Photoprotective Properties of Eumelanin: Computational Insights into the Photophysics of a Catechol:Quinone Heterodimer Model System

: Melanins are skin-centered molecular structures that block harmful UV radiation from the sun and help protect chromosomal DNA from UV damage. Understanding the photodynamics of the chromophores that make up eumelanin is therefore paramount. This manuscript presents a multi-reference computational study of the mechanisms responsible for the experimentally observed photostability of a melanin-relevant model heterodimer comprising a catechol (C)–benzoquinone (Q) pair. The present results validate a recently proposed photoinduced intermolecular transfer of an H atom from an OH moiety of C to a carbonyl-oxygen atom of the Q. Photoexcitation of the ground state C:Q heterodimer (which has a π -stacked “sandwich” structure) results in population of a locally excited ππ * state (on Q), which develops increasing charge-transfer (biradical) character as it evolves to a “hinged” minimum energy geometry and drives proton transfer (i.e., net H atom transfer) from C to Q. The study provides further insights into excited state decay mechanisms that could contribute to the photostability afforded by the bulk polymeric structure of eumelanin. heterodimers. The present results support earlier suggestions, from analysis of


Results and Discussion
An isolated gas-phase multi-reference computational study of the UV photoinduced chemistry of C:Q heterodimers is presented, the results of which support and extend conclusions reached in recent transient absorption studies of this system in a weakly interacting solvent (cyclohexane) [95]. This section is sub-divided into sections addressing the minimum energy structures of the heterodimer and its biradical tautomer, the electronic spectroscopy of the former and then the topography of the PE surfaces sampled following photoexcitation of the heterodimer.

Minimum Energy Geometries of the Ground State Heterodimer and Its Biradical Tautomer
As a reminder, the ωB97XD functional was used in order to achieve an appropriately balanced description of the dominant π-π interactions between the C and Q chromophores and the long-range correlation effects. As Figure 1a,b show, the ground state minimum energy geometry of the C:Q heterodimer exhibits a π-stacked configuration. (The Cartesian coordinates of all atoms in this minimum energy structure are provided as Supplementary Materials, as are (harmonic) normal mode wavenumbers for the ground state C:Q heterodimer and the bare C monomer.) Alternative side-on hydrogen bonding could also be expected to provide a strong intermolecular interaction, but the π-stacked ground state configuration shown in Figure 1 offers both π-π and hydrogen-bonding, and the stability of this π-stacked structure can be understood by recognizing two stabilizing interactions. One is a π-π interaction between bonding π electrons on the catechol moiety and the antibonding π* orbital localized on the benzoquinone. These orbitals are reasonably wellmatched in energy. The second is the inter-chromophore hydrogen-bonding between an O-H donor, local to catechol, and a carbonyl oxygen acceptor, localized on the benzoquinone chromophore. This deduced C:Q heterodimer structure, involving one intra-and one intermolecular H-bond, is fully consistent with that derived by analysis of the Fourier transform infrared spectrum of mixed solutions of (the di-tert-butyl substituted forms of) C and Q in cyclohexane [95]. benzoquinone chromophore. This deduced C:Q heterodimer structure, involving one intra-and one intermolecular H-bond, is fully consistent with that derived by analysis of the Fourier transform infrared spectrum of mixed solutions of (the di-tert-butyl substituted forms of) C and Q in cyclohexane [95].  (The Cartesian coordinates of all atoms in this minimum energy structure are also provided in the Supplementary Materials). The "hinged" structural arrangement of the C and Q chromophores is very different from the π-stacked configuration of the ground state heterodimer, though it again displays one intermolecular and one intramolecular hydrogen bond. The breakdown of the π-stacking upon biradical formation can be understood by recognizing that the lowest energy biradical configuration has ππ* character, wherein the π-and π*-orbitals are localized on, respectively, the C and Q moieties (vide infra). The ensuing electron-electron repulsion destroys the π-π interaction inherent to the ground state parent structure, leaving inter-chromophore H-bonding as the dominant non-covalent interaction in the biradical tautomer.
We note that the experimentally studied C:Q heterodimer contains bulky tert-butyl substituents which may affect the π-stacking. That said, we do not expect this to have a serious impact on the excited state photophysics deduced here, as the tert-butyl group is a σ-perturbing substituent, while the dominant effects observed in the photophysics of C:Q are π-centered.  (The Cartesian coordinates of all atoms in this minimum energy structure are also provided in the Supplementary Materials). The "hinged" structural arrangement of the C and Q chromophores is very different from the π-stacked configuration of the ground state heterodimer, though it again displays one intermolecular and one intramolecular hydrogen bond. The breakdown of the π-stacking upon biradical formation can be understood by recognizing that the lowest energy biradical configuration has ππ* character, wherein the πand π*-orbitals are localized on, respectively, the C and Q moieties (vide infra). The ensuing electron-electron repulsion destroys the π-π interaction inherent to the ground state parent structure, leaving inter-chromophore H-bonding as the dominant non-covalent interaction in the biradical tautomer.
We note that the experimentally studied C:Q heterodimer contains bulky tert-butyl substituents which may affect the π-stacking. That said, we do not expect this to have a serious impact on the excited state photophysics deduced here, as the tert-butyl group is a σ-perturbing substituent, while the dominant effects observed in the photophysics of C:Q are π-centered. Table 1 lists the vertical excitation energies (VEEs) to the first few singlet and triplet excited states of the C:Q heterodimer from the π-stacked minimum energy configuration. The active space orbitals used in the CASPT2 computations, shown in Figure 2, may be used along with Table 1 to identify the dominant orbital promotions involved in preparing these various excited states. Table 1 lists the vertical excitation energies (VEEs) to the first few singlet and triplet excited states of the C:Q heterodimer from the π-stacked minimum energy configuration. The active space orbitals used in the CASPT2 computations, shown in Figure 2, may be used along with Table 1 to identify the dominant orbital promotions involved in preparing these various excited states. Active space orbitals used in the CASSCF/CASPT2 calculations of the C:Q heterodimer. Q sits above C in all depictions, but the orientation of the heterodimer structure is varied to allow better visualization of the various occupied (1-5) and virtual (6-10) orbitals. The orbital numbering aligns with that used to describe the dominant promotions associated with forming the various excited states in Table 1. Table 1. Calculated vertical excitation energies (VEE) and oscillator strengths (f) to the lowest singlet and lowest triplet excited states of the C:Q heterodimer. The entries in the "Character" column show the dominant electron promotions between the active space orbitals shown in Figure 2, with the respective contributions (i.e., the squares of the associated coefficients) shown in parentheses.

Electronic State
Character Vertical excitation to the S1 state from the π-stacked ground state minimum energy geometry is dominated by electron promotion from a largely non-bonding (n) orbital, localized on the carbonyl oxygen atom, to an antibonding π* orbital, both of which are localized on the benzoquinone moiety. The S1 state is best viewed as a locally excited state (i.e., the excitation is concentrated on a common chromophore) with nπ* character, and optically "dark" (i.e., the S1-S0 transition has a low oscillator strength-reflecting the poor spatial overlap of the n and π* orbitals).
The S2 state is best described by a mixture of two configurations. The 6←5 orbital promotion involves excitation of an electron from the π highest occupied molecular orbital (HOMO) (which is mainly localized on Q but extends over the C moiety also) to the Figure 2. Active space orbitals used in the CASSCF/CASPT2 calculations of the C:Q heterodimer. Q sits above C in all depictions, but the orientation of the heterodimer structure is varied to allow better visualization of the various occupied (1-5) and virtual (6-10) orbitals. The orbital numbering aligns with that used to describe the dominant promotions associated with forming the various excited states in Table 1. Table 1. Calculated vertical excitation energies (VEE) and oscillator strengths (f ) to the lowest singlet and lowest triplet excited states of the C:Q heterodimer. The entries in the "Character" column show the dominant electron promotions between the active space orbitals shown in Figure 2, with the respective contributions (i.e., the squares of the associated coefficients) shown in parentheses.

Electronic State
Character VEE/eV f Vertical excitation to the S 1 state from the π-stacked ground state minimum energy geometry is dominated by electron promotion from a largely non-bonding (n) orbital, localized on the carbonyl oxygen atom, to an antibonding π* orbital, both of which are localized on the benzoquinone moiety. The S 1 state is best viewed as a locally excited state (i.e., the excitation is concentrated on a common chromophore) with nπ* character, and optically "dark" (i.e., the S 1 -S 0 transition has a low oscillator strength-reflecting the poor spatial overlap of the n and π* orbitals).
The S 2 state is best described by a mixture of two configurations. The 6←5 orbital promotion involves excitation of an electron from the π highest occupied molecular orbital (HOMO) (which is mainly localized on Q but extends over the C moiety also) to the π* lowest unoccupied molecular orbital (LUMO) localized on Q. 6←4 promotion, in contrast, involves excitation from a bonding π orbital, largely localized on the C chromophore, to the Q-localized π* antibonding orbital. Both promotions can be pictured as π*←π transitions: The S 2 state is thus best viewed as having ππ* character but, even in the vertical region, formation of the S 2 state of the heterodimer involves some electron transfer from C to Q-which likely contributes to the high oscillator strength reported in Table 1. For completeness, excitation energies to the first two triplet excited states, both of which are also best described as locally excited ππ* states, are also included in Table 1.
The predicted oscillator strengths and VEEs for the (weak) S 1 -S 0 and (strong) S 2 -S 0 transitions reported in Table 1 match well with the maxima evident (at λ~595 nm and 400 nm) in the UV absorption spectrum of the (t-butyl substituted) C:Q heterodimer in cyclohexane [95], lending further support to our expectation that the tert-butyl substituents have little effect on the electronic properties (and excited state photophysics) of the heterodimer.

Photophysics of the C:Q Heterodimer
We now consider the possible fate(s) of the C:Q heterodimer following UV photoexcitation and the potential role of such photophysics in explaining the photostability of eumelanin. Motivated by the work of Kohler and co-workers [95], the ensuing discussion focuses on the mechanism of photoinduced HAT. Figure 3a displays PE profiles of the ground and first two excited singlet electronic states along the LIIC between the ground state minimum energy geometry and that of the optimized biradical tautomer formed by HAT (henceforth Q LIIC )-the structures of both of which are reproduced again as insets in Figure 3a. We caution that the use of a LIIC almost inevitably means that the present calculations do not capture the true minimum energy path from reactant to product, but they are expected to identify key topographical features of the PE surfaces under study. The ground state (black) PE profile increases, reaching a maximum at Q LIIC~0 .5, beyond which it decreases en route to the biradical tautomer. The electronic wavefunction of the adiabatic ground state switches at Q LIIC~0 .5, as illustrated in Figure 3b,c, which illustrates the increasing ππ* character of the ground state configuration of the biradical tautomer (which correlates diabatically with the S 2 state of the parent C:Q heterodimer). The stability of the biradical structure can be understood by considering the change in electronic character. In the ππ* configuration, the O-atom donor of the pre-existing OH moiety contains a doubly occupied p orbital which, when viewed from the biradical minimum, provides a long-range repulsive interaction in the reverse HAT direction. As Figure 3b,c show, the singly occupied molecular orbitals in the S 0 state of the biradical are localized on different chromophores: the S 0 state at Q LIIC > 0.5 is best described as a charge-separated (or charge transfer) state.
The S 1 state (red in Figure 3a) has nπ* character in the vertical region and is bound with respect to initial motion along Q LIIC -reflecting the fact that the π* ← n transition is localized on the benzoquinone moiety and shows no net driving force for HAT. The S 2 state, in contrast, has ππ* character at the Franck-Condon geometry and shows net reactivity with respect to the "hinge-like" geometry change along Q LIIC . This can be understood by recognizing that the transition involves π and π* orbitals that are initially largely localized on a single chromophore but then develop increasing charge-separated character, as shown in Figure 3b,c. The diabatic 1 ππ* state progressively develops charge transfer (CT) character as Q LIIC → 1, which is neutralized by proton transfer from the C to the Q moiety. Such photoinduced HATs are also frequently termed proton-coupled electron transfer (PCET) or electron-driven proton transfer (EDPT) processes. Upon increasing Q LIIC from the Franck-Condon region, the diabatic CT state crosses both the 1 nπ* state and the ground state. This is the origin of the evolution of the ground state electronic wavefunction described above. As in many related systems [4,102,103], these diabatic crossing points will surely be CIs when motion along orthogonal modes are considered, and represent regions of configuration space where internal conversion between electronic states is favorable (i.e., where population is funneled efficiently to the lower PE surface).
Given the foregoing descriptions of the various electronic states of the C:Q heterodimer, the reported photophysics can be rationalized as follows: Photoexcitation populates the "bright" 1 ππ* state, which is initially largely localized on Q but evolves spontaneously along the coordinate associated with HAT. Internal conversion is likely to occur at both the S 2 /S 1 and S 1 /S 0 CIs (see Figure 3a). The former may well lead to some population becoming temporarily trapped in the 1 nπ* state-as has been proposed in the case of oxybenzone [72]-while non-adiabatic interaction at the latter CI will promote efficient internal conversion back to the S 0 state. Having accessed the S 0 PE surface, population may bifurcate to reform the ground state heterodimer (thereby demonstrating photostability) or evolve towards the biradical tautomer and thence to two (potentially harmful) semiquinone free radical species. Thus, the extent to which the C:Q heterodimer offers photoprotection and photostability will be sensitively dependent on the non-adiabatic dynamics prevailing at the S 1 /S 0 CI-which will be sensitive to the detailed topography of the CI and the nuclear momenta within the evolving population. Such details, in turn, are likely to be sensitively dependent upon the natures of any (less benign than tert-butyl) substituents within the C and Q moieties and, in any condensed phase application, to the prevailing solvent [104]. Extrapolating to eumelanin itself, any such competition between reformation of the minimum energy ground state structure and biradical (and thence radical) formation might well be influenced by the extent (or otherwise) to which the system is able to distort away from any structural layering imposed by more extensive π-stacking between polymer strands. The S1 state (red in Figure 3a) has nπ* character in the vertical region and is bound with respect to initial motion along QLIIC-reflecting the fact that the π* ← n transition is localized on the benzoquinone moiety and shows no net driving force for HAT. The S2 state, in contrast, has ππ* character at the Franck-Condon geometry and shows net reactivity with respect to the "hinge-like" geometry change along QLIIC. This can be understood by recognizing that the transition involves π and π* orbitals that are initially largely localized on a single chromophore but then develop increasing charge-separated character, as shown in Figure 3b,c. The diabatic 1 ππ* state progressively develops charge transfer (CT) character as QLIIC → 1, which is neutralized by proton transfer from the C to the Q moiety. Such photoinduced HATs are also frequently termed proton-coupled electron transfer (PCET) or electron-driven proton transfer (EDPT) processes. Upon increasing

General Discussion and Conclusions
This study, which is limited to the isolated heterodimer only, adds to the growing body of computational research aimed at exploring possible excited state decay paths in organic acid-base heterodimers. The present results support earlier suggestions, from analysis of transient absorption measurements in a weakly interacting solvent [95], that HAT from an OH moiety of a catechol sub-unit to the carbonyl-oxygen atom of a quinone unit arranged in a π-stacked C:Q heterodimer could contribute to the pool of photoprotection mechanisms available to eumelanin upon exposure to UV radiation.
From the photophysical perspective, the π-stacked chromophores in the C:Q heterodimer exhibit similarities and differences with the excited state decay mechanisms identified for the chromophores in double-stranded DNA. UV excitation of a, base pair starts with a π*←π promotion localized on the purine (adenine (A) or guanine (G)), which is dissipated by PCET to the pyrimidine (thymine (T) or cytosine (C)) partner and subsequent coupling via a CI to the S 0 state [105][106][107][108][109]. The H atom in these cases is transferred within an H-bonded base pair wherein the individual bases are parts of complementary strands (i.e., an inter-strand HAT process). UV photoinduced intra-strand electron transfer between stacked nucleobases-more reminiscent of the present situation-has been identified also, but the subsequent charge-separation (and ultimate photostability) is again achieved by an inter-strand proton transfer in the resulting radical anion base-pair [110,111].
As noted in the Introduction, eumelanin is a heterogeneous macromolecule, and much remains to be learned both about its exact structure and the mechanisms of the photoprotection it affords. Several studies of intramolecular processes contributing to the decay of excited states of monomers (and oligomers) of various of the proposed key sub-units of eumelanin, like DHI and DHICA, have been reported [75,82,84], along with some studies of their intermolecular interactions with solvent molecules [104]. The present work supports another inter-chromophore excited state decay pathway wherein HAT facilitates non-radiative coupling to, and reformation of, the ground state C:Q heterodimer. But, as experimental studies of (the di-tert-butyl substituted form of) this heterodimer also show, the biradical structure at the asymptote of the HAT coordinate can decompose to two semiquinone radicals [95]. While it is notable from an energetic perspective that absorption of one photon with an energy less than that required to break an O-H bond in bare catechol [88] could result in the formation of two semiquinone radicals, it is unlikely that nature would have adopted eumelanin as a skin pigment if such heterodimers could act as significant light-driven radical generation centers. Clearly, much further work will be needed in order to establish the importance (or otherwise) of the excited state decay pathways identified thus far for small constituent parts to the overall photoprotection afforded by bulk eumelanin.
Supplementary Materials: The following are available online at https://www.mdpi.com/2673-7 256/1/1/3/s1, Cartesian coordinates associated with the various optimized structures of C:Q and (harmonic) normal mode wavenumbers for the ground states of bare C and the C:Q heterodimer.  Data Availability Statement: The data supporting this study are available from the corresponding author on reasonable request.

Conflicts of Interest:
The authors declare no conflict of interest.