiquinone electron transfer with the generation of a proton gradient for ATP synthesis, was the most severely affected in old E50K mice [29,30]. Damage to that complex is associated with ROS generation [29], triggering oxidative stress and gradual cell death, thereby partly explaining the previous nding of E50K mutation selectively inducing ROS-mediated RGC death [31]. Another factor behind this mechanism is complex IV disruptions, of which 16 DEPs were found in old OPTN E50K mice. RGCs, unlike other neurons, are directly exposed to light, thus cytochrome oxidase (COX) in complex IV is important to help absorb light and avoid retinal damage [32], which was lost with complex IV dysfunction. A key molecule in energy homeostasis is the AMPK complex, which senses low cellular ATP levels to trigger autophagy and inhibit protein synthesis in response to energetic stress and mitochondrial insults [33,34]. However, this capacity is attenuated under aging and pathological conditions, where severe mitochondrial damage disturbs AMPK signaling, causing apoptosis [33]. The nding of these associations is in agreement with our IPA results in old OPTN E50K mice indicating signi cant inhibition in OXPHOS and AMPK signaling.
Autophagy is a fundamental cellular process mediating cellular component degradation, and defects in this process plays an important role in neurodegenerative disorder pathogenesis [35,36]. mTOR is its central regulator, found in our IPA analysis to target various downstream effectors, such as PI3K/AKT signaling, mitochondrial protein synthesis, and mitochondrial-lysosomal homeostasis. Recently, mitochondria and lysosomes are considered mutually functional in maintaining homeostasis [37,38], where mitochondrial dysfunction leads to lysosomal impairment and autophagy by-product buildup, whereas lysosomal defects trigger functional and morphological mitochondrial defects. This is supported by our IPA analysis showing that mTOR signaling-associated canonical pathways of mitochondria function and phagosome maturation were both disrupted. This connection between mTOR, mitochondria, and lysosomes is facilitated by V-ATPase, which ensures low lysosomal pH to maintain lysosomal hydrolase activity and autophagic vesicle fusion [39,40]. Loss of V-ATPase subunits triggers accumulation of non-functional lysosomes and inhibition of autophagy, as shown in a previous study where downregulation of the Atp6v0c subunit is observed to promote accumulation of autophagosomes with high basal LC3 II levels, thereby adversely affecting autophagy-lysosome pathway (ALP) function [41]. This was consistent with our ndings showing Atp6v0c down-regulation, coupled with higher LC3 II levels, indicating autophagy dysfunction, in old OPTN E50K mice. Furthermore, our ndings indicating SNARE family inhibition and RAB family downregulation are in accordance with reduced autophagy activity being present in E50K mutants, as both protein families are associated with membrane fusion. We thus speculate that E50K mutant autophagy blockage stems from the impairment of multiple cellular pathways.
Our results showed the OPTN E50K mutation inducing several age-related proteomic features, even at younger ages. Similar DEP distribution was found between proteins in young OPTN E50K and old WT retinas, particularly those associated with EIF2, eIF4, p70S6K, and mTOR signaling, as well as those found in the ribosome, all of which are involved in protein synthesis and energy metabolism, suggesting a common energy stress response occurring in both retinal types. This association, particularly with ribosomal proteins, is supported by recent studies highlighting that ribosomal dysfunction may cause neuronal synaptic functional and plasticity impairment [42,43]. The similarities for DEPs are further supported by our GSEA analysis nding 16 biological processes, associated with transcription, translation and protein folding in endoplasmic reticulum, being affected in both young OPTN E50K and old WT retinas, with the same trends. Interestingly, 8 processes for young OPTN E50K mice were discovered to also be affected with old mutants, but with opposite trends. We postulate that aging-driven dysfunctions in transcriptional and protein homeostasis play an important role in old-age glaucoma, which is aggravated by the E50K mutation. This is consistent with clinical ndings showing that NTG patients with E50K mutation had early and severe visual dysfunction.
The retina responds similarly to the brain under pathological conditions, owing to it being an extension of the CNS [44,45]. Being the only place allowing direct visualization of neurons and blood vessels, early non-invasive detection and monitoring of cerebral changes can take place, resulting in greater interest in identifying ocular indicators for neurodegenerative diseases, such as AD [44,[46][47][48]. Various studies have proven AD ocular involvement, including RGC loss, retinal thinning, astrogliosis, in ammatory changes, and Aβ plaques [48,49]. Some of these processes, particularly RGC loss, GCL thinning, and Aβ presence, are also present in glaucoma [49,50]. A recent study also observed OPTN within AD neuro brillary tangles and dystrophic neurites [9], supporting our usage of the OPTN E50K mouse model to investigate underlying common pathologies between NTG and AD. We rst examined Aβ deposits within NTG retinas, and observed its age-related elevation and disease-speci c enhancement in OPTN E50K mice, even at early stages. This aggregation was also observed at 2-3 months in AD mice model retinas, prior to their brain appearance [51]. Based on this observation, it is likely that the OPTN E50K mutation may be connected to Aβ accumulation for both NTG and AD through its impairment of autophagy, which would otherwise remove Aβ peptides [52][53][54]. Another important pathologic process for both diseases is neuroin ammation triggered by astrocytes and microglia in response to protein aggregates and degenerated neurons [49,55]. Its occurrence was supported by our ndings where astrocytes and microglia were signi cantly activated in old OPTN E50K mice. Both AD and NTG share other pathological biochemical alterations, such as mitochondria, mRNA splicing and crystallin protein dysfunction [27,56]. In particular, alternative splicing dysregulation is a common characteristic for several nervous system age-related illnesses, and glaucoma is no exception, as our GSEA analysis showed that the spliceosome was signi cantly up-regulated in old OPTN E50K mice, involving 45 DEPs [57][58][59]. As for crystallin proteins, involved in eye development and protein folding, their levels were decreased in both NTG and AD mice [26,27,[60][61][62][63]. Indeed, a reduction in abundance was observed in previous studies for 7 such proteins in association with high IOP [26,64], as well as here for young OPTN E50K mice. In particular, 2 crystallin proteins within those 7, Crybb3 and Cryab, were also found to have signi cantly decreased in AD mice retina [27], further supporting ndings highlighting biomolecular similarities between glaucoma and AD. We also found this crystallin decrease to be associated with neuro bril formation and glial cell in ammation, consistent with previous studies [61][62][63]. Overall, the shared pathological alterations between NTG and AD provide promising targets for glaucoma treatment, as well as possible markers to diagnose CNS disease in patients.

Conclusion
Our study elucidated that the OPTN E50K mutation is associated with aggravating several age-related changes at the early stages of NTG, resulting in young mutants sharing similar phenotypes as old WT mice. These changes, such as with mitochondrial energy metabolism, autophagy, and mTOR signaling, were further accentuated in OPTN E50K retinas, suggesting glaucoma resulting from a complex interaction of multiple cellular processes. We also demonstrated the role of energy defects and ALP impairments via AMPK-mTOR pathway in NTG retinal degeneration, and further outlined the biochemical connections with the neurodegenerative disease AD, which may aid in nd certain retinal biomarkers for accessible and non-invasive CNS disease detection.