Photoinduced water–chromophore electron transfer causes formation of guanosine photodamage

Abstract

UV-induced photolysis of aqueous guanine nucleosides produces 8-oxo-guanine and Fapy-guanine, which can induce various types of cellular malfunction. The mechanistic rationale underlying photodestructive processes of guanine nucleosides is still largely obscure. Here, we employ accurate quantum chemical calculations and demonstrate that an excited-state non-bonding interaction of guanosine and a water molecule facilitates the electron-driven proton transfer process from water to the chromophore fragment. This subsequently allows for the formation of a crucial intermediate, namely guanosine photohydrate. Further (photo)chemical reactions of this intermediate lead to the known products of guanine photodamage.

---

Introduction

Ultraviolet-driven chemical reactions in nucleic acid components could entail irreversible damage to their molecular structure, and consequently may result in undesirable mutagenic processes in biological systems.^1–7 Accumulation of such alterations in the genome can induce cellular malfunction, tumorigenesis, and even cell death.^8,9 Therefore, understanding the photochemical mechanisms directly enabling the formation of DNA/RNA lesions is a fundamental issue for cellular biology and disease prevention. 8-oxo-guanine and 2,6-diamino-4-oxo-5-formamidopyrimidine (Fapy-guanine) are among the most frequent DNA/RNA photolesions, which are produced upon UV excitation of guanine nucleosides in relative yields strongly dependent on the local chemical environment.^10–14 However, despite considerable efforts to scrutinize the molecular mechanisms underlying UV-induced damage of pyrimidine nucleobases,^15–21 photochemical processes causing generation of guanine nucleoside lesions (such as 8-oxo-guanine or Fapy-guanine) are still obscure. In this paper, we provide a plausible explanation based on the reinterpretation of available experimental data and our state-of-the-art ab initio calculations.

In recent decades, the photodynamics of guanine nucleosides was extensively investigated through time-resolved (TR) spectroscopic measurements that were often supplemented by quantum-chemical calculations.^22–28 Joint experimental–theoretical studies demonstrated that the photoinduced dynamics of aqueous guanine nucleosides is characterized by sub-picosecond excited-state lifetimes that were assigned to the population of the ¹ππ* excited state. Indeed, the proposed ballistic photorelaxation pathways most likely enable the observed ultrafast deactivation of nucleosides, but this does not explain the origin of guanine photolesions. In addition to the recorded short-lived signals, a complementary longer-lived and weakly-emissive (∼2 ps) excited state has been identified and tentatively assigned to the charge-transfer ¹πσ* excited state.^22,23 However, the accessibility of such a repulsive state in guanine nucleosides seems rather unlikely as there are no distinctive signatures in the TR absorption spectra^22,23 originating from solvated electrons (expected in the range of 600–700 nm) that could be produced through the population of the ¹πσ* electronic state.²⁹ Importantly, this puzzling excited state efficiently quenches the strongly emissive ¹ππ* states and its lifetime varies in different polar solvents such as water and methanol.^22,23

Recent advances in mechanistic studies of aqueous photochemistry of pyrimidine and purine nucleobases^30–32 have shown that the dark ¹nπ* excited states can serve as a doorway to UV-induced water-to-chromophore electron transfer. This process is enabled by an excited-state interaction of the lone electron pairs of a heteroatom in the aromatic ring and the neighbouring water molecule. Consequently, the photoinduced electron transfer and the subsequent proton transfer may lead to the formation of reactive hydroxyl radicals causing damage to DNA/RNA bases.^30–32

Our working hypothesis is that the population of the dark ¹n_Oπ* state in a complex of guanine nucleoside and an explicit water molecule (Guo-H₂O, see Fig. 1) is attainable through the excited-state intermolecular C6–O⋯OH₂ chalcogen bonding interaction, which triggers the long-sought photochemical pathway to the formation of both 8-oxo-guanosine and Fapy-guanosine.

Fig. 1 The geometries of the Guo-H₂O complex corresponding to the S₀ and S₁ (n_Oπ*) minimum-energy structures are shown in the first row. Below, the molecular orbitals involved in the S₁ (n_Oπ*) transition are shown, which demonstrate a charge-transfer character of the process in which 0.11 e⁻ is transferred from the water molecule to the nucleoside. The assumed numbering of atoms and the relevant distances between atoms are indicated.

The results of our quantum-chemical calculations are supported by comparison with previously recorded time-resolved femtosecond transient absorption spectra obtained for aqueous guanine nucleosides.^22,23

Computational methods

To find the most stable structures of nucleosides in an aqueous environment, the conformational space was explored using the Conformer-Rotamer Ensemble Sampling Tool³³ (CREST) at the semi-empirical GFN2-xTB³⁴ level of theory. The exploration of the conformational space was performed using the iMTD-GC scheme along with the analytical linearized Poisson–Boltzmann (ALPB) implicit solvent model representing the water solvent. The equilibrium ground-state geometries and harmonic vibrational frequencies of the lowest-energy conformer were then obtained using Kohn–Sham density functional theory (KS-DFT) employing the range-separated hybrid ωB97X-D functional³⁵ and the def2-TZVPP³⁶ basis set. The KS-DFT calculations were performed using the Gaussian 16³⁷ package.

Vertical excitation energies were obtained using the spin-component scaled variant³⁸ of the second-order algebraic diagrammatic construction^39,40 [SCS-ADC(2)] with the cc-pVTZ basis set⁴¹ [SCS-ADC(2)/cc-pVTZ], assuming the equilibrium ground-state structures optimized as described above. The orbital character of each electronic transition was assigned based on natural transition orbitals (NTOs). The estimation of solvent-to-chromophore transferred charge was obtained using the one-electron transition density matrix (1TDM) and the Löwdin style analysis. Tg 1TDM analysis was performed employing the TheoDore⁴² 2.4 package. Excited-state minimum-energy structures were located at the SCS-ADC(2)/cc-pVTZ level of theory. The optimization of minimum-energy crossing points (MECPs) between two electronic states was performed employing the sequential penalty constrained function implemented by Levine et al.⁴³ in the CIOpt package. To locate MECPs, the energies and analytical nuclear gradients for the electronic excited and ground states were computed at the SCS-ADC(2) and SCS-MP2 levels, respectively, with the cc-pVTZ basis set.

To verify excited-state minimum energy structures and the S₁/S₀ state crossing obtained using the single reference method, we optimized these critical excited-state geometries with the extended multi-state complete active space second-order perturbation theory^44,45 (XMS-CASPT2), employing the state-averaged complete active space self-consistent field (SA-CASSCF) method with the cc-pVDZ basis set. To locate the S₁ (n_Oπ*) minimum-energy structure at the XMS-CASPT2 level, the complete active space (CAS) was composed of molecular orbitals (MOs) having natural orbital occupations in the range of 0.02–1.98,⁴⁶ augmented by n_O, σ_O, and orbitals to facilitate the C6–O⋯OH₂ excited-state interaction in the S₁ (n_Oπ*) state and the possible elongation of the carbonyl bond. Consequently, the active space was built out of one σ_O, three π and one n_O occupied MOs along with three π* and one virtual MOs (10 electrons in 9 orbitals, shown in Fig. S10, ESI†). The SA-CASSCF wave function was averaged over four lowest-lying electronic states. The optimization of the S₁/S₀ minimum-energy conical intersection was performed assuming the CAS constructed of the π_O, two π, two n_O, and four π* MOs (10 electrons in 9 orbitals, shown in Fig. S11, ESI†). In the latter calculations, we included two lone electron pair n_O orbitals in the CAS to allow for an appropriate description of the hydroxyl radical. The SA-CASSCF wave function was averaged over three lowest-lying electronic states. The vertical shift, i.e., the empirical correction applied to the zeroth-order Hamiltonian, was equal to 0.4 Hartree, and the XMS-CASPT2 calculations were conducted using the BAGEL 1.2.0⁴⁷ package. To estimate the emission energy of the S₁ (n_Oπ*) structure, SCS-ADC(2) theory was used alongside the conductor-like screening model (COSMO) as an implicit solvent, assuming the equilibrium solvation limit for the S₁ excited state, and SCS-ADC(2) calculations were performed using the TURBOMOLE 7.3⁴⁸ package.

Excited-state absorption spectra of S₁ minimum-energy structures were simulated based on calculations of 14 excited states, which provided the excitation energies and corresponding oscillator strengths at the SCS-ADC(2)/cc-pVTZ level of theory. Absorption lineshapes, assuming the full width at half-maximum equal to 3000 cm⁻¹, were generated using the GaussSum 3.0⁴⁹ program.

The potential energy (PE) profiles for the singlet electron-driven proton transfer (EDPT) photorelaxation pathway were computed at the SCS-ADC(2)/cc-pVTZ level of theory using the key excited-state structures, namely S₂/S₁ and S₁/S₀ MECPs and the S₁ minimum-energy structure. The PE profiles between the Franck–Condon region and the S₁ minimum were obtained by single-point calculations of intermediate geometries constructed by the image dependent pair potential (IDPP) interpolation⁵⁰ between the optimized geometries. To show PE profiles between the S₁ minimum and S₁/S₀ MECP, relaxed excited-state geometry optimization was performed constraining the H–O bond length (0.97–1.27 Å) in the N7⋯H–OH moiety at the SCS-ADC(2)/cc-pVTZ level. The IDPP interpolation was also employed to obtain intermediate structures between the optimized constrained S₁ geometry, having a H–O bond distance of 1.27 Å, and the S₁/S₀ MECP. The optimization of ground-state structures of photoproducts was performed using the ωB97X-D functional³⁵ with the def2-TZVPP⁵¹ basis set. Relative energies of the photoproducts were determined using the polarizable continuum model (PCM) as an implicit solvent model, assuming their gas-phase optimized geometries. The IDPP interpolations were performed using the Orca 4.2.1.⁵² package.

Results and discussion

Ground-state structures

Based on the conformational analysis of Guo, performed using the semi-empirical GFN2-xTB method³⁴ (see Fig. S1 in the ESI†), we found that the structure having the syn orientation of the nucleobase and the C2′-endo arrangement of the sugar ring is energetically preferred in an aqueous environment. Our results agree with previous molecular dynamics simulations of guanosine monophosphate that showed dominant population of the syn-like conformer (occurring in 93% of simulations).²⁶ It is worth noting that the syn orientation enables the formation of the OH⋯N3 hydrogen bond between the sugar moiety and the nucleobase (see Fig. 1). This intramolecular hydrogen bond was previously shown to allow for an excited-state forward–backward proton transfer mechanism between the nucleobase and sugar, which could serve as a radiationless deactivation channel of excited purine nucleosides.^53–55

We added a single explicit water molecule to the selected structure of Guo, to saturate hydrogen bonds with potentially photoreactive sites of the guanine moiety, namely, the C6–O and N7 atoms. Finally, we optimized the ground-state geometry of this Guo-H₂O complex (see Fig. 1) using the ωB97X-D/def2-TZVPP method. The resulting equilibrium S₀ geometry of the Guo-H₂O complex has the water molecule lying in the plane of the aromatic purine ring, to which it is attached by two hydrogen bonds with the carbonyl group (2.19 Å) and the N7 atom (2.12 Å; see Fig. 1). We expect that upon UV excitation, the C6–O⋯H₂O hydrogen bond would break and thus facilitate the accessibility of the ¹n_Oπ* dark state.³¹ It is worth adding that we constructed a similar water complex with deoxyguanosine (dGuo-H₂O, see ESI†), and the optimization procedure yielded a virtually identical structure.

Vertical excitation energies

We further computed the vertical excitation energies (shown in Table 1) of the low-lying electronically excited states in the Franck–Condon region of Guo-H₂O using the SCS-ADC(2)/cc-pVTZ method. The obtained data show that the absorption of UV photons can result in population of the two locally-excited (LE) ¹ππ* states (at 5.13 and 5.76 eV) marked by relatively high oscillator strengths (see Fig. S3 in ESI†). The population of either of the ¹ππ* states enables dissipation of the excitation energy through puckering of the aromatic ring and out-of-plane motion of the amino group. These ring-puckering pathways allow for ultrafast non-destructive deactivation to the electronic ground state and are often referred to as ballistic photorelaxation mechanisms.^22,26,56,57 However, in the context of this study, we focus on the ¹n_Oπ* excitation at 5.63 eV (Table 1) that is characterized by a transition from the non-bonding lone electron pair molecular orbital located on the carbonyl oxygen atom that is also partially localized on the water molecule (see Fig. S3 in ESI†). The ¹n_Oπ* state in the Franck–Condon region exhibits a moderate charge-transfer (CT) character, associated with the donation of 0.06 e⁻ from the water molecule to the chromophore moiety. We expect that even more pronounced water-to-chromophore electron transfer would occur on the ¹n_Oπ* potential energy (PE) surface due to geometry relaxation. It is worth noting that the photophysical properties of dGuo-H₂O (see Table S1 in the ESI†) are entirely consistent with those of the Guo-H₂O system.

Table 1 Vertical excitations energies (in eV) of Guo-H₂O and Guo-OH obtained at the SCS-ADC(2) level of theory using the cc-pVTZ and aug-cc-pVDZ basis sets, respectively. These equilibrium ground-state structures were found using the ωB97X-D/def2-TZVPP method

State/transition		E exc [eV]	f osc	λ [nm]
Guo-H2O
S1		5.13	1.71 × 10^−1	241.7
S2	nOπ*	5.63	1.37 × 10^−4	220.2
S3		5.76	3.26 × 10^−1	215.3
Guo-OH
S1		4.08	1.80 × 10^−1	303.9
S2	πσ*	4.23	1.95 × 10^−2	293.1
S3	πσ*	4.52	1.81 × 10^−2	274.3

---

Water-to-chromophore electron transfer

To explore the ¹n_Oπ* potential energy surface, we performed excited-state geometry optimization for the Guo-H₂O complex. The located S₁ (n_Oπ*) minimum-energy structure (see the upper panel in Fig. 1) shows an intriguing intermolecular chalcogen bonding interaction between the oxygen atoms of the chromophore and the water molecule (C6–O⋯OH₂), with an O⋯O distance of 2.46 Å. To the best of our knowledge, such a UV-induced intermolecular interaction between two oxygen atoms has not been reported in the literature so far, although it resembles other excited-state water-to-chromophore contacts reported previously.^30,31,58 However, the intramolecular O⋯O chalcogen bonding in the electronic ground state has already been confirmed experimentally by Fellowes et al.⁵⁹ in oxime. The excited-state chalcogen bonding complex is stabilized through the interaction of the non-bonding orbitals n_O located on the carbonyl group and the H₂O molecule (see the bottom panel in Fig. 1). The formation of the O⋯O chalcogen contact entails the elongation of the C6–O bond by 0.2 Å and a higher degree of charge transfer (0.11 e⁻ in the S₁ minimum) when compared to the Franck–Condon region. In addition, there is a slight pyramidalization of the N1 atom and an out-of-plane distortion of the amino group at the C2 position of the purine ring. We also found the analogous S₁ (n_Oπ*) minimum-energy geometry for dGuo-H₂O (see Fig. S4 in ESI†) that closely resembles the charge-transfer mechanism described for Guo-H₂O.

Interestingly, previous TR infrared experiments revealed that the carbonyl group actively contributes to the energy dissipation process of UV-excited guanine nucleosides and that the C6–O bond vibrations are also coupled to the vibrational modes of the solvent.^24,25 Thus, the elongation of the C6–O bond associated with the described chalcogen O⋯O intermolecular interaction (Fig. 1) serves as a reasonable mechanistic rationale for this experimental observation.

The formation of the O⋯O chalcogen contact in the S₁ minimum substantially lowers the vertical energy gap between the S₁ (n_Oπ*) and S₀ states, which amounts to merely 2.56 eV in the corresponding S₁ minimum. Cheng et al.²³ demonstrated that in the photodynamics of guanine nucleosides, the puzzling longer-lived excited state exhibits a very weak broadband emission (420–600 nm) with a maximum at 520 nm. The SCS-ADC(2) S₁ (n_Oπ*) minimum can be associated with the weak emission due to its dipole-forbidden character, and an energy gap of 2.56 eV (484 nm) agrees very well with the experimentally determined fluorescence spectrum. We further validated this result by calculations involving the COSMO solvent model of bulk water. The results of COSMO/SCS-ADC(2)/cc-pVTZ calculations indicate that the S₁ minimum-energy structure is only slightly destabilized by ∼0.01 eV in bulk water and that the emission wavelength and the oscillator strength of S₁ (n_Oπ*) amount to 482 nm (or 2.57 eV) and 1.92 × 10⁻⁴, respectively. Therefore, we postulate that the previously recorded experimental weak emission band could be another fingerprint of the discussed dark ¹n_Oπ* state.

Excited-state absorption spectra

To further support our hypothesis about the population of the ¹n_Oπ* excited state during the photodynamics of guanine nucleosides, we simulated excited-state absorption spectra (ESA) that are overlayed with experimental transient absorption spectra, taken from previous studies,^22,23 and are shown in Fig. 2. For this purpose, we used the S₁ (n_Oπ*) minimum-energy structures of the Guo-H₂O complex (Fig. 1) and the analogous complex of deoxyguanosine with one water molecule (dGuo-H₂O, Fig. S4 in ESI†).

Fig. 2 The excited-state absorption (ESA) spectra were simulated using the SCS-ADC(2)/cc-pVTZ method, assuming the S₁ (n_Oπ*) minimum-energy structures for Guo-H₂O (solid blue line) and dGuo-H₂O (solid red line) obtained at the SCS-ADC(2)/cc-pVTZ level. The digitalized experimental femtosecond transient absorption spectra of aqueous GMP (dashed blue line) and dGMP (dashed red line) recorded at 240 and 600 fs, respectively.^22,23

The experimental and simulated excited-state absorption spectra having maxima at 298 and 325 nm, respectively, exhibit reasonably consistent lineshapes in the range of 290–350 nm, which confirms that the UV-induced O⋯O chalcogen bonding structure may indeed be attainable in the photodynamics of guanine nucleosides.^22,23 The discrepancies between the positions and shapes of the simulated and experimental absorption bands are expected and can be ascribed to the following reasons. First, the ESA spectra for the S₁ (n_Oπ*) structures are simulated using model systems containing a single water molecule that do not allow to fully take into account the effects of bulk water on the position of the absorption bands, and thereby the discussed maxima differ by 27 nm. Second, the recorded TA bands possess two different spectral regions, namely below and above 350 nm, which are fingerprints of different photochemical processes. The acquired TA spectrum in the range of 350–500 nm has been previously ascribed to the ¹ππ* deactivation channels.^22,23 Since our simulated ESA spectra are obtained using only one specific S₁ minimum representing a single UV-induced process, it cannot reproduce the entire range of the recorded TA spectrum. Moreover, the ¹ππ* state of guanosine that is responsible for the ultrafast ballistic photorelaxation channel would be rather strongly emissive owing to the bright character.^22,26,56,57 Accordingly, it seems very unlikely that the puzzling longer-lived excited state having a very weak broadband emission could originate from the ballistic photorelaxation channels.

To further support our interpretation, we simulated the ESA spectrum (see Fig. S7 in the ESI†) for the S₁ (nπ*) minimum of Guo in the gas phase located without the explicit water molecule. The absorption maxima (380 and 490 nm) and lineshape for the latter structure strongly differ from the experimental spectra (Fig. 2), and thus it reinforces our hypothesis about the fingerprint of the excited-state chalcogen bonding complex in the recorded transient absorption band.

Furthermore, dynamics of the bleaching band associated with the recovery of the electronic ground state of guanine nucleoside has the same lifetime (∼2.0 ps) as the discussed transient absorption band (290–350 nm), which we assigned to the S₁ (n_Oπ*) structure.^22,23 This shows that the ¹n_Oπ* dark state can indeed be responsible for a large part of the photodynamics of guanine nucleosides, thus shedding an entirely new perspective on the aqueous photochemistry of heterocycles having a carbonyl group.

Methanol-to-chromophore electron transfer

In addition to the TR measurements in water, Cheng et al.²³ investigated also the photoinduced dynamics of deoxyguanosine in methanol. They observed that the lifetime of the transient absorption band (290–350 nm) and of the weak emission band (at 520 nm) increased to ∼4.2 ps in this medium, when compared to the analogous bands in water.²³ This indicates that S₁ (n_Oπ*) minimum is even more stabilized in methanol than in water.

To provide a mechanistic rationale for the TR measurements in methanol, we explored the ¹n_Oπ* PE surface for a system containing deoxyguanosine and an explicit CH₃OH molecule (dGuo-CH₃OH) using the same SCS-ADC(2)/cc-pVTZ level of theory. We located a very similar S₁ minimum-energy structure having the C6–O⋯OHCH₃ chalcogen bonding interaction between dGuo and the methanol molecule (see Fig. 3 and Fig. S8 in the ESI†). The O⋯O distance at the S₁ minimum-energy geometry of dGuo-CH₃OH amounts to 2.07 Å (see Fig. S8 in ESI†) and the aromatic purine ring accepts 0.41 e⁻ from the explicit methanol molecule. The excited-state O⋯O chalcogen bond is noticeably shorter by 0.4 Å, and the associated charge transfer from the solvent molecule is roughly twice as large when compared to the S₁ minimum of the analogous Guo-H₂O complex. Importantly, the excited-state absorption (ESA) spectrum of the S₁ (n_Oπ*) minimum of dGuo-CH₃OH (see Fig. S9 in ESI†) closely resembles the simulated ESA presented in Fig. 2 and is consistent with the transient absorption band (in the range of 290–350 nm) recorded in methanol.²³ Accordingly, methanol solvent molecules can stabilize the ¹n_Oπ* excited state more strongly than water molecules, and as a result, the associated excited-state lifetime in methanol is doubled. Thus, excited-state chalcogen bonding between deoxyguanosine and methanol is now supported by both experimental²³ and theoretical results. It is worth adding that the S₁ (n_Oπ*) minimum of the dGuo-CH₃OH complex was also reoptimized using the XMS-CASPT2/SA-CASSCF(10,9)/cc-pVDZ level of theory (see Fig. S8 in ESI†). The latter calculations confirmed the reliability of the excited-state interaction between the methanol molecule and deoxyguanosine.

Fig. 3 The S₁ (n_Oπ*) minimum-energy structure of dGuo-CH₃OH found at the SCS-ADC(2)/cc-pVTZ level of theory alongside the delocalized n_O and π* molecular orbitals demonstrating a charge-transfer process of 0.41 e⁻ from CH₃OH to the nucleobase part.

Electron-driven proton transfer deactivation pathway

To elucidate the role of S₁ (n_Oπ*) minimum (Fig. 1) in the photochemistry of guanine nucleosides, we located the relevant potential energy (PE) surface crossings and key structures at the SCS-ADC(2)/cc-pVTZ level of theory, and we present the associated potential energy profiles in Fig. 4.

Fig. 4 The potential energy profiles (PE) demonstrate a photochemical channel, involving the S₂/S₁ minimum-energy crossing point (MECP), S₁ minimum, and S₁/S₀ MECP, that enables a water-to-chromophore electron-driven proton transfer (EDPT) mechanism in Guo-H₂O. The PE profiles were obtained at the SCS-ADC(2)/cc-pVTZ level of theory.

Upon UV excitation of the ¹ππ* excited state (orange line), the system can readily reach the ¹ππ*/¹n_Oπ* MECP at 4.74 eV (see the left inset in Fig. 4) that is marked by the elongation of the C6–O carbonyl bond (1.33 Å) as well as weakening of the C6–O⋯H₂O hydrogen bond (2.48 Å). Importantly, the located S₂/S₁ MECP shows that a slight elongation of the carbonyl bond by ∼0.1 Å significantly stabilizes the energy of the ¹n_Oπ* excited state with respect to the Franck–Condon region. The accessibility of S₂/S₁ MECP enables the population of the ¹n_Oπ* excited state (blue line) and, in turn, the Guo-H₂O complex may reach the shallow S₁ minimum at 4.42 eV (shown in Fig. 1) in a barrierless manner.

The associated charge transfer found for the S₁ minimum-energy structure could trigger the electron-driven proton transfer (EDPT) from the water molecule to the N7 atom of guanosine. The EDPT mechanism having an energy barrier of 0.41 eV could then lead to S₁/S₀ MECP at 3.95 eV (see the right inset in Fig. 4), which allows for radiationless deactivation of guanosine. Notably, an energy barrier of 0.41 eV could be overcome due to excess excitation energy. Thus, the discussed ¹n_Oπ*-mediated photorelaxation channel allows for a water-splitting process leading to the formation of hydrogenated guanosine radical and a hydroxyl (˙OH) radical.

Benchmarking of relevant excited-state structures

Single reference ab initio methods such as SCS-ADC(2) are widely used to investigate photochemical processes, which occur in organic chromophores. However, Marsili et al.⁶⁰ have demonstrated that the ADC(2)-s method can lead to finding nonphysical S₁/S₀ surface crossings for carbonyl-containing compounds. To address a potential issue with applying the SCS-ADC(2) method to the excited-state optimization of Guo-H₂O, we reoptimized the SCS-ADC(2) S₁(n_Oπ*) minimum and the SCS-ADC(2) S₁ (n_Oπ*)/S₀ minimum-energy crossing point using the XMS-CASPT2/SA-CASSCF(10,9)/cc-pVDZ level of theory. The molecular orbitals included in the complete active space for both excited-state structures are shown in the ESI† (Fig. S10 and S11).

The superimposed excited-state minimum-energy structures obtained at the SCS-ADC(2) (black) and XMS-CASPT2 (cyan) levels are depicted in Fig. 5. The root-mean-square deviations (RMSDs) evaluated for those structures amount to 0.082 and 0.135 Å for the S₁ minimum and S₁/S₀ surface crossing, respectively. This indicates that the geometries are virtually identical. Thus, the XMS-CASPT2 calculations confirmed that the excited-state structures provided by SCS-ADC(2) method are reliable.

Fig. 5 The superimposed excited-state minimum-energy structures optimized at the SCS-ADC(2) S₁ minimum, SCS-ADC(2) S₁/S₀ surface crossing (black) and XMS-CASPT2/SA-CASSCF(10,9)/cc-pVDZ level (cyan) are presented with the root-mean-square deviation (RMSD) calculated for the corresponding structures.

Regarding the relevant structural parameters, the most significant difference between the S₁ minimum-energy geometries is found for the O⋯O intermolecular distances that amount to 2.43 and 2.46 Å at the XMS-CASPT2 and SCS-ADC(2) levels, respectively. Furthermore, the C6–O bond length is shortened to 1.35 Å (by 0.05 Å) at the XMS-CASPT2 level in comparison with the SCS-ADC(2) geometry. Comparing XMS-CASPT2 and SCS-ADC(2) S₁/S₀ MECP, the noticeable structural difference is a slightly different position of the hydroxyl radical. Since the multireference method (XMS-CASPT2) predicted the excited-state O⋯O intermolecular interaction and confirmed the existence of the electron-driven proton transfer S₁/S₀ surface crossing, we conclude that the SCS-ADC(2) method is suitable for the description of the electronic structure of our model system.

Photolesions of guanine nucleosides

As described above, the EDPT photorelaxation mechanism generates two reactive radical species close to one another, that is, the ˙OH and GuoH˙ radicals, with the unpaired electron located on the C8 atom in the latter case. These radicals may readily recombine in the hot electronic ground state and produce a guanosine-OH (Guo-OH) photohydrate intermediate, the minimum-energy structure of which lies 1.96 eV (45.2 kcal mol⁻¹) below the EDPT S₁/S₀ minimum-energy crossing point (Fig. 6 and Fig. S12 in the ESI†). The vertical excitation energies calculated for the Guo-OH photohydrate (see Table 1) show that there is a low-lying ¹ππ* locally excited state at 4.08 eV having a substantial oscillator strength, which indicates that the photohydrate can easily absorb longer UV wavelengths (UV-B region) than the parent Guo molecule. In other words, the formation of Guo photohydrate redshifts the UV absorption spectrum of the chromophore and increases the photochemical activity of the molecule since it can be excited with much lower energies. More importantly, Guo-OH is also characterized by the low-lying ¹πσ* excited state (at 4.23 eV, Table 1) that could enable the photodetachment of an electron from the intermediate, thus forming a radical cation that was previously proposed as the critical intermediate necessary for the formation of 8-oxo-guanosine.¹³ Therefore, we propose that Guo-OH photohydrate could be the structural precursor of 8-oxo-guanosine, which is an important photolesion that can be formed in aqueous solution of guanine nucleosides.¹³

Fig. 6 Possible photoproducts generated from the biradical intermediate produced in the EDPT process, along with the relative energies of their ground-state structures with respect to EDPT S₁/S₀ state crossing.

On the other hand, the main observed photoproduct upon UV excitation of guanosine in water is Fapy-guanosine (Fapy-Guo).¹³ Given the above results, this may be explained by the C8–N9 bond rupture (see Fig. 6 and Fig. S12 in ESI†) of the Guo-OH photohydrate and subsequent proton transfer to the N9 atom, which allows for the formation of the more stable Fapy-Guo. Furthermore, the transition-state structure leading to the formation of Fapy-Guo is located 0.66 eV (15.2 kcal mol⁻¹, see Fig. S12 in ESI†) lower than the located S₁/S₀ minimum-energy crossing point. Consequently, photochemically formed and still vibrationally hot Guo-OH can be readily transformed to Fapy-Guo after passing through this modest energy barrier. It is worth noting that the Fapy-Guo photoproduct is lower in energy than the EDPT S₁/S₀ surface crossing by 2.71 eV (62.4 kcal mol⁻¹). Interestingly, both Fapy-Guo and Guo-H₂O systems are nearly isoenergetic. We expect that Fapy-Guo should be a stable photoproduct that would not undergo a reverse ground-state conversion to Guo, particularly since such a reaction would involve a considerable energy barrier, exceeding 47.2 kcal mol⁻¹ (cf. Fig. S12 in ESI†). The same photoproduct was also detected in the presence of electron scavengers (H₃O⁺ and N₂O), suggesting that solvated electrons mediated by the ¹πσ* state do not participate in the formation of Fapy-Guo.¹³ This is consistent with our proposed mechanism for forming guanosine photolesions, which result from a direct water-to-chromophore electron transfer rather than the attachment of free solvated electrons to the chromophore.

Conclusions

In conclusion, we have demonstrated the long-sought plausible photochemical pathway leading to the formation of photolesions of aqueous guanine nucleosides, namely 8-oxo-guanosine and Fapy-guanosine. It involves the formation of an excited-state chalcogen bonded complex with a water molecule that enables the population and stabilization of the dark ¹nπ* state. This excited-state complex shows a substantial charge-transfer character, which can entail an electron-driven proton transfer process from a water molecule to the guanine ring leading to the formation of a reactive hydroxyl radical that may damage the canonical structure of guanine. The results of our quantum chemical calculations are strongly supported by reinterpretation of the time-resolved spectra recorded previously for aqueous guanine nucleosides, indicating that the intermolecular chalcogen bonding complex can be easily reached upon UV excitation. The proposed mechanism is the first that explains both the formation of Fapy-Guo as the main photoproduct in an aqueous solution and 8-oxo-guanine as a concomitant minor photoproduct. Furthermore, our results point out that the reactive dark ¹nπ* state, which seems to be hardly accessible owing to high excitation energy, can be substantially stabilized beyond the Franck–Condon region. This is in consequence of the formation of a specific and short-lived excited-state interaction that can lower the energy of the nπ* state.

Our results also indicate that the SCS-ADC(2) method can be used as a more accurate alternative to the ADC(2)-s approach for studying the photochemical properties of carbonyl-containing compounds. However, we also strongly encourage benchmarking results of this single-reference method using multireference wavefunction methods, such as the XMS-CASPT2 level of theory.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

M. J. J. acknowledges the support of the ‘Diamond Grant’ (0144/DIA/2017/46) from the Polish Ministry of Science and Higher Education, a doctoral scholarship from the National Science Center Poland (2020/36/T/ST4/00564), and the START Fellowship from the Foundation for Polish Science. J. S. acknowledges support by the Czech Science Foundation (21-23718S) and the project SYMBIT reg. number: CZ.02.1.01/0.0/0.0/15_003/0000477 financed by the ERDF.

References

1. M. L. Kripke, P. A. Cox, L. G. Alas and D. B. Yarosh, Proc. Natl. Acad. Sci. U. S. A., 1992, 89, 7516–7520.
2. J. S. Taylor, Acc. Chem. Res., 1994, 27, 76–82.
3. V. Pagès and R. P. Fuchs, Oncogene, 2002, 21, 8957–8966.
4. R. P. Sinha and D.-P. Häder, Photochem. Photobiol. Sci., 2002, 1, 225–236.
5. J. Cadet, E. Sage and T. Douki, Mutat. Res., 2005, 571, 3–17.
6. J. Cadet and J. R. Wagner, Cold Spring Harbor Perspect. Biol., 2013, 5, a012559.
7. J. Cadet and T. Douki, Photochem. Photobiol. Sci., 2018, 17, 1816–1841.
8. S. P. Jackson, Carcinogenesis, 2002, 23, 687–696.
9. R. Madabhushi, L. Pan and L.-H. Tsai, Neuron, 2014, 83, 266–282.
10. C. E. Crespo-Hernández, S. Flores, C. Torres, I. Negrón-Encarnación and R. Arce, Photochem. Photobiol., 2000, 71, 534–543.
11. C. E. Crespo-Hernández and R. Arce, Photochem. Photobiol., 2000, 71, 544–550.
12. G. A. Papadantonakis, R. Tranter, K. Brezinsky, Y. Yang, R. B. van Breemen and P. R. LeBreton, J. Phys. Chem. B, 2002, 106, 7704–7712.
13. C. E. Crespo-Hernández and R. Arce, J. Photochem. Photobiol., B, 2004, 73, 167–175.
14. M. Gomez-Mendoza, A. Banyasz, T. Douki, D. Markovitsi and J.-L. Ravanat, J. Phys. Chem. Lett., 2016, 7, 3945–3948.
15. D. B. Hall, R. E. Holmlin and J. K. Barton, Nature, 1996, 382, 731–735.
16. W. J. Schreier, T. E. Schrader, F. O. Koller, P. Gilch, C. E. Crespo-Hernandez, V. N. Swaminathan, T. Carell, W. Zinth and B. Kohler, Science, 2007, 315, 625–629.
17. K. V. Nguyen and C. J. Burrows, J. Am. Chem. Soc., 2011, 133, 14586–14589.
18. R. Szabla, J. Campos, J. E. Šponer, J. Šponer, R. W. Góra and J. D. Sutherland, Chem. Sci., 2015, 6, 2035–2043.
19. D. B. Bucher, C. L. Kufner, A. Schlueter, T. Carell and W. Zinth, J. Am. Chem. Soc., 2016, 138, 186–190.
20. R. Szabla, H. Kruse, P. Stadlbauer, J. Šponer and A. L. Sobolewski, Chem. Sci., 2018, 9, 3131–3140.
21. R. Szabla, M. Zdrowowicz, P. Spisz, N. J. Green, P. Stadlbauer, H. Kruse, J. Šponer and J. Rak, Nat. Commun., 2021, 12, 3018.
22. V. Karunakaran, K. Kleinermanns, R. Improta and S. A. Kovalenko, J. Am. Chem. Soc., 2009, 131, 5839–5850.
23. C. C.-W. Cheng, C. Ma, C. T.-L. Chan, K. Y.-F. Ho and W.-M. Kwok, Photochem. Photobiol. Sci., 2013, 12, 1351–1365.
24. D. A. McGovern, G. W. Doorley, A. M. Whelan, A. W. Parker, M. Towrie, J. M. Kelly and S. J. Quinn, Photochem. Photobiol. Sci., 2009, 8, 542–548.
25. J. B. Nielsen, J. Thøgersen, S. K. Jensen, S. B. Nielsen and S. R. Keiding, Phys. Chem. Chem. Phys., 2011, 13, 13821–13826.
26. S. F. Altavilla, J. Segarra-Martí, A. Nenov, I. Conti, I. Rivalta and M. Garavelli, Front. Chem., 2015, 3, 29.
27. S. E. Krul, S. J. Hoehn, K. J. Feierabend and C. E. Crespo-Hernández, J. Chem. Phys., 2021, 154, 075103.
28. J. Ortín-Fernández, J. González-Vázquez, L. Martínez-Fernández and I. Corral, Molecules, 2022, 27, 989.
29. T. A. A. Oliver, Y. Zhang, A. Roy, M. N. R. Ashfold and S. E. Bradforth, J. Phys. Chem. Lett., 2015, 6, 4159–4164.
30. M. Barbatti, J. Am. Chem. Soc., 2014, 136, 10246–10249.
31. R. Szabla, H. Kruse, J. Šponer and R. W. Góra, Phys. Chem. Chem. Phys., 2017, 19, 17531–17537.
32. X. Wu, J. Ehrmaier, A. L. Sobolewski, T. N. V. Karsili and W. Domcke, Phys. Chem. Chem. Phys., 2019, 21, 14238–14249.
33. P. Pracht, F. Bohle and S. Grimme, Phys. Chem. Chem. Phys., 2020, 22, 7169–7192.
34. S. Grimme, J. Chem. Theory Comput., 2019, 15, 2847–2862.
35. J.-D. Chai and M. Head-Gordon, Phys. Chem. Chem. Phys., 2008, 10, 6615.
36. F. Weigend and R. Ahlrichs, Phys. Chem. Chem. Phys., 2005, 7, 3297–3305.
37. M. J. Frisch, G. W. Trucks, H. B. Schlegel, G. E. Scuseria, M. A. Robb, J. R. Cheeseman, G. Scalmani, V. Barone, G. A. Petersson, H. Nakatsuji, X. Li, M. Caricato, A. V. Marenich, J. Bloino, B. G. Janesko, R. Gomperts, B. Mennucci, H. P. Hratchian, J. V. Ortiz, A. F. Izmaylov, J. L. Sonnenberg, D. Williams-Young, F. Ding, F. Lipparini, F. Egidi, J. Goings, B. Peng, A. Petrone, T. Henderson, D. Ranasinghe, V. G. Zakrzewski, J. Gao, N. Rega, G. Zheng, W. Liang, M. Hada, M. Ehara, K. Toyota, R. Fukuda, J. Hasegawa, M. Ishida, T. Nakajima, Y. Honda, O. Kitao, H. Nakai, T. Vreven, K. Throssell, J. A. Montgomery Jr., J. E. Peralta, F. Ogliaro, M. J. Bearpark, J. J. Heyd, E. N. Brothers, K. N. Kudin, V. N. Staroverov, T. A. Keith, R. Kobayashi, J. Normand, K. Raghavachari, A. P. Rendell, J. C. Burant, S. S. Iyengar, J. Tomasi, M. Cossi, J. M. Millam, M. Klene, C. Adamo, R. Cammi, J. W. Ochterski, R. L. Martin, K. Morokuma, O. Farkas, J. B. Foresman and D. J. Fox, Gaussian-16 Revision B.03, 2016, Gaussian Inc., Wallingford CT.
38. A. Hellweg, S. A. Grün and C. Hättig, Phys. Chem. Chem. Phys., 2008, 10, 4119.
39. C. Hättig, Adv. Quantum Chem., 2005, 50, 37–60.
40. A. Dreuw and M. Wormit, Wiley Interdiscip. Rev.: Comput. Mol. Sci., 2015, 5, 82–95.
41. R. A. Kendall, T. H. Dunning and R. J. Harrison, J. Chem. Phys., 1992, 96, 6796–6806.
42. F. Plasser, J. Chem. Phys., 2020, 152, 084108.
43. B. G. Levine, J. D. Coe and T. J. Martínez, J. Phys. Chem. B, 2008, 112, 405–413.
44. K. Andersson, P. A. Malmqvist, B. O. Roos, A. J. Sadlej and K. Wolinski, J. Phys. Chem., 1990, 94, 5483–5488.
45. T. Shiozaki, W. Györffy, P. Celani and H.-J. Werner, J. Chem. Phys., 2011, 135, 081106.
46. V. Veryazov, P. Å. Malmqvist and B. O. Roos, Int. J. Quantum Chem., 2011, 111, 3329–3338.
47. T. Shiozaki, Wiley Interdiscip. Rev.: Comput. Mol. Sci., 2018, 8, e1331.
48. TURBOMOLE V7.3 2018, a development of University of Karlsruhe and Forschungszentrum Karlsruhe GmbH, 1989-2007, TURBOMOLE GmbH, since 2007; available from http://www.turbomole.com.
49. N. M. O’boyle, A. L. Tenderholt and K. M. Langner, J. Comput. Chem., 2008, 29, 839–845.
50. S. Smidstrup, A. Pedersen, K. Stokbro and H. Jónsson, J. Chem. Phys., 2014, 140, 214106.
51. J. Zheng, X. Xu and D. G. Truhlar, Theor. Chem. Acc., 2011, 128, 295–305.
52. F. Neese, Wiley Interdiscip. Rev.: Comput. Mol. Sci., 2012, 2, 73–78.
53. D. Tuna, A. L. Sobolewski and W. Domcke, J. Phys. Chem. A, 2014, 118, 122–127.
54. D. Tuna and W. Domcke, Phys. Chem. Chem. Phys., 2016, 18, 947–955.
55. L. Zhao, G. Zhou, B. Jia, G. Teng, K. Zhan, H. Zheng, J. Luo and B. Liu, J. Photochem. Photobiol., A, 2020, 401, 112753.
56. L. Serrano-Andrés, M. Merchán and A. C. Borin, J. Am. Chem. Soc., 2008, 130, 2473–2484.
57. B. Heggen, Z. Lan and W. Thiel, Phys. Chem. Chem. Phys., 2012, 14, 8137–8146.
58. M. J. Janicki, R. Szabla, J. Šponer and R. W. Góra, Chem. Phys., 2018, 515, 502–508.
59. T. Fellowes, B. L. Harris and J. M. White, Chem. Commun., 2020, 56, 3313–3316.
60. E. Marsili, A. Prlj and B. F. E. Curchod, Phys. Chem. Chem. Phys., 2021, 23, 12945–12949.

---

Footnote

† Electronic supplementary information (ESI) available: Computational procedures, conformational analysis, and characteristics of excited-state crucial points. See DOI: 10.1039/d2cp00801g

---