Mitochondrial dysfunction triggers a maladaptive peroxisomal response driving lipid accumulation Authors: Patrícia Coelho, Pedro Dionísio, Vilma Sardão Oliveira, Roberto A Dias, Matthias E Futschik, Robin Klemm, Ira Milosevic, Nuno Raimundo Published: 2026 ABSTRACT AbstractMitochondria and peroxisomes communicate to maintain lipid homeostasis, but how the latter adjust to mitochondrial dysfunction remains unclear. Here, we show that loss of complex I subunit NDUFS4 in mouse fibroblasts leads to impaired mitochondrial fatty acid oxidation, resulting in the accumulation of triacylglycerol and lipid droplet (LD) expansion. In this context, peroxisomal biogenesis is upregulated, but their β-oxidation capacity is impaired, suggesting an adaptive yet ineffective response. Additionally, lipid overload using a very-long-chain fatty acid (VLCFA) leads to peroxisomal proliferation but prevents LD expansion when peroxisomal β-oxidation is compromised. The data demonstrated that proper peroxisomal processing is necessary for lipid storage under mitochondrial stress conditions. Our findings reveal a peroxisomal maladaptive remodelling response that fails to compensate for mitochondrial dysfunction, leading to disruptions in LD homeostasis. We propose a critical axis involving peroxisomes-LD-mitochondria that buffers metabolic stress in mitochondrial diseases.Keywords: Mitochondria-peroxisome crosstalk, Complex I dysfunction, NDUFS4-KO, Peroxisomes, Lipid homeostasis Introduction Mitochondria coordinate energy production, lipid metabolism, redox signalling, and play a crucial role in cellular homeostasis. Impaired mitochondrial function triggers compensatory adaptations across multiple organelles [1,2], yet it remains poorly understood how other organelles adapt to mitochondrial dysfunction. Among mitochondrial impairments, respiratory complex I (NADH:ubiquinone oxidoreductase) dysfunction [3,4] is the most prevalent cause of mitochondrial disease and underlies a broad spectrum of severe clinical phenotypes, including Leigh syndrome, an incurable paediatric encephalopathy characterized by neurometabolic decline, brainstem lesions, and premature death [5,6]. Complex I is the largest and most intricate complex of the electron transport chain (ETC), responsible for electron transfer and proton pumping [5]. NDUFS4 subunit is a small accessory protein essential for the final assembly and stability of the fully functional complex. Loss of NDUFS4 leads to incomplete assembly of the complex, and is associated with metabolic impairments, specifically defective oxidative phosphorylation (OXPHOS) composed by decreased ETC activity, decreased mitochondrial membrane potential and adenosine triphosphate (ATP) production and decreased NAD+/NADH ratio, all contributing to a metabolic switch from OXPHOS to glycolysis [[7], [8], [9], [10], [11], [12], [13]]. NDUFS4 knockout (KO) cells and mice represent well-established models of complex I deficiency and mitochondrial disease, as they recapitulate key biochemical and metabolic features of Leigh syndrome, including defective respiration, NADH accumulation, altered redox state, and metabolic rewiring [5]. Despite this well-established clinical relevance, the metabolic consequences of complex I deficiency and the mechanisms by which they reshape inter-organelle signalling remain incompletely understood. Notably, mitochondrial complex I dysregulation results in lipid imbalance, characterized by accumulation of triacylglycerols (TAGs) and augmented lipid droplets (LDs), hallmarks of disrupted fatty acid (FA) metabolism [[14], [15], [16], [17], [18]], and inhibition of cholesterol synthesis [19]. These features are also observed across other forms of mitochondrial stress [20,21]. Notably, recent work in Ndufs4 mutant mice demonstrated that LD accumulation is a conserved and early response to neuronal mitochondrial dysfunction [17]. However, the mechanisms underlying metabolic rewiring and the consequent accumulation of LDs remain poorly defined. Furthermore, it remains unclear if the accumulation of LDs is a beneficial adaptation or if it contributes to the pathology. This gap is particularly important given that additional organelles share central roles in FA handling and stress adaptation, yet their contribution to cellular resilience in cells with respiratory chain dysfunction is not well characterized. FA catabolism is a pathway shared between mitochondria and peroxisomes, which together coordinate cellular β-oxidation [22,23]. Mitochondria primarily oxidize short-chain fatty acids (SCFAs, C2–C6), medium-chain fatty acids (MCFAs, C6–C12), and long-chain fatty acids (LCFAs, C12–C20), whereas very-long-chain fatty acids (VLCFAs, >C22) are initially shortened in peroxisomes before being transported to mitochondria for complete degradation. This inter-organelle cooperation prevents lipotoxicity and maintains lipid homeostasis [24]. Peroxisomes are highly dynamic organelles capable of adapting to metabolic and environmental stress by altering their number, morphology, and enzymatic function [25,26]. Besides their cooperation on cellular β-oxidation, peroxisomes and mitochondria share the membrane fission machinery and their biogenesis is triggered by common transcription factors [[27], [28], [29]]. Furthermore, assembly of respiratory chain super complexes depends on a lipid species that is synthesized in peroxisomes [30] and peroxisomal biogenesis is dependent on trafficking of pre-peroxisomal vesicles originated at mitochondria [31]. Yeast studies demonstrate that impaired mitochondrial respiratory chain can trigger peroxisomal proliferation, suggesting a possible compensatory role [32,33]. In mammals, mitochondrial inhibition by rotenone similarly alters peroxisomal dynamics, indicating that this crosstalk is conserved [34]. However, whether these peroxisomal changes represent a functional adaptation or a maladaptive response remains unclear. Notably, abnormal peroxisomal function impair mitochondrial homeostasis and alter lipid metabolism by inadequately processing VLCFAs [[35], [36], [37]]. Nevertheless, the impact of complex I deficiency on peroxisomal dynamics, and how this mitochondria-peroxisome communication contributes to altered lipid homeostasis in mitochondrial complex I dysfunctional models, has not been systematically investigated. Therefore, peroxisomal failure may constitute a central, yet overlooked, mechanism amplifying metabolic stress in complex I deficient cells. In this study, we show that NDUFS4-KO cells accumulate TAGs and display LD expansion in a context of impaired mitochondrial and peroxisomal fatty acid oxidation (FAO). Although peroxisomes increase in number, they remain functionally immature, indicating that this proliferative response fails to restore FA-handling capacity under complex I deficiency. Importantly, pharmacological and metabolic challenges reveal that functional peroxisomal β-oxidation is essential for supporting mitochondrial organization and for enabling LD remodelling, underscoring a reciprocal and previously underrated dependency between these organelles. Our work uncovers a maladaptive mitochondria-peroxisome crosstalk mechanism that drives lipid dysregulation on respiratory chain defects. It further identifies peroxisomal inefficiency as a key amplifier of metabolic and redox stress, underscoring peroxisomal FA processing as a crucial effector on cellular resilience in mitochondrial disease, including Leigh syndrome. Results Respiratory chain deficiency alters lipid homeostasis and promotes triacylglycerol storage in lipid droplets We took advantage of mouse fibroblasts lacking NDUFS4 (NDUFS4-KO) to test the impact of respiratory chain deficiency on lipid homeostasis. First, we confirmed that there was a mitochondrial impairment in these cells, with less activity of the respiratory chain, as expected (Fig. S1A–B), also reflected in decreased activity of complexes I to IV (Fig. S1C–D). We further confirmed that complex I enzymatic activity was specifically decreased in NDUFS4-KO cells (Fig. S1E). Mitochondrial membrane potential was decreased in NDUFS4-KO cells (Fig. S1F), while mitochondrial hydrogen peroxide levels (H2O2) were not changed (Fig. S1G). The respiratory chain defect observed in NDUFS4-KO cells had no effect on the mitochondrial morphology (Fig. S2). Given the pivotal role of mitochondria in lipid catabolism, we sought to test how loss of NDUFS4 affected the cellular lipid profile, by performing whole cell lipidomics (Fig. 1A). We found a distinct lipid pattern between the WT and NDUFS4-KO cells (Fig. 1B), with 83 increased (Table S1) and 30 decreased species (Table S2) in NDUFS4-KO (Fig. 1C). We then performed a gene set enrichment analysis (GSEA)-like analysis for several lipid classes and observed that triacylglicerides (TAG) were enriched, together with several structural lipids (phospholipids and ether-linked phospholipids), cholesterol esthers (CE), sphingomyelins (SM) and hexosylceramides (Fig. 1D). Moreover, overall lipid class abundance showed differences in CE, SM, ether-linked phospholipids and TAG groups, corroborating the GSEA-like findings (Fig. 1E). The decreased levels of CE observed in our lipidomic analysis of NDUFS4-KO (Fig. 1E) are in agreement with our previous data showing decreased cholesterol synthesis in cells with respiratory chain deficiency [19]. NDUFS4 deficiency leads to lipidome remodelling and triacylglycerol accumulation in MEFs. Mouse embryonic fibroblasts (MEF) cells were used 48 h post-plating for experiments. (A) Schematic representation of the global untargeted lipidomics workflow. Whole-cell extracts from WT and NDUFS4-KO MEFs were analyzed by mass spectrometry to generate lipid profiles. (B) Principal component analysis (PCA) of lipidomic datasets showing clear separation between WT and NDUFS4-KO samples. [N = 4]. (C) Volcano plot showing log fold-changes and significance of individual lipid species between NDUFS4-KO and WT cells. 30 lipid species were significantly decreased (blue) and 83 were significantly increased (red) in KO cells relative to WT. [N = 4]. (D) Lipid class enrichment analysis based on significantly altered lipid species. Bubble size indicates the number of significant hits per class; colour indicates the FDR-adjusted p-value. (E) Relative abundance of lipid classes in WT and NDUFS4-KO cells. Bars represent the mean percentage of total lipid abundance per class. [N = 4]. (F) Bubble plot showing fold-changes (KO vs. WT) of individual lipid species within each lipid class. Dot size is proportionate to statistical significance, where blue dots represent significantly altered lipid species. Values are shown as quantitative log2-fold changes. [N = 4]. (G) Quantification of total cellular triacylglycerol (TAG) levels normalized to protein content following the instructions in Triglyceride Assay Kit. [n1 = 4-5, n2 = 3-4, n3 = 5, n4 = 4; N = 4]. (H) Representative images of lipid droplets stained with BODIPY 493/503 (green) and nuclei (Hoechst; blue) in WT and NDUFS4-KO MEFs. Scale bar = 10 μm. (I) Quantification of lipid droplet number per cell and average lipid droplet area (μm2) from images in (H), using FIJI software for analysis. [n1 = 22-24, n2 = 15-19, n3 = 19; N = 3]. Data are presented as mean ± S.E.M. in bar graphs. n, number of 96-well evaluated (G) or number of cells (I); N, independent experiments; p-values were calculated using two-tailed t-tests in Microsoft Excel (C, F). Multiple testing correction (Benjamini-Hochberg FDR) used in (B, D), with significant hits at FDR ≤0.05. (C, F) Represent lipids with raw p-values ≤0.05 and |log2 fold change| ≥ 0.5, considered significant for biological relevance. Further functional validation supported the candidates; two-tailed Welch's t-test with significance from p < 0.05 and exact p-values indicated (C, E, G, I). n.s. means nonsignificant. We focused our attention on metabolically labile lipid species such as TAG, the main form of lipid storage in cells. The species of TAG that were significantly changed between WT and NDUFS4-KO were all up-regulated (Fig. 1F) supporting increased levels of TAG in NDUFS4-KO cells. To confirm the TAG increase, we measured TAG using a fluorometric assay and observed increased levels in NDUFS4-KO (Fig. 1G), in agreement with the lipidomics data. Cells can deal with excess of free FAs by incorporating them into TAGs, which are stored in LDs [38]. For that reason, we assessed LD abundance in WT and NDUFS4-KO cells (Fig. 1H) and observed that, while the number of LDs was not changed, there was increased LD area in NDUFS4-KO (Fig. 1I), in agreement with the increased levels of TAGs. These findings show that the loss of NDUFS4 disrupts mitochondrial function, triggering a metabolic shift towards lipid storage characterized by TAG accumulation in LDs. Mitochondrial FA import is impaired in NDUFS4-KO cells Because mitochondria are central to the process of FA β-oxidation and given the observed shift toward lipid storage in NDUFS4-KO cells, we hypothesized that the accumulation of TAGs and the expansion of LDs is due to impaired mitochondrial fatty acid degradation. Therefore, we investigated whether the loss of NDUFS4 disrupts the machinery needed for FA import and oxidation. To test this possibility, we assessed the levels of proteins involved in FA metabolism (Fig. 2A). Mitochondrial fatty acid β-oxidation starts with the activation of fatty acids by Acyl-CoA synthetase long-chain family member 1 (ACSL1). This activation allows the fatty acids to be transported to the mitochondria through the carnitine shuttle system, which involves carnitine palmitoyltransferase I (CPT-I) and carnitine palmitoyltransferase II (CPT-II). We found that ACSL1, CPT-I and CPT-II protein levels were decreased in NDUFS4-KO (Fig. 2B–D). To test if the changes at protein level were due to transcriptional regulation, we measured the expression of genes involved in FA catabolism, which were not significantly changed, except for a slight decrease in ACADVL (involved in the early steps of mitochondrial beta-oxidation) (Suppl Fig. S3A). This suggests that the low levels of FA metabolizing enzymes are due to a post-translational mechanism. Consistently, lipidomic analysis of the FA pool showed a selective accumulation of polyunsaturated fatty acids with 4 or more double bonds and of VLCFA at the C22 length (Suppl. Fig. S4A and B). Moreover, detailed analysis of TAG species abundance demonstrated a broad increase in TAG molecular species in NDUFS4-KO cells (Suppl. Fig. S4C), particularly those composed of longer chain lengths and higher degrees of unsaturation. This shift toward higher-mass TAGs indicates preferential sequestration of polyunsaturated and long chain FAs into storage lipids. These data further support a defect in mitochondrial FA catabolism, that promotes the accumulation and remodelling of TAG species, and possibly driving the expansion of LDs observed in NDUFS4-KO cells. NDUFS4-KO impairs mitochondrial lipid metabolism enzyme expression and ACLY activity. Mouse embryonic fibroblasts (MEF) cells were used 48 h post-plating for experiments. (A) Schematic representation of central carbon and lipid metabolism related to mitochondrial function highlighting enzymes investigated. (B–G) Immunoblot quantification of mitochondrial fatty acid oxidation and lipid metabolism enzymes in whole-cell extracts from WT and NDUFS4-KO cells. Densitometric analysis for ACSL1 (B) [N = 7], CPTI (C) [N = 5], CPTII (D) [N = 6], pACLY/ACLY (E) [N = 5], ACSS2 (F) [N = 5], and FASN (G) [N = 10], normalized to GAPDH and expressed relative to WT. (H) ACLY enzymatic activity measured fluorometrically with or without the ACLY inhibitor SB204990 (30 μM, 24h) [n1 = 4, n2 = 4, n3 = 4; N = 3]. (I–J) Immunoblot quantification of CPTI (I) [N = 4], and CPTII (J) [N = 4], protein levels in whole-cell extracts from WT and NDUFS4-KO cells following SB204990 incubation. (K) Representative images of lipid droplets stained with BODIPY 493/503 (green) and nuclei (Hoechst; blue) in WT and NDUFS4-KO MEFs with or without the ACLY inhibitor SB204990 (30 μM, 24h). Scale bar = 10 μm. (L) Quantification of lipid droplet number per cell and average lipid droplet area (μm2) from images in (K), using FIJI software for analysis. [n1 = 14-17, n2 = 13-16, n3 = 13-16; N = 3]. Data are presented as mean ± S.E.M. n, number of 96-well evaluated (H); N, independent experiments; two-tailed Welch's t-test group (B-G, I); two-way ANOVA followed by two-tailed Tukey's multiple comparison test (H, J, L). All statistical analyses with significance from p < 0.05 and exact p-values indicated; n.s. means nonsignificant. We then looked at proteins involved in lipid synthesis. First, we measured ACLY, which is the enzyme that metabolizes mitochondrial-derived citrate towards lipid biogenesis. ACLY levels were decreased in NDUFS4-KO cells (Fig. 2E). Because ACLY can be inhibited by phosphorylation at Ser455, we also checked the levels of this phosphorylation and found that, while there is less ACLY-Ser455-P, the ratio between phosphorylated and total ACLY is not affected in NDUFS4-KO (Fig. 2E). This result suggests that NDUFS4-KO cells are less able to mobilize citrate towards lipid biosynthesis. We then measured the proteins levels of Acetyl-CoA Synthase 2 (ACSS2), which converts acetate into acetyl-CoA, and observed no difference between WT and NDUFS4-KO (Fig. 2F). Next, we tested a later stage of the FA synthesis pathway and assessed the levels of fatty acid synthase (FASN), which were increased in NDUFS4-KO cells (Fig. 2G). These changes in the protein levels of lipid synthesis enzymes were not due to transcriptional regulation (Suppl Fig. 3B). Altogether, these results suggest decreased conversion of citrate to acetyl-CoA in NDUFS4-KO cells, and overall decreased lipid synthesis. In NDUFS4-KO cells, the decreased OCR can be a consequence of a reduced tricarboxylic acid (TCA) cycle activity and OXPHOS, thereby limiting citrate export to the cytoplasm [39,40]. This reduced citrate availability could diminish the demand for ACLY, suggesting that the decrease in ACLY levels may represent an adaptive response to metabolic flux. Given that ACLY links mitochondrial-derived citrate to cytosolic acetyl-CoA production for lipid synthesis, and that ACLY levels are decreased in NDUFS4-KO cells, we hypothesized that reduced lipid synthesis may serve as a compensatory mechanism to limit FA availability in a context of excess FAs. We first tested whether impaired FA oxidation is a consequence of decreased ACLY activity. We inhibited ACLY pharmacologically, both in WT and NDUFS4-KO cells, using SB-204990 (30 μM, 24h; see Suppl. Fig. S5 for optimization of concentration and incubation time) and assessed its impact on FA import proteins [41]. As expected, this treatment decreased ACLY activity, in WT cells, while having no effect in NDUFS4-KO cells, where the activity was already low (Fig. 2H). When ACLY was inhibited, there was an increase in CPT-I protein levels in both WT and NDUFS4-KO cells (Fig. 2I), suggesting that the decrease in FA mobilization observed in NDUFS4-KO cells is not due to lower ACLY activity. In contrast, CPT-II protein levels were significantly reduced upon ACLY inhibition in both WT and NDUFS4-KO cells (Fig. 2J). Moreover, treatment with SB-204990 did not affect LD size or number in either WT or NDUFS4-KO cells (Fig. 2K and L). These data suggest that the lower levels of ACLY are not driving the accumulation of LDs in NDUFS4-KO cells. Overall, these findings indicate that NDUFS4 deficiency disrupts mitochondrial FA oxidation at the level of FA import and processing, resulting in FA accumulation. Interestingly, this occurs despite a downregulation of key enzymes for de novo lipid synthesis (e.g., ACLY, SREBF1 [19] and without compensation of ACSS2) indicating that the observed buildup of TAGs and expansion of LDs is not driven by increased anabolic flux. Instead, the data support a model in which impaired mitochondrial respiratory chain limits FA catabolism, leading to the diversion of excess FAs into neutral lipid storage as TAGs. Peroxisomal biogenesis is activated but functionally impaired in NDUFS4-deficient cells Given the decrease in FA mobilization to mitochondrial β-oxidation and decreased mitochondrial aerobic capacity observed in NDUFS4-KO cells, we investigated whether peroxisomes, which also carry out β-oxidation of VLCFAs, might undergo adaptive changes in response to mitochondrial dysfunction, i.e. if peroxisomal β-oxidation might be activated as a compensatory mechanism (Fig. 3A). NDUSF4-KO have adaptive yet functionally inefficient peroxisomes. Mouse embryonic fibroblasts (MEF) cells were used 48 h post-plating for experiments. (A) Schematic model illustrating fatty acid oxidation involving coordinated mitochondrial and peroxisomal activity. (B) mRNA was extracted from WT and NDUFS4-KO cells, converted to cDNA and analyzed by quantitative PCR for peroxisomal-related genes, grouped by functional category [N = 3–8]. mRNA levels are normalized to housekeeping genes and expressed relative to WT. (C–G) Immunoblot quantification of peroxisomal biogenesis and import machinery proteins in whole-cell extracts from WT and NDUFS4-KO cells. Densitometric analysis for PEX3 (C) [N = 8], PEX19 (D) [N = 5], PEX11β (E) [N = 5], PEX14 (F) [N = 6], and PEX5 (G) [N = 6], normalized to GAPDH, Vinculin or Calnexin and expressed relative to WT. (H) Representative images of peroxisomes transfected with GFP-SKL (green) and stained with Hoechst (nuclei; blue) in WT and NDUFS4-KO MEFs. Scale bar = 10 μm. (I) Quantification of peroxisome number per cell and average area (μm2) from images in (H), using FIJI software for analysis. [n1 = 15-16, n2 = 16-30, n3 = 16-17; n4 = 14; N = 4]. (J–L) Immunoblot quantification of peroxisomal β-oxidation proteins in whole-cell extracts from WT and NDUFS4-KO cells. Densitometric analysis for ACOX1 (J) [N = 6], PMP70 (K) [N = 5], ACBD5 (L) [N = 4], normalized to GAPDH or Vinculin and expressed relative to WT. Data are presented as mean ± S.E.M.; n, number of cells; N, independent experiments; two-tailed Welch's t-test group with significance from p < 0.05 and exact p-values indicated (B-L). n.s. means nonsignificant. First, we assessed the transcriptional program of peroxisomal biogenesis by measuring the transcript levels of several genes involved in peroxisomal structure and function. We observed that genes related with peroxisomal biogenesis and maturation were up-regulated (Fig. 3B). The transcripts related to pexophagy or encoding peroxisomal β-oxidation enzymes were globally not changed (Fig. 3B). We then tested if the changes in peroxisomal biogenesis were also manifesting at protein level. Pex3, a peroxin essential for peroxisome biogenesis and maturation and which participates in mitochondrial-dependent de novo peroxisome formation [42], was significantly increased (Fig. 3C). Similarly, Pex11β, involved in peroxisomal elongation and division trough fission [43], was also significantly increased in NDUFS4-KO cells (Fig. 3D). Pex5, the cytosolic receptor that shuttles matrix proteins containing a peroxisomal targeting signal (PTS) 1 to the peroxisomal membrane [44,45], was also increased (Fig. 3E), whereas Pex14, a docking protein that forms the translocation channel by interacting with Pex5 [46], remained unchanged (Fig. 3F). These data indicate a shift toward increased peroxisome biogenesis and import activity, suggesting activation of the peroxisomal compartment as a compensatory response to mitochondrial dysfunction. To confirm if this potential for increased peroxisomal biogenesis is translating into actual biogenesis, we used the GFP-SKL construct, a recombinant green fluorescent protein (GFP) fused to the peroxisomal targeting signal 1 (PTS1) sequence and composed by the tripeptide Ser-Lys-Leu (SKL). This construct is imported to the peroxisomal matrix, serving as reporter to peroxisomal biogenesis and import capacity [47]. We observed that GFP-SKL is imported in both WT and NDUFS4-KO cells (Fig. 3G), with increased number of GFP-positive peroxisomal particles in NDUFS4-KO cells (Fig. 3H). The area of peroxisomes is not affected (Fig. 3H). These results are also supported by increased PMP70+- and Pex14+-positive puncta in KO cells, corroborating increased peroxisomal number using multiple markers (Suppl. Fig. 6). No differences in pexophagy were observed (Suppl. Fig. 7A and B), using colocalization between GFP-SKL and the lysosomal integral membrane protein lysosome-associate membrane protein 1 (LAMP1) as proxy for pexophagy. These results are compatible with increased peroxisomal biogenesis as compensatory to mitochondrial aerobic malfunction. We next tested if the increase in peroxisome biogenesis materialized into functional peroxisomes. Pex19, a critical cytosolic chaperone and import receptor for peroxisomal membrane proteins (PMPs) [48], was significantly decreased in NDUFS4-KO (Fig. 3I). As Pex19 plays a dual role in early peroxisomal biogenesis by facilitating both membrane protein targeting and insertion, its reduction likely impairs the proper assembly of newly formed peroxisomes. Although the transcript levels of most peroxisomal β-oxidation genes were unchanged or even slightly increased (e.g., HSD17B4), ABCD3, encoding the PMP transporter PMP70 [49], was decreased (Fig. 3B), consistent with the reduced protein levels observed (Fig. 3J). Given the essential role of PMP70 in importing VLCFAs into the peroxisomal matrix [50], this downregulation likely contributes to compromised peroxisomal β-oxidation capacity. This incomplete maturation is further supported by the downregulation of acyl-CoA oxidase 1 (ACOX1; the rate-limiting enzyme of peroxisomal β-oxidation [51]), and ACBD5 (a tethering and lipid-binding protein involved in fatty acid metabolism and import, and organelle contact sites [52]) (Fig. 3J–L). These molecular features suggest that, although peroxisomal proliferation is activated via increased de novo biogenesis (Pex3) and fission (Pex11β), the functional competence of these organelles in NDUFS4-KO cells appear compromised. These results suggest that peroxisomal β-oxidation is impaired in NDUFS4-KO cells. The dichotomy between increased peroxisomal biogenesis and impaired function suggests a maladaptive response, in which the expansion of peroxisome number fails to restore FA catabolic capacity. Acute inhibition of peroxisomal β-oxidation and lipid overload disrupts lipid homeostasis and mitochondrial integrity The previous results showing an increase in peroxisome biogenesis in NDUFS4-KO cells imply a regulatory circuit between mitochondria and peroxisomes, while the decrease in peroxisomal β-oxidation in those cells suggests that the regulatory circuit directly affects lipid metabolism. Therefore, we investigated the nature of this regulatory circuit, particularly if peroxisomal perturbations can also impact mitochondria, and how this circuit contributes to the lipid accumulation we observed in the NDUFS4-KO cells. Since peroxisomes work together with mitochondria to break down lipids, especially by shortening VLCFA, we first tested how acute disruption of peroxisomal β-oxidation affects cellular lipid balance and mitochondrial function, both under normal conditions and lipid overload. We acutely inhibited peroxisomal β-oxidation with enoximone (300 μM, 1h) (Fig. 4A), an inhibitor of peroxisomal β-oxidation [[53], [54], [55]], without significantly affecting the levels of ACOX1 (Suppl. Fig. 8A). Peroxisomal β-oxidation inhibition by Enoximone increases peroxisome number without affecting mitochondrial function in NDUSF4-KO MEF cells. Mouse embryonic fibroblasts (MEF) cells were used 48 h post-plating for experiments, with or without treatment with DMSO or enoximone (300 μM; 1 h). (A) Schematic model illustrating Enoximone's mechanism of action: its classical role as a PDE inhibitor and its role in suppression of peroxisomal β-oxidation, presently not fully elucidated. (B) Representative images peroxisomes stained with PMP70 (red) and Hoechst (nuclei; blue) in WT and NDUFS4-KO MEFs. Scale bar = 10 μm. (C) Quantification of PMP70-positive peroxisome number per cell and average area (μm2) from images in (B), using FIJI software for analysis. [n1 = 15-21, n2 = 11-14, n3 = 14; N = 3]. (D) Mitochondrial membrane potential (Δψm) was evaluated using JC-1 (200 μM) fluorescent probe, and basal fluorescence was recorded for 2 min and then CCCP (2 μM) was added to completely depolarize mitochondria. Basal Δψm was quantified [n1 = 5, n2 = 5, n3 = 5, N = 3]. (E) Representative images of lipid droplets stained with BODIPY 493/503 (green) and nuclei (Hoechst; blue) in WT and NDUFS4-KO MEFs. Scale bar = 10 μm. (F) Quantification of lipid droplet number per cell and average lipid droplet area (μm2) from images in (H), using FIJI software for analysis. [n1 = 14-19, n2 = 10-18, n3 = 16-18; N = 3]. Data are presented as mean ± S.E.M.; n, number of 96-well evaluated (D) or number of cells (C, F); N, independent experiments; two-way ANOVA followed by two-tailed Tukey's multiple comparison test with significance from p < 0.05 and exact p-values indicated (C, D, F). n.s. means nonsignificant. First, we tested how inhibition of peroxisomal β-oxidation with enoximone affects peroxisome abundance. As shown above, NDUFS4-KO cells displayed a higher basal number of peroxisomes compared to WT (Fig. 4B–C). Importantly, acute inhibition of peroxisomal β-oxidation did not alter peroxisome number in WT cells but led to a further significant increase in peroxisome number in NDUFS4-KO cells (Fig. 4B–C). This suggests that control cells do not readily respond to peroxisomal β-oxidation inhibition, but when mitochondrial function is already compromised, cells promote peroxisomal proliferation in response to impairment of peroxisomal β-oxidation. Because 1h is not sufficient for de novo peroxisomal biogenesis, it is likely that this effect relies on peroxisomal fission, which is in agreement with the increased levels of Pex11β in NDUFS4-KO. We then tested if this acute inhibition of peroxisomal β-oxidation had impacts on mitochondrial aerobic metabolism. No significant effect was observed on mitochondrial oxygen consumption rate (Suppl. Fig. 8B and C), on mitochondrial membrane potential (Fig. 4D), or on mitochondrial morphology (Suppl. Fig. 8D and E). As the acute inhibition of peroxisomal β-oxidation is sufficient to impact peroxisome number in NDUFS4-KO cells, we sought to test if it could also impact LD homeostasis. We found that enoximone treatment did not change the number of LDs in either genotype; however, it significantly increased LD area in WT, while having no effect in NDUFS4-KO cells (Fig. 4E–F). This suggests that when peroxisomal function is impaired in a background with functional mitochondria, cells may shift excess fatty acids toward LD expansion. In contrast, NDUFS4-KO cells which already have higher LD mass are unable to adequately intake further lipids upon peroxisomal impairments. These data suggest that the inability of peroxisomes to adequately process FAs contributes to their diversion towards neutral lipid storage and LD expansion. Fatty acid substrate overload reveals remodelling of mitochondrial bioenergetics and lipid homeostasis in a respiratory chain-dependent manner Given the rapid response of cellular lipid homeostasis to inhibition of peroxisomal β-oxidation in the presence of multiple sources of carbon (e.g., FAs, glucose), we sought to test what happens if the cells rely only on a carbon substrate that requires peroxisomal function, such as a very long-chain FA. To test this, we added a very long-chain C24:0 FA, lignoceric acid (50 μM, 2h; henceforth LA), to cellular medium lacking glucose (Fig. 5A). We observed that when LA is added to the medium in WT cells, there is no effect on peroxisomal number (Fig. 5B–C). However, when LA is added to WT cells with the inhibitor of peroxisomal β-oxidation, there is an increase in the number of peroxisomes (Fig. 5B–C). In NDUFS4-KO cells, we observed more peroxisomes, in agreement with previous results, and this number increased further when cells were treated with LA, regardless of enoximone treatment (Fig. 5B–C). This result suggests that cells can handle excess LA when peroxisomal β-oxidation is working, but if peroxisomal function is impaired they trigger a retrograde response leading to increased peroxisomal numbers. Furthermore, this pattern is mimicked in NDUFS4-KO cells, which already have lower peroxisomal β-oxidation capacity: when these cells are fed LA, they immediately respond by increasing peroxisomal numbers. Notably, when peroxisomal β-oxidation is inhibited and LA is added to WT cells, there is a decrease on mitochondrial membrane potential (Fig. 5D), which can be attributed to the excess of LA, since enoximone treatment alone did not change mitochondrial membrane potential, as shown above (Fig. 4D). To test if the effect on mitochondrial membrane potential is really due to excess LA, we tested several LA concentrations and observed that most of them decreased mitochondrial membrane potential, but the magnitude of the effect was proportional to the concentration of LA (Suppl. Fig. 9A). Lipid overload by VLCFA exposure and peroxisomal β-oxidation inhibition synergistically increase pperoxisomal biogenesis and disrupt mitochondrial morphology in NDUFS4-KO cells. Mouse embryonic fibroblasts (MEF) cells were used 48 h post-plating for experiments, with or without treatment with BSA or lignoceric acid (LA; 50 μM; 2 h) or LA + Enoximone (50 μM of LA 2h and 300 μM of Enoximone 1h). (A) Schematic representation of the experimental timeline, where WT and NDUFS4-KO MEF cells were treated with 50 μM lignoceric acid (LA) for 2 h and 300 μM Enoximone for 1 h before fixation for immunocytochemistry or functional assays. (B) Representative images of peroxisomes stained with PMP70 (red) and Hoechst (nuclei; blue) in WT and NDUFS4-KO MEFs. Scale bar = 10 μm. (C) Quantification of PMP70-positive peroxisome number per cell and average area (μm2) from images in (B), using FIJI software for analysis. [n1 = 18-22, n2 = 17-22, n3 = 13-17; N = 3]. (D) Mitochondrial membrane potential (Δψm) was evaluated using JC-1 (200 μM) fluorescent probe, and basal fluorescence was recorded for 2 min and then CCCP (2 μM) was added to completely depolarize mitochondria. Basal Δψm was quantified [n1 = 5, n2 = 5, n3 = 5, n4 = 5, n5 = 5; N = 5]. (E) Representative images of mitochondria network (HSP60) and nuclei (Hoechst) in WT and NDUFS4-KO MEFs. Scale bar = 10 μm. (F) Mitochondrial morphology analysis was performed using FIJI software and the plugin Mitochondria Analyzer, detailing morphological parameters (number, perimeter, area, aspect ratio, form factor) and branch parameters (total branch length) [n1 = 18-20, n2 = 16-19, n3 = 11-16; N = 3]. (G) Representative images of lipid droplets stained with BODIPY 493/503 (green) and nuclei (Hoechst; blue) in WT and NDUFS4-KO MEFs. Scale bar = 10 μm. (F) Quantification of lipid droplet number per cell and average lipid droplet area (μm2) from images in (G), using FIJI software for analysis. [n1 = 18-23, n2 = 14-17, n3 = 12-15; N = 3]. Data are presented as mean ± S.E.M.; n, number of 96-well evaluated (D) or number of cells (C, F, H); N, independent experiments; two-way ANOVA followed by two-tailed Tukey's multiple comparison test with significance from p < 0.05 and exact p-values indicated (C, D, F, H). n.s. means nonsignificant. In parallel, we assessed how cells adapt bioenergetically to VLCFA challenge under glucose-free assay conditions that minimize glycolytic contribution while promoting reliance of fatty acids as carbon sources for mitochondrial metabolism (Suppl. Fig. 9B–C). In WT cells, LA + enoximone treatment significantly decreased basal respiration, maximal respiration, and ATP-linked respiration, indicating that inhibition of peroxisomal β-oxidation aggravates LA-induced mitochondrial dysfunction. Additionally, non-mitochondrial respiration was significantly increased in WT cells treated with LA, suggesting enhanced oxidative stress or activation of oxygen-consuming processes outside the electron transport chain (Suppl. Fig. 9C). In contrast, NDUFS4-KO cells, which already display reduced basal respiration compared to WT, showed no further decrease in basal, maximal, or ATP-linked respiration upon LA or LA + enoximone treatment. OCR parameters remained similarly reduced across all KO conditions (Suppl. Fig. 9B–C). These findings indicate that mitochondrial respiration in NDUFS4-deficient cells is already severely compromised and cannot be further suppressed by VLCFA overload. To distinguish peroxisome-dependent effects from those related to direct mitochondrial FAO, we next examined the effects of oleic acid (20-200 μM, 2h or 4h; henceforth OA), a long-chain fatty acid predominantly metabolized through mitochondrial β-oxidation (Suppl. Fig. 10). In contrast to LA, OA required prolonged exposure (4h) and multiple concentrations to induce a decrease in mitochondrial membrane potential, and this effect was observed exclusively in WT cells (Suppl. Fig. 10A–B). NDUFS4-KO cells displayed reduced basal membrane potential but did not show further changes upon OA treatment at any concentration tested. Interestingly, OA overload showed a bioenergetic profile opposite to that observed with LA (Suppl. Fig. 10C–D). In WT cells, OA and OA + enoximone significantly increased basal respiration compared to WT basal conditions (Suppl. Fig. 10C–D). This was accompanied by increased proton leak and elevated non-mitochondrial respiration. However, maximal respiration and ATP-linked respiration were not significantly altered in WT cells. In NDUFS4-KO cells, basal respiration remained lower than WT under all conditions, where OA did not further modify basal or maximal respiration (Suppl. Fig. 10C–D). Unexpectedly, ATP-linked respiration was significantly increased in KO cells treated with OA or OA + enoximone compared to the NDUFS4-KO control condition (Suppl. Fig. 10C–D). This suggests that, despite complex I deficiency, mitochondrial oxidation of OA can partially enhance ATP-coupled respiration in NDUFS4-deficient cells. Collectively, these data reveal a striking substrate-specific response. VLCFA overload by LA, which requires peroxisomal shortening prior to mitochondrial oxidation, impairs mitochondrial respiration in WT cells and exacerbates dysfunction when peroxisomal β-oxidation is inhibited, while having no additional impact on already compromised KO mitochondria. In contrast, OA, a fatty acid oxidized by mitochondria in a peroxisome-independent manner, stimulates basal respiration in WT cells and enhances ATP-linked respiration in NDUFS4-KO cells without further depressing mitochondrial function. This result further supports the existence of coordination between peroxisomal β-oxidation and mitochondrial aerobic metabolism. To test if the mitochondrial network dynamics were affected, we stained the cells for Hsp60, a mitochondrial matrix protein (Fig. 5E) and observed that while WT mitochondrial morphology does not seem to be affected by the conjugated incubation of LA + enoximone, mitochondria in NDUFS4-KO show fragmentation under these conditions. We hypothesized that this may reflect a broader disturbance in lipid handling and metabolic stress. Therefore, we next examined whether LD homeostasis was also affected under these conditions (Fig. 5G). While LA or LA + enoximone treatment did not alter the number of LDs in either genotype (Fig. 5H), LD area increased significantly in WT cells treated with LA, but not in NDUFS4-KO cells (Fig. 5H). This suggests that, in WT cells, LA is at least partially metabolized via peroxisomal β-oxidation into shorter FA species that may be subsequently stored in LDs, thereby increasing their size. In contrast, NDUFS4-KO cells, which already display compromised peroxisomal β-oxidation, may be unable to efficiently process LA or to further increase LD mass. Interestingly, LD area also increased in WT cells treated with enoximone alone, in the absence of exogenous LA (Fig. 4F). One possible explanation is that acute inhibition of peroxisomal β-oxidation causes a transient accumulation of endogenous VLCFAs or other lipid species, which may be rerouted toward neutral lipid synthesis and LD expansion as a buffering mechanism. As demonstrated in Fig. 1, Fig. 2, LD area is already increased in KO cells, due to the increase in TAGs. This might also be causing an inability to deal with further overload of lipids. Therefore, this compensatory increase in LD size may not occur in NDUFS4-KO cells, which already operate under chronic peroxisomal dysfunction and may have adapted to rewire their lipid storage and degradation pathways. To determine whether OA overload also alters lipid storage or peroxisomal adaptation in a similar manner to LA, we next examined the homeostasis of LDs (Suppl. Fig. 11A–B) and peroxisomes (Suppl. Fig. 11C–D). LD area was significantly increased in WT cells upon OA treatment compared to WT control. In contrast, NDUFS4-KO cells already displayed elevated basal LD area relative to WT control, consistent with previous results, and OA treatment did not further increase LD size in KO cells. These findings further corroborate the phenotype described in WT cells but, most importantly that there is no further response in KO cells. Importantly, OA treatment did not alter peroxisomal number (Suppl. Fig. 11C–D). As previously shown, NDUFS4-KO cells display a higher basal number of peroxisomes compared to WT controls; however, OA exposure did not further modify peroxisomal abundance in WT or NDUFS4-KO cells. This contrasts with the response to LA, which requires peroxisomal shortening and triggers peroxisomal proliferation when β-oxidation is impaired. Thus, OA-induced metabolic remodelling occurs independently of changes in peroxisomal abundance, reinforcing that OA is primarily handled by mitochondrial β-oxidation rather than peroxisomal processing and emphasizing the important role of peroxisomes in mitochondrial handling. These data indicate that the acute inhibition of peroxisomal β-oxidation affects LD dynamics and mitochondrial organization in a genotype-dependent manner. WT cells respond to enoximone and lipid overload by increasing LD area while maintaining mitochondrial integrity. In contrast, NDUFS4-KO cells do not adapt to these conditions; they do not adapt LD morphology to accommodate excess lipids and present mitochondrial fragmentation. These findings suggest that peroxisomal dysfunction, when combined with mitochondrial impairment or lipid overload, disrupts lipid organization and organelle homeostasis. Discussion Mitochondrial function and lipid metabolism are closely integrated [24]. Mitochondrial and peroxisomal metabolism are tightly coordinated to maintain lipid homeostasis, yet how peroxisomes adapt to pre-existing mitochondrial dysfunction remains unclear, with most of the published work focusing on the reciprocal problem (peroxisomal loss causing secondary respiratory-chain defects) [[56], [57], [58], [59]]. Here, we show that loss of the complex I subunit NDUFS4 in cells triggers a selective remodelling of lipid metabolism characterized by TAG accumulation, LD expansion, and altered CE profiles. This phenotype emerges despite suppression of de novo lipid synthesis, indicating that lipid accumulation is driven primarily by defective catabolism rather than anabolic activation, in a context of aerobic dysfunction. Our lipidomic data revealed significant enrichment in TAGs, ether-linked phospholipids, and CE species in NDUFS4-KO cells. Ether-linked phospholipids, including ether-linked phosphatidylcholine (PC O-) and ether-linked phosphatidylethanolamine (PE O-), are lipids synthesized through a pathway involving both peroxisomal and endoplasmic reticulum (ER) steps, with ACBD5-VAPB tethering facilitating lipid transfer between these organelles [60,61]. In NDUFS4-KO cells, we detected a significant decrease in PC O- species and a trend toward increased PE O-, coinciding with reduced ACBD5 expression. This suggests that peroxisomal-ER communication and ether lipid trafficking may be perturbed in complex I deficiency, also contributing to the altered lipid profile [62]. Also, a recent study shows that protein kinase c (PKC) inactivates glycogen synthase kinase 3 β (GSK3β), coordinating peroxisomal-ER communication [63]; however, we previously found that GSK3β signalling is increased in these cells [19], contributing to the idea of peroxisomal-ER impaired interaction. Moreover, since ether lipid synthesis starts in peroxisomes, impaired peroxisomal β-oxidation and handling of VLCFA could further affect the availability of precursors, worsening the consequences of defective tethering. As these ether lipids are known to be involved in membrane architecture, organelle dynamics, and oxidative stress protection [64], further studies are needed to dissect how mitochondrial-peroxisomal dysfunction reshapes these lipid pools. TAG accumulation likely represents a protective sequestration mechanism against excess FFAs as already observed in several other models [38,[65], [66], [67], [68]], as both mitochondrial and peroxisomal β-oxidation pathways are compromised. Here, excess FFAs cannot be efficiently degraded and are instead esterified into neutral TAGs for storage in LDs. This re-direction limits the negative effects of unesterified FAs, thereby buffering the cell against lipotoxic stress. Interestingly, the observed changes occurred independently of increased lipid synthesis since we showed decreased levels of ACLY and mature SREBF1 [19]. Instead, impaired mitochondrial FA β-oxidation was associated with reduced expression of key import and oxidation components, including CPT-I, CPT-II, and ACSL1, indicating defective lipid catabolism as the primary driver of TAG accumulation. Strikingly, complex I impairment is met by robust upregulation of peroxisomal biogenesis markers (e.g., PEX3, PEX11β, and PEX5) and increased peroxisome number. This proliferative response mirrors compensatory peroxisomal induction observed in various mitochondrial oxidative stress disorders and yeast respiratory mutants, suggesting a conserved adaptation between the two organelles [[69], [70], [71], [72]]. However, this compensation is incomplete. Indeed, NDUFS4-KO cells exhibit deficient peroxisomal maturation and function, corroborated by decreased levels of PEX19, ACOX1, PMP70, and ACBD5. These alterations are also seen in peroxisomal biogenesis disorders (PBDs), where peroxisomes are structurally intact but fail to properly import or assemble β-oxidation machinery [[73], [74], [75], [76], [77]]. Therefore, peroxisomal induction in NDUFS4-KO cells represents a structural adaptation without functional gain, being a maladaptive attempt to buffer metabolic stress. Acute inhibition of peroxisomal β-oxidation using enoximone further show the importance of the metabolic communication between peroxisomes and mitochondria. Enoximone increased peroxisome number in KO but not WT cells, a response likely mediated by fission rather than de novo biogenesis. In WT cells, this inhibition increased LD area and tendentially modulated mitochondrial respiration, reflecting metabolic flexibility. In contrast, NDUFS4-KO cells failed to exhibit further changes in LDs or respiratory adaptation, suggesting a rigid and saturated metabolic state. This metabolic inflexibility was particularly evident when cells were overload with LA, a VLCFA that requires peroxisomal processing [78,79]. In WT cells, LA promoted LD expansion, an effect abolished when peroxisomal β-oxidation was blocked, showing that intact peroxisomal processing is essential for lipid storage. In contrast, in NDUFS4-KO cells, where peroxisomal function is already impaired, LA failed to induce LD expansion and impaired mitochondrial morphology when conjugated with enoximone. This suggests that VLCFAs are incompletely metabolized and not effectively sequestered, exacerbating organelle stress and further confirming the necessity for peroxisomal correct function. Importantly, the differential response to OA further reinforces the concept that mitochondrial respiratory competence determines how lipid substrates are handled. In WT cells, OA overload promoted LD expansion together with increased basal respiration, consistent with an adaptive mechanism to buffer excess fatty acids while maintaining respiratory capacity. In contrast, NDUFS4-KO cells, which already exhibit impaired ETC activity and reduced basal respiration, failed to further increase LD area despite showing selective enhancement of ATP-linked respiration. When considered alongside the LA experiments, these findings indicate that substrate-specific lipid overload reveals distinct bioenergetic remodelling depending on mitochondrial functionality. WT cells retain the ability to dynamically coordinate fatty acid oxidation, LD expansion, and respiratory adaptation under both mitochondrial and peroxisome-dependent challenges. Conversely, KO cells, with chronically compromised complex I activity, display restricted plasticity in both lipid storage and respiratory remodelling. Thus, the capacity to modulate LD expansion in response to fatty acid availability appears tightly coupled to functional mitochondrial respiration. Moreover, several data also pin-points the importance of correct peroxisomal function to correct lipid metabolism [[80], [81], [82], [83]]. These results suggest a peroxisome-LD-mitochondria buffering axis that is compromised in complex I deficiency, leading to an inability to safely re-direct excess lipids. The inability of NDUFS4-KO cells to adapt to lipid overload supports a model in which peroxisomal and mitochondrial β-oxidation pathways form an integrated metabolic axis, with LDs serving as dynamic buffers to prevent lipotoxicity, contributing to a peroxisome-LD-mitochondria axis. This axis is required for lipid buffering under metabolic stress. In the absence of functional FAO, excess FFAs accumulate and are diverted toward storage in LDs; however, in metabolically saturated cells, such as NDUFS4-KO, this storage mechanism becomes insufficient. The resulting imbalance between FA supply, degradation, and storage leads to a feedforward loop of lipid accumulation, organelle dysfunction, and impaired homeostasis. Notably, the cells respond differently to overload of fatty acids that require peroxisomal beta-oxidation versus those that only require mitochondrial beta-oxidation, suggesting that sensing the size of the carbon chains of the overloading lipids is part of the mechanism that triggers the cellular homeostatic adaptation to such overload. Although our experiments functionally assess mitochondrial respiratory adaptation to FA substrates of different chain lengths, they do not directly measure β-oxidation flux. Future studies focusing on direct measurements of FAO activity will be important to further dissect the relative contributions of mitochondrial and peroxisomal FAO pathways to the metabolic adaptations observed here. Importantly, although these findings were obtained in NDUFS4-deficient cells, mitochondrial and peroxisomal β-oxidation pathways, lipid metabolism, and NAD+/redox homeostasis are highly conserved across cell types. The NDUFS4-KO model reproduces a canonical complex I defect typical of mitochondrial diseases, supporting the relevance of our observations beyond this specific cell model. Our data are in line with recent in vivo studies showing altered lipid metabolism in NDUFS4-KO mice [7,9], and identify peroxisomal inefficiency as a central contributor to the observed lipid dysregulation. Interestingly, such studies identified different lipid remodelling patterns in throughout brain areas and different organs, suggesting tissue-specific adaptation. Moreover, our results are in line with in vivo evidence showing that peroxisomal dysfunction alone is sufficient to drive lipid imbalance and mitochondrial abnormalities in ACOX1-deficient models [84,85], resembling what we observe in KO cells exposed to enoximone. Our findings establish lipid dyshomeostasis as a direct consequence of impaired FA oxidation arising from complex I deficiency and uncover a maladaptive organelle crosstalk that have significant implications for diseases where both mitochondrial and peroxisomal dysfunctions are present, such as Leigh syndrome, depicting a new mitochondria-peroxisome crosstalk mechanism that coordinate lipid homeostasis. Targeting peroxisomal maturation/function or restoring inter-organelle lipid trafficking could represent therapeutic strategies to alleviate lipid stress in mitochondrial disorders. Conclusion Complex I deficiency triggers a maladaptive peroxisomal response characterized by organelle proliferation but impaired β-oxidation capacity. This metabolic dysfunction disrupts mitochondrial-peroxisome-LD coordination, leading to impaired FA handling, TAG accumulation, and susceptibility to lipid stress. Our findings identify peroxisomal inefficiency as a key amplifier of metabolic imbalance in respiratory-chain defects and highlight peroxisomal maturation and inter-organelle lipid trafficking as potential therapeutic targets in mitochondrial disease. Methods For detailed methodology, please see “Detailed Methods” Section in Supplementary Materials. Cell culture and treatments Mouse embryonic fibroblasts (MEFs) from WT and NDUFS4-KO cells were immortalized using E6E7 viral particles as previously described [86]. Cells were cultured in Dulbecco's modified Eagle's medium (DMEM, Gibco) supplemented with 10% v/v fetal bovine serum (FBS) and 1% Penicillin/Streptomycin (Corning). Cell lines were maintained at 37 °C in a 5% CO2 controlled atmosphere. Trypsin (Gibco) was employed to detach cells and generate subcultures. When at 80% of confluency, cells were treated with Enoximone (300 μM, 1h; Tocris Bioscience #1693), with SB204049 (30 μM, 24h; Tocris Bioscience #4962), with lignoceric acid (20-150 μM, 2h; Sigma #L6641) and with Oleic Acid (20-150 μM, 2h and 4h; Sigma #O1283). Cells were used 48 h after plating. Transfection of immortalized cell lines MEF cells were transfected 24 h after plating, at 70% confluence, with 0.75 μg DNA/cm2 of growth area in opti-MEM without FBS or antibiotics, following the Lipofectamine 3000 (ThermoFisher Scientific) manufacturer instructions. Medium was changed 4 h after transfection and cells were cultured for 48 h. In the transfection procedure, GFP-SKL plasmid [47] was used. Cell lysis and immunoblotting Cells were harvested in RIPA lysis buffer supplemented with protease and phosphatase inhibitor cocktail (Thermo Scientific). Lysates were incubated on ice for 30 min and cleared by centrifugation at 14000×g for 10 min at 4 °C. Supernatants were collected and stored at −20 C° until used. Protein concentration in the samples was evaluated using the Pierce BCA Protein Assay Kit (Thermo Scientific) following the manufacturer's instructions. 40 μg of protein lysates were boiled for 5 min at 95 C° and loaded in 10% Tris-glycine polyacrylamide gels in SDS/Tris-glycine running buffer. PageRuler Plus pre-stained protein ladder (Thermo Scientific) was used for protein size estimation. The resolved proteins were transferred onto nitrocellulose membranes (GE Healthcare) through a Tetra Blot transfer system (Bio-Rad). Membranes were subsequently blocked in Tris-buffered saline plus 0.1% Tween (TBS-T) plus 5% bovine serum albumin (BSA) for 1 h at room temperature (RT) and then incubated over-night at 4 °C with primary antibodies diluted in TBS-T plus 5% BSA. Membranes were then washed in TBS-T (3 × 10 min) at RT and subsequently incubated for 1h at RT with horseradish peroxidase (HRP)-conjugated secondary antibodies. Membranes were then washed in TBS-T (3 × 10 min) at RT and rinsed in TBS-T. Immunoreactive proteins were visualized using Immobilon® Forte Western HRP Substrate (Merck Millipore) at the ImageQuant 800 (Amersham, Cytiva). Densitometric analysis was carried out using the Image Lab 6.0.1 software (Bio-Rad). The following antibodies were used for western blotting: rabbit ACSL1 (Proteintech, #13989-1-AP), mouse CPT-I (Santa Cruz, #SC-393070), mouse CPT-II (Santa Cruz, #SC-377294), rabbit ACSS2 (GeneTex, #GTX30020), rabbit FASN (CST, #3180P), rabbit ACLY (CST, #4332S), rabbit pSer455_ACLY (CST, #4331S), rabbit Pex3 (Thermo Fisher, #PA5-115740), rabbit Pex19 (Thermo Fisher, #PA5-22129), rabbit Pex11b (Assay Biotech, #C17630), rabbit Pex14 (Millipore, #ABC142), rabbit Pex5 (SigmaPrestige, #HPA039259), mouse PMP70 (Merck, #SAB4200181), rabbit ACBD5 (Sigma, #HPA011861), rabbit ACOX1 (Abcam, #ab59964), mouse Vinculin (Millipore, #MAB3574), mouse GAPDH (Millipore, #MAB374), rabbit Calnexin (Enzo, #ADI-SPA-860-D). Immunocytochemistry and confocal imaging Cells were cultured onto 10 mm glass coverslips (Thermo Fisher Scientific) coated with Poly-l-Lysine (Sigma-Aldrich). After three washes with PBS (in mM: 137 NaCl, 2.7 KCl, 1.8 KH2PO4, 10 Na2HPO4.2H2O, pH 7.4), cells were fixed with 4% paraformaldehyde (pre-warmed at 37 °C) for 15 min and permeabilized in 0.1% Triton X-100 in PBS, for 10 min, at RT. Then, cells were blocked for 1 h at room temperature with 3% (w/v) BSA in PBS and further incubated with primary antibody overnight at 4 °C. Subsequently, cells were then washed with PBS and incubated with the adequate secondary antibody (Alexa Fluor-568 donkey anti-mouse) at 1:200 in blocking solution, for 1 h at room temperature. Nuclei were stained with 1 μg/mL Hoechst 33342 in PBS (Invitrogen, CA, USA) for 10 min and coverslips were mounted using Mowiol 40-88 (Sigma Chemical, St. Louis, MO, USA). Confocal images were obtained using a Plan-Apochromat/1.4NA 63x lens on an Axio Observer.Z1 confocal microscope (Zeiss Microscopy, Germany) with Zeiss LSM 710 software. The following primary antibodies were used: mouse HSP60 (1:250; BD Biosciences, #611563), mouse PMP70 (1:2000; Merck, #SAB4200181), LAMP1 (1:100; DSHB; #1D4B), rabbit Pex14 (1:100; Millipore, #ABC142). Statistical analyses Data were analyzed using Excel (Microsoft, Seattle, WA, USA), GraphPad Prism 10 (GraphPad Software, San Diego, CA, USA), and Python software (Python Software Foundation, USA), and are expressed as the mean ± S.E.M. of the number of independent experiments indicated in figure legends. Comparisons among two groups were performed using the two-tailed unpaired t-test followed by Welch's correction, when appropriate. Comparisons among two nonparametric groups were performed using two-tailed Mann-Whitney testing. One-way ANOVA was used for comparisons among multiple groups under a single experimental condition. Two-way ANOVA was applied when assessing the effect of two independent variables and their interaction, followed by Tukey's multiple comparison test as post hoc analysis. The statistical test used is detailed on figure legends. Statistical significance was defined as p < 0.05. Funding This project has received funding from the European Union’s Horizon2020 research and innovation programme under grant agreement No 857524. FCT – Fundação para a Ciência e Tecnologia grant 2022.11614.BD (P.C.), ERC-Portugal, 2022.09311.PTDC. FCT – Fundação para a Ciência e Tecnologia grant UID/04539/2025, LA/P/0058/2020 [DOI: 10.54499/LA/P/0058/2020], UID/PRR/4539/2025 [DOI: 10.54499/UID/PRR/04539/2025] (M.E.F.) CRediT authorship contribution statement Patrícia Coelho: Data curation, Formal analysis, Investigation, Methodology, Visualization, Writing – original draft, Writing – review & editing. Pedro Dionísio: Data curation, Formal analysis, Investigation, Supervision, Writing – review & editing. Vilma Sardão Oliveira: Formal analysis, Investigation, Supervision, Writing – review & editing. Roberto A. Dias: Formal analysis, Investigation, Writing – review & editing. Matthias E. Futschik: Formal analysis, Visualization, Writing – review & editing. Robin Klemm: Formal analysis, Methodology, Supervision, Writing – review & editing. Ira Milosevic: Formal analysis, Methodology, Supervision, Writing – review & editing. Nuno Raimundo: Conceptualization, Formal analysis, Investigation, Methodology, Resources, Supervision, Visualization, Writing – original draft, Writing – review & editing. Declaration of competing interest The authors have no conflict of interest to declare. Acknowledgments We would like to thank Dr. Paulo Oliveira (CiBB - Centre for Innovative Biomedicine and Biotechnology, University of Coimbra, Portugal) for providing access to the Seahorse apparatus and for his support during the experiments conducted in his laboratory. We thank Dr. Rita Gaspar and MSc. Joana Pires for technical assistance. Footnotes Supplementary data to this article can be found online at https://doi.org/10.1016/j.redox.2026.104166. Supplementary data The following is the Supplementary data to this article: Multimedia component 1 Data availability Data will be made available on request. References Associated Data Supplementary Materials Multimedia component 1 Data Availability Statement Data will be made available on request.