The binding energy to the substrate will affect the Km (Michaelis-Menthen constant), but not Kcat. It may also affect product inhibition, if both the affinity to the substrate and the product are increased.

To improve the catalytic efficiency, you would have to selectively increase the binding energy to the transition state of the reaction relative to the binding energy to substrate and product, thereby decreasing the activation energy of the reaction.

Hi,

Annemarie Honegger has provided a good answer to your question.

I want to comment on whether you should use a computational approach for this question:

Theoretically, it is possible that you can predict enzyme kinetics using docking and simulations. There are many caveats to this approach.

1) You cannot predict solely based on Rg, SASA etc. These are preliminary analyses that your simulation is sane and nothing went wrong.

2) It is a time-consuming and you need to run multiple replicas to make sure the process/energies are somewhat reproducible.

3) you need to simulate the ligand binding and converting to product and then product release from the active site. You can speed up the process by enhanced sampling such as Umbrella sampling etc. During this process, you need to focus n Annemarie Honegger answer that your ligand/substrate binds with mutant enzyme quickly, processed in a short time, and the product should promptly be released without inhibiting the enzyme.

I don't recommend this approach, however, if you are at the beginning of your project, you should focus on machine learning that can give you better and (maybe) more reliable answers.

Good luck.

Of course, preferential binding to the transition state is only one way for enzymes to do catalysis, the one that can be approached by studying/optimizing binding interactions through docking and molecular dynamics. Depending on the type of reaction, active site residues play a role in acid/base catalysis, providing or abstracting protons, with second sphere ligands and helix dipoles fine-tuning the pK values of the catalytic residues , substrate and water molecules involved in the catalytic process. Zn2+ ions can also be used for this purpose.

Redox enzymes rely on prosthetic groups as electron donors and acceptors, eg copper and iron ions, with protein residues and organic compounds restricting the type of reactions these ions catalyse to the one the enzyme catalyses, and NADox/red, FADox/red cofactors. Therefore to really understand what happens in the catalytic center of an enzyme, you have to look at it in much finer details than run-of-the-mill macromolecular docking and md programs do.

Thank you for the answers, Annemarie Honegger and Ayaz Anwar

Is there a specific kind of MD simulation that allows us to visualize the conversion of substrate to its intermediate in the transition state and finally the product instead of simulating the protein-substrate, protein-intermediate and protein-product complexes separately?

Could you comment on the machine learning that can give us better and (maybe) more reliable answers? Are there any good published papers or established protocols for our reference?

There are people to do that kind of things in chemistry to understand reaction coordinates. Programs like GAUSSIAN ( https://gaussian.com ) and GAMESS ( https://www.msg.chem.iastate.edu/gamess/ ) come to mind. These ab initio quantum chemistry packages look at molecules at a much more detailed levels than macromolecular simulations do. I have no practical experience with these type of programs, you would have to talk with a computational chemist for these. They are computationally far more expensive, as they have to simulate the electronic structure of the molecules rather than the corse approximations we use in macromolecular simulations.

From the description of GAMESS's capabilities:

"GAMESS is a program for ab initio molecular quantum chemistry. Briefly, GAMESS can compute SCF wavefunctions ranging from RHF, ROHF, UHF, GVB, and MCSCF. Correlation corrections to these SCF wavefunctions include Configuration Interaction, second order perturbation Theory, and Coupled-Cluster approaches, as well as the Density Functional Theory approximation. Excited states can be computed by CI, EOM, or TD-DFT procedures. Nuclear gradients are available, for automatic geometry optimization, transition state searches, or reaction path following. Computation of the energy hessian permits prediction of vibrational frequencies, with IR or Raman intensities. Solvent effects may be modeled by the discrete Effective Fragment potentials, or continuum models such as the Polarizable Continuum Model. Numerous relativistic computations are available, including infinite order two component scalar relativity corrections, with various spin-orbit coupling options. The Fragment Molecular Orbital method permits use of many of these sophisticated treatments to be used on very large systems, by dividing the computation into small fragments. Nuclear wavefunctions can also be computed, in VSCF, or with explicit treatment of nuclear orbitals by the NEO code."

I have to confess that I do not understand half of that, it is too far out of my field of expertise

As to the use of machine learning - This is mainly a question of whether you can define and assemble good training sets. Macromolecular structure determination is usually not able to look at catalysis in progress, it usually looks at complexes with non-reactive substrate analogs or under conditions where catalysis is not possible e.g. for lack of a co-substrates.

You may want to Google "simulation of enzymatic catalysis"

Some interesting hits:

Article EnzyHTP: A High-Throughput Computational Platform for Enzyme Modeling

Article Modeling Catalysis in Allosteric Enzymes: Capturing Conforma...

Two Sides of Quantum-Based Modeling of Enzyme-Catalyzed Reactions: Mechanistic and Electronic Structure Aspects of the Hydrolysis by Glutamate Carboxypeptidase

https://www.mdpi.com/pdfhttps://www.mdpi.com/1420-3049/26/20/6280/pdf

Hi,

Just repeating an already great answer by Annemarie Honegger is of no use.

I want to share my two cents here.

MD is good (no one is perfect) in terms of reaction understanding, how the evolving reaction is evoling from substrate to product; involvement of different residues, how they react, and how those interactions supports the conversion of substrate to the product.

Machine learning (ML), on the other hand, will predict whether the mutant is more efficient than the WT (pleaser bear in mind that it is not perfect either). Plus, you need good starting (training) data (rule of thumb: more data is proportional to more reliable results). If the catalytic efficiency is the target of your study, then ML is better.

Good luck.