1. Introduction

In 2012, Brent R. Stockwell first described a novel form of programmed cell death distinct from apoptosis, necrosis, and autophagy, and coined the term “ferroptosis”^[1]. This groundbreaking discovery opened a new frontier in cell death research, and the field of ferroptosis has witnessed explosive growth over the past decade. Ferroptosis is biochemically defined by the lethal peroxidative accumulation of phospholipid-polyunsaturated fatty acids (PL-PUFAs)^[1,2]. Morphologically, ferroptotic cells exhibit characteristic mitochondrial abnormalities, including shrinkage, cristae reduction or complete loss, and outer membrane rupture, features considered hallmarks of the irreversible stage of ferroptosis^[3]. At the molecular level, ferroptosis involves dysregulation of cellular iron metabolism, reprogramming of lipid metabolism, and disruption of redox homeostasis. Specifically, excess intracellular free iron ions react with reactive oxygen species (ROS) via the Fenton reaction to generate highly reactive hydroxyl radicals (·OH). These potent oxidants attack membrane-resident PL-PUFAs, producing phospholipid radicals that initiate a self-propagating chain reaction of lipid peroxidation, leading to the continuous generation of toxic lipid peroxides. When the accumulation of lipid peroxides exceeds the clearance capacity of the cellular antioxidant defense system, plasma membrane integrity is compromised, ultimately resulting in cell swelling and membrane rupture^[4,5].

In recent years, with the rapid advancement of spatial biology technologies, accumulating evidence has reframed ferroptosis not as a simple cytoplasmic biochemical event, but as a spatially orchestrated cascade process executed cooperatively across multiple subcellular organelles, with strict tissue specificity, spatial compartmentalization, and microenvironment dependence. The initiation and progression of ferroptosis result from the dynamic trafficking and interplay of free iron, remodeled lipids, and ROS across different subcellular compartments. Organelles communicate extensively through physical membrane contact sites (MCSs), vesicular trafficking, and metabolic flux transfer. Lysosomes act as the gatekeeper of iron metabolism, regulating iron uptake and release to initiate and amplify oxidative stress. The endoplasmic reticulum (ER) serves as the central hub for lipid assembly, providing substrates for the synthesis of lethal PL-PUFAs and acting as the primary site of lipid peroxidation initiation. Mitochondria, as the core hub of cellular redox homeostasis, generate ROS via the tricarboxylic acid (TCA) cycle and electron transport chain (ETC) to further amplify oxidative stress. The plasma membrane is the final execution site of ferroptosis, where excessive accumulation of peroxidized lipids directly increases membrane permeability and triggers irreversible rupture. This inter-organelle communication and functional coordination collectively govern the initiation and execution of ferroptosis.

Given the critical role of ferroptosis in a wide spectrum of pathological processes, including cancer, ischemia-reperfusion injury (IRI), neurodegenerative diseases (NDDs) and cardiovascular diseases, its clinical translational potential has attracted enormous attention^[6]. For cancer cells, their hyperactive metabolism and high iron demand confer survival advantages but also expose a profound metabolic vulnerability to ferroptosis. In contrast, in NDDs and IRI, ferroptosis drives irreversible tissue damage and accelerates disease progression. Therefore, a comprehensive understanding of the spatiotemporal-specific roles of subcellular organelles and the inter-organelle communication networks in ferroptosis is essential for translating ferroptosis research into clinical applications, including overcoming tumor drug resistance and developing precision therapeutic interventions for human diseases.

2. The Biochemical Basis of Ferroptosis

Ferroptosis results from the convergence of intracellular iron metabolism dysregulation, lethal accumulation of PUFAs, and global collapse of antioxidant defense systems. Extracellular ferric iron (Fe³⁺) is internalized via the transferrin receptor (TFRC), reduced to ferrous iron (Fe²⁺), and enters the cytosolic labile iron pool (LIP)^[7,8]. Excess free Fe²⁺ catalyzes the decomposition of peroxides via the Fenton reaction to generate highly destructive ·OH^[9], which represents one of the major sources of ROS during ferroptosis. Early studies established that PLs are the key structural components of cell membranes^[10]. PUFAs, the oxidizable substrates whose membrane esterification is a prerequisite for ferroptosis, are first activated to PUFA-CoA by long-chain-fatty-acid-CoA ligase 4 (ACSL4) and subsequently incorporated into membrane lysophospholipids to form PUFA-PLs by lysophosphatidylcholine acyltransferase 3 (LPCAT3)^[11,12]. The accumulated ROS sustains the propagation of the lipid peroxidation chain reaction, with PUFA-PLs serving as the preferential substrates for both enzymatic (lipoxygenase (LOX)-mediated) and non-enzymatic (Fenton-driven) lipid oxidation.

To counteract ferroptosis, cells have evolved a suite of antioxidant defense systems distributed across multiple subcellular compartments, including mitochondria, ER, lysosomes, cytosol, and the plasma membrane, collectively forming a multi-layered protective network. The four most well-characterized ferroptosis defense mechanisms are the glutathione (GSH)-glutathione peroxidase 4 (GPX4) axis^[13], ferroptosis suppressor protein 1 (FSP1)-coenzyme Q10 (CoQ10) axis^[14], mitochondrial dihydroorotate dehydrogenase (DHODH) regulatory axis^[15], and guanosine triphosphate cyclohydrolase 1 (GCH1)-tetrahydrobiopterin (BH4)-dihydrofolate reductase (DHFR) axis^[16].

The GSH-GPX4 system was the first identified ferroptosis defense pathway. GSH acts as an essential cofactor for GPX4, which catalyzes the conversion of toxic cytosolic lipid hydroperoxides (L-OOH) to non-toxic lipid alcohols (L-OH), thereby inhibiting lipid peroxidation^[13,17]. It is important to note that the availability of cysteine, the rate-limiting precursor for GSH biosynthesis, is not solely determined by system Xc^--mediated cystine import. Cells also synthesize cysteine endogenously via the transsulfuration pathway. In the transsulfuration pathway, cystathionine β-synthase (CBS) catalyzes the condensation of homocysteine and serine to form cystathionine, which is subsequently converted to cysteine by cystathionine γ-lyase. This CBS-dependent alternative cysteine supply route becomes particularly critical in the tumor microenvironment. Indeed, our group discovered that, under cystine deprivation stress, colorectal cancer (CRC) cells engage the ATF3-CBS signaling axis as a primary adaptive feedback pathway to maintain GSH levels and evade ferroptosis^[18]. These findings fundamentally expand the mechanistic landscape of ferroptosis resistance beyond canonical system Xc^--GPX4 signaling. The FSP1-CoQ10 system functions as a parallel, GPX4-independent defense mechanism: FSP1 is targeted to the plasma membrane via N-myristoylation, where it reduces oxidized CoQ10 to ubiquinol (CoQ10H2). As a potent lipophilic antioxidant, CoQ10H2 directly scavenges lipid peroxyl radicals at the plasma membrane^[19,20]. In 2021, Mao et al. first identified a novel subcellular compartment-specific ferroptosis defense system, demonstrating that inner mitochondrial membrane (IMM)-localized DHODH regulates ferroptosis by suppressing mitochondrial lipid peroxidation^[15]. Furthermore, the GCH1-BH4-DHFR axis mediates the synthesis of the lipid-soluble antioxidant BH4, which not only directly blocks the lipid peroxidation chain reaction^[21], but also enhances the antioxidant capacity of the FSP1-CoQ10 pathway by regulating CoQ10 biosynthesis^[16]. Additionally, cells have evolved a range of non-canonical ferroptosis defense and regulatory mechanisms that are highly dependent on specific subcellular compartments and organelles; these are discussed in detail in Section 2, “Organelle-Regulated Ferroptosis Processes” of this review (Figure 1).

Display Full Size

Download

Figure 1. The subcellular landscape of antioxidant defense systems in ferroptosis. Fe³⁺ is internalized into lysosomes and reduced to Fe²⁺, triggering early LPO in lysosomes. Released Fe²⁺ enters the cytosolic LIP, propagates across multiple subcellular compartments and initiates the lipid peroxidation chain reaction, ultimately leading to plasma membrane rupture and ferroptosis. Each organelle has evolved distinct antioxidant defense systems to counteract ferroptosis: mitochondria defend against lipid peroxidation via the mGPX4-GSH axis and DHODH-CoQ10H2 axis; the ER exerts protective effects through lipid remodeling, the GPX1-OSBPL8 axis, and iPLA2β-mediated repair of oxidized membrane lipids; LDs buffer oxidative damage via the FSP1-CoQ10H2 antioxidant system; and the plasma membrane inhibits ferroptosis progression through lipid remodeling and the FSP1-CoQ10H2 pathway. Created in BioRender. Zheng, H. (2026) https://BioRender.com/5heqmfa. Fe³⁺: ferric iron; Fe²⁺: ferrous iron; LPO: local lipid peroxidation; LIP: labile iron pool; mGPX4: mitochondrial glutathione peroxidase 4; GSH: glutathione; DHODH: dihydroorotate dehydrogenase; CoQ10H2: coenzyme Q10-hydroquinone; ER: endoplasmic reticulum; GPX1: glutathione peroxidase 1; OSBPL8: oxysterol binding protein-like 8; iPLA2β: calcium-independent phospholipase A2β; LDs: lipid droplets; FSP1: ferroptosis suppressor protein 1.

Emerging evidence further reveals that the intracellular metabolic network controlling cysteine bioavailability, and hence GSH biosynthesis, extends well beyond the canonical system Xc^--mediated cystine import pathway. A recent study from our group demonstrated that hydrogen sulfide (H₂S), a gaseous signaling molecule derived from the transsulfuration pathway, sensitizes non-small cell lung cancer (NSCLC) cells to ferroptosis by reprogramming homocysteine metabolism. Mechanistically, H₂S covalently persulfidates the 195th cysteine residue (Cys195) of S-adenosylhomocysteine hydrolase (SAHH), the enzyme responsible for hydrolyzing S-adenosylhomocysteine (SAH) to homocysteine, thereby impeding homocysteine generation and subsequently reducing the transsulfuration-derived cysteine pool. Under conditions of concurrent cystine deprivation, this dual depletion of cysteine supply pathways synergistically collapses intracellular GSH levels and dramatically amplifies ferroptosis susceptibility in NSCLC. Combining exogenous H₂S donors with pharmacological cystine depletion (e.g., system Xc^- inhibitors) therefore represents a mechanistically rational and therapeutically actionable strategy to potentiate ferroptosis-based cancer therapy^[22].

In summary, cellular susceptibility to ferroptosis is not determined by a single factor but rather by the dynamic balance between multiple pro-ferroptotic drivers and layered anti-ferroptotic defense systems. This delicate equilibrium ultimately dictates cell fate decisions under stress conditions.

3. Organelle-Regulated Ferroptosis Processes

In recent years, with the successive discovery of multiple non-canonical antioxidant mechanisms, the spatial regulatory network of ferroptosis is being redefined. Specifically, at the subcellular initiation level, studies from the Curie Institute in France have demonstrated that lysosomes are a critical initiation site for ferroptosis. The release of labile iron from lysosomes triggers the initial free radical chain reaction, which further propagates to adjacent organelles, a process that confers ferroptosis with distinct spatial propagation characteristics^[23]. Building on this, the ER has also been established as a major platform for lipid peroxidation. Gu’s team revealed a non-canonical ferroptosis pathway initiated by ROS-induced peroxidation of ER phosphatidic acid and regulated by the glutathione peroxidase 1 (GPX1)-oxysterol binding protein-like 8 (OSBPL8) axis, indicating that ER membrane lipids can also serve as important initiation sites for ferroptosis^[24].

Dynamic remodeling of membrane lipid composition has further been proven to be a key determinant of cellular ferroptosis susceptibility. Jiang’s group found that membrane-bound glycerophospholipid O-acyltransferase 2 (MBOAT2) directly restricts lipid peroxidation at the substrate level by converting oxidation-prone PL-PUFAs into relatively stable PL-monounsaturated fatty acids (PL-MUFAs)^[25].

Meanwhile, ferroptosis does not solely depend on local oxidative events; its magnitude is also regulated by the cellular metabolic network. Bian’s and Bi’s teams integrated polyamine metabolism into the ferroptosis regulatory framework, discovering that polyamine catabolism enhances ROS production and amplifies lipid peroxidation reactions, thereby acting as a metabolic “amplifier” to potentiate ferroptosis progression^[26,27]. Metabolites themselves can also directly participate in ferroptosis regulation. For example, Li, Florencio Porto Freitas, and others found that 7-dehydrocholesterol (7-DHC), an intermediate in the cholesterol biosynthesis pathway, is an endogenous antioxidant molecule that effectively inhibits ferroptosis; conversely, its metabolizing enzyme, 7-dehydrocholesterol reductase, significantly increases cellular sensitivity to ferroptosis by reducing intracellular 7-DHC accumulation^[4,28]. Furthermore, the spatial regulation of ferroptosis has extended from membrane systems to lipid droplets (LDs). The team led by James A. Olzmann reported that FSP1 protects neutral lipids from peroxidation by regulating LD quality, a finding that demonstrates LDs are not only cellular lipid storage depots but also critical defensive compartments that buffer lipid peroxidation^[29].

Collectively, these studies have systematically revealed the multidimensional network of ferroptosis regulation at different spatial levels, shifting the field from the study of individual antioxidant pathways to a multi-level spatial integration model.

Current research in the ferroptosis field has generally proposed two mechanistic models: the sequential cascade amplification model and the parallel, and independent occurrence model. In this review, we adopt the sequential cascade model as a hypothetical conceptual framework to systematically delineate the subcellular regulatory progression of ferroptosis.

3.1 Lysosomes: The gatekeeper of iron metabolism

For decades, lysosomes have been regarded as the “waste disposal stations” of cells. Their repertoire of hydrolytic enzymes engulfs and degrades senescent or damaged organelles and digests invading pathogens, thereby maintaining intracellular homeostasis and normal metabolic processes. However, within the molecular regulatory network of ferroptosis, lysosomes have emerged as the central regulator of iron homeostasis and a critical signaling hub governing cell fate. Ferroptosis is an iron-dependent form of cell death whose initiation and execution are highly dependent on the abundance and mobilization rate of lysosomal labile iron. By precisely regulating iron uptake, conversion, and release, lysosomes play a pivotal role as the iron metabolism gatekeeper in the lipid peroxidation cascade of ferroptosis.

Cellular acquisition of free iron relies heavily on the vesicle trafficking-mediated transferrin (TF)-TFRC endocytic pathway. Under physiological conditions, iron primarily exists as Fe³⁺ bound to TF to form a complex. This complex is specifically recognized by TFRC on the cell membrane^[30-32] and internalized into early endosomes via clathrin-mediated endocytosis^[33]. As early endosomes mature progressively into late endosomes and eventually fuse with lysosomes, vacuolar-type H⁺-ATPase (V-ATPase) continuously pumps protons into the lumen, causing dramatic acidification of the lysosomal interior and triggering the dissociation of Fe³⁺ from TF^[34]. The dissociated Fe³⁺ is reduced to highly bioactive Fe²⁺ by the lysosome-resident six-transmembrane epithelial antigen of prostate 3, and finally released into the cytosolic LIP via the membrane-localized divalent metal transporter 1 (DMT1)^[30,31].

Under normal physiological conditions, cells safely sequester excess free iron in ferritin “storage depots” assembled from ferritin heavy chain 1 (FTH1) and ferritin light chain^[35]. However, under stress conditions, nuclear receptor coactivator 4 on the surface of autophagosomes specifically recognizes and binds to the FTH1 subunit, triggering ferritinophagy. Ferritin is then transported to lysosomes and degraded by acidic hydrolases, releasing large amounts of free Fe²⁺ into the cytosolic LIP and significantly promoting the Fenton reaction and the accumulation of oxidative damage^[36-39]. Collectively, lysosomes serve as the critical “metabolic gatekeeper” in iron import, storage, and export, maintaining a dynamic balance between the physiological demand for iron and the lethal risk of iron overload.

The release of Fe²⁺ in the acidic lysosomal environment is a prerequisite for ferroptosis. Studies have shown that free Fe²⁺ efficiently catalyzes the Fenton reaction, generating abundant ROS on the lysosomal membrane. This causes lethal lipid peroxidation to initiate and become highly enriched at lysosomes^[23,40]. This process induces increased lysosomal membrane permeability, leading to the massive leakage of intraluminal free iron and cathepsins (e.g., cathepsin B) into the cytosol. The leaked lysosomal iron and proteases further trigger a cell-wide lipid peroxidation storm, ultimately resulting in plasma membrane rupture and cell death^[34,40]. In summary, free radical generation at lysosomes acts as the “igniter” of early lipid peroxidation, which amplifies oxidative damage signals in a cascade manner via lysosomal membrane permeabilization, culminating in extensive ferroptosis.

3.2 ER: The hub for lipid remodeling and stress responses

Ferroptosis is highly dependent on lipid metabolism, and the ER, as the central site for cellular lipid synthesis and metabolism, plays an indispensable role in the ferroptotic process. By dynamically regulating the ratio of less oxidizable MUFAs to highly oxidizable PUFAs in membrane systems, the ER directly determines the threshold of cellular ferroptosis susceptibility^[41-44]. ER-resident ACSL4 preferentially activates free long-chain PUFAs (e.g., arachidonic acid (AA)) to form PUFA-CoA^[11,45-47], which are subsequently esterified and incorporated into membrane phospholipids by LPCAT3 or membrane-bound glycerophospholipid O-acyltransferase 5 to generate PUFA-PLs^[12]. The ER possesses the most extensive membrane surface area within the cell and is highly enriched in highly oxidizable PUFA-PLs, providing abundant substrates for ferroptosis initiation^[2,48].

Conversely, MUFAs are activated to MUFA-CoA by long-chain-fatty-acid-CoA ligase 3 (ACSL3) and then selectively incorporated into phospholipids by ER-resident membrane-bound glycerophospholipid O-acyltransferase 1/2 (MBOAT1/2). Since MUFAs are less susceptible to oxidation, their incorporation competitively reduces the proportion of PUFA-PLs in cell membranes, thereby significantly inhibiting ferroptosis in a GPX4-independent manner^[25,49]. Furthermore, ER-resident cytochrome P450 oxidoreductase and cytochrome b5 reductase 1 utilize electron transfer from nicotinamide adenine dinucleotide phosphate (NADPH) to generate ROS locally at the ER, directly catalyzing lipid peroxidation of ER membranes^[50,51]. Although the mechanism by which local ER membrane lipid peroxidation signals propagate to the plasma membrane and ultimately execute ferroptosis remains incompletely understood, it is undeniable that the ER has emerged as the “fuel reservoir” for ferroptotic lipid peroxidation by virtue of its abundant PUFA-PL substrates, locally resident oxidoreductases, and extreme sensitivity to antioxidant depletion.

To defend against ferroptosis, the ER has also evolved dedicated defense systems. Traditionally, GPX4 was considered the only glutathione peroxidase capable of reducing complex lipid peroxides and inhibiting ferroptosis. However, the team led by Wei Gu innovatively identified an ER-localization-dependent non-canonical ferroptosis defense pathway, the GPX1-OSBPL8 axis. During ROS-induced ferroptosis, OSBPL8 specifically recruits the antioxidant enzyme GPX1 to the ER membrane, scavenging highly enriched peroxidized phospholipids in the ER compartment and thereby suppressing the cascade activation of non-canonical ferroptosis^[24]. The ER also harbors an oxidized lipid repair network mediated by calcium-independent phospholipase A2β (iPLA2β). This enzyme precisely recognizes and specifically cleaves the oxidatively damaged tails of PUFA-PLs, effectively preventing the propagation of the lipid peroxidation chain reaction and potently suppressing p53-driven non-canonical ferroptosis^[52-54]. Additionally, recent studies have revealed that inactivation of histone lysine demethylase 6A (KDM6A) under hypoxic microenvironments significantly downregulates the transcription of key ER lipid synthesis enzymes ACSL4 and ethanolamine kinase 1 (ETNK1), leading to a substantial reduction in PUFA-phosphatidylethanolamine production. This globally remodels the ER phospholipid synthesis network and confers robust non-canonical ferroptosis resistance on cells^[55]. Other studies have shown that the E3 ubiquitin ligase MARCHF6 functions as an NADPH sensor on the ER membrane, participating in the regulation of lipid synthesis, metabolic sensing, and ferroptosis susceptibility by sensing intracellular NADPH metabolic status^[56].

3.3 Mitochondria: The ROS engine and independent defense island

The biochemical basis of ferroptosis is the dysregulation of iron homeostasis. As the primary iron storage organelle in cells, mitochondria harbor 20% to 50% of total cellular iron^[57] and play a critical role in iron metabolism regulation. Fe²⁺ from the LIP sequentially traverses the outer mitochondrial membrane (OMM) and IMM via DMT1 and mitoferrin-1/2, and is pumped into the mitochondrial matrix^[2,58,59]. One of the major uses of iron in mitochondria is the de novo biosynthesis of iron-sulfur (Fe-S) clusters, whose levels serve as a key indicator of cellular iron sensing^[60]. In this process, cysteine desulfurase (NFS1) abstracts sulfur from cysteine, frataxin assists in Fe²⁺ loading, and ferredoxin 2 provides electrons, ultimately completing the de novo assembly of Fe-S clusters on the scaffold protein Fe-S cluster assembly protein U^[57,60,61]. Assembled Fe-S clusters are widely integrated into mitochondrial respiratory chain complexes, key TCA cycle enzymes, and proteins involved in DNA synthesis and repair, acting as indispensable cofactors for electron transport and enzymatic catalysis^[60,61]. A portion of Fe-S cluster intermediates is transported across the mitochondrial membrane to the cytosol via the mitochondrial membrane transporters ATP binding cassette subfamily B 7/8, supporting the normal function of the cellular Fe-S protein network^[57,59]. Excess free iron in the mitochondrial matrix is specifically sequestered in mitochondrial ferritin, achieving effective isolation that significantly reduces mitochondrial susceptibility to the Fenton reaction, potently protects mitochondria from oxidative damage, and thereby confers ferroptosis resistance on cells^[6,57,59,62]. Conversely, when NFS1 expression or activity is inhibited, cells receive erroneous iron deficiency signals, leading to mitochondrial free iron overload via upregulation of TFRC or downregulation of ferritin, ultimately markedly sensitizing cells to ferroptosis^[2,61,63].

ROS are the core driver and metabolic engine of ferroptosis, and are predominantly produced by mitochondria^[57]. During highly active oxidative phosphorylation, the mitochondrial ETC inevitably undergoes electron leakage. Leaked electrons combine with oxygen to form superoxide anion radicals (·O₂^-), which are catalyzed by superoxide dismutase (SOD) to generate hydrogen peroxide (H₂O₂). H₂O₂ then reacts with Fe²⁺ via the Fenton reaction to produce ·OH^[2,64]. The TCA cycle continuously provides energy substrates and electron carriers for ETC activity^[65]. Notably, even under stress conditions such as cystine deprivation, the coordinated operation of the TCA cycle and ETC remains the primary source of cellular lipid peroxides^[57,65]. Intriguingly, the indirect TCA cycle donor glutamate and the intermediate metabolite α-ketoglutarate also drive ferroptosis progression^[59,66-68]. A growing body of evidence indicates that the mitochondrial epitranscriptome constitutes an additional, previously underappreciated, layer of ferroptosis defense regulation. Our group recently identified that the mitochondrial methyltransferase METTL17 governs mitochondrial gene expression and respiratory chain function in CRC cells through epigenetic regulation of mitochondrial RNA methylation, encompassing m⁴C, m⁵C, m³C, m⁷G, and m⁶A modifications. Bioinformatic analysis of cancer cell line datasets establishes that METTL17 expression is upregulated in CRC and positively correlates with resistance to ferroptosis inducers. Mechanistically, depletion of METTL17 impairs mitochondrial translation fidelity, disrupts ETC complex assembly, reduces ATP production, and enhances intracellular and mitochondrial lipid peroxidation and ROS accumulation under ferroptotic stress. Combined targeting of METTL17 and ferroptosis-inducing agents produced synergistic anti-tumor efficacy in vivo, highlighting METTL17-mediated mitochondrial RNA methylation as a novel, therapeutically targetable axis of ferroptosis resistance in cancer^[69]. In general, the spatiotemporal colocalization of high concentrations of free iron and ROS within mitochondria makes mitochondria the “powder keg” that accelerates lipid peroxidation and drives ferroptosis progression.

Meanwhile, mitochondria possess a cytosol-independent antioxidant defense system comprising two key pathways: GPX4 and DHODH. The canonical antioxidant defense system GPX4 has a mitochondrial-specific isoform (mGPX4), which is the core molecule for mitochondrial ferroptosis defense^[59]. Using GSH as a cofactor, mGPX4 reduces destructive polyunsaturated fatty acid phospholipid hydroperoxides on mitochondrial membranes to non-toxic lipid alcohols, thereby blocking the lipid peroxidation chain reaction^[2,30,70]. In addition, mitochondria establish a second ferroptosis defense line via IMM-localized DHODH^[15]. During the catalysis of dihydroorotate (DHO) oxidation to orotate, DHODH transfers electrons to CoQ10 on the IMM, reducing it to CoQ10H2^[15,57,71]. As a potent RTA, CoQ10H2 scavenges lipid peroxyl radicals locally in mitochondria, protecting cells from ferroptosis^[15,72,73]. The sustained operation of the DHODH-CoQ10 antioxidant system depends on mitochondrial respiratory chain complex III. After DHODH-mediated reduction of CoQ10 to CoQ10H2, complex III reoxidizes CoQ10H2 back to CoQ10. This cyclical process maintains the robust antioxidant capacity of the IMM^[15,74,75]. Furthermore, the IMM-resident deubiquitinase ubiquitin-specific peptidase 24 directly interacts with and deubiquitinates DHODH, protecting it from proteasomal degradation and thereby ensuring the sustained antioxidant activity of the DHODH-CoQ10 axis^[76].

Mitochondria exert bidirectional regulatory effects on ferroptosis through selective mitophagy. Basal mitophagy functions as an intrinsic cellular defense: it removes damaged mitochondria in a timely manner to limit excessive mitochondrial ROS (mtROS) production and free iron leakage, thereby suppressing ferroptosis initiation. In the tumor microenvironment, tumor-associated macrophages secrete extracellular vesicles highly enriched in peroxiredoxin 6 (PRDX6). Upon internalization by tumor cells, PRDX6 exerts glutathione peroxidase activity to scavenge mitochondrial superoxide, mitigate mitochondrial damage, and downregulate BNIP3-mediated mitophagy, collectively reducing lipid peroxidation and rendering tumor cells resistant to ferroptosis^[77,78]. Conversely, dysregulation of PTEN-induced putative kinase 1 or FUN14 domain-containing 1-dependent mitophagy leads to the collapse of the mitochondrial regulatory network, which amplifies oxidative stress and disrupts iron homeostasis, ultimately sensitizing cells to ferroptosis^[79,80].

Intriguingly, the dual roles of mitochondria in ferroptosis, exerting antioxidant defense and driving pro-oxidant cell death, are not statically opposed states. Instead, they represent a continuous functional spectrum governed by the dynamic balance between the magnitude of incoming oxidative damage and the capacity of the mitochondrial antioxidant defense system. The activities of the mGPX4-GSH and DHODH-CoQ10H2 axes, mitochondrial iron burden, and ROS levels are the core determinants of this functional transition. When stress intensity does not breach the defense threshold, mitochondria predominantly act as “defense islands” that counteract ferroptosis. Once accumulated damage surpasses the critical tipping point and the defense system is inactivated, mitochondria switch to become “ROS engines” that propagate ferroptosis progression.

3.4 Plasma membrane: The ultimate executor of ferroptosis

The plasma membrane serves as the first barrier for material exchange between the intracellular and extracellular compartments, and its dynamic lipid composition directly determines the baseline susceptibility of cells to ferroptosis. Similar to lipid remodeling in the ER, the plasma membrane mediates the formation of PUFA-PLs via the ACSL4-LPCAT3 axis on one hand^[11,45,81], and dynamically constructs a ferroptosis resistance network centered on MUFAs on the other. Recent studies have found that membrane-bound glycerophospholipid O-acyltransferase 7 (MBOAT7) acts synergistically with ACSL4 to incorporate activated AA into phosphatidylinositol (PI), further increasing the abundance of PUFA-PLs on the plasma membrane^[46]. In addition, LOXs directly catalyze the peroxidation of plasma membrane PUFA-PLs to produce toxic L-OOH, thereby directly driving ferroptosis^{[50,63,82,83]}. Through this series of enzymatic reactions, cells actively convert exogenously acquired PUFAs into lethal ferroptotic signals on the plasma membrane.

Upon extensive peroxidation of plasma membrane PUFA-PLs, they decompose to generate toxic lipid electrophiles such as 4-hydroxynonenal. These electrophiles cause plasma membrane thinning and increased curvature via cross-linking, disrupting the stability and fluidity of the lipid bilayer^[84,85], and ultimately leading to the formation of nanoscale pores in the membrane^[86,87]. Concurrently, elevated membrane tension aberrantly activates mechanosensitive ion channels on the plasma membrane, including Piezo1 and transient receptor potential channels, triggering a lethal massive influx of sodium ions and calcium ions (Ca²⁺) and efflux of potassium ions^[86,88]. This severe disruption of transmembrane ion homeostasis leads to irreversible cell swelling, and ultimately results in complete plasma membrane rupture mediated by membrane proteins such as nerve injury-induced protein 1, marking the physical execution of ferroptosis^[89,90].

However, in the final stages before plasma membrane rupture, cells do not passively await death but rapidly activate membrane repair systems. Ca²⁺ influx triggered during late-stage ferroptosis acts as a danger signal, recruiting the endosomal sorting complex required for transport III (ESCRT-III) (comprising CHMP5 and CHMP6 subunits) to damaged plasma membrane regions. The ESCRT-III complex potently counteracts plasma membrane permeabilization by physically cleaving, shedding, and patching damaged membrane fragments, thereby significantly delaying or even partially reversing ferroptosis execution^[91,92]. Concurrently, the plasma membrane-resident FSP1-CoQ10 axis constitutes the second major ferroptosis defense system. FSP1 is specifically recruited and anchored to the plasma membrane via N-myristoylation, and uses NAD(P)H to reduce membrane-associated CoQ10 to CoQ10H2. As a potent lipophilic antioxidant, CoQ10H2 directly neutralizes and scavenges lipid peroxyl radicals on the membrane, blocking the lipid peroxidation chain reaction and potently inhibiting ferroptosis^{[14,19,20,93,94]}.

In summary, the plasma membrane is not only a “receiver” of ferroptosis-inducing signals but also a highly dynamic platform for regulating lipid metabolism and redox homeostasis. The dynamic balance between “pro-ferroptotic PUFA lipid remodeling” driven by ACSL4/LPCAT3 and “anti-ferroptotic MUFA lipid remodeling” driven by ACSL3/MBOAT on the plasma membrane ultimately determines the redox homeostasis of the plasma membrane lipid bilayer and the fate of the cell.

3.5 Hidden players: Nucleus, LDs, Golgi apparatus and peroxisomes

3.5.1 Nucleus: The global regulatory command center and stress signal integration hub

Although the nucleus is not traditionally recognized as a direct executor of lipid peroxidation, it functions as a global integration hub for inter-organelle stress signals during ferroptosis. The nucleus receives and integrates stress inputs from the ER, lysosomes, and mitochondria, thereby dynamically reprogramming the cellular antioxidant defense network.

Specifically, the protein kinase R-like ER kinase (PERK)-eukaryotic translation initiation factor 2α (eIF2α)-activating transcription factor 4 (ATF4)-nuclear factor erythroid 2-related factor 2 (NRF2) axis activated by ER and mitochondrial stress^[95], along with the F-box and leucine-rich repeats protein 5 (FBXL5)-iron regulatory protein 2 (IRP2) axis driven by cytosolic and lysosomal iron metabolism^[96,97], transduce organelle-level stress signals to the nucleus and trigger global defensive responses. In addition, transcription factors including p53^[98], HIF-1/2α^[99], and YAP/TAZ^[100,101] regulate the expression of key ferroptosis regulators via direct or indirect mechanisms, participating in the modulation of ferroptosis sensitivity in pathological contexts such as tumors and NDDs. Beyond canonical transcriptional responses, the nucleus exerts remote control over the ER lipid synthesis network through the KDM6A-ACSL4-ETNK1 axis, profoundly shaping organelle-specific lipid remodeling^[55].

Currently, understanding of the nuclear regulatory role in ferroptosis remains centered on transcriptional homeostatic reprogramming. Direct non-transcription-dependent regulatory mechanisms still lack dynamic evidence from high-spatiotemporal-resolution approaches and definitive functional validation, representing a critical gap to be addressed in future research on subcellular ferroptosis regulation.

3.5.2 LDs: The buffer and trigger zone of ferroptosis

LDs are highly dynamic intracellular organelles consisting of a hydrophobic core of neutral lipids, including triacylglycerols (TAGs) and cholesteryl esters, surrounded by a phospholipid monolayer^[102,103]. LDs play a classic “double-edged sword” role in the regulation of ferroptosis. On one hand, via the diacylglycerol acyltransferase (DGAT)-dependent biosynthetic pathway, intracellular free PUFAs are esterified into the fatty acid side chains of neutral lipids and sequestered within LDs^[104-106]. This physical sequestration of PUFAs away from ROS significantly reduces the probability of lipid peroxidation, exerting an oxidative buffering effect. Furthermore, FSP1 can also specifically localize to the LD surface, where it is confirmed to regenerate antioxidant molecules such as CoQ10H2 in situ, protecting the stored lipids within LDs from oxidative damage^[14,20,29].

On the other hand, under stress conditions such as nutrient deprivation, fatty acids sequestered in LDs in the form of TAGs are mobilized via lipophagy and lipolysis for energy metabolism and biosynthesis^[107,108]. Concurrently, AMP-activated protein kinase (AMPK) is activated, triggering chaperone-mediated autophagy to specifically degrade the protective coat proteins PLIN2 and PLIN3 on the LD surface^[109-112]. Subsequently, a lipase cascade localized to the LD surface initiates lipolysis, driving the breakdown of PUFA-rich neutral lipids within LDs into free PUFAs that are released into the cytosol^[107,113]. Meanwhile, the reduction in LD volume caused by lipolysis further enhances the degradation efficiency of lipophagy^[107,114].

Lipophagy occurs in two main forms: macrolipophagy and microlipophagy. In macrolipophagy, autophagosomes engulf intact LDs and transport them to lysosomes to form autolysosomes, facilitating LD degradation^{[107,114,115]}. In microlipophagy, lysosomes directly engulf LD fragments via membrane invagination, or form transient “kiss-and-run” contacts with LDs to transfer lipids directly into the lysosomal lumen for degradation^[116-118]. Massive degradation of LDs leads to a sharp increase in cytosolic free PUFA levels, rendering cells hypersensitive to ferroptosis. Notably, in specific tumor types such as clear cell renal cell carcinoma, excessive LD accumulation paradoxically forms the metabolic basis for heightened ferroptosis sensitivity, promoting the cascade propagation of ferroptotic signals^[82,99].

In summary, LDs exhibit a distinct bidirectional role in ferroptosis regulation: they exert anti-ferroptotic protective effects via “physical sequestration” and “antioxidation in situ”, while promoting ferroptosis through “uncontrolled release under stress” and “lethal excessive accumulation under pathological conditions”.

3.5.3 The Golgi apparatus: Stress sensing and trafficking hub

The Golgi apparatus, the central hub for protein/lipid processing, sorting, and vesicular trafficking, engages in extensive crosstalk with diverse subcellular organelles. Emerging evidence identifies it as a critical regulator of ferroptosis, acting through three distinct mechanisms.

First, Golgi stress directly drives ferroptosis. Pharmacological disruption of Golgi structure by Brefeldin A, Golgicide A, or AMF-26 (also known as M-COPA), triggers a massive accumulation of lipid peroxides and initiates the ferroptosis cascade across various cell lines. Strikingly, this lethality can be fully abrogated by antioxidant overexpression (e.g., SLC7A11 or GPX4) or ROS scavengers^[119], underscoring the tight functional coupling between Golgi integrity and ferroptosis surveillance. A recent paradigm-shifting study by Wang’s team, published in Nature Cancer, revealed that early ferroptosis is characterized by rapid Golgi membrane lipid peroxidation and a concomitant, dramatic alkalinization of the Golgi lumen. In this context, the Golgi-resident transmembrane protein 87A emerges as a critical homeostatic rheostat, buffering the luminal pH fluctuations to dynamically modulate the ferroptosis progression^[120].

Second, Golgi-derived vesicular trafficking may spatially coordinate the global distribution of RTAs. CoQ10 is a redox-active lipid whose reduction to CoQ10H2 serves as a potent lipid detoxification pathway. The mobilization of CoQ10 from its biosynthetic organelles to the plasma membrane inevitably requires vesicular transport. Given the specific targeting of Golgi-derived vesicles to ER-endosome MCSs^[121] and the dependence of CoQ10 trafficking on mitochondria-ER contact sites (MERCs)^[122], it is tantalizing to postulate that Golgi-derived vesicles are recruited to MERCs to facilitate the acquisition and subsequent transmembrane shuttling of CoQ10.

Furthermore, Golgi-resident glycosyltransferases fine-tune ferroptosis vulnerability via post-translational modifications. For instance, β-1,3-N-acetylglucosaminyltransferase 3 stabilizes system Xc^- through N-glycosylation of its 4F2hc subunit, enhancing cystine uptake and ferroptosis resistance^[123]. Similarly, the presence of α-1,3-mannosyltransferase acts as a critical molecular brake on sterol regulatory element-binding protein 1-dependent lipogenesis, effectively preventing the lethal hyper-accumulation of pro-ferroptotic lipids^[124].

3.5.4 Peroxisomes: An autonomous lipid metabolism engine

Historically, ferroptosis research has centered on mitochondria and the ER as the primary regulators of oxidative stress and lipid remodeling. However, a rapidly evolving paradigm now identifies peroxisomes as critical determinants of cellular ferroptosis susceptibility, functioning through two interconnected mechanisms: lipid biosynthetic coupling with the ER and redox signaling crosstalk with mitochondria.

Within the lipid metabolic network, peroxisomes harbor a dedicated pro-ferroptotic pathway that acts synergistically with the ER. Using genome-wide CRISPR screening, Zou et al. made the groundbreaking discovery that peroxisomal genes, including PEX10, PEX3, alkylglycerone phosphate synthase (AGPS), and fatty acyl-CoA reductase 1 (FAR1), are essential regulators of ferroptosis^[125]. Distinct from the canonical ACSL4-LPCAT3 pathway that generates PUFA-containing phospholipids, peroxisomes drive the de novo biosynthesis of the plasmalogen precursor 1-O-alkyl-glycerol-3-phosphate (AGP) via the sequential actions of AGPS, FAR1, and glyceronephosphate O-acyltransferase. This precursor is then transported to the ER, where the ER-resident 1-acylglycerol-3-phosphate O-acyltransferase 3 and plasmanylethanolamine desaturase 1 (also known as TMEM189) convert it into PUFA-rich ether phospholipids (PUFA-ePLs), which potently sensitize cells to ferroptosis^[125]. More recently, Cui et al. delineated a novel ferroptosis pathway governed by the FAR1-ether lipid-TMEM189 axis, reinforcing the central importance of peroxisomal ether lipid biosynthesis in ferroptosis regulation^[126].

In summary, each organelle performs distinct and indispensable roles in the regulation of ferroptosis: lysosomes initiate the process, the ER dominates substrate synthesis, mitochondria mediate signal amplification, the nucleus acts as a global inter-organelle stress signal integration hub, LDs exert bidirectional regulatory effects, the plasma membrane executes the final cell death, and the Golgi apparatus and peroxisomes assist in lipid metabolism, collectively governing the progression of ferroptosis (Figure 2).

Display Full Size

Download

Figure 2. The global regulation of ferroptosis by inter-organelle crosstalk. Extracellular Fe³⁺ bound to TF is internalized into early endosomes via clathrin-dependent endocytosis. In lysosomes, Fe³⁺ is reduced to Fe²⁺ by STEAP3 and exported to the cytosolic LIP via DMT1. Under stress, NCOA4-mediated ferritinophagy releases ferritin-sequestered Fe²⁺ into the LIP, driving the Fenton reaction. LIP-derived Fe²⁺ traffics to mitochondria, where it is either sequestered by MtFT for safe storage, used for de novo Fe-S cluster assembly, or drives local accumulation of PUFA-PL-OOH. Mitochondria exert autonomous antioxidant defenses via the mGPX4-GSH and DHODH-CoQ10H2 axes. The ER determines ferroptosis susceptibility via bidirectional lipid remodeling: the ACSL4-LPCAT3 cascade produces pro-ferroptotic PUFA-PLs, whereas the ACSL3-MBOAT1/2 axis generates anti-ferroptotic MUFA-PLs. The non-canonical GPX1-OSBPL8 axis at the ER membrane further inhibits lipid peroxidation propagation. The ER also orchestrates LD biogenesis. LDs function as metabolic buffers by sequestering toxic PUFAs in neutral lipids, and their surface recruits the FSP1-CoQ10H2 system for membrane protection. Conversely, stress-induced lipolysis and lipophagy mobilize LDs to release pro-ferroptotic PUFAs. The plasma membrane suppresses lipid peroxidation via lipid remodeling and the RTAs generated by the FSP1-CoQ10H2 axis, while the ESCRT-III machinery provides a critical fail-safe by repairing damaged membranes to delay ferroptosis. Created in BioRender. Zheng, H. (2026) https://BioRender.com/d2ugadu. Fe³⁺: ferric iron; TF: transferrin; Fe²⁺: ferrous iron; vSTEAP3: six-transmembrane epithelial antigen of prostate 3; LIP: labile iron pool; DMT1: divalent metal transporter 1; NCOA4: nuclear receptor coactivator 4; MtFT: mitochondrial ferritin; Fe-S: iron-sulfur; ABCB7/8: ATP binding cassette subfamily B 7/8; PUFA-PL-OOH: polyunsaturated fatty acid phospholipid hydroperoxides; mGPX4: mitochondrial glutathione peroxidase 4; GSH: glutathione; DHODH: dihydroorotate dehydrogenase; CoQ10H2: coenzyme Q10-hydroquinone; ER: endoplasmic reticulum; ACSL4: long-chain-fatty-acid-CoA ligase 4; LPCAT3: lysophosphatidylcholine acyltransferase 3; PUFA-PL: polyunsaturated fatty acid phospholipid; ACSL3: long-chain-fatty-acid-CoA ligase 3; MBOAT1/2: membrane-bound glycerophospholipid O-acyltransferase 1/2; MUFA-PL: monounsaturated fatty acid phospholipid; GPX1: glutathione peroxidase 1; OSBPL8: oxysterol binding protein-like 8; LD: lipid droplet; FSP1: ferroptosis suppressor protein 1; RTA: radical-trapping antioxidant; ESCRT-III: endosomal sorting complex required for transport III.

4. Inter-Organelle Communication and Global Regulatory Network

Ferroptosis regulation is not the result of independent operation by a single organelle, but rather relies heavily on a multi-layered, dynamically integrated global regulatory network constructed through interorganellar communication. A landmark study has demonstrated that lysosomes, the ER, mitochondria, LDs, the Golgi apparatus, and peroxisomes can achieve precise and efficient exchange of lipids, metabolites, and signaling molecules via MCSs^[127-129]. The dynamic trafficking of iron ions, lipids, and ROS across different organelles collectively determines cellular ferroptosis susceptibility and ultimate cell fate.

4.1 Iron homeostasis

Iron ions function as the catalytic drivers of the Fenton reaction that initiates the chain of lipid peroxidation underlying ferroptosis. The maintenance of iron homeostasis depends on a highly coordinated and tightly coupled communication network among various organelles. As previously described, iron ions are internalized through physical connections between the plasma membrane and early endosomes, undergo conversion from Fe³⁺ to Fe²⁺ in lysosomes, and ultimately enter the cytosolic LIP^[31,34]. During this process, V-ATPase maintains the lysosomal pH at 4.5-5.0, enabling conformational dissociation of Fe³⁺ from TF^[33,34].

To prevent excessive iron release caused by lysosomal over-acidification, the core ferroptosis defense protein SLC7A11, in addition to its canonical plasma membrane localization, also resides on the lysosomal membrane^[130]. It mediates proton leakage to achieve a dynamic balance and precisely regulate the rate of iron release^[33,34,130]. This non-canonical lysosomal function reveals a dual-layered ferroptosis defense mechanism of SLC7A11, expanding the prior view that SLC7A11 acts solely via redox modulation. As the core catalytic subunit of system Xc^-, the canonical plasma membrane pool mediates cystine uptake to sustain GSH biosynthesis and lipid peroxide clearance, whereas the lysosomal pool tunes upstream iron homeostasis by regulating lysosomal acidity. These two functionally distinct pools act in concert to determine cellular ferroptosis sensitivity, refining our mechanistic understanding of SLC7A11-mediated defense.

Studies have shown that, to avoid non-specific oxidative damage caused by excessive accumulation of Fe²⁺ in the cytosolic LIP, early endosomes carrying exogenous iron form transient physical contacts with mitochondria, bypassing the LIP to directly transfer iron ions into the mitochondrial matrix^[131]. Such direct organellar contacts effectively bypass the bulk cytosolic LIP, providing a spatial mechanism for maintaining safe, low-level cytosolic iron concentrations.

Iron ions taken up by mitochondria are used for two primary purposes: the biosynthesis of Fe-S clusters and the synthesis of heme. Intriguingly, the heme-regulated inhibitor pathway triggered by impaired heme synthesis due to mitochondrial iron homeostasis dysregulation, and the PERK pathway triggered by abnormal ER lipid folding, achieve precise integration of inter-organelle signals at the core node ATF4^[132].

Meanwhile, iron flux is tightly regulated by the cytosolic ubiquitin-proteasome system (UPS). The substrate receptor subunit of the SCF-type ubiquitin ligase, FBXL5, acts as a dual sensor for iron and oxygen, recognizing and mediating the ubiquitination and degradation of IRP2^[96,133-135]. This cascade reaction of the FBXL5-IRP2 signaling axis, triggered by fine-tuning of lysosomal iron export, directly sends strong signals across compartments to the nucleus to reprogram the global iron metabolism transcriptional network, forming the first line of defense that determines cellular ferroptosis susceptibility.

The regulation of mitochondrial iron homeostasis also relies heavily on NEET family proteins located at MERCs, primarily including mitochondria-localized CISD1 and ER-localized CISD2^[136-138]. CISD1 limits excessive mitochondrial iron uptake^[139] and also interacts with the OMM voltage-dependent anion channel (VDAC) to further fine-tune mitochondrial iron metabolism^[140]. Furthermore, NEET proteins mediate inter-organelle transport of Fe-S clusters at MERCs, while simultaneously sensing local ROS concentrations to potently neutralize and inhibit local lipid peroxidation^[137,141]. Thus, they maintain cellular iron homeostasis by mediating material exchange and signal transmission between mitochondria and the ER.

Collectively, the maintenance of intracellular iron homeostasis is a multi-organellar relay process: from the directional transfer of iron ions via endosome-mitochondria direct contacts, to the nuclear-cytoplasmic “pseudo-iron deficiency” transcriptional reprogramming triggered by impaired mitochondrial Fe-S cluster synthesis; from the iron level sensing coordinated by lysosomal iron export and the UPS system, to the iron metabolism regulatory barrier formed by NEET proteins at MERCs. Dysfunction of any inter-organelle communication node will release free iron, this lethal catalyst, into unprotected membrane lipids, ultimately driving the outbreak of ferroptosis through a tightly regulated cascade reaction (Figure 3).

Display Full Size

Download

Figure 3. Inter-organelle coordinated regulation of cellular iron homeostasis in ferroptosis. Extracellular Fe³⁺ bound to TF is internalized into early endosomes, subsequently transported to lysosomes, and reduced to Fe²⁺. Early endosomes can also directly transport Fe²⁺ to mitochondria via DMT1 for Fe-S cluster and heme biosynthesis. Mitochondrially synthesized Fe-S clusters are transported to ER via NEET family proteins (mitochondria-localized CISD1 and ER-localized CISD2), while CISD1 concurrently limits excessive mitochondrial iron uptake. Impaired mitochondrial heme synthesis activates the HRI pathway, and ER protein folding dysfunction activates the PERK pathway; both converge on eIF2α phosphorylation to regulate ATF4, integrating mitochondrial and ER stress signals. In addition, cytosolic FBXL5 functions as a dual sensor for iron and oxygen, recognizing and mediating the ubiquitination and degradation of IRP2. This FBXL5-IRP2 signaling axis transmits signals across compartments to the nucleus, reprogramming the global transcriptional network of iron metabolism. Created in BioRender. Zheng, H. (2026) https://BioRender.com/k2nr1t9. Fe³⁺: ferric iron; TF: transferrin; Fe²⁺: ferrous iron; DMT1: divalent metal transporter 1; Fe-S: iron-sulfur; ER: endoplasmic reticulum; CISD1/2: CDGSH iron-sulfur domain-containing protein 1/2; HRI: heme-regulated inhibitor; PERK: protein kinase R-like ER kinase; eIF2α: eukaryotic translation initiation factor 2α; ATF4: activating transcription factor 4; FBXL5: F-box and leucine-rich repeats protein 5; IRP2: iron regulatory protein 2.

4.2 Lipid remodeling

As previously described, the ER, as the largest lipid synthesis factory in cells, generates highly oxidizable PUFA-PLs via its resident ACSL4-LPCAT3 axis, providing lethal substrates for ferroptosis^[45,46]. Meanwhile, peroxisomes independently synthesize the plasmalogen precursor AGP, which contains higher levels of PUFAs^[125]. These precursors cannot complete their final maturation within peroxisomes and must be transported to the ER via MCSs for subsequent processing into PUFA-ePLs^[125]. Recent studies have further revealed that the Golgi apparatus is also deeply involved in ferroptosis-related lipid remodeling. The Golgi-resident scaffold protein monocyte-to-macrophage differentiation-related forms an inter-organelle functional complex with ER-localized ACSL4 and MBOAT7, specifically driving the spatial remodeling of polyunsaturated PIs^[46,142]. This peroxisome-ER-Golgi lipid biosynthesis pipeline fundamentally determines the oxidative susceptibility of the entire cellular membrane system.

To safely store these lethal ferroptotic substrates, cells have evolved a highly dynamic inter-organelle crosstalk network among the ER, LDs, and lysosomes. Within the ER bilayer, esterification reactions mediated by enzymes such as DGAT1/2 and acyl-CoA cholesterol acyltransferase 1/2 catalyze the synthesis of neutral lipids. When the concentration of neutral lipids exceeds a critical threshold, phase separation occurs, forming physical aggregates that mix with surrounding phospholipids to create neutral lipid lens structures^{[103,143-146]}. These neutral lipid aggregates then bud from the ER membrane and form the LD assembly complex (LDAC) through the interaction of seipin with LD assembly factor 1 (also known as TMEM159)^[147,148]. A conformational change in seipin (from a closed to an open state) opens the dome-shaped LDAC structure toward the cytosol, facilitating the release of nascent LDs into the cytoplasm^[146]. On one hand, highly oxidizable PUFAs synthesized in the ER are stored as neutral lipids in LDs, which fundamentally prevents excessive incorporation of PUFAs into the phospholipid bilayers of membrane systems, reduces the substrate pool for lipid peroxidation, and enhances cellular ferroptosis resistance^[29,145]. On the other hand, under severe stress conditions such as hypoxia, excessive autophagy activation, or ferroptosis induction, LDs undergo lipolysis or lipophagy, are engulfed and fused with autophagosomes and lysosomes, and release large amounts of free PUFAs^[107]. These free PUFAs are then recruited back to membrane systems such as the ER, where they are activated and incorporated into membrane phospholipids, significantly increasing cellular ferroptosis susceptibility^[16,149,150].

These pro-ferroptotic lipids must encounter ROS to actually trigger ferroptosis, a process that is highly dependent on the lipid transport network at MERCs. In the narrow gap between the ER and the OMMs, lipid transfer proteins such as oxysterol-binding protein-related proteins 5/8 and NIR2 rapidly transport newly synthesized ER lipids to the OMM via non-vesicular transport^[151,152]. Mitochondria continuously generate abundant ROS through their active ETC and TCA cycle. When ER-derived PUFA-PLs are transported to mitochondria via MERCs and encounter high concentrations of ROS, a local lipid peroxidation storm is triggered^[153]. The resulting lipid peroxides not only disrupt mitochondrial morphology and function but also transmit oxidative signals back to the ER via the phosphorylated RMDN3-VAPB complex at MERCs^[154].

As previously described, during the progression of the lipid peroxidation chain reaction, cells also attempt to counteract oxidative damage through membrane lipid remodeling. On one hand, the ACSL3-MBOAT1/2 axis synthesizes MUFA-PLs to competitively replace PUFA-PLs in membranes^[25,49]. On the other hand, the calcium-independent phospholipase A2β (iPLA2β)-mediated oxidized lipid repair network specifically cleaves oxidatively damaged fatty acid tails^[52-54]. The mainstream view is that when the synthesis of anti-ferroptotic lipids and the repair of damaged lipids are insufficient to counteract lipid peroxidation damage, toxic oxidized phospholipids generated and remodeled in the ER are primarily delivered to the plasma membrane via vesicular trafficking and membrane fusion. Accumulation of these toxic lipids beyond the cellular repair threshold drives permeabilization and rupture of the plasma membrane, which constitutes the execution phase of ferroptosis^[2,48].

In summary, from the perspective of lipid metabolism regulation, ferroptosis is the collective outcome of lethal substrate synthesis in the peroxisome-ER axis, directional lipid sorting in the Golgi-ER axis, stress-induced lipid release from the lysosome-LD axis, lipid-ROS signal integration at the ER-mitochondria interface, and final mechanical rupture of the plasma membrane. No single organelle acts independently; rather, physical contacts and material exchanges among organelles collectively determine the ultimate fate of the cell (Figure 4).

Display Full Size

Download

Figure 4. Inter-organelle coordinated regulation of lipid metabolism in ferroptosis. In peroxisomes, DHAP and acyl-CoA are catalyzed by GNPAT, FAR1, and AGPS to generate AGP. AGP is subsequently transported to the ER, where it is processed by AGPAT3 and TMEM189 to form PUFA-ePLs. Meanwhile, the ER converts PUFAs into PUFA-PLs via the ACSL4-LPCAT3 cascade. Golgi-resident scaffold protein MMD promotes the ER-localized ACSL4-MBOAT7 complex to specifically remodel the spatial distribution of PUFA-PIs. PUFA-CoAs are catalyzed by DGATs and ACATs to form TAGs and CEs. The LDAC, formed by seipin and TMEM159, mediates the biogenesis of LDs with TAGs and CEs as the core. Under physiological conditions, LDs sequester PUFAs to inhibit ferroptosis, whereas under stress, they release free PUFAs via lipolysis and lipophagy, which are then recruited back to the ER for membrane phospholipid synthesis. ER-synthesized PUFA-PLs are transported to mitochondria via ORPs at MERCs, leading to massive accumulation of PUFA-PL-OOH in mitochondria. Conversely, the ER generates MUFA-PLs via the ACSL3-MBOAT1/2 cascade, which competitively replace PUFA-PLs in membranes and reduce cellular ferroptosis susceptibility. Created in BioRender. Zheng, H. (2026) https://BioRender.com/y5jcj1k. DHAP: dihydroxyacetone phosphate; acyl-CoA: acyl-coenzyme A; GNPAT: glyceronephosphate O-acyltransferase; FAR1: fatty acyl-CoA reductase 1; AGPS: alkylglycerone phosphate synthase; AGP: 1-O-alkylglycerol-3-phosphate; AGPAT3: acylglycerol-3-phosphate O-acyltransferase 3; TMEM189: transmembrane protein 189; PUFA-ePLs: polyunsaturated fatty acid-rich ether phospholipids; PL: phospholipids; ACSL4: long-chain-fatty-acid-CoA ligase 4; LPCAT3: lysophosphatidylcholine acyltransferase 3; MBOAT1/2/7: membrane-bound glycerophospholipid O-acyltransferase 1/2; PI: phosphatidylinositols; DGAT: diacylglycerol acyltransferase; ACAT: acetyl-CoA acetyltransferase; TAG: triacylglycerol; CE: cholesteryl esters; LDAC: lipid droplet assembly complex; TMEM159: transmembrane protein 159; ORP: oxysterol-binding protein-related protein; MERC: mitochondria-ER contact site; PUFA-PL-OOH: polyunsaturated phospholipid hydroperoxide; MUFA: monounsaturated fatty acid; ACSL3: long-chain-fatty-acid-CoA ligase 3.

4.3 Redox homeostasis

At its molecular core, ferroptosis represents a global oxidative catastrophe arising from a breakdown in the balance between cellular pro-oxidant drivers and antioxidant defense systems. The ER, mitochondria, peroxisomes, the Golgi apparatus, lysosomes, and LDs collectively maintain redox homeostasis through intercompartmental shuttling of reducing equivalents via vesicular trafficking, MCSs, and signal transduction. Under pathological conditions, once this delicate balance is disrupted, oxidative stress signals undergo cascade amplification through the inter-organelle crosstalk network, ultimately leading to the onset of ferroptosis.

Within the inter-organelle redox regulatory network, MERCs occupy an irreplaceable central position. MERCs serve as key organellar exchange hubs for cellular ROS regulation, lipid transport, and ion exchange^[155-157]. Multiple redox-related enzymes, including calnexin, ER oxidoreductase 1α (Ero1α), and SUMO-specific protease 1, localize and function at these sites to modulate ROS signal transduction^[158,159]. The ER lumen is the primary Ca²⁺ storage reservoir in mammalian cells^[160]. Under oxidative stress conditions, the ER releases Ca²⁺ via the sigma-1 receptor-inositol 1,4,5-trisphosphate receptor (IP3R) cascade, and the tight structural architecture of MERCs directs and concentrates the surge of Ca²⁺ for rapid mitochondrial uptake^{[155,161-164]}. Mitochondrial Ca²⁺ overload accelerates the TCA cycle and hyperactivates the ETC, directly triggering mitochondrial depolarization and a lethal burst of mtROS^{[156,165-167]}. More critically, when mitochondrial oxidative stress exceeds the local antioxidant capacity, mitochondrially generated H₂O₂ propagates to the adjacent ER lumen via MERCs, rather than via active vesicular transport. Mitochondrial oxidative-stress signals can be transmitted to the ER, inducing over-activation of the ER-resident disulfide-forming oxidase Ero1α, which in turn increases intraluminal H₂O₂ accumulation, disrupts disulfide bond formation, and promotes the buildup of misfolded proteins^{[153,168,169]}. Reciprocally, sustained ER stress enhances the Ca²⁺-releasing activity of IP3R, which triggers mitochondrial Ca²⁺ overload and metabolic dysfunction through MERCs, further augmenting mtROS release and establishing a deleterious feedforward vicious cycle^[170-173]. This inter-organelle signal transmission continuously amplifies the initial mild oxidative stress, eventually generating a ROS storm that overwhelms the cellular antioxidant defense system.

In response to redox imbalance, multiple organelles coordinately mount a global ferroptosis defense network. First, CoQ10 precursors are synthesized via the cytosolic-ER-associated mevalonate pathway, then transported to the mitochondrial matrix and inner membrane for head group modifications and final assembly^[93,174]. Assembled CoQ10 not only participates in the DHODH antioxidant defense axis at the IMM^[15], but also is delivered to LDs and the plasma membrane via specific transport systems such as StAR-related lipid transfer protein 7, contributing to the FSP1-mediated ferroptosis defense system^[14,20,175]. In parallel, reducing equivalents carried by NADPH are exchanged between mitochondria and the cytosol through metabolic shuttle systems, supporting the antioxidant function of FSP1 and GPX4^[176].

Peroxisomes represent another critical node in the inter-organelle redox network, engaging in bidirectional crosstalk with mitochondria to shape cellular redox status. On the one hand, they drive pro-oxidative events by generating ROS via the interconversion of xanthine dehydrogenase and xanthine oxidase^[177,178]. On the other hand, they confer antioxidant protection via enzymes such as catalase, SOD1, and peroxiredoxin 5^{[176,179,180]}. Strikingly, recent studies reveal that peroxisomes establish direct physical MCSs with mitochondria via the tethering of the peroxisomal membrane protein ACBD5 and the mitochondrial outer membrane protein PTPIP51. Under basal conditions, these contacts facilitate the anterograde transfer of peroxisome-derived ROS into mitochondria. Conversely, under mitochondrial oxidative stress, these MCSs mediate the retrograde transfer of mitochondrial ROS into peroxisomes, where they are efficiently neutralized by peroxisomal antioxidant enzymes, thereby maintaining mitochondrial redox homeostasis^[181].

Furthermore, when the ER is severely damaged by oxidative stress, it activates PERK, which significantly upregulates the expression of the nuclear transcription factor ATF4 via phosphorylation of eIF2α^[182,183]. In response to redox imbalance signals, ATF4 mediates compensatory upregulation of system Xc^- to increase cystine uptake^[184]. Meanwhile, it cooperates with the antioxidant transcription factor NRF2 to transcriptionally activate a large battery of ferroptosis defense genes involved in iron metabolism, NADPH production, and GSH synthesis, attempting to reestablish cellular antioxidant defenses^[185,186]. Notably, in parallel to the ATF4-SLC7A11 cystine import axis, our group discovered that the stress-responsive transcription factor ATF3 is activated under cystine deprivation and directly upregulates CBS transcription to promote transsulfuration-dependent cysteine synthesis as a compensatory adaptive response^[18]. Beyond transcriptional regulation, our group demonstrated that transsulfuration-derived H₂S modulates intracellular cysteine pools via post-translational modification. Mechanistically, H₂S persulfidates SAHH at Cys195 to inhibit its activity and reduce homocysteine production, thereby restricting substrate supply for the transsulfuration cascade and diminishing endogenous cysteine biosynthesis^[22].

In summary, from the redox perspective, the initiation and progression of ferroptosis are governed by an interwoven complex network comprising the ER-mitochondria oxidative signaling loop, intercompartmental shuttling of reducing equivalents among the ER, mitochondria, plasma membrane, and LDs, oxidative buffering and homeostasis maintenance by the peroxisome-mitochondria axis, and global transcriptional regulation by the nucleus. As the core signaling molecules, ROS mediate cell-wide redox signal transmission and integration through their dynamic trafficking and signal amplification across organelles (Figure 5).

Display Full Size

Download

Figure 5. Inter-organelle coordinated regulation of cellular redox homeostasis in ferroptosis. The ER synthesizes CoQ10 precursors via the mevalonate pathway, which are transported to mitochondria for maturation. Mature CoQ10 supports both the mitochondrial DHODH antioxidant axis and the FSP1-CoQ10 antioxidant pathway on LDs and the plasma membrane. ER-released Ca²⁺ is rapidly taken up by mitochondria through MERCs, triggering a burst of mtROS and lipid peroxidation. The generated mtROS oxidizes adjacent IP3R Ca²⁺ channels to trigger calcium release. Phosphorylated RMDN3 promotes MERCs assembly and shuttles mitochondrial lipid radical to the ER, aggravating ER oxidative stress. Under ER oxidative stress, Ero1α further potentiates IP3R‑dependent calcium efflux, culminating in a positive oxidative‑stress‑amplifying feedback loop between mitochondria and the ER. In addition, peroxisomes generate ROS via the interconversion between XDH and XO, which synergistically exacerbates mitochondrial oxidative damage. At the transcriptional level, ER stress activates the PERK pathway, which upregulates ATF4 via phosphorylation of eIF2α. ATF4 increases cystine uptake by upregulating SLC7A11. Intracellular cysteine is also generated via the methionine cycle and transsulfuration pathway. H₂S-mediated persulfidation inactivates the key enzyme SAHH, thereby blocking this metabolic pathway. Meanwhile, the antioxidant transcription factor NRF2 transcriptionally activates genes related to iron metabolism, NADPH production, and GSH synthesis, collectively constructing the global cellular antioxidant defense system. Created in BioRender. Zheng, H. (2026) https://BioRender.com/2b4xpt7. CoQ10: coenzyme Q10; DHODH: dihydroorotate dehydrogenase; FSP1: ferroptosis suppressor protein 1; MERC: mitochondria-ER contact site; H₂O₂: hydrogen peroxide; mtROS: mitochondrial reactive oxygen species; IP3R: inositol 1,4,5-trisphosphate receptor; RMDN3: regulator of microtubule dynamics protein 3; Ero1α: ER oxidoreductase 1α; XDH: xanthine dehydrogenase; XO: xanthine oxidase; PERK: protein kinase R-like ER kinase; ATF4: activating transcription factor 4; eIF2α: eukaryotic translation initiation factor 2α; SAH: S-adenosylhomocysteine; SAHH: adenosylhomocysteinase; NRF2: nuclear factor erythroid 2-related factor 2; NADPH: nicotinamide adenine dinucleotide phosphate; GSH: glutathione.

In essence, the core mechanism by which inter-organelle communication regulates ferroptosis is the precise global control of the “sequestration” and “release” of three key metabolic hazards: labile free iron, oxidizable polyunsaturated lipids, and ROS. When inter-organelle communication favors the physical compartmentalization and safe storage of these high-risk metabolites, cells can build robust antioxidant defenses and effectively resist ferroptosis. However, when external stress or pathological factors disrupt this delicate balance, the organellar communication network shifts to promote the massive release of catalytic iron ions and lethal lipid substrates, while continuously amplifying oxidative stress signals across organelles. This spatiotemporal convergence of “fuel” (lipid substrates) and “ignition source” (ROS) ultimately triggers an irreversible lipid peroxidation cascade storm, leading to plasma membrane rupture and cell death.

5. Organelle-Targeted Therapeutic Strategies for Ferroptosis

5.1 Cancer

Ferroptosis is increasingly recognized as an intrinsic tumor suppressor mechanism. Although tumor cells have evolved diverse mechanisms to resist ferroptosis, their inherent ferroptosis susceptibility provides a vulnerable therapeutic target for cancer treatment.

Among all subcellular organelles, lysosomes, acting as iron storage reservoirs and core compartments of hydrolytic enzymes, have emerged as one of the most promising intervention targets in the ferroptosis pathway. The team led by Raphaël Rodriguez demonstrated that salinomycin specifically binds iron ions and sequesters them in lysosomes, inducing local Fenton reactions and lipid peroxidation to selectively eliminate drug-resistant, self-renewing cancer stem cells^[187]. In recent years, the same team leveraged the characteristic that some drug-resistant cancer cells overexpress CD44 to increase iron uptake for their proliferative demands. They delivered the synthetic small molecule fentomycin-1 specifically to lysosomes of target cells via CD44-mediated endocytosis, which activated stored free iron in lysosomes and triggered a lipid peroxidation cascade in situ, resulting in potent tumor cell killing^[23]. Furthermore, based on the lysosomal autophagic degradation mechanism, autophagy-targeting chimeras (AUTACs) developed by Yan Cheng’s team can specifically recruit the key ferroptosis defense protein GPX4 to lysosomes for autophagic degradation, exhibiting robust in vitro and in vivo antitumor efficacy in breast cancer models^[188].

Mitochondria serve as the central hub for cellular energy metabolism and ROS production, and their metabolic status directly determines cellular ferroptosis sensitivity. Therefore, targeting mitochondrial metabolic reprogramming represents a critical strategy for ferroptosis-based cancer therapy. The DHODH-CoQ10 antioxidant axis at the IMM constitutes a major, GPX4- and FSP1-independent ferroptosis defense system. To date, translational clinical development targeting DHODH has centered on two small-molecule inhibitors: the classic potent agent brequinar, and the new-generation orally bioavailable, highly selective inhibitor BAY2402234 (orludodstat). As the earliest developed DHODH inhibitor, brequinar has undergone systematic preclinical validation both in vitro and in vivo across a broad spectrum of malignancies, including melanoma^[189], lung cancer^[190], acute myeloid leukemia^[191], glioblastoma^[192] and so on. It has also been advanced into multiple clinical cohorts to evaluate its efficacy in patients with advanced melanoma^[193], head and neck squamous cell carcinoma^[194], advanced lung cancer^[195], acute myeloid leukemia^[196], and advanced gastrointestinal cancer^[197]. In contrast, BAY2402234, as a next-generation drug candidate, exerts its pharmacological action by specifically binding to the ubiquinone-binding pocket of DHODH to block de novo pyrimidine biosynthesis. Its clinical development has thus far been largely restricted to myeloid malignancies; however, its phase I trial (NCT03404726) enrolling patients with myeloid neoplasms such as relapsed/refractory acute myeloid leukemia was terminated prematurely as the expected clinical benefit was not achieved^[198]. Importantly, the tumor growth-suppressive effects of DHODH inhibitors are mechanistically primarily attributed to impaired tumor cell proliferation secondary to blockade of the de novo pyrimidine biosynthesis pathway. To date, no clinical or translational study has established a ferroptosis-specific therapeutic mechanism for these agents. Additionally, compounds developed to target mitochondrial iron homeostasis, including dexrazoxane^[199], pioglitazone^[200], and resveratrol^[201], have been shown to stabilize Fe-S clusters by targeting CISD1, exerting protective effects against ferroptosis-related organ damage.

In the field of anticancer nanomedicine, researchers have developed ER-targeted photodynamic therapy and photothermal therapy nanoplatforms. These nanoplatforms trigger robust oxidative stress in situ in the ER of tumor cells, induce immunogenic cell death, and activate systemic antitumor immune responses while directly eliminating tumor cells^[202].

The plasma membrane, as the terminal sensor and executor of ferroptotic signals, is also an important therapeutic target. Classical ferroptosis inducers such as erastin and multiple U.S. Food and Drug Administration (FDA)-approved drugs (e.g., sulfasalazine and sorafenib) have been demonstrated to induce tumor cell ferroptosis by inhibiting system Xc^- on the plasma membrane^[1,203-205]. Furthermore, small-molecule inhibitors including iFSP1 and FSEN1 can specifically target and block the FSP1-CoQ10 antioxidant defense system on the plasma membrane, thereby overcoming ferroptosis resistance in various cancer types^[20,206,207].

5.2 IRI

During IRI, ferroptosis is the core pathological mechanism driving massive cell necrosis and tissue damage in vital organs including the heart, brain, liver, and kidney. Organelle-based targeted intervention strategies can effectively alleviate IRI by restoring cellular antioxidant capacity and reestablishing lipid and iron homeostasis.

Currently, clinical interventions targeting ferroptosis in IRI primarily focus on inhibiting the LIP-mediated Fenton reaction. In ischemic or hemorrhagic brain injury, deferoxamine (DFO) has demonstrated a favorable safety profile in phase II clinical trials^[208] and shows potential to improve 180-day long-term neurological functional recovery in patients^[209]. Furthermore, clioquinol, which exhibits excellent blood-brain barrier (BBB) permeability, not only chelates free iron but also upregulates plasma membrane ferroportin to promote iron efflux, exhibiting superior therapeutic potential over conventional iron chelators in alleviating cerebral edema and ROS production^[210].

Another therapeutic strategy involves the use of antioxidants and organelle-targeted RTAs to directly block the membrane lipid peroxidation chain reaction. Among these, N-acetylcysteine (NAC), a key precursor for GSH synthesis, has been widely used in clinical studies of cardiovascular, renal, and hepatic IRI as well as hypoxic-ischemic encephalopathy (NCT01223326, NCT00564642, NCT04643821)^[6]. In myocardial ischemia-reperfusion models, lipophilic antioxidants such as Lip-1 downregulate the expression of the OMM protein VDAC1 and restore GPX4 expression, exerting significant cardioprotective effects^[211].

In addition, multiple approved clinical drugs have been shown to exert organ-protective effects by regulating ferroptosis. Perioperatively, dexmedetomidine attenuates myocardial ferroptosis by activating the AMPK/GSK-3β/NRF2 axis^[212]; propofol pretreatment potently inhibits ferroptosis and alleviates IRI via the SLC16A13-AMPK-GPX4 signaling pathway^[213]; and low-dose olaparib, an anticancer poly(ADP-ribose) polymerase inhibitor, improves cardiac dysfunction induced by ischemia or sepsis by accelerating mitophagy flux to promptly clear damaged, ROS-overloaded mitochondria^[214].

5.3 NDDs

The initiation and progression of NDDs are highly correlated with age-related pathological iron accumulation, exacerbated lipid peroxidation, and collapse of the antioxidant defense system in specific brain regions. Accumulating evidence indicates that ferroptosis is the important mechanism driving progressive neuronal loss.

Targeting the release of free iron from lysosomes, early small phase II clinical trials reported that injection of the iron chelator DFO could slow the rate of cognitive decline in patients with Alzheimer’s disease (AD)^[215]. However, a large phase II randomized double-blind clinical trial for Parkinson’s disease found that another iron chelator, deferiprone, although reducing iron levels in brain regions such as the substantia nigra, significantly accelerated the deterioration of motor symptoms^[216]. This intervention strategy of chelating lysosomal free iron has failed due to non-specific, indiscriminate depletion of iron, which severely disrupts the maintenance of normal neurological function.

Therefore, the development of specific organelle-targeted intervention strategies is imperative, and clinical research focus has shifted to RTAs that target lipid peroxidation in the plasma membrane and mitochondrial membrane. Edaravone, a moderate-potency RTA that scavenges lipid peroxyl radicals, has been approved by the FDA for long-term intravenous administration in amyotrophic lateral sclerosis. Clinical studies have also verified that it delays motor neuron degeneration through antioxidant and anti-ferroptosis effects^[217]. Another mitochondria-targeted RTA, CoQ10, failed to advance beyond phase II clinical trials due to poor BBB penetration^[218]. NAC and selenium preparations, which target GSH synthesis and GPX4 expression in mitochondria/cytosol respectively, have also not been successfully translated into clinical therapies due to narrow therapeutic windows and limited efficacy^[219,220].

Targeted lipid remodeling of the ER offers new hope for the clinical treatment of NDDs. In addition to its use in type 2 diabetes mellitus, pioglitazone has been found to potentially inhibit the expression of ER-localized ACSL4^[45]. Epidemiological evidence indicates that long-term pioglitazone use is associated with a reduced risk of dementia in patients with type 2 diabetes^[221]. It has been proposed that pioglitazone may provide long-term neuroprotective effects in NDDs by inhibiting the synthesis of ER lipid peroxidation substrates. In recent years, to overcome the limitations of traditional iron chelators such as non-specificity and low BBB penetration, nanomedicine delivery systems and multi-targeted organelle-specific smart delivery technologies have been rapidly developed. For example, researchers constructed a triphenylphosphonium-modified quercetin-derived smart nanomedicine using the self-assembly properties of plant polyphenols. The nanomedicine precisely targets mitochondria and sequesters excess iron ions in situ, significantly improving cognitive dysfunction in AD model mice^[222]. Furthermore, polycatechols, PDP and PLDP, consisting of dopamine and L-Dopa, respectively, can simultaneously achieve “subcellular iron/ROS buffering” and “regulation of pathogenic protein phase separation”, regulating ferroptosis progression in NDDs through multi-modal synergistic effects^[223].

5.4 Challenges in clinical translation

Despite promising progress in preclinical and early clinical studies, the translation of organelle-targeted ferroptosis therapies into clinical practice remains fraught with formidable challenges.

First, insufficient targeting specificity and off-target toxicity remain the primary bottlenecks. Most currently available ferroptosis modulators are multi-target compounds, making it difficult to establish a definitive causal relationship between ferroptosis induction and therapeutic efficacy. Furthermore, key ferroptosis regulatory proteins are ubiquitously expressed and essential for normal tissue homeostasis. Systemic intervention may indiscriminately disrupt cellular redox and iron homeostasis, thereby eliciting severe dose-limiting toxicities such as acute renal failure or profound immunosuppression.

Second, the organelle-targeted delivery efficiency and in vivo bioavailability of these agents are generally suboptimal. Many small-molecule modulators suffer from poor aqueous solubility and metabolic instability, which prevent their effective accumulation in specific subcellular compartments. For CNS disorders, these molecules face an additional formidable barrier: BBB penetration. Although emerging nanomedicine delivery systems have shown promise in achieving organelle-specific targeting, their in vivo biosafety, metabolic clearance profiles, and feasibility for large-scale manufacturing still require rigorous clinical validation.

Third, compensatory defense mechanisms and acquired drug resistance are highly prevalent. The multi-organelle ferroptosis regulatory network exhibits remarkable metabolic plasticity. When a defense node in a specific organelle is disrupted by pharmacological intervention, tumor cells often mount adaptive responses by upregulating parallel defense pathways in other subcellular compartments, ultimately leading to acquired resistance. A paradigmatic example is that targeted inhibition of the mitochondrial DHODH axis typically drives cells to rely on the plasma membrane-anchored FSP1-CoQ10 pathway. Overcoming this plasticity requires the design of rational combinatorial strategies.

Fourth, reliable clinical biomarkers for ferroptosis assessment are currently lacking. To date, no validated non-invasive biomarkers are available for the dynamic monitoring of in vivo ferroptosis levels and organelle functional status. Although emerging markers such as altered plasma membrane trafficking of TfR1 or hyperoxidized mitochondrial PRDX3 provide specific subcellular readouts, there is an urgent need for clinically feasible surrogate endpoint markers, such as signature circulating metabolites or targeted molecular imaging probes. This deficiency severely hampers patient stratification, efficacy evaluation, and dose optimization in clinical trials.

Looking ahead, the successful clinical translation of organelle-targeted ferroptosis therapies will hinge on three key advances: the development of highly selective organelle-targeted agents (e.g., PROTACs or AUTACs), the identification of ferroptosis-specific circulating or imaging biomarkers, and the exploration of combination therapies that block multiple inter-organelle interaction nodes. Precision therapeutic strategies tailored to the unique organelle metabolic profile of each patient will ultimately maximize therapeutic efficacy while minimizing off-target adverse effects.

6. Conclusions and Future Perspectives

Ferroptosis is the external manifestation of a lipid peroxidation chain reaction caused by ROS accumulation in the context of iron homeostasis dysregulation, and results from a spatial cascade reaction coordinated by multiple organelles including lysosomes, ER, mitochondria, plasma membrane, and LDs. Among them, lysosomes act as the central hub of iron metabolism to regulate iron storage and release; the ER dominates lipid synthesis and remodeling; mitochondria are responsible for ROS signal amplification and establish an independent antioxidant defense line; the Golgi apparatus participates in lipid sorting and trafficking; LDs play a bidirectional metabolic buffering role; and the plasma membrane serves as the final execution terminal of ferroptosis. Deciphering the multi-organelle spatial regulatory network of ferroptosis provides novel intervention directions for precision targeted therapy of cancer, IRI, and NDDs.

However, several key scientific questions remain to be elucidated in this field. First, does lipid peroxidation at the organelle level follow a “sequential cascade amplification” or “parallel and independent occurrence” pattern? Is it propagated from ER lipid peroxide accumulation to mitochondrial and plasma membrane rupture, or does global collapse of the antioxidant system lead to parallel and independent ferroptosis outbreaks in multiple PUFA-rich organelles? Although this review, drawing on currently available evidence, has systematically delineated the inter-organelle coordinated regulatory mechanisms using the sequential cascade amplification model as the organizing framework, it is critical to note that this model has yet to be validated by advanced imaging modalities with higher spatiotemporal resolution. Furthermore, the molecular underpinnings and pathophysiological contexts of the parallel and independent occurrence model remain to be thoroughly elucidated. Second, within the complex inter-organelle communication network, which organelle crosstalk events are the true lethal drivers of global ferroptosis, and which are local compensatory responses accompanying ferroptosis? Elucidating these questions has critical reference value for fundamentally determining clinical strategies for targeted intervention.

The rapid development of emerging technologies has also provided unprecedented opportunities for research on the organelle spatial regulation of ferroptosis. With the continuous advancement of single-molecule imaging technology, live-cell super-resolution dynamic tracking, cryo-electron tomography, spatial metabolomics, and high-sensitivity mass spectrometry, it is expected to achieve dynamic tracking and nanoscale spatial localization of metabolites and signaling molecules at the in vivo level. On this basis, the clinical translation of ferroptosis will gradually break through the bottlenecks of traditional drugs such as non-specific targeting and off-target effects. More precise combination therapeutic strategies will be developed around multi-organelle coordinated regulatory mechanisms, ultimately realizing precision ferroptosis intervention based on individual differences.

In this review, we provide a comprehensive overview of how direct crosstalk between mitochondria, ER, lysosomes, and LDs coordinates ferroptosis. The important contribution of nuclear transcriptional regulation to ferroptosis is acknowledged but not covered in depth, as this topic merits a separate dedicated review.

Authors contribution

Cai Y: Writing-original draft, writing-review & editing, visualization.

Zheng H: Visualization.

Zhang Y: Funding acquisition, supervision, writing-review & editing.

Conflicts of interest

The authors declare no conflicts of interest.

Ethical approval

Not applicable.

Consent to participate

Not applicable.

Consent for publication

Not applicable.

Availability of data and materials

Not applicable.

Funding

This work was supported by Fujian Provincial Natural Science Foundation of China (Grant No. 2024J01023 to Yongyou Zhang), and the Shenzhen Science and Technology Program (Grant No. JCYJ20240813145615020 to Yongyou Zhang).

Copyright

© The Author(s) 2026.