Item

Abstract

Mouesca et al. (1) criticize our analysis (2) of the superoxidized proximal [4Fe-3S]⁵⁺ cluster in O₂-tolerant hydrogenases (A) that was also studied in Volbeda et al. (3), albeit from a different organism. Both analyses were based on broken-symmetry (BS) density functional theory (DFT) calculations (2, 3), which presently is the only feasible alternative for the treatment of complex systems like A. However, BS-DFT is a crude approach and we regard the differences in refs. 2 and 3 as subtle rather than major.
The critique raised in Mouesca et al. (1) that our cluster model was too small and incorporated spurious constraints is unfounded. In Volbeda et al. (3), only three additional amino acid residues in the third coordination sphere were included, but their importance was not demonstrated. Furthermore, our constraints allowed for sufficient structural relaxation and the link atoms used in the quantum mechanical/molecular mechanical investigation (3) imposed almost identical constraints.
The allegation that our selection of BS solution Ox2_24 over OX2_14 (favored in ref. 3) on the basis of “chemical intuition” and “weak isomer shift considerations” is factually wrong. Instead, we analyzed relative energies and four Mössbauer parameters (δ, ΔE_Q, η, A) for each iron site, whereas in Volbeda et al. (3) only energetics and unsigned ΔE_Q values were used. The energy difference in favor of OX2_14 (3) is below 2 kcal/mol, which is well within the uncertainty of BS-DFT. Thus, we have presented more solid evidence for our choice than the authors of ref. 3.
Because our quantum mechanical model was not inferior to that of Volbeda et al. (3), we have no reason to revoke deprotonation of GLU82. Our choice was derived from a comprehensive comparison of a wide range of spectroscopic parameters, and no conclusive evidence against it was presented in Mouesca et al. (1).
The authors of refs. 1 and 3 argue in favor of OX2_14 as leading to an explanation of the cluster’s reactivity. However, no sound conclusions on transition states or mechanisms can be drawn from the analysis of the magnetic properties of one reactant only. Thus, the argument is invalid.
Mouesca et al. (1) argue that OX2_24 emerges from unrecognized, unwanted charge migration, which leads to “local spin state 3/2, instead of 5/2, for Fe2,” and therefore is an “artificially trapped electronic state.” We note that: (i) the magnetic properties are well explained by OX2_24; (ii) the reactivity argument is invalid; and (iii) the local spin is not an observable and BS-DFT does not conserve the total spin but only its projection onto the z axis. The assignment of oxidation states and local spins in refs. 1 and 3 relies on differences as small as 0.1 electrons in Mulliken spin populations. In our experience, differences of this order are not conclusive. Importantly, the lower spin population on Fe2 in OX2_24 is simply a result of spin-canting, as well as (partial) formation of covalent chemical bonds between spin-carrying fragments. Related effects are predominant in BS-DFT calculations on antiferromagnetic clusters. The claim that our OX2_24 has S_Fe2 = 3/2 is therefore unsupported and unrealistic.
In conclusion, although we do not exclude future refinements of our understanding of the [4Fe-3S] cluster, the arguments brought forward in Mouesca et al. (1) are insufficient to invalidate our original analysis (2).