Accurate Stabilities of Laccase Mutants Predicted with a Modified FoldX Protocol

Abstract

Fungal laccases are multicopper enzymes of industrial importance due to their high stability, multifunctionality, and oxidizing power. This paper reports computational protocols that quantify the relative stability (ΔΔG of folding) of mutants of high-redox-potential laccases (TvLIIIb and PM1L) with up to 11 simultaneously mutated sites with good correlation against experimental stability trends. Molecular dynamics simulations of the two laccases show that FoldX is very structure-sensitive, since all mutants and the wild type must share structural configuration to avoid artifacts of local sampling. However, using the average of 50 MD snapshots of the equilibrated trajectories restores correlation (r ∼ 0.7−0.9, r 2 ∼ 0.49−0.81) and provides a root-mean-square accuracy of ∼1.2 kcal/mol for ΔΔG or 3.5 °C for T 50 , suggesting that the time-average of the crystal structure is recovered. MD-averaged input also reduces the spread in ΔΔG, suggesting that local FoldX sampling overestimates free energy changes because of neglected protein relaxation. FoldX can be viewed as a simple "linear interaction energy" method using sampling of the wild type and mutant and a parametrized relative free energy function: Thus, we show in this work that a substantial "hysteresis" of ∼1 kcal/mol applies to FoldX, and that an improved protocol that reverses calculations and uses the average obtained ΔΔG enhances correlation with the experimental data. As glycosylation is ignored in FoldX, its effect on ΔΔG must be additive to the amino acid mutations. Quantitative structure−property relationships of the FoldX energy components produced a substantially improved laccase stability predictor with errors of ∼1 °C for T 50 , vs 3−5 °C for a standard FoldX protocol. The developed model provides insight into the physical forces governing the high stability of fungal laccases, most notably the hydrophobic and van der Waals interactions in the folded state, which provide most of the predictive power.

Towards a “Golden Standard” for computing globin stability: Stability and structure sensitivity of myoglobin mutants

“…The sensitivity of a method's output to the protein structure used for calculation is a major issue, since a choice of crystal structure can substantially affect the computed results 25,26,28 . Since crystal structures vary substantially in terms of resolution and refinement quality, missing residues, heteroatom presence, crystal symmetry, and conditions (pH, salt, T), this is a major challenge to protein modeling that is not yet widely explored.…”

“…A C C E P T E D M A N U S C R I P T ACCEPTED MANUSCRIPT 4 There are particular advantages with using myoglobin as a test case: Due to its special role in protein science, a large range of high-resolution crystal structures are available with reduced noise, which otherwise impairs both routine computation of protein stability and benchmarks on larger, more diverse data sets 25,26 . Also, myoglobin is small (153 residues) and is an all-helix protein, providing perspective to the recently studied all-beta protein superoxide dismutase 1, which also contains 153 residues and is a widely studied protein with several high-resolution crystal structures available 28 .…”

Structural investigation of the cold-adapted acylaminoacyl peptidase from Sporosarcina psychrophila by atomistic simulations and biophysical methods

“…In particular, dynamically coupled residues are involved in protein structural stability and activity [26][27][28][29][30][31][32] as also shown by mutational studies [30,31,[33][34][35][36]. NMR data indicate that the motions on the nanosecond (ns) or longer timescales are equally important.…”

Exploring the Protein Stability Landscape: Bacillus subtilis Lipase A as a Model for Detergent Tolerance

“…[25][26][27] The FoldX algorithm was developed on the basis of a database of 1030 thermo-stabilizing single-site substitutions [34] (for applications see refs. [28][29][30]). The 3DM database uses a combination of tools to collect available data across superfamilies of proteins.…”