There are a huge number of techniques in the protein world that relay on tying down some binding partner onto some kind of solid support. When you’re talking about immobilizing proteins, that’s one thing – they’re large beasts, and presumably there’s some tether that can be bonded to them to string off to a solid bead or chip. It’s certainly not always easy, but generally can be done, often after some experimentation with the length of the linker, its composition, and the chemistry used to attach it.
But there are also plenty of ideas out there that call for doing the same sort of thing to small molecules. The first thing that comes to mind is affinity chromatography – take some small molecule that you know binds to a given protein or class of proteins well, attach it to some solid resin or the like, and then pour a bunch of mixed proteins over it. In theory, the binding partner will stick to its ligand as it finds it, everything else will wash off, and now you’ve got pure protein (or a pure group of related proteins) isolated and ready to be analyzed. Well, maybe after you find a way to get them off the solid support as well.
That illustrates one experimental consideration with these ideas. You want the association between the binding partners to be strong enough to be useful, but (in many cases) not so incredibly strong that it can never be broken up again. There are a lot of biomolecule purification methods that rely on just these sorts of interactions, but those often use some well-worked-out binding pair that you introduce into the proteins artificially. Doing it on native proteins, with small molecules that you just dreamed up, is quite another thing.
But that would be very useful indeed, if you could get it work reliably. There are techniques available like surface plasmon resonance, which can tell with great sensitivity if something is sticking close to a solid surface. At least one whole company (Graffinity) has been trying to make a living by (among other things) attaching screening libraries of small molecules to SPR chips, and flowing proteins of interest over them to look for structural lead ideas.
And Stuart Schreiber and his collaborators at the Broad Institute have been working on the immobilized-small-molecule idea as well, trying different methods of attaching compound libraries to various solid supports. They’re looking for molecules that disrupt some very tough (but very interesting) biological processes, and have reported some successes in protein-protein interactions, a notoriously tempting (and notoriously hard) area for small-molecule drug discovery.
The big problem that people tend to have with all these ideas – and I’m one of those people, in the end – is that it’s hard to see how you can rope small molecules to a solid support without changing their character. After all, we don’t have anything smaller than atoms to make the ropes out of. It’s one thing to do this to a protein – that’ll look like a tangle of yarn with a small length of it stretching out to the side. But on the small molecule scale, it’s a bit like putting a hamster on a collar and leash designed for a Doberman. Mr. Hamster is not going to be able to enjoy his former freedom of movement, and a blindfolded person might, on picking him up, have difficulty recognizing his essential hamsterhood.
There's also the problem of how you attach that leash and collar, even if you decide that you can put up with it once it's on. Making an array of peptides on a solid support is all well and good - peptides have convenient handles at both ends, and there are a lot of well-worked-out reactions to attach things to them. But small molecules come in all sorts of shapes, sizes, and combinations of functional groups (at least, they'd better if you're hoping to see some screening hits with them). Trying to attach such a heterogeneous lot of stuff through a defined chemical ligation is challenging, and I think that the challenge is too often met by making the compound set less diverse. And after seeing how much my molecules can be affected by adding just one methyl group in the right (or wrong) place, I’m not so sure that I understand the best way to attach them to beads.
So I’m going to keep reading the tethered-small-molecule-library literature, and keep an eye on its progress. But I worry that I’m just reading about the successes, and not hearing as much about the dead ends. (That’s how the rest of the literature tends to work, anyway). For those who want to catch up with this area, here's a Royal Society review from Angela Koehler and co-workers at the Broad that'll get you up to speed. It's a high-risk, high-reward research area, for sure, so I'll always have some sympathy for it.
College chemistry, 1983
The 2002 Model
After 10 years of blogging. . .
1. Nadia on March 6, 2009 9:08 AM writes...
'hamsterhood'
Good word.
Permalink to Comment2. milkshake on March 6, 2009 9:37 AM writes...
My first US employer was a small combichem company built originaly around the idea one-bead-one compound, on resin testing. The problem was a fairly high number of false-positive beads, and much-lower-than-predicted confirmed hit yield from the on-bead assay. When they purposefully spiked the library with a known inhibitor attached to resin they usually could not find it in the assay.
As far as I am aware they tried a number of hydrophylic supports and long linkers, before giving up on this completely and going in the direction of macro-beads with a linker cleavable chemically, before switching to parallel synthesis and screening of purified compounds in solution.
I have also made some immobilized Sutent analogs for a toxicology group, analogs linked to Tentagel by an additional liker, through the aminosidechain (where we knew it would not affect the in vitro activity). I spent about two weeks on it and they took it and we never heard from them again. I felt from the beginning it was a futile shot, trying to fish out proteins responsible for side effects by affinity chromatography because the real toxicology problem was related to high organ accumulation of the drug.
Also I should mention that we got a new academic hire here at the institute who did his best work by screening cell adhesion to beads with a peptoid library, with differentially fluorescent-labeled cells, and it worked for him so maybe it really depends on the used techniques...
Permalink to Comment3. HelicalZz on March 6, 2009 9:39 AM writes...
All true, but there is an advantage as well. After all, this is just a screening of a different color. It will have hits, and misses, and misdirections. But ... a hit with a tethered small molecule does have the advantage (again in theory) of handing the development chemists a structure that may have activity and also provides for a portion of the molecule i.e. the linkage part, that they are free to 'mess with' to improve drugability down the road.
As a technique, it will unquestionably miss a lot. How could it not? But it may (perhaps - and no, I can't hedge this enough) result in some more practical hits when it does find them.
Zz
Permalink to Comment4. JAB on March 6, 2009 10:22 AM writes...
I think this group of techniques is not really something you can do in a screening, turn the crank mode. I think you have to first understand the SAR of your small molecule to find the best attachment point, then link it up and test all of the constructs for bioactivity in your primary assay. Even then you may not be able to fish out what you're looking for. And you need really tight controls in your experiments. That said, I'm intrigued with click chemistry as a way to do this, if one can get a terminal acetylene or azide onto one's small molecule.
Permalink to Comment5. JAB on March 6, 2009 10:24 AM writes...
And the best example I can think of where this worked really well was with the glycopeptide antibiotics like vancomycin. Smith Kline did some nice stuff with that - you could send your extract through an affinity column and elute off your glycopeptides very neatly.
Permalink to Comment6. Ty on March 6, 2009 11:43 AM writes...
I thought small molecule microarray was a bs when I saw it during the interview at one of Schreiber's former companies way back when. Only recently, I began to think that "FRAGMENT" microarray would make perfect sense. I have even looked to see if I can file a patent on this idea (seriously) and, alas, that's when I came across Graffinity, who had already been doing exactly what I had in mind...
Permalink to Comment7. Hap on March 6, 2009 4:41 PM writes...
NYT says that Roche bumped its offer for Genentech to $93/share. I guess dropping the original offer price didn't work so well.
Permalink to Comment8. Sili on March 7, 2009 9:14 AM writes...
King of the simile.
Permalink to Comment9. Lucifer on March 9, 2009 5:06 AM writes...
Merck buys Schering-Plough for 44 Billion..
Permalink to Comment10. Profiler on March 9, 2009 7:01 AM writes...
Affinity-purification approaches work well with enzymes which have solvent-accessible but not too deep catalytic pockets. It is really neat in combination with using the compound in its free form for competition with the immobilized ligand and a quantitative read-out for target binding. For target classes where pan-inhibitors exist, one can use the pan-inhibitor for immobilization on the resin and the compound of interest in its free form. This is the principle of the "Kinobeads" approach for Kinases which was published in Nature Biotechnology not long ago.
Permalink to Comment11. anon the II on March 9, 2009 7:58 AM writes...
I went looking at these "Kinobeads" and I'm trying to figure out how they differ from the approach of Serenex, a small biotech in Durham, NC.
Serenex was bought by Pfizer last year and no longer exists.
Permalink to Comment12. Seenitbefore on March 9, 2009 1:13 PM writes...
Just another attempt by a famous professor to get rich off an IPO, um didn't Sunesis try and sling this BS too?
Permalink to Comment