Carotenoids are natural pigmented compounds that are widely used as food additives and pharmaceuticals. They are produced mainly by bacteria, fungi, and plants but often cannot be purified in useful quantities from their natural sources. Thus, there is much interest in finding other ways to make them. Now, in the July issue of Nature Biotechnology (18, 750–753) Claudia Schmidt-Dannert, Daisuke Umeno, and Frances Arnold report the successful ‘molecular breeding’ of a particular carotenoid biosynthetic pathway in Escherichia coli. This is the first example of directed evolution applied to a biosynthetic pathway, rather than a single enzyme.

Carotenoids are composed of long carbon chains, with various modifications. They contain different numbers of carbon–carbon double bonds, and the number of double bonds can be increased by enzymes called desaturases. Other enzymes, called cyclases, also increase the diversity of carotenoids by cyclizing the ends of the carbon chains. Carotenoids with many double bonds and cyclic ends are darker in color than their simpler counterparts.

Schmidt-Dannert et al. took advantage of these color differences to create new pathways for making specific carotenoids in E. coli, which does not normally produce these compounds. They generated a library of desaturase genetic variants in vitro — by ‘shuffling’ related desaturase genes from two carotenoid-producing species of Erwinia bacteria — and transformed this library into an E. coli strain that had been engineered to make the appropriate precursor compound, phytoene.

Next, they screened through the colonies, looking for ones displaying darker pigmentation than a colony containing a wild type desaturase gene. The expectation was that some of the desaturase variants in the library, unlike the wild type desaturase enzyme, would be capable of producing additional, darkly colored carotenoids. The picture shows an artistic rendition of individual E. coli cells producing a sampling of carotenoids, ranging in color from pale yellow to reddish-purple.

This procedure worked. Schmidt-Dannert et al. were able to isolate a pink colony containing the carotenoid tetradehydrolycopene, which has six more double bonds than phytoene, the starting substrate. Using the same basic approach, but with a different enzyme, this group was then able to extend the biosynthetic pathway to include a cyclase. They obtained a mutant cyclase that could act on the new tetradehydrolycopene substrate to produce torulene, a bright red carotenoid.

The selected desaturase and cyclase genes have been sequenced, but it is not yet understood how the observed mutations affect the structures of the proteins. Nonetheless, it is evident that the torulene-producing E. coli strain uses a brand new pathway to make the carotenoid — in organisms, such as yeasts, that naturally produce torulene, it is made using different enzymes and starting substrates. These results suggest the exciting possibility that entirely novel compounds could be produced by these combinatorial molecular breeding methods.