Ensemble-based enzyme design can recapitulate the effects of laboratory directed evolution in silico

Abstract

The creation of artificial enzymes is a key objective of computational protein design. Although de novo enzymes have been successfully designed, these exhibit low catalytic efficiencies, requiring directed evolution to improve activity. Here, we use room-temperature X-ray crystallography to study changes in the conformational ensemble during evolution of the designed Kemp eliminase HG3 (k_cat/K_M 146 M^-1s^-1). We observe that catalytic residues are increasingly rigidified, the active site becomes better pre-organized, and its entrance is widened. Based on these observations, we engineer HG4, an efficient biocatalyst (k_cat/K_M 103,000 M^-1s^-1) containing key first and second-shell mutations found during evolution. HG4 structures reveal that its active site is pre-organized and rigidified for efficient catalysis. Our results show how directed evolution circumvents challenges inherent to enzyme design by shifting conformational ensembles to favor catalytically-productive sub-states, and suggest improvements to the design methodology that incorporate ensemble modeling of crystallographic data.

Fig. 1. HG series of Kemp eliminases.

a HG enzymes catalyze the Kemp elimination reaction using a catalytic dyad consisting of a base (Asp127) that deprotonates 5-nitrobenzisoxazole, and an H-bond donor (Gln50) that stabilizes negative charge buildup on the phenolic oxygen at the transition state (^‡). This reaction yields 4-nitro-2-cyanophenol. b Directed evolution of HG3, a higher-activity mutant (S265T) of the in silico design HG2. A total of 17 mutations (shown as spheres) were introduced during evolution, including 11 at positions within or close to the active site (green) and 6 at distal sites (magenta). c Angles describing the hydrogen bonding interaction between the transition-state analogue 6-nitrobenzotriazole (6NT) and Asp127 in the HG2 (PDB ID: 3NYD) and HG3.17-E47N/N300D (PDB ID: 4BS0) crystal structures are indicated in gray and black, respectively. Values in italics are optimal angles calculated for hydrogen bonding interactions between acetamide dimers. d Cut-away view of the active-site pocket shows that its structural complementarity with 6NT (spheres) is improved in the higher-activity variant HG3.17-E47N/N300D. Key active-site residues are shown as sticks. e The HG4 variant engineered in this study contains 8 first- and second-shell mutations found during evolution.

Fig. 2. Crystal structures of HG-series Kemp eliminases.

In all cases, only atoms from chain A are shown. a Binding pose of the 6-nitrobenzotriazole (6NT) transition-state analogue (orange). Hydrogen bonds are shown as dashed lines. The red sphere represents a water molecule. The 2Fo-Fc map is shown in volume representation at two contour levels: 0.5 and 1.5 eÅ⁻³ in light and dark blue, respectively. b 6NT (orange) is sandwiched between the hydrophobic side chains of Trp44 and Met237. c The peptide bond between residues 83 and 84 can adopt cis or trans conformations. Hydrogen bonds are shown as dashed lines. The 2Fo-Fc map is shown in volume representation at two contour levels: 0.5 and 1.5 eÅ⁻³ in light and dark blue, respectively. d Conformational changes to the loop formed by residues 87–90 over the course of the evolutionary trajectory. The 2Fo-Fc map is shown in volume representation at two contour levels: 0.5 and 1.5 eÅ⁻³ in light and dark blue, respectively. e Superposition of the 6NT-bound structure (white) with the highest (magenta) and lowest (green) occupancy conformers of the unbound structure for each Kemp eliminase. From HG3 to HG3.14, the unbound state is never pre-organized for catalysis as both Trp44 and Met237 adopt conformations that would prevent the productive binding of the transition state. In HG3.17 and HG4 however, only Trp44 adopts a non-productive conformation in the unbound state, with an occupancy of 62% or 26%, respectively. f Cut-away view of the active site shows that its entrance (top) becomes widened during evolution, as indicated by an increasing bottleneck radius (reported as the average radius ± s.d. calculated using the highest occupancy conformers from both chain A and B, except for HG3.17, which contains a single chain). 6NT is shown as orange spheres. Bottleneck radii were calculated using the PyMOL plugin Caver 3.0.

Fig. 3. Conformational heterogeneity.

a B-factor Z-scores for the residue at position 50 in the absence of bound 6-nitrobenzotriazole (6NT) decrease over the course of the evolutionary trajectory, while those for Asp127 do not change significantly. Z-scores of individual side-chain heavy atoms are shown as dots (values averaged over both chain A and B for all structures except that of HG3.17, which contains a single chain in the asymmetric unit), while the average Z-score for the whole side chain is indicated by the bar. Positive and negative Z-scores indicate increased flexibility or rigidity relative to the average residue in the protein, respectively. b B-factor Z-scores of each protein residue (average of all side-chain heavy atoms) in the absence of bound 6NT plotted on a model backbone for each Kemp eliminase. Thickness of the sausage plot increases with the B-factor Z-score, indicating increased flexibility. The loop formed by residues 87–90 (boxed) becomes more rigid during evolution. Source data are provided as a source data file.

Fig. 4. Computational design of HG4 on various backbone templates.

The HG4 crystal structure with bound 6-nitrobenzotriazole (white) is overlaid on the HG4 design models (colored) obtained using the crystal structure of (a) HG4 with bound 6-nitrobenzotriazole, b Thermoascus aurantiacus xylanase 10 A (PDB ID: 1GOR), c HG3 with bound 6-nitrobenzotriazole, or (d) HG3 without 6-nitrobenzotriazole. e–h the HG4 design models obtained using the template prepared by ensemble refinement or unconstrained molecular dynamics (MD) that gave the best energy following repacking. PHOENIX energies of design models after repacking are indicated at the bottom right. For reference, the energy of the HG4 crystal structure with a bound transition state is −186.7 kcal/mol. In all cases, the transition state and transition-state analogue are shown at the center of the barrel. Side chains of all residues forming the binding pocket are shown with the exception of Ala21 and Pro45, which were omitted for clarity. The sphere shows the alpha carbon of Gly83. Asterisks indicate residues that adopt side-chain rotamers varying by >20 degrees around one or more side-chain dihedrals between the design model and crystal structure.