Quantum-chemical cluster modeling is challenged in the limit of large, soft systems by the effects of dispersion and solvent, and well as other physical interactions. Adenosylcobalamin (AdoCbl, coenzyme B-12), as one of the most complex cofactors in life, constitutes such a challenge. The cleavage of its unique organometallic Co-C bond has inspired multiple studies of this cofactor. This paper reports the fully relaxed potential energy surface of Co-C cleavage of Ado Cbl, including for the first time all side-chain interactions with the dissociating Ado group. Various methods and corrections for dispersion, relativistic effects, solvent polarity, basis set superposition error, and thermal and vibrational effects were investigated, totaling more than SSO single-point energies for the large model. The results show immense variability depending on method, including solvation, functional type, and dispersion, challenging the conceived accuracy of methods used for such systems. In particular, B3LYP-D3 seems to severely underestimate the Co-C bond strength, consistent with previous results, and BP86 remains accurate for cobalamins when dispersion interactions are accounted for.
Journal of Physical Chemistry Part A: Molecules, Spectroscopy, Kinetics, Environment and General Theory, 2014, Vol 118, Issue 34, p. 7104-7117