Computational Optimization of Electric Fields for Improving Catalysis of a Designed Kemp Eliminase

Abstract

Here we report a computational method to improve efficiency of a de novo designed Kemp Eliminase enzyme KE15, by identifying mutations that enhance electric fields and chemical positioning of the substrate that contribute to free energy stabilization of the transition state.Starting from the design that has a kcat/KM of 27 M -1 s -1 , the most improved variant introduced 4 computationally targeted mutations to yield a kcat/KM of 403 M -1 s -1 , with almost all of the enzyme improvement realized through a 43-fold improvement in kcat, indicative of a direct impact on the chemical step. This work raises the prospect of computationally designing enzymes that achieve better efficiency with more minimal experimental intervention using electric field optimization as guidance. † authors contributed equally

Computational optimization of electric fields for better catalysis design

“…8 Markland and co-workers confirmed that the scaffold exerts an overall stabilizing electric field from ab initio molecular dynamics (AIMD) simulations that include Nuclear Quantum Effects (NQE) 16 13,16 ; this is because the electric field calculated in the reactant and transition states are different and cannot be factorized in front of the reaction dipole difference. 8,22 Starting from the original design with a kcat/KM of 27 M -1 s -1 , we created a computationally improved variant through introduction of 4 targeted mutations to yield a kcat/KM of 403 M -1 s -1 .…”

“…Shown is the location of the 4 mutations of KE15 best variant (left) and the total electric field contributions to the transition state relative to the original design (right). 22 The positive direction for the bond dipole is the same as for the electric field (from C to H, from N to C and from O to N for the CH, CN and ON bond respectively with magnitude +1 , +0.4 and +2.3 Debye). The efficiency of these computationally designed variants was measured and confirmed experimentally.…”

“…The most desirable solution is a full ab initio molecular dynamics (AIMD) treatment of large catalytic environments for thousands of atoms and hundreds of nanoseconds (or longer), but this is well beyond current capabilities of AIMD. Bridging this length and time scale gap will be realized through algorithmic developments that would decrease the number of cycles in the self-consistent cycle of AIMD 74 , substantially increase the time step in a stable manner 75 , achieving linear scaling in an accurate and reliable way 76 , as well as coupling AIMD to the Grand Canonical ensemble 77 , and use of advanced sampling methods 78 such as the low energy alternative side chain arrangements of an enzyme using Monte Carlo (MC) to create QM/MC approaches to determine fluctuating electric fields 22,79 . The emergence of more detailed molecular in situ and operando catalytic experiments 80…”

Impact of long-range electrostatic and dispersive interactions on theoretical predictions of adsorption and catalysis in zeolites

“…The results summarized here may also potentially be relevant to the in silico design [156] of zeolite adsorbents and catalysts, where different zeolite frameworks, could be compared based whether they provide the ideal long-range electrostatic and dispersion interactions, which lead to the most effective stabilization of adsorbed substrates and TSs. In a similar vein, recent studies on electric field optimization in enzymes have emphasized on the importance of electrostatics [157] in the protein scaffolding surrounding the active site, and have even reported computational strategies to improve the efficiency of electric field catalysis in de novo designed Kemp eliminases [158]. Lastly, we note from a theoretical perspective, that Energy Decomposition Analysis (EDA) and Principal Component Analysis (PCA) [118] are promising tools with regards to deconvoluting the contributions of longrange electrostatics and dispersion in selective TS stabilization.…”

What Are We Missing by Not Measuring the Net Charge of Proteins?

“…[4] The systematic absence of experimentally determinedv alues of Z has likely impeded ar igorous understanding of most chemicalp rocesses in which proteinsa re involved including aggregation and self-assembly, [20][21][22][23][24][25][26] ligand binding, [27][28][29][30][31][32][33][34] catalysis, [35][36][37][38][39] electron transfer, [3,6,[40][41][42][43][44][45][46][47] protein crystallization, [14,48] analytical separation, [49,50] and protein engineering. [51][52][53][54][55][56] It is tempting to assume that the formal net chargeo faprotein predicted from generalized residue pK a values (Z seq )issosimilar to the actual net charge that any difference is irrelevant, and the isoelectric point tells us all we need to know about ap rotein's net charge. Thep K a of individual amino acid residues in folded proteins can vary by severalu nits (possibly as many as 10 units) from textbook values because of solventb urial, dipole-dipole interactions, or metal coordination.…”