Reactivity of the Binuclear Non-Heme Iron Active Site of Δ⁹ Desaturase Studied by Large-Scale Multireference Ab Initio Calculations

Abstract

The results of density matrix renormalization group complete active space self-consistent field (DMRG-CASSCF) and second-order perturbation theory (DMRG-CASPT2) calculations are presented on various structural alternatives for the O-O and first C-H activating step of the catalytic cycle of the binuclear nonheme iron enzyme Δ(9) desaturase. This enzyme is capable of inserting a double bond into an alkyl chain by double hydrogen (H) atom abstraction using molecular O2. The reaction step studied here is presumably associated with the highest activation barrier along the full pathway; therefore, its quantitative assessment is of key importance to the understanding of the catalysis. The DMRG approach allows unprecedentedly large active spaces for the explicit correlation of electrons in the large part of the chemically important valence space, which is apparently conditio sine qua non for obtaining well-converged reaction energetics. The derived reaction mechanism involves protonation of the previously characterized 1,2-μ peroxy Fe(III)Fe(III) (P) intermediate to a 1,1-μ hydroperoxy species, which abstracts an H atom from the C10 site of the substrate. An Fe(IV)-oxo unit is generated concomitantly, supposedly capable of the second H atom abstraction from C9. In addition, several popular DFT functionals were compared to the computed DMRG-CASPT2 data. Notably, many of these show a preference for heterolytic C-H cleavage, erroneously predicting substrate hydroxylation. This study shows that, despite its limitations, DMRG-CASPT2 is a significant methodological advancement toward the accurate computational treatment of complex bioinorganic systems, such as those with the highly open-shell diiron active sites.

Pathways for Arene Oxidation in Non-Heme Diiron Enzymes: Lessons from Computational Studies on Benzoyl Coenzyme A Epoxidase

“…For example, in methane monooxygenase (MMO) oxidizing methane to methanol, 27 a bis(μ-oxo)-diiron(IV) species (termed intermediate Q) has been found to be responsible for methane C-H activation. 28 In contrast, soluble Δ 9 desaturase, 29 inserting a double bond into an alkyl chain by double hydrogen abstraction, seems to operate via a 1,1-μ-hydroperoxodiiron(III) intermediate, 30 whereas for myo-inositol oxygenase, 31 a superoxo-diiron(III) species is held responsible for the initial C-H cleavage. 31,32 Methane monooxygenase can oxidize π-systems of olefins and aromatics, 33 and so can several synthetic model non-heme diiron systems.…”

Protein effects in non-heme iron enzyme catalysis: insights from multiscale models

“…[40] There are also large differences between functionals when it comes to reaction barriers, more than 10 kcal/mol between B3LYP and M06 in both AlkB and Δ 9 desaturase (Δ 9 D). [41,42] The alternatives to DFT are the multiconfigurational wavefunction methods. The complete active space self-consistent field (CASSCF) method has been used to calculate structures for oxygen binding in hypoxia-inducible factor asparaginyl hydroxylase (FIH-1) and to get accurate excitation energies, meaning errors of approximately 4 kcal/mol, for a catechol dioxygenase with the inclusion of second-order perturbation theory (PT2).…”

“…[43,44] For reaction mechanisms, the outstanding example is the modeling of a large number of reaction steps in Δ 9 D using the density-matrix renormalization group (DMRG), which offers an approximate solution to a system with a large number of active orbitals. [41] However, accurate energies in transition metal systems are highly challenging also for wavefunction methods. [45] This is partly due to the need for a large basis sets.…”

“…Also Δ 9 D showed large static MM effects increasing two of the barriers by almost 10 kcal/mol, and even if the origin is not analyzed in detail, the effect would be to slow catalysis. [41] A calculation with extremely large electrostatic effects is for a proton transfer reaction in nitrile hydratase where the reaction barrier decreases by 40 kcal/mol, [82] but as other reaction steps are stabilized by an unreasonable 200 kcal/mol, without any clear protein origin, it is possible that these effects were not correctly evaluated. Other studies of QM/MM of nitrile hydratase also give more reasonable energy profiles.…”

A state interaction spin-orbit coupling density matrix renormalization group method

“…In transition metal systems, active spaces with up to 50 orbitals can be routinely converged to chemical accuracy. 26,28,29,31,34,35,[38][39][40][41]43,44,48 In this work, we describe a method to combine a treatment of spin-orbit coupling with a density matrix renormalization group description of the electronically near-degenerate individual states. We use a state interaction approach, [53][54][55][56][57][58] whereby DMRG wavefunctions are determined for spinpure electronic states, and the Hamiltonian with spin-orbit coupling is recomputed within this basis and diagonalized.…”