Vitamin B2's Dark Side: How a Common Nutrient May Be Helping Cancer Cells Survive

In a cancer research lab in Würzburg, a small molecule built by soil bacteria has done something that dietary advice columnists would rather not publicise: it has helped kill cancer cells, in petri dishes and mice, by sabotaging a shield those cells build out of vitamin B2. The work, published in March 2026 in Nature Cell Biology, does not suggest that vitamin B2 is dangerous. It does suggest that tumours are unusually dependent on it, and that researchers now have a clean, drug-shaped way to exploit that dependence.

The shield in question is a protein called FSP1, short for ferroptosis suppressor protein 1. FSP1 was first described as a defender of cancer cells in 2019, in a study that placed it as a parallel defender alongside a better-known enzyme called GPX4 and the glutathione system [1]. The biological term for the damage that FSP1 prevents is ferroptosis, a form of cell death driven by the runaway oxidation of phospholipids in cell membranes. FSP1's job is to keep a small lipid called ubiquinone in its reduced, protective form, ubiquinol, using the cellular fuel NAD(P)H as a source of electrons. A 2023 review in Trends in Molecular Medicine mapped the broader FSP1-CoQ10-NAD(P)H axis and noted that the same pathway has been implicated in neurodegeneration and ischemia-reperfusion injury, not only in cancer [2].

For more than a decade, researchers have hunted for ways to disable FSP1 in tumours. The challenge has been that FSP1 is a stable, well-behaved protein. Standard drug-discovery screens tend to find inhibitors of enzymes with obvious pockets. FSP1 looked, on paper, hard to drug. The Würzburg team, led by PhD student Vera Skafar and professor José Pedro Friedmann Angeli at the Rudolf Virchow Centre, decided to ask a different question. Instead of looking for chemicals that bind FSP1 directly, they asked which human genes a cancer cell needs in order to keep FSP1 alive in the first place.

A vitamin with a hidden second job

The screen covered roughly 3,000 druggable genes in HT1080 fibrosarcoma cells, a line that depends heavily on FSP1 for survival. The gene that rose to the top of the list was not an obvious drug target. It encoded an enzyme called riboflavin kinase, or RFK, whose known job is to attach a phosphate group to riboflavin, the chemical name for vitamin B2, to make a cofactor called FMN. A second enzyme, FADS (also called FLAD1), then converts FMN into FAD, an energy-carrying helper molecule that FSP1 and many other enzymes need in order to do their chemistry [3].

The implication was elegant and slightly uncomfortable. To make functional FSP1, a cancer cell first has to take up riboflavin from its environment, phosphorylate it to FMN, and convert FMN to FAD. Knock out RFK or FADS, and the cell runs out of FAD. Without FAD, FSP1 protein levels collapse, the cell loses its shield, and standard GPX4-blocking drugs become lethal. In experiments, depleting RFK or FADS in the presence of the GPX4 inhibitors RSL3 or ML210 killed cancer cells far more effectively than the GPX4 inhibitors alone [3]. Adding riboflavin back to the culture medium rescued FSP1 protein levels and reduced the lipid peroxidation that drives ferroptosis.

The dietary analogue is misleading, and worth putting down early. The Recommended Daily Allowance for riboflavin in the United States is 1.3 mg per day for men, 1.1 mg for women, 1.4 mg in pregnancy, and 1.6 mg while nursing [4]. The body treats the vitamin as water-soluble. Excess is filtered by the kidneys, the urine turns a vivid yellow, and there is no established upper intake level because toxicity has not been documented at any realistic dose [5]. The Würzburg finding operates at a different scale. It concerns the intracellular metabolism of riboflavin inside tumour cells, the rate at which those cells convert the vitamin into FAD, and whether that conversion can be disrupted with a drug. It is not a reason to change a normal diet or to take, or stop taking, any supplement.

Poisoning the shield without touching the cell

If the FSP1 shield requires FAD to function, then a molecule that slips into the FAD-binding slot of FSP1 and refuses to do chemistry should in theory collapse the shield. That, in essence, is what roseoflavin appears to do. Roseoflavin is a rose-coloured antimetabolite first isolated from Streptomyces davawensis, a soil bacterium. It looks enough like riboflavin to fool the cell's salvage enzymes, so it gets phosphorylated and adenylated into a modified FAD analog. The Skafar group showed that this analog still binds FSP1 well enough to stabilise the protein, but cannot perform the redox chemistry that FSP1 needs to keep ubiquinone reduced [3,6].

The result is a Trojan horse. From the cancer cell's point of view, FSP1 is present and apparently functional. In reality, the protein has been disarmed. In HT1080 cells and in A375 melanoma cells, nanomolar (billionths of a mole per litre) concentrations of roseoflavin dramatically increased the killing power of GPX4 inhibitors, but only in cells that actually expressed FSP1. Cells that lacked FSP1 were unaffected by roseoflavin, which is the expected result if the drug's only job is to corrupt the FSP1 shield [3]. The Skafar preprint on bioRxiv includes the full supplementary methods for the roseoflavin work and confirms the basic mechanism in animal models, though the Nature Cell Biology paper is the more comprehensive record [6].

A press release from the University of Würzburg put the stakes plainly:

Vitamin B2 plays a crucial role in protecting cancer cells from ferroptosis. An inhibitor that can do this is still missing.

Skafar said in a statement accompanying the publication [7]. The first half is now a published result. The second half is the open question the lab is racing to close.

From a soil bacterium to a clinical candidate

The work is preclinical. It has been done in cell lines and in mice. No human trials have started, and there is no guarantee that any will. The funding behind the project, a Deutsche Forschungsgemeinschaft priority programme on ferroptosis and an ERC Consolidator Grant called "DeciFerr" worth roughly two million euros, takes a discovery from petri dish to first-in-human studies, not all the way through them. Independent coverage has framed the result as a "double-edged sword," a vitamin that both feeds and, in the right hands, can be made to expose the very tumour it was protecting [8]. News-Medical covered the same paper at publication [9].

There are several practical questions a clinical programme would have to answer. Roseoflavin is a natural product, and natural products can be chemically finicky. The team will need to show that a drug-like version of the molecule reaches tumour tissue at the right concentration, that it does not simply shut down FAD-dependent enzymes indiscriminately (there are many of them, in mitochondria and elsewhere), and that healthy tissues, which also need FAD, can tolerate the disruption. The paper's most reassuring detail is the dependency on FSP1: roseoflavin only worked in cells that actually had the shield. That gives a rough hint of selectivity, but selectivity in cell culture is a promise, not a guarantee.

Why this is not a story about food

The most important sentence in the press release, for any reader who has been told to fear a vitamin, is one the scientists did not have to say. Healthy cells that do not lean on FSP1 should be largely indifferent to the new strategy, and a normal dietary intake of riboflavin will neither raise nor lower the risk of any cancer anyone has studied, as far as the existing evidence goes. The NIH Office of Dietary Supplements lists sore throat, cheilosis, glossitis, and dermatitis as the established consequences of riboflavin deficiency, and notes that the body simply excretes what it does not need [5]. The Harvard T.H. Chan School of Public Health makes the same point from the other direction. The bright-yellow urine people sometimes notice after a multivitamin is just the kidneys doing their job [4].

What the Würzburg work changes is a researcher's picture of a tumour, not a reader's grocery list. A cell that is addicted to FSP1 is also, it turns out, addicted to riboflavin. That is a useful piece of information for drug design, and a slightly unsettling one for anyone who likes a tidy story about vitamins. It is, in any case, a story about what cancer cells do with their food, not about what we should do with ours.

What to watch next

Three threads are worth following. The first is medicinal chemistry. Can roseoflavin, or a close analog, be made into a compound that survives the bloodstream, enters tumour cells at useful concentrations, and tolerates long-term dosing? The second is combination design. GPX4 inhibitors have so far been hard to develop because of on-target toxicity in tissues such as the kidney. A drug that selectively disarms FSP1 would let researchers revisit GPX4 inhibition in a stratified, biomarker-driven patient population. The third is mechanism. The paper shows that FAD is what keeps FSP1 both stable and catalytically alive, which is a more general claim than the roseoflavin result alone. Other antimetabolites that target riboflavin metabolism, or that mimic FAD, may turn out to be useful in cancers beyond HT1080 and A375.

For now, the headline is restrained on purpose. A common vitamin, it turns out, is part of the machinery that lets some cancer cells refuse to die. A bacterium, of all things, has shown scientists how to break that machinery. The rest is chemistry, patience, and time.