Tests on Mandelate Racemase

Two model reactions:
We have performed several series of tests on the same model of Mandelate Racemase enzyme studied in section 2.3. In particular, we took as starting points the mechanism found with the propargylglycolate substrate. Between the different steps of the three mechanisms we have selected two of them as representative examples of opposite cases. The first one is step 1 of mechanism II, which consists of a proton transfer from Glu317 to a carboxylate oxygen of propargylglycolate. The second one is step 4 of mechanism I, which essentially involves the configuration inversion of the stereogenic carbon of propargylglycolate.

The analysis of the components of the transition vectors associated with the respective transition state structures shows the difference between both steps, which, for the sake of clarity, will be called from here on the proton transfer step and the carbon configuration inversion step. Few atoms are expected to rearrange during the proton transfer step, whose transition vector mainly includes the motion of the transferring proton and a small contribution of the proton donor oxygen atom of Glu317. That is, the changes along this step are restricted to a small local zone involving a very reduced number of atoms. Conversely, many atoms are involved in the components of the transition vector corresponding to the carbon configuration inversion step: the stereogenic carbon atom and the atoms of propargylglycolate directly attached to it, along with the proton of His297 that will be transferred to the substrate. In addition, some participation of Lys166 at the first stages of the step is also expected. Then, this step involves a rather global motion of several groups and residues including an important number of atoms.

Procedure:
In this test series we have looked for the transition state structures of the two above mentioned steps starting from the highest energy point of the profile built up along the adequate reaction coordinate as described in section 2.3.

We have preferred to test our algorithm on the location of transition states rather than for minima. The reason is because although the algorithm can be applied to both cases, the transition state search is always more problematic and this will help us to discriminate and discuss between the several options considered here.

In what follows we compare the results obtained with the micro-iterative method using the different options mentioned above. Prior to this we have to stress that some uncertainty accompanies the quantitative value of most of the results obtained in this work.^4.5The number of iterations required to reach the transition state structures is an example. The convergence of each minimization of the environment or each core search is fulfilled when the suitable convergence criterion of the gradient norm is reached. However, sometimes the low-gradient region can be attained fast, but the algorithm can spend some time in this quasi-converged region conferring some aleatory character to the crossing of the gradient norm threshold (RMS=0.005 kcal/(mol·Å)). Indeed this fact also affects the total CPU time spent in each transition state structure location. In spite of that, we think that several important qualitative trends emerge from the analysis of our numerical results. The tests were run in a 2.0 GHz Pentium IV computer.