Reversing Mitochondrial Dysfunction in Optineurin E50K Glaucoma: A Metabolic Approach to Neuroprotection

Abstract

Mutations in optineurin (OPTN) are linked to neurodegenerative diseases such as normal tension glaucoma (NTG) and amyotrophic lateral sclerosis. The E50K-OPTN mutation is the most common genetic cause of NTG, where it disrupts mitophagy and leads to the accumulation of dysfunctional mitochondria. To understand how cellular metabolism is altered in these persistent mitochondria, and whether any pathological state can be reversed, we investigated NTG-patient-derived fibroblasts carrying the E50K-OPTN mutation. We identified a form of mitochondrial leak metabolism driven by elevated levels of the ATP synthase c-subunit leak channel (ACLC). These cells exhibit reversed F1FO ATP synthase activity, increased mitochondrial proton leak, and fragmented mitochondria, resulting in inefficient oxidative phosphorylation and a shift toward aerobic glycolysis and high protein synthesis rate. The ratio of ATP synthase c-subunit to β-subunit was markedly elevated, suggesting open ACLC pores. Treatment with dexpramipexole normalized ATP synthase function and cellular metabolism, promoted ATP synthesis rather than hydrolysis and reduced protein synthesis rates. Dexpramipexole reduced p62 levels in E50K fibroblasts, consistent with a reduced mitophagic burden from decreased accumulation of damaged mitochondrial cargo. These findings identify ACLC-mediated leak as a central driver of metabolic dysfunction in E50K-OPTN glaucoma and suggest ACLC closure as a viable therapeutic strategy.

Introduction

Mitochondrial dysfunction is increasingly recognized as a central driver of neurodegenerative diseases, including Parkinson’s disease, Alzheimer’s disease, and amyotrophic lateral sclerosis, where it contributes to energy failure, oxidative stress, and progressive neuronal loss^1–3. Despite this, few therapies directly target mitochondrial mechanisms. Primary open angle glaucoma (POAG), a neurodegenerative optic neuropathy and leading cause of irreversible blindness⁴, shares these pathological features⁵. Whereas raised eye pressure is the principal driver of progressive retinal ganglion cell (RGC) loss, loss occurs even at normal eye pressure, as seen in normal tension glaucoma (NTG), signposting the role of intrinsic metabolic vulnerability. Our recent work shows that mitochondrial respiratory function is a key determinant of glaucoma progression, independent of intraocular pressure (IOP), underscoring the centrality of bioenergetic failure⁶.

Mitochondria possess several previously described quality-control mechanisms to counteract dysfunction, including biogenesis, dynamics (fusion and fission), the mitochondrial unfolded protein response (UPR), and mitophagy^7,8. Among these, mitophagy is the most extensively characterised and has emerged as a tightly regulated pathway essential for mitochondrial turnover and neuronal health⁹.

The E50K mutation in Optineurin (OPTN) is the most commonly reported variant associated with NTG¹⁰, causing a severe form of familial NTG and is known to impair mitophagy via dysregulation of the PINK1-Parkin pathway¹¹. Dysfunctional or depolarized mitochondria accumulate PINK1 on their outer membrane, which recruits Parkin to selectively ubiquitinate them for mitophagy¹². Through asymmetric fission, dysfunctional mitochondrial parts are separated from healthy ones, enabling the targeted clearance of depolarized fragments and preservation of mitochondrial integrity¹³. However, in E50K cells, mitophagic flux is impaired, as evidenced by elevated LC3-II/I ratios and p62 accumulation, allowing dysfunctional mitochondria to persist¹⁰. Mitophagy is initiated by the loss of membrane potential or other mitochondrial dysfunction, such as mitochondrial UPR¹⁴. Recent studies have proposed that the ATP synthase c-subunit channel (ACLC), a latent ion-conductance pathway, may contribute to these early disturbances^15–18. The ACLC can mediate proton leak through the c-ring both within the intact F₁F₀ complex and when present as free c-rings in the inner mitochondrial membrane (IMM), either as disassembled c-rings released from F₁F₀, or as unassembled c-rings formed from newly synthesized c-subunits in contexts where c-subunit expression exceeds that of other ATP synthase components^15,17,19,20. Under specific physiological or stress conditions, such as transient rises in mitochondrial Ca²⁺ or changes in membrane voltage, the ACLC can open to permit a regulated proton leak, partially uncoupling respiration from ATP synthesis^{15,16,18,19,21}. Elevated matrix Ca²⁺ promotes the transition of ATP synthase from its dimeric to monomeric state and can bind to the F₁ sector, inducing conformational changes that increase the probability of c-ring opening²². This controlled conductance leads to mild depolarisation, which in turn can drive ATP synthase reversal, pumping protons out of the matrix to sustain the membrane potential. Under sustained depolarising conditions, however, the F₁ and F₀ sectors may dissociate, generating free c-rings that remain constitutively open¹⁵. These dissembled c-rings produce persistent proton leak, loss of potential, and a metabolic shift towards ATP hydrolysis and glycolytic compensation¹⁷. Under physiological conditions, increased ratio of ATP synthase c-subunit/ATP synthase drives a faster electron transport, tricarboxylic cycle (TCA), glycolysis and glucose uptake. This metabolism is good for developing or repairing cells because it increases the rate of generation of amino acids and lipids.^{15,17–20,23}. In uncontrolled proton leak under pathological conditions, however, ACLC has been shown drive large-conductance mitochondrial permeability transition (mPT) in neurons and other cells under stress^{15,17,18,23–25}.

We find here in fibroblasts derived from individuals with E50K optineurin mutation (E50K-OPTN cells) may engage in a similar metabolic adaptation, where controlled proton conductance brought on by extra free c-subunit rings causes mild depolarisation that drives ATP synthase reversal, allowing the enzyme to hydrolyse ATP to sustain mitochondrial membrane potential. By maintaining sufficient ΔΨm despite underlying dysfunction, we find that these mitochondria reduce the cues that normally trigger mitophagic recognition, enabling a metabolically altered yet persistent mitochondrial population. This ATP synthase reversal reduces net ATP production and drives a shift toward glycolysis to meet cellular demands. In turn, this aerobic glycolytic shift supports increased protein and lipid biosynthesis, potentially to facilitate repair or promote survival, a metabolic state reminiscent of the Warburg effect observed in cancer and other proliferative conditions^26,27. While this leak-driven glycolytic state may offer short-term protection, we hypothesize that it is metabolically unsustainable, potentially exacerbating disease progression by rendering neurons more vulnerable to additional insults. We therefore sought to test whether closing the ACLC could reverse this maladaptive state. Dexpramipexole (KNS-760704) is a small molecule previously shown to bind to the ATP synthase F1 portion, closing the ACLC of ATP synthase, and restoring mitochondrial function in models of neurodevelopmental and neurodegenerative disorders, including Fragile X syndrome^17,21,28,29. However, whether dexpramipexole can restore ATP synthase forward activity, support oxidative metabolism, and reduces mitophagic burden in glaucomatous cells remains unknown.

To address this, we used the patient-derived fibroblasts from individuals carrying the E50K-OPTN mutation to examine the metabolic and functional consequences of ACLC activation. We demonstrate that dexpramipexole reverses the leak-driven metabolic phenotype, restores oxidative phosphorylation, normalises ATP synthase activity, and reduces the mitophagic burden. These findings uncover a potentially treatable form of mitochondrial dysfunction and suggest a therapeutic strategy for glaucoma and other neurodegenerative diseases characterised by leak-associated mitochondrial disorder.

Results

Mitochondrial morphology and function are altered in patient-derived Optineurin E50K fibroblasts

We examined fibroblast mitochondria from six members of the same family, all of whom carry the E50K mutation and have NTG, along with six non-glaucomatous participants without the E50K mutation (Figure 1A). Two of the six control participants were first-degree relatives of the E50K carriers. Participants were age-matched, with a median (IQR) age of 59 years (51–75) in the E50K cohort and 61 years (51–79) in controls. The E50K group included 3 males and 3 females, while the control group comprised 2 males and 4 females. Supplementary Table 1 illustrates the main demographic features of each group. All E50K participants had advanced glaucoma in their worst eye at the time of skin biopsy (mean (IQR) mean deviation −13.0dB (−6.8 to −18.8)).

Figure 1. E50K mitochondria show structural abnormalities.

(A) Family tree showing individuals with normal tension glaucoma (NTG) and the E50K OPTN mutation (black-filled symbols; squares = males, circles = females). Red outlines indicate family members recruited for the study. These include two first degree relatives who do not carry the E50K mutation and do not have NTG. Four additional unrelated participants without glaucoma or a family history of the disease were included in the study. (B) Representative electron microscopy (EM) images from fibroblast cell lines derived from one E50K participant and one control. Red arrows indicate mitochondria; blue arrows indicate endoplasmic reticulum (ER). (C) Quantitative analysis of EM images. A total of 48 images were analyzed from two E50K and two control fibroblast lines. The number of mitochondria quantified was n = 77 for controls and n = 64 for E50K cells. Each dot represents one mitochondrion for measurements of circularity, length, and cristae area; one mitochondria-ER contact (MERC) for MERC distance analysis (some mitochondria have multiple MERCs; others have none); one individual crista for cristae length measurements (i.e., multiple cristae per mitochondrion are shown). Dot colour indicates the participant of origin. Mitochondria from E50K fibroblasts were significantly more circular, shorter, had darker matrices, smaller cristae area/unit mitochondrial area, shorter cristae, and were found in closer proximity to the ER compared to controls. MERCs were defined as contacts with a mitochondria-ER distance of <100 nm. (D) Representative Airyscan images of fibroblasts stained with MitoTracker^™ Red, showing mitochondrial networks in control and E50K cells. Left to right: Raw fluorescence image -> image converted to 8-bit grayscale and binarized -> binarized image skeletonized for analysis. (E) Boxplots show mitochondrial morphology and connectivity analysis from 20 Control cells, and 27 E50Kcells across two control and two E50K fibroblast participants. Each dot represents one cell, with dot colour indicating the individual participant-derived cell line. E50K cells exhibited fewer mitochondrial branches and junctions per mitochondrion, shorter mean branch length, and reduced individual mitochondrial area and form factor. Higher form factor values indicate more complex mitochondrial shapes, while lower values suggest more compact and regular structures. Statistical comparisons were performed per cell (F) Representative Airyscan confocal images from one Control and one E50K fibroblast stained with MitoTracker Red and transfected with mito-GFPand merged image. MitoTracker Red accumulates in mitochondria with intact membrane potential, while GFP labels all mitochondria regardless of membrane potential. Mitochondria with overlapping red and green signal indicate mitochondria with a membrane potential, while green only on the merged image indicates depolarized mitochondria. (G) Boxplots showing quantification of mitochondrial area (footprint) based on MitoTracker^™ Red and GFP signals in fibroblasts from two E50K patients and two controls (n = 19 cells per group). Each dot represents a single cell, colour-coded by participant/cell line. E50K cells exhibited a larger GFP-positive mitochondrial area, suggesting increased total mitochondrial mass potentially due to impaired mitophagy. The ratio of MitoTracker Red to GFP area was reduced in E50K cells, indicating depolarized mitochondria in the E50K cells. Significance is indicated as follows: P < 0.05 (), P < 0.01 (**), P < 0.0001 (***), P < 0.00001 (****); ns: not significant.

To assess mitochondrial morphology and network integrity in E50K patient-derived fibroblasts, we used transmission electron microscopy (TEM) and live-cell confocal imaging. TEM revealed striking morphological differences in mitochondria from E50K fibroblasts (Figure 1B–C). Compared to controls, E50K mitochondria were more circular and shorter in length, features consistent with increased mitochondrial fragmentation. Cristae were also notably shorter and disorganized, with reduced overall area and altered inner membrane structure. The mitochondrial matrix appeared darker and denser than in controls, suggestive of altered protein composition or water content. Additionally, we observed an increase in mitochondria–ER contact sites (MERCs), a feature often associated with mitochondrial stress, but that we will attribute to high protein synthesis rates.

To further characterize mitochondrial morphology beyond TEM, we performed live-cell confocal imaging using MitoTracker^™ Red, which stains mitochondria in a membrane potential–dependent manner (Figure 1D–E). E50K fibroblasts displayed significantly more fragmented mitochondrial networks, with reduced branch length, smaller mitochondrial area, and lower form factor, consistent with the increased mitochondrial fragmentation seen on TEM.

To complement the morphological data, we quantified the mitochondrial footprint using live-cell imaging with MitoTracker^™ Red (membrane potential dependent) and mitochondrially targeted, transfected GFP construct (membrane potential-independent) (Figure 1F–G). The total MitoTracker^™ Red-stained membrane area (μm²) was similar between E50K and control fibroblasts, indicating comparable levels of mitochondria with intact membrane potential. In contrast, the GFP-positive mitochondrial membrane area was significantly greater in E50K cells (p < 0.01), suggesting an overall increase in mitochondrial mass or number. Consequently, the ratio of MitoTracker^™ Red to GFP area was significantly lower in E50K fibroblasts (p < 0.0001), reflecting a relative accumulation of mitochondria with reduced or lost membrane potential. This finding suggests the accumulation of dysfunctional, depolarized mitochondria in E50K cells, consistent with a mitophagy bottleneck, a defining feature of the E50K mutation^10,30–35.

We assessed mitochondrial and glycolytic function using Seahorse respirometry and tetramethylrhodamine-methyl-ester (TMRM) based membrane potential analysis. TMRM fluorescence showed no significant difference between E50K and control cells (Figure 2A–B). However, Seahorse analysis, normalized to cell numbers, revealed significantly elevated basal oxygen consumption and proton leak in E50K fibroblasts, with no difference in maximal or spare respiratory capacity (Figure 2C–E). Because E50K cells exhibit increased mitochondrial mass, OCR was also normalized to total mitochondrial footprint (GFP-based), which revealed slightly higher OCR in controls (Supplementary Figure 1A). In contrast, normalization to MitoTracker^™ Red (membrane potential-positive mitochondria) revealed markedly higher basal respiration and proton leak in E50K cells (Supplementary Figure1B). This suggests that membrane-potential retaining mitochondria in E50K cells are working disproportionately harder, likely reflecting uncoupled respiration and compensatory activity.

Figure 2.

E50K mitochondria have altered bioenergetics. (A) Representative Airyscan confocal images of mitochondria stained with TMRM to assess membrane potential in control and E50K fibroblasts. (B) Quantification of TMRM mean fluorescence intensity (MFI) from 5 control and 5 E50K cell lines, with a total of 44 control and 42 E50K cells analysed, showing no significant difference between groups. (C) Seahorse XFe24 analysis of oxygen consumption rate (OCR),, normalized to cell number, showing overall higher OCR in E50K cells. (D, E) Boxplots of basal respiration and proton leak, both elevated in E50K fibroblasts (p < 0.01). (F) Seahorse glycolytic assay using antimycin A to inhibit mitochondrial respiration followed by 2-deoxyglucose (2-DG) to inhibit glycolysis, showing proton efflux rate (PER), a measure of glycolytic activity, higher in E50K cells. (G, H) Basal and compensatory glycolysis (rotenone/antimycin), both increased in E50K (p < 0.01). (I) Boxplot showing LDHB (catalyzes the conversion of lactate ot pyruvate), lower in E50K cells (p < 0.01). (J) Lactate levels in media from fibroblast cultures at day 3, higher in E50K (p < 0.05) compared to control. (K) NADH/NAD+ ratio elevated in E50K cells (p < 0.01). Experiment done on 6 control and 4 E50K cell lines (L) Barplots showing total, mitochondrial, and glycolytic ATP production, demonstrating higher total ATP in E50K, unchanged mitochondrial ATP, and elevated glycolytic ATP following oligomycin treatment. (M) Pyruvate dehydrogenase (PDH) activity (control n=5, E50K n=4), and (N) citrate synthase (CS) activity (control n=5, E50K n=5), both reduced in E50K fibroblasts (p < 0.05). (O) Cell growth curves from day 1 to 7, with 8 technical replicates per cell line, showing mean and SEM; E50K (n=4) cells lack a lag phase under basal conditions, while controls (n=6) display a typical lag-exponential growth pattern. Oligomycin (50 nM) treatment restores the lag phase in E50K cells. (P) Quantification of lag phase growth rate across conditions, with E50K cells exhibiting the highest rate that decreases to control levels following oligomycin treatment. Significance is indicated as follows: P < 0.05 (), P < 0.01 (**), P < 0.0001 (***), P < 0.00001 (****); ns: not significant. Assays were performed on 6 control and 6 E50K cell lines, unless otherwise specified.

To further explore metabolic adaptation, we performed a Seahorse glycolysis stress test. Both basal and compensatory glycolysis were significantly increased in E50K cells (Figure 2F–H). To add support this observation, we measured the activity of lactate dehydrogenase B (LDHB), which catalyzes the conversion of lactate to pyruvate. LDHB activity was significantly lower in E50K cells (Figure 2I), consistent with reduced pyruvate recycling and a shift toward lactate production.

Correspondingly, lactate levels in the culture medium were significantly higher in E50K cells after 72 hours (Figure 2J). These findings indicate glycolytic phenotype. This glycolytic shift was accompanied by a significantly higher NADH:NAD⁺ ratio in E50K cells (Figure 2K) suggesting increased glycolytic flux to sustain ATP production. Together with the elevated OCR, this implies that both glycolysis and oxidative phosphorylation are upregulated, but the balance of energy metabolism is shifted toward glycolysis to compensate for mitochondrial inefficiency. This may reflect a state where ATP synthase reversal and proton leak require increased glycolysis to generate ATP.

To assess the contribution of glycolysis to total cellular ATP production, we measured ATP levels before and after oligomycin treatment. Total ATP levels were comparable between groups; however, following oligomycin, which inhibits mitochondrial ATP synthesis or hydrolysis, E50K cells exhibited significantly higher glycolytic ATP production, consistent with greater reliance on glycolysis to meet energetic demands (Figure 2L). Mitochondrial ATP production was similar between groups. Although the activity of key TCA cycle enzymes, including pyruvate dehydrogenase (PDH) and citrate synthase (CS), was reduced in E50K cells (Figure 2 M–N), we interpret this not as a suppression of the TCA cycle, but rather as evidence of TCA rewiring. This may reflect a metabolic adaptation in which TCA intermediates are diverted toward anabolic processes such as amino acid and lipid biosynthesis, rather than being fully oxidised for ATP generation by ATP synthase. To assess growth dynamics, we monitored fibroblast proliferation over 7 days. Control cells showed a typical lag phase (days 1–3) followed by exponential growth. In contrast, E50K cells entered the log phase immediately, indicating accelerated proliferation. Strikingly, treatment with low-dose oligomycin restored the lag phase in E50K cells, suggesting that their rapid growth is at least partly driven by F1FO-ATPase activity (Figure 2 O–P).

Optineurin E50K mutant mitochondria have a large conductance inner membrane leak

To determine whether ATP synthase operates in reverse in E50K mitochondria, which is a hallmark of inner membrane proton leak¹⁶; we used TMRM live-cell imaging to monitor mitochondrial membrane potential dynamics. At baseline, TMRM fluorescence intensity was similar in E50K and control cells. Upon treatment with oligomycin, controls hyperpolarised, whereas E50K cells depolarised consistent with pre-oligomycin ATP synthase reversal and proton leak. Subsequent FCCP addition fully depolarised both groups (Figure 3A–B). Quantification of TMRM mean fluorescence intensity confirmed a decrease in E50K cells following oligomycin treatment (p < 0.0001, Figure 3C). To directly assess inner membrane conductance, we performed patch-clamp recordings on submitochondrial vesicles (SMVs). At rest, E50K SMVs exhibited higher ratio of channel open to closed durations and peak conductance compared to controls (p < 0.01). ATP application, which closes the ACLC^15,17, reduced conductance in both groups. Subsequent addition of WEHI-539, a selective Bcl-xL inhibitor that displaces Bcl-xL from ATP synthase and opens the c-subunit channel^36,37, caused a significantly greater increase in conductance in E50K SMVs which showed higher open/closed channel ratios (p < 0.05) and elevated peak conductance (p < 0.05) compared to controls (Figure 3D–F). Finally, Western blot analysis of mitochondrial lysates revealed higher ATP synthase c-subunit levels in E50K mitochondria (p < 0.01), while β-subunit levels were only slightly elevated (Figure 3G). The c:β ratio was therefore increased in E50K cells, suggesting accumulation of free c-subunit embedded in the IMM. These findings collectively indicate that E50K mitochondria possess an enhanced ATP synthase c-subunit leak, driven by channel opening and subunit imbalance, and support the hypothesis of an ACLC-mediated proton leak contributing to altered mitochondrial metabolism.

Figure 3. E50K mitochondria have a large conductance inner membrane leak.

(A) Representative Airyscan confocal images of mitochondria in control and E50K fibroblasts stained with TMRM to assess membrane potential at baseline, after oligomycin (2 μM), and after FCCP (2 μM) injection. (B) Time-course plot showing percentage change in TMRM fluorescence intensity relative to baseline, with oligomycin injected at 4 min and FCCP at 6 min. Control cells exhibited hyperpolarization following oligomycin, while E50K fibroblasts showed a rapid depolarization. Both groups depolarized following FCCP. Lines represent group means and bars indicate SEM; 6 control and 6 E50K cell lines were analysed, with a total of 44 control and 42 E50K cells. (C) Boxplot quantifying TMRM fluorescence change post-oligomycin, showing significantly reduced (depolarized) response in E50K compared to control fibroblasts (p < 0.0001). (D) Representative patch-clamp recordings from submitochondrial vesicles (SMVs) held at −70 mV. At baseline, control SMVs exhibited low conductance (−10 pA), while E50K SMVs showed elevated conductance (−70 pA), indicating increased channel activity. Addition of ATP closed the leak in both groups, followed by WEHI-539, a selective Bcl-xL inhibitor, which re-opened the channel—markedly more in E50K. Bcl-xL normally binds ATP synthase to prevent c-subunit leak; WEHI-539 inhibits BCL-xL from binding to ATP synthase, exposing latent channel activity. (E) Quantification of open/closed channel ratio in SMVs at baseline, post-ATP, and post-WEHI-539, showing higher ratios in E50K cells at rest and after WEHI-539 exposure. (F) Peak conductance values under the same conditions, also elevated in E50K SMVs, confirming enhanced ATP synthase leak channel activity. For both panels (E and F): Control SMVs (n = 6) from 3 participants (2 SMVs per participant); E50K SMVs (n = 7) from 3 participants. (G) Representative Western blot of mitochondrial lysates is shown followed by box plots of group data from 6 control and 6 E50K cell lines showing elevated ATP synthase c-subunit levels in E50K mitochondria, normalized to SDHA (p < 0.01). ATP synthase β-subunit levels were slightly elevated in E50K; however, the β/c-subunit ratio was increased in E50K cells. Significance is indicated as follows: P < 0.05 (), P < 0.01 (**), P < 0.0001 (***), P < 0.00001 (****); ns: not significant.

E50K cells have a leak dependent mitochondrial metabolism

Our findings show reversed ATP synthase activity and increased ACLCs in E50K mitochondria. To assess associated metabolic changes, we used puromycin labeling to measure active protein synthesis combined with proteomic profiling in fibroblasts, combined with targeted metabolomics in patient sera. We labeled newly synthesized proteins with puromycin and immunoprecipitated with an anti-puromycin antibody. We then evaluated the newly synthesized proteome by liquid chromatography with tandem mass spectrometry (LC/MS/MS). Principal component analysis (PCA) of the proteomic profiles revealed clear separation between E50K and control samples, suggesting distinct translational signatures (Supplementary Figure 1C). This analysis also revealed anincrease in global protein synthesis in E50K cells, with a greater number of upregulated proteins compared to controls (log₂FC > 1 and p < 0.05) (Figure 4A). Western blotting confirmed increased puromycin incorporation relative to GAPDH (p < 0.05) (Figure 4B). Notably, upregulated proteins included α-tubulin isoforms (TUBA1A, TUBA1C, TUBA3C, TUBA3E), elongation factor eEF2, stress response proteins (STAT1, ALDH7A1, PA2G4, PDIA4), whilst downregulated proteins included MARS1 and PTK7. Newly synthesized proteins in E50K cells also included several large and small subunit components, including RPL38, RPL12, RPSA, and RPS3A, consistent with increased translation. Interestingly, this was not a uniform response, some ribosomal proteins (RPL13A, RPL13AP3, RPL6) were downregulated, suggesting remodeling of ribosome composition rather than bulk ribosome biogenesis. This pattern aligns with emerging models of ribosome specialization, in which altered ribosomal protein stoichiometry supports stress-adapted or disease-specific translational programs³⁸. Gene set enrichment analysis (GSEA) of the nascent proteome revealed enrichment for glycolysis, mTORC1 signalling, unfolded protein response, apoptosis, mitotic spindle assembly, protein secretion, and hypoxia (p < 0.05) (Figure 4C–D).

Figure 4.

E50K cells have mitochondrial leak metabolism (A) Volcano plot of newly synthesized proteins quantified via LC–MS/MS after anti-puromycin immunoprecipitation (n = 3 control, n = 3 E50K). Red dots indicate significantly upregulated proteins (log₂FC > 1, p < 0.05), blue dots show significantly downregulated proteins (log₂FC < −1, p < 0.05). The increase in rightsided dots show increased protein synthesis rate in E50K fibroblasts. (B) Western blot of puromycin-labeled proteins in fibroblast lysates, showing increased signal in E50K versus controls, normalized to GAPDH (n = 6 per group, p < 0.01). This indicates increased protein synthesis rate in E50K cells. (C) Gene Set Enrichment Analysis (GSEA) of the newly synthesized proteome using hallmark pathways and showing significant enrichment in glycolysis, mTORC1 signaling, unfolded protein response (UPR), apoptosis, mitotic spindle, protein secretion, and hypoxia. (D) Tile plots of the top seven enriched pathways across all samples (3 control, 3 E50K), highlighting protein-level contributions. Color scale reflects the normalized enrichment score (NES) per protein in each pathway. More proteins in each pathways shown are upregulated in E50K cells compared to controls. (E) Targeted metabolomic profiling of serum from E50K patients (n = 6) and controls (n = 20) reveals enrichment in NAD⁺ signaling, nicotinate metabolism, purine degradation, TCA cycle, and sirtuin signaling. (F) Integrated pathway analysis combining significantly altered proteins (fibroblasts) and metabolites (sera), using MetaboAnalyst. Pathway impact (x-axis) and significance (y-axis) identify convergent dysregulation in glycolysis, aminoacyl-tRNA biosynthesis, arginine biosynthesis, TCA cycle, pentose phosphate pathway, pyruvate metabolism, and glutathione metabolism.

Targeted serum metabolomics showed enrichment in NAD⁺ and nicotinate metabolism, purine degradation, TCA cycle intermediates, amino acid transport, and sirtuin signaling (Figure 4E). Orthogonal partial least squares discriminant analysis (orthoPLS-DA) of the metabolomic profiles revealed clear separation between E50K and control samples (Supplementary Figure 1D), consistent with distinct metabolic signatures. Joint pathway analysis further identified convergence on glycolysis, the TCA cycle, pentose phosphate pathway, pyruvate metabolism, glutathione metabolism, and aminoacyl-tRNA biosynthesis (Figure 4F).

Together, these findings are consistent with a model where chronic mitochondrial leak channel opening drives an anabolic and glycolytic shift, supporting the notion of an ACLC-dependent metabolic state in E50K cells^17,20,28.

Dexpramipexole restores mitochondrial structure and function in E50K cells and reduces aberrant protein synthesis

Dexpramipexole is a cytoprotective compound known to reduce mitochondrial leak by binding directly to the ATP synthase^21,39–41. To test its effect on E50K fibroblasts, we examined mitochondrial ultrastructure using TEM. After 24 hours of dexpramipexole treatment, E50K mitochondria showed markedly improved morphology: reduced circularity, increased cristae area, fewer MERCs, and less dense matrix, all consistent with healthier organelles (Figure 5A–B). Confocal imaging with MitoTracker^™ Red confirmed recovery. Although branching tended to increase, mitochondrial length and area returned to control levels, and form factor exceeded controls, indicating a more fused, bioenergetically competent network (Figure 5C–D). We next tested whether dexpramipexole treatment restored correct ATP synthase directionality by closing the ATP synthase leak channel. In TMRM-based membrane potential assays, untreated E50K mitochondria depolarised upon oligomycin exposure, as expected from ATP synthase reversal and a persistent leak. By contrast, dexpramipexole-treated E50K cells hyperpolarised in response to oligomycin, similar to controls, indicating a switch back to forward ATP synthase activity and pore closure (Figure 5E–G).

Figure 5.

ATP synthase leak channel inhibitor dexpramipexole restores mitochondrial structure and function in E50K cells (A) Representative transmission electron micrographs (TEM) of control and E50K fibroblasts pre- and post-treatment with dexpramipexole (Dex, 24 h). Red arrows indicate mitochondria; blue arrows denote endoplasmic reticulum. E50K mitochondria exhibit rounded morphology, reduced cristae area, and darker matrix pre-treatment; Dexpramipexole restores ultrastructure toward control appearance. (B) Quantification of mitochondrial morphology from TEM images: E50K mitochondria are more circular than controls pre-treatment (p < 0.01), with reduced cristae area and darker matrix. Post-dexpramipexole treatment, E50K mitochondria become less circular, with increased cristae area and matrix opacity comparable to controls. Data are from 2 control and 2 E50K participants (colored separately); each dot represents a single mitochondrion. The number of mitochondria quantified was n = 77 for controls, n = 98 for controls + Dex, n = 64 for E50K and n = 121 for E50K +Dex. (C) Representative confocal live-cell imaging with MitoTracker^™ Red for control and E50K cells pre and post 24 hr Dex treatment. Images show improved mitochondrial network morphology in E50K fibroblasts after 24 h Dex. (D) Boxplots showing mitochondrial morphology and connectivity analysis from Airyscan confocal live-cell images, stained with MitoTracker^™ Red (Control = 20, E50K = 27, Control + Dex = 17, E50K + Dex = 19) across two control and two E50K fibroblast cell lines, coloured separately. While the number of branches per cell remains unchanged, mean branch length, mitochondrial area, and form factor are significantly increased in E50K post-treatment. (E) Representative Airyscan confocal images of TMRM-stained live fibroblasts at baseline, post-oligomycin (2 μM), assessing ATP synthase activity directionality. In untreated E50K cells, TMRM signal decreases after oligomycin, consistent with ATP synthase reversal and mitochondrial depolarization; this is prevented by 8 h Dexpramipexole treatment, with TMRM signal maintained post-oligomycin, indicating ATP synthase operating in the forward direction. (F) Time-course of TMRM fluorescence showing mean fluorescence intensity (MFI, % change from baseline) following oligomycin injection at 2 minutes. E50K fibroblasts exhibit acute depolarization, whereas both control groups and Dex-treated E50K cells maintain mitochondrial polarization. (n = 8 images from 2 control cell lines; n = 6 images from 3 Dex-treated control cell lines; n = 8 images from 3 E50K cell lines; n = 12 images from 3 Dex-treated E50K cell lines). (G) Quantification of TMRM response at peak post-oligomycin across all four groups. E50K fibroblasts show a significant drop in TMRM signal after oligomycin, while Dex-treated E50K cells maintain signal, indicating rescue of ATP synthase directionality.

Given the aberrantly increased protein synthesis rate observed in E50K cells at baseline, we next evaluated whether dexpramipexole treatment could reverse this hyperactive biosynthesis. Using puromycin-labelled proteomics, we quantified newly synthesized proteins to assess the impact of dexpramipexole on global protein synthesis. PCA of the proteomic profiles revealed clear separation among control, E50K, and E50K + Dex samples, indicating distinct translational signatures and a partial rescue effect following Dex treatment (Supplementary Figure 1E). After 8 hours of dexpramipexole treatment, E50K cells showed significantly reduced global protein synthesis rate, confirmed both by western blotting of puromycin/GAPDH ratios and proteomics (Figure 6A–C). GSEA showed that hallmark pathways previously enriched in untreated E50K cells; including glycolysis, mTORC1 signaling, unfolded protein response, apoptosis, and hypoxia, no longer reached statistical significance post-dexpramipexole treatment and exhibited decreased normalized enrichment scores, indicating a reduction in their activation (Figure 6D–E). To validate rescue of mitochondrial protein stoichiometry, we assessed ATP synthase subunit levels. The elevated c-subunit:β-subunit ratio observed in untreated E50K cells was reduced after 8 hours of dexpramipexole (p < 0.01), approaching control levels (Figure 6F).

Figure 6.

Dexpramipexole reverses aberrant leak metabolism of E50K cells (A) Representative Western blot of newly synthesized proteins (puromycin-labeled) with β-tubulin as a loading control, across four conditions: control, E50K, control + Dex, and E50K + Dexpramipexole (Dex, 8 h). E50K cells show increased puromycin incorporation, indicating elevated global protein synthesis. (B) Quantification (boxplots) of puromycin/β-tubulin signal intensity. Control samples are set at 100%. E50K fibroblasts exhibit over a twofold increase in newly synthesized proteins relative to controls. Following Dexpramipexole treatment, protein synthesis levels in both E50K and control fibroblasts return to control levels. Data are from 3 control and 3 E50K cell lines. (C) Volcano plot of differentially expressed newly synthesized proteins (LC/MS/MS). Red: log₂FC > 1 and p < 0.05; blue: log₂FC < −1 and p < 0.05. E50K fibroblasts show broad upregulation of protein synthesis compared to controls; in plot at right, after Dexpramipexole treatment (8 h), global protein synthesis is reduced. (D) Gene Set Enrichment Analysis (GSEA) of hallmark pathways in the nascent proteome comparing E50K pre and post treatment (n = 3 cell lines per group). E50K fibroblasts show enrichment of glycolysis, mTORC1 signaling, unfolded protein response (UPR), and other pathways, which decrease following Dex. Only TNFα signaling via NF-κB remains significantly enriched after treatment. NES (Normalized Enrichment Score) is reflected in the color intensity. (E) Bar plots showing signal-to-noise ratio (SNR, difference in mean expression signal/standard deviation across samples) for the top three enriched pathways: glycolysis, mTORC1 signaling, and UPR. Red: E50K vs control; green: E50K + Dexpramipexole vs control. All three pathways are upregulated in E50K and suppressed following treatment. (F) Western blot of ATP synthase subunits c and β in 3 E50K participants pre- and post-dexpramipexole treatment. E50K fibroblasts show a higher c/β ratio at baseline (p < 0.01), which decreases after Dex, due to selective reduction of subunit c. (G) Representative western blots of LC3-I, LC3-II, p62, PFK1, and MDH2, with β-actin as the loading control, in control and E50K fibroblasts at baseline and following dexpramipexole treatment at 4, 8, and 24 hours. (H) Quantification of protein expression over time from 3 control and 3 E50K participants, shown as line graphs with mean ± SEM. LC3-II/I ratio is elevated in E50K cells at baseline and remains high following Dex treatment. p62 levels are significantly increased in E50K cells (p < 0.001) and normalize by 8–24 hours post-treatment, consistent with reduced mitophagic burden. PFK1 and MDH2 (Glycolysis and TCA markers) are also elevated in E50K cells (p < 0.05) and return to control levels after Dex treatment, indicating recovery of metabolic balance.

Finally, we examined mitophagy dynamics via time-course western blotting (Figure 6G–H). LC3-II/I ratios were elevated in E50K cells at baseline and remained increased at 4-, 8-, and 24-hours post-dexpramipexole (p < 0.05), indicating sustained mitophagy activation under basal conditions. However, p62 levels, initially elevated in E50K cells (p < 0.01), decreased progressively over time and normalised by 8–24 hours. This reduction may reflect either a decreased mitophagic burden due to improved mitochondrial integrity or, alternatively, partial restoration of mitophagic flux. Similarly, glycolytic enzyme PFK1 and mitochondrial malate dehydrogenase 2 (MDH2), a key component of the malate-aspartate shuttle (MAS) and TCA were elevated at baseline and 4 hours post-treatment (p < 0.05). Both enzymes trended toward control levels by 8–24 hours, suggesting that dexpramipexole normalizes not only glycolysis but also mitochondrial redox shuttling and overall metabolic activity over time.

Discussion

Our findings confirm previous reports that the OPTN-E50K mutation impairs mitophagy, resulting in the accumulation of dysfunctional mitochondria. Building on this, we show for the first time that the E50K mutant mitochondria feature an abnormally active ACLC. This channel activity dissipates membrane potential, forcing ATP synthase to reverse and hydrolyze cytosolic ATP to sustain mitochondrial membrane potential and ensure cell survival. This reversal increases glycolytic ATP production and shifts cellular metabolism toward aerobic glycolysis, consistent with a Warburg-like metabolic adaptation. Supporting this, we observed elevated glycolytic and biosynthetic pathways alongside increased protein synthesis rate in E50K cells. DEX, an ATP synthase modulator that stabilizes and closes the ACLC, restores forward ATP synthase activity, normalizes mitochondrial morphology and function, and reduces maladaptive metabolic signaling. It also reduces the mitophagic burden, consistent with improved mitochondrial integrity. Together, these findings identify the ACLC as a key driver of metabolic and mitochondrial pathology in E50K cells and establish ATP synthase modulators as promising therapeutics to reverse these maladaptive changes. Although DEX previously failed to show efficacy in the Phase III ALS EMPOWER trial⁴⁰, the vast majority of ALS patients do not carry OPTN mutations (fewer than 1% are OPTN-associated) and the mechanisms driving sporadic ALS are highly heterogeneous and largely undefined. By contrast, E50K-OPTN fibroblasts exhibit a specific, well-defined mitochondrial defect: elevated c-subunit levels, open ACLC pores, reversed ATP synthase activity, and leak-driven metabolism. This provides a mechanistically matched context for DEX, a compound known to stabilize the F₁ domain and close the ACLC^21,39. Thus, unlike in heterogeneous ALS cohorts, the E50K model offers a genetically and mechanistically targeted setting in which DEX’s mitochondrial action can be directly evaluated.

We first examined mitochondrial morphology in E50K patient-derived fibroblasts, as structural changes often provide early clues to mitochondrial dysfunction. Using TEM and confocal microscopy, we observed mitochondrial fragmentation, cristae disorganization, and a dense matrix, all indicative of bioenergetic and proteostatic stress. Fragmentation, a hallmark of mitophagy initiation that facilitates removal of damaged mitochondria, appeared incomplete in E50K cells. This was supported by elevated LC3-II/I ratios and increased p62 levels, markers of impaired autophagic flux and accumulation of undegraded autophagosomes. The accumulation of small, rounded mitochondria alongside disrupted cristae suggests an imbalance between mitochondrial fission and clearance. Notably, E50K mitochondria exhibited significantly shortened cristae and altered morphology, consistent with inner membrane remodeling⁴². Since cristae structure organizes oxidative phosphorylation complexes and ATP synthase dimers, their disruption likely contributes to inefficient, leak-prone respiration^43–45. The mitochondrial matrix appeared darker and denser, a feature linked to increased mitochondrial protein synthesis and activation of the mitochondrial UPR, both markers of proteostatic stress^17,46–48. We also observed increased MERCs, specialized junctions essential for calcium signaling⁴⁹, lipid exchange⁵⁰, increased protein synthesis rate¹⁷ and mitophagy initiation^51,52. This likely reflects enhanced mitophagy activation as the cell attempts to clear damaged mitochondria, consistent with chronic mitochondrial stress. Despite this, E50K fibroblasts showed paradoxically increased OCR, indicating that the mitochondria are working harder to maintain membrane potential and consuming more oxygen, consistent with uncoupled respiration. This bioenergetic inefficiency was accompanied by a marked shift towards glycolytic ATP production, underscoring a metabolic adaptation to mitochondrial dysfunction. To reconcile these metabolic changes with the apparently unchanged TMRM signal, we considered how membrane potential was quantified. Because TMRM signal was normalised to total cell area, this measurement reflects membrane potential per unit area rather than per mitochondrion. E50K cells contain a larger mitochondrial mass (increased mito-GFP area relative to Mitotracker Red), meaning the TMRM signal likely represents a mixture of both fully depolarised and hyperpolarised mitochondria. This heterogeneous population would dilute detectable differences and may explain why overall TMRM values appear unchanged.

E50K fibroblasts showed elevated basal and compensatory glycolysis, increased lactate accumulation, and a lower LDHB activity, suggesting impaired pyruvate recycling and a reliance on the conversion of pyruvate to lactate. This shift coincided with a higher NADH:NAD⁺ ratio, pointing to inefficient NADH oxidation at complex I (CI), most likely indicating other complex use to increase oxygen consumption; a feature of mitochondrial uncoupling⁵³. (could possibly also be increased TCA production of NADH with hyperpolarized membrane potential, so no pressure to use the NADH in CI?) The CI bottleneck may be compensated by activation of the Glycerol-3-Phosphate Shuttle (G3PS), which regenerates cytosolic NAD⁺ via glycerol-3-phosphate dehydrogenase 1 (GPD1) to maintain the observed high glycolytic flux^54,55. Electrons are then transferred by the flavoprotein GPD2 directly to the ubiquinone pool, bypassing CI and feeding into Complex III. This strategic diversion provides two advantages: it limits CI-dependent reactive oxygen species (ROS) production⁵⁶ and ensures rapid electron transport and oxygen consumption, which are critical for sustaining membrane potential during uncoupled phosphorylation. Impaired Complex I function has been shown previously in glaucoma⁵⁷. Reduced activity of TCA enzymes such as PDH and CS, in the absence of overt mitochondrial ATP production deficits, points to TCA rewiring, possibly favoring anabolic outputs. This pattern, which reflects a Warburg-like phenotype, is further supported by the enhanced proliferation of E50K fibroblasts and their unique sensitivity to low-dose oligomycin, which slowed their growth and restored a lag phase typical of non-transformed cells.⁵⁸

E50K cells show a marked shift towards glycolytic ATP production, reflecting a metabolic adaptation to mitochondrial dysfunction. Despite unchanged total ATP levels, this reliance on glycolysis suggests impaired oxidative phosphorylation. Live-cell imaging confirmed that ATP synthase operates in reverse in E50K mitochondria, hydrolyzing ATP to preserve membrane potential, indicative of energy-inefficient, uncoupled respiration. Prior work from our group and others has shown that a regulated proton leak can occur through the membrane-embedded c-subunit ring of ATP synthase (ACLC), which plays a critical role in controlling mitochondrial efficiency under stress conditions^{17,18,20,23,24,59,60}.

Electrophysiological recordings revealed that E50K submitochondrial vesicles exhibit increased ACLC activity, with higher open-to-closed channel ratios and peak conductance. This leak was suppressed by ATP, suggesting its origin from the ATP synthase ACLC, and reactivated by WEHI-539, a Bcl-xL inhibitor known to promote ATP synthase ACLC opening. We observed elevated c-subunit to β-subunit ratios in E50K mitochondria, suggesting an increased pool of free (disassembled or unassembled) c-rings. Together, these findings implicate dysregulated ACLC activity as a key mechanism underlying mitochondrial inefficiency in E50K cells.

Our proteomic and metabolomic analyses reveal a coordinated metabolic and biosynthetic reprogramming in E50K fibroblasts. Hallmark pathways for glycolysis, mTORC1 signalling, UPR, hypoxia, and apoptosis were significantly upregulated, consistent with a Warburg-like adaptation to mitochondrial dysfunction. In particular, mTORC1 activation suggests enhanced anabolic signalling, supporting protein synthesis and cell growth. Notably, the activation of the hypoxia pathway, while not indicating low oxygen here, likely reflects stabilization of HIF1α, which can occur in normoxic conditions via mitochondrial leak²⁰. Recent work has shown that proton leak through ATP synthase can drive HIF1α accumulation independently of hypoxia, thereby reprogramming metabolism toward glycolysis and cell survival²⁰. This leak-driven pseudo-hypoxic signaling, alongside impaired mitophagy and biosynthetic dysregulation, identifies ATP synthase as a central pathological hub in E50K fibroblasts.

Analysis of newly synthesized proteins revealed changes in pathways beyond energy metabolism. Downregulation of PTK7, a pseudokinase involved in non-canonical Wnt signalling, is notable in light of previous studies showing that Wnt inhibition exacerbates glaucoma-like changes in the trabecular meshwork⁶¹. In parallel, enrichment of NAD⁺ metabolism and sirtuin-associated pathways in patient serum indicates a systemic response to mitochondrial dysfunction. Members of the sirtuin family (e.g., SIRT1, SIRT3, SIRT7) are known to coordinate the mitochondrial UPR and regulate redox and stress signalling under metabolic strain^62,63. Their involvement in neurodegenerative diseases, including glaucoma, reinforces the relevance of the NAD⁺/sirtuin axis in E50K pathology⁶⁴.

Previous studies have demonstrated that dexpramipexole (DEX) can effectively close the ACLC, thereby improving mitochondrial coupling and overall bioenergetic efficiency. Building upon these findings, we sought to determine whether DEX treatment could not only halt the leak but also reverse the broader mitochondrial dysfunction characteristic of E50K mutant fibroblasts. Our investigations revealed that DEX treatment robustly restored mitochondrial ultrastructure, and eliminated the aberrant ATP synthase reversal that compromises energy production. Importantly, DEX normalized the ratio of ATP synthase c-subunit to β subunit, re-established proper mitochondrial membrane potential and normalized dysregulated protein biosynthesis. The divergent effects of DEX on LC3-II and p62 offer two plausible, non-exclusive interpretations regarding its impact on mitochondrial quality control in E50K fibroblasts. First, the marked reduction in p62 following DEX treatment suggests a decrease in ubiquitinated, damaged mitochondrial cargo, consistent with improved mitochondrial integrity and a reduced requirement for mitophagic clearance. In this scenario, DEX primarily enhances mitochondrial health upstream of autophagy, thereby lowering p62 recruitment while leaving the global autophagy tone (reflected by LC3-II) unchanged. An alternative explanation is that DEX partially rescues mitophagic flux itself, facilitating more efficient cargo turnover, while persistent elevation of LC3-II reflects broader autophagy activation or residual defects in autophagosome dynamics intrinsic to the E50K mutation. Given that OPTN E50K is known to disrupt autophagosome trafficking and lysosomal engagement, improved mitochondrial quality may “buy time” for the existing autophagy machinery rather than fully normalising it. Thus, the post-DEX profile, p62 reduction with sustained LC3-II elevation, likely reflects a combination of enhanced mitochondrial function with either maintained or only partially restored autophagic processing. These improvements in mitochondrial integrity and function are accompanied by broad changes in cellular metabolism. Consistent with this, GSEA pathway analysis of newly synthesized proteins revealed significant downregulation of hallmark metabolic and biosynthetic pathways following DEX treatment, including glycolysis, mTORC1 signalling, UPR, and hypoxia pathways, indicating reversal of the E50K-induced metabolic reprogramming. Notably, DEX treatment also significantly reduced the elevated protein synthesis seen in E50K cells, which is critical because excessive protein synthesis can exacerbate cellular stress and energy demand under conditions of mitochondrial dysfunction^17,65. By downregulating protein synthesis, DEX may help re-balance cellular metabolism, reducing proteotoxic stress and promoting cell survival. Interestingly, the only hallmark pathway with increased enrichment following DEX treatment was TNFα signalling via NF-κB. This may represent a protective response, as NF-κB mediates the induction and upregulates the expression of BCL-xL⁶⁶, which binds to the β-subunit of the F₁ sector of ATP synthase and acts to close the leak through the c-subunit channel^36,37, thereby restoring mitochondrial coupling and improving membrane potential stability.

Among the glycolysis-associated proteins, PRPS1 (phosphoribosyl pyrophosphate synthetase 1) exhibited the highest signal-to-noise ratio (SNR), with levels significantly elevated in E50K fibroblasts (SNR ≈ 7.2, FDR < 0.05) and reduced following DEX treatment (SNR ≈ 3.0, FDR > 0.05). While not a direct glycolytic enzyme, PRPS1 catalyzes the synthesis of PRPP from ribose-5-phosphate, a metabolite derived from the pentose phosphate pathway (PPP), which branches from glycolysis. PRPP is a critical precursor for de novo synthesis of both RNA and DNA⁶⁷. Its upregulation suggests a metabolic shift towards increased transcription, which may support increased protein synthesis that we found in theE50K cells. The marked reduction following DEX treatment indicates a normalization of this drive and supports the role of DEX in restoring metabolic homeostasis.

In E50K fibroblasts, Calreticulin (CALR), was the most upregulated protein involved in mTORC1 and UPR pathways, showing a strong upregulation pre-DEX treatment (SNR ~14.0, FDR <0.001), with a marked reduction post-treatment (SNR ~5.0, FDR <0.01). CALR functions as a Ca²⁺ binding chaperone within the ER, playing a critical role in protein folding and maintaining ER calcium homeostasis⁶⁸. By regulating Ca²⁺ levels, CALR supports proper protein quality control and modulates cellular stress responses such as the UPR⁶⁹. The observed upregulation of CALR in E50K cells suggests elevated protein synthesis rate, which could increase ER stress and disrupted Ca²⁺ signaling; DEX treatment alleviated this, indicating restoration of proteostasis and cellular homeostasis.

Collectively, these results highlight the therapeutic potential of DEX to reverse key pathological features of E50K-driven mitochondrial dysfunction and restore cellular homeostasis.

This study utilizes patient-derived fibroblasts, which recapitulate key mitochondrial and metabolic defects seen in E50K glaucoma but may not fully mirror the in vivo environment of retinal ganglion cells. However, fibroblasts are considered a viable model for studying neurodegenerative changes due to their metabolic and biochemical similarities to neurons. Several studies have shown that patient-derived fibroblasts recapitulate mitochondrial bioenergetic impairments also observed in affected neuronal tissues from neurodegenerative diseases such as Alzheimer’s⁷⁰, Parkinson’s⁷¹ and Huntington’s disease⁷².

Our study identifies a mitochondrial leak channel through the ATP synthase c-subunit as a central dysfunction in glaucoma patient cells with the OPTN-E50K mutation and demonstrates that this defect is reversible with DEX treatment. By restoring mitochondrial efficiency, re-engaging mitophagy, and dampening maladaptive metabolic signaling, ATP synthase modulators emerge as promising therapeutic candidates for energy-deficient cells. While DEX improved mitochondrial function and mitophagy in vitro, further validation in neuronal and in vivo models is needed to confirm its translational potential.

Although the E50K-OPTN mutation accounts for a small fraction of glaucoma cases, monogenic forms of disease offer uniquely tractable systems in which to define pathogenic mechanisms with precision. Insights from such rare variants have historically uncovered fundamental pathways, such as mitochondrial dysfunction^73,74, autophagy defects⁷⁵, and axonal vulnerability⁷⁶, that extend beyond the mutation itself. By delineating how E50K drives ACLC-mediated leak metabolism, our findings identify a mitochondrial mechanism that may be shared more broadly across glaucoma subtypes in which metabolic impairment is increasingly recognized. Given recent studies implicating mitochondrial dysfunction as a common feature in POAG^{6,57,73,77–86}, these findings offer a mechanistic framework that could inform therapeutic strategies for mitochondrial rescue in glaucoma.

Methods

Transmission electron microscopy

Two E50K fibroblast cell lines and two control lines, picked at random, were cultured to 60–80% confluency in culture plates. Subsequently, the cells were fixed in a solution containing 1.5% glutaraldehyde in 0.1M sodium cacodylate buffer (pH 7.4) for 30 minutes at room temperature (RT), followed by an additional hour at 4°C. After fixation, cells were thoroughly rinsed three times with sodium cacodylate buffer supplemented with arsenic. Following the fixation step, electron micrographs were acquired using a FEI Tecnai Biotwin transmission electron microscope (TEM) operating at 80 kV. Images were captured with a Morada CCD camera and iTEM (Olympus) software. Subsequent analysis of the electron micrographs was conducted using ImageJ software developed by the National Institute of Health (NIH). For experiments involving dexpramipexole, fibroblasts were pre-treated with 10 μM DEX for 24 hours prior to fixation.

Confocal microscopy

For the investigation of mitochondrial morphology in live fibroblasts, MitoTracker^™ Red was selected due to its distinct spectral characteristics and advantageous properties (membrane potential dependant, can be used on live cells and resistant to bleaching)⁸⁷. MitoTracker^™ Red, has an excitation wavelength of approximately 581 nm and an emission wavelength of around 644 nm. The choice of MitoTracker^™ Red over other dyes such as Tetramethylrhodamine Methyl Ester (TMRM), was deliberate. Whilst TMRM is more sensitive to mitochondrial membrane potential, MitoTracker^™ Red is membrane potential sensitive, however it is more resistant to bleaching than TMRM⁸⁷. Fibroblasts were cultured in 35mm glass-bottom plates, specifically designed for confocal imaging, containing 2ml of complete growth media. On the day of the experiment, cells were washed with 1 ml of pre-warmed imaging buffer (5mM K+ in 50 mM HEPES buffer, pH = 7.4). Following the wash, 2 ml of 50nM MitoTracker^™ Red in imaging buffer was added to each plate and the cells were subjected to a 30-minute incubation. After this incubation, cells underwent another round of washing with imaging buffer, followed by the addition of 2 ml of fresh imaging buffer. Images were obtained using an LSM-710 confocal laser scanning microscope (Carl Zeiss, Inc.) equipped with a 63 × /1.49 numerical aperture oil-immersion objective. Cells were imaged in a 37 °C stage incubator connected to 5% CO2 gas inlet to buffer the media. The experiment was repeated three times, with an average of 3 images per cell line during each run. MitoTracker Red is sensitive to mitochondrial membrane potential. However, mitochondria green fluorescent protein (mito-GFP) is not. Mito-GFP is therefore able to comprehensively label the entire mitochondrial network, regardless of its viability. As such, the ratio of MitoTracker^™ Red mitochondrial footprint to mito-GFP is a good indicator of the proportion of mitochondria in a cell that are damaged (depolarised). Cells were stained with MitoTracker^™ Red as described in the previous section, on the day of imaging. Three days before imaging, cells were transfected with CellLight^® mito-GFP, BacMam 2.0. This construct, once transfected, expresses GFP fused to the leader sequence of E1 alpha pyruvate dehydrogenase, and serves as a universal mitochondrial marker, allowing the visualization of both live and compromised mitochondria within the cell. Transfection was carried out when cells reached approximately 30% confluency, adhering to the manufacturer’s instructions. To calculate the appropriate volume of CellLight^® reagent for the number of cells the following formula was used:

$$\mathit{\text{Volume}}of\mathit{\text{CellLight}}\circledR \mathit{\text{Rengent}}\left(mL\right)=\frac{\mathit{\text{Namber}}of\mathit{\text{cells}}\times \mathit{\text{Desired}}\mathit{\text{PPC}}}{1\times {10}^8\mathit{\text{CellLight}}\circledR \mathit{\text{particles/mL}}}$$

After several trial runs, it was determined that the optimal conditions for transfection involved using 30 particles per cell (PPC); the transfection process was conducted over a period of 3 days. Consequently, the appropriate volume of CellLight^® reagent utilized was 6 μL per plate. This dual-labelling approach, employing MitoTracker^™ Red and Mitochondria-GFP, facilitated a comprehensive assessment of mitochondrial footprint, enabling the discrimination between functional and non-functional mitochondrial populations. Confocal imaging was done on the same cell lines used for investigating mitochondrial morphology (TEM and MitoTracker^™ Red). Images were obtained using an LSM-710 confocal laser scanning microscope (Carl Zeiss, Inc.) as described in the previous section. For experiments involving dexpramipexole, fibroblasts were pre-treated with 10 μM DEX for 24 hours prior to imaging.

Confocal Image Processing and Analysis

Fibroblasts were stained with MitoTracker^™ Red and, where indicated, transfected with mito-GFP. Confocal images were acquired using an Airyscan detector on a Zeiss LSM 880 microscope. Raw images were converted to 8-bit grayscale and binarized in ImageJ using a consistent threshold across all cells. For mitochondrial network analysis, binarized images were skeletonized using the ImageJ 2D Mitochondrial Analysis plugin to quantify parameters such as branch length, number of branches, junctions, network complexity (form factor), and mitochondrial area. For comparison of MitoTracker Red and GFP signals, binarized areas were measured for each channel separately to determine the mitochondrial footprint and to calculate the ratio of MitoTracker Red to GFP area, reflecting the proportion of mitochondria with intact membrane potential relative to total mitochondrial mass. All analyses were performed on a per-cell basis to account for intra-line heterogeneity.

Mitochondrial membrane potential imaging and time-course experiments (TMRM live-cell imaging)

Mitochondrial membrane potential was assessed using the potentiometric dye tetramethylrhodamine methyl ester (TMRM). Fibroblasts (five control lines and five E50K lines) were plated into 96-well black-walled, clear-bottom imaging plates (PerkinElmer) at a density of 10,000 cells/well one day prior to imaging. On the day of the experiment, cells were incubated with TMRM (20 nM final concentration) prepared in modified Hank’s balanced salt solution (HBSS: NaCl 137 mM, Na₂HPO₄ 0.35 mM, KCl 5.4 mM, KH₂PO₄ 1 mM, MgSO₄ 0.81 mM, CaCl₂ 0.95 mM, glucose 5.5 mM, NaHCO₃ 25 mM, HEPES 20 mM, pH 7.4) for 30 min at 37 °C, 5% CO₂. Following incubation, cells were washed twice with warm HBSS and imaged immediately. Initial live-cell imaging was performed on the Opera Phenix^™ Plus high-content imaging system (PerkinElmer) at 37 °C and 5% CO₂. For each cell, four z-stacks were acquired (step size ~2 μm), and maximum intensity projections were generated in Fiji/ImageJ. ROIs corresponding to individual cells were defined using brightfield images, and mean fluorescence intensity (MFI) of TMRM was measured per ROI, with background subtraction.

For time-course experiments, oligomycin (2 μM) was added at 4 minutes and FCCP (10 μM) at 6 minutes, with imaging every minute. Four pre-oligomycin, two post-oligomycin, and three post-FCCP time points were analyzed. MFI was expressed as percent change from baseline, defined as the mean of the four pre-oligomycin measurements.

Mitochondrial membrane potential imaging and time-course experiments using dexpramipexole and Zeiss Airyscan.

To assess the effect of dexpramipexole on mitochondrial membrane potential dynamics, the time-course experiment was repeated on a Zeiss Airyscan confocal system. Oligomycin (final 2 μM) was added at 2 minutes, and FCCP (final 10 μM) was added at 5 minutes. MFI measurements were processed as described above. Sample sizes were: n = 8 images from 2 control lines; n = 6 images from 3 Dex-treated control lines; n = 8 images from 3 E50K lines; and n = 12 images from 3 Dex-treated E50K lines. For experiments involving dexpramipexole, fibroblasts were pre-treated with 10 μM DEX for 8 hours prior to imaging.

Application of Seahorse XFe24 to measure oxygen consumption in fibroblasts

Oxygen consumption rate (OCR) was quantified using the XFe24 Analyzer (Agilent Technologies) according to the manufacturer’s instructions.

Day before the experiment

The XFe24 Sensor Cartridge (Agilent) was hydrated overnight at 37 °C in a non-CO₂ incubator. Based on optimization experiments, fibroblasts were plated at 20,000 cells per well in XFe24 cell culture plates. Cells were harvested by trypsinisation and resuspended in complete tissue culture medium. Cell number and viability were determined by trypan blue exclusion using a haemocytometer. Cell suspensions were then diluted to the required concentration and re-counted to confirm accuracy. For plating, 100 μl of the cell suspension was added to each well of the XFe24 culture plate. The plate was incubated at 37 °C in a CO₂ incubator for 1 hour to allow cells to adhere, after which 150 μl of warm complete growth medium was added to each well. The plate was incubated overnight at 37 °C with CO₂.

Day of the experiment

On the day of the assay, culture medium was aspirated and cells were washed twice with pre-warmed Seahorse Base medium (Agilent) supplemented with 10 mM glucose, 1 mM sodium pyruvate, and 2 mM glutamine (final concentration), adjusted to pH 7.4. After washing, each well received 500 μl of the same assay medium, and the plate was incubated for 60 minutes in a 37 °C non-CO₂ incubator. OCR was measured under basal conditions and following sequential injections of oligomycin (1.5 μM), carbonylcyanide-4-trifluoromethoxyphenylhydrazone (FCCP, 1.5 μM), and antimycin A/rotenone (1.6 μM / 16 μM). Concentrations of inhibitors were determined from prior titration experiments.

Experimental design

A total of 12 fibroblast lines were assessed (6 control, 6 E50K). Each cell line was tested in two independent experiments performed on separate culture preparations. Within each run, every cell line was plated in four technical replicate wells.

Normalization

Following the Seahorse assay, 0.5 μl of DAPI (final concentration 8 μg/ml) was added to each well of the XFe24 plate, which was incubated at 37 °C for 20 minutes. Cells were imaged and counted using the Cytation1 Imaging plate reader (Biotek Instruments Inc., Agilent). OCR values were normalized to the number of cells per well and are reported as pmol/min/10,000 cells.

Glycolytic rate assay – Seahorse XFe24 Analyzer

Glycolytic rate was measured using the same cell preparation and plating conditions as the OCR assay described above. The assay was performed using the XFe24 Analyzer (Agilent Technologies) with sequential injections of antimycin A and rotenone (0.5 μM final) to inhibit the electron transport chain, followed by 2-deoxyglucose (2-DG, 50 mM final) to inhibit glycolysis. These concentrations have been previously shown to fully inhibit mitochondrial respiration and glycolysis^88,89. Proton efflux rate (PER) was calculated as an indicator of glycolytic activity. Basal glycolytic rate, as well as compensatory glycolysis following ETC inhibition, were derived from these measurements. Each cell line (6 control, 6 E50K) was assayed in two independent experiments, with a minimum of three technical replicate wells per run. With the exception of the inhibitors used, all other experimental steps and conditions were identical to those of the OCR assay.

Lactate dehydrogenase B activity

LDHB activity was measured using the Sigma LDHB Assay Kit (Sigma-Aldrich, Poole, UK, MAK066), which quantifies the reduction of NAD⁺ to NADH by LDHB⁹⁰. Experiments were performed in duplicate. Fibroblasts were trypsinised, and cell pellets were rapidly homogenized on ice in 500 μL of cold LDHB assay buffer. Samples were centrifuged at 10,000 × g for 15 minutes at 4 °C to remove insoluble material. 1 μL of the soluble fraction was diluted in 9 μL of LDHB assay buffer. From this, 1 μL and 2 μL were transferred into separate wells of a 96-well clear, flat-bottom plate and brought to 50 μL with assay buffer. Next, 50 μL of master reaction mix (48 μL LDHB assay buffer + 2 μL LDHB substrate mix) was added to each well and mixed by pipetting. A standard curve (1.25 mM NADH, in duplicate) was included in each run. Plates were protected from light and absorbance measured every 5 minutes at 450 nm using the Cytation 1 (Biotek Instruments Inc., Agilent) preheated to 37 °C. The assay was stopped once the highest-activity sample exceeded the top standard (12.5 nmol/well). Background values from blank wells were subtracted, and LDHB activity was calculated as nmol/min/mL/mg protein (milliunit/mL/mg protein). Protein content was measured using the Pierce^™ BCA Protein Assay Kit. This protocol was titrated and validated for fibroblasts as described by Protasoni and Taanman (2023)⁹¹.

L-lactate measurement in fibroblast media

L-lactate levels were measured in the cell culture media of six E50K and six control fibroblast lines. Each line was cultured in triplicate (three separate plates), and media samples (100 μL) were collected on days 1, 3, and 7. Media from three cell-free plates was collected in parallel as blanks. Samples were promptly frozen at −80 °C until analysis one week later. Initial tests were performed to determine the linear range of the assay (to establish appropriate media dilution) and to identify the optimal time point for measurement, which was determined to be day 3. For the main experiment, 1.5 μL of frozen media per sample was used. Each sample run (from three plates) included two technical replicates as serial dilutions, yielding a total of six replicates per cell line. On the day of the assay, 3 μL of media was mixed with 47 μL of assay buffer, and 25 μL of this mixture was dispensed into two wells of a 96-well clear, flat-bottom plate. A standard curve was prepared in each run using a fresh 1 mM L-lactate stock. All wells, including blanks (media from plates without cells), were topped with 25 μL of reaction mix (23 μL assay buffer, 1 μL substrate mix, 1 μL enzyme mix). Blanks received 2 μL substrate mix in place of enzyme. Plates were mixed and incubated at room temperature for 30 minutes, and absorbance was measured at 450 nm using the Cytation 1 (Biotek Instruments Inc., Agilent). L-lactate concentration was calculated using the formula:

$$\mathit{\text{Lactate}}\mathit{\text{concentration}}=\left(\frac{La}{Sv}\right)*D$$

La = amount of Lactic acid in the sample well calculated from standard curve (nmol).

Sv = volume of sample added into the well (μL).

D = sample dilution factor.

Citrate Synthase (CS) activity assay

CS activity was measured in five E50K and five control fibroblast lines. A minimum of two serial dilutions per cell line/sample was performed, and the experiment was repeated twice using cells grown in separate plates. Cells were trypsinised, collected as pellets, and washed twice with PBS. Pellets were resuspended in 500 μL of dH₂O, and 5 μL and 10 μL of this suspension were dispensed into wells of a 96-well clear, flat-bottom plate and topped up to 250 μL with reaction mixture. The reaction mixture consisted of 25 μL of 1 M Tris-HCl (pH 8), 5 μL of 10 mM Acetyl Coenzyme A, 2.5 μL of 10% Triton X-100, 5 μL of 10 mM DTNB, 2.5 μL of 10 mM oxaloacetate (OAA), and ddH₂O to reach the final volume. Absorbance at 412 nm was recorded every minute for 30 minutes using the Cytation 1 (Biotek Instruments Inc., Agilent) preheated to 30 °C. CS activity was determined by the rate of color development of TNB, generated from the reaction of DTNB during citrate synthesis. TNB formation was quantified by its absorbance at 412 nm. CS activity was normalized to total protein content, measured using the Pierce^™ BCA Protein Assay Kit. For each cell line, technical replicates were averaged, and results are reported as μM of citrate formed per minute per mg of protein

Pyruvate dehydrogenase (PDH) activity

PDH activity was measured using the Abcam PDH assay kit (Abcam, ab287837). PDH catalyzes the conversion of pyruvate to an intermediate product, which reduces the developer to a colored product detectable at 450 nm. This reaction is accompanied by the reduction of NAD⁺ to NADH, allowing PDH activity to be expressed as nmol of NADH produced over time. The assay was conducted on four E50K and five control fibroblast lines, with a minimum of two technical replicates per cell line. The experiment was repeated twice using cells grown in separate plates. Cells were trypsinised, washed with ice-cold PBS, and pellets were resuspended in 60 μL of ice-cold Assay Buffer, followed by 10 min on ice. Samples were centrifuged at 10,000 × g for 15 min at 4 °C, and 25 μL of supernatant was transferred into two wells of a 96-well clear, flat-bottom plate and topped up to 50 μL with Assay Buffer. A standard curve was generated using 0, 2, 4, 6, 8, and 10 μL of 1.25 mM NADH standard, topped up to 50 μL with Assay Buffer. Finally, 50 μL of reaction mixture (48 μL Assay Buffer, 2 μL Developer, 2 μL PDH substrate) was added to each well. Absorbance at 450 nm was recorded immediately in kinetic mode every minute for 45 minutes at 37 °C using the Cytation 1 (Biotek Instruments Inc., Agilent). Protein content was measured using 5 μL of the original cell suspension with the Pierce^™ BCA Protein Assay Kit. PDH activity was normalized to protein and expressed as nmol/min of NADH produced per mg of protein (nmol/min/mg protein).

ATP levels

ATP levels were measured using the ATPlite assay (Perkin Elmer, Waltham, MA, USA) according to the manufacturer’s instructions. Six control and six E50K fibroblast lines were plated in 96-well tissue culture plates at 10,000 cells/well in 200 μL of complete culture media and incubated overnight at 37 °C, 5% CO₂. The following day, half of the wells for each cell line were treated with Oligomycin (5 μM final) for 20 minutes to inhibit ATP synthase, allowing ATP production to be predominantly glycolytic, while the other half remained untreated. After incubation, media was aspirated, wells were washed twice with PBS, and 100 μL of PBS plus 50 μL of lysis buffer were added per well. Plates were shaken at 700 rpm for 5 minutes to lyse cells and stabilize ATP. Plates were then centrifuged at maximum speed for 1 minute, followed by addition of 50 μL substrate solution and an additional 5 minute shake at 700 rpm. After a final centrifugation at maximum speed for 1 minute, contents were transferred to a 96-well white flat-bottom plate, dark-adapted for 10 minutes, and luminescence was recorded using the Cytation 1 (Biotek Instruments Inc., Agilent). A standard curve was prepared using a 10 mM ATP stock to convert luminescence to ATP concentration (nM). ATP values were normalized to protein content measured from 5 μL of the original lysate using the Pierce^™ BCA Protein Assay Kit. Each cell line was assayed with a minimum of two technical replicates, and the experiment was independently repeated twice.

Cell growth assay

Cell growth curves were generated for E50K (n = 4) and control (n = 6) fibroblast cell lines over 7 days, with measurements taken on days 1, 3, and 7. Each cell line was represented by 8 technical replicates per experiment. Cells were seeded in 96-well plates with 200 μL of complete, phenol red–free DMEM per well (supplemented with 10% FBS, 1:200 Penicillin/Streptomycin, 1:100 MEM Non-Essential Amino Acids, 1:100). Cells were stained with Hoechst 33342 (8 μg/mL final concentration) and imaged using the Cytation 1 (Biotek Instruments Inc., Agilent). Oligomycin treatment (50 nM final) was applied to test the effect of ATP synthase inhibition on growth.

Submitochondrial vesicle (SMV) preparation and electrophysiology

Fibroblasts were grown in 20-cm culture dishes and harvested by trypsinisation. All steps were performed at 4 °C or on ice unless otherwise stated. After washing once with PBS, cells were detached and collected by centrifugation at 1,000 × g for 5 min. The pellet was washed with PBS and centrifuged again at 1,000 × g for 5 min, before being resuspended in five volumes of ice-cold extraction buffer (0.25 M sucrose, 20 mM HEPES-KOH pH 7.5, 10 mM KCl, 1.5 mM MgCl₂, 1 mM EDTA, 1 mM EGTA, 1 mM dithiothreitol, 0.1 mM PMSF). Cells were homogenized in a 5-mL Teflon/glass Potter-Elvehjem homogenizer (clearance 0.1–0.15 mm) with 10–20 strokes of the pestle. The homogenate was centrifuged at 750 × g for 10 min and the supernatant collected, while the pellet was re-homogenised and centrifuged under the same conditions. Supernatants from both spins were pooled and centrifuged at 10,000 × g for 15 min to yield a crude mitochondrial pellet, which was resuspended in extraction buffer (~20–25 μL) and stored at −80°C. Crude mitochondria were resuspended in 200 μL of Isolation Buffer (250 mM sucrose, 20 mM HEPES, 1 mM EDTA, 0.5% BSA, pH 7.4) combined with an equal volume of 1% digitonin and incubated on ice for 15 min. Additional Isolation Buffer was added and the sample centrifuged at 16,000 × g for 10 min; this wash step was repeated once. The pellet was resuspended in 200 μL Isolation Buffer, and 2 μL of 10% Lubrol PX (C12E9) was added. After mixing, the suspension was incubated on ice for 15 min. The mixture was layered onto Isolation Buffer and centrifuged in a SW-50.1 rotor at 182,000 × g for 1 h. The final pellet was washed once in Isolation Buffer (16,000 × g for 10 min) and resuspended in Isolation Buffer. Samples were either used immediately for single-channel electrophysiological recordings or stored at −80°C until use.

For recording, 2 μl of the SMVs suspension was placed in a 2 cm plate, left to settle for 10 minutes, and then topped up with 1 ml of mitochondrial recording buffer (120 mM KCl, 8 mM NaCl, 0.5 mM EGTA, 10 mM HEPES). The electrophysiology rig included an Axopatch 200B amplifier (Axon Instruments), a PC equipped with a Digidata 1440A A/D converter and pClamp10.0 software, manipulators, microscope, vibration isolation table, and Faraday cage. Recording electrodes were pulled from borosilicate glass capillaries (WPI) using a Flaming/Brown Micropipette Puller Model P-87, optimized to generate pipettes with resistances of ~50–100 MΩ. Patch-clamp recordings of SMVs were performed by first forming a giga-ohm seal on the SMV membrane. Recordings were made at room temperature (22–25 °C) in voltage-clamp mode, and current signals were filtered at 5 kHz using the amplifier circuitry. During recordings, ATP (1 mM final, added to the bath) and the Bcl-xL inhibitor WEHI-539 (5 μM final, added sequentially) were applied without perfusion. pCLAMP-10 software was used for electrophysiology data acquisition and analysis (Molecular devices). All current measurements were adjusted for the holding voltage assuming a linear current-voltage relationship: The resulting conductances are expressed in pS according to the equation G = I/V where G is conductance in pS, V is the membrane holding voltage in mV, and I is the peak membrane current in pA after subtraction of the baseline electrode leak current (the current remaining after ATP addition). Group data were quantified in terms of peak conductance and channel open to closed ratio. Control SMVs (n = 6) from 3 participants (2 SMVs per line); E50K SMVs (n = 7) from 3 participants.

Mitochondrial lysate preparation for Western Blotting

Mitochondrial lysates were prepared using the Qproteome Mitochondria Isolation Kit (Qiagen, USA). Fibroblasts were cultured in 20 cm plates to ~80% confluency, then trypsinised and centrifuged at 500 × g for 10 min at 4 °C. The cell pellet was washed with 1 mL ice-cold 0.9% NaCl, resuspended in 1 mL ice-cold lysis buffer containing 1/1000 of 100× protease inhibitor, and incubated for 10 min at 4 °C on an end-over-end shaker. Lysates were centrifuged at 1000 × g for 10 min at 4 °C to separate the cytosolic fraction, which was collected and stored at −80 °C. The remaining pellet was resuspended in 1.5 mL ice-cold Disruption Buffer and homogenized using a pipette followed by a blunt-ended needle and syringe. The suspension was centrifuged at 1000 × g for 10 min at 4 °C, and the supernatant was transferred to a fresh tube and centrifuged at 6,000 × g for 10 min at 4 °C. The resulting pellet contained mitochondria, which were washed with 1 mL Mitochondria Storage Buffer and centrifuged at 6,000 × g for 20 min at 4 °C. Finally, the mitochondrial pellet was resuspended in 100 μL 1× RIPA buffer with 1× protease inhibitor (Halt^™ Protease Inhibitor Cocktail (100X), Cat: 78430) and stored at −80 °C for downstream Western Blot analysis.

Cell Lysate Preparation for Western Blotting

Fibroblasts were grown to ~80% confluency in 10–20 cm tissue culture dishes. Cells were washed twice with ice-cold PBS and lysed in RIPA buffer (50 mM Tris-HCl, pH 7.4; 150 mM NaCl; 1% NP-40; 0.5% sodium deoxycholate; 0.1% SDS) supplemented with protease inhibitor cocktail (Halt^™ Protease Inhibitor Cocktail (100X), Cat: 78430). Lysates were incubated on ice for 30 min with intermittent mixing and clarified by centrifugation at 14,000 × g for 15 min at 4 °C. The supernatant containing total cellular protein was collected, protein concentration determined using the Pierce^™ BCA Protein Assay Kit, and aliquots stored at −80 °C until Western blot analysis. For experiments involving dexpramipexole, fibroblasts were pre-treated with 10 μM DEX for 4, 8 and 24 hours prior to lysing.

Western blot

Mitochondrial and total cell lysates were prepared as described above. Protein concentrations were determined using the Pierce^™ BCA Protein Assay Kit (Cat: 23227, ThermoFisher). Lysates (20–30 μg per lane) were separated on 4%–20% gradient SDS-PAGE gels (Bio-Rad, USA) and transferred overnight to PVDF membranes (0.2 μm pores, Bio-Rad, USA). Membranes were blocked in 5% bovine serum albumin (BSA) in Tris-buffered saline with 0.1% Tween 20 (TBST) for 1 h at room temperature. BSA serves as a protein-based blocking agent to prevent non-specific binding of antibodies.

Membranes were incubated overnight at 4 °C with the following primary antibodies:

• Anti-ATP synthase C subunit [EPR13907] (Abcam, ab181243, 1:1000)

• Anti-ATP β subunit [3D5] (Abcam, ab14730, 1:1000)

• Anti-SDHA [2E3GC12FB2AE2] (Abcam, ab14715, 1:1000)

• LC3B (Cell Signaling Technologies, #2775, 1:1000)

• SQSTM1/p62 (Cell Signaling Technologies, #5114, 1:1000)

• MDH2 (Genetex, GTX105870, 1:1000)

• PFKP (ThermoFisher, PA5–28673, 1:1000)

• β-actin (ThermoFisher, PA1–183, 1:5000)

Following three washes in TBST (10 min each), membranes were incubated for 1 h at room temperature with HRP-conjugated secondary antibodies (1:10,000, Cell Signaling, Anti-rabbit IgG #7074 and Anti-mouse IgG #7076). Protein bands were visualized using the SuperSignal^™ West Pico PLUS Chemiluminescent Substrate (Cat: 34580, ThermoFisher) and imaged. Band intensities were quantified using ImageJ, with grey densities measured for each band.

Puromycin Labelling for Protein Synthesis

To measure protein synthesis, fibroblasts were incubated in medium containing puromycin dihydrochloride (10 μg/mL, Sigma-Aldrich) for 15 minutes at 37 °C. Following incubation, cells were washed twice with cold PBS and lysed in 1× RIPA buffer supplemented with 1/1000 volume of 100× protease inhibitor cocktail (Halt^™, ThermoFisher). Cells were scraped, collected into 1.5 mL Eppendorf tubes, and centrifuged at 14,000 rpm for 10 minutes at 4 °C. The supernatant was transferred to a fresh tube, and protein concentration was determined using the Pierce^™ BCA Protein Assay Kit (ThermoFisher). Remaining lysates were stored at −80 °C until use for Western blotting. Western blotting was performed as described above. Lysates from cells grown in three independent plates were collected for each cell line. Primary and secondary antibodies used for detection were:

• Anti-puromycin [3RH11], Mouse IgG1, Kappa – Absolute Antibody (Ab02366–1.1, 1:1000)

• Anti-GAPDH, Rabbit polyclonal – Abcam (ab9485, 1:1000)

• Anti-mouse IgG, HRP-linked Antibody – Cell Signaling Technologies (#7076, 1:10,000)

• Anti-rabbit IgG, HRP-linked Antibody – Cell Signaling Technologies (#7074, 1:10,000)

Puromycin incorporation into nascent polypeptides was detected via chemiluminescence, with GAPDH used as a loading control. For experiments involving dexpramipexole, fibroblasts were pre-treated with 10 μM DEX for 8 hours prior to fixation.

Immunoprecipitation of puromycin labelled peptides.

Puromycin immunoprecipitation was performed using a protocol developed by Pawel Licznerski in the Jonas laboratory¹⁷. Following puromycin labelling as described above, an appropriate volume of cell lysate in 1× RIPA buffer containing 80 μg of protein was used for immunoprecipitation with 1 μg of anti-puromycin antibody (Absolute Antibody, Ab02366–1.1) overnight at 4 °C on a rotating shaker. On the following day, 40 μL of Pierce^™ Protein A/G magnetic beads (ThermoFisher, Cat. 88802) were added to each sample and incubated overnight at 4 °C with gentle rotation. After incubation, the beads were separated using a magnetic rack, the supernatant carefully removed, and the bead–antibody–protein complexes washed three times with 600 μL of ice-cold 1× RIPA buffer containing protease inhibitor cocktail. Beads were resuspended in 20 μL of Laemmli buffer supplemented with β-mercaptoethanol and heated at 95 °C for 5 minutes. Samples were loaded on SDS-PAGE gels and electrophoresed at 150 V until the protein front had just entered the separating gel. At this point, gels were stopped, cut, and stored at −80 °C before submission for proteomic analysis at the Yale/Keck MS & Proteomics Resource center. For experiments involving dexpramipexole, fibroblasts were pre-treated with 10 μM DEX for 8 hours prior to puromycin labelling.

Mass Spectrometry-proteomics

Raw MS/MS data were processed in Proteome Discoverer v3.0.1.27 using a precursor-based quantification and label-free workflow with CHIMERYS (inferys_2.1_fragmentation) for peptide identification and Percolator validation. Spectra were searched against the Homo sapiens UniProt reference proteome (October 2024 release) with an added contaminant database. Trypsin digestion was specified with up to two missed cleavages, precursor mass tolerance of 20 ppm, fragment mass tolerance of 0.02 Da, carbamidomethylation of cysteine as a static modification, and methionine oxidation as a dynamic modification. Peptide spectral matches (PSMs) were validated using Percolator with a concatenated target–decoy strategy, achieving a false discovery rate (FDR) of 1% at the PSM, peptide, and protein levels. Only peptides of ≥6 amino acids with high confidence were considered, and protein grouping was restricted to “leaf” master proteins after removal of contaminants. In total, 78,189 PSMs (of 440,546 total), 8,394 peptide groups (of 85,301 total), and 2,041 protein groups passed high-confidence filters, yielding 1,877 master proteins. Across all runs, 494,531 MS/MS spectra were acquired. Label-free quantification was performed using the Minora Feature Detector and Precursor Ions Quantifier nodes, considering only unique peptides, with normalization to total peptide abundance per channel and protein-level rollup using summed intensities; average peptide and protein abundance was scaled to 100. Quantification ratios were calculated for the following contrasts: (Control + Dex)/Control, E50K/Control, (E50K + Dex)/(Control + Dex), and (E50K + Dex)/E50K, with log₂ fold changes computed and |log₂FC| ≥1 (≥2-fold) considered biologically relevant. The resulting normalized abundance table of 2,378 proteins was exported and analyzed using GSEA v4.3.2 with MSigDB Hallmark gene sets, using gene symbols as identifiers, 1,000 permutations, and the classic enrichment statistic. Significance of enrichment was assessed using the normalized enrichment score (NES) and false discovery rate (FDR), with pathways considered enriched at FDR q < 0.25. Comparisons were performed for E50K versus control and E50K + dexpramipexole versus control samples.

Serum metabolomics

Blood was collected via venipuncture from 6 E50K participants and 20 non-glaucomatous controls into BD Vacutainer serum tubes (no anticoagulant). Tubes were allowed to clot at room temperature for around 60 minutes. Following clotting, tubes were centrifuged at 800 × g for 10 minutes at 4 °C. The clear serum was carefully transferred into two separate 1.5 mL Eppendorf tubes. Serum was centrifuged a second time at 1000× g for 10 minutes at 4 °C. The resulting clear serum was aliquoted and stored at −80 °C until further use. A 100 μL aliquot from each sample was sent for metabolomics analysis. Metabolic profiling by LC-MS was performed at the Swedish Metabolomics Centre in Umeå, Sweden. Solvents and eluent additives: Methanol and Acetonitrile, HPLC-grade were obtained from Fischer Scientific (Waltham, MA, USA), Ammonium formate and Medronic acid (Methylenediphosphonic acid) were obtained from Sigma (St. Louis, MO, USA). Reference and tuning standards: Purine, 4 μM, HP-0921 (Hexakis(1H, 1H, 3H-tetrafluoropropoxy)phosphazine), 1 μM, Calibrant, ESI-TOF, ESI-L Low Concentration Tuning Mix, HP-0321 (Hexamethoxyphosphazine), 0.1 mM were obtained from Agilent Technologies (Santa Clara, CA, USA). Stable isotope internal standards: 13C9-Phenylalanine, D4-Cholic acid, Salicylic acid-D6, L-glutamic acid-13C5,15N and D-sucrose-13C12 were obtained from Sigma (St. Louis, MO, USA). L-proline-13C5 and alpha-ketoglutarate-13C4 were obtained from Cil (Andover, MA, USA). Sample preparation of 100 μl serum was preformed according to A. et al. (2005)⁹². In detail, 900 μl of extraction buffer (90/10 v/v methanol: water) including internal standards were added to the samples. The samples were shaken at 30 Hz for 2 minutes in a mixer mill and proteins were precipitated at −20 °C for 2 hours. The samples were centrifuged at +4 °C, 14 000 rpm (18 620 g), for 10 minutes. 200μL supernatant was transferred to micro vials and evaporated to dryness in a speed-vac concentrator (P0776 plasma: evaporated under a stream of nitrogen instead of a speed-vac concentrator) and the samples were stored at −80 °C until analysis. Small aliquots of the remaining supernatants were pooled and used to create quality control (QC) samples. Before analysis the samples were resuspended in 10 + 10 μL methanol and elution solvent A. The samples were analyzed in batches according to a randomized run order. Each batch of samples was first analyzed in positive mode. After all samples within a batch had been analyzed, the instrument was switched to negative mode and a second injection of each sample was performed. The chromatographic separation was performed on an Agilent 1290 Infinity UHPLC-system (Agilent Technologies, Waldbronn, Germany). 2 μL (1 μl for P0748) of each sample were injected onto an Atlantis Premier BEH-Z-HILIC VanGuard FIT (1.7 μm, 2.1 × 50 mm) column (Waters Corporation, Milford, MA, USA) held at 40 °C. The HILIC gradient elution solvents were A) H2O, 10 mM ammonium formate, 5μM Medronic acid, pH 9, B) 90:10 Acetonitrile: H2O, 10 mM ammonium formate, pH 9 (5 μM Medronic acid also in elution solvent B for P0789 and P0803). Chromatographic separation was achieved using a linear gradient (at a flow rate of 0.4 mL min-1) starting at 90% B decreasing to 80% B over 6 minutes; B was decreased to 20 % over 3.5 minutes and held at 20 % for 1.5 minutes; B was increased to 90 % for 0.5 minutes and the flow-rate was increased to 0.7 mL min-1 for 2 minutes; these conditions were held for 0.5 minutes, after which the flow-rate was reduced to 0.4 mL min-1 for 0.5 minutes before the next injection. The compounds were detected with an Agilent 6546 Q-TOF mass spectrometer equipped with a jet stream electrospray ion source operating in positive or negative ion mode. A reference interface was connected for accurate mass measurements; the reference ions purine (4 μM) and HP-0921 (Hexakis(1H, 1H, 3H-tetrafluoropropoxy)phosphazine) (1 μM) were infused directly into the MS at a flow rate of 0.05 mL min-1 for internal calibration, and the monitored ions were purine m/z 121.05 and m/z 119.03632; HP-0921 m/z 922.0098 and m/z 966.000725 for positive and negative mode respectively. The gas temperature was set to 150°C, the drying gas flow to 8 L min-1 and the nebulizer pressure 35 psi. The sheath gas temp was set to 350°C and the sheath gas flow 11 L min-1. The capillary voltage was set to 4000 V in both positive and negative ion mode. The nozzle voltage was 300 V. The fragmentor voltage was 120 V, the skimmer 65 V and the OCT 1 RF Vpp 750 V. The collision energy was set to 0 V. The m/z range was 70 – 1700, and data was collected in centroid mode with an acquisition rate of 4 scans s-1. MSMS analysis was run on the QC samples for identification purposes. All data pre-processing was performed using the Agilent MassHunter Profinder version B.10.0 SP1 (Agilent Technologies Inc., Santa Clara, CA, USA). The data pre-processing was performed in a targeted fashion. A pre-defined list of metabolites commonly found in plasma and serum were searched for using the Batch Targeted feature extraction in MassHunter Profinder. An-in-house LC-MS library built up by authentic standards run on the same system with the same chromatographic and mass-spec settings, were used for the targeted processing.

Multivariate modeling was performed using Orthogonal partial least squares discriminant analysis (orthoPLS-DA) to evaluate global metabolic separation between E50K and control participants. Metabolomics data were first processed and normalized prior to statistical analysis. Differential expression analysis was performed using the LIMMA (Linear Models for Microarray Data) method, which identified metabolites with significant fold changes between groups. Log2 fold change (FC) values and associated adjusted p-values were obtained from this analysis. The resulting list of significantly altered metabolites was then uploaded into Ingenuity Pathway Analysis (IPA, Qiagen) for functional enrichment and pathway analysis. IPA computes canonical pathways based on both the p-value of overlap (Fisher’s exact test) and the pathway ratio (number of input molecules relative to total molecules in the pathway). In addition, integrated analysis of metabolites and proteins was performed using the. MetaboAnalyst 5.0 platform. Differentially expressed metabolites and proteins were combined, and features with absolute fold change ≥ 1.5 (corresponding to log2_22FC ≥ 0.58) were selected. Pathway topology analysis was then conducted, reporting both pathway impact scores and associated p-values for enrichment significance.

Supplementary Material

Supplementary Figure 1. (A) Mitochondrial respiration in control and E50K fibroblasts assessed using Seahorse XFe24 extracellular flux analysis. Oxygen consumption rate (OCR) traces represent mean ± SEM from six independent control and six E50K cell lines, with values normalized to both cell number and mitochondrial mass. Mitochondrial mass was quantified based on expression of a mitochondrially targeted GFP construct, which localizes to the mitochondrial matrix independently of membrane potential. Sequential injections of oligomycin (ATP synthase inhibitor) and FCCP (carbonyl cyanide-4-(trifluoromethoxy)phenylhydrazone, a mitochondrial uncoupler) and a combination of antimycin A and rotenone (complex III and I inhibitors, respectively) were used to evaluate key respiratory parameters. E50K fibroblasts exhibited lower baseline respiration (pre-oligomycin) and reduced maximal respiratory capacity (post-FCCP) compared to controls. (B) Mitochondrial respiration in control and E50K fibroblasts, assessed by Seahorse XFe24 extracellular flux analysis and normalized to MitoTracker^™ Red fluorescence. Oxygen consumption rate (OCR) traces represent mean ± SEM from six independent control and six E50K cell lines. MitoTracker^™ Red accumulates in mitochondria proportionally to membrane potential, providing a functional readout of polarized, active mitochondria. Sequential injections of oligomycin (ATP synthase inhibitor), FCCP (carbonyl cyanide-4-(trifluoromethoxy)phenylhydrazone, a mitochondrial uncoupler), and a combination of antimycin A and rotenone (complex III and I inhibitors, respectively) were used to evaluate key respiratory parameters. Baseline respiration (pre-oligomycin), maximal respiration (post-FCCP), and proton leak, calculated as the difference between OCR after oligomycin and OCR after antimycin A/rotenone, were all significantly elevated in E50K fibroblasts compared to controls. OCR values normalized to MitoTracker^™ Red were markedly higher than those normalized to cell number alone, indicating that mitochondria retaining membrane potential in E50K cells are hyperactive. This suggests a compensatory response, in which a subset of polarized mitochondria exhibit elevated respiratory activity despite global mitochondrial dysfunction and increased uncoupling. (C) Principal component analysis (PCA) of newly synthesized proteomes in control and E50K fibroblasts. Proteomic profiles were obtained from three independent control and three E50K participants. The plot shows clear separation between groups, with principal component 1 (Dim 1) and principal component 2 (Dim 2) explaining 50.6% and 23.8% of the total variance, respectively, highlighting distinct translational signatures in E50K cells. (D) Orthogonal partial least squares discriminant analysis (orthoPLS-DA) of targeted serum metabolomic profiles in 6 E50K patients and 18 age-matched non-glaucomatous controls. The x-axis shows the predictive component (T score 1 = 6.6%) representing class-discriminating variance, while the y-axis reflects orthogonal variation (orthogonal T score = 13.7%) unrelated to class separation. Despite modest explained variance, the model reveals clear group separation, supporting distinct metabolic signatures in E50K serum. (E) Principal component analysis (PCA) of puromycin-labeled newly synthesized proteomes from three groups: control, E50K, and E50K + dexpramipexole (Dex) treated fibroblasts (3 cell lines per group). The plot shows clear group separation, with principal component 1 (Dim 1) explaining 53% of the variance and principal component 2 (Dim 2) explaining 16%, indicating distinct translational profiles among the groups.

Availability of Data and material:

Human fibroblasts are available from some of the authors (BP and DFGH), only for non-commercial use and only for research purposes, upon reasonable request, other data and materials available after publication.