Metabolomic Profiling of Aqueous Humor From Glaucoma Patients Identifies Metabolites With Anti-Inflammatory and Neuroprotective Potential in Mice

Abstract

Purpose: Metabolomic profiling of aqueous humor from primary open-angle glaucoma (POAG) patients using targeted metabolomics analysis and assessment of the potential anti-neuroinflammatory and neuroprotective roles of key dysregulated metabolites in a mouse model of retinal neuroinflammation.

Methods: A targeted metabolomics was performed on aqueous humor from POAG patients (n = 19) and healthy subjects (n = 10) via LC-MS/MS. In vitro neuroprotection studies were performed using a mouse cone photoreceptor cell line (661W) exposed to oxidative stress. For in vivo therapeutic studies, a few key dysregulated metabolites were delivered either topically via extracellular vesicle (EV)-mediated delivery or intravitreally into a C57BL/6 mouse model of retinal neuroinflammation. The neuroprotective and anti-neuroinflammatory properties were determined in the presence and absence of metabolites through pattern electroretinography, TUNEL, and quantitative PCR analyses.

Results: Among the 135 endogenous metabolites identified, 31 metabolites showed significant dysregulation in POAG. Metabolite set enrichment analysis revealed that these altered metabolites were associated with dysregulation of multiple key cellular pathways, including glycolysis, pentose phosphate pathway, short-/long-chain fatty acid metabolism, mitochondrial β-oxidation, and electron transport chain under glaucomatous conditions. Among these differentially expressed metabolites, a putative neuromodulator (agmatine) and a vitamin (thiamine) significantly decreased in POAG patients. Intravitreal or EV-mediated topical delivery of agmatine and thiamine significantly reduced the inflammatory response and protected retinal ganglion cell function against neuroinflammatory damage in the mouse retina. Agmatine and thiamine treatment also significantly protected photoreceptor cells from oxidative stress-induced cell death and attenuated the inflammatory cytokine response.

Conclusions: Our results revealed significant metabolic alterations in POAG that affect key cellular functions. Agmatine and thiamine could be potential immunomodulatory or neuroprotective drugs to treat or prevent neuroinflammatory damage to the retina during glaucoma.

Figure 1.

Metabolic profiling of aqueous humor distinguishes healthy controls from patients with POAG. (A) Spearman's correlation heatmap of pairwise sample comparisons within the aqueous humor metabolomes of healthy controls (n = 10, C1–C10) and POAG patients (n = 19, G1–G19). The color gradient reflects the strength and direction of the correlations. (B) PCA and PLS-DA plots demonstrate the separation of the control (CT) and POAG (GL) groups on the basis of their distinct metabolic profiles.

Figure 2.

Hierarchical clustering and random forest analyses of metabolites. (A) Heatmap displaying hierarchical clustering of the top 25 metabolites, ranked by the variable importance in projection (VIP) score, in the control (CT, red) and POAG (GL, green) samples. Clustering was performed using Euclidean distance and Ward's linkage algorithm. The x-axis represents samples, and the y-axis represents metabolites. The red/blue color gradient indicates metabolite levels across samples. The respective VIP scores are displayed in parentheses after the metabolite labels. (B) The random forest classification model emphasizes the significance of the top 15 DEMs, ranked according to their impact on classification accuracy.

Figure 3.

Glaucoma-induced dysregulation of amino acids, vitamins, and nucleic acid metabolites. (A) Volcano plot illustrating differentially abundant metabolites in glaucoma patients compared with healthy controls. The x-axis represents the log₂FC, and the y-axis represents the −log₁₀(P value) from the Student's t-test. Red circles highlight metabolites exceeding the FC threshold of 1.5 (log₂FC ≥ 0.5) and a P value threshold of 0.05 (−log₁₀[P value] ≥ 1.3). (B) Box plots showing the relative abundance of significantly altered metabolites (FC > 2; P < 0.05, Student's t-test) in the control (green) and glaucoma (red) samples. Boxes represent the interquartile range, and the line within each box indicates the median value. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Figure 4.

Dysregulated metabolites in glaucoma impact diverse essential metabolic pathways. (A) MSEA effectively illustrates the modulation of multiple pathways influenced by these DEMs, indicating the broad-reaching consequences of metabolite dysregulation in glaucoma. The size of the nodes represents the degree of influence on the relevant pathway, whereas the node color corresponds to the P value derived from a pathway enrichment analysis. (B) Interaction network of the pathways identified via MSEA analysis.

Figure 5.

Agmatine and thiamine treatment attenuates the TLR-induced neuroinflammatory response and protects neuronal cells from oxidative stress-induced cell death. (A) Mouse cone photoreceptor cells (661W) were challenged with the TLR agonists Pam3 (TLR2), poly I:C (TLR3), and LPS (TLR4) (100 ng/mL) in the presence or absence of agmatine or thiamine (100 ng/mL) for 24 hours. The cells were harvested and subjected to qPCR to measure the transcript levels of the inflammatory cytokines TNF-α, IL-1β, and IL-6. (B) To assess the neuroprotective effects of agmatine and thiamine, 661W cells were challenged with H₂O₂ (100 µM) in the presence or absence of agmatine or thiamine (100 ng/mL) for 24 hours. The cells were fixed and subjected to TUNEL staining (green, TUNEL-positive cells; blue, DAPI nuclear stain). *P < 0.05, **P < 0.005, ***P < 0.0005, ****P < 0.0001 (one-way ANOVA).

Figure 6.

Agmatine and thiamine provide neuroprotection in a murine model of retinal neuroinflammation. C57BL/6 WT mice (n = 6) were intravitreally injected with Pam 3, poly I:C, or LPS (0.1 µg/eye) to induce retinal inflammation. After 6 hours, the eyes were treated with either agmatine or thiamine via intravitreal injection. (A) RGC function was assessed via pERG. The bar graph represents the pERG (P1–N2) amplitude of treated and untreated mice. (B) The neural retina was harvested and subjected to qPCR to measure the transcript levels of the inflammatory cytokines TNF-α, IL-1β, and IL-6. *P < 0.05, **P < 0.005, ***P < 0.0005, ****P < 0.0001 (one-way ANOVA).

Figure 7.

EV-mediated topical delivery of agmatine mitigates retinal neuroinflammation. (A–C) EVs were purified from HTMCs. EV size (A), shape (scale bar = 100 nm) (B), and enrichment for EV markers (CD63⁺) (C) were assessed via NanoTracker, transmission electron microscopy, and immunoblotting analysis, respectively. (D) Agmatine loading was optimized using room temperature or heat shock incubation methods. (E) C57BL/6 mice were intravitreally injected with LPS to induce retinal inflammation. Six hours after LPS injection, agmatine-loaded EVs were applied as eye drops once a day. In another group of mice, 6 hours after LPS injection agmatine was injected intravitreally. After 48 hours of agmatine treatment, the mouse neural retina was harvested and subjected to qPCR to measure the transcript levels of the inflammatory cytokines TNF-α, IL-1β, and IL-6. *P < 0.05, **P < 0.005, ***P < 0.0005, and ****P < 0.0001 (one-way ANOVA).