Computational redesign of a mononuclear zinc metalloenzyme for organophosphate hydrolysis

Abstract

The ability to redesign enzymes to catalyze noncognate chemical transformations would have wide-ranging applications. We developed a computational method for repurposing the reactivity of metalloenzyme active site functional groups to catalyze new reactions. Using this method, we engineered a zinc-containing mouse adenosine deaminase to catalyze the hydrolysis of a model organophosphate with a catalytic efficiency (k(cat)/K(m)) of ~10(4) M(-1) s(-1) after directed evolution. In the high-resolution crystal structure of the enzyme, all but one of the designed residues adopt the designed conformation. The designed enzyme efficiently catalyzes the hydrolysis of the R(P) isomer of a coumarinyl analog of the nerve agent cyclosarin, and it shows marked substrate selectivity for coumarinyl leaving groups. Computational redesign of native enzyme active sites complements directed evolution methods and offers a general approach for exploring their untapped catalytic potential for new reactivities.

Figure 1. Computational active site redesign

Based on the mechanism of metal ion-catalyzed OP hydrolysis, we superimposed the model organophosphate TS on a set of mononuclear zinc metalloenzyme active sites such that the zinc ion acts as (a) an activator of the nucleophilic hydroxyl moiety or (b) stabilizes the developing negative charge on the phosphate moiety, or (c) both. We generated similar alignments for tetrahedral zinc sites. (d) Additional hydrogen bonding interactions were found in the wild type (e.g. H238 in PT3) or introduced (e.g. Q58 in PT3) using the RosettaMatch algorithm. Shape complementary interactions to maximize TS affinity were introduced using RosettaDesign. Residue numbers shown correspond to the PT3 design (see text). (e) The wild type adenosine deamination and the designed OP hydrolysis reactions.

Figure 2. Kinetic characterization of PT3

(a) Michaelis-Menten analysis on the computationally designed PT3 variant and the wild type adenosine deaminase (1A4L) indicates a k_cat/K_m=4 M⁻¹s⁻¹ for PT3. Measurements were made at an enzyme concentration of 2.5 μM for both the designed and wild type proteins. Measurements were made in triplicate, and a representative trace is shown (b) The E217Q mutation in the evolved variant PT3.1 shows 1A4L-level OP hydrolysis activity, pointing to the catalytic importance of E217. (c) Multiple turnovers (>140) are observed with PT3.3 with an enzyme concentration of 350 nM and DECP concentration of 50 μM. For estimating the maximum fluorescence signal corresponding to complete hydrolysis of DECP, bacterial phosphotriesterase (PTE) at 10 μM was used to hydrolyze the substrate under identical conditions.

Figure 3. Spatial clustering of wild type and activity-enhancing residues

Residues where computationally designed simultaneous substitutions were essential for the emergence of OP hydrolysis activity are highlighted in blue. Wild type residues retained in the most active variant of PT3 (purple), and positions where activity-enhancing mutations occur during directed evolution (green) form two separate spatial clusters. Residue sidechain identities are from the wild type crystal structure (PDB accession code: 1A4L), and the TS model is shown in grey sticks.

Figure 4. Design model and crystal structure of apo-PT3.1

(A) Superposition of the PT3.1 design model (gold) and the crystal structure (green) showing that while the overall backbone similarity is high (backbone RMSD=0.65 A), two active site proximal loops show conformational differences. (B) The PT3 design model showing alignment of the TS in the redesigned enzyme. (C) Sidechains of designed residues S19, W65, I183 and A296 observed in the crystal structure (green) adopt rotamers predicted in the designed model (gold), whereas (D) the sidechain of residue Q58 adopts a different rotamer in the crystal structure (green) compared to the designed rotamer (gold) which corresponds to a change in its χ² dihedral angle.

Figure 5. PT3 variants stereo-selectively catalyze the hydrolysis of a cyclosarin analog

(a) Release of the coumarin leaving group from ~10 μM racemic mixture of CMP-coumarin with the indicated enzyme in was monitored by measuring absorbance at 400nm (OD₄₀₀). After incubating 0.002 μM PT3 with CMP-coumarin so that OD₄₀₀ reached a plateau, the variant 3B3 of the recombinant mammalian-like serum paraoxonase that selectively hydrolyzes the R_P form of CMP-coumarin, was added. When OD₄₀₀ in presence of 3B3 reached the ~0.2 OD₄₀₀ plateau, the reaction mixture was spiked with 3D8 variant of the paraoxonase that hydrolyzes both the R_P and S_P isomers^, to detect the presence of the intact S_P isomer. The resulting increase in OD₄₀₀ to ~0.4 demonstrates the stereo-preference of PT3.3 for the R_P isomer. The observed burst phase (resulting in OD₄₀₀~0.2) and biphasic behavior with a PT3 concentration of 0.1 μM suggests that the S_P isomer is hydrolyzed by PT3, albeit at a much slower rate compared to the R_P isomer. All data were collected in triplicate and a representative trace is shown. (b) Model of the CMP-coumarin TS docked into the PT3 crystal structure. The cyclohexyl moiety occupies the open pocket of PT3 consistent with the observed preference for the R_P-isomer.