Roche and their partner, Trimeris are developing an anti-HIV compound (called T-20) that has some remarkable features. It's the first to target a cell protein (gp-41) that is a key binding step in HIV's mode of infection, for one thing. But to be more precise, it has 106 more interesting things about it - that's the total number of synthetic steps in the manufacturing process.

My synthetic-organic readers are probably saying just what everyone in the field does when they first hear about that one - it's usually a variation of "no (procreating) way," followed by laughter. Those of us who've done natural product synthesis get the shivers, remembering those 20- or 30-step build-the-pyramids projects from our pasts. Trying to extrapolate those experiences up the asymptotic curve of awfulness to 106 steps is a scary exercise.

Fortunately, the compound is a peptide (36 amino acids,) which means that the chemistry is well-established and can be done largely by machine. It also means that you don't do all 106 steps in a row. Well, you could try, if you had a good excuse like having had a big recent dose of LSD or something, but it's not recommended.

For starters, it probably wouldn't work. Peptide synthesizers work by hanging one end of the chain off a solid resin support. As the molecule gets longer and snakier, it tends to ball up on you. Eventually, the far end, the one that is getting new amino acids added to it, isn't sticking out into the solution any more, but is tucked back into its own folds. End of synthesis, by default. The second problem with having that many linear steps is a mathematical one. A to-the-hundredth-power term in an equation is a bomb waiting to go off, and if your synthetic yields aren't just about perfect, you're going to be in bad shape.

For example, if every step works in 95% yield (which would be a mighty fine result across 106 steps,) then you're going to come staggering across the finish line with a 0.4% overall yield. What if you average an 85% yield? Admittedly, a peptide synthesizer will beat that, but it had damn well better, because that pace will leave you with 3-millionths of a percent yield. Ugly things, those exponents.

Roche, not being a gang of idiots, will certainly be making T-20 in small chunks, then stitching those together. The tricky part is figuring out where to break up a molecule that size. What combination of fragment assembly schemes has the best chance of high reproducible yields? The life expectancies of humans being what they are, you can't try all the permutations and pick the best. Some educating guessing using known pitfalls of protein design has apparently given them a route that works.

But it's going to be a honker of a synthesis, no matter how well it goes, because it's going to have to be done on monstrous scale. Peptides, as I never tire of pointing out, have every chance of making crappy drugs, and this is worse than most. You can't take a 36 amino-acid peptide orally and expect much to happen; your digestive tract will rip it to shreds like it does every other protein. T-20 has to be injected, twice a day, 100 mg per shot. Roche is going to have to make thousands of kilos of this stuff, which should handily break any records for peptide synthesis. It already looks like it'll be tough, at least at first.

While T-20 works quite well, it's going to be extremely costly to produce. Roche has reportedly already spent nearly $500 million on the manufacturing facility in Colorado. T-20, then, is also going to be extremely costly to take. There's room to doubt how long it'll take to pay off, especially for Trimeris. If resistance to the drug shows up ahead of expectations, it could never pay off at all. (The companies are developing some related peptides that might get around this problem, which is a good strategy.)

I'm still in awe of the decision to go ahead with this drug. It's a major risk, and I can only hope that it has major rewards for both the patients and the companies. And I can also be glad that I'm not having to keep those peptide synthesizers out in Boulder stocked with solvents!