TCS Daily

Let's Hope This 'Clicks'

By Derek Lowe - June 14, 2002 12:00 AM

Nobel laureate Barry Sharpless has just published some very interesting chemistry, with implications for pharmaceuticals and several other fields. It could lead to faster and cheaper ways to find drug candidates, which the industry could certainly use, given its current dry spell. And it's a good example of how (when the time finally comes) an idea can just sort of condense out of the air, looking as if it should have been there the whole time.

Sharpless and his colleagues have for several years been advocating what he calls "click" chemistry, which he named after the sound of two plastic building blocks coming together. To be a real "click" reaction, the two starting materials have to react cleanly and with no added reagents -- just mix them and make a new product.

Why is this important? Because it saves time and trouble. Most organic reactions need costly cleaning up at the end to remove the other components and unwanted side products. With pharmaceutical companies making larger and larger libraries of new compounds for screening as leads to possible new drugs, Sharpless has proposed click reactions as a faster and cleaner way to produce them.

But that's just one part of the latest work.

The second part (which at first seems unrelated to the first) is the use of the pockets and cavities of large molecules to influence the reactions of smaller ones. Again, this is important as a time saver. Many chemical reactions suffer from being slow or unselective. But if the reactants can be guided together in the right way by the presence of a larger molecule, the speed and selectivity improves dramatically. Sharpless has long been a leader in this field of improving the selectivity of chemical reactions. His use of new guiding catalysts to produce chiral molecules, products that are purely right-handed or left handed instead of a mixture of the two, is what won him his Nobel Prize. Biologically active compounds tend to be pure chiral isomers, and finding ways to make them cleanly has been a hot area for many years now.

Reactions that are normally very slow can be accelerated, too. William Mock (at the University of Illinois - Chicago) studied a classic one in the 1980s. Two classes of molecules, azides and acetylenes, were already well known to react and form a cyclic five-membered product, called a triazole. But this only worked if you heated them together for days. Mock's group showed that if you picked the right azide and acetylene partners, and ran them in the presence of a ring-shaped protein, the ring would force the two starting materials into just the right position to react. When everything lined up correctly, the reactions started working at room temperature. This was an interesting result, but was considered by most to be a laboratory curiosity. Cyclic proteins of the sort the UIC group used aren't very easy to obtain, and it wasn't obvious which particular reactions would be accelerated and which wouldn't.

And now for the insight that led to the latest work: Sharpless knew that this type of reaction was one of the "click" processes he'd advocated, but it wasn't one of the best ones, since it worked so slowly. It struck him that if you were going to use proteins to speed up the reaction, you would be better off using enzymes. And that way, you wouldn't be stuck trying to find some protein that would cause a particular triazole to be formed. Instead you'd run the process in reverse: run a whole variety of possible triazole-forming partners to find one combination that would react with a given enzyme.

The reason that makes sense has to do with the importance of enzymes and of molecules that bind to them. Enzymes exist to speed up chemical reactions in living systems, by lining up the starting molecules inside special binding pockets formed in their three-dimensional structures. Designing molecules to block enzyme binding pockets has been one of the classic medicinal chemistry techniques. Hugely successful drugs, such as the statins for lowering cholesterol, the ACE inhibitors for lowering blood pressure (and Viagra for selectively raising it!) are all inhibitors of various enzymes.

Use of the built-in binding pockets of enzymes would speed up reactions, but only for things that fit perfectly (otherwise, the two reacting partners would never get close enough.) Whatever new triazole compounds that formed could be tailor-made enzyme blockers.

This is an ingenious inversion of classic medicinal chemistry: Instead of making thousands of compounds to get one that works, you set things up so that the ones that work are the only things that get made at all. The unaccelerated reaction is so slow that essentially nothing happens unless everything's set up for success.

And that's exactly what happened in the experiment. Sharpless and his co-workers tried a system with 98 possible combinations. Only one "clicked," and it made the all-time tightest-binding inhibitor for its enzyme. That's an extremely high hit rate. Random screening methods normally could have gone through hundreds of thousands of compounds without finding a hit at all, and through many millions without finding one anywhere near this potent.

Some have argued that this means that the technique is going to be a very powerful tool for making new enzyme inhibitors, while others contend that the team got very lucky. I'd say it mainly shows that they did a very careful job of choosing their enzyme and their chemical building blocks.

That's not a criticism, really: for an idea like this, proof of principle is key. You need to use the conditions that give it the best chance to work, just to show that it can be done at all. The researchers chose acetylcholinesterase as their enzyme, a very well-studied one indeed. The size of its binding pocket is known to be rather large -- large enough so that there are separate classes of compounds already known to bind to either end of it. (In fact, other groups had tried to string these two structures together to make a double-barreled enzyme inhibitor before, although without achieving this kind of affinity.)

So they had a leg up, chemically. They took these known structural motifs and strung the reactive ends (actylenes and azides) from them, hanging off carbon chains of several lengths and in each possible variation. This was far from a random screen.

But that could be where things are headed next. The big test of this idea will be when it's used on a system that isn't so well worked out. What that will mean is, first, you'll probably have to try a lot more than 98 variations when you don't have as good an idea of what fits into your enzyme. There are some possible pitfalls, too: if the enzyme's binding pocket isn't very large, then you run the risk of both your starting materials binding (just like you wanted,) clicking to form the triazole, just like you wanted ... but then having the new molecule float right out of the binding pocket because its shape has changed too much to fit inside any more.

That leads to a worst-case situation, one that would make this technique of little use for the pharmaceutical industry. If the method only works for proteins with really large binding pockets, it drastically cuts down the number of systems it'll be good for. It also means that the molecules that you make could well be too large to ever be drugs.

Most successful drugs have to work when taken by mouth, for the simple reason that if they only work when they're injected, it's a severe problem getting them to patients and getting patients to accept them.

And to get an oral pill to succeed, you have to get a good amount of it into the circulating bloodstream after eating it. That is a lot harder than it looks. The drug first has to dissolve and get absorbed out of the gut, a complex process. One thing everyone agrees on is that really large compounds usually have trouble. If the drug does make it out to the bloodstream, it next gets sent right through the liver, which is there to rip to shreds any molecules it doesn't recognize. The larger the molecule, the more chances the liver has to demolish it. These crucial hurdles send a lot of good drug candidates packing.

The cholinesterase inhibitor that Sharpless's team made is a hefty-sized one. Though it's impressively active in a flask, it would be predicted to have very poor "drug-like" properties in an animal test or in a human. To be fair, the group certainly wasn't trying to optimize for these characteristics, but we'll need to find out if smaller molecules can be made this way or not.

There's also the question of how the triazole structures will behave as drugs in general. Not many of these have ever been taken into the clinic, but that's partly because industrial chemists aren't crazy about the azide compounds you usually need to make them. Azides have a bad reputation for exploding when concentrated or heated, although this seems to be less common than the laboratory lore has it. So it could be that the triazoles themselves are fine, but just haven't been explored as well as they might have been.

How will this all play out? Even if the problems I've outlined are real, I believe that this will still go down as a really good idea. And on the other hand, if it turns out to be a general method that catches on, it could change the way medicinal chemistry is done. We could spend less time finding leads for new drugs, for one thing. In many cases, we'd be starting our new projects with much more potent compounds, too, which could shorten the whole drug development process. If this really "clicks," it'll be big.



TCS Daily Archives