Protein Modeling

Introduction

Due to the extensive nature of plant metabolic flux and limited available energy, improving production efficiency is one of the key challenges when synthesizing resource products using the natural chassis of plants. Enhancing enzyme catalytic activity is a common and effective strategy. The catalytic activity of HCT2, derived from red clover, is limited. Its affinity for caffeoyl-CoA is weak, and it tends to bind to feruloyl-CoA, producing a large amount of feruloyl malate byproducts. Enhancing the enzyme's catalytic activity towards caffeoyl-CoA can also increase substrate specificity, directing more metabolic flux towards the production of caffeoyl malate.

On the other hand, resveratrol has significant potential in medicine, and its market is rapidly expanding. However, its relatively low yield limits further development. STS uses p-coumaroyl-CoA and malonyl-CoA as substrates to produce resveratrol, functioning in plant synthesis chassis. Optimizing STS can enhance resveratrol production efficiency, further unlocking its medical value.

To optimize HCT2 and STS and increase the production efficiency of caffeoyl malate and resveratrol, we applied a semi-rational design approach to model these two proteins. We identified beneficial mutants with improved catalytic activity and thermal stability.

HCT2(HCT-M) and STS Structure

HCT2 Structure

(1) Catalytic Mechanism: The catalytic mechanism of HCT2 shares similarities with plant acyltransferases. Acyl acceptors and donors first bind to the active site via a solvent channel. The basic histidine residue (His-159) deprotonates the hydroxyl or amino group of the acyl acceptor, forming an oxyanion. The oxyanion then nucleophilically attacks the carbonyl carbon atom of the acyl donor, forming a tetrahedral intermediate. Finally, the intermediate releases a CoA-SH molecule, forming a new ester or amide. In this process, His-159 plays a key role in initiating the reaction, while Thr-384 and Trp-386 stabilize the tetrahedral transition state through hydrogen bonding. Other amino acids also contribute to maintaining the protein’s conformation.

(2) Key Conserved Domain: The BAHD acyltransferase domain is crucial for HCT2. Studies have shown that directed evolution of this domain can enhance the binding affinity between mutants and ligands, improving the enzyme's stability in vitro.

(3) Active Pocket: Using Protein Plus, we calculated HCT2's active pocket, discovering a large cavity in the center of its structure. The surrounding amino acid sequences are close to the BAHD-AT domain, indicating that this cavity is likely the active pocket of HCT2.

STS Structure

STS has a conserved cysteine residue at its center, which is the binding site for p-coumaroyl-CoA. Research has shown that the differences in STS activity are determined by the N-terminal domain (Residues 5-228). The active site is located between residues 156-172.

Screen for Mutation Hotspots

We used pre-trained protein language models and a shared protein interaction design platform to screen for single-point mutation hotspots. Files were uploaded to Hotspot Wizard 3.0, which used homology modeling to generate mutant structures. The model quality was evaluated using WHAT_CHECK, PROCHECK, and MolProbity, and mutants' thermodynamic stability was assessed using Rosetta and FoldX.

We also uploaded the wild-type protein files to SaprotHub, which performed computer-simulated single-point saturation mutations. Scores were based on a combination of protein stability, homology, and ligand affinity using FLIP, TAPE, and PEER methods. We prioritized mutants with higher SaprotHub scores, as this scoring system better reflects real enzymatic reactions. We carefully examined the domain locations of these mutants, excluding those far from the active pocket.

Molecular Docking

After identifying mutation hotspots, we performed molecular docking to evaluate the rationality of the docked structures, Vina scores, and the number of polar bonds in the protein-ligand complex. Mutants with unreasonable docking results (e.g., ligands docking outside the active pocket or p-coumaroyl-CoA docking too far from STS's central cysteine residue) were excluded.

Two HCT2 mutants were eliminated from this round of screening, T162F and A419E, T162F was eliminated due to low Vina Score and too few polar bonds, and A419E was eliminated due to the fact that the docking results after the mutation showed that caffeoyl-coenzyme A has a stronger affinity for the docking pocket around the molecule than with the docking pocket at the center of the molecule, which may result in the inability of caffeoyl-coenzyme A to interact properly with the BAHD-AT structural domain interaction, resulting in the inability of HCT2 to catalyze the substrate reaction; three STS mutants were also eliminated in this round, namely A205H, P269E, and A182Q; A182Q and P269E were eliminated because of too few polar bonds, and A205H was eliminated because of the stronger affinity of p-coumaroyl-coenzyme A for docking to the outer than to the central pockets, while the The p-coumaroyl binding site of STS is in the center of the molecule proper, and thus may result in the inability of the enzymatic reaction to occur properly.

Molecular Dynamics Simulation

To further refine the scale of STS and HCT2 mutants and reduce the burden of wet-lab experiments, we conducted molecular dynamics simulations to better reflect energy changes in the system. Stability was measured using RMSD and RMSF, and system activity was represented by the total potential energy changes and their convergence time. Systems with excessive RMSD, RMSF, or total potential energy were eliminated.

By combining molecular docking and dynamics simulations, we narrowed down the final selection of mutants for further experimental testing.