Introduction
Most biochemical building blocks are chiral—they exist in nonsuperposable mirror-image forms (enantiomers), just like shoes, screws and snail shells. Although both forms are possible, nature generally chooses one as a building block; for example, only the L forms of amino acids are used to construct proteins. This choice makes the enantiomeric D-amino acids perfect components for specialized applications such as bacterial cell walls, antibiotics and pharmaceuticals because proteases (most of which have evolved to target the L-amino acid subunits) cannot degrade compounds made from the D-amino acid subunits. There are no natural sources for most D-amino acids, and current chemical and biochemical routes to D-amino acids are limited in the range of materials accessible and by the high cost of preparation1. To address this problem, Vedha-Peters et al.2 created a promiscuous D-amino acid dehydrogenase by expanding the substrate range of a natural single-substrate enzyme. The resulting variant efficiently accepts more than 16 substrates and promises to be a useful synthetic tool.
The ideal catalyst for synthesizing D-amino acids or other building blocks requires contradictory features. First, the catalyst must have high enantioselectivity so that it makes only the desired D-amino acids, and second, it must have broad substrate specificity so it can create different D-amino acids for different applications. How could one adapt an existing enzyme to synthesize a broad range of D-amino acids? It is possible to imagine starting with an enzyme that forms L-amino acids and then reversing the enantioselectivity to favor D-amino acids. Some previous attempts to reverse enantioselectivity have succeeded3, whereas others yielded nonenantioselective enzymes, which suggests that this is a difficult approach4. An alternate strategy is to start with an enzyme that forms one D-amino acid and engineer the active site so it accepts many substrates. Such a substrate-broadening approach has succeeded previously for two other synthetically useful reactions: amino-acid oxidation and sugar phosphorylation5, 6. Vedha-Peters et al. applied this second approach in two stages and generated an especially useful reductive-amination enzyme to create D-amino acids.
The authors started with the enzyme meso-2,6-diaminopimelic acid D-dehydrogenase (DAPDH), a highly substrate-specific enzyme involved in lysine biosynthesis. The goal was to broaden the range of side chains that the enzyme accepts while maintaining its high enantioselectivity at the reaction site (Fig. 1a). A previous X-ray crystal structure had identified three amino acid residues that bind to the carboxylate group on the side chain of a meso-2,6-diaminopimelic acid (m-DAP) analog7 (Fig. 1b). The authors envisioned that replacing these amino acids might broaden the range of side chains tolerated by the enzyme. To choose the replacement amino acids, they created random substitutions at all three sites and tested the resulting variants with analogs of the natural substrate. They identified one variant with substitutions at all three sites (Fig. 1b). It no longer accepted the original substrate, but it did accept D-lysine, which lacks the carboxylate side chain of m-DAP.
Figure 1: Converting a single-purpose enzyme to a multipurpose enzyme for synthesis.
(a) Five amino acid changes converted a dehydrogenase that accepts only one substrate to one that accepts over 16 different substrates. In the first stage, random substitution of three amino acids in the active site followed by screening revealed a triple mutant that converted a substrate analog to D-lysine but no longer reacted with the starting substrate. In the second stage, random mutagenesis throughout the enzyme identified two mutations outside the active site that further broadened the substrate specificity. (b) Structure showing the location of the changes relative to a substrate analog (magenta sticks) and reaction partner (NADP+, gray sticks). Three of the changes (magenta balls) touch the side chain of the substrate analog, whereas two others (yellow balls) do not touch the substrate analog and would be difficult to predict with existing modeling methods. Protein graphic created using PyMOL (http://pymol.sourceforge.net)9.
Katie Ris
Full size image (63 KB)Starting with this variant, the authors then sought to further broaden the range of substrates accepted. There was no obvious next location for further changes, so the authors made random substitutions throughout the enzyme and again tested the resulting variants. They found a variant with two additional changes, this time just outside the binding site for the side chain (Fig. 1b). A helix forms part of the binding site, and the changes were on the away face of this helix. These more distant changes would have been hard to predict using molecular modeling or directed approaches. The final DAPDH variant contains five amino acid substitutions and achieves the goal of being useful for synthetic applications. It shows good activity and high enantioselectivity and now accepts more than 16 substrates efficiently, though it still does not accept some more-hindered substrates.
Creating highly selective enzymes having broad substrate specificity has been an elusive goal for protein engineers. Modeling and rational design can enhance the substrate specificity toward a particular new substrate8, but strategies to engineer broad specificity have been lacking. The current success of Vedha-Peters et al. and previous successes5, 6 all include changes outside direct contact with the substrate. It remains to be seen whether such second-sphere changes are often the key to broad specificity and what the molecular basis for this broad specificity is; perhaps it is protein dynamics. In the future, it may be possible to predict the changes needed to broaden the substrate specificity, but in the short term the authors' approach of stepwise broadening may be the fastest route to new useful enzymes.
-Amino Acids