Catalytic machinery of enzymes expanded

The amino-acid side chains found in enzymes contain at most one chemical group, and are crucial for molecular recognition. But fewer than half of these side chains contain groups that can act as acids, bases or nucleophiles (electron-pair donors) in enzyme catalytic cycles. None of the side chains can act as electrophiles (electron-pair acceptors), which could also be useful for catalysis. The introduction of unnatural amino-acid residues that bear potentially catalytic side chains could therefore open up a wide range of new enzymatic reactions.

Conventional catalysts are a fertile source of inspiration for chemical groups that would expand the catalytic repertoire of enzymes: both small-molecule organic catalysts (organocatalysts) and transition-metal catalysts can activate substrate molecules in ways that enable a variety of reactions that are useful for organic synthesis. To enable enzymes to access this exciting reactivity, methods are required for the efficient site-specific incorporation of amino acids that bear new chemical groups. Methods for the directed evolution of the resulting modified enzymes are also required to optimize catalysis in active sites.

Artificial enzymes have previously been constructed by attaching transition-metal catalysts to a small molecule known as biotin, which in turn binds non-covalently with extremely high affinity to the protein streptavidin, thus anchoring the catalyst in a protein framework^2,3. Metal catalysts have also been covalently attached to the side chains of unnatural amino-acid residues that have been incorporated into proteins using modified biological protein-synthesis machinery⁴. With both of these strategies, directed evolution was used to greatly improve the catalytic efficiency and turnover (the average number of reactions catalysed by each enzyme) of the initially produced artificial enzymes, and, in some cases, to increase the selectivity of the enzyme for a particular mirror-image isomer of the product (enantioselectivity). Artificial enzymes have thus been produced that catalyse reactions not found in nature, including silicon–carbon bond-forming reactions⁴, and carbon–carbon bond-forming reactions known as cyclopropanations⁴and ring-closing metathesis reactions².

Burke et al. took a different approach. They started from an enzyme⁵(BH32) that had been computationally designed to catalyse a particular type of carbon–carbon bond-forming reaction, but which also weakly catalyses an unrelated transformation: the hydrolysis of compounds known as 2-phenylacetate esters (Fig. 1). The authors therefore decided to remodel the enzyme to make it an effective catalyst for these hydrolyses.

The researchers determined that a histidine amino-acid residue (His23) in BH32 forms an intermediate called an acyl–enzyme compound during the catalytic cycle. This intermediate is then hydrolysed to yield the product of the enzymatic reaction. However, the catalytic turnover was poor because the hydrolysis of the acyl–enzyme intermediate was slow.

To address this issue, Burke and colleagues replaced His23 with a genetically encodable, unnatural amino acid called N_™-methylhistidine (Me-His; Fig. 1). Me-His is an analogue of histidine in which a methyl group is attached to one of the nitrogen atoms in the side chain. The authors observed that catalytic turnover for the modified enzyme (OE1) was higher than for BH32, an effect that they ascribed to more rapid hydrolysis of the acyl–enzyme intermediate.

Burke et al. then used directed evolution to optimize the function of Me-His in the enzyme’s active site. A wide range of strategies was used to introduce mutations, ultimately resulting in the discovery of a variant, OE1.3, that had improved catalytic efficiency. This variant differed from OE1 by having six mutations, in which one amino-acid residue has been replaced by another. The authors found that OE1.3 hydrolyses a range of analogues of 2-phenylacetate esters in which only hydrogen atoms are attached to the carbon atom adjacent to the carbonyl (C=O) group in the molecules. However, analogues in which a methyl group is attached next to the carbonyl group were poor substrates. The authors therefore carried out further directed evolution to generate OE1.4, an enzyme that has improved catalytic activity with this class of substrate, and which predominantly hydrolyses one of the two mirror-image isomers of each substrate.

The Me-His residue in the modified enzymes acts as a nucleophilic catalyst that is broadly analogous to the nucleophilic residues found in serine hydrolase and cysteine hydrolase enzymes. But how might organocatalysis⁶ in general inspire the discovery of enzymes that are more distant from those found in nature? Organocatalysts speed up many different reactions using just a few generic mechanisms (activation modes), but the catalysis is often inefficient, requiring rather high catalyst loadings (typically 5–20 mole %)⁶. Some of these activation modes are also widely used by enzymes; for example, enamine catalysis is used by class I aldolases⁷. But other activation modes are less widely used enzymatically, despite the fact that they can enable many potentially useful synthetic reactions.

Organocatalysts have been introduced into proteins in various ways, for example by using an attached biotin group as an anchor that binds to streptavidin⁸, or by chemically modifying genetically encoded unnatural amino-acid residues⁹. However, to realize the full power of an expanded range of catalytic chemical groups, substantial optimization is likely to be needed to generate catalytically efficient active sites. Burke et al.have shown that directed evolution can improve enzymes that contain an unnatural organocatalytic group. Their approach might also provide a route to efficient enzymes that use activation modes not found in nature, and which are much more efficient than organocatalysts themselves.