Ancestral Sequence Reconstruction retrofits proteins to try to make better drugs.

Rational drug design seems so promising. So, you know, rational. How could it not work? Here’s the premise:
•1. Identify a drug that has a desired effect. This is usually done either by screening a library of possible drugs or even of random chemicals, or by actively searching for a molecule hypothesized to be active against a cellular target of interest.
•2. Figure out how the candidate drug works. What is its target is in the cell? How does it interact with said target?
•3. Design a drug that optimizes its attraction to the target so that the final drug binds more strongly to its target, or stays bound for longer, or doesn’t accidentally also bind to something else.

Optimizing can be done by tweaking a drug’s interaction with its target: changing its shape, size, or electric charge so that it fits better. These modifications can be designed and assayed on a computer before a drug is synthesized and tested in the real world—first against a drug’s isolated target, then in cultured cells, and then animals. If all those work out, the modified drug may eventually be tested in humans.

The strategy has yielded some spectacular successes, most notably HIV protease inhibitors. But rational drug design is costly, in terms of time, money, materials, and effort. And too often it yields end products that are too toxic to be used pharmacologically, even if they bind their target well.

Now, a bunch of scientists writing in Nature Biotechnology have suggested that, when the drug itself is a protein, there's a better option than trying to design and chemically optimize drug candidates. Instead, we can take a look back through time to see how evolution has already optimized proteins to perform in different contexts.

The scientists call their method Ancestral Sequence Reconstruction. It is based on the idea that evolution has optimized proteins to function optimally in species that may have very different physiologies. Ancestral Sequence Reconstruction highlights the notion that evolution is toward fitness in a specific context, rather than toward an absolute idea of “highest” or “best.” As the name implies, the scientists predict ancient protein sequences based on proteins currently present across different species. Using data from gene sequencing, they can reconstruct these potential ancestral proteins.

As a proof of concept they looked at factor VIII (FVIII), a clotting factor that, when deficient, causes hemophilia A. We currently treat this with a recombinant human FVIII, but it has a short half-life, tends to generate an immune response, and is not particularly efficient. Based on the sequence present in other mammals—not just primates, but mice, rats, hamsters, rabbits, chinchillas, water buffalo, killer whales, snow leopard, shrews, hedgehogs, manatees, alpacas, ferrets, mouse-eared bats, etc.—these scientists generated 78 putative earlier, ancestral versions that would have been present in ancient mammals.

When put into cell lines, three of these possible ancestral FVIII proteins were expressed at higher levels, were exported more efficiently, and were more potent than the proteins we’ve got. The researchers in Nature Biotechnology speculate as to why human FVIII may have evolved toward reduced coagulation activity, but regardless of the reason, these retrofitted, hyperactive variants could make better drugs.

One limitation of this approach is that it won’t work on all classes of drugs. As yet it can only work on proteins, since this is the only type of molecule that is under evolutionary pressure in the biological systems where evolution holds sway. But the authors conclude their piece in Nature Biotechnology by noting that the “protein design space has already been refined by natural selection for beneficial properties.” Not taking advantage of that would be irrational.