Caltech researchers use the "unnatural selection" of directed evolution to alter a bacterial enzyme

In a novel process that makes the evolution of species look like an engineering design contest, California Institute of Technology scientists have forced a bacterial strain to "evolve" a beta caratenoid enzyme . The evolved enzyme can carry out reactions that normally require other proteins and expensive agents. These reactions are important for making drugs and chemicals.

The enzyme, called a cytochrome P450, is one of a class of enzymes that inserts oxygen atoms into a huge number of compounds, according to Caltech chemical engineering and biochemistry professor Frances Arnold. In an article appearing in the June 17 issue of the journal Nature, Arnold and her team demonstrate an evolutionary process to alter the enzyme and overcome several of the natural limitations that make it inefficient and expensive to use. "The P450s do some great chemistry, but they are complex and ill-behaved," says Arnold, who helped pioneer directed evolution some years ago. "We hope to create stable P450s that need no expensive external cofactors to work. We'd like to pare them down to the absolute minimum and see how well they can do. They may well be much better catalysts without all the fancy machinery."

Nature helps these enzymes along by providing a retinue of protein assistants and complex chemicals that are either impossible or very expensive to reproduce outside a cell. In the Nature paper, the Arnold team reports the laboratory evolution of a P450 that no longer needs any of this help to catalyze its reaction.

Directed evolution has been heralded recently as a means of creating new enzymes and even whole organisms with new or vastly improved characteristics. In contrast to natural evolution, in which the survival of the fittest dictates the direction of change, directed evolution engineers enzymes for specific purposes. These purposes may have nothing to do with what the enzymes do in their natural organisms. For example, the enzymes may be better suited for removing laundry stains, or they may be used to treat diseases.

In directed evolution, the scientists dictate which enzyme characteristics will be selected in each generation, very much the way plant breeders use mutation and selective breeding to create new corn varieties or animal breeders have introduced new varieties of cattle or sheep.

In the case of the P450 enzymes, Arnold and her team wanted an enzyme that would work without any additional help. To do this, they made use of a known feature of the P450s—that hydrogen peroxide could support the reaction in the absence of the helper proteins and cofactor. Their goal was to evolve what Arnold calls this "biochemical oddity" to make it the primary pathway for the enzyme to work.

Arnold is interested in creating enzyme catalysts that could be used to manufacture drugs and chemicals. "The ideal is a rock-stable, very fast, very active enzyme that you could put in a bottle," she says. "Over the next five to 10 years, we're aiming at enzymes that the chemical industry could use." These improved enzymes could also perhaps lead eventually to new technologies that remove toxic wastes in soil or water.

"In this paper, we present the results of two generations, which makes the enzyme 20 times better than it originally was," she says. "We don't know how far we'll be able to go. But we hope that 10 more generations could result in something remarkably good. With other enzymes, for example, we have seen improvements of 500-fold over 10 generations."

Arnold's procedure for directed evolution is to first target an interesting enzyme. In the case of the current paper, the enzyme is from Pseudomonas putida, a bacterial strain that uses the P450 enzyme as a sort of "digestive aide" to eat camphor that is found in soil. The researchers then take the gene that codes for the enzyme and create millions of mutants, which they put back into a bacterial "workhorse" that generates the mutant enzymes. The scientists then screen these mutants for the desired qualities.

The best candidates from that generation can either be "bred" together to obtain further improvements, or the process can be started over again with new mutations.

Arnold says that each generation of the P450 enzyme takes about a week or two. Practically, a total effort in directed generation takes at least several months.

The other authors of the paper are Hyun Joo and Zhanglin Lin, both postdoctoral scholars in chemical engineering at Caltech.