Seven designs were subsequently selected for experimental characterization based on (i) hydrophobic packing of the naphthyl group, (ii) supporting interactions for the Asp/Glu residue (salt bridges or interactions to stabilize the unfavorable charge with good packing of the side chain), and (iii) packing of the lysine side chain to restrict its movement
Seven designs were subsequently selected for experimental characterization based on (i) hydrophobic packing of the naphthyl group, (ii) supporting interactions for the Asp/Glu residue (salt bridges or interactions to stabilize the unfavorable charge with good packing of the side chain), and (iii) packing of the lysine side chain to restrict its movement. variants of the two best designs, RA114 and RA117, exhibited among the highest design process proceeds in three stages: (i) construction of an idealized active-site description, or theozyme; (ii) placement of the theozyme in a suitable protein scaffold; and (iii) optimization of the surrounding sequence for transition-state binding. Choosing an appropriate theozyme is critical as the catalytic mechanism and the chemical composition of the catalytic residues and their interactions with the transition state must be made the decision upon. Each different theozyme represents GNF-6231 a hypothesis about how catalysis can be achieved, which can be evaluated using quantum mechanical calculations [6] and ultimately by the experimentally observed activity of the designed enzymes. Multistep retroaldol reactions, which are subject to amine catalysis, were among the first transformations tackled by computational design [2,7]. Catalysis is initiated by attack of a reactive lysine around the carbonyl group of the -hydroxy-ketone substrate to form a tetrahedral carbinolamine intermediate that subsequently breaks down to give a protonated Schiff base. The latter serves as an electron sink, facilitating cleavage of the adjacent carbonCcarbon bond to generate an aldehyde and an enamine. Protonation and hydrolysis of the enamine prospects finally to release of acetone and regeneration GNF-6231 of the enzyme. This mechanism, which is usually exploited by natural type I aldolases [8], has been successfully mimicked by lysine-rich helical peptides [9C11] and proteins [12], as well as catalytic antibodies selected against 1,3-diketones [13,14] and -keto sulfones [15]. The first computationally designed retroaldolases were obtained by explicitly modeling the structure of the carbinolamine intermediate and flanking transition states, the most sterically demanding species along the reaction coordinate. These designs also included an ordered water molecule, bound by two hydrogen-bonding side chains, to promote carbinol-amine formation and breakdown. It was envisaged that this water would additionally aid proton transfer from your -alcohol in Rabbit Polyclonal to PLA2G4C the cleavage step. The designed catalysts exhibited significant retroaldolase activity, with rate accelerations of up to 4 orders of magnitude over background [2,7]. Detailed mutagenesis and structural studies of representative designs have confirmed the importance of the reactive lysine, but a significant catalytic role for the explicit water has not been observed [16]. Although naturally occurring class I aldolases such as D-2-deoxyribose-5-phosphate aldolase often make use of a water molecule for acid/base catalysis, this water is typically oriented and activated by an extensive network of polar side chains that is hard to emulate with current computational protein design methodologies [17,18]. We speculated that, in the absence of such a network, amino acid side chains interacting directly with bound ligands at the designed active sites might provide better control over the reaction coordinate than a loosely bound water molecule and thus afford higher activity. Here we describe the results of design calculations in which the explicit water in the earlier theozymes GNF-6231 is replaced by the carboxylic acid GNF-6231 side chain of glutamic or aspartic acid, to function as a general acid/base, plus a serine or threonine residue, to provide additional hydrogen-bonding interactions. We also describe approaches to increase the activity of the designed catalysts by computational loop remodeling and by protein evolution using yeast display with a mechanism-based inhibitor. Results Computational design strategy As in our previous work [2,7], we focused on amine catalysis of the retroaldol reaction of 4-hydroxy-4-(6-methoxy-2-naphthyl)-2-butanone [19] to give 6-meth-oxy-2-naphthaldehyde and acetone (Fig. 1). However, the water molecule in the original theozyme was replaced with the side chains of two amino acids, an aspartic or glutamic acid plus a serine or threonine, which can make hydrogen-bonding interactions directly with the carbinolamine. We hypothesized that such residues would be better suited for acid/base chemistry than a loosely bound water molecule. For example, carboxylic acids are effectively utilized for acid/base catalysis in aspartyl proteases and glycosidases. The carboxylic acid side chain of Asp/Glu could promote several actions in the retroaldolase reaction (Fig. 1), including (i) formation of the carbinolamine intermediate by protonation of the oxyanion, (ii) breakdown of the carbinolamine with release of water to generate a protonated Schiff base, (iii) proton abstraction from your -hydroxyl group of the intermediate to initiate CCC bond cleavage, (iv) protonation of the producing enamine, and (v) general-base-assisted hydrolysis of the protonated Schiff base to release acetone and regenerate the catalyst..