A method has been devised that allows a ketoreductase enzyme to catalyse reactions other than its natural ones. The key is to excite the enzyme's cofactor using light – an approach that might work for other enzymes. See Letter p.414
Enzymes have several advantages over conventional catalysts for organic synthesis. For example, their ability to perform reactions at room temperature in water makes them suitable for environmentally friendly chemical processes. But many synthetically useful reactions cannot be catalysed by naturally occurring enzymes. The quest to expand nature's enzymatic repertoire of transformations is therefore a crucial area of research. On page 414, Emmanuel et al.1 report a strategy that allows ketoreductase enzymes to perform completely different reactions from the ones that they evolved to catalyse.
Living organisms are without doubt the best chemists on Earth — a plethora of reaction types is catalysed by the thousands of different enzymes present in every cell. The reactions take place with excellent selectivity (forming solely the desired product), astonishing efficiency (performing hundreds of catalytic reactions per second at a single catalytic site) and at ambient temperatures and pH values. By contrast, chemists have developed methods that allow a range of reactions with no enzymatic counterpart to be easily performed. In many cases, developing an enzyme that can perform such reactions is desirable.
Researchers have therefore devised a range of concepts for creating or modifying proteins to catalyse reactions unknown in nature2,3,4. One approach is to incorporate chemical transition-metal catalysts into protein scaffolds. As early as 1978, the protein avidin was modified to incorporate a rhodium catalyst5, producing an enzyme that catalyses asymmetric hydrogenations — transformations in which hydrogen reacts with organic molecules to produce products predominantly as one mirror-image isomer (enantiomer). This year, a system in which a ruthenium catalyst was incorporated into the streptavidin protein enabled olefin metathesis (a carbon–carbon bond-formation reaction; Fig. 1a) in vivo in the bacterium Escherichia coli6.
A second general approach is to use protein engineering to redesign enzymes to catalyse reactions other than the native one. This strategy is exemplified by the engineering of P450 monooxygenase enzymes7 to catalyse carbon–carbon bond-formation reactions (cyclopropanations; Fig. 1b), rather than the analogous carbon–oxygen bond-formation reactions (epoxidations) that occur naturally. A third approach is computational de novo protein design, which has been used to make an enzyme that catalyses the Kemp elimination reaction, in which a hydrogen ion (H+) is removed from a carbon atom in an organic molecule (Fig. 1c)8. Subsequent extensive protein engineering through directed evolution of the enzyme resulted in catalytically efficient mutants9.
Emmanuel et al. now report a striking new concept for generating enzymes that catalyse unnatural reactions: the authors use light to excite a cofactor (NAD(P)H) bound in the active site of a ketoreductase (KRED) enzyme. The resulting photoexcited cofactor generates a radical intermediate that serves as a hydrogen source. Furthermore, this hydrogen source is chiral — it has a 'handedness' that can potentially be passed on to other molecules during reactions. The authors find that, when KRED contains a photoexcited cofactor, it catalyses a reaction in which a halogen atom is removed from molecules known as halolactones, forming products predominantly as one enantiomer (Fig. 1d). Moreover, the enantiomer that is formed depends on the preference of the KRED that is used. The authors show that this unnatural reaction can be used to generate either of the enantiomers of products formed from a broad range of halolactones, demonstrating the synthetic usefulness of this approach.
The KRED fulfils two functions in this reaction. First, it ensures productive, coordinated binding of the photoexcited NAD(P)H with the halolactone in its active site. But it also recycles the spent cofactor by reacting it with isopropanol (a component of the reaction mixture), regenerating NAD(P)H. This efficient recycling enables a KRED molecule to mediate multiple catalytic cycles, as would be needed for the enzyme to be used to make gram or kilogram quantities of product for industrial applications.
Not all the KREDs investigated by the authors could catalyse the reaction; Emmanuel and colleagues found that certain point mutations in the enzyme are needed to promote the productive binding of NAD(P)H within the enzyme's scaffold. However, the catalytically active KREDs bind the halolactones perfectly, even though they do not resemble the enzymes' natural substrates. Furthermore, the authors proved that the unnatural reaction occurs only when NAD(P)H is tightly bound to KRED and is irradiated with blue light.
The authors proposed a mechanism for the reaction in which light irradiation causes an electron to be transferred between the NAD(P)H and the substrate, triggering cleavage of the substrate's carbon–halogen bond, and thus generating a radical intermediate that accepts a hydrogen atom to form the final product enantioselectively (see Fig. 3d of the paper1). They nicely confirm this mechanism by generating a deuterium donor from NAD(P)H in situ in KRED, and observing where the deuterium is incorporated into the reaction products.
Emmanuel et al. have demonstrated a completely new strategy for accessing unnatural enzymatic reactions by exploring the interface between photochemistry and protein science. Other synthetic transformations can be envisaged with this approach, by using light-induced changes in NAD(P)H analogues or other cofactors. For instance, the well-studied flavin cofactors (flavin adenine dinucleotide, flavin mononucleotide and their artificial analogues) could be prime candidates for investigation, because various flavin-dependent enzymes are important biological catalysts used by synthetic chemists10. In combination with modern tools for protein engineering11, the authors' concept is likely to have a strong impact on the use of various enzyme classes in biocatalysis.