In March, Technology Review publish a piece about some enzymes that the lab of Dr. David Baker (the head of Rosetta@home, one of my favorite scientific distributed computing projects) designed from scratch.
In a major step forward for computational protein design, scientists have built from scratch a handful of enzymes that successfully catalyze a specific chemical reaction. These proteins have no naturally occurring counterparts, and the reaction–which breaks down a man-made chemical–has no natural catalyst.
As you can see from the comparisons between catalyzed and un-catalyzed reactions in the graph above, when it comes to enzymes, it’s all about speed.
Of the 72 proteins selected, 32 successfully helped along the reaction. The most efficient proteins sped up the reaction to 10,000 times the rate without an enzyme.
While that’s an impressive feat compared with earlier enzyme design attempts, the synthesized enzymes pale in comparison to naturally occurring ones. “It’s not very good at all,” says Baker. “Naturally occurring enzymes can increase the rate of reactions by much, much greater amounts”–as much as a quadrillion-fold.
So while that might seem like a very small step in the right direction when you compare it to evolved enzymes, it helps to keep in perspective just how hard what computational protein design is trying to do is. Alberts’ Molecular Biology of the Cell (5th ed.) cites the following example: In one particular enzyme, if you replace a glutamic acid (an amino acid) with an aspartic acid, that shifts the position of the catalytic carboxylate ion by only 1 ångström (0.1 nanometer, or about the radius of an hydrogen atom). That’s enough to decrease the activity of that enzyme by a thousandfold. If that’s not precision work, I don’t know what is.
The Baker Lab is taking two approaches to refine its enzyme designs: Further refinements in their software, which is updated often on Rosetta@home and is currently being tested against other approaches in the CASP 8 international protein prediction competition. The second approach is using the enzyme designs as starting points for directed evolution, to see if natural selection can do some of the heavy lifting of optimization now that they work.
This will be very exciting to watch in the following months and years.