Computational Stabilization of a Non-Heme Iron Enzyme Enables Efficient Evolution of New Function

Abstract

Deep learning tools for enzyme design are rapidly emerging, and there is a critical need to evaluate their effectiveness in engineering workflows. Here we show that the deep learning-based tool ProteinMPNN can be used to redesign Fe(II)/αKG superfamily enzymes for greater stability, solubility, and expression while retaining both native activity and industrially relevant non-native functions. This superfamily has diverse catalytic functions and could provide a rich new source of biocatalysts for synthesis and industrial processes. Through systematic comparisons of directed evolution trajectories for a non-native, remote C(sp³)-H hydroxylation reaction, we demonstrate that the stabilized redesign can be evolved more efficiently than the wild-type enzyme. After three rounds of directed evolution, we obtained a 6-fold activity increase from the wild-type parent and an 80-fold increase from the stabilized variant. To generate the initial stabilized variant, we identified multiple structural and sequence constraints to preserve catalytic function. We applied these criteria to produce stabilized, catalytically active variants of a second Fe(II)/αKG enzyme, suggesting that the approach is generalizable to additional members of the Fe(II)/αKG superfamily. ProteinMPNN is user-friendly and widely accessible, and our results provide a framework for the routine implementation of deep learning-based protein stabilization tools in directed evolution workflows for novel biocatalysts.

Keywords: Biocatalysis; Directed evolution; Hydroxylation; Protein design; non-heme iron(II) α-ketoglutarate-dependent oxygenase.

Figure 1.

Initial whole-cell reaction screen data with a panel of Fe(II)/αKG amino acid hydroxylases and free acid substrate analogues. Each enzyme was screened against all three substrates. A white panel indicates that product was not detectable (n.d.), which is the case for all reactions except tP4H with substrate 1. Reactions were performed in whole cell from 50 mL expression cultures where whole cell volume was 1/20^th the expression volume. Reactions were carried out in MOPS (pH 7.0, 50 mM) with 20 mM substrate, 60 mM αKG (as disodium salt), 1 mM ferrous ammonium sulfate, and 1 mM L-ascorbic acid.

Figure 2.

Validation of tP4H activity with free acid 1. (A) Reaction of tP4H with substrate 1 to form trans-4-hydroxycyclohexane carboxylic acid 4. (B) Yield of 4 after reaction of 1 with tP4H variants, as well as negative control reaction with Fe(II) and bovine serum albumin (BSA). The Fe 1 mM control was run in the absence of added enzyme. Purified enzyme concentration was varied between 10–40 μM with 20 mM 1, 40 mM αKG, 1 mM ferrous ammonium sulfate, and 1 mM L-ascorbic acid in MES buffer (50 mM, pH 6.8). Reactions were carried out for 24 hours at 25 °C and quantified with analytical LC-MS. (C) Structural model of the tP4H showing key active site residues. Fe(II), αKG, and L-Pip were modeled in Chimera.^[43]

Figure 3.

Stability and activity of wild-type tP4H and ProteinMPNN design R2_11. (A) Flowchart of the ProteinMPNN workflow that successfully produced stabilized variants that retained catalytic activity. See Supporting Information for detailed guidelines for each computational step. (B) tP4H structure (Alphafold2 model, Figure S9) color-coded to show sites fixed in the design process (blue, Supplementary spreadsheet – ProteinMPNN sequences_metrics) and sites mutated in the ProteinMPNN R2_11 redesign (orange-red). Sites colored black were neither fixed nor redesigned in the R2_11 variant. Side chains for first shell active site residues (Table S4) are shown in blue. (C) Temperature-dependent CD spectroscopy of wild-type tP4H and R2_11. T_m values were calculated using the Boltzmann sigmoid function in GraphPad Prism. (D) Activity-stability analysis of wild-type tP4H. (E) Activity-stability analysis of R2_11. For D and E, relative activity was determined using PBP assay described in Supplementary Information. Values are mean ± SD for three replicates.

Figure 4.

(A) Reaction scheme for C-H hydroxylation of substrate 1 with tP4H, R2_11, and associated variants. (B) Directed evolution of wild-type tP4H. (C) Directed evolution of stabilized variant R2_11. Reactions were carried out for 24 hr at 25 °C using purified enzyme (10–20 μM) in MES buffer (50 mM, pH 6.8), with 20 mM cyclohexane carboxylic acid 1, 40 mM αKG, 1 mM ferrous ammonium sulfate, and 1 mM ascorbic acid. Concentration of 4 in quenched reaction samples was quantified by analytical LC-MS analysis. For B and C, values are mean ± SD for three replicates.

Figure 5.

(A) Temperature dependent CD of the R2_11 and tP4H parent enzymes and triple mutants. T_m values were calculated using the Boltzmann sigmoid function in GraphPad Prism. (B) TTN for formation of 4 (d.r. 4:1, Figure S4) with the R2_11 and tP4H triple mutants at two different temperatures. Reactions were carried out for 6 hrs. at 25 °C and 35 °C using purified enzyme (15 μM) in MES buffer (50 mM, pH 6.8), with 20 mM cyclohexane carboxylic acid 1, 40 mM αKG, 1 mM ferrous ammonium sulfate, and 1 mM ascorbic acid. Values are mean ± SD for three replicates.