Optimization of Conical Intersections in Biological Systems with the ONIOM Scheme: Projected Gradient Implementation in the Microiteration Process

Abstract

In recent years, the photochemical and photophysical properties of DNA have been a widely studied topic, as this gives a new insight at explaining DNA mutation by light. One fundamental theoretical element for the characterization of these processes are the conical intersections. The intersections between the ground and excited states which have been described up to now for the isolated nucleobases but not for the bases embedded in the large DNA system. For this reason, in this work a conical intersection optimization algorithm has been implemented in the ONIOM scheme, which will allow the study of biological systems such as DNA or proteins. First, we have studied the photophysics of thymidine-phosphate in the gas phase and in solution, and we conclude that it is a good model system to test the application. The procedure used to improve the original conical intersection optimization method was to use a projected gradients in the microiteration steps of the ONIOM(QM:MM) scheme in order to avoid losing the energy degeneracy during the optimization. The gradient projection was first tested on an independent subroutine and then implemented in a development version of the Gaussian program. The CASSCF and AMBER methods are employed for the QM and MM calculations, respectively. This implementation was tested on the thymidine-phosphate model. Finally, we have evaluated the results obtained for the implementation during the first optimization step, which includes the different microiteration cycles. The implementation performs properly. At the end of the optimization step, the energy difference is smaller when the projected gradient is used, in line with our goal. At the same time, the absolute energy obtained in the first optimization step including the projected gradient is somewhat higher than in the case where the gradient was not projected. The present results provide a first test of the new algorithm, which will be tested in the future with the more demanding case of thymidine-phosphate in solution