How Does Thymine DNA Survive Ultrafast Dimerization Damage?

The photodimerization reaction between the two adjacent thymine bases within a single strand has been the subject of numerous studies due to its potential to induce DNA mutagenesis and possible tumorigenesis in human skin cells. It is well established that the cycloaddition photoreaction takes place on a picosecond time scale along barrierless or low barrier singlet/triplet pathways. However, the observed dimerization quantum yield in different thymine multimer is considerable lower than might be expected. A reasonable explanation is required to understand why thymine in DNA is able to survive ultrafast dimerization damage. In this work, accurate quantum calculations based on the combined CASPT2//CASSCF/AMBER method were conducted to map the excited state relaxation pathways of the thymine monomer in aqueous solution and of the thymine oligomer in DNA. A monomer-like decay pathway, induced by the twisting of the methyl group, is found to provide a bypass channel to ensure the photostability of thymine in single-stranded oligomers. This fast relaxation path is regulated by the conical intersection between the bright SCT(1ππ*) state with the intra-base charge transfer character and the ground state to remove the excess excitation energy, thereby achieving the ground-state recovery with high efficiency.


(b)
Scheme S1. The chosen QM/MM partitioning: (a) the QM1 subsystem includes the thymine monomer; (b) the QM2 subsystem includes two adjacent thymine bases, while the MM subsystem includes the rest of the DNA bases, amino acid residues, crystal water molecules, and counterions. See the text for details.  The calculations of the QM parts were conducted at the complete active space self-consistent field (CASSCF) level of theory [9,10]. with the cc-PVDZ basis set. For the thymine monomer, the ab initio calculations were primarily performed at the CASSCF level of theory with a total of 14 electrons in 10 orbitals (14e/10o). The active orbitals include the O4 lone-pair orbital, all π orbitals and their corresponding π* orbitals (see Figure S2). For the thymine oligomer, 14 electrons in 11 orbitals were chosen as the active space, which includes C5-C6 (C5′-C6′) π/π* orbitals, C4-O8 (C4′-O8′) π/π* orbitals, O8 lone-pair n orbital and the delocalized π orbitals on the 5′-thymine. All of these orbitals in the active space are shown schematically in Figures S3. Geometry optimizations were performed using a 2-root state-averaged CASSCF approach (S0 and SCT, equal weights) for the SCT state and a state-specific approach for the S0 and T1 state. To consider dynamic electron correlation effects, the single-point energy of the optimized geometries in the above computations was recalculated at the multi-configuration second-order perturbation (CASPT2) level of theory [11,12] based on the zeroth-order six roots state-averaged CASSCF wave functions. These calculations were performed without an ionization potential-electron affinity (IPEA) shift but included an imaginary energy-level shift of 0.2 a.u. to avoid intruder state problems.

Vertical Excitation Energies
Vertical excitation energies, oscillator strengths and transition dipole moments to the lowest five excited singlet states of the QM part at the Franck-Condon (FC) point were computed using the CASPT2//CASSCF and CASSI//CASSCF methods at the CASSCF-optimized S0 minimum.

Optimizations of Minima, Conical Intersections and Paths
Local minima on the excited and ground states were obtained by CASSCF optimizations. The location of conical intersections and singlet/triplet crossings was assessed on the basis of the computed energy gaps for the optimized structures. At the same computational levels, the minimum energy profiles (MEPs) were mapped by intrinsic reaction coordinate (IRC) computations [13,14] to connect above critical points in several possible excited and ground states. The single point energy calculations were carried out at the CASPT2 level of theory, based on optimized geometries using the CASSCF method. Therefore, the MEPs were eventually computed at the CASPT2//IRC/CASSCF level of theory along the unbiased reaction coordinates to gain insight into how the deactivation for the thymine monomer and thymine oligomer takes place.

Packages
The CASSCF calculations were performed using GAUSSIAN03 [15]. The CASPT2 and CASSI calculations were performed using MOLCAS7.6 [16], whereas the MM calculations were conducted under the AMBER99 [3] force field using TINKER4.2 package [4]. The interface between the QM and MM parts was coded by Ferré et al. and included in the Molcas program [17].

Charge Translocation Calculations
To further explore the properties of thymine monomer and thymine oligomer in the excited state, a charge translocation calculation was performed based on Mulliken charge population and an appropriate fragment partitioning strategy. As shown in Figure S1, for the thymine monomer, the link nitrogen group and its adjacent -CH group are defined as part I, while the rest part in the ring are defined as part II. For the thymine oligomer, the unexcited thymine base are included as part II. The charge distributions were obtained using a full Mulliken population analysis at the CASPT2//CASSCF level of theory. Table S1 presents the Mulliken charge distributions of part I and II in the ground (S0) and SCT( 1 ππ*) state upon the photo-excitation of thymine monomer and thymine oligomer.

Thymine monomer
Thymine oligomer Figure S1. The scheme of fragment partition for charge translocation is shown, for the thymine monomer, the link nitrogen group and its adjacent -CH group are defined as part I, while the rest part in the ring are defined as part II. For the thymine oligomer, the unexcited thymine base are included as part II.

Selected Orbitals in the Active Space
Diagram of selected orbitals in the active space for the thymine monomer and thymine oligomer.

Optimized Structures
(a) (b) Figure S4. The structures optimized for the ground and excited states are schematically shown below: (a) the thymine monomer obtained at the CASSCF level of theory; (b) the thymine oligomer obtained at the CASSCF level of theory. Selected key bond lengths (Å) are given (see Section 7 for full Cartesian coordinates obtained at the CASSCF level of theory).  Table S5. Absolute energies (A.E., hartree), relative energies (R.E., eV/mol) and MM energies (hartree) for the optimized structures of thymine monomer along the relaxation pathway in the singlet excited state. The corresponding energy profiles are plotted in Figure 1a of the main article.  Table S6. Absolute energies (A.E., hartree), relative energies (R.E., eV/mol) and MM energies (hartree) for the optimized structures of thymine monomer along the relaxation pathway in the triplet excited state. The corresponding energy profiles are plotted in Figure 1b of the main article.  Table S9. Absolute energies (A.E., hartree), relative energies (R.E., eV/mol) and MM energies (hartree) for the optimized structures of thymine oligomer along the relaxation pathway in the triplet excited state. The corresponding energy profiles are plotted in Figure 2b in the main article.