The October 29th issue of Science has an interesting article from a team at Stanford on a possible approach for Alzheimer's therapy. The dominant Alzheimer's hypothesis, as everyone will probably have heard, is that the aggregation of amyloid protein into plaques in the brain is the driving force of the disease. There's some well-thought-out dissent from that view, but there's a lot of evidence on its side, too.

So you'd figure that keeping the amyloid from clumping up would be a good way to treat Alzheimer's, and in theory you'd be correct. In practice, though, amyloid is extremely prone to aggregation - you could pick a lot of easier protein-protein interactions to try to disrupt, for sure. And protein-protein targets are tough ones to work on in general, because it's so hard to find a reasonable-sized molecule that can disrupt them. It's been done, in a few well-publicized cases, but it's still a long shot. Proteins are just too big, and in most cases so are the surfaces that they're interacting with.

The Stanford team tried a useful bounce-shot approach. Instead of keeping the amyloid strands off each other directly, they found a molecule that will cause another unrelated protein to stick to them. This damps down the tendency of the amyloid to self-aggregate. The way they did this was, by medicinal chemistry standards, simplicity itself. There's a well-known dye, the exotically named Congo Red, that stains amyloid very powerfully - which must mean that it has a strong molecular interaction with the protein. They took the dye structure and attached a spacer group coming off one end of it, and at the other end they put a synthetic ligand which is known to have high affinity for the FK506 binding protein (FKBP). That one is expressed in just about all cell types, and there are a number of small molecules that are known to bind to it.

The hybrid molecule does just what you'd expect: the Congo Red end of it sticks to amyloid, and the other end sticks to FKBP, which brings the two proteins together. And this does indeed seem to inhibit amyloid's powerful tendency for self-aggregation. And what's more the aggregates that do form appear to be less toxic when cells are exposed to them. It's a fine result, although I'd caution the folks involved not to expect things to make this much sense very often. That stich-em-together technique works sometimes, but it's not a sure thing.

So. . .(and you knew that there was going to be a paragraph like this one coming). . .do we have a drug here? The authors suggest that "Analogs based on (this) model may have potential as therapeutics for Alzheimer's disease." I hate to say it, but I'd be very surprised if that were true. All the work in this paper was done in vitro, and it's a big leap into an animal. For one thing, I'm about ready to eat my own socks if this hybrid compound can cross the blood-brain barrier. Actually, I'm about ready to sit down for a plateful of hosiery if the compound even shows reasonable blood levels after oral dosing.

It's just too huge. Congo Red isn't a particularly small molecule, and by the time you add the linking group and the FKBP ligand end, the hybrid is a real whopper - two or three times the size of a reasonable drug candidate. The dye part of the structure has some very polar sulfonate groups on it, as many dyes do, and they're vital to the amyloid binding. But they're just the sort of thing you want to avoid when you need to get a compound into the brain. No, if this structure came up in a random screen in the drug industry, we'd have to be pretty desperate to use it as a starting point.

Science's commentary on the paper quotes a molecular biologist as saying that this approach shows how ". . .a small drug becomes a large drug that can push away the protein. . ." But that's wrong. You can tell he's from a university, just by that statement. I'm not trying to be offensive about it, but neither Congo Red nor the new hybrid molecule are drugs. Drugs are effective against a disease, and this molecule isn't going to work against Alzheimer's unless it's administered with a drill press. If that's a drug, then I must have single-handedly made a thousand of them. The distance between this thing and a drug is a good illustration of the distance between academia and industry.

To be fair, this general approach could have value against other protein-protein interaction targets. I think that it's worth pursuing. But I'd attack something other than a CNS disease, and I'd pick some other molecule than Congo Red as a starting point.