Clarifying the molecular mechanisms underlying the lethal in vitro cellular effects and mild organismal phenotype of SOD1 loss

Abstract

SOD1 is the superoxide dismutase localized principally to the cytoplasm. It is found in all cell types, is extremely abundant, and highly efficient at converting superoxide (O₂^·−) to hydrogen peroxide (H₂O₂). SOD1 and its homologues in the mitochondrial matrix (SOD2) and extracellular environment (SOD3) are the only enzymes that act directly on superoxide. Surprisingly, Sod1^−/− knockout mice exhibit only a mild phenotype. In contrast, Sod1^−/− cells cannot survive and multiply in vitro. One possibility for the mild in vivo phenotype is that a subtle genome-wide mechanism of developmental compensation attenuates the phenotype. Conversely, many different mechanisms have been proposed to explain the lethality in vitro and the difficulties of rescuing it with antioxidants. Here, we aim to determine the cause of the severe in vitro phenotype of Sod1^−/− cells and whether the mild in vivo phenotype of Sod1^−/− mice reflects developmental compensation. We provide evidence that the lethality in vitro is directly due to superoxide toxicity and not to more indirect mechanisms. Furthermore, we created a floxed Sod1 gene to compare acute adult-onset organism-wide removal of the protein with constitutive absence in mice and derived cells. We observed no phenotypic differences between acute and chronic loss of SOD1. We discuss how these observations can be reconciled with the evolutionary conservation of high levels of expression of a broadly expressed, high catalytic rate enzyme.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1038/s41598-026-59950-5.

Introduction

Reactive oxygen species (ROS) are inevitable by-products of aerobic metabolism, arising from various cellular redox reactions. ROS can oxidize biological molecules, and excessive ROS can potentially threaten the structural and functional integrity of the cell. On the other hand, by acting as signalling molecules, ROS are involved in the regulation of various biological activities and stress adaptation responses. Among endogenous antioxidant defence mechanisms that control and regulate cellular ROS levels by neutralizing them, superoxide dismutases (SODs) are considered one of the most crucial front-line defences. They specifically scavenge superoxide anions (O₂^·−), which are the source of most other ROS, catalyzing their disproportionation into molecular oxygen and hydrogen peroxide (H₂O₂) in a highly efficient manner^1,2. In mammals, there are three isoforms of SOD that differ by their binding metal cofactors and subcellular localization. SOD1 (Cu/ZnSOD) is the predominant SOD isoform, present largely in the cytosol but also found in the mitochondrial intermembrane space (IMS) as well as in the nucleus under certain conditions^3–7. In contrast, SOD2 (MnSOD) is located within the mitochondrial matrix, and SOD3 (EC-SOD)—also a Cu- and Zn-dependent SOD—is a secretory extracellular enzyme present on cell surfaces and within the extracellular matrix^8,9.

SOD1 is estimated to comprise 1–2% of total cellular proteins in many mammalian cell types¹⁰, and its cytosolic concentration is estimated to be 4–40 µM¹¹. This high concentration of SOD1 is thought to be necessary for maintaining controlled intracellular superoxide levels. Excessive superoxide can damage cellular components, most notably by rapidly oxidizing iron–sulphur ([Fe–S]) clusters in proteins¹². Diverse cellular mechanisms can produce superoxide across various sites, but superoxide is considered to be diffusion-restricted and to act only locally¹³. Therefore, maintaining a high concentration of SOD1 might be necessary to ensure the immediate clearance of excessive superoxide before it reacts with vulnerable targets. On the other hand, SOD1 plays a critical role in redox signalling by generating hydrogen peroxide^2,14. For example, SOD1-derived hydrogen peroxide has been implicated in the oxidative inactivation of GAPDH and protein tyrosine phosphatases^15,16. Furthermore, several other functions of SOD1 beyond superoxide dismutation have been discovered, including involvement in copper and zinc metabolism, acting as a nuclear transcription factor, regulating RNA metabolism, modulating metabolism, and contributing to sulfide detoxification^17–21. The high cellular abundance of SOD1 may be particularly important for the optimal performance of some of these functions. This possibility is supported by work in yeast showing that redox control of casein kinase stability requires higher SOD1 levels than are needed for protection against oxygen toxicity²². However, whether a similar dose dependency exists in mammalian systems has yet to be established.

Surprisingly, Sod1^−/− knockout mice completely lacking SOD1 from the start of life exhibit far milder phenotypes than might be expected for the loss of a constitutively expressed and highly abundant frontline enzymatic antioxidant. In fact, Sod1^−/− mice are viable, fertile, and develop normally without overt abnormalities, though elevated oxidative stress, accelerated muscle wasting, increased susceptibility to hepatocarcinogenesis, and reduced lifespan have been observed in these mutants, among other changes^14,23–33. The viable and fertile phenotype of Sod1^−/− mice has been suggested to indicate induction of compensatory mechanisms during early development that help mitigate the loss of what would appear to be a critical antioxidant^25,34–38. In contrast to the mild whole-organism phenotype, embryonic and dermal fibroblasts isolated from Sod1-null mice were reported to fail to survive and grow in vitro^39,40.

In this study, we aimed to address two related questions raised by the discrepancy between the mild phenotype of Sod1^−/− mice in vivo and the severe growth defect of Sod1^−/− cells in vitro: what primarily causes the severe phenotype of Sod1 knockout (KO) cells and whether the mild phenotype of Sod1 KO mice reflects developmental adaptation or compensation in the germline mutant. We generated inducible Sod1 KO models for mice and cultured cells and demonstrated that induced Sod1 loss in cultured cells or adult mice yielded phenotypes essentially indistinguishable from those of conventional (non-inducible) models. Similar to primary mouse embryonic fibroblasts (MEFs) derived from Sod1-null mice, acute ablation of SOD1 in wild-type MEFs caused growth arrest and cell death. Whole-body SOD1 KO induced in adult mice produced no overt abnormalities, mirroring the phenotype of germline Sod1-null mice. Moreover, we show that the only antioxidant treatment that significantly rescues the viability of Sod1 KO cells is vitamin C (ascorbic acid, AscH₂; ascorbate, AscH⁻) supplementation, indicating that elevated superoxide levels per se, rather than other downstream consequences of SOD1 loss, are primarily responsible for the toxicity associated with SOD1 loss. Taken together, these findings argue against development reprogramming or compensatory mechanisms induced by loss of the key antioxidant enzyme as the explanation for the mild in vivo phenotype of SOD1-deficient mice. Instead, they support that the level of superoxide generation under laboratory conditions renders SOD1 non-essential for acute animal survival.

Results

A compound screen for the rescue of Sod1^−/− MEF viability

To begin to revisit and address the question of the source of the lethality of Sod1^−/− cells, we isolated mouse embryonic fibroblasts (MEFs) from Sod1^−/− embryos and, as consistent with previous findings^40–42, found that under standard culture conditions (21% O₂ and 5% CO₂ at 37 °C) they died shortly after isolation. To identify treatments capable of restoring the viability of Sod1 KO cells, we screened the SCREEN-WELL REDOX library (BML-2835, Enzo Life Sciences) for compounds that could rescue Sod1^−/− MEFs. The library contains 83 compounds with defined prooxidant or antioxidant activity. They were added at a concentration of 50 µM to 48-well plates pre-seeded with Sod1^−/− MEFs (> 90% confluence) and incubated for 4 days before cell viability was measured using a resazurin viability assay. Using a threshold of a 20% viability increase, we identified 5 primary hits, including ascorbyl palmitate, a lipophilic derivative of vitamin C (Table 1)⁴³.

Table 1.

Compounds from the SCREEN-WELL Redox Library identified for their ability to improve the survival of Sod1^−/−MEFs.

Compound	CAS #	Activity
Epigallocatechin gallate	989-51-5	Natural polyphenol
Ascorbyl palmitate	137-66-6	Lipophilic ascorbate
U74389G maleate	153190-29-5	Synthetic 21-aminosteroid
Terbinafine·HCl	78628-80-5	Antifungal, free radical quencher in vitro
N-Propyl gallate	121-79-9	Phenolic antioxidant

The other four compounds, besides ascorbyl palmitate, were also antioxidants known to have free-radical scavenging activity, including against superoxide^44–47. Epigallocatechin gallate (EGCG) is a natural polyphenol. Its best-known biological and pharmacological property is its scavenging activity against free radicals, including superoxide anions and hydroxyl radicals^46,48,49. U-74389G maleate is a synthetic compound belonging to a class of drugs known as 21-aminosteroids (or lazaroids). Its primary antioxidant action involves the inhibition of lipid peroxidation, and it has been demonstrated to reduce extracellular superoxide concentration in injured rat brain^44,50. Terbinafine hydrochloride (HCl) is an allylamine antifungal agent. Its antioxidant potential has been noted in some in vitro studies, including the direct quenching of ROS (e.g., superoxide) and the inhibition of their production^45,51. N-propyl gallate is a phenolic antioxidant that has been shown to act as a free radical scavenger, metal chelator, and SOD mimic^47,52. The most apparent shared property of the 5 screen hits is their superoxide-quenching antioxidant activity, suggesting that superoxide buildup is the primary determinant of Sod1^−/− MEF death in vitro.

Vitamin C promotes Sod1^−/− cell survival but not other compounds acting more indirectly on the consequences of the loss of SOD1

In a separate experiment, we tested several compounds of interest based on their known activities that could counteract the cell death induced by SOD1 loss. Those include vitamin C, coenzyme Q (a.k.a. ubiquinone), N-acetylcysteine (NAC), carbobenzoxy-valyl-alanyl-aspartyl-[O-methyl]-fluoromethylketone (Z-VAD-FMK), N(ω)-nitro-l-arginine methyl ester (l-NAME), l-arginine, diethylenetriamine/nitric oxide adduct (DETA NONOate) and 5,10,15,20-Tetrakis(4-sulfonatophenyl)porphyrinato iron(III) chloride (FeTTPS). In addition to its primary role in mitochondrial respiration, coenzyme Q (CoQ) is an essential antioxidant in lipid membranes. We used a water-soluble formulation of CoQ₁₀ (the human form of CoQ) that shows enhanced cellular uptake (CF/CoQ₁₀)⁵³. NAC enhances the cell’s capacity to neutralize hydrogen peroxide and lipid hydroperoxides, mainly by replenishing GSH⁵⁴. Z-VAD-FMK is a well-known pan-caspase inhibitor that is cell-permeable and capable of inhibiting apoptosis in many situations^55,56. l-NAME and FeTTPS were used to establish whether the reaction with nitric oxide (^·NO) plays a role in superoxide-derived toxicity in Sod1^−/− cells. Superoxide reacts with ^·NO extremely rapidly, at a nearly diffusion-limited rate. In fact, the rate constant for the reaction between superoxide and ^·NO is estimated to be 5–10 times higher than that of SOD-catalyzed superoxide clearance^11,57,58. The reaction has two potentially significant consequences. One is the generation of peroxynitrite (ONOO⁻), a potent, versatile oxidant and a powerful nitrating agent that can inflict widespread damage on protein, DNA and membrane lipids⁵⁹. Given that SOD1 deficiency elevates O₂^·− levels, the enhanced reaction of superoxide with ^·NO could lead to excessive ONOO⁻ generation, thereby triggering cellular damage and ultimately leading to cell death⁶⁰. l-NAME inhibits the production of ^·NO, and FeTTPS is a ONOO^- decomposition catalyst⁶¹. Thus, both can protect against ONOO⁻ toxicity. Furthermore, to test whether reduced ^·NO bioavailability, another consequence of O₂^·− accumulation and its subsequent reaction with ^·NO, may contribute to the inviable phenotype of Sod1^−/− MEFs, we treated the cells with the NO synthesis precursor l-arginine and the well-known ^·NO donor DETA NONOate.

Consistent with the identification of an ascorbate derivative in the Sod1^−/− MEF rescue screen, we found that vitamin C indeed can markedly improve the viability of Sod1^−/− MEF (Fig. 1A,B). However, none of the other treatments we attempted improved Sod1^−/− MEF survival (Fig. 1A,B and Supplementary Fig. 1A). Of note, the test doses for each compound, except vitamin C and CF/CoQ₁₀, were chosen as the highest nontoxic concentrations established using wild-type Sod1^+/+ MEFs (Supplementary Fig. 1B). Figure 1C shows that the rescue of the viability of Sod1^−/− MEFs by vitamin C is dose-dependent. The growth curves in Fig. 1D show that, following a one-time addition of 0.2 mM vitamin C to the growth medium, Sod1^−/− MEFs showed active proliferation during the first 3 days and then underwent growth arrest, most likely due to the depletion or instability of vitamin C, as vitamin C is known to be chemically unstable in aqueous culture media^62,63. In fact, with daily refreshing of vitamin C-supplemented medium, Sod1^−/− MEFs were able to proliferate for 4–5 weeks, but still eventually underwent growth arrest or death (Supplementary Fig. 1C). Vitamin C supplementation also slightly enhanced wild-type cell proliferation (Fig. 1D and Supplementary Fig. 1D).

Fig. 1.

Vitamin C rescues short-term viability of Sod1^−/− MEFs. (A) Viability measurement of Sod1^−/− MEFs after 5-day treatment with indicated agents added immediately post-isolation. Sod1^−/− MEFs were seeded at a density of 1.5 × 10⁵ cells (~ 90% confluency) per well in 48-well plates, and viability was assessed by the resazurin viability assay after 5 days. Values are mean ± SEM (n = 5). (B) Cell plate images showing viable, proliferating vitamin C-treated Sod1^−/− MEFs at ~ 2 weeks post-isolation vs. dead untreated controls. Lower panel: resazurin viability assay (pink colour indicates the reduction of blue resazurin dye by metabolically active cells. (C) Viability measurement of Sod1^−/− MEFs after 5-day culture without or with increasing concentrations of vitamin C. Sod1^−/− cells were seeded into 48-wells at an initial density of 2 × 10⁴ cells/well (~ 20% confluency). Viable cells were assessed after 5 days by crystal violet staining. Values are mean ± SEM (n = 6). (D) Growth curves of Sod1^+/+ and Sod1^−/− MEFs treated with or without vitamin C. Relative proliferation was quantified at the indicated time points over 8 days using a resazurin viability assay. Data points represent metabolic activity normalized to the day 0 measurement taken 2 h post-seeding. Values are mean ± SEM (n = 6). ****p < 0.0001 versus untreated control (by one-way ANOVA followed by Dunnett’s multiple comparison test).

We also tested l-ascorbic acid 2-phosphate (Asc-2P), a stable phosphorylated derivative of vitamin C. In this molecule, a phosphate group is attached to the second carbon atom of a vitamin C molecule, protecting it from oxidation and thus enhancing stability against heat and light until it is activated by cellular phosphatase^63,64. It has been shown that a mixture of vitamin C and Asc-2P provides a more constant concentration of ascorbate in the culture medium⁶³. As anticipated, Asc-2P also rescued Sod1^−/− MEFs from acute death after isolation, and it showed greater efficacy compared to vitamin C, while the combination of both was even superior to either alone (Supplementary Fig. 1E). Compared to vitamin C treatment (0.25 mM) alone, the combination of vitamin C and Asc-2P (0.25 mM each) extended Sod1^−/− MEF growth beyond 5 weeks (Supplementary Fig. 1F). However, the effect was not indefinite either as the treatment was unable to sustain proliferation and viability past 8 weeks.

Protection against paraquat toxicity by vitamin C

We studied Sod1^+/− heterozygous MEFs for haplo-insufficiency and for the possibility of revealing consequences of SOD1 loss that might be hidden by the severe phenotype of Sod1^−/− cells. Compared to WT MEFs, Sod1^+/− MEFs exhibit greater sensitivity to paraquat (PQ), a well-known superoxide generator (Fig. 2). Vitamin C treatment conferred significant protection against this PQ cytotoxicity, whereas the antioxidants NAC and vitamin E showed no effect (Supplementary Fig. 2A,B). These observations align with the established primary mode of action of PQ, which involves the continuous production of superoxide via redox cycling^65,66. l-NAME, which inhibits NO synthesis and thereby inhibits the superoxide-^·NO reaction and the resultant formation of ONOO⁻, was also found to lack a protective effect (Supplementary Fig. 2C). These findings corroborate the observed lack of a rescue of the cell death of Sod1^−/− MEFs by these compounds.

Fig. 2.

Vitamin C protects against increased sensitivity to paraquat (PQ) toxicity in Sod1^−/− cells. (A) Viability measurement of Sod1^+/+ and Sod1^+/− MEFs after 6-day PQ exposure. Cells were seeded at a density of 2 × 10⁴ cells/well in 48-well plates (~ 20% initial confluency). Following 6 days of culture in the absence or presence of PQ, cell viability was assessed by the resazurin assay. Data points represent viability measurement normalized to the day 0 measurement taken 2 h post-seeding. Values represent mean ± SEM (n = 6). **p < 0.01, ****p < 0.0001 (by one-way ANOVA and Sidak’s multiple comparison test versus respective no-treatment control. ns: no significant difference compared to Sod1^+/+ MEFs. (B) Vitamin C protects against PQ toxicity. Sod1^+/+ and Sod1^+/− MEFs were seeded at a density of 1.5 × 10⁵ cells/well (~ 90% confluency) in 48-well plates. After 3 days of the indicated treatments, cell viability was assessed by resazurin assay. Data are expressed as a percentage of the untreated control (mean ± SEM, n = 6). ****p < 0.0001 comparing vitamin C-treated versus non-vitamin C-treated group; ^####p < 0.0001 versus the respective untreated control (one-way ANOVA with Sidak’s multiple comparison test).

Induction of Sod1 knockout in vitro and in vivo

As previously noted, germline Sod1 KO (Sod1^−/−) mice present only a relatively mild phenotype. A primary hypothesis to explain the surprisingly mild phenotype of the constitutive Sod1 KO mice is that SOD1 absence from the start of life triggers developmental compensatory mechanisms that mitigate the effects of its loss^{25,34–37,67}. However, this has not yet been substantiated. We generated a conditional Sod1 floxed strain to study the impact of acute Sod1 inactivation in vitro and in vivo. This approach minimizes the likelihood of the development of compensatory and adaptive responses and allows us to revisit the compensation hypothesis.

We isolated primary MEFs from the conditional Sod1 KO strain (Sod1^flox/flox) and, using retrovirus-mediated Cre delivery as previously described⁶⁸, induced Sod1 deletion in vitro. As shown in Fig. 3A, at 5 days after retroviral transduction, no SOD1 protein was detectable in the floxed cells transduced with Cre-expressing viruses. Concomitantly, these cells exhibited a marked reduction in total SOD activity. Of note, the residual activity is expected to derive from mitochondrial SOD2, the expression of which remained unchanged following SOD1 loss (Fig. 3A). While SOD1 accounts for most of the total cellular SOD activity in most mammalian tissues, SOD2 typically contribute approximately 20%^69,70.

Fig. 3.

Cre-mediated SOD1 inactivation causes growth arrest and cell death in Sod1^flox/flox MEFs. (A) Left panel: Western blot analysis showing complete loss of SOD1 protein 5 days post-transduction with pBabe-Cre-puro retrovirus (+ Cre) compared to mock-transduced (pBabe-puro retrovirus) control (− Cre). No significant change in SOD2 levels was observed. GAPDH serves as a loading control. Right panel: measurement of SOD activity in whole cell lysate (mean ± SEM, n = 4). **p < 0.01 (unpaired Student’s t-test). (B) Cell pictures taken 10 days post-retroviral transduction. (C) Recovery of SOD1 protein expression in Cre-transduced Sod1^flox/flox MEF cultures 4 weeks post-transduction. MnTMPyP (SOD mimic) treatment prevented the restoration of SOD1 expression. Blots in A and C are cropped; original uncropped blots are provided in Supplementary Figs. 3 and 4, respectively.

The inducible Sod1 KO MEFs (Sod1^flox/flox; pBabe-puro-Cre) showed severe growth arrest compared with the empty vector–infected control (Fig. 3B). And they appeared to be gradually lost from the population during continuous culture. By 4 weeks of culture, there was no visible difference in cell growth between Cre-transduced plates and empty-vector controls. Concomitantly, we observed that SOD1 protein levels were almost fully restored in the Cre-transduced population, indicating a loss of the KO phenotype over time (Fig. 3C). We interpret this to indicate that, because the loss of SOD1 severely compromises MEF growth and viability, MEFs carrying unrecombined alleles of Sod1^flox/flox (Sod1^flox/flox or Sod1^flox/+), though initially representing only a very small fraction of the population of cells, outcompeted the Sod1 KO population and eventually took over the entire population. This is supported by the observation that 4 weeks after Cre virus transduction there was a lesser degree of SOD1 protein restoration in the plates treated with the SOD mimic manganese (III) tetrakis (1-methyl-4-pyridyl) porphyrin (MnTMPyP), most likely because the compound conferred a survival benefit to the KOcells, thereby lowering the competitive advantage of the incompletely or unrecombined cells (Fig. 3C). Lastly, it is worth noting that neither vitamin C nor the SOD mimic could prevent the loss of the Sod1 KO population, likely due to the overwhelming growth advantage of cells retaining SOD1 expression.

To examine the impact of acute systemic SOD1 deletion in vivo, we generated an inducible Sod1 conditional KO strain by crossing Sod1 floxed mice with the CAG-CreER^T2 transgenic line, which expresses a tamoxifen-inducible Cre recombinase under the control of the CMV early enhancer/chicken beta actin (CAG) promoter. To induce Sod1 KO, tamoxifen was administered by intraperitoneal injection at approximately 6 weeks of age. As shown in Fig. 4A and B, loss of SOD1 protein expression and marked reduction of total SOD activity were confirmed by immunoblotting analysis and enzymatic activity assay, respectively. There was still 18–28% residual activity remaining in the KO tissues, which is presumed to originate from SOD2, as noted above. Of note, the CAG-Cre transgene is known to potentially exhibit incomplete deletion of target genes in the liver, as we and others previously observed^71–73.

Fig. 4.

Generation and characterization of inducible whole-body Sod1 knockout (KO) mice. (A) Western blot analysis confirms the near-complete loss of SOD1 protein in all major tissues except the liver. Blots are cropped; original uncropped blots are provided in Supplementary Fig. 5. (B) SOD1 enzymatic activity in various tissues. The remaining 18–28% residual activity in KO tissues likely originates primarily from mitochondrial SOD2. (C) Body weight at ~ 2 months after induction of Sod1 KO. No significant difference was found between inducible Sod1 KO and control animals. Data in (B) and (C) are expressed as mean ± SEM (n = 4). ****p < 0.0001 (unpaired Student’s t-test).

Despite a complete loss of SOD1 in most tissues, the inducible Sod1 KO mice showed no obvious phenotypic abnormalities. No weight loss was observed in the inducible Sod1 KO mice at 2 months post-induction, compared with vehicle-administered controls (Fig. 4C). One prominent phenotype of constitutive Sod1 KO mice is skeletal muscle pathology and accelerated functional loss. Characterized skeletal muscle changes in Sod1^−/− mice include accelerated loss of muscle mass and contractile force, myofiber atrophy with increased central nucleation, deterioration of neuromuscular junctions (NMJs) and denervation^{26,32,34,74,75}. We examined muscle histology approximately 3 months post-induction of systemic Sod1 KO. It found no evidence of significant pathology (Supplementary Fig. 6).

Discussion

Complete loss of SOD1 has been reported to severely halt proliferation and lead to cell death in MEFs, mouse dermal fibroblasts and an in vitro conditional SOD1 KO model generated using chicken lymphocyte DT40 cells^40–42. This proliferative defect and viability loss were reported to result from the activation of apoptosis as well as accelerated senescence^39,40. To understand why cells cultured in vitro, unlike those in vivo, cannot survive without SOD1, both successful and unsuccessful rescue attempts have been reported. In Sod1^−/− MEFs and conditional SOD1 KO cells from DT40 cells, lowering the oxygen concentration in the cell cultures was reported to partially mitigate the lethality of SOD1 deletion, yet failed to fully restore their proliferative capacity^39,42. For antioxidant treatments, two stabler vitamin C derivatives, magnesium ascorbyl phosphate (APM) and l-ascorbyl 2-phosphate 6-palmitate (APPS), were shown to enhance the viability and growth of Sod1^−/− dermal fibroblasts and SOD1 KO DT40 cells^40,42,76. In Sod1^−/− MEFs, 2-mercaptoethanol (2-ME) and NAC were shown to exert a slight protection from immediate cell death in hypoxic culture, while Tiron, (2,2,6,6-tetramethylpiperidin-1-yl)oxyl (TEMPO), and manganese(III) tetrakis (4-benzoic acid)porphyrin chloride (MnTBAP) failed to improve survival³⁹.

In this study, we report that, like Sod1^−/− MEFs isolated from constitutive KO mice, inducible KO of Sod1 in vitro also results in growth arrest and lethality (Fig. 3B). NAC, whose antioxidant function primarily targets H₂O₂, rather than superoxide, shows no effect in either case (Fig. 1A and Supplementary Fig. 1A). As noted above, increased apoptosis has been reported in Sod1^−/− primary dermal fibroblasts⁴⁰. However, we observe no effect of Z-VAD-FMK, a widely used pan-caspase inhibitor, on Sod1^−/− MEFs, suggesting that apoptosis is probably not the primary driver of the viability loss in Sod1 KO cells cultured in vitro (Fig. 1A).

Vitamin C retards the lethality of Sod1^−/− MEFs significantly (Fig. 1). It also protects against PQ toxicity (Fig. 2). The primary consequence of SOD1 deficiency or PQ exposure is the superoxide accumulation. Indeed, previous studies have reported elevated superoxide levels and increased oxidative stress in Sod1^−/− embryonic and dermal fibroblasts cultured under standard conditions^39,40. Sod1^−/− dermal fibroblasts were also reported to exhibit elevated mitochondrial superoxide levels and a loss of mitochondrial membrane potential, a sign of mitochondrial dysfunction⁴⁰. The beneficial effect of vitamin C observed is, most likely, a direct result of its superoxide scavenging activity. Vitamin C can directly reduce superoxide and is kinetically superior compared to other antioxidants such as glutathione and CoQ. However, the rate constant for the reaction of vitamin C with superoxide (2.7–3.3 × 10⁵ M⁻¹ S⁻¹) is roughly half that of the spontaneous dismutation of superoxide and about 10⁵-fold lower than the rate at which superoxide reacts with SOD^77–80. Thus, in most wild-type cell types, endogenous vitamin C concentrations are likely too low to complement SOD in superoxide removal. However, in cells either lacking SOD1 or exposed to PQ, where there is a marked accumulation of superoxide, vitamin C supplementation can be expected to offer critical protection by enhancing superoxide clearance. In line with this, the improved viability of Sod1^−/− mouse dermal fibroblasts treated with the vitamin C derivative APPS was shown to correlate with a reduction in intracellular superoxide levels⁴⁰. MEFs do not synthesize vitamin C, and standard culture media lack an exogenous source of this vitamin^81,82. Therefore, at the rescue doses used in this study (0.025–0.4 mM), though not very high, vitamin C supplementation could meaningfully offset toxic superoxide buildup in Sod1^−/− MEFs, thus mitigating their growth defect and lethality, as well as protecting wild-type and Sod1 heterozygous MEFs against PQ toxicity.

One critical mechanism of superoxide toxicity is the direct oxidation and destabilization of Fe–S clusters in various essential enzymes, resulting in their inactivation and release of catalytic iron. The cytosolic aconitase IRP1/ACO1 is among the most vulnerable. In fact, cytosolic aconitase and xanthine oxidase were reported to decrease significantly in the liver of Sod1^−/− mice, whereas mitochondrial aconitase remained unaffected^83,84. However, because an IRP1/ACO1 knockout is compatible with viability, its functional loss may contribute to but is unlikely to be the primary driver of Sod1^−/− cell lethality⁸⁵. Other superoxide-labile Fe–S enzymes include those critical for cell proliferation and genomic maintenance. For instance, the cytosolic ATP-binding cassette subfamily E member 1 (ABCE1) is essential for protein synthesis; its inhibition has been shown to reduce mammalian cell proliferation in culture^86,87. The nuclear DNA primase subunit 2 (PRIM2) is involved in DNA replication, and its loss is synthetically lethal with sod1∆ in yeast⁸⁸. The exact contribution of these enzymes to the non-viability of Sod1 KO cells in vitro remains to be studied.

As previously noted, another pathway for superoxide damage involves its direct reaction with ^·NO. In rat PC12 neuronal cells, SOD1 knockdown was shown to trigger apoptosis that could be counteracted by inhibiting NO synthesis with l-NAME, pointing to the reaction between superoxide and ^·NO as a major driver of SOD1-deficient toxicity in these cells^89–91. However, neither l-NAME nor FeTTPS improved the viability of Sod1^−/− MEFs and also lacked a protective effect against PQ toxicity (Fig. 1A, Supplementary Figs. 1A and 2C). The reaction also presumably consumes ^·NO significantly. However, boosting ^·NO levels through supplementation with the NO precursor l-arginine or the ^·NO donor DETA NONOate similarly failed to improve Sod1^−/− MEF viability (Supplementary Fig. 1A). Taken together, our data suggest that, in Sod1 KO MEFs, the reaction between superoxide and ^·NO plays a negligible role in their unviability. This discrepancy between our findings and the previous report may be attributed to potential variations in basal ^·NO production levels between these distinct cell types. Although undifferentiated PC12 cells reportedly have low NO production, they probably maintain enough endogenous activity to facilitate ONOO⁻ generation upon the rise of superoxide levels after SOD1 loss^92,93. In contrast, ^·NO availability in MEFs may be too low for ONOO⁻ to be a significant toxicity pathway⁹⁴.

Beyond elevating cytosolic superoxide, SOD1 loss also compromises the clearance of superoxide in other subcellular compartments locations where a fraction of the enzyme has been reported to localize, namely the IMS, nucleus, and peroxisome^5,18,95,96. Notably, targeting SOD1 exclusively to the IMS was shown to be sufficient to prevent motor neuropathy in Sod1^−/− mice, suggesting the importance of IMS-localized redox protection⁹⁷. However, how disrupted ROS homeostasis across these distinct compartments contributes to other specific Sod1 KO phenotypes, such as the in vitro growth defect, and the precise mechanism by which vitamin C confers protection (whether strictly via direct superoxide scavenging or additional actions) remains to be fully elucidated. On this note, consistent with the nuclear localization of the enzyme, SOD1-depleted cells have been reported to have increased genomic damage^39,40,42. In Sod1^−/− dermal fibroblasts, DNA damage was observed alongside elevated p53 protein levels, an effect significantly inhibited by APPS treatment⁴⁰. Similarly, in a human SOD1 conditional KO model generated from DT40 cells, DNA lesions and activation of the base-excision repair pathway were observed, and the vitamin C derivative APM protected against both⁴². However, a lack of genomic protection was postulated not to be the primary cause of cell lethality following SOD1 loss, as in the later model, it was shown that extranuclear SOD1 expression alone was sufficient to suppress cell death⁴².

Beyond its canonical role in detoxifying superoxide, other functions have been described for SOD1, including acting as a redox sensor, modulating mitochondrial function and dynamics, coordinating cellular stress responses, and serving as a cytosolic copper reservoir^18,98–102. Could impairment of some of those auxiliary functions contribute to the viability loss of SOD1-depleted cells? With continuous vitamin C supplementation, Sod1^−/− MEFs still eventually cease growth after 4–5 weeks of in vitro culture (Supplementary Fig. 1C). This incomplete rescue is most likely due to the relatively low scavenging efficiency of vitamin C and its inherent instability in culture medium, rather than indicating that the loss of a non-canonical function of SOD1 contributes to the viability loss. The fact that Asc-2P, a more stable derivative of vitamin C, enhanced and prolonged the growth of Sod1^−/− MEFs more effectively is consistent with this interpretation (Supplementary Fig. 1E,F). If superoxide levels are not sufficiently removed, sustained high levels could lead to cumulative damage that gradually overwhelms the cell’s turnover capacity, ultimately triggering growth arrest and death; however, the exact nature of these damages remains to be determined. Aligned with this speculation, the partial phenotypic rescue of Sod1 KO MEFs observed under hypoxic culture conditions was shown to be consistent with an incomplete attenuation of elevated oxidative stress³⁹.

The notion that superoxide toxicity primarily drives SOD1-deficiency phenotypes also offers a rationale for the striking discrepancy between in vitro and in vivo outcomes of Sod1 KO. In vivo, oxygen levels in most mammalian tissues, with a few exceptions (for example, the lung), generally range from 2 to 6%, approximately one order of magnitude lower than the 21% atmospheric oxygen level¹⁰³. This difference in oxygen tension may reduce the superoxide burden imposed on many cells in vivo and has been proposed to account for the viability and relatively mild phenotype of Sod1^−/− mice³⁹. Supporting this idea, pathological conditions known to provoke excess production of ROS, such as ischemia reperfusion and exposure to the redox-cycler diquat, have been shown to cause greater impairment in Sod1^−/− mice compared with heterozygous mutants and wild-type controls^104–106. Skin wound healing is notably delayed in these mutants, likely because the outer skin layers are directly exposed to high atmospheric oxygen concentrations¹⁰⁷. However, it is worth noting that even though our finding support the hypothesis that SOD1 is mostly dispensable in mice for the reasons that we just discussed, metabolic state, substrate availability, mitochondrial activity, antioxidant capacity, and other physiological variables differ substantially between cultured cells and intact tissues, as well as among tissues. Thus, the extent to which lower tissue oxygen tension accounts for the tolerance of Sod1 KO in vivo remains to be directly demonstrated.

On the other hand, a lower physiological oxygen tension and the consequently reduced superoxide burden in vivo may not fully account for the mild phenotype of Sod1^−/− mice. It has been suggested that developmental adaptation and compensatory mechanisms, often triggered by germline genetic mutations, likely provide additional protection that mitigates the loss of SOD1 in vivo^23,26,39. To name a few findings in favour of this notion, the skeletal muscle of Sod1^−/− mice was reported to exhibit upregulation of genes or proteins involved in redox control, metabolism or induction of proteolytic systems^34,38,67. The kidneys of young Sod1^−/− mice exhibit increased expression and activity of glutathione S-transferase Alpha 4 (GSTA4), considered a compensatory mechanism that protects renal cells from ROS generated upon iron exposure³⁶. Furthermore, adult-onset neuron-specific Sod1 KO led to motor neuron loss in young mice, followed by neuromuscular junctions (NMJs) disruption, muscle atrophy, and weakness past middle age³⁷. In comparison, embryonic neuron-targeted Sod1 deletion resulted in only a very mild NMJ phenotype and minimal muscle atrophy, even at the advanced age of 20 months¹⁰⁸. Nevertheless, this compensation hypothesis requires more testing to be validated.

We generated a novel, whole-body inducible Sod1 KO model and found that virtually complete SOD1 loss in all tissues (except the liver) does not result in acute pathology (Fig. 4, and Supplementary Fig. 6). Our findings align with a previous study using a different inducible model (Sod1^flox/flox; UBC-Cre^ERT2) which similarly reported essentially no phenotype, except for mild liver steatosis and a slight reduction in body weight during the first 9 months¹⁰⁹. As that study focuses on the roles of SOD1 and ROS in non-small cell lung cancer (NSCLC), it did not include a detailed molecular characterization of the systemic inducible model, such as validation of SOD1 expression loss across tissues. Nonetheless, that study and our present findings challenge the developmental compensation hypothesis. Future studies involving head-to-head phenotypic comparisons between inducible and constitutive models, including longitudinal monitoring into old age when accelerated aging phenotypes typically emerge in Sod1^−/− mice, are warranted to address this question rigorously.

Finally, it should be noted that mice can synthesize vitamin C, mainly in the liver¹¹⁰. Compensation by endogenously synthesized vitamin C has been proposed as an explanation for the mild phenotype of Sod1^−/− mice. The primary supporting evidence comes from genetic studies showing that combined Sod1 KO and impaired vitamin C biosynthesis (Akr1a^−/−; Sod1^−/−) produce more severe phenotypes than Sod1 KO alone, and that vitamin C supplementation improves survival in this setting¹¹¹. Additionally, plasma ascorbate has been shown to be decreased in Sod1^−/− mice^35,111. However, aldehyde reductase (AKR1A) has functions beyond vitamin C biosynthesis, and vitamin C has multiple biological roles in addition to antioxidant defence; thus, a definitive causal link remains to be established. Moreover, it is relevant to point out that human patients with homozygous SOD1 loss-of-function variants and absent SOD1 activity, who develop severe infantile-onset motor neuron disease, have plasma ascorbate levels similar to those of heterozygous individuals or healthy controls¹¹². This finding raises questions about vitamin C compensation as the key mechanism underlying the surprisingly benign phenotypes of Sod1^−/− mice.

In summary, the present study demonstrates that there is no significant discrepancy between inducible and constitutive Sod1 KO in both cell and mouse models. These results suggest that compensatory and adaptive mechanisms, which might be triggered during embryonic development, may not fundamentally mask or alter the phenotypic outcome of SOD1 loss. Rather, they point to a substantial functional reserve or biochemical buffer for SOD1 in animals, at least under controlled and non-challenged laboratory conditions. The identification of human patients with a predominantly infantile-onset SOD1 deficiency syndrome, caused by biallelic loss-of-function variants resulting in absent or near-absent enzymatic activity, has further reinforced the importance of better understanding superoxide homeostasis and the pathophysiological consequences of SOD1 loss^112–119.

We are left with the question of why there is such a high evolutionary conservation of the presence at high levels and in all cells of an enzyme that is not essential for acute survival. The explanation could lie in the fact that superoxide dismutases are the only enzymes that directly act on the poorly diffusible superoxide, and that superoxide production is both unavoidable and indispensable as a crucial signalling molecule. In this view, the evolutionary pressure of retaining SOD1 comes from an effect on general cellular health and not on an acute need, at least not under sheltered laboratory conditions. Under natural conditions, there might of course also be situations in which the need for SOD1 is acute. The need for SOD1 for quick skin healing alone would be such a requirement.

Methods

Chemical and reagents

Caspofungin (CF) was obtained from MedChemExpress (MCE), and the CF/Coenzyme Q₁₀ complex was prepared as previously described⁵³. N(ω)-nitro-l-arginine methyl ester (l-NAME) and diethylenetriamine/nitric oxide adduct (DETA NONOate) were purchased from Cedarlane Labs. All other chemicals were obtained from Sigma-Aldrich.

Cell culture and treatment

All cells were cultured at 37 °C under standard conditions in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% fetal bovine serum and 1% antibiotic/antimycotic mix (Wisent, Inc.). Mouse embryonic fibroblasts (MEFs) were isolated from day 12.5 mouse embryos obtained from heterozygous (Sod1^+/−) or homozygous(Sod1^flox/flox) matings using a standard procedure⁶⁸. To induce Sod1 deletion in Sod1^flox/flox MEFs, they were infected with the pBabe-puro-Cre retrovirus as previously described⁶⁸. Briefly, infection of Sod1^flox/flox MEFs with pBabe-puro-Cre or control pBabe-puro retroviral vectors was carried out in the presence of 10 μg/ml polybrene. Forty-eight hours post-infection, cells were selected in complete medium containing 10 μg/ml puromycin. For vitamin C treatment, a freshly prepared stock solution was added directly to the culture medium, and the medium was refreshed every 24–48 h.

Drug screen

Sod1^−/− MEFs were seeded in 48-well plates to reach > 90% confluence the following day. Compounds from the SCREEN-WELL REDOX library (BML-2835, Enzo Life Sciences) were added at a final concentration of 50 µM, with DMSO serving as the vehicle control. Cell viability was assessed using the resazurin assay 4 days after treatment. Compounds that increased viability by ≥ 20% relative to the DMSO control were considered hits.

Cell viability measurement

Cell viability measurements were taken as previously described¹²⁰. Briefly, for the resazurin assay, the medium was replaced at the end of treatment with fresh growth medium containing 0.15 mg ml⁻¹ resazurin. After 2 h of incubation at 37 °C with 5% CO₂, absorbance at 570 nm was measured using a plate reader (TECAN Infinite M1000), with 600 nm as the reference wavelength. For the crystal violet assay, cells were washed with PBS before adding a crystal violet solution [0.05% (w/v) crystal violet, 1% formaldehyde, and 1% methanol in 1× PBS]. Following overnight incubation at room temperature with gentle agitation, all traces of dye were removed, and the cells were washed thoroughly with distilled water. The bound stain was then solubilized with 10% acetic acid, and absorbance was measured at 590 nm.

Western blot

Cultured MEFs and mouse organs were lysed in RIPA buffer (Cell Signalling Technology). After quantifying protein concentrations via BCA assay (Thermo Fisher Scientific), the supernatants were collected for Western blot analysis. Protein (20 µg) was loaded onto a 16% SDS-PAGE gel and transferred to a 0.2 μm PVDF membrane (Bio-Rad). The primary antibodies used were anti-SOD1 (1:4000; 10269-1-AP; Proteintech), anti-SOD2 (24127-1-AP; 10269-1-AP; Proteintech), and anti-GAPDH (1:8,000; #2118; Cell Signaling Technology). HRP-conjugated anti-rabbit secondary antibody (#7074; Cell Signalling Technology) was used for detection. Blots were developed using ECL substrates (Froggabio Inc.) and visualized by exposure to X-ray film.

Mice

Germline Sod1 KO mice were obtained from MMRRC (Mutant Mouse Resource & Research Centers). A conditional allele of Sod1 was created by inGenious Targeting Laboratory (Ronkonkoma, NY, USA) via homologous recombination in C57BL/6 × 129/SvEv hybrid embryonic stem (ES) cells. Targeted ES cells were microinjected into C57BL/6 blastocysts, and the resulting chimeras were bred to C57BL/6 FLP mice to remove the neomycin resistance cassette. In the final Sod1 floxed allele, exons 2 and 3 are flanked by loxP sites. To generate tamoxifen-inducible Sod1 KO mice, Sod1 floxed mice were mated with the CAG-CreER^T2 transgenic line as previously described⁷¹. Tamoxifen was administered via intraperitoneal injection at approximately 6 weeks of age to induce Sod1 deletion.

Mice were group-housed by sex (2–5 per cage) at the Animal Facility of McGill University. They were maintained in a controlled environment at 18–24 °C with 30–70% humidity under a standard 12/12 h light/dark cycle (7:00–19:00 light period) and given ad libitum access to standard laboratory chow and water. At weaning (postnatal day 21), mice were ear‐notched for identification. The Cre transgene was detected by PCR with primers Cre-F (5′-GCCAGCTAAACATGCTTCATC-3′) and Cre-R (5′-ATTGCCCCTGTTTCACTATCC-3′). Sod1 WT and floxed alleles were detected with primers: Sod1-F (5′-CTCCACAGGCAGTAGGACAAAGG-3′) and Sod1-R (5′-ACAACTGGTTCACCGCTTGCCTTG-3). All experimental protocols involving mice were approved by the Animal Care Committee of McGill University. All experiments were conducted in accordance with the guidelines of the Canadian Council on Animal Care.

SOD activity assay

SOD activity was quantified using colorimetric assays according to the manufacturers’ instructions. For activity measurement in cultured cells, CelLytic™ M lysis buffer (Sigma-Aldrich) was used to lyse the cells, and SOD activity in the resulting lysates was measured using Sigma-Aldrich’s SOD Determination Kit (Cat# 19160). For activity measurement in mouse tissues, tissues were homogenized in a cold buffer containing 20 mM HEPES, 1 mM EGTA, 210 mM mannitol, and 70 mM sucrose, and SOD activity was measured using Cayman’s Superoxide Dismutase Assay Kit (Cat # 706002). In all samples, protein concentration was determined using a BCA Protein Assay Kit (Thermo Fisher Scientific) to normalize SOD activity levels.

Statistical analysis

Prism 10.0 (GraphPad Software, Inc) was used to perform statistical analyses and create graphs. All quantitative data are expressed as mean ± standard error of the mean (SEM). The specific types of statistical tests performed are indicated for each graph in the figure legends, and p < 0.05 was used as the cut-off for significant differences of group means.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

S.H.: Conceptualization, investigation, writing—review and editing. Y.W.: Investigation, writing—original draft, Writing—review and editing. A.N. Investigation. L.T. Investigation.

Funding

This work was supported by the Canadian Institutes of Health Research, FDN-159916.

Data availability

Datasets used and/or analyzed during the current study are available from the corresponding author upon reasonable request.

Declarations

Competing interests

The authors declare no competing interests.