Ferritin loss exposes an iron-toxicity vulnerability in medulloblastoma

What the work claims

Segui, Durivault, Pagnuzzi, Vial and colleagues argue that ferritin is not merely a storage depot for excess iron in medulloblastoma (MB); it sets the upper limit of iron tolerance.¹ Deleting the ferritin heavy chain (FTH1) is tolerated in standard two-dimensional culture because MB cells remodel iron uptake and light-chain ferritin expression. Under iron stress, however, the same deletion exposes a vulnerability that is stronger than loss of the canonical ferroptosis defender xCT (SLC7A11). Pharmacological vitamin C (VitC) can bypass genetic deletion: by reducing Fe³⁺ to Fe²⁺, high-dose VitC expands the labile iron pool and triggers rapid, iron-dependent, but non-ferroptotic death in MB cells and in MB-like tumor organoids.

How it works

Ferritin is a 24-subunit nanocage whose heavy chain oxidizes Fe²⁺ to Fe³⁺ for safe sequestration. The authors used CRISPR-Cas9 to knock out FTH1 in two MB cell lines: DAOY, a sonic-hedgehog (SHH)-subgroup line with a TP53 mutation, and HD-MB03, a Group 3 line with c-Myc amplification. Two independent clones per line were validated by immunoblotting and Sanger sequencing.

Under basal conditions, FTH knockout did not change viability, ferrous iron levels, reactive oxygen species, or lipid hydroperoxides in either line. The cells compensated: DAOY upregulated ferritin light chain (FTL), iron regulatory protein 2 (IRP2), GPX4, and ferroportin, while HD-MB03 chiefly upregulated FTL. This suggests that MB cells can reroute iron homeostasis, but the rerouting is subgroup-specific.

The phenotype sharpened under stress. In clonogenic assays with the iron donor ferric ammonium citrate (FAC), FTH knockout clones from both lines formed far fewer colonies. In DAOY the defect was fully rescued by the ferroptosis inhibitor ferrostatin-1; in HD-MB03 it was only partly rescued, pointing to additional death mechanisms. Consistent with this, FTH knockout cells were also more sensitive to the ferroptosis inducer erastin, and this sensitivity was reversed by ferrostatin-1.

The in vivo result was even more pronounced. Orthotopic xenografts of DAOY FTH knockout cells into the cerebellum of nude mice produced roughly three-fold longer survival than wild-type DAOY tumors (n = 10 per group). HD-MB03 FTH knockout showed only a non-significant trend, again indicating subgroup dependence. Notably, xCT knockout had a weaker anti-tumor effect than FTH knockout despite a far stronger ferroptotic phenotype in vitro, implying that ferritin-mediated iron buffering matters more than canonical cystine defense for MB tumor fitness in vivo.

For the organoid work, the authors used the SCTi003-A human iPSC line to generate cerebral organoids, then electroporated them at day 35 with PiggyBac vectors expressing OTX2, c-Myc, and an mVenus reporter to drive Group 3 MB-like transformation. Tumors appeared by 29 days post-electroporation. Treatment with 1 mM VitC for 10 days followed by 10 mM VitC for another 10 days eliminated mVenus-positive tumor tissue; 10 mM VitC alone, started after tumors were established, also eradicated them within 10 days. Tumor-organoid cells (TOCs) detached from these organoids remained highly sensitive to 10 mM VitC, showing about 40 percent increased cell death within 4 hours. The death was iron-dependent (deferoxamine rescued it) but not canonical ferroptosis: lipid hydroperoxides did not rise, ROS fell rather than rose, and ferrostatin-1, necrostatin-1, Q-VD-OPh, bafilomycin A1, and N-acetylcysteine all failed to rescue viability.

The authors also tied sensitivity to cell phenotype. DAOY cells are mesenchymal-like (high N-cadherin, vimentin, Twist, ZEB1), whereas HD-MB03 cells are more epithelial-like. Mesenchymal-like states correlated with greater iron toxicity across public MB transcriptomic data, and culturing DAOY cells in 3D to induce a more epithelial phenotype reduced VitC sensitivity. Thus, ferritin buffering capacity and cellular phenotype together determine how much iron stress a tumor cell can tolerate.

Where a skeptic should push

The central claim rests on two cell lines. DAOY and HD-MB03 are established lines and may not reflect the full heterogeneity of SHH or Group 3 MB; neither is patient-derived. The orthotopic experiments used immunodeficient mice, and the dramatic survival benefit for DAOY FTH knockout did not cleanly replicate in HD-MB03, so generalizing across MB subgroups is premature.

The vitamin C pharmacology is also model-dependent. In the orthotopic model, systemic VitC at 3 g/kg twice daily for two weeks had no significant survival benefit, which the authors attribute to poor blood-brain-barrier penetration. Mice also synthesize their own vitamin C, so their systemic antioxidant handling differs from humans. The organoid experiments used escalating 1 mM and 10 mM concentrations; while intravenous VitC can reach millimolar plasma levels in patients, intratumoral brain concentrations are uncertain.

The non-ferroptotic iron death pathway is descriptive rather than mechanistically resolved. The authors show that the hallmarks of ferroptosis are absent and that metabolic flux collapses, but the exact executioner remains unidentified. Without that, translating the finding into a druggable target beyond iron modulation is speculative.

Implications for brain-tumor organoid drug screens

For organoid-based drug discovery, the most useful lesson is that tumor organoids can reveal vulnerabilities that two-dimensional assays and even xenografts obscure. Basal culture conditions showed FTH loss as benign; only the three-dimensional organoid and in vivo stress environments exposed ferritin buffering as a genuine liability. This is a reminder that an organoid screen read out under routine conditions may miss mechanisms that appear only when the model is pushed metabolically.

The study also gives a concrete example of how to stress-test a tumor organoid: combine a genetic lesion (here FTH1 deletion or OTX2/c-Myc transformation) with a metabolic perturbation (iron overload or high-dose VitC) and look for synthetic-lethal effects. Such designs could be ported to patient-derived tumor organoid panels to ask whether ferritin-low or mesenchymal-like MB samples are selectively sensitive to iron-redox therapy.

The threat is over-interpretation. Vitamin C has a long history of being hailed as an anti-cancer agent and then failing in randomized trials. This preprint adds a mechanistic rationale tied to ferritin, but the lack of in vivo efficacy after systemic administration warns against assuming that organoid killing translates directly to patients. Any drug-discovery program built on this result would need to measure whether candidate compounds achieve sufficient intratumoral iron-redox stress, not just cell death in an organoid.

Finally, the subgroup differences matter. If Group 3 MYC-amplified MB is less dependent on ferritin than SHH MB, a ferritin-targeted therapy would need patient stratification. Organoid biobanks that retain molecular subtype annotations are the right tool to test that stratification before it reaches the clinic.

The bottom line

This preprint establishes ferritin heavy chain as a conditional dependency in MB: dispensable at rest, essential under iron stress. It identifies a distinct iron-dependent cell death program that is not classical ferroptosis, and it uses a genetically transformed cerebral organoid model to show that high-dose VitC can selectively kill MB-like tumor cells. The work is a promising starting point for patient-derived organoid screens and for designing iron-redox combination therapies, but the in vivo pharmacology and the execution mechanism of the non-ferroptotic death pathway remain open questions.