Iron-addicted colorectal cancers exploit heme-complex II axis to resist oxidative cell death Authors: Chesta Jain, Muqit Essani, Roshan Kumar, Nupur K. Das, Rashi Singhal, Nicholas J. Rossiter, Brandon Chen, Wesley Huang, Zheng Hong Lee, Sumeet Solanki, Yuezhong Zhang, Peter Sajjakulnukit, Li Zhang, Prarthana J. Dalal, David A. Hanna, Cristina Castillo, Harrison S. Greenbaum, Shannon E. McCollum, Elena M. Stoffel, Joel K. Greenson, L. James Maher, Costas A. Lyssiotis, Ruma Banerjee, Yatrik M. Shah Published: 2026 ABSTRACT SUMMARYColorectal cancer (CRC) cells are addicted to iron, which fuels nucleotide synthesis, mitochondrial respiration, and proliferation. Yet paradoxically, high intracellular iron is cytotoxic to most cells, raising the question of how CRC cells tolerate and exploit iron-rich environments. Ferroptosis, an iron-dependent form of cell death, is thought to mediate iron toxicity. However, whether most ferroptosis regulators, identified through synthetic chemical screens or small molecule activators, play a role in modulating iron toxicity, particularly in vivo, remains unclear. Here, using multi-omics profiling, CRISPR screening, and in vivo models, we uncover a heme-succinate dehydrogenase (SDH)-coenzyme Q (CoQ) axis that enables CRC cells to buffer iron-induced oxidative stress. Heme-dependent SDH reduces CoQ, which redistributes to mitochondrial and plasma membranes to detoxify lipid reactive oxygen species (ROS) as a radical-trapping antioxidant. These findings reveal that CRCs co-opt metabolic cofactors both for growth and for survival under physiologically toxic iron levels, uncovering new vulnerabilities for therapy. INTRODUCTION Colorectal cancer (CRC) is the third most diagnosed cancer globally and the third leading cause of cancer-related mortality.1 CRC cells are remarkably resilient, displaying profound metabolic adaptations that enable survival and growth in nutrient- and oxygen-deficient microenvironments.2 This metabolic rewiring is essential for optimal nutrient utilization and the prioritization of cellular energetics for proliferation.2 While such adaptations support tumor growth, they also create dependencies on specific metabolites and micronutrients. Although much attention has been focused on the role of altered metabolism in tumorigenesis, the contribution of micronutrients to cancer progression remains poorly understood. Iron is a critical micronutrient required by nearly all cells. Both systemic and cellular iron homeostasis are tightly regulated through transcriptional and translational control of iron uptake, utilization, and storage.3 Dysregulation of these mechanisms can lead to iron deficiency or iron overload, manifesting clinically as anemia or hemochromatosis, respectively.4 At the cellular level, labile iron supports proliferation, ATP production, and mitochondrial respiration through its redox activity. The role of iron and its regulated pathways in cancer initiation, progression, and metastasis is well established and continues to be a highly active area of investigation.5 In CRC, epidemiological studies have linked elevated systemic iron with increased CRC incidence.6–8 CRC cells accumulate significantly higher levels of iron compared with adjacent normal tissue, largely by enhancing iron uptake and limiting export.9–12 We and others have shown that restricting iron availability reduces proliferation, lowers tumor burden, and improves survival in mouse models.9,12 However, therapeutic iron depletion remains clinically challenging. CRC cells can access iron from both dietary sources and systemic circulation, and lowering intracellular iron primarily slows proliferation rather than eliminating transformed cells.9,13 Excess intracellular iron is toxic, driving a form of lipid peroxidation-dependent, non-apoptotic cell death known as ferroptosis.14,15 This creates a striking paradox. CRC cells sequester and accumulate iron and resist its toxicity. Previous work has pointed to elevated antioxidant defenses in CRC models and patient tissues, but the specific metabolic mechanisms that allow cancer cells to tolerate such iron overload remain unknown.16 This represents a fundamental gap in our understanding of CRC biology. How do CRC cells withstand and even thrive under conditions that should be lethal due to iron-induced oxidative stress? Addressing this question is critical, as uncovering these adaptive mechanisms could expose new therapeutic vulnerabilities. We show that the canonical anti-ferroptosis response mediated by GPX4 and SLC7A11 is dispensable for CRC growth and progression in both sporadic and colitis-associated cancer (CAC) models. A metabolism-focused CRISPR screen revealed a protective role of cellular heme against iron toxicity. As a critical cofactor for mitochondrial electron transport chain (ETC) enzymes, heme depletion caused profound mitochondrial defects, including loss of complex II (CII) activity. Through multi-omics profiling of in vitro and in vivo CRC models, we identify a previously unrecognized role of mitochondrial CII in buffering iron-induced cell death by regulating coenzyme Q (CoQ) biosynthesis. Strikingly, pharmacological or genetic inhibition of CII sensitized CRC cells to iron-induced toxicity in vitro and in vivo. Collectively, these findings uncover a non-canonical antioxidant function of CII in protecting CRC cells from iron-induced toxicity. By defining how CRC cells adapt to survive under iron overload, our study addresses a major gap in the field and highlights metabolic strategies for the development of new therapies. RESULTS GPX4 is dispensable for CRC initiation and progression A defining paradox in CRC biology is its ability to accumulate and tolerate exceptionally high levels of intracellular iron.9 Iron is indispensable for DNA synthesis, mitochondrial metabolism, and rapid proliferation.13,17 Yet in most cells, sustained iron loading is toxic, as it accelerates reactive oxygen species (ROS) production and triggers ferroptosis, an iron-dependent cell death program driven by lipid peroxidation.14,15 The prevailing paradigm is that cancer cells buffer this stress primarily through the GPX4-SLC7A11 antioxidant axis.18,19 We first assessed the expression of SLC7A11 and GPX4 in tumor tissues and matched normal mucosa from individual CRC patients and in enteroids generated from these same tumors and matched normal tissues. In all cases, RNA transcripts for both were significantly elevated in the tumor-derived material compared with matched normal controls (Figures 1A and 1B), indicating that up-regulation of these pathways is intrinsic to the tumor epithelium and maintained ex vivo. These findings were consistent with The Cancer Genome Atlas (TCGA) datasets, where GPX4 and SLC7A11 transcripts are elevated in CRC tumors relative to normal colon tissue (Figure S1A). Despite this consistent upregulation, Kaplan-Meier survival analysis revealed a weak correlation between high GPX4/SLC7A11 expression and overall patient survival (Figure 1C). However, as a single gene association using TCGA dataset, GPX4 was associated with poor prognosis. To test this functionally, we generated intestine-specific Gpx4 knockout mice (Figures 1D–1F). Intestine-specific Gpx4 deletion does not manifest as any significant histological or physiological changes and does not alter overall GI health or overall survival (Figure 1G).20 In the CAC model, Gpx4 loss did not significantly alter tumor number, size, or overall burden (Figures S1B–S1E). While these findings suggest that CRC cells rely on metabolic adaptations beyond the canonical SLC7A11-GPX4-dependent anti-ferroptotic response, CAC is rare in the human population, and the majority of the sporadic CRC cases are APC, P53, and KRAS mutation driven. To examine the role of GPX4 in sporadic CRC, we crossed the Gpx4F/F mice with a series of genetically engineered colon cancer models: single mutant (ApcF/F), double mutant (ApcF/F; Tp53F/F), or triple mutant (ApcF/F; Tp53F/F; KrasG12D).21 All mutations were specifically targeted to the colon epithelium using CDX2-CreERT2 and showed effective recombination and GPX4 deletion in all four genotypes (Figures 1H and S1F). If GPX4 were essential to buffer tumoral iron and resist iron-induced toxicity, its deletion would be expected to slow tumor progression and improve survival. Surprisingly, Gpx4 deficiency did not improve survival in any of these models (Figures 1I–1L). While the percent tumor area corresponded to the mutational burden with a significant increase in histologically advanced tumors in the triple mutant mice, no differences were observed in the colon between the Gpx4F/F and Gpx4ΔIE (Figures 1M, S1G, and S1H). Notably, in mice harboring Apc or Apc/Tp53 mutations, Gpx4 deficiency slightly shortened lifespan, which may reflect altered inflammatory responses in the absence of GPX4.22 This indicates that CRC does not depend on the canonical SLC7A11-GPX4 ferroptosis defense to tolerate iron stress. Instead, CRC must exploit alternative metabolic adaptations that enable survival under high iron. CRC cells resist iron toxicity through metabolic adaptations To understand if the resistance to high iron levels is a cell-autonomous mechanism, we first compared proliferation under millimolar concentrations of ferric ammonium citrate (FAC) in CRC cell lines versus normal intestinal epithelial cells and non-CRC cancer lines. CRC cells maintained robust proliferation, whereas non-CRC cells rapidly succumbed to iron toxicity (Figures 2A–2C). Sensitivity to iron varied across CRC lines, reflecting heterogeneity in intrinsic metabolic adaptations (Figure S2A). Chronic high-iron exposure reduced CRC growth, sensitized cells to ferroptosis inducers (RSL3 and erastin), and induced cell death, which was partially rescued by ferroptosis inhibitors and antioxidants such as liproxstatin (Lip-1) and N-acetyl cysteine (NAC) (Figures 2D and S2B). By contrast, iron-induced death in non-CRC lines (fibroblasts CD1, NIH3T3, and fibrosarcoma HT1080) was fully reversed by ferroptosis inhibition (Figures S2C and S2D). This raises the question: what metabolic adaptations allow CRC cells to survive under iron loads that are lethal for normal intestinal epithelial cells or non-CRC cancer cells? Intracellular iron homeostasis depends on coordinated regulation of iron import/export and ferritin-mediated storage,3 which collectively maintain a free iron pool also called the labile iron pool (LIP). The LIP pool increases significantly by directly treating cells with FAC; however, iron-sensitive cells such as HT1080 and RKO show a significant increase in the LIP with a much lower concentration of FAC (Figure S2E). This indicates that the iron resistance could be due to the cellular capacity to maintain a steady-state LIP potentially through coordinated iron uptake, export, and storage. CRC cells have increased expression of iron importers and hepcidin, an iron export repressor, via sustained HIF2α transcriptional activity.9,10,13 Ferritin acts as both a bioavailable iron reservoir and a buffer against iron overload (Figure 2E).23,24 To test its role in CRC iron tolerance and maintenance of the LIP, we generated CRC lines with doxycycline-inducible short hairpin RNA (shRNA) targeting ferritin heavy chain (FTH1) (Figure 2F). FTH1 knockdown (KD) was lethal, markedly reducing proliferation, and growth defects were not rescued by lipid ROS scavengers (Figures 2G, 2H, and S2F). Interestingly, FTH1 protein levels saturated at low FAC concentrations and did not increase further under higher FAC, and overexpression did not improve iron tolerance (Figures 2I, S2G, and S2H). These data indicate that while FTH1 maintains basal iron homeostasis, it is insufficient to buffer the extreme LIPs in CRC. To uncover additional mechanisms, we performed a metabolism-focused CRISPR-Cas9 screen.25 Cells were maintained under chronic FAC exposure to mimic iron overload or under HIF2α activation via FG4592 treatment, which we and others have shown increases intracellular iron levels and susceptibility to iron-induced toxicity (Figure 2J).26 This approach allowed us to probe metabolic pathways critical for CRC survival under physiologically relevant iron stress conditions. The screen identified depletion of several mitochondrial metabolism-related genes, including those involved in purine/pyrimidine biosynthesis, cholesterol biosynthesis, and heme biosynthesis, under both FAC and HIF2α activation (Figures 2K, 2L, and S2I). FTH1 was selectively depleted under FAC but not FG4592, consistent with HIF2α-mediated transcriptional regulation of ferritin (Table S1). These results suggest that CRC survival under high iron depends on mitochondrial and heme-dependent metabolic pathways, beyond canonical ferritin-mediated iron storage. Heme biosynthesis buffers iron-induced toxicity The metabolism-focused CRISPR-Cas9 screen (Figure 2L) identified uroporphyrinogen decarboxylase (UROD), a key enzyme in heme biosynthesis, as a top-scoring gene significantly depleted under both FAC and HIF2α activation (FG4592) treatments. Heme, an iron-containing cofactor, is essential for numerous cellular processes, including mitochondrial respiration, Fe-S cluster biogenesis, and nucleotide synthesis.27 While dietary heme is a major source, all cells, including CRC cells, express the full complement of enzymes required for de novo heme synthesis. Increasing intracellular iron elevates cellular heme pools in CRC cells (Figures 3A and 3B). To validate the screen, we generated HCT116 and SW480 cells with Tet-inducible KD of UROD using two independent shRNAs (shRNA1 and shRNA2) with empty vector (Scr) controls (Figure S3A). Consistent with the previous findings, depleting intracellular heme genetically by knocking down UROD or pharmacologically using succinyl acetone (SA), a competitive inhibitor of delta-aminolevulinate dehydratase (ALAD), sensitizes CRC cells to iron-induced toxicity (Figures 3C and S3B). Heme depletion upon SA treatment-induced iron sensitivity is completely rescued by supplementing heme back in the culture media (Figures 3D and 3E). KD of ferrochelatase (FECH), which inserts Fe2+ into protoporphyrin IX in the mitochondrial matrix, similarly sensitized CRC cells to iron toxicity, with rescue upon heme supplementation (Figures S3C–S3E). These results indicate that heme functions as an essential cofactor supporting antioxidant and metabolic responses, rather than acting as a passive iron store. This response is conserved in other CRC cell lines as well as other cancer lines (Figure S3F). Heme depletion increased both general cellular ROS and lipid ROS under FAC treatment, effects that were reversed by either heme supplementation or ferroptosis inhibitor Lip-1 (Figures 3F, 3G, and S3G). Interestingly, only heme supplementation rescued SA-induced growth suppression, indicating a non-ferroptotic, heme-dependent oxidative vulnerability (Figure 3H). Furthermore, despite increased lipid ROS and general ROS, heme depletion or supplementation did not alter the sensitivity to other ferroptosis inducers such as RSL3 and IKE (Figure 3I). These data show that iron-induced toxicity is aggravated by the induction of general ROS and lipid ROS; however, quenching these species is not sufficient to reverse iron toxicity, suggesting a non-ferroptotic iron-dependent oxidative cell death. Heme depletion also reduced oxygen consumption rates (OCRs) and decreased protein levels of ETC complexes (CII, CIII, and CIV) in CRC cells; both parameters were restored by heme supplementation (Figures 3J, 3K, and S3H). Increased oxidative stress in the cells can destabilize Fe-S and therefore the stability of Fe–S-containing proteins, which could explain the defects in OCR and stability of the ETC complexes. Heme depletion showed significant reduction in the Fe–S-dependent catalytic activity of aconitase; however, the loss of ETC complexes could not be rescued with antioxidants (Figures S3I and S3J). These findings are consistent with the known role of heme in mitochondrial respiration and suggest a protective role of heme for buffering iron-induced oxidative stress. Overall our findings align with the epidemiological data that high expression of genes involved in heme uptake and absorption correlates with reduced overall survival in CRC patients, consistent with the role of heme in promoting tumor cell fitness under high iron conditions (Figure 3J).8 Collectively, these data demonstrate that an increase in cellular heme through biosynthesis is a key metabolic adaptation that enables CRC cells to tolerate high intracellular iron, bridging the gap between iron accumulation and oxidative stress resistance. Mitochondrial CII protects CRC cells from iron-induced oxidative cell death Heme depletion through both pharmacological and genetic approaches induced oxidative stress; however, it did not sensitize cells to ferroptosis inducers (Figures 3F–3I). We wanted to investigate whether depletion induces sensitivity to other oxidative stressors beyond iron. Interestingly, heme depletion did not sensitize CRC cells to other metals, copper and selenium, or oxidants, hydrogen peroxide, and chemotherapeutic agent 5-fluorouracil, demonstrating specificity for redox stress (Figures S4A and S4B). Next, we wanted to understand the specific role of heme in regulating iron sensitivity in CRC cells. Heme itself can be degraded by hemoxygenase (HO)-1 into the byproducts labile iron, bilirubin, and biliverdin.28 The latter two are hydrophobic molecules with potent antioxidant properties in membranes, where they can protect cells from iron-induced toxicity.29 However, this process simultaneously releases free iron, complicating the interpretation of their protective effects. We did not see enhanced iron sensitivity upon alteration of cellular bilirubin and biliverdin levels by inhibition of HO-1 or protection by directly supplementing with products of HO-1 (Figure S4C). Heme deficiency activates HRI kinase activity through autophosphorylation, triggering the ATF4-dependent integrated stress response and reducing translation of mitochondrial proteins, including the ETC complexes.30 We observed no changes in canonical ATF4 target gene expression across FAC, FAC + SA, and FAC + SA + heme treatments (Figure S4D). Notably, mRNA levels of both CHOP and ATF4 were increased in FAC + SA treatment. Since CHOP is broadly induced during cell death, this increase is likely reflective of enhanced cell death rather than a specific heme-regulated signaling effect. Beyond providing structural and catalytic support to mitochondrial proteins, heme availability has been reported to regulate antioxidant responses through BACH1 (BTB and CNC homology 1) and cap-dependent translation via the heme-regulated kinase (HRI).30,31 Free heme can bind to BACH1, a translational repressor of other master antioxidant response transcription factors such as nuclear factor erythroid 2-related factor 2 (NRF2).32 We did not see a significant reduction in the mRNA transcripts of NRF2 targets in SA-treated cells (heme deficient) (Figure S4E). Because ~20% of total heme-containing proteins function in the ETC and ~50% serve catalytic roles in diverse metabolic reactions, we next investigated whether heme deficiency disrupts CRC metabolic programs beyond respiration.33 Our prior work demonstrated that CRC cells rely on iron-dependent nucleotide synthesis, but it was unclear whether this phenotype was strictly iron- or heme-dependent.13 Metabolomics revealed that SA-induced heme depletion caused a broad metabolic reprogramming, which was only partially corrected by heme supplementation (Figure 4A). TCA and glycolytic intermediates, including lactate, fructose-1,6-bisphosphate (F-1,6BP), glucose-6-phosphate (G6P), and dihydroxyacetone phosphate (DHAP), were markedly reduced in SA-treated cells and restored upon heme repletion, while succinate accumulated in SA-treated cells (Figure 4B). Under nutrient- and oxygen-limiting conditions, resembling the tumor microenvironment, amino acids such as glutamine can fuel the TCA cycle to maintain cell metabolism independent of oxygen-driven mitochondrial respiration.34,35 Consistent with this, glutamine and α-ketoglutarate, as well as glutaminolysis-dependent intermediates including aspartate, orotate, and uridine, accumulated in both SA- and SA + heme-treated cells (Figures 4B–4D; Table S2). To probe whether elevated iron enhances flux through these pathways, we performed 13C5-L-glutamine tracing (Figure 4E). FAC treatment did not significantly increase incorporation of the label into metabolites altered by heme deficiency, whereas iron chelation with deferoxamine reduced flux (Figure 4F; Table S3). Nevertheless, both FAC and heme supplementation increased OCR in iron-resistant (HCT116 and SW480) but not in iron-sensitive (HT1080 and RKO) cells, suggesting that the heme-dependent metabolic rewiring of the TCA cycle and nucleotide synthesis occurs primarily through respiratory chain-mediated processes (Figures 4G–4I). Given the observed ETC defects upon heme depletion, we hypothesized that disruption of specific complexes may sensitize CRC cells to iron stress. Indeed, pharmacologic inhibition of succinate dehydrogenase (SDH) selectively sensitized CRC cells to iron-induced toxicity, while inhibition of other ETC complexes produced modest effects (Figures 4J and 4K). These findings point to the heme-CII axis as a central metabolic safeguard that enables CRC cells to tolerate high iron and sustain proliferation. SDH-CoQ axis integrates mitochondrial metabolism with iron detoxification CII is composed of four nuclear-encoded SDH subunits that mediate electron transfer from the TCA cycle into the ETC through succinate oxidation to fumarate. Electrons released from this reaction pass sequentially through FAD, three iron-sulfur clusters, and a heme cofactor before ultimately reducing CoQ, which then serves as an electron carrier to CIII36 (Figure 5A). Similar to pharmacological inhibition of CII by (2-thenoyltrifluoroacetone) TTFA, CRC cells deficient in the structural subunits SDHA or SDHB exhibited heightened sensitivity to iron toxicity but not to other stressors such as copper, H2O2, or 5-fluorouracil (Figures 5B, 5C, and S6A–S6E). Since heme depletion did not alter the succinate levels in the cells, we next considered whether CII-dependent CoQ reduction is critical for modulating tolerance to iron in CRC cells (Figure 4D). Interestingly, both total and reduced CoQ levels were increased in cells exposed to elevated iron in a CII-dependent manner (Figure 5D). Deletion of CI subunits (NDUFS1 and NDUFS2) also capable of reducing CoQ10 did not alter iron sensitivity, underscoring the specificity of the heme-SDH-CoQ axis in protecting against iron toxicity (Figure S6F). Inhibition of CII-dependent CoQ reduction increased both lipid ROS and total ROS, effects that could not be rescued by heme supplementation (Figures 5E and 5F). Given prior data that CoQ can translocate from mitochondria to other cellular membranes, where it contributes to detoxification of lipid ROS, we examined CoQ distribution under iron stress (Figure 5G).37,38 Remarkably, we observed an iron-induced, CII-dependent increase in plasma membrane CoQ levels, further supporting its role as a lipid antioxidant defense system (Figures 5H and 5I). Unlike heme depletion, CII inhibition did sensitize cells to ferroptosis inducers, reversed by supplementation of CoQ analog, idebenone (Figure 5J). Notably, CII inhibition-induced iron sensitivity was rescued by exogenous CoQ analogs and the ferroptosis inhibitor liproxstatin-1, but only the membrane-permeable CoQ analog idebenone, and not the mitochondria-targeted MitoQ restored viability (Figure 5K). The iron-dependent increase in the total CoQ levels was observed in other iron-resistant cells such as DLD1 and Huh7 and not in iron-sensitive cells such as RKO, HT1080, and MC38. However, inhibition of CII significantly depleted CoQ levels in all cell lines irrespective of their sensitivity to iron (Figures S5G and S5H). The pancreatic cancer cell line Miapaca was an exception, as it showed increased CoQ levels upon iron treatment even though it’s highly sensitive to iron toxicity (Figure S5I). Consistently, we observed CII inhibition-induced iron toxicity in these non-CRC iron-sensitive cells, suggesting that the ability to sustain and induce CII-dependent CoQ reduction may confer resistance against iron and other ferroptotic stimuli (Figure 5L). Furthermore, CII inhibition did not affect cell proliferation at the 24 h time point measured by bromodeoxyuridine (BrdU) staining (Figure S5J). FSP-1-dependent redox of cycling of CoQ in the plasma membrane is shown to modulate ferroptosis sensitivity in parallel to the canonical Slc7a11-GPX4-GSH axis.39 Although the metabolism-focused CRISPR screen did not identify loss of other CoQ-reducing enzymes, including FSP-1, as a significant vulnerability, we wanted to reevaluate this in the context of impaired CII activity. Therefore, we developed a custom CRISPR library targeting 40 genes involved in lipid peroxidation, iron metabolism, and cellular antioxidant response. Consistent with our previous results, FSP-1 was not significantly depleted in the FAC- and FG4592-treated group; however, it emerged as a top hit in cells treated with TTFA (Figure S5K; Table S4). Together, these findings identify SDH-dependent CoQ reduction as a central metabolic defense against iron-induced oxidative stress, linking TCA cycle electron flow, heme metabolism, and extramitochondrial antioxidant capacity across mitochondrial and plasma membrane compartments. Heme-dependent SDHC expression is essential for protection against iron-induced toxicity in CRC cells Heme depletion led to a loss of multiple ETC subunits, diminished respiratory output, and a profound shift in the cellular metabolome. In erythroid progenitors, heme availability is known to control essential processes transcriptionally and translationally, such as proliferation, differentiation, and intercellular communication. However, whether similar heme-dependent regulation operates in cancer cells has remained unclear. To address this, we performed quantitative proteomics using tandem mass tag labeling of mitochondrial-enriched fractions from HCT116 cells treated with FAC, SA, or their combination for 72 h (Figures 6A, 6B, and S6A–S6C).40 Among the most striking changes, CII subunits and assembly factors emerged as iron-induced, heme-dependent proteins. SDHC (succinate dehydrogenase subunit C) was significantly enriched under FAC (heme sufficiency) but strongly depleted with FAC + SA (Figure 6C). Heme-dependent SDHC stabilization was required to sustain CII activity and maintain cellular CoQ pools, thereby mitigating iron toxicity in CRC cells. Consistently, heme depletion and CII inhibition yielded highly concordant proteomic signatures, with ~48% overlap in differentially altered proteins (Figures 6D and 6E; Table S5). Notably, heme biosynthetic enzymes were largely unaffected by SA treatment, with the exception of ALAS1, which is regulated by cellular iron via IREs (Figure S6D).41 Importantly, other CoQ-reducing enzymes that have also been implicated in mediating mitochondrial redox balance (DHODH, SQOR, and ETFDH)42,43 were upregulated in FAC + SA cells, underscoring the central role of reduced CoQ as an antioxidant defense (Figure S6E). Beyond CII, protein abundance across CI–CV subunits and iron-sulfur cluster-containing proteins was increased under FAC and reduced with SA, whereas Fe-S biosynthetic enzymes remained unchanged (Figures S6F–S6H). Interestingly, our data suggest that iron itself directly increases SDHC protein abundance in CRC cells; however, large-scale patient datasets demonstrate the opposite. Analysis of TCGA and Clinical Proteomic Tumor Analysis Consortium revealed that all SDH subunits (SDHA–SDHD) are decreased at the mRNA and protein levels in CRC tumors relative to matched normal tissue (Figure 6F).44 Our immunohistochemistry analyses demonstrated a clear increase in SDHC protein specifically within tumor epithelial cells of COAD patients as compared with the normal adjacent tissue (Figures 6G and 6H) (Table S6). These findings demonstrate that bulk tumor profiling underestimates tumor intrinsic SDH expression as invasive CRC lesions are intermixed with surrounding stromal tissue, which expresses negligible SDH. To understand the mechanism by which the iron-heme pathway increases SDHC levels, we first assessed SDHC mRNA expression but did not observe any significant changes (Figure S6I). Because SDHC has been reported to be a heme-binding protein, we next tested whether heme could stabilize the SDHC protein. To determine whether heme binding influences SDHC stability, we performed a temperature-dependent solubility assay followed by western blot analysis. In this assay, cell lysates were exposed to increasing temperatures to induce protein denaturation and aggregation, and the soluble fraction remaining after centrifugation was analyzed by western blot.45 In the presence of heme, SDHC exhibited increased thermal stability, indicating that heme binding stabilizes the protein (Figure S6J). Inducible KD of SDHC in HCT116 and SW480 cells enhanced iron toxicity, phenocopying pharmacological inhibition of CII with TTFA, and reversal with RTAs such as IDB and Lip-1 but not MitoQ (Figures 6I, 6J, and S6K). Knocking down SDHC concomitantly reduced iron-induced CoQ10 levels in the whole cell as well as the plasma membrane (Figures 6K and S6L). Metabolically, SDHC-deficient cells, like TTFA-treated cells, showed moderate succinate accumulation and a more robust decrease in other TCA intermediates (Figure S6M; Table S7). Together, these data reveal a heme-dependent regulatory mechanism that sustains SDHC expression and CoQ10 homeostasis, positioning the heme-SDH-CoQ axis as a critical antioxidant pathway that enables CRC cells to thrive under iron-rich conditions. SDHC inhibition sensitizes CRC cells to iron-induced toxicity and reduces tumor progression To evaluate the in vivo efficacy of SDHC inhibition in promoting iron-induced toxicity, we generated CT26 mouse CRC cells with a doxycycline-inducible shRNA targeting Sdhc and implanted them subcutaneously into BALB/c mice (Figure S7A). Once tumors were established, doxycycline-containing chow was administered to induce Sdhc KD (Figure S7B). SDHC-deficient tumors were significantly smaller and exhibited elevated lipid ROS, although the magnitude of this effect was moderate (Figures S7C–S7E). These results and our in vitro data suggest that SDHC deficiency induces oxidative stress, sensitizing cells to iron-induced toxicity, and treatment with antioxidants should completely rescue the growth defect. To test this, we implanted HCT116, SW480, and CT26 cells stable for SDHC-targeting shRNA, followed by doxycycline-containing chow since dox alone did not significantly affect the tumor size and daily treatment with vehicle or Lip-1 through intraperitoneal (i.p.) injections (Figures 7A and S7D). SDHC-deficient cells result in smaller tumors with high lipid ROS and cell death, which was completely reversed in the Lip-1-treated group (Figures 7B, 7C, and S7F). Because subcutaneous tumor models do not fully recapitulate the colonic tumor microenvironment or the extent of iron accumulation observed in patients, we next crossed SdhcF/F mice with a tamoxifen-inducible intestinal epithelial Cre line (Villin-ERT2Cre mice) to generate intestine-specific Sdhc-deficient mice (Sdhc-IntKO) (Figure 7D).9,46 Immunoblot analysis confirmed near-complete loss of SDHC in Sdhc-IntKO mice, whereas mice retaining one wild-type allele (Sdhc-IntHet) showed an ~50% reduction in Sdhc levels with no histological changes in the intestinal epithelium (Figures 7E and S7G). Similar to in vitro findings, Sdhc-deficient mice showed signs of GI toxicity when fed a high 1,000 ppm iron-containing diet (Figures 7F and 7G). Sdhc-deficient mice on a high-iron diet for 1 week lost up to 25% of total body weight with significant damage to the epithelial lining and accumulation of lipid peroxides as assessed by 4-hydroxynonenal staining in both the colon and small intestine (Figures S7H–S7J). Because complete Sdhc loss in the intestine is predicted to compromise epithelial viability and confound tumorigenesis studies, we focused on the heterozygous model to test whether partial loss of Sdhc is sufficient to sensitize tumors to iron-induced stress.46 To this end, mice were treated with azoxymethane (AOM) and subjected to three cycles of dextran sodium sulfate (DSS) to induce CAC. After completion of the three DSS cycles, at a stage when tumors are reliably established, tamoxifen was administered to delete one copy of Sdhc (Figure 7H). Strikingly, Sdhc-IntHet mice developed markedly fewer tumors than their littermate controls, accompanied by increased lipid ROS, overall ROS accumulation, and accumulation of 4-hydroxynonenal (Figures 7I–7K and S7K–S7M). Collectively, these findings define a previously unrecognized heme-SDHC axis that regulates iron-induced toxicity in vivo models (Figure 7L). DISCUSSION CRC cells are highly dependent on iron to support proliferation by fueling nucleotide synthesis and mitochondrial respiration. Yet excess iron generates ROS and lipid peroxidation, which can trigger ferroptotic cell death. How CRCs thrive under these iron-rich conditions, which are cytotoxic to most other cell types, has remained poorly understood. While prior studies have largely focused on iron import, storage, and export via transferrin receptor, ferroportin, and ferritin, little is known about the cellular mechanisms that actively buffer high intracellular iron.5,8 Our study identifies a previously unrecognized heme-SDH-CoQ axis that mediates this protective function. Cellular iron availability sustains SDHC protein abundance, which in turn supports CoQ biosynthesis, which together contribute to lipophilic antioxidant defense, neutralizing iron-induced ROS and enabling CRC cells to exploit iron-rich environments for growth. Iron-dependent lipid ROS-driven cell death, or ferroptosis, has been extensively characterized in vitro. Canonical regulators such as SLC7A11 and GPX4 were largely discovered through chemical genetic screens and through synthetic compounds.47,48 However, whether these pathways explain iron toxicity in physiological contexts remains unclear. In CRC, although canonical ferroptosis regulators are transcriptionally upregulated, deletion of these genes fails to sensitize cells to iron or alter tumor progression. This suggests that alternative, physiologically relevant protective mechanisms exist. At the same time, several features of our data remain consistent with ferroptotic stress. Inhibition of the heme-SDH-CoQ axis increases lipid ROS, and the resulting iron sensitivity can be mitigated by the ferroptosis inhibitor liproxstatin-1 both in vitro and in vivo. Nevertheless, additional observations suggest that this pathway regulates a broader oxidative stress response. Heme supplementation confers resistance to iron and oxidants such as H2O2 and selenium, which are not canonical ferroptosis inducers, and lipid peroxidation markers detected in vivo, including 4-hydroxynonenal (4-HNE), are not specific to ferroptosis and may reflect more general oxidative injury. Similarly, the intestinal toxicity observed in Sdhc-deficient mice exposed to high dietary iron may arise from mitochondrial dysfunction or global redox imbalance in addition to ferroptotic cell death. Together, these findings suggest that the heme-SDH-CoQ pathway functions as an iron-responsive antioxidant system that intersects with ferroptosis while more broadly protecting cells from iron-driven oxidative stress. Several intrinsic (oncogenic mutations in KRAS and Tp53) and extrinsic (nutrient and oxygen availability) factors affect ferroptosis sensitivity in cancers. Mechanistically, these factors all converge to altered cellular iron homeostasis.49–52 In PDAC, KRAS activation not only increases the LIP via ferritinophagy to fuel mitochondrial metabolism via iron-sulfur cluster biogenesis but also sensitizes cells to iron toxicity and consequently ferroptosis.53,54 Interestingly, loss of iron-sulfur clusters sensitizes lung cancer cells to ferroptosis via IRP stabilization, which increases the LIP.55,56 Our work positions SDH and CoQ as critical iron-specific antioxidant defenses in CRC cells, functioning in parallel to canonical ferroptosis pathways. By demonstrating that mitochondrial CoQ metabolism directly mitigates iron-induced lipid ROS, we expand the conceptual framework of ferroptosis and iron toxicity, particularly in vivo and in intestinal tissues where canonical ferroptosis effectors are dispensable. SDH has classically been studied as a metabolic hub, integrating TCA cycle flux with ETC activity, and as a tumor suppressor when mutated in paragangliomas, GIST, and renal cancers.36,57–59 It remains unclear whether these tumor types experience significant iron accumulation. If not, SDH mutations in this context may have a limited impact on iron buffering. In such cases, the signaling and metabolic advantages conferred by SDH loss may outweigh any detrimental effects of decreased protection against iron-induced stress. By contrast, our work identifies a previously unrecognized role for SDH in safeguarding cells against iron toxicity. Notably, SDH also serves as a key regulator of CoQ redox status through its bidirectional catalytic capacity to oxidize succinate or reduce fumarate.60,61 This property underlies tissue-specific variation in SDH function. Some tissues such as liver, kidney, and brain constitutively engage more in fumarate reduction than succinate oxidation, suggesting that cancers arising in these contexts may be less protected against iron overload by CII activity.62 Indeed, both liver and kidney are highly susceptible to iron-induced oxidative damage and ferroptosis, consistent with the notion that limited SDH-dependent succinate oxidation may compromise CoQ redox buffering capacity in these tissues.63–65 Inhibition of SDHC-dependent succinate ubiquinone reductase (SQR) activity has also been shown to increase sensitivity to BCL-2 antagonist venetoclax in multiple myeloma, while heme supplementation induces resistance.66,67 Whereas targeting SDHA-dependent succinate oxidation reduces de novo purine synthesis, sensitizing cancer cells to inhibition of purine salvage pathways.68 These findings also highlight the multifaceted role of mitochondrial ETC complexes, where specific catalytic activity of the same complex can have vastly different effects. At present, the mechanisms by which iron or heme regulates SDHC remain unresolved. One possibility is that heme or iron regulates SDHC stability through mitochondrial quality-control pathways, which are tightly coupled to respiratory complex integrity.69 Another intriguing possibility is that heme binding to SDHC or associated CII subunits stabilizes the holoenzyme, thereby protecting it from proteolytic degradation. Indeed, our thermal stability assays suggest that heme binding directly stabilizes SDHC. Moreover, why this process is only activated in subset cells and how SDHC regulation leads to a net increase in the CoQ pool is also unclear. Future structural or biochemical studies will be required to determine how heme occupancy regulates SDHC stability and CII activity. Beyond its canonical role as an electron carrier in the ETC, CoQ functions as a glutathione-independent, membrane-specific antioxidant.39,70,71 Its redistribution to non-mitochondrial membranes leads to protection from iron-induced lipid peroxidation.37,38 CoQ has been described as a radical-trapping antioxidant in the membrane, where it is reduced by FSP-1 and DHODH.39,42,70 In this sense, the heme-SDH-CoQ axis may operate as a mitochondrial counterpart to the FSP1-CoQ system, with mitochondria actively supplying CoQ to the plasma membrane and other cellular compartments. This model is supported by the increased abundance of other mitochondrial CoQ-reducing enzymes under heme deficiency, where SDH is no longer functional. However, it remains unclear how CoQ is redistributed to non-mitochondrial membranes in response to iron. Our proteomic data did not detect changes in STARD7, a recently described CoQ transporter, suggesting that other factors may mediate this process.72 These findings highlight a broader principle: cancer cells co-opt redox-active cofactors, such as heme and CoQ, not only to sustain metabolism but also to adapt to oxidative stress. Collectively, these findings uncover a novel mitochondrial antioxidant defense mechanism, redefine the landscape of ferroptosis regulation in vivo, and illuminate how CRCs buffer iron toxicity. By delineating the heme-SDH-CoQ axis, our study reveals a previously unknown metabolic adaptation that underlies CRC survival in iron-rich environments and highlights potential therapeutic opportunities to exploit iron-dependent synthetic vulnerabilities. While no highly selective CII inhibitors have yet been approved as cancer therapies, targeting mitochondrial metabolism is an active area of clinical and translational research. Several mitochondria-targeted agents that disrupt electron transport, such as metformin (complex I) and other ETC modulators, are under investigation in clinical studies. Moreover, agents like lonidamine have demonstrated partial CII inhibitory activity in preclinical models, and mitochondria-directed vitamin E analogs (e.g., mitochondrially targeted vitamin E succinate) act on SDH-related sites and have shown anti-cancer effects in experimental systems. In CRC specifically, strategies that transiently impair SDHC function or synergize CII inhibition with iron modulation or ferroptosis-inducing agents could enhance tumor selectivity and overcome systemic toxicity, motivating further preclinical optimization and early clinical evaluation of CII-targeted interventions. Limitations of the study As discussed above, the precise biochemical mechanism by which heme stabilizes SDHC and why this is activated in only a subset of cancer cells remains incompletely defined. Our thermal stability assays suggest that heme binding directly stabilizes SDHC, but future studies are needed to delineate the exact molecular mechanisms underlying SDHC stability and the signaling pathways that confer this process in iron-resistant cells compared with iron-sensitive cells. Relatedly, while our data implicate iron, heme, and SDHC in regulating total and reduced CoQ levels and their accumulation at the plasma membrane under iron stress, it remains unclear whether these factors primarily influence CoQ biosynthesis, redox state, stability, or trafficking. The mechanisms governing the subcellular distribution of CoQ in mammalian cells are poorly understood, and it is possible that iron-dependent enzymes, including SDHC, contribute to the stabilization of CoQ biosynthetic complexes or putative CoQ transporters. Further studies are required to clarify these pathways and refine the model. Lastly, while translational relevance is suggested by existing mitochondria-targeted drugs and preclinical ETC modulators, the development of clinically viable, selective CII inhibitors for CRC remains an open challenge, and further preclinical optimization is necessary to evaluate dosing, tumor specificity, and potential combination therapies. RESOURCE AVAILABILITY Lead contact Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Dr. Yatrik M. Shah, who can be reached at shahy@umich.edu. Materials availability All unique and stable reagents, including mouse lines that were generated in the course of this study, are available from the lead contact with a completed materials transfer agreement. Data and code availability Data: data generated in this study are available in Tables S1, S2, S3, S4, S5, S6, S7, S8, and S9. Any raw data requested that were generated from this paper will be shared by the lead contact upon request. Data: data generated in this study are available in Tables S1, S2, S3, S4, S5, S6, S7, S8, and S9. Any raw data requested that were generated from this paper will be shared by the lead contact upon request. Code: this paper does not report any original code. Code: this paper does not report any original code. Additional information: any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request. Additional information: any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request. STAR★METHODS EXPERIMENTAL MODEL AND STUDY PARTICIPANT DETAILS Mice All mice used in this study are predominantly C557BL/6 background unless mentioned otherwise. 6–8 weeks old male and female mice were used for all the experiments and littermates were randomized in all experimental conditions. All the mice were housed at Unit for Laboratory Animal Management at University of Michigan in standard temperature controlled specific pathogen free environment with 12h light/dark cycle and had excess to standard chow and water ad libitum. Mouse lines used are Villin-ERT2Cre;Gpx4F/F.20 To generate GPX4 deficient mice with Apc, Tp53 and Kras mutation, Gpx4F/F mice were crossed CDX2-ERT2Cre;ApcF/F mice (single mutant), CDX2-ERT2Cre;ApcF/F,Tp53F/F mice (double mutant-p53), CDX2-ERT2Cre;ApcF/F,KrasG12D mice (double mutant-KRAS) and CDX2-ERT2Cre;ApcF/F, Tp53F/F,KrasG12D mice (triple mutant).21 And finally to generate intestine specific SDHC deficient mice, Villin-ERT2Cre mice(Jax strain 020282) were crossed with SdhcF/F mice.46 All animal studies were performed in accordance with the Association of Assessment and Accreditation of Laboratory Animal Care International guidelines, approved by the University Committee on the Use and Care of Animals, University of Michigan. Tamoxifen induction Mice were intraperitoneally administered with 100mg/kg body weight tamoxifen dissolved in corn oil every other day 3 times. Mice were used for experiments 5 days post last tamoxifen dose unless mentioned otherwise. AOM-DSS induced colitis associated cancer (CAC) model Mice were injected with 10mg/kg body weight azoxymethane (AOM) intraperitoneally on Day 0. Day 3 mice were cycled on and off with 2% dextran sodium sulfate (DSS) in drinking water for 7 days, while body weights were recorded every other day.20 5 days after the last DSS cycle, mice were induced with tamoxifen as described above and euthanized 4 weeks after last DSS cycle, tissues were either flash frozen, fixed and Swiss rolled or processed live for molecular measurements, histological analysis for ROS measurements respectively. Syngeneic and Xenograft studies Wild type BalbC mice were anesthetized with 2–2.5% isoflurane and 2 million CT26 cells stable (Scr, SDHC-sh2/3) were implanted into the lower flanks subcutaneously. Mice were then placed on standard or doxycycline containing chow 7 days after the implantation after the tumors were palpable. Mice were euthanized roughly 2 weeks after or the tumors reached 2cm3 in size, the humane end point.26 Tumor volume/weight was measured, and the tissues were either flash frozen, fixed and Swiss rolled or processed live for molecular measurements, histological analysis for ROS measurements respectively. High Iron Diet For high iron studies mice were given tamoxifen to induce Cre-mediated recombination as described above. After administering the last tamoxifen dose, mice were fed with standard diet chow or high iron containing chow i.e., AIN-93G Purified Rodent Diet chow with 350 ppm Fe (Dyets, 115180) or 1000ppm Fe (Dyets, 115842) added from Ferric Citrate respectively for 7 days. Body weight was recorded every day and % body weight change was calculated i.e. (body weight on day 7/body weight on day 1) *100. Human Samples Paraffin embedded blocks of de-identified human adult colon biopsies were collected with approval from the University of Michigan Institutional Review Board. 10uM sections were cut from the blocks and uses for Immunohistochemistry (IHC). Deidentified human CRC enteroids and matched control enteroids were established and cultured as previously described.20 Cell lines Cancer cell lines HCT116, SW480, DLD1, KM12, RKO-1, HT29, MC38, CT26, CD-1, NIH3T3, HT1080, DANG, ASPC, UOS2 were obtained from ATCC and maintained at 37°C, 5% CO2, 21% O2 humidified environment in DMEM supplemented with 10% heat inactivated fetal bovine serum (FBS) and 1% antibiotic/antimycotic agent). METHOD DETAILS Clonogenic Assay 250–500 cells were seeded in a 12 well plate in biological triplicates with 1mL media. The cells were treated as described in the legends for Figures 2C, 2G, 3C, 3D, 5B, S2H, S3D, S3F, S4, S5B–S5D, and S5F). 2 weeks after the treatment, cells were fixed in 10% buffered formalin and stained with 0.5% crystal violet, 20% methanol solution. Colonies were then blindly counted using Cytation 5 Imaging Multi-Mode reader. Cell Growth Assays Cell growth was measured using live cell imaging by the Cytation 5 Imaging Multi-Mode reader with attached BioSpa. 1000 cells/well were seeded in 96 well clear flat bottom plates and treated 24h later in biological triplicates or quadruplicate as described in the legends for Figures 2A, 2B, 3H, S2B, and S4C. Images were taken every 24h and normalized to the baseline values counted from images taken right before the treatments (0h). Relative growth was plotted using Graphpad Prism software. Cell Death Assay Cell death was measured using SYTOX Green Nucleic Acid stain (20nM, Invitrogen #S7020). 3500–5000 cells / well were seeded in 96 well clear flat bottom plates and treated 24h later in biological triplicates or quadruplicate as described in the legends for Figures 2D, 2E, 3I, 4J, 4K, 5J–5L, 6J, S2C, S2D, and S6K. Imaged were taken every 8h using Cytation 5 Imaging Multi-Mode reader with attached BioSpa and % cell death was calculated by dividing number SYTOX+ cells by total number of cells. Histology Colonic and tumor tissues were rolled and fixed in 10% buffered formalin overnight, dehydrated in 70% ethanol and embedded in paraffin. 5μM thick sections were then stained for H&E and mounted with Permount Mounting Medium (Thermo Fisher Scientific). The slides were imaged using Leica Aperio Slide Scanner and representative images are shown at 10x or 20x magnification. Histological scoring was performed evaluating inflammation severity (0–3), depth of injury (0–3), and crypt damage (0–4). Each parameter was multiplied by the percentage of tissue involvement (1, 0–25%; 2, 26–50%; 3, 51–75%; 4, 76–100%), and the values were summed to yield a total injury score (maximum 40). SDHC and 4HNE staining and quantification For SDHC staining, 5μM thick sections obtained from the paraffin blocks containing human CRC tissues were rehydrated and antigen retrieval was done using 10mM sodium citrate buffer. Followed by blocking with 5% goat serum in PBS and sections were then probed with primary antibody against SDHC (Proteintech, 14575-AP) or 4HNE (R&D Systems, MAB3249-SP) diluted at 1:500 in the blocking solution. After 3 washes with PBS, slides were incubated HRP-conjugated anti-rabbit IgG at 1:500 dilution in the blocking solution for 1h. Sections were washed with PBS and incubated with DAB substrate and monitored for the color reaction (~5minutes). The sections were then counter staining with hematoxylin, dehydrated and mounted with Permount mounting medium. The slides were imaged using Leica Aperio Slide Scanner and representative images are shown at 10x or 20x magnification. Qupath software was used for quantification and cells were categorized as SDHC negative, low, medium, or high based on mean cell DAB OD and represented as % of the total positive cells in Figure 6H. The annotated images show SDHC negative, low, medium, and high cells as blue, yellow, orange, and red respectively (Figure 6G). Western Blotting Protein lysates were prepared by lysing the cells or tissues in RIPA buffer supplemented with protease and phosphatase inhibitors. Lysates were then mixed appropriate volume of the 5X SDS loading buffer, boiled at 95°C for 5 minutes, allowed to cool to RT and centrifuged at 15,000g for 5 minutes. After lysis solubilized proteins were resolved on 12% SDS-polyacrylamide gels and transferred to PVDF membrane, blocked with 5% milk in TBST and probed overnight with indicated primary antibodies diluted at 1:1000 in the blocking buffer for GPX4 (Proteintech, 67763–1-Ig), FTH1 (Cell Signaling, 3998S), UROD (Santa Cruz, SC-365297), FECH (Proteintech, 14466–1-AP), SDHA (Cell Signaling, 11998), SDHB (Proteintech, 10620–1-AP), SDHC (Proteintech, 14575-AP), β-actin (Proteintech 66009–1-Ig). Protein bands were then visualized by electrochemical luminescence using HRP conjugated secondary antibody against mouse (Cell Signaling, 7076S) or rabbit (Cell Signaling 7074S). Real Time Quantitative PCR RNA was isolated from cells or tissues using TRIzol chloroform method. RNA was reverse transcribed using MMLV reverse transcriptase using primers listed in the Table S8. Quantitative PCR was done with Radiant Green qPCR Master Mix (Alkali Scientific Inc.) and primers listed in Table S8. Quantification cycle values were normalized to β-actin and expressed and log2 fold change. CRISPR screen and Analysis Metabolic CRISPR library The human CRISPR metabolic gene knockout library was a gift from David Sabatini containing 29,790 sgRNAs targeting 2981 human metabolic genes. A total of 75 × 106 cells were seeded at 0.5×106 cells/mL in 6 well plates and transfected with the CRISPR library virus through spinfection at 1200g for 1h in presence of 10μM polybrene. 24h after spinfection the virus was removed by changing the media and after additional 24h, 1μM/mL puromycin was added transfected cells were selected for 3 days. About 30 million cells were pelleted and saved as doubling 0 population for normalization purposes described later. 30 million cells/ condition were then treated with 2.5mM FAC, 100μM FG4592 or vehicle control and passaged every 3–4 days until 14 doublings. Genomic DNA was isolated from 15 million cells from doubling 0 control and doubling 14 FSC, FG4592 and vehicle treated cells using DNeasy kit (Qiagen). sgRNA inserts were PCR amplified using Takara Ex Taq polymerase, purified, and sequenced using NextSeq Illumina. The sequence reads were mapped to the sgRNA reference library and abundance of sgRNA was quantified (normalized to doubling 0 abundances). For gene level differential score the log2 fold change abundance between vehicle, FAC and FG4592 for all sgRNA targeting the same gene was averaged. Finally, the gene score for each condition was calculated as median log2 fold change in abundance between FAC or FG492 and vehicle treated population (Table S1). Ferroptosis CRISPR library The custom CRISPR library targeting genes involved in ferroptosis was generated based on an integrated ferroptosis-CRISPR analysis as described previously.74 A total of 6 × 106 HCT116 cells expressing Cas9 were seeded at 3 × 106 cells per 15cm plates and transfected with the CRISPR library virus in presence of polybrene. 48h after transfection virus was removed and after additional 24h transfected cells were selected with 1μM/mL puromycin for 3 days. 3 × 106 cells were saved as doubling 0 for normalization purposes and 0.5 × 106 cells per condition were treated with vehicle control, RSL3 (1μM), TTFA (50μM), IDB (4μM), FAC (0.5mM) and FG4592 (100 μM). Cells were maintained for upto 14 doublings followed by DNA purification, PCR amplification and sequenced as described above. Differential gene scores were calculated by taking the difference between the treatment condition and corresponding vehicle control (cells. Differential scores were then standardized within a condition by subtracting the mean and dividing by the standard deviation (Table S4). Generation of shRNA/sgRNA stable cell lines shRNA targeting UROD, FECH and SDHC were cloned into pLKO.1-Tet on plasmid to generate doxycycline inducible knock down. While shRNA targeting SDHA and SDHB were cloned into pLKO.1 plasmid to generate constitutively active knockdown cells. Corresponding empty vectors were used as controls. shRNA sequences were sequence validated and packaged into lentivirus by University of Michigan vector core. Cells were transfected through spinfection at 900g for 1h and selected for 2 weeks in puromycin (2μM/mL). All clones were validated by measuring protein expression 24, 48 and 72h post doxycycline (250ng/mL) induction. Cell ROS and Lipid ROS measurement For cells, 1million cells/well were seeded in 6 well plates and 24h later were treated as indicated in the legends of Figures 3F, 3G, 5E, 5F, and S3G for 72h and 24h respectively. Cells were harvested using PBS-EDTA (5mM) buffer, washed with HBSS and stained for lipid ROS or cell ROS with 5μM C11-BODIPY (Thermo Fisher, D3861) and 10μM cell permeable free radical sensor carboxy-H2DCFDA (Thermo Fisher, C10444) for 30minutes at 37°C respectively. Cells were pelleted, washed, and resuspended in FACS buffer (1x PBS, 2%BSA, 1mM EDTA) for analysis on Cytek aurora spectrum flow analyzer. Fluorescence intensity was measured in the FITC channel for a minimum of 20,000 cells, data analyzed using FlowJo software and values expressed as percentage of positive cells.26 For tumor and colon tissue: Middle region of the colon or tumors were collected from mice as indicated in the legend of Figures 7C, 7K, S7E, and S7M. The tissues were digested with 1mg/mL collagenase type IV dissolved in RPMI for 30 minutes at 37°C. Cell suspension was then filtered, collected, washed, and resuspended in FACS buffer (1x PBS, 2%BSA, 1mM EDTA). A single cell suspension was stained with 7-AAD (1:200; BD Pharmingen, catalog number 559925) and APC anti-mouse CD326 (EpCAM) antibody (1:200; BioLegend, catalog number 118214) for 30 min on ice. Subsequently, the single suspension was stained with 5μM C11-BODIPY (Thermo Fisher, D3861) and 10μM cell permeable free radical sensor carboxy-H2DCFDA (Thermo Fisher, C10444) for 30minutes at 37°C respectively. On Cytek aurora spectrum flow analyzer, forward and side scatter was used to identify single cells while 7AAD negative, APC positive cells were identified as live epithelial or tumor cells. Fluorescence intensity for C11-BODIPY or carboxy-H2DCFDA was measured in the FITC channel for previously identified live cell population.20 The data was collected for a minimum of 20,000 cells, analyzed using the FlowJo software and values expressed as percentage of positive cells. Labile Iron measurement For cells, 1million cells/well were seeded in 6 well plates and 24h later were treated as indicated in the legends of Figure 2E. Cells were harvested using PBS-EDTA (5mM) buffer, washed with HBSS and stained for labile iron with 5μM BioTracker FerroOrange Live Cell Dye (Millipore Sigma SCT210-175nmol) for 30minutes at 37°C respectively. Cells were pelleted, washed, and resuspended in FACS buffer (1x PBS, 2%BSA, 1mM EDTA) for analysis on Cytek aurora spectrum flow analyzer. Fluorescence intensity was measured using absorbance at 542nm and emission at 572nm for a minimum of 20,000 cells, data analyzed using FlowJo software and values expressed as percentage of positive cells.75 Cell death measurement Subcutaneous tumors were collected from mice as indicated in the legend of Figure 7A. The tissues were digested with 1mg/mL collagenase type IV dissolved in RPMI for 30 minutes at 37°C. Cell suspension was then filtered, collected, washed, and resuspended in FACS buffer (1x PBS, 2%BSA, 1mM EDTA). A single cell suspension was stained with 2μM Propidium Iodide for 30 min on ice. On Cytek aurora spectrum flow analyzer, forward and side scatter was used to identify single cells APC positive cells were identified as live epithelial or tumor cells. PI positive dead cells measured using absorbance at 532nm and emission at 620nm for a minimum of 20,000 cells, data analyzed using FlowJo software and values expressed as percentage of positive cells (Figure S7F). Metabolomics HCT116 cells were seeded at 1 million cells/well seeding density and treated with FAC (100μM), Succinyl Acetone (500μM), heme (10μM) and TTFA (50μM) for 72h, 24h post seeding as described in Figures 4A–4D and S6M. HCT116 cells stable for SDHC targeting shRNA treated with doxycycline (250ng/mL) for 72h as described in Figure S6M. Metabolites were extracted in 80% methanol, lyophilized, and reconstituted in 50:50 methanol water mixture, normalized based on protein concentrations obtained from a well processed in parallel. Samples were then analyzed on Agilent 1290 Infinity II Bio LC ultra-high performance liquid chromatography system paired with an Agilent 6470 QQQ.76 Data was processed and analyzed via MassHunter Quantitative Analysis v10. Metabolite abundance was median centered across samples and statistical significance was determined by one-way ANOVA with a significance threshold of 0.05 (Tables S2 and S7). 13C5 L- Glutamine Labeling HCT116 cells were seeded at 1million cells/ well in a 6 well plate and treated 24h later with 0.5mM FAC or 150μM DFO for 24h. For stable isotope tracing, glutamine free DMEM (Thermo Fisher) was supplemented with 2mM 13C5-L glutamine (Sigma), 1mM pyruvate, 10% FBS, and 1% antibiotic/antimycotic solution. A parallel plate was treated similarly except it was supplemented with regular 12C5-L glutamine for normalization purposes. The cells were labeled for 8h and metabolites were extracted as described above. Samples were run on a Thermo Scientific IQX OrbiTrap LCMS system using a Waters Acquity UPLC BEH amide 1.7 um 2.1 × 100 mm column with a Waters UPLC BEH Amide 1.7 um VanGuard Pre-Column column of amide for high throughput acquisitions. Solvent A consists of 10 mM ammonium acetate in water with 10 mM Ammonia pH 9.2. Solvent B consists of acetonitrile. Pump Seal wash and autosampler wash is 50% isopropanol with 0.1% formic acid. The LC profile is 0.2 ml/min of 85% B at 0–0.5 min: 0.5–15 min, 15%B; 15–17 min, 15% B: 17.1 min, 85%B; 22 min, 85% B. The column compartment temperature is maintained at 30 °C and the autosampler is at 5 °C. The injection volume is 3 μl. Data was analyzed using Compound Discoverer v3.4 (Table S3). Heme measurement Total cellular heme was measured using fluorometric assay that measures protoporphyrin IX fluorescence upon release of free iron from heme as previously described.77 1×106 cells were seeded in 6 well plate and 24h later treated with FAC, Heme and Succinyl acetone for 72h as described in the legend for Figure 3B. Cells were washed with PBS, harvested, and resuspended in 250μL 20mM oxalic acid and incubated overnight at 4°C in dark. Next day, 250uL of supersaturated prewarmed 2M oxalic acid was added to the cells to achieve a final 1M oxalic acid concentration. The samples were then split in half and boiled at 100°C for 1h while the other half was kept at room temperature for background normalization. The samples were centrifugated at full speed for 5inutes at room temperature, 50uL of the supernatant was transferred to 96 well black wall, clear bottom plates and fluorescence was measured at 400nM excitation and 604nM emission. The fluorescence of unboiled sampled representing background was subtracted from the fluorescence of boiled sample and the heme concentration was calculated using a standard curve of hemin chloride as μM. Oxygen Consumption Rate Cells were seeded (5000 cells/ well) in a 96 clear flat bottom plate and treated 24h later as described in the legend of Figures 3J, 4G, 4H, and S3H. Plate was then interfaced with Resipher system by placing the Resipher sensors over the plate as per manufacturer’s instructions for live, continuous monitoring of oxygen consumption. OCR values computed by the Lucid Lab software were normalized to cell numbers. Proteomics Mitochondria were isolated using 3 million cells were seeded in a 15cm cell culture plate in 25mL media and treated with 0.5mM FAC, 0.5mM SA for 72h and 50μM TFFA for 24h in biological triplicates. Cells were washed, harvested in ice cold PBS and resuspended in homogenization buffer (containing 200mM sucrose, 10mM TrisCl pH7.5, 10mM EDTA and protease inhibitor cocktail). Cells were then homogenized with 30–40 strokes in the dounce homogenizer. Homogenate was the centrifuged twice at 1000g for 10minutes at 4°C to removed nuclear pellet and the mitochondrial pellet was obtained by centrifuging the supernatant at 9000g for 20 minutes at 4°C. The obtained mitochondrial pellet was washed with the homogenization buffer twice to remove any cellular debris and nuclear contaminants. 100ug of mitochondrial proteins were then digested into peptides with trypsin, labeled with tandem mass tag labeling kit (ThermoFisher) according to manufacturer’s instructions, analyzed on mass spectrometer by the University of Michigan proteomics core. The values are represented as log2 fold change in abundance compared vehicle treated group (Table S5). Co-enzymeQ10 measurement For whole cell CoQ measurements, 1.5million cells were seeded in 10cm cell culture plates in 20mL media in biological triplicates, 24h later they were treated as indicated in the legends of Figures 5D, 5I, 6K, S5G–S5I, and S6L. Cells were then washed, harvested, and pelleted in ice cold PBS in pre-weighed tubes and wet weight of cell pellets was recorded. Cell pellets were then rapidly resuspended in n-propanol for CoQ extraction, where volume of n-propanol was normalized based on cell pellet weight. Samples were allowed to stand at room temperature and centrifuged at 12,000g for 5 minutes at 10°C to avoid CoQ precipitation and stored at −80°C. HPLC analysis for CoQ measurement was performed on a Hypersil ODS column (150 × 4.6mm, 3μm bead size, thermo Fischer) as described previously described.78 The samples were run twice, first to measured oxidized CoQ, followed by oxidation of reduced CoQ by p-benzoquinone to measure total CoQ. Reduced CoQ was calculated by total CoQ – Oxidized CoQ. For plasma membrane CoQ levels, 3 million cells were seeded in 15cm cell culture plate in 25mL media, treated 24h later and homogenized as described above for mitochondrial extraction. The remaining supernatant after pelleting mitochondrial content was centrifuges at 100,000g for 1h at 4°C to get plasma membrane pellet. Wet weight of the plasma membrane pellet was determined, samples were prepared and analyzed as described above. BrdU staining 25,000 cells were seeded on glass coverslips coated with poly D-Lysine (Sigma Aldrich P7280–5MG). Cells were treated with FAC and TTFA as described in legend Figure S5J. Cells were labeled with 10μM BrdU (Cayman Chemicals 15580) for 2h and fixed and permeabilized with 3.7% formaldehyde and 0.1% Triton X for 20 minutes at RT respectively. DNA was denatured with 1N HCl for 10 minutes on ice followed by 2N HCl at RT and neutralized with the phosphate/citric acid buffer. Cells were then stained with anti-BrdU-FITC conjugated antibody (ebioscience 11-6071-42) overnight at RT, counter stained with nuclear stain Hoechst 33342 (Invitrogen H3570), mounted and images using DAPI and GFP channels. Percent BrdU staining was calculated using imageJ and normalized using nuclear staining. Aconitase Assay Cells were seeded (10,000 cells/ well) in a 96 clear flat bottom plate and treated 24h later as described in the legend of Figure S3I. The Assay was performed as described by the kit manufacturer (Cayman chemical technology, 705502). Temperature dependent solubility assay 3 million cells were seeded in a 15cm cell culture dish and cells and treated with SA (500μM) for 72h and heme (10μM) for 24h (as described in the legend of Figure S6J). Cells were washed and gently scraped in 10mL PBS collected and centrifuged at 600g for 15 minutes at 4°C to pellet the cells. To obtain the crude extract, cell pellet was resuspended in 1.5mL of 0.5% Triton X diluted in PBS. Cell debris were pelleted by spinning the tubes at 2000g for 10 minutes and the supernatants were quantified using BCA assay, aliquoted into 12 PCR tubes and heated at their designated temperature for 3 minutes. To obtain the soluble fraction, the samples were centrifuged at 15,000g for 1.5h at 4°C.45 For western blotting 40μL of the supernatant was mixed with 10μL of the 5x SDS loading buffer, boiled at 95°C for 5 minutes and resolved as described previous under the western blot. QUANTIFICATION AND STATISTICAL ANALYSIS In vitro experiments were validated in multiple cell lines. Each condition of each cell line experiment was performed in technical replicates and repeated three times to ensure reproducibility. Figures show a representative biological replicate unless otherwise indicated. Blinding was performed whenever appropriate. Sample description and identification was unavailable to the core personnel during data collection and analysis. Statistical details of all experiments can be found in the figure legends. The sample numbers are mentioned in each figure legend and denote biological replicates. Statistical details are reported in the figure legends. Results are expressed as the mean +/− standard error of the mean for all figures unless otherwise noted. Significance between 2 groups was tested using a 2 tailed unpaired t test. Significance among multiple groups was tested using a one-way ANOVA. For all ANOVAs the Tukey-Kramer test was used for posthoc multiple comparison, as this allows for the possibility of unequal sample sizes and takes into account the scatter of both groups. GraphPad Prism 10.0 was used for the statistical analysis. Statistical significance is described in the figure legends as: * p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001. Supplementary Material Supplemental information can be found online at https://doi.org/10.1016/j.cmet.2026.04.020.