Role of Electron-Driven Proton-Transfer Processes in the Ultrafast Deactivation of Photoexcited Anionic 8-oxoGuanine-Adenine and 8-oxoGuanine-Cytosine Base Pairs

It has been reported that 8-oxo-7,8-dihydro-guanosine (8-oxo-G), which is the main product of oxidative damage of DNA, can repair cyclobutane pyrimidine dimer (CPD) lesions when incorporated into DNA or RNA strands in proximity to such lesions. It has therefore been suggested that the 8-oxo-G nucleoside may have been a primordial precursor of present-day flavins in DNA or RNA repair. Because the electron transfer leading to the splitting of a thymine-thymine pair in a CPD lesion occurs in the photoexcited state, a reasonably long excited-state lifetime of 8-oxo-G is required. The neutral (protonated) form of 8-oxo-G exhibits a very short (sub-picosecond) intrinsic excited-state lifetime which is unfavorable for repair. It has therefore been argued that the anionic (deprotonated) form of 8-oxo-G, which exhibits a much longer excited-state lifetime, is more likely to be a suitable cofactor for DNA repair. Herein, we have investigated the exited-state quenching mechanisms in the hydrogen-bonded complexes of deprotonated 8-oxo-G− with adenine (A) and cytosine (C) using ab initio wave-function-based electronic-structure calculations. The calculated reaction paths and potential-energy profiles reveal the existence of barrierless electron-driven inter-base proton-transfer reactions which lead to low-lying S1/S0 conical intersections. The latter can promote ultrafast excited-state deactivation of the anionic base pairs. While the isolated deprotonated 8-oxo-G− nucleoside may have been an efficient primordial repair cofactor, the excited states of the 8-oxo-G−-A and 8-oxo-G−-C base pairs are likely too short-lived to be efficient electron-transfer repair agents.


Introduction
The photoinduced dynamics of biological chromophores have been extensively studied in the past two decades. Within this class of organic chromophore systems, the most notable include DNA and RNA nucleobases [1][2][3][4][5][6][7][8][9][10][11][12][13][14][15], nucleosides [1,[16][17][18][19][20][21][22][23][24][25][26], and base pairs [27][28][29][30][31][32][33][34][35][36]. Despite strongly absorbing in the near-UV, DNA and RNA nucleobases exhibit a remarkable degree of photostability, although the generation of photoinduced lesions in DNA strands is not fully suppressed [37][38][39]. It is generally believed that the mechanism of the observed photostability of the building blocks of DNA is ultrafast internal conversion of excited state populations to the electronic ground state with the excess energy being dissipated to the surrounding environment as heat. For the isolated nucleobases, there is consensus that internal conversion is mediated by low-lying conical intersections (CIs) which involve excited singlet states of ππ* and/or nπ* character as well as the S 0 state and become accessible by out-of-plane deformations of six-membered aromatic rings. These CIs dominate . The optimized ground-state energy of the HG2 base pair is found to be lower than that of the HG1 base pair by 0.05 eV, which indicates a Boltzmann population of 13:87 for HG1:HG2. For the 8-oxo-G − -C base pair, three lowenergy H-bonded configurations have been optimized. Among these, the structure shown in Figure  1f is the lowest-energy conformer. 8-oxo-G − can pair with A via Hoogsteen (HG) base pairing in two low-energy conformations [67]: HG1 and HG2 (see Figure 1). 8-oxo-G − can also form a stable pair with cytosine in a structure involving two hydrogen bonds, see Figure 1f. EDPT processes in the neutral 8-oxo-G-A and 8-oxo-G-C base pairs were investigated by Kumar and Sevilla with time-dependent density functional (TD-DFT) calculations [67]. In the present work, we focus on EDPT reactions in the anionic 8-oxo-G − -A and 8-oxo-G − -C base pairs. Since anionic 8-oxo-G − has been shown to have a substantially longer excited-state lifetime than neutral 8-oxo-G, the former appears better suited for light-driven DNA repair reactions than the short-lived neutral 8-oxo-G. It is therefore of interest to explore how base-pairing with A or C affects the excited-state lifetime of anionic 8-oxo-G − . As well as base pairing with cytosine (i.e., the complementary base to G), 8-oxoG − is also well-known to form mismatched base pairing with A via a Hoogsteen configuration [68][69][70]. This propensity for forming Hoogsteen base pairs with adenine is due to the enhanced redox potential of 8-oxoG − compared with that of natural guanine.
Using ab initio wave-function based electronic-structure calculations, we identify the EDPT reaction paths leading to CIs through which the excited-state population can internally convert to S 0 . Our findings provide evidence for barrierless EDPT reaction paths and therefore likely highly efficient excited-state deactivation of the 8-oxo-G − -A and 8-oxo-G − -C base pairs. The efficient excited-state deactivation of the base pairs enhances their photostability, but inevitably also lowers their repair efficiency. Figure 1 presents the MP2/cc-pVDZ-optimized ground-state structures of 9H-adenine (a); 8-oxo-G − (b); HG1 (c) and HG2 (d) 8-oxo-G − -A conformers; cytosine (e); and the 8-oxo-G − -C base pair (f). In both 8-oxo-G − -A HG base pairs, all atoms are in a common plane with the exception of the wagging angle of the amino group of 8-oxo-G − . Adenine retains a planar geometry since the amino group of adenine is involved in the inter-base hydrogen bonding, while in the structure of isolated adenine there is some pyramidization of the amino group.

Ground State Geometries
In the HG1 and HG2 8-oxo-G − -A base pairs, 8-oxo-G − and A act both as hydrogen-bond donors and as hydrogen-bond acceptors. There are two hydrogen bonds in the HG1 base pair: The HG2 base pair also has also two hydrogen bonds, N10−H12(A)•••O11(8-oxo-G − ) and N7−H13(8-oxo-G − )•••N7(A). The calculated hydrogen-bond lengths are included in Figure 1. In the HG1 base pair, the length of N10−H11•••O11(1.596 Å) is shorter than that of N7−H13•••N1(1.842 Å). In contrast, in the HG2 base pair, N10−H12•••O11(1.817 Å) is longer than N7−H13•••N7(1.694 Å). The optimized ground-state energy of the HG2 base pair is found to be lower than that of the HG1 base pair by 0.05 eV, which indicates a Boltzmann population of 13:87 for HG1:HG2. For the 8-oxo-G − -C base pair, three low-energy H-bonded configurations have been optimized. Among these, the structure shown in Figure 1f is the lowest-energy conformer. Table 1 lists the calculated vertical excitation energies and corresponding oscillator strengths (in parentheses) of the lowest four singlet excited states of the 8-oxoG − containing base pairs presently studied. For comparison, the analogous vertical excitation energies of isolated cytosine, 9H-adenine and 8-oxo-G − are presented in Table 2. In addition to Table 1, the reader is directed to Figure 2, which depicts the orbitals and orbital promotions associated with the formation of the lowest four excited electronic states of isolated 9H-adenine, cytosine, 8-oxo-G − , as well as the 8-oxo-G − -A and 8-oxo-G − -C base pairs.  As is well known, the lowest four excited states of 9H-adenine are of 1 nπ*, 1 ππ*(L b ), 1 ππ*(L a ), and 1 nπ* character. The S 1 and S 4 states involve promotion of an electron from an in-plane nitrogen 2p y orbital to a ring-centered π* orbital, while the S 2 and S 3 states involve π*←π orbital promotions which are delocalized over the aromatic rings. For 8-oxo-G − , the lowest four excited states are 1 ππ*(S 1 ), 1 nπ*(S 2 ), 1 ππ*(S 3 ), and 1 nπ*(S 4 ) in nature. As shown in Figure 2, the S 2 and S 4 states involve the promotion of an electron from the nitrogen 2p y orbital to a ring-centered anti-bonding π* orbital. The S 1 and S 3 states involve excitation from a ring-centered π orbital to the lowest π* orbital.

Vertical Excitation Energies
In the two 8-oxo-G − -A HG base pairs, the orbital promotions are almost the same and the lowest four excited states are of 1 ππ* character. The S 1 state involves electron promotion from the ring-centered π HOMO localized on 8-oxo-G − to the ring-centered π* LUMO localized on adenine, leading to a charge-separated state of CT character. The S 2 state involves a π*←π promotion, whereby both orbitals are localized on 8-oxo-G − . The S 3 state involves a π*←π promotion localized on adenine. The S 2 and S 3 states are therefore locally-excited (LE) states on 8-oxo-G − and A, respectively. The vertical excitation energies of the S 2 and S 3 states are comparable to that of isolated 8-oxo-G − and adenine, respectively. As can be seen in Table 1, the vertical excitation energy of the S 2 ( 1 ππ*) state of the HG1/HG2 base pair (4.85 eV/4.89 eV) is nearly equal to that of the analogous LE state (i.e., the first 1 ππ* state) of isolated 8-oxo-G − (4.92 eV), suggesting that pairing of 8-oxo-G − with adenine has little effect on the lowest LE 1 ππ* state energy. In contrast, the vertical excitation energies of all electronic states show a significant red-shift upon complexation relative to that of isolated adenine. Similar to the S 1 state, the S 4 state also is of CT character, involving the transition from a π orbital of 8-oxo-G − to an π* orbital of adenine, with a much higher excitation energy. The vertical excitation energies of the two HG base pairs are very similar and lower than those of isolated adenine and 8-oxo-G − . Compared with the TD-DFT results for neutral 8-oxo-G-A HG base pair [67], the lowest 1 ππ* transition localized on 8-oxo-G − is blue-shifted, while the lowest 1 ππ* transition localized on adenine is red-shifted. Figure 2f depicts the orbital promotions associated with the 8-oxoG − -C base pair. As shown, the electronic excitation to S 1 involves a π to π* electron promotion in which the former is localized on the 8-oxoG − moiety, whereas the latter is localized on the C moiety. As with the 8-oxo-G − -A base pairs, electronic excitation to S 1 involves a significant charge separation and is thus of CT character. In contrast, electronic excitation to the S 2 , S 3 , and S 4 states involves electron promotions between π/n to π* orbitals that are localized on the same nucleobase within the base pair. The observed orbital ordering is very similar to that of the WC-type G-C base pair which also has an S 1 state of CT character, whilst the higher-lying states are of LE character [34,35].

Electron-Driven Proton-Transfer Decay Paths
In order to study the intrinsic photophysical properties of the two HG base pairs, we explored the details of the potential-energy (PE) profiles along possible inter-base electron and proton transfer paths. The HG1 and HG2 base pairs have two potential reaction paths for proton transfer, one involving the transfer of a proton from adenine to 8-oxo-G − along the N10−H11•••O11 or

Electron-Driven Proton-Transfer Decay Paths
In order to study the intrinsic photophysical properties of the two HG base pairs, we explored the details of the potential-energy (PE) profiles along possible inter-base electron and proton transfer paths. The HG1 and HG2 base pairs have two potential reaction paths for proton transfer, one involving the transfer of a proton from adenine to 8-oxo-G − along the N10−H11•••O11 or N10−H12•••O11 hydrogen bonds, the other involving the transfer of a proton from 8-oxo-G − to adenine along the N7−H13•••N1 or N7−H13•••N7 hydrogen bonds (as indicated by the arrows in Figure 1c,d).
In order to study the energetics associated with a particular excited-state electron/proton transfer reaction, we computed the PE profiles along the R N10−H11 , R N7−H13 bond-stretching coordinates for the HG1 base pair and along the R N10−H12 , R N7−H13 bond-stretching coordinates for the HG2 base pair. The results are depicted in Figures 3 and 4, respectively. In these figures, the filled black circles represent the S 0 energy profile calculated along the reaction path optimized in the S 0 state for the specific R N−H driving coordinate. The energy of the unrelaxed 1 ππ* CT state (henceforth 1 ππ*(uCT)) of the base pairs, calculated at the S 0 -relaxed geometries, is designated by the profile plotted with the open red circles. The filled red circles represent the energy of the lowest inter-base CT state along the proton-transfer relaxed scan optimized for this state. The curve plotted with open black circles gives the energy of the S 0 state calculated along the minimum-energy reaction path determined in the CT state.
path. The corresponding energy profile is shown in Figure 3b. This energy profile exhibits no barrier, which ensures that the minimum-energy path connecting the 1 ππ*(uCT) and 1 ππ*(CT) states is barrierless. The relaxed 1 ππ*(CT) profile (Figure 3a, full red circles) shows a strong decrease of the energy as a function of RN7−H13 stretching; this represents the driving force towards proton transfer that results from the charge-separated character of the 1 ππ*(CT) state, which is characteristic of EDPT [71]. The S0 energy computed at the 1 ππ*(CT)-relaxed geometries (Figure 3a, black open circles) increases as a function of the RN7−H13 stretching coordinate. As a result, the energies of the 1 ππ*(CT) state and the S0 state cross at RN7−H13 ≈ 1.45 Å. This S1/S0 crossing becomes a CI when the appropriate coupling modes are taken into account. Depending on the topography of the PE surfaces and the nonadiabatic coupling at the CI, the reaction can lead to internal conversion to the S0 state of the complex (adiabatic path) or to a biradical.
The other possible pathway by which inter-base proton transfer can occur in the HG1 base pair is along the N10−H11•••O11 hydrogen bond (Path 2, see inset in Figure 3c). Figure 3c shows the PE profiles of the S0, 1 ππ*(uCT) and 1 ππ*(CT) states along the RN10−H11 driving coordinate. The LIIC path connecting the 1 ππ*(uCT) and 1 ππ*(CT) states (not shown) exhibits no barrier. As for Path 1, the energy profile along the minimum-energy path from the 1 ππ*(uCT) state to the 1 ππ*(CT) state is barrierless. However, the 1 ππ*(uCT) and 1 ππ*(CT) energies do not cross along RN10−H11 (Figure 3c), in contrast to the energies along RN7−H13. This result can easily be rationalized. While the electron transfer occurs from 8-oxo-G − to A, the proton has to move in the opposite direction, from A to 8-oxo-G − , which is not energetically favorable. The S1(CT) state is therefore not stabilized by the transfer of the proton and the EDPT mechanism does not apply for Path 2 in the HG1 base pair.  The PE profiles of the lowest excited states of 1 ππ* (uCT) and 1 ππ* (CT) character of the HG2 base pair as a function of the RN7−H13 are shown in Figure 4a. In the HG2 base pair, there likewise exists a proton-transfer path (N7−H13•••N7), which leads to a low-lying S1/S0 CI, and a proton-transfer path (N10−H12•••O11) which does not lead to a CI. The mechanistic details of the N7−H13•••N7 reaction path (Path 1, see inset in Figure 4a) are similar to those described for the Path 1 in the HG1 base pair. The relaxed 1 ππ* (CT) state exhibits a distinct driving force for proton transfer and its energy crosses the S0 energy along this path at RN7−H13 ≈ 1.30 Å, giving rise to a CI, representing a route by which either ultrafast IC to the ground state or biradical formation can occur. The linearly interpolated reaction path connecting the 1 ππ* (uCT) and 1 ππ* (CT) states is shown in Figure 4b. The energy profile exhibits a substantial barrier of approximately 0.5 eV, which represents an upper limit to the reaction barrier along the minimum-energy path. The access of the photoexcited HG2 8-oxo-G − -A base pair to the S1/S0 CI may thus be kinetically hindered.
The proton-transfer path along the RN10−H11 driving coordinate in the HG2 base pair is similar to Path 2 in the HG1 base pair. Figure 4c shows the PE profiles of the S0, 1 ππ*(uCT) and 1 ππ*(CT) states along the RN10−H11 driving coordinate. While the LIIC path connecting the 1 ππ*(uCT) and 1 ππ*(CT) states (not shown) exhibits no barrier, the 1 ππ*(uCT) and 1 ππ*(CT) energies do not exhibit a crossing along RN10−H11, as is shown in Figure 3c. As in the HG1 base pair, the Coulomb attraction after electron transfer from 8-oxo-G − to adenine renders the proton transfer from 8-oxo-G to the adenine anion (Path 1) favorable, while it renders the proton transfer from the adenine anion to 8-oxo-G − (Path 2) unfavorable.   Figure 3a) in the HG1 base pair. The S 0 energies calculated along the reaction path relaxed in the S 0 state rise steadily upon R N7−H13 bond extension, showing that proton transfer is unfavorable in this electronic state. When the energy of the 1 ππ*(uCT) state is optimized for fixed R N7−H13 = 1.2 Å, the electronic character of this state changes from LE character to CT character, which implies the transfer of an electron localized on 8-oxo-G − to the π* orbital localized on 9H-adenine, resulting in an electronic charge separation. The path connecting the 1 ππ*(uCT) state at R N7−H13 = 1.0 Å to the 1 ππ*(CT) state at its optimized geometry for R N7−H13 = 1.2 Å was constructed as a linearly interpolated reaction path. The corresponding energy profile is shown in Figure 3b. This energy profile exhibits no barrier, which ensures that the minimum-energy path connecting the 1 ππ*(uCT) and 1 ππ*(CT) states is barrierless. The relaxed 1 ππ*(CT) profile (Figure 3a, full red circles) shows a strong decrease of the energy as a function of R N7−H13 stretching; this represents the driving force towards proton transfer that results from the charge-separated character of the 1 ππ*(CT) state, which is characteristic of EDPT [71]. The S 0 energy computed at the 1 ππ*(CT)-relaxed geometries (Figure 3a, black open circles) increases as a function of the R N7−H13 stretching coordinate. As a result, the energies of the 1 ππ*(CT) state and the S 0 state cross at R N7−H13 ≈ 1.45 Å. This S 1 /S 0 crossing becomes a CI when the appropriate coupling modes are taken into account. Depending on the topography of the PE surfaces and the nonadiabatic coupling at the CI, the reaction can lead to internal conversion to the S 0 state of the complex (adiabatic path) or to a biradical.
The other possible pathway by which inter-base proton transfer can occur in the HG1 base pair is along the N10−H11•••O11 hydrogen bond (Path 2, see inset in Figure 3c). Figure 3c shows the PE profiles of the S 0 , 1 ππ*(uCT) and 1 ππ*(CT) states along the R N10−H11 driving coordinate. The LIIC path connecting the 1 ππ*(uCT) and 1 ππ*(CT) states (not shown) exhibits no barrier. As for Path 1, the energy profile along the minimum-energy path from the 1 ππ*(uCT) state to the 1 ππ*(CT) state is barrierless. However, the 1 ππ*(uCT) and 1 ππ*(CT) energies do not cross along R N10−H11 (Figure 3c), in contrast to the energies along R N7−H13 . This result can easily be rationalized. While the electron transfer occurs from 8-oxo-G − to A, the proton has to move in the opposite direction, from A to 8-oxo-G − , which is not energetically favorable. The S 1 (CT) state is therefore not stabilized by the transfer of the proton and the EDPT mechanism does not apply for Path 2 in the HG1 base pair. Figure 3. PE profiles of the ground state and the lowest singlet excited states of the HG1 8-oxo-G − -A base pair along RN7−H13 (a) and RN10−H11 (c) proton-transfer coordinates; (b) shows the energy profiles of the LIIC path connecting S1(uCT) with S1(CT) in (a).
The PE profiles of the lowest excited states of 1 ππ* (uCT) and 1 ππ* (CT) character of the HG2 base pair as a function of the RN7−H13 are shown in Figure 4a. In the HG2 base pair, there likewise exists a proton-transfer path (N7−H13•••N7), which leads to a low-lying S1/S0 CI, and a proton-transfer path (N10−H12•••O11) which does not lead to a CI. The mechanistic details of the N7−H13•••N7 reaction path (Path 1, see inset in Figure 4a) are similar to those described for the Path 1 in the HG1 base pair. The relaxed 1 ππ* (CT) state exhibits a distinct driving force for proton transfer and its energy crosses the S0 energy along this path at RN7−H13 ≈ 1.30 Å, giving rise to a CI, representing a route by which either ultrafast IC to the ground state or biradical formation can occur. The linearly interpolated reaction path connecting the 1 ππ* (uCT) and 1 ππ* (CT) states is shown in Figure 4b. The energy profile exhibits a substantial barrier of approximately 0.5 eV, which represents an upper limit to the reaction barrier along the minimum-energy path. The access of the photoexcited HG2 8-oxo-G − -A base pair to the S1/S0 CI may thus be kinetically hindered.
The proton-transfer path along the RN10−H11 driving coordinate in the HG2 base pair is similar to Path 2 in the HG1 base pair. Figure 4c shows the PE profiles of the S0, 1 ππ*(uCT) and 1 ππ*(CT) states along the RN10−H11 driving coordinate. While the LIIC path connecting the 1 ππ*(uCT) and 1 ππ*(CT) states (not shown) exhibits no barrier, the 1 ππ*(uCT) and 1 ππ*(CT) energies do not exhibit a crossing along RN10−H11, as is shown in Figure 3c. As in the HG1 base pair, the Coulomb attraction after electron transfer from 8-oxo-G − to adenine renders the proton transfer from 8-oxo-G to the adenine anion (Path 1) favorable, while it renders the proton transfer from the adenine anion to 8-oxo-G − (Path 2) unfavorable.

8-oxo-G − -C
We now turn our attention to the 8-oxoG − -C base pair.The PE profiles along the RN-H driving coordinate are depicted in Figure 5. As with 8-oxoG − -A, the base pair under consideration exhibits two possible inter-molecular proton-transfer paths along hydrogen bonds as depicted in the insets in Figure 5a,c. Path 1, which involves PT from the N-H donor group of 8-oxo-G − to the N acceptor group of C, shows a barrierless profile with respect to EDPT on S1 (Figure 5a). Along this coordinate, the decreasing energy of the S1 state is accompanied by an increasing energy of the S0 state, which leads to an S1/S0 curve crossing at RO-H ≈ 1.2 Å. At this crossing, the excited-state population can return to the S0 state-providing enhanced photostability of the 8-oxoG − -C base pair-or a radical pair can be formed. The energy profiles along the LIIC path connecting the 1 ππ*(uCT) state to the 1 ππ*(CT) state are shown in Figure 5b. This energy profile exhibits no barrier and leads in fact to an S1/S0 energy crossing. This result ensures that the minimum-energy path connecting the 1 ππ*(uCT) and 1 ππ*(CT) states is barrierless and that the S1/S0 crossing seam is easily accessible from the Franck-Condon region of the S1(uCT) state.
An S1/S0 crossing is not observed for the second possible proton-transfer path (energy profiles depicted in Figure 5c), although the overall gradients of the S0 and S1 energy profiles mimick those The PE profiles of the lowest excited states of 1 ππ* (uCT) and 1 ππ* (CT) character of the HG2 base pair as a function of the R N7−H13 are shown in Figure 4a. In the HG2 base pair, there likewise exists a proton-transfer path (N7−H13•••N7), which leads to a low-lying S 1 /S 0 CI, and a proton-transfer path (N10−H12•••O11) which does not lead to a CI. The mechanistic details of the N7−H13•••N7 reaction path (Path 1, see inset in Figure 4a) are similar to those described for the Path 1 in the HG1 base pair. The relaxed 1 ππ* (CT) state exhibits a distinct driving force for proton transfer and its energy crosses the S 0 energy along this path at R N7−H13 ≈ 1.30 Å, giving rise to a CI, representing a route by which either ultrafast IC to the ground state or biradical formation can occur. The linearly interpolated reaction path connecting the 1 ππ* (uCT) and 1 ππ* (CT) states is shown in Figure 4b. The energy profile exhibits a substantial barrier of approximately 0.5 eV, which represents an upper limit to the reaction barrier along the minimum-energy path. The access of the photoexcited HG2 8-oxo-G − -A base pair to the S 1 /S 0 CI may thus be kinetically hindered.
The proton-transfer path along the R N10−H11 driving coordinate in the HG2 base pair is similar to Path 2 in the HG1 base pair. Figure 4c shows the PE profiles of the S 0 , 1 ππ*(uCT) and 1 ππ*(CT) states along the R N10−H11 driving coordinate. While the LIIC path connecting the 1 ππ*(uCT) and 1 ππ*(CT) states (not shown) exhibits no barrier, the 1 ππ*(uCT) and 1 ππ*(CT) energies do not exhibit a crossing along R N10−H11 , as is shown in Figure 3c. As in the HG1 base pair, the Coulomb attraction after electron transfer from 8-oxo-G − to adenine renders the proton transfer from 8-oxo-G to the adenine anion (Path 1) favorable, while it renders the proton transfer from the adenine anion to 8-oxo-G − (Path 2) unfavorable.

8-oxo-G − -C
We now turn our attention to the 8-oxoG − -C base pair.The PE profiles along the R N-H driving coordinate are depicted in Figure 5. As with 8-oxoG − -A, the base pair under consideration exhibits two possible inter-molecular proton-transfer paths along hydrogen bonds as depicted in the insets in Figure 5a,c. Path 1, which involves PT from the N-H donor group of 8-oxo-G − to the N acceptor group of C, shows a barrierless profile with respect to EDPT on S 1 (Figure 5a). Along this coordinate, the decreasing energy of the S 1 state is accompanied by an increasing energy of the S 0 state, which leads to an S 1 /S 0 curve crossing at R O-H ≈ 1.2 Å. At this crossing, the excited-state population can return to the S 0 state-providing enhanced photostability of the 8-oxoG − -C base pair-or a radical pair can be formed. The energy profiles along the LIIC path connecting the 1 ππ*(uCT) state to the 1 ππ*(CT) state are shown in Figure 5b. This energy profile exhibits no barrier and leads in fact to an S 1 /S 0 energy crossing. This result ensures that the minimum-energy path connecting the 1 ππ*(uCT) and 1 ππ*(CT) states is barrierless and that the S 1 /S 0 crossing seam is easily accessible from the Franck-Condon region of the S 1 (uCT) state.

8-oxo-G − -C
We now turn our attention to the 8-oxoG − -C base pair.The PE profiles along the RN-H driving coordinate are depicted in Figure 5. As with 8-oxoG − -A, the base pair under consideration exhibits two possible inter-molecular proton-transfer paths along hydrogen bonds as depicted in the insets in Figure 5a,c. Path 1, which involves PT from the N-H donor group of 8-oxo-G − to the N acceptor group of C, shows a barrierless profile with respect to EDPT on S1 (Figure 5a). Along this coordinate, the decreasing energy of the S1 state is accompanied by an increasing energy of the S0 state, which leads to an S1/S0 curve crossing at RO-H ≈ 1.2 Å. At this crossing, the excited-state population can return to the S0 state-providing enhanced photostability of the 8-oxoG − -C base pair-or a radical pair can be formed. The energy profiles along the LIIC path connecting the 1 ππ*(uCT) state to the 1 ππ*(CT) state are shown in Figure 5b. This energy profile exhibits no barrier and leads in fact to an S1/S0 energy crossing. This result ensures that the minimum-energy path connecting the 1 ππ*(uCT) and 1 ππ*(CT) states is barrierless and that the S1/S0 crossing seam is easily accessible from the Franck-Condon region of the S1(uCT) state.
An S1/S0 crossing is not observed for the second possible proton-transfer path (energy profiles depicted in Figure 5c), although the overall gradients of the S0 and S1 energy profiles mimick those observed in Figure 5a. The respective decrease and increase of the energies of S1(CT) and S0 are too weak to lead to a degeneracy of the S1 and S0 energies. As in the HG1 and HG2 base pairs of 8-oxo-G − -A, there exists no substantial driving force for EDPT from cytosine to the 8-oxo-guanine anion.

General Discussions and Conclusions
We explored the excited-state reaction paths and PE profiles associated with coupled electron/proton transfer reactions in the two most stable hydrogen-bonded conformers of the 8-oxo-G − -A base pair as well as in the lowest-energy conformer of the 8-oxo-G − -C base pair. In both cases, the 8-oxo-G moiety was assumed to be in its deprotonated (anionic) form which is found in aqueous An S 1 /S 0 crossing is not observed for the second possible proton-transfer path (energy profiles depicted in Figure 5c), although the overall gradients of the S 0 and S 1 energy profiles mimick those observed in Figure 5a. The respective decrease and increase of the energies of S 1 (CT) and S 0 are too weak to lead to a degeneracy of the S 1 and S 0 energies. As in the HG1 and HG2 base pairs of 8-oxo-G − -A, there exists no substantial driving force for EDPT from cytosine to the 8-oxo-guanine anion.

General Discussions and Conclusions
We explored the excited-state reaction paths and PE profiles associated with coupled electron/proton transfer reactions in the two most stable hydrogen-bonded conformers of the 8-oxo-G − -A base pair as well as in the lowest-energy conformer of the 8-oxo-G − -C base pair. In both cases, the 8-oxo-G moiety was assumed to be in its deprotonated (anionic) form which is found in aqueous solution at pH > 7. In the 8-oxo-G − -A HG1 base pair as well as in the 8-oxo-G − -C base pair, the calculated PE profiles reveal the existence of a barrierless path for EDPT from 8-oxo-G − to A or C, leading to a low-lying S 1 /S 0 conical intersection which can promote ultrafast excited-state deactivation. In the 8-oxo-G − -A HG2 base pair, on the other hand, a low barrier may exist on the S 1 PE surface which may possibly kinetically hinder the access of this conformer to the S 1 /S 0 CI. We did not find evidence for the existence of S 1 /S 0 conical intersections along reaction paths for proton transfer from adenine or cytosine to the 8-oxo-G − anion in any of the three base pairs. The EDPT reactions revealed in the present work for the 8-oxo-G − -A HG1 and 8-oxo-G − -C base pairs are rather similar to those identified earlier in the G-C and A-T WC base pairs [33,34].
These results are of relevance for the current discussion on the potential role of 8-oxo-G as a photo-repair agent in DNA, possibly being a precursor of modern flavine cofactors [54,55,72]. It is firmly established that the photo-excited state of neutral 8-oxo-G has a sub-picosecond lifetime in aqueous solution, while deprotonated 8-oxo-G − exhibits a much longer fluorescence lifetime of 43 ps [26]. The drastic shortening of the excited-state lifetime of 8-oxo-G − relative to neutral 8-oxo-G has been explained by either CIs intrinsic to guanine, which are more easily accessible in the neutral than in the anionic form [57], or by an EDPT reaction along the H-bond between guanine and ribose in 8-oxo-guanosine, which is available in the neutral form, but not in the anionic form [23]. It has been speculated that the long lifetime of anionic 8-oxo-G − should be favorable for repair by electron transfer in the excited state, while the very short excited-state lifetime of neutral 8-oxo-G should be detrimental in this respect [26]. Herein, we have found computational evidence for presumably very efficient excited-state deactivation via barrierless EDPT reactions leading to S 1 /S 0 conical intersections in the 8-oxo-G − -A and 8-oxo-G − G base pairs which call the concept of repair of CPD lesions via electron transfer from excited-state 8-oxo-G − in DNA oligomers into question. Kumar and Sevilla investigated the corresponding EDPT paths in the neutral 8-oxo-G-A and 8-oxo-G-C base pairs and found a path with a barrierless PE profile en route to a low-lying S 1 /S 0 conical intersection in the 8-oxo-G-C base pair, while no such path was found for the 8-oxo-G-A base pair [67]. This finding led Kumar and Sevilla to the conclusion that the 8-oxo-G-A base-pair, due to its longer excited-state life time, should allow for efficient repair of CPD lesions. However, the very short intrinsic lifetime of neutral 8-oxo-guanosine, not considered by Kumar and Sevilla, renders it unlikely that the neutral 8-oxo-G-A base pairs are efficient repair agents in DNA oligomers. Notwithstanding, we do however stress that isolated nucleobasic or nucleosidic forms of 8-oxoG-may be efficient at repairing CPD lesions-as advocated by Matsika and co-workers [26] and Tuna et al. [23].
There exist additional complexities in a bulk DNA environment which are not taken into account in the present calculations. Electrostatic and dispersive interactions between stacked DNA bases may modify the topographies of the PE profiles and the locations and energies of CIs. Nonetheless, the present calculations for isolated base pairs are useful as they can serve as a starting point for forthcoming studies which include the effect of complex environments, albeit at a more approximate level of theory.

Computational Methods
The ground-state minimum-energy geometries of the 8-oxo-G − -A base pairs, in the two HG conformations, and of the 8-oxo-G − -C base pair were optimized at the MP2/cc-pVDZ level of theory [73,74]. At these ground-state minimum-energy geometries, the vertical excitation energies and oscillator strengths of the lowest four singlet excited states were calculated using the ADC(2) method. [75]. In the MP2 and ADC(2) calculations, the resolution of the identity (RI) approximation was employed in the evaluation of the electron repulsion integrals [76].
The reaction path for inter-base hydrogen-atom transfer from 8-oxo-G − to adenine in the electronic ground state was calculated as a relaxed scan at the MP2 level using R N-H of the hydrogen-bonded NH group of 8-oxo-G − as the driving coordinate. This involves scanning of the appropriate R N-H driving coordinate, while allowing the rest of the nuclear framework to relax. The energies of the 1 ππ* excited states along the relaxed ground-state path were computed using the ADC(2) method.
Relaxed scans along R N-H were also computed for the lowest excited state of charge-transfer (CT) character using the ADC(2) method. In these calculations, the energies of the electronic ground state and the uCT 1 ππ* states were computed at the relaxed geometries of the CT state using the MP2 and ADC(2) methods, respectively. When calculations of relaxed scans were not possible due to failure of excited-state geometry optimization, an approximate reaction path was constructed by linear interpolation in internal coordinates (LIIC). All calculations were carried out with Turbomole [77].