One of the more unusual drugs on the market is Tecfidera (dimethyl fumarate). I went into its history a bit in this post, if you’re wondering how a molecule that small and unfunctionalized became a multiple sclerosis drug. As that shows, it went into trials for the disease with quite a bit of clinical rationale, but no clear molecular mechanism. So how does it work?
This new paper in Science has what could be the solution. It’s long been suspected that there’s a covalent mechanism at work, since the compound is an unsaturated small methyl ester (and thus potentially reactive with nucleophiles like cysteine). The history of allergic reactions in some patients fits in with that idea as well – you’d have to imagine that if it’s reacting with Cys residues that it’s capable of reacting with quite a few of them. One suspect had been a protein called KEAP1, which (along with its partner Nrf2) is a key player in inflammation, but that alone doesn’t seem to account for the drug’s effects. But now it looks like a more mechanistically important one has been tracked down: Cys150 of the enzyme GAPDH (glyceraldehyde 3-phosphate dehydrogenase.
That one, as many will recognize, is responsible for a step in the glycolysis pathway. Why should messing with metabolism help out MS patients (and psoriasis patients as well, the drug’s original market)? Glycolysis is actually important for immune cell function: if you block it, macrophages cannot activate. The cell types affected after Tecfidera treatment are just those that are the most glycolitic (such as effector T cells), while Treg cells (among others) are left alone. Overall, the population of immune cells gets shifted around in ways that benefit autoimmune disease.
The paper demonstrates this in a number of convincing ways: if you incubate the enzyme with dimethyl or monomethyl fumarate, you can see the new mass adduct. And you see the same modified protein if you isolate GAPDH from the blood mononuclear cells of patients who have been taking Tecfidera for months as well. You get the same results from cell cultures and by dosing mice. The isolated enzyme shows dose- and time-dependent inhibition on treatment with either fumarate in vitro. Looking more closely at the cells, the compound shows its effects on glycolysis in activated macrophages and lymphocytes, but not in resting cells, as shown by several downstream markers. This treatment also abrogates their immune functions (as shown by their response to the classic LPS antigen), and you can reproduce all these effects with a known GAPDH inhibitor (heptelidic acid). So yeah, this appears to be pretty nailed down.
It’s also noted that fumarate is actually a downstream product of glycolysis, so it may be that dimethyl fumarate is just a prodrug for getting product inhibition of the pathway as well. None of this rules out other enzymes being modified by it, though – in fact, given its size and simplicity, other enzymes surely are. And the GAPDH-driven effects in other cells may also be important (it’s already been suggested that it has metabolic effects in oligodendrocytes, although the mechanism wasn’t figured out). So what we have here is a small, simple, promiscuous covalent-inhibitor molecule that inactivates a ubiquitous enzyme that’s crucial to metabolic activity. And it’s a beneficial drug with a useful therapeutic index. Go figure!
Update: here’s a recent chemical proteomics effort on dimethyl fumarate, which identified a number of potential binding sites – but not this one on GAPDH! That’s worth thinking about (from both directions) and these sorts of disconnects may turn into a future blog post. . .