Interaction between cell stress and energy control as a new approach to eye disease treatment

The recognition that UPR transducers exert non‐canonical functions at MAMs, where ER and mitochondrial membranes are apposed is a major advancement in this field.17, 42 At these contact sites, UPR components directly affect mitochondrial Ca²⁺ handling, bioenergetics, and cell‐fate decisions, linking ER proteostasis to mitochondrial function.43

Beyond its canonical role in XBP1 splicing, IRE1α acts as a structural regulator of ER–mitochondria contacts. Loss of IRE1α disrupts the spatial organization of inositol 1,4,5‐trisphosphate receptors (IP₃Rs) at the MAMs, which results in impaired mitochondrial Ca²⁺ uptake and altered basal bioenergetics. These effects are independent of canonical UPR transcriptional outputs.17, 44, 45

PERK is also enriched at MAMs and contributes to adaptive and maladaptive ER–mitochondria signaling.9 Through interactions with mitochondrial tethering proteins, such as mitofusin‐2 (MFN2), PERK stabilizes ER–mitochondria contacts during stress conditions.16 Sustained PERK activation propagates oxidative stress and apoptosis through CHOP expression and disrupted Ca²⁺ homeostasis.15, 46

ATF6 is a transcriptional regulator of ER folding capacity and lipid metabolism; however, it has emerging roles in the epithelial–mesenchymal transition (EMT) and stress adaptation in ocular cells.24 These affect mitochondrial function indirectly by reshaping cellular metabolism and redox balance.

Taken together, these results support a model in which UPR sensors act not only as ER stress detectors, but also as functional nodes at ER–mitochondria interfaces, by translating the proteostasis imbalance into mitochondrial dysfunction during chronic stress.

Instead of being uniform or static interfaces, MAMs are organized into discrete and functionally specialized molecular modules whose composition, spacing, and signaling output are dynamically regulated. Structural, biochemical, and imaging studies indicate that ER–mitochondria contacts can be divided into three interdependent, but mechanistically distinct axes: Ca²⁺ transfer, mitochondrial dynamics, and cell‐death/quality‐control signaling. Together, they determine mitochondrial fitness and cell fate under certain physiological and pathological conditions (Figure 1).

During sustained metabolic, oxidative, or proteotoxic stress, ER–mitochondria contact sites undergo rapid and compositional remodeling, rather than existing as static interfaces. High‐resolution imaging and single‐molecule tracking revealed that contact sites consist of dynamic subdomains and can expand or reorganize to accommodate changing physiological demands, such as nutrient stress.71 Consistent with this plasticity, UPR transducers directly contribute to MAM homeostasis through non‐canonical structural functions: IRE1α acts as a scaffold to control InsP₃R distribution at the MAMs and tunes mitochondrial Ca²⁺ uptake and bioenergetics,17 whereas PERK localizes to contact sites and sustains ER–mitochondria apposition to enhance reactive oxygen species (ROS) propagation during chronic stress.15 Simultaneously, stress‐responsive post‐translational regulation of tethering complexes (e.g., VAPB–PTPIP51) reshapes coupling strength and Ca²⁺ transfer, providing a route through which disease‐linked signals, including kinase activation, disrupt ER–mitochondria communication.72, 73 Oxidative lipid stress can trigger minute‐scale expansion of ER–mitochondria contact interfaces and bias these sites toward pro‐oxidant outcomes, such as mitochondrial ROS amplification and fission, which implicates pathological MERC remodeling as an early and potentially reversible amplifier that lowers the threshold for mitochondrial failure and regulated cell death.74, 75

Cataract is characterized by the loss of proteostasis and redox balance in LECs (Figure 2A). While the general cascades of UPR activation and MAM destabilization are summarized in Table 1, specific evidence in LECs highlights a unique vulnerability to organelle‐interface disruption.9, 10, 11 At the organelle level, stress‐induced MAM destabilization is associated with mitochondrial Ca²⁺ overload and ROS amplification, a paradigm increasingly invoked as a causal mechanism for LEC apoptosis.10, 11, 12

The involvement of ER stress is supported by robust activation of BiP/GRP78 and CHOP in models such as the sodium‐selenite rat.76 Beyond apoptosis, ER stress has been implicated in the regulation of the EMT, which underlies anterior subcapsular cataract and posterior capsular opacification.77 Genetic models provide the most direct evidence of organelle crosstalk; for instance, αA‐crystallin mutations (R49C and Y118D) engage the ER–mitochondria axis, contributing to nuclear cataract, and suggesting that single‐gene misfolding may facilitate systemic organelle failure.78, 79 Human data mirror these findings, with aged donor lenses and aqueous humor proteomics showing enriched ER‐stress and metabolic dysfunction pathways.80, 81

Recent translational efforts focus on disrupting the ER‐stress–fibrosis axis. Chemical chaperones, including 4‐phenylbutyrate (PBA) and tauroursodeoxycholic acid (TUDCA), have been shown to attenuate high‐glucose–induced EMT in human LECs.22 A key mechanistic highlight is the ATF6–SNAI1 self‐amplifying loop; its inhibition via AAV‐shATF6 or melatonin attenuates fibrotic changes, positioning ATF6 as a specific druggable node at the ER–mitochondria interface.24 Additional pharmacological leads, such as SESN2 up‐regulation and mitochondria‐targeted peptides (e.g., elamipretide), offer a roadmap for non‐surgical therapy by safeguarding ATP sufficiency and avoiding CHOP‐dominated apoptosis.25, 82, 83, 84, 85 Ultimately, cataract represents a prototypical proteostasis‐driven disorder where maladaptive UPR and MAM remodeling are closely linked to terminal fibrotic remodeling.9, 12 Taken together, cataract represents a prototypical proteostasis‐driven ER–mitochondria disorder, in which maladaptive UPR activation in LECs precedes mitochondrial dysfunction and promotes cell death and fibrotic remodeling.

Glaucoma is a progressive optic neuropathy marked by the selective vulnerability of RGCs, whose long axons and high metabolic demand render them exquisitely dependent on efficient mitochondrial ATP production and QC18, 19 (Figure 2B). Beyond the general UPR signaling summarized in Table 1, the glaucomatous retina undergoes a critical shift where sustained stress converts adaptive responses into pro‐apoptotic programs, resulting in synaptic failure and axonal degeneration.9, 20 At the ER–mitochondria interface, MAMs coordinate Ca²⁺ transfer and lipid exchange. Destabilization of the MAMs during glaucomatous stress results in mitochondrial Ca²⁺ overload, ROS amplification, and the permeability transition, which reduces the threshold for RGC apoptosis.21 A second, tightly linked axis is mitochondrial dynamics, in which pathologic activation of DRP1‐mediated fission with impaired MFN1/2‐ and OPA1‐dependent fusion fragments the mitochondrial network. This reduces respiratory capacity and sensitizes RGCs to death.18, 26, 27

Evidence from rodent models suggests that increasing IOP induces mitochondrial fission and ultrastructural disruption in RGC axons prior to cell loss, positioning organelle reprogramming as an early contributing event.28 Actions such as the genetic or pharmacologic inhibition of DRP1 have been shown to preserve mitochondrial integrity and rescue RGC function, supporting the hypothesis that excessive fission is a targetable bottleneck in disease progression.27 Recent single‐cell analyses further implicate a broader failure of QC, including mtDNA damage and defective mitophagy.19 Importantly, the ER–mitochondria unit acts as an “integrated operative locus” rather than two independent organelles, where ER stress and mitochondrial failure maybe mutually reinforced via MAMs.21 For example, TrkB pathway augmentation reveals that preserving axonal transport is closely linked to the maintenance of mitochondrial output, further linking organelle crosstalk to functional survival.29

Human postmortem and surgical data are consistent with these preclinical patterns, showing widespread mitochondrial compromise and increased ER‐stress markers in the optic nerve head and inner retina.19, 30 Two complementary therapeutic strategies are emerging. First, tuning ER stress: small‐molecule PERK inhibitors (e.g., LDN‐0060609) suppress maladaptive UPR outputs in human trabecular meshwork and astroglial models associated with increased pressure and neuroinflammation. This indicates a translatable entry point upstream of CHOP‐driven death programs.31, 32 Second, stabilizing mitochondrial morphology through DRP1 inhibitors or enhancing mitophagy is associated with vision protection in preclinical systems.27, 86 Finally, the MAMs perspective suggests an integrated screen for candidates that normalize ER–mitochondria calcium homeostasis and lipid transfer. By preserving the fidelity of this physical interface, such agents could reduce ER stress, improve ATP supply, and increase the injury threshold of RGCs, even when IOP is medically controlled but progression continues.21 In summary, the disruption of ER–mitochondria signaling in RGCs represents a primary mechanism associated with axonal energy failure and progressive neurodegeneration.

Chronic hyperglycemia in the diabetic retina initiates a self‐reinforcing loop of oxidative stress, UPR activation, and MAM‐mediated mitochondrial collapse (Figure 2C). MAMs serve as a site for Ca²⁺ and lipid exchange. Under metabolic stress, altered MAM composition is associated with changes in ER‐to‐mitochondria Ca²⁺ transfer, potentially contributing to the permeability transition and bioenergetic collapse. This mechanism is supported by proteomics in diabetic retinas and mechanistic studies showing IRE1α′s scaffolding role at the MAMs.17, 87 Furthermore, hyperglycemia disrupts the PINK1/Parkin‐mediated mitophagy pathway, compounding ROS and inflammasome signaling across vascular, glial, and neuronal compartments.88, 89, 90, 91

Illustrative mechanistic studies highlight specific molecular nodes that govern this organelle crosstalk. Retinal MAM proteomics reveal extensive interface remodeling involving over 180 differentially regulated proteins.87 In Müller cells, the ER‐resident antioxidant PRDX4 represents a key node. PRDX4 deficiency is linked to accelerated STZ‐induced retinal pathology, whereas PRDX4 overexpression is associated with reduced reactive gliosis, apoptosis, ER stress, oxidative stress, and mitochondrial dysfunction. These findings suggest that strengthening ER redox control may support mitochondrial stability in the diabetic retina.33 Beyond traditional UPR signaling, immune co‐stimulation via the CD40–TRAF2/3 axis intersects with ER stress to drive vascular leakage, linking inflammatory tone to organelle‐mediated barrier failure.92 Additionally, targeting the PERK branch with inhibitors like GSK2606414 or modulating NR2F2 has been shown to improve retinal structure and function in preclinical models of ischemic and microvascular injury.34, 93

Human biomarker studies increasingly mirror the experimental results. Quantitative proteomics of aqueous humor across DR stages (non‐proliferative diabetic retinopathy → proliferative diabetic retinopathy) has revealed stage‐dependent shifts in inflammation, complement, oxidative stress, and metabolic networks. This indicates that redox/ER‐stress pathways are engaged in vivo and are scalable as fluid biomarkers.94, 95 Clinical observations of increased GRP78 and mitochondrial functional markers in patient fluids correlate with VEGF levels and macular edema, supporting the involvement of these organelle axes in human disease.17, 33, 88, 96, 97 Integrating these mechanism‐based approaches with standard anti‐VEGF therapy may facilitate earlier detection and more effective intervention for DR as a metabolically driven ER–mitochondria stress syndrome.94 Thus, DR reflects a metabolically driven ER–mitochondria stress syndrome, in which chronic hyperglycemia induces sustained MAM remodeling and impaired mitophagy, amplifying oxidative stress, inflammation, and vascular–neural dysfunction.

In AMD, the RPE endures chronic oxidative and proteotoxic stress, which activates the UPR and is associated with the progressive coupling of ER dysfunction to mitochondrial failure9 (Figure 2D). While the basic UPR‐MAM signaling framework is summarized in Table 1, the RPE exhibits unique pathomechanisms such as A2E‐induced phototoxicity and ZBP1‐mediated PANoptosis, an integrated inflammatory cell death pathway that coordinates key features of apoptosis, pyroptosis, and necroptosis.12, 98, 99 Concurrently, a decline in PINK1/Parkin and BNIP3/NIX‐dependent mitophagy is associated with impaired the removal of damaged mitochondria, potentially contributing to a feed‐forward loop of ROS and inflammatory signaling.100, 101

Regarding multi‐omic and model systems of the ER–mitochondria axis, iPSC‐derived RPE from AMD donors (and those harboring risk variants) show impaired oxidative phosphorylation and an increased susceptibility to mitochondrial toxins. This provides a platform in which mitochondria‐targeted drugs can restore bioenergetics, which functionally validates mitochondria as a target in AMD.102 Mechanistic studies using RGR‐deletion or PGC‐1‐related mouse models associate autolysosomal defects and mitophagy failure with RPE degeneration in vivo.103, 104 A complementary data map revealed how aging and altered redox status reshape mitochondrial dynamics (DRP1/MFN/OPA1), crosstalk with complement, and regulate cell death programs during geographic atrophy. This situates organelle stress early in the AMD cascade.105, 106

Human donor studies consistently document an organelle‐linked bioenergetic crisis in AMD. Mitochondrial DNA damage is significantly accentuated in RPE from carriers of the complement factor H Y402H risk variant, aligning genetic susceptibility with bioenergetic failure.107, 108 Beyond cell‐intrinsic defects, mitochondria‐associated autoantibodies are enriched in AMD and correlate with disease mechanisms, suggesting an immune amplification layer that intersects with organelle stress.109 Treatment strategies involve: (i) UPR tuning to avoid CHOP‐dominated maladaptation, while sustaining proteostasis,9 (ii) MAM stabilization/Ca²⁺ control to preserve ATP supply and limit ROS‐inflammation coupling,12 and (iii) mitochondrial quality‐control restoration (enhancing PINK1/Parkin or BNIP3/NIX mitophagy) to eliminate dysfunctional organelles.100 Translational momentum includes mitochondria‐targeted drug screening in iPSC‐RPE and early‐stage concepts, such as mitochondrial transplantation for retinal degenerations, which supports the feasibility of reinforcing the ER–mitochondria unit pharmacologically or biologically in human tissue models.102, 110 Integrating these organelle‐targeted strategies with high‐resolution imaging and fluid biomarkers will enable the earlier detection of RPE failure and guide the next generation of AMD combination therapies.105

FECD is hallmarked by the progressive attrition of mitochondria‐rich, minimally proliferative CECs111 (Figure 2E). Chronic oxidative and proteotoxic stress in these cells triggers a transition from adaptive proteostasis to metabolic rewiring and apoptosis.111, 112 While the canonical UPR branches and general MAM‐mediated Ca²⁺ flux dysregulation are summarized in Table 1, the FECD phenotype is uniquely defined by a rapid loss of mitochondrial membrane potential (ΔΨm) and ATP sufficiency.112, 113

A hierarchical signaling study indicated that p38 mitogen‐activated protein kinase (MAPK) is downstream of PERK–eIF2α, yet upstream of ATF4–CHOP. Inhibiting p38 preserves ΔΨm and suppresses apoptosis in multiple ER‐stress scenarios (TGF‐β–conditioned FECD cells, thapsigargin, MG‐132). This suggests a potential signaling switch involved in the transition of adaptive UPR into death signaling in the corneal endothelium.114 Besides signaling, mitochondrial QC is a central failure point. Reviews and experimental studies have documented mitophagy/autophagy dysregulation in FECD, suggesting that the impaired clearance of dysfunctional mitochondria compounds ROS and ER stress.115 In Slc4a11−/− models, mitochondrial ROS has been shown to lie upstream of ER stress, reinforcing the concept that mitochondrial stabilization is a primary requirement for maintaining the corneal endothelial pump.116

Translational strategies are coalescing around restoring organelle fidelity. Tuning the UPR through p38 MAPK effectors or PERK‐proximal outputs may help rebalance ER signaling away from CHOP‐dominated death.114 On the mitochondrial front, the antioxidant MitoQ and NAD⁺‐precursor support have demonstrated efficacy in protecting CECs from oxidative and ultraviolet B‐induced injury.117, 118 Most notably, the exogenous supplementation of mitochondria (mitochondrial transplantation) into FECD patient explants has been shown to reduce oxidative stress and restore ΔΨm, providing a proof‐of‐concept for organelle‐based biological rescue.119, 120 Targeting the ER–mitochondria interface to normalize MAM‐mediated flux remains a promising, mechanism‐based strategy to preserve the minimally proliferative CEC population.119, 121

As discussed in Sections 2 and 4, sustained activation of the PERK–eIF2α–ATF4–CHOP axis represents a pro‐apoptotic switch in LECs, RGCs, retinal Müller cells, RPE, and CECs during conditions of chronic stress.9, 112 Therefore, therapeutic strategies aimed at attenuating maladaptive UPR outputs, while preserving adaptive proteostasis, have gained interest. Chemical chaperones, such as 4‐phenylbutyrate (4‐PBA) and TUDCA, reduce ER stress by enhancing protein folding capacity, and they exert protective effects in models of cataract‐related EMT, corneal endothelial degeneration, and metabolic retinal stress.22, 112

Beyond global chaperoning, pathway‐selective modulation of UPR sensors offers greater precision. The inhibition of PERK signaling ameliorates RGC death in glaucomatous models31, 32 and reduces pathological angiogenesis and neuronal loss in ischemic and DR paradigms.93 Similarly, targeting ATF6‐dependent transcriptional loops, such as the ATF6–SNAI1 axis implicated in lens fibrosis, attenuates EMT and capsular opacification, which highlights the disease‐specific utility of branch‐selective UPR intervention.24 These results support a therapeutic paradigm in which UPR signaling is not globally suppressed, but dynamically rebalanced to prevent ER stress from causing mitochondrial failure.

MAMs constitute the physical and functional hub of ER–mitochondria crosstalk, as discussed in Section 3. They regulate Ca²⁺ transfer, lipid metabolism, and apoptotic signaling.3, 12 Disruption of MAM architecture through altered MFN2, PACS2, VAPB–PTPIP51, or InsP₃R–GRP75–VDAC1 complexes has been implicated in glaucoma,21 DR,87 AMD,99 and FECD.113 Restoring balanced ER‐to‐mitochondria Ca²⁺ flux represents an effective strategy to prevent mitochondrial Ca²⁺ overload, permeability transition pore opening, and ROS amplification. Although pharmacological stabilizers of MAMs are still in early development, indirect modulation of Ca²⁺ handling has shown promise. Sigma‐1 receptor agonists, which stabilize InsP₃R function at the MAMs and prolong adaptive mitochondrial Ca²⁺ uptake, exert neuroprotective effects in retinal neurons and are mechanistically consistent with preserving ER–mitochondria coupling.122 Similarly, targeting excessive mitochondrial Ca²⁺ influx by modulating mitochondrial calcium uniporter (MCU) activity or upstream ER Ca²⁺ release may prevent stress‐induced bioenergetic collapse in retinal and corneal cells.17, 123 These approaches are consistent with the evidence that MAM dysfunction represents an early and reversible event in ocular disease progression.

As discussed in Sections 3.1 and 4, pathological remodeling of mitochondrial dynamics, which is characterized by excessive DRP1‐mediated fission and impaired MFN/OPA1‐dependent fusion, contributes directly to neuronal and epithelial vulnerability in glaucoma,26, 27, 28 AMD,105, 106 and FECD.112 Pharmacological or genetic inhibition of DRP1 preserves mitochondrial network integrity, maintains ATP production, and promotes RGC survival and axonal function in experimental glaucoma.27 These results support mitochondrial dynamics as a therapeutic target downstream of ER–mitochondria stress signaling.

Complementary strategies focus on reinforcing mitochondrial bioenergetics and redox capacity. Mitochondria‐targeted antioxidants, such as MitoQ, can restore mitochondrial membrane potential, reduce cytochrome c release, and enhance cell survival in corneal endothelial and retinal stress models.116, 117 Similarly, NAD⁺ precursor supplementation promotes mitochondrial resilience against ultraviolet‐ and oxidative‐stress‐induced damage in corneal and retinal tissues, thus dampening ER stress propagation.118 Taken together, these strategies indicate that preserving mitochondrial function can interrupt the vicious cycle linking ER stress to cell degeneration.

Impaired mitochondrial QC is a recurring theme among ophthalmic diseases, particularly in DR,90, 91 AMD,103, 104 glaucoma,86 and FECD,119 in which defective PINK1/Parkin‐ or BNIP3/NIX‐mediated mitophagy results in the accumulation of dysfunctional mitochondria. Because of the intimate coupling between mitophagy, ER stress resolution, and MAM signaling, therapeutics aimed at restoring mitophagic flux have emerged as a key downstream component of ER–mitochondria–targeted intervention.

In metabolic and degenerative retinal diseases, such as DR and AMD, evidence indicates that impaired mitophagy amplifies mitochondrial ROS production, inflammasome activation, and inflammatory cell death, thereby reinforcing ER stress and MAM dysfunction. Pharmacological or genetic enhancement of mitophagy can reduce oxidative injury, suppress inflammatory signaling, and stabilize mitochondrial bioenergetics in retinal endothelial cells, Müller glia, and RPE, which support mitophagy as a disease‐modifying target in preclinical models.90, 91, 103, 104 However, translation to these diseases is constrained by challenges in cell‐type‐specific delivery and the need to preserve mitochondrial mass in tissues with high energetic demand. This highlights the importance of stage‐ and context‐dependent modulation, rather than indiscriminate activation of mitochondrial clearance.

In glaucoma, mitophagy dysfunction has been implicated in RGC vulnerability, in which impaired clearance of damaged mitochondria exacerbates bioenergetic failure and axonal degeneration.18, 19 Although direct therapeutic manipulation of mitophagy in glaucoma is at an early experimental stage, studies suggest that restoring mitochondrial QC may synergize with interventions targeting mitochondrial dynamics and ER stress to preserve RGC survival during the early stages of disease.86

Notably, the strongest translational evidence for targeting mitochondrial QC is in FECD. Studies using exogenous mitochondrial supplementation in patient‐derived corneal endothelium revealed that direct restoration of mitochondrial capacity reduces oxidative stress, normalizes mitochondrial membrane potential, and rebalances excessive mitophagy in human tissues ex vivo.117, 119 These results establish organelle‐level mitochondrial intervention as a feasible treatment strategy and position FECD as a model disease for translating ER–mitochondria–centered therapies toward clinical application.

Taken together, these observations indicate that while mitophagy restoration represents a shared therapeutic axis across multiple ophthalmic diseases, optimal intervention differs by tissue context and disease stage. FECD provides a near‐clinical paradigm for direct mitochondrial rescue, whereas DR, AMD, and glaucoma may benefit from more nuanced approaches that integrate mitophagy modulation with upstream ER stress control and mitochondrial dynamics stabilization.

Data sharing not applicable to this article as no datasets were generated or analyzed during the current study.