Quantum Mechanical Insights into Biological Processes at the Electronic Level - Computational Modeling of Biological Systems - page 138

Fig. 6 Theo-zymes predicted by DFT calculations for the catalysis of the Kemp elimination
reaction. These structures were then grafted onto existing protein scaffolds, and tested for catalytic
activity. The transition state of the reaction is shown in yellow .( a ) Glu is playing the role of the
catalytic base, and ( b ) the Asp-His diad is the base in this case. Adapted with permission from [ 18 ]

the specificity to the given transition state, and also the hydrophobic environment.
In Fig. 6 , two types of theo-zymes predicted on the basis of these ideas are shown.
In Fig. 6 a, Glu plays the role of the catalytic base in the action. Phe is placed for
-stacking purposes, to stabilize the negative charge developing in the transition
state and orient the substrate. Lys is placed to form a H-bond to the O atom that
acquires the negative charge. The structure in Fig. 6 b has Asp as the base, Ser as a
H-bond donor, and Trp as a -stacker. Theo-zymes such as these were installed into
existing proteins and catalytic activity was confirmed with a considerable success
rate. This demonstrates that small models of the active sites calculated using ab
intio techniques are highly relevant to the chemical actuality in the binding site of
proteins, even though long-range effects are obviously not included in the model.
Highly correlated electronic structure methods, especially multireference meth-
ods are often not needed for the ground state calculations. They are, however,
extensively used in more sophisticated calculations, such as those of excited states
and conical intersections. We withhold examples of their performance until Sects. 3
and 4 .

3 Mixed QM/MM Techniques

3.1 Theoretical Foundation

An obvious deficiency of pure ab initio calculations that one might point out
immediately is that calculations are always done on a small model, and the effect of
the environment is considered negligible. However, sometimes excluding the effect
of the larger biomolecule and solvent surrounding the model is highly undesirable,
while it is still needed to describe the small reactive part of the system with quantum
mechanical accuracy. The answer to this dilemma is often found in mixed QM/MM
approaches [ 19 , 20 ]. In QM/MM, the system is partitioned into what is thought to
be the most chemically significant part, e.g., a reactive center, and the rest of the