Emerging principles of de novo catalyst design

Abstract

Considerable progress has been made in the understanding of how to exploit hydrophobic and charge—charge interactions in forming binding sites for peptides and small molecules in folded polypeptide catalysts. This knowledge has enabled the introduction of feedback and control functions into catalytic cycles and the construction of folded polypeptide catalysts that follow saturation kinetics. Major advances have also been made in the design of metalloproteins and metallopeptides, especially with regards to understanding redox potential control.

Introduction

The design of proteins that fold into well-defined tertiary structures with the properties of native proteins 1., 2., 3., 4., 5. has recently been demonstrated for sequences of modest size and complexity. Key examples include four-helix bundles, 6., 7., 8., 9. triple-stranded β sheets 10., 11. and a ββα-motif [12]. Protein scaffolds for the de novo design of efficient folded polypeptide catalysts have therefore become available that are readily synthesized and sufficiently complex to accommodate a wide-range of designs and chemistries 13., 14., 15.. The de novo design of proteins that fold into native-like structures capable of recognizing substrates, intermediates and transition states while repelling reaction products, however, remains a challenge. The capacity of enzymes for exploiting the modest reactivity of the amino acids in catalyzing chemical reactions is based on their ability to control interresidue distances and conformational freedom to ensure cooperativity between several residues in individual reaction steps. Enzymes are also capable of excluding solvent water from reactive-site cavities to tune the reactivity of catalytic groups. Smaller peptides are unable to provide the necessary structural rigidity and are therefore limited in these respects in comparison with natural enzymes [16]. Not only can proteins function as scaffolds where several residues bind a single molecule using a combination of hydrogen bonds, salt bridges, metal ion coordination and hydrophobic interactions, but from a mechanistic point of view the enzymes have an unequalled talent for combining several catalytic functions in a single chemical transformation. For example, in the formation of the tetrahedral intermediate in amide bond hydrolysis, chymotrypsin simultaneously performs general-base and nucleophilic catalysis while providing efficient stabilization of the developing oxyanion through finely tuned hydrogen bonding [17]. Although the function of chymotrypsin at this level of description is well understood, the design of new catalysts capable of efficiently hydrolyzing amide bonds remains a formidable and so far unequalled challenge. The challenge that lies ahead is, therefore, to construct reactive sites capable of several catalytic functions simultaneously, as illustrated schematically in Fig. 1. Although de novo enzyme design is still an unrealistic target, designed proteins are excellent model systems for tailoring novel reaction systems of high complexity that are difficult to engineer in any other molecular scaffold, in particular for reactions not catalyzed by nature. So far, designed catalysts are far from enzymatic in terms of efficiency, K_M values in the millimolar to micromolar range have been achieved [18] but k_cat values lag behind evolved enzymes by several orders of magnitude. Considerable progress has, however, been made in the understanding of how to incorporate partial features of enzymes such as substrate binding, metal ion, general-acid and nucleophilic catalysis, and how to achieve cooperativity. In particular, impressive results have been obtained in systems that are capable of recognition, feedback and control over reaction pathways and products 15., 19.. The incorporation of cofactors 20., 21. is an alternative way of obtaining practically useful and competent catalysts. The selectivity of protein binding sites can be combined with the reactivity of natural and non-natural cofactors devised by organic chemists to form catalysts for which there is no natural role model. So far, little use has been made of this opportunity. So, although rate enhancements that depend on exquisite fine tuning of protein scaffold structures so far remain modest in model catalysts, our ability to implement complexity has grown rapidly [22]. There are inherent limitations in using simple motifs as scaffolds, which are due to the regularity of interresidue distances in secondary structures and the difficulties in constructing cavities. It is therefore unlikely that rational design will bring enzymatic rate enhancements, except in selected cases, until more sophisticated constructs have been developed. However, the exploration of new catalytic constructs in model systems coupled to their refinement in more complex proteins using the toolbox of molecular biology may turn out to be a viable approach in the design of new enzymes. A highly successful strategy for refining a designed non-natural catalytic site using in vivo selection and DNA shuffling was reported by Fersht and colleagues [23•]. Thus important advances in incorporating new chemistry into ‘old’ scaffolds have been shown to be a powerful route towards new enzymes, although the difficulties in designing in vivo selection strategies is likely to limit the scope of this particular approach. Nevertheless, the combination of rational design in non-natural scaffolds with selection and screening in natural proteins is an interesting strategy in the engineering of new enzymes (see the article by Soumillion and Fastrez in this issue pp 387–394. Some optimism is justified due to recent developments in the de novo design of complex coupled catalytic systems and progress in the exploitation of metal ions for catalytic functions that are the focus of this review.