A decade of ferroptosis research used chemical induction to study the pathway. The chemicals targeted GPX4, a lipid repair enzyme. Inhibit GPX4, lipid peroxides accumulate, the cell membrane collapses — ferroptosis. This worked in culture. In animal models, GPX4 inhibition proved lethal: too toxic for therapy, killing healthy cells alongside cancerous ones.
Columbia researchers found the natural ferroptosis pathway. It doesn't use GPX4. It uses GPX1, through the GPX1-OSBPL8 axis — a completely different enzyme that is dispensable under normal conditions but essential when reactive oxygen species are high. Cancer cells live in exactly that high-ROS state. GPX1 inhibitors would selectively kill cancer cells while leaving healthy tissue alone.
The through-claim: the method of induction replaced the mechanism it was meant to reveal. Chemical induction of ferroptosis bypassed the natural sensing cascade and activated a different enzyme entirely. The researchers weren't studying the cell's ferroptosis pathway — they were studying a chemical's ferroptosis pathway that happened to share the endpoint (cell death via lipid peroxidation).
The artificial and natural pathways converge at the same phenotype — membrane collapse — but diverge at the target. Ten years of drug development aimed at GPX4 because the tool said GPX4 was the target. The tool was answering a different question than the one being asked: not “how does the cell trigger ferroptosis?” but “how does this chemical trigger ferroptosis?”
The broader pattern: when you use an artificial perturbation to study a natural process, you risk discovering the perturbation's mechanism rather than the cell's. The endpoint looks the same. The pathway is entirely different. And the therapeutic implications — selective vs. toxic — diverge at exactly the point the methodology obscured.