Research

Biological ammonia formation at the iron-molybdenum cofactor of nitrogenase

Despite intense research for the last decades, the mechanism for biological nitrogen reduction remains elusive. Only recently was the identity of the interstitial atom of the FeMo cofactor (FeMoco) clarified but many questions remain about the electronic structure of the cofactor and its connection to substrate interaction and catalysis.

In a joint experimental-theoretical study [1] the molybdenum ion in FeMoco was reassigned as Mo(III) rather than the common Mo(IV) assignment that has been in the literature since the 1980s. Using Mo XAS spectroscopy of MoFe protein and selected model compounds, combined with TDDFT-calculated spectra and analysis of the electronic structure using broken-symmetry DFT we were able to assign the molybdenum ion as a d3 Mo(III). Furthermore theoretical calculations revealed Mo in an unusual non-Hund configuration, apparently arising due to strong spin-coupling to the Fe atoms in an unusual (and currently not completely understood). A Mo XAS L-edge study further supported the Mo(III) oxidation state [2]. A computational study revisiting the Mössbauer properties of FeMoco demonstrated conclusively the charge of the cofactor as [MoFe7S9C]1- [3].

Crucial to this recent research into the FeMoco cofactor has been the comparison of the complex cofactor to heterometallic model compounds. A recent review on molecular and electronic structure aspects of FeMoco highlights the molecular and electronic similarity of FeMoco to synthetic cubanes by Holm et al. and Coucouvanis et al. [4].

Going beyond cluster models, we now employ hybrid QM/MM approaches where a large part of the MoFe protein is described, removing the need for artificial constraints on amino acids near the cofactor. In a study on MoFe protein we demonstrated that unprecedented agreement between broken-symmetry DFT calculations and the crystal structure can be obtained. Not only could the charge of FeMoco be confirmed as [MoFe7S9C]1-, but via comparison of metal-metal distances and an understanding of the electronic structure we made a case for the cofactor predominantly populating a specific electronic state, described via a broken-symmetry solution labelled BS7-235, where Fe ions 2, 3 and 5 are "spin-down" [5], see figure on the left.

Additionally vanadium nitrogenase is being studied in our group, and recently Fe XES experiments and calculations were together able to demonstrate the presence of a carbide in FeVco [6]. Furthermore Fe XAS experiments and calculations revealed a more reduced Fe component of FeVco compared to FeMoco, despite the same spin state of both cofactors [7]. This can be explained by the heterometal substitution of a d3 Mo(III) ion for a d2 V(III) ion requiring a 1-electron reduced Fe part of FeVco compared to FeMoco.

An unusual crystal structure of VFe protein recently showed a light-atom ligand replacing a sulfide bridge on FeVco, under turnover conditions (Sippel et al. Science, 359, 1484-1489). The ligand was proposed as either NH or OH based on the crystallographic analysis. Via our QM/MM approach we could demonstrate conclusively that the ligand is an OH group, likely derived from a water molecule as shown below [8].

Many mysteries remain about nitrogenase such as the site of N2 binding, the nature of FeMoco/FeVco redox states and the mechanism of N2 reduction and associated H2 evolution. These questions are currently being explored by computations in our group, together with experimental efforts going on in the department.

[1] Bjornsson, R., Lima, F. A., Spatzal, T., Weyhermueller, T., Glatzel, P., Bill, E., Einsle, O., Neese, F., and DeBeer, S. (2014) Identification of a spin-coupled Mo(III) in the nitrogenase iron-molybdenum cofactor, Chem. Sci., 5, 3096-3103.

[2] Bjornsson, R., Delgado, M., Lima, F. A. , Einsle, O., Neese, F., DeBeer. S. (2015) Molybdenum L-edges of molybdenum-dependendent nitrogenase, ZAAC, 641, 65-71.

[3] Bjornsson, R., Neese, F., DeBeer S., (2017) Revisiting the Mössbauer isomer shifts of the FeMoco cluster of nitrogenase and the cofactor charge, Inorg. Chem., 56, 1470-1477.

[4] Bjornsson, R., Neese, F., Schrock, R. R., Einsle, O., and DeBeer, S. (2015) The discovery of Mo(III) in FeMoco: reuniting enzyme and model chemistry, J. Biol. Inorg. Chem., 20, 447-460.

[5] Benediktsson, B., Bjornsson, R. (2017) QM/MM Study of the Nitrogenase MoFe Protein Resting State: Broken-Symmetry States, Protonation States, and QM Region Convergence in the FeMoco Active Site, Inorg. Chem., 57, 218-230.

[6] Rees, J. A., Bjornsson, R., Schlesier, J., Sippel, D., Einsle, O., and DeBeer, S. (2015) The Fe-V Cofactor of Vanadium Nitrogenase Contains an Interstitial Carbon Atom, Angew. Chem. Int. Ed. 54, 13249-13252.

[7] Rees, J. A., Bjornsson, R., Kowalska, J. K., Lima, F. A., Schlesier, J., Sippel, D., Weyhermueller, T., Einsle, O., Kovacs, J. A., and DeBeer, S. (2017) Comparative electronic structures of nitrogenase FeMoco and FeVco, Dalton T. 46, 2445-2455.

[8] Benediktsson, B., Thorhallsson, A. Th., Bjornsson, R. (2018) QM/MM calculations reveal a bridging hydroxo group in a vanadium nitrogenase crystal structure, Chem. Comm., DOI: 10.1039/C8CC03793K


Accounting for environmental effects in Quantum Chemistry:

Environmental effects have important effects on the molecular or electronic structure of molecular systems and usually some account of the solution or solid environment is necessary in calculations. Some molecular species are only stable in solution (e.g. zwitterions) or the solid phase and the solvent contribution is an integral part of solution properties like redox potentials or pKa values. Continuum solvation models are typically employed for including solution phase effects. For large biomolecules like proteins the QM/MM approach has been very successful and has e.g. been applied to the nitrogenase enzymes successfully by us (see above), demonstrating considerable improvement in the description of the cofactor over cluster models. For the solution and solid phases, it is often less clear how to apply QM/MM methodology due to questions of availability of forcefield parameters and accounting for adequate sampling.

A QM/MM approach for computing local structure and properties of molecules inside molecular crystals was previously described [1] and was found to satisfactorily describe molecular geometries as well as solid-state NMR properties. We are currently improving the methodology and expanding its use beyond molecular crystals.

Molecular redox potentials in solution are an example of where accounting for solvent effects is crucial. Continuum solvation models suffer from considerable errors in computed potentials, particularly for aqueous solution, likely due to the lack of explicit solvent molecules. Explicit solvation methodology, on the other hand suffers from lack of established methodology. We are currently developing an explicit solvation QM/MM-based protocol for describing redox potentials of molecules in aqueous solution, see figure to the right. The main focus is currently on accuracy and robustness and improvement over continuum models, with the next target to bring the computational cost down. This will be followed by applications to redox chemistry and spectroscopy of molecular inorganic catalysts. Our first article on this was recently published [2] where we highlighted the importance of accounting for both short-range and long-range polarization and bulk effects for such redox chemistry. Sampling was performed via efficient semi-empirical QM/MM MD. We demonstrated that our approach performs well for a large selection of 1-electron oxidation potentials of organic molecules, with a mean absolute error of 0.13 V compared to mean absolute errors of >0.2 V for continuum solvation models.

[1] Modelling Molecular Crystals by QM/MM: Self-Consistent Electrostatic Embedding for Geometry Optimizations and Molecular Property Calculations in the Solid

R. Bjornsson and M. Bühl, J. Chem. Theory Comput., 2012, 8, 498-508.

[2] A Multi-Step Explicit Solvation Protocol for Calculation of Redox Potentials

Cody M. Sterling, Ragnar Bjornsson*, J. Chem. Theory Comput. accepted, DOI:10.1021/acs.jctc.8b00982