Design of a switchable eliminase

Abstract

The active sites of enzymes are lined with side chains whose dynamic, geometric, and chemical properties have been finely tuned relative to the corresponding residues in water. For example, the carboxylates of glutamate and aspartate are weakly basic in water but become strongly basic when dehydrated in enzymatic sites. The dehydration of the carboxylate, although intrinsically thermodynamically unfavorable, is achieved by harnessing the free energy of folding and substrate binding to reach the required basicity. Allosterically regulated enzymes additionally rely on the free energy of ligand binding to stabilize the protein in a catalytically competent state. We demonstrate the interplay of protein folding energetics and functional group tuning to convert calmodulin (CaM), a regulatory binding protein, into AlleyCat, an allosterically controlled eliminase. Upon binding Ca(II), native CaM opens a hydrophobic pocket on each of its domains. We computationally identified a mutant that (i) accommodates carboxylate as a general base within these pockets, (ii) interacts productively in the Michaelis complex with the substrate, and (iii) stabilizes the transition state for the reaction. Remarkably, a single mutation of an apolar residue at the bottom of an otherwise hydrophobic cavity confers catalytic activity on calmodulin. AlleyCat showed the expected pH-rate profile, and it was inactivated by mutation of its active site Glu to Gln. A variety of control mutants demonstrated the specificity of the design. The activity of this minimal 75-residue allosterically regulated catalyst is similar to that obtained using more elaborate computational approaches to redesign complex enzymes to catalyze the Kemp elimination reaction.

Scheme 1.

Kemp elimination reaction.

Fig. 1.

Summary of the computational procedure. A) Identification of potential points to introduce a catalytic residue into cCaM (coordinates from X-ray structure 1EXR). (B) Examination of the Glu residues in the identified positions/docking of the substate. (C) Transition state docking based on a search of a superrotamer library for the transition state.

Fig. 2.

(A) Comparison of the benzisoxazole elimination activity of AlleyCat (circles) with cCaM-F92Q (diamonds), [Protein] = 16 μM, [Substrate] = 0.5 mM. (B) Dependence of the activity of AlleyCat on [Ca²⁺] at pH 6.9, solid line shows the simulated two-site binding isotherm with an average K_d of 10^-5 M. (C) pH profile of activity of AlleyCat. (D) Chemical (guanidinium hydrochloride, GdmCl) denaturation profiles of AlleyCat (circles) and cCaM (triangles).

Fig. 3.

NMR structure of AlleyCat. (A) Representation of protein’s surface showing E92 pointing into the hydrophobic binding pocket; carbons and the backbone atoms are shown in green. (B) The 20 conformers with the lowest energy representing the solution structure shown after superposition of the backbone heavy atoms N, C^α, and C′ atoms of the secondary structures. (C) “Sausage” representation of backbone and best-defined side chains: A spline function was drawn through the mean positions of C^α atoms with the thickness proportional to the mean global displacement of C^α atoms in the 20 conformers after superposition as in B. Calcium atoms are shown in yellow.