Fragment-Based Calculations of Enzymatic Thermochemistry Require Dielectric Boundary Conditions

Abstract

Electronic structure calculations on enzymes require hundreds of atoms to obtain converged results, but fragment-based approximations offer a cost-effective solution. We present calculations on enzyme models containing 500–600 atoms using the many-body expansion, comparing to benchmarks in which the entire enzyme–substrate complex is described at the same level of density functional theory. When the amino acid fragments contain ionic side chains, the many-body expansion oscillates under vacuum boundary conditions but rapid convergence is restored using low-dielectric boundary conditions. This implies that full-system calculations in the gas phase are inappropriate benchmarks for assessing errors in fragment-based approximations. A three-body protocol retains sub-kilocalorie per mole fidelity with respect to a supersystem calculation, as does a two-body calculation combined with a full-system correction at a low-cost level of theory. These protocols pave the way for application of high-level quantum chemistry to large systems via rigorous, ab initio treatment of many-body polarization.

Water-Mediated Charge Transfer and Electron Localization in a Co3Fe2 Cyanide-Bridged Trigonal Bipyramidal Complex

“…Crystallographic data demonstrate that the temperature-dependent magnetic properties of Co 3 Fe 2 are associated with changes in oxidation state at the Co 2 and Co 3 sites, whereas Co 1 (located at the bottom in Figure ) remains high-spin Co(II). , In accord with the crystal structure that was used as a starting point, we obtain Co–N bond lengths that are ∼0.2 Å longer at the Co 1 site as compared to Co 2 or Co 3 , characteristic of a class I or class II system. , However, this could also be consistent with a class II/III system, specifically with readily switchable spin states based on small fluctuations in geometry, which are not sampled in these static calculations. Despite concerns that vacuum-exposed anionic moieties (such as CN – ) can cause DFT convergence problems, , or that vacuum boundary conditions exaggerate charge delocalization (tending toward class III), we obtain similar geometries when using a polarizable continuum model − to add low-dielectric boundary conditions.…”

Eliminating Imaginary Vibrational Frequencies in Quantum-Chemical Cluster Models of Enzymatic Active Sites

“…With that in mind, we note that it is often suggested that the influence of dielectric boundary conditions wanes as the size of the QM-cluster model increases, ,, especially for models with ≳150 atoms . In our own work using sizable enzyme models, , including some with ionic side chains, we observe that enthalpy changes and barrier heights computed using ε = 2 or ε = 4 are virtually indistinguishable from results obtained using much larger dielectric constants. However, results for ε = 2 are distinguishable from those obtained using vacuum boundary conditions (ε = 1).…”

“…For some calculations, the conductor-like polarizable continuum model (C-PCM) was used to incorporate dielectric boundary conditions. − For these calculations, the solute cavity was constructed from atomic spheres with radii 1.2 times larger than those in the modified Bondi set. − This cavity was discretized using the switching/Gaussian algorithm, ,, with 110 Lebedev grid points for each hydrogen and 194 Lebedev points for other atomic spheres. Note that low-dielectric boundary conditions (with ε = 2–4) can be critically important for converging SCF calculations on large protein models with ionic side chains. , …”

Roadmap on electronic structure codes in the exascale era

“…Alongside the conventional results are data from a fragment-based approach called pp-GMBE(2) [119], which does not require any single calculation that is larger than four amino acids yet provides conformational energy profiles that are faithful to the macromolecular result. Thermochemistry [120] and non-covalent interactions [121,122] can also be reproduced with high fidelity, via fragmentation. By the nature of the approximation, wall-clock time for the protein calculations in figure 23 can be reduced to the cost of a single subsystem calculation if sufficient hardware is available.…”