Finding. Metastasis-derived lung-adenocarcinoma cells in a KRAS G12D, p53-knockout mouse model already carry mitochondrial dysfunction. Munteanu, Friedmann Angeli, Gruner and colleagues barcode a focused metabolic-drug library, run it in vivo, and watch artesunate come back as a selective inhibitor of metastatic seeding. The same dose leaves primary tumor growth alone. The drug throws reactive oxygen species and lipid peroxidation. Transcriptomes say the metastatic cells have rewritten redox homeostasis. Block glutathione synthesis and those cells die in large numbers. Primary tumor-derived cells stay resistant.
Why this paper matters
Most "metabolic vulnerability" papers kill the primary and hope metastasis follows. This screen was built to do the opposite: find a compound that stops seeding while the original tumor keeps growing. Artesunate is not a new chemical. It is an artemisinin antimalarial with a long redox resume. Putting it on a mitochondrial-dysfunction-plus-glutathione logic in metastatic KRAS-p53 lung adenocarcinoma is the news.
The mitochondrial framing is disciplined. The organelle defect is the predisposition that made the screen worth running. The killing mechanism, as stated, is oxidative and peroxidative stress under a glutathione budget the metastatic cells can no longer afford. That is a redox paper sitting on a mitochondrial observation, which is still enough for this corpus if you do not over-claim a Complex I inhibitor.
What they actually measured
Two related line sets from the same genetically engineered model, split by metastatic ability. An in vivo, multiplexed DNA-barcode metabolic-drug screen, then a dedicated artesunate validation on metastatic burden versus primary growth. Mechanistic layer: ROS, lipid peroxidation, redox transcriptomes, and a glutathione-synthesis interaction that is selective for the metastatic cells. The abstract does not hand you a named mitochondrial lesion beyond "dysfunction," and it does not need to for the selectivity claim.
How to read the score
Mid 80s. Metastasis-selective in vivo phenotype, already-used drug, glutathione interaction, mitochondrial starting state. Confidence is high for the mouse split. It is not a human dosing paper.
Caveats
Mouse only. Mitochondrial dysfunction is not mapped to a subunit or an mtDNA variant here. Artesunate plus glutathione blockade is combination toxicity. Lipid peroxidation can be ferroptosis-adjacent (Friedmann Angeli is on the author list); the abstract says lipid peroxidation and glutathione, not an explicit ferroptosis genetic rescue.
What to do with it
If you screen metabolic drugs against metastasis, copy the barcode-in-vivo design and the primary-versus-seeding split. If you already use artesunate or other artemisinins, look at glutathione-synthesis inhibitors as a metastatic-cell sensitizer, not as a primary-tumor recipe. Pull the redox transcriptomes before you label this ferroptosis. Do not write a human lung-cancer protocol from this brief.