Electronic Spectra, Excited State Structures and Interactions of Nucleic Acid Bases and Base Assemblies: A Review

Journal of Biomolecular Structure & Dynamics, ISSN 0739-1102 Volume 25, Issue Number 1, (2007) Adenine Press (2007) Abstract Electronic Spectra, Excited State Structures and Interactions of Nucleic Acid Bases and Base Assemblies: A Review http://www.jbsdonline.com A comprehensive review of recent theoretical and experimental advances in the singlet electronic transitions, excited state structures and dynamics of nucleic acid bases (NABs) and base assemblies are presented. It is well known that NABs absorb ultraviolet radiation, but the absorbed energy is efficiently dissipated in the form of ultrafast internal conversion processes believed to occur in the subpicosecond time scale and, therefore, enabling NABs highly photostable. It is not known how much evolutionary role was played in evolving these molecules and the ultimate selection by nature as genetic materials, but it is well accepted that survival-of-fittest prevails. Recently, significant efforts have been continuously paid to understand the mechanism of electronic excitation deactivation, but universally acceptable mechanism is still elusive. However, recent investigations reveal that electronic excited state geometries of DNA bases are usually nonplanar and this structural nonplanarity may facilitate nonradiative deactivation. Investigation of excited state structures is challenging and, therefore, it is not surprising that despite the impressive theoretical and computational advances, this research area is still hampered by the methodological and computational limitations. Further, stacking has significant influence on the emission properties of molecules. The 2- aminopurine, a fluorescent adenine derivative frequently used in studying DNA dynamics, shows significant attenuations in fluorescence quantum yield when incorporated in the DNA. Theoretical and computational bottlenecks limit a thorough theoretical understanding of effect of stacking interactions on the excited state dynamics of NABs. Despite these limitations the investigations of excited state properties are progressing in the right direction and our better understanding of excited state structure and dynamics of NABs and nucleic acids may help to design preventive strategy for radiation induced illness and photostable materials. M. K. Shukla Jerzy Leszczynski * Computational Center for Molecular Structure and Interactions Department of Chemistry Jackson State University Jackson, Mississippi 39217, USA Key words: Nucleic acid bases; Base pairs; Electronic transitions; Excited states; and Nonradiative decay. Introduction The genetic code in deoxynucleic acid (DNA) is stored in the form of hydrogen bonded purine and pyrimidine bases, the specific patterns of which is unique for each individual. Alteration in DNA structure may lead to mutation by producing a permanent change in the genetic code (1). It has long been speculated that proton transfer may lead to mispairing of bases and thus causing point mutations. Some theoretical investigations on model systems have suggested that the excited state proton transfer proceeds through small barrier height and in some cases it is even barrierless (2, 3). Computational studies on adenine, guanine, and hypoxanthine, on the other hand, have suggested that excited state proton transfer barrier height is significantly large and, therefore, electronic excitation may not facilitate such processes in the excited state for these species (4-6). The exact cause for mutation is not known, but several factors, e.g., environment, irradiation, et cetera, may contribute towards it. It is well known that nucleic acid bases (NABs) absorb ultraviolet (UV) * Phone: 601-979-3723 Fax: 601-979-7823 Email: jerzy@ccmsi.us 93

94 Shukla and Leszczynski radiation efficiently. The formation of pyrimidine dimers between adjacent thymine bases on the same strand results in the most common UV-induced DNA damage (7, 8). Recent investigations suggest that low energy radiation (even less than 3 ev) may also be fatal for the stability of nucleic acid polymers (9, 10). However, the high photostability of NABs is perhaps the reason for their selection as genetic species by nature. The high photostability of NABs is associated with the ultrafast nonradiative decay of absorbed radiation and, therefore, these species show very poor fluorescence; the fluorescence quantum yield being in the order of 10-4 (11-14). Recently, impressive progress has been made investigating the excited state dynamics of NABs at the picosecond and femtosecond time domains (12). These studies clearly show that the excited state life-times of genetic molecules are in the sub-picosecond order and they show very complex excited state dynamics (12). Different possible mechanisms for the ultrafast nonradiative decay in nucleic acid bases have been suggested and they will be discussed in detail latter. They include the out-of-plane vibronic coupling of closed lying electronic ππ* and nπ* states (15, 16) and conical intersection between excited and ground states through some reaction coordinates (12, 17-24). It is clear that excited state geometries of NABs are generally nonplanar and this nonplanarity plays pivotal role in assisting the ultra-fast nonradiative decay (12-14, 17-25). Further, we have also shown that molecular environments, e.g., base pairing, hydration, et cetera, have significant effect on the characteristics of excited state structural nonplanarity and, thus, excited state dynamics would have significant dependency on the molecular environment (26-28). It should be noted that excited state geometries of these molecules are yet not known experimentally; only few studies have indicated the possibility of nonplanar excited state geometry (29-32). Billinghurst and Loppnow (31) have studied the excited state structural dynamics of cytosine using the resonance Raman spectroscopy and time-dependent wave packet analysis and found excited state structural changes consequent to electronic excitation. These authors also computed the distribution of reorganization energy consequent to electronic excitation and found that among pyrimidine bases, the thymine has the largest (66%) and uracil has the lowest (13%) contribution of the reorganization energy along the photochemical relevant coordinates while the contribution for cytosine was revealed to be 31%. The percent contribution of reorganization energy was predicted in agreement with the photodimeric activities of these bases, according to which the thymine shows the most and uracil shows the least UV induced photodimerization reaction. The fluorescence for purine bases (adenine and guanine) are known to originate from the rare tautomer (keto-n7h for guanine and amino-n7h for adenine) (11). However, there is at least one low temperature study which shows that the fluorescence excitation and emission spectra of guanine do not agree with that of the 7-methylguanine; thus, suggesting that the fluorescence in guanine sample does not originate from the minor tautomeric form (33). The positions of substitutions have been found to have profound effect on the photophysical properties of purine bases. For example, the parent molecule, purine, is well know to exhibit strong phosphorescence and insignificant fluorescence. On the other hand 2-aminopurine shows very strong fluorescence and no phosphorescence (11, 34). The photophysical properties of adenine (6-aminopurine) are in between that of purine and 2-aminopurine. Consequently, adenine shows weak fluorescence and weak phosphorescence. The last substantial review on excited state properties of nucleic acid systems was done by Callis in 1983 (11). In this review, he performed an excellent analysis of experimental and theoretical results of electronic transitions of nucleic acid bases and related analogues. But it should be noted that in the early eighties theoretical results were limited to semiempirical methods (the ab initio calculation for this class of molecules were almost impossible at that time). An excellent review article on nucleic acid bases has also appeared recently from Kohler s group (12) and it is mostly devoted to the ultrafast excited state dynamics of bases and base assemblies. The

present review focuses on the recent theoretical and experimental advances in the excited state structures and interactions of nucleic acid bases and base assemblies. Ground State Structures and Properties of Nucleic Acid Bases and Base Pairs The nucleic acid bases are well known to exhibit various tautomeric phenomena in different environments. Although, the presence of sugar in nucleic acid polymers blocks the prototropic tautomerism (N9 N7 in adenine and guanine and N1 N3 in cytosine), it does not stop the possibility of the formation of other tautomeric forms (enol and imino). Different ground state properties (e.g., geometries, tautomerism, transition states corresponding to the proton transfer from the canonical form to the rare tautomeric form, base pair formation, stacking interactions, interactions with metal ions, and hydration) have been discussed in detail in recent review articles (35-37). Therefore, only brief description of ground state properties of NABs and base pairs would be presented here. 95 Electronic Spectra, Excited State Structures, and Interactions of Nucleic Acid Bases and Base Assemblies Earlier experimental investigations have suggested the existence of only two tautomers (N9H and N7H) of adenine; the N9H tautomer is the major form while the relative population of the N7H tautomer has been found to depend upon the environment (38-40). Recent experimental investigation supplemented with theory suggests the existence of three tautomers [N9H (major), N7H and N3H (both minor)] of adenine in dimethyl sulfoxide solution (41). The theoretical calculations also show that although the N9H tautomer is the global minima, the relative stability of the N3H tautomer is very close to that of the N7H tautomer (42, 43). High level experimental and theoretical investigations performed recently suggest that the tautomeric equilibria of guanine are very complex. The existence of up to four tautomers of guanine (keto-n9h, keto-n7h, enol-n9h, and enol-n7h) has initially been suggested using the jet-cooled resonance enhanced two photon ionization (R2PI) spectroscopic investigations (44, 45). However, Choi and Miller (46), on the basis of the comparison of the experimental IR spectra of guanine trapped in the helium droplet and theoretically computed frequencies at the MP2 level using the 6-311++G(d,p) and aug-cc-pvdz basis sets, have recently assigned the presence of only keto-n9h, keto-n7h, and cis- and trans forms of the enol-n9h tautomer of guanine. Based on the results of Choi and Miller (46), Mons et al. (47) have reassigned their experimental findings and accordingly the enol-n9h-trans, enol-n7h, and two rotamers of the keto-n7h-imino tautomers of guanine are present in the supersonic jet-beam. However, it is surprising, since imino tautomers are much less stable than the canonical form of guanine and probably they are formed during the laser desorbtion of guanine in the experiments. It is generally believed that the pyrimidine bases uracil and thymine exist mainly in the keto form (35-37). However, the existence of a small amount of the enol tautomer in aqueous solutions of 5-chlorouracil at room temperature has been suggested by Suwaiyan et al. (48). A trace amount of the enol form of thymine in aqueous solutions has also been suggested by Morsy et al. (49) on the basis of extensive UV/ Vis absorption and fluorescence measurements. Although, the Hobza group (50) does not support the utility of experiments used by Morsy et al. (49) in tautomer detection, but do not completely ruled out the presence of trace amount of minor tautomers in the water solution. Cytosine exists as a mixture of the amino-hydroxy and amino-oxo (N1H) tautomeric forms with the equilibrium being shifted slightly towards the former tautomeric form in the argon and nitrogen matrices (51, 52). A matrix isolation study of 1-methylcytosine and 5-methylcytosine indicates the existence of the imino-oxo tautomeric form (53, 54). Three tautomers (amino-oxo, imino-oxo, and amino-hydroxy) of cytosine have been found in a microwave study (55). In aqueous solutions both of the amino-oxo forms (N1H and N3H) are present (56). In crystals, mainly the N1H amino-oxo form is found (57). The theoretical results for cytosine are available up to the CCSD(T) level of theory using a complete basis set approach in which the energies are obtained by applying an extrapolation

96 Shukla and Leszczynski technique (58). It has been found that the coupled-cluster approach with single, double, and triple excitations [CCSD(T)] is necessary to predict the relative stability of cytosine tautomers (59). In the gas phase the amino-hydroxy tautomer is predicted to be the most stable; however, under aqueous solvation tautomeric stability is found to be shifted to the canonical amino-oxo form (58, 60). It is established that the amino groups of NABs are pyramidal due to the partial sp 3 pyramidalization of the amino nitrogen (35-37, 61). The amino group pyramidalization of guanine is highest among the nucleic acid bases (35-37). Experimental evidence for the nonplanarity of adenine and cystosine has been recently indicated in the vibrational transition moment direction measurement study by Dong and Miller (62). Further, it has also been revealed theoretically that the pyrimidine ring in the NABs possesses high conformational flexibility (63, 64). The electron (proton) affinity of a molecule is measured in terms of the amount of energy released when an electron (proton) is added to the molecule. It is computed as the energy difference between the neutral and anionic (cationic) forms of the molecule. Ionization potential on the other hand is defined as the amount of energy required to remove an electron from a molecule. It is computed as the energy difference between the cationic and neutral forms of the molecule. In a recent theoretical study, Li et al. (65) with the help of available experimental data have estimated the value of adiabatic valence electron affinities to be in the range of 0-0.2 ev for pyrimidines and about -0.35 and -0.75 ev for adenine and guanine, respectively. The purines have lower and pyrimidines have higher ionization potentials and it is clear that guanine has the lowest ionization potential among the nucleic acid bases and, therefore, is the most susceptible for oxidation under irradiation (66-71). Experimental (72-74) and high level theoretical investigations (75-77) were also performed to determine the protonation and deprotonation (basicity and acidicity) properties of the different sites of nucleic acid bases. Our group (75) has computed proton affinities of all nucleic acid bases up to the MP4(SDTQ) level and found that the computed proton affinities are very close to the experimental data; the computed error was found to be within the 2.1%. The Watson-Crick (WC) base pair geometries are generally planar including the amino group at the HF and DFT levels (35, 37, 78-80). At the MP2 level with smaller basis sets, the amino groups of the WC GC and AT base pairs are pyramidal, but with larger basis sets the corresponding group of the AT base pair was revealed almost planar (80, 81). It has been suggested that the nonplanarity of GC base pair may enhance the stacking of bases on the strand and may increase the stability of the helix (81). The structural properties of different reverse Watson-Crick (RWC), Hoogsteen (H), and reverse Hoogsteen (RH) base pairs have also been investigated, and the geometries of some of them have been found to be nonplanar (35, 37, 82). Recently, the energetics of hydrogen bonded and stacked base pairs were studies up to the CCSD(T) level (83-85). Kumar et al. (86, 87) have recently investigated the adiabatic electron affinities of GC, AT, and hypoxanthine-cytosine base pairs at the DFT level and found the significant increase in the electron affinity of the AT base pair under the polyhydrated environments. A comprehensive investigation of structure and properties of deprotonated GC base pair was recently performed by Schaefer and coworkers (88). Excited State Properties of Nucleic Acid Bases Ground state geometries of nucleic acid bases are planar (except the amino group, which is pyramidal) (35-37, 61), while the corresponding excited state geometries are generally nonplanar (4-6, 12-14, 25-28, 89-97). The excited state structural nonplanarities may facilitate the ultrafast nonradiative decay in bases and base assemblies (12-14, 17-25, 96). The modes of interaction of NABs with water molecules are also found to be different in the electronic excited states compared to the ground state (26,

28, 98-101). The hydrogen bond accepting sites under the nπ* excitations provide repulsive potential for hydrogen bonding interactions (100, 101). Consequently base pairs are destabilized under such excitations (78, 79). In femtosecond spectroscopic investigations of adenine-water clusters, the adenine-water hydrogen bonds were found to be dissociated on the nπ* potential energy surfaces of adenine (98, 99). Electronic Transitions 97 Electronic Spectra, Excited State Structures, and Interactions of Nucleic Acid Bases and Base Assemblies Adenine: In 1954 Mason first suggested that the main absorption band of adenine observed near 260 nm (4.77 ev) consists of two electronic transitions differing with respect to the relative intensity and the transition moment directions measured according to the DeVoe-Tinoco convention (Fig. 1) (102). However, in the vapor phase and in a trimethyl phosphate (TMP) solution of adenine, these transitions are not resolved (103, 104). In the water solution, a stronger transition appearing at 261 nm (4.75 ev) is short axis polarized, while another transition appearing as a weak shoulder near 267 nm (4.64 ev) is long axis polarized (105). Similar results were also found in the linear dichroism (LD) spectra of 9-methyladenine partially oriented in stretched polymer poly(vinyl alcohol) films (106), in the polarized absorption spectra of 9-methyladenine in crystal environments (105), and in the photoacoustic spectra of the evaporated film of adenine (107). However, the splitting between these two transitions appreciably increases in a crystal environment compared to solution (105). In the photoacoustic spectra (107), four absorption peaks were revealed in the 300-180 nm region with stronger transition found near 270 nm (4.59 ev) and a weaker transition detected near 290 nm (4.28 ev). It is interesting to note that the splitting of the 260 nm band is observed generally in all experiments, in linear dichroism (LD) (106, 108, 109), in magnetic circular dichroism (MCD) (110, 111), in single crystal absorption (105, 112), in fluorescence polarization (113), and H62 H61 O6 N6 Φ Φ H1 C6 N7 H8 H21 C4 C4 C2 N2 N9 N9 N3 N3 H9 H22 H9 Adenine (A) Guanine (G) H42 H41 O4 N4 Φ C4 C6 O2 H62 H1 O4 N7 H9 C5 R5 C4 C6 H3 C2 C2 H2 C5 C6 N3 N1 C4 N3 Cytosine (C) H61 N6 O2 AU(T) base pair H6 N1 Uracil (U) (Thymine (T)) N9 C6 C2 H6 N1 H1 C8 C5 N3 C2 H8 H5 C4 C5 N3 Figure 1: Structure and atomic numbering schemes of nucleic acid bases and Watson-Crick base pairs. In uracil, R5/R5 =H and in thymine, R5/R5 =CH3. The Φ represents the transition moment direction according to the DeVoe-Tinoco convention (11). Φ R5 H3 O2 H8 C8 C8 C2 N7 C5 N1 C5 N1 H2 C6 N7 H8 C8 H6 N9 H9 N1 H1 C5 H42 H41 O6 C6 C4 H5 N4 C4 N1 C5 N3 C6 H1 N3 C2 C2 N2 H22 H21 N1 O2 GC base pair H1 H6

98 Shukla and Leszczynski in substituent effects (114), but it has not been found in the CD spectra (115-117). The transition moment direction (according to the DeVoe-Tinoco convention, Fig. 1) for the stronger component is found to be -3º for 9-methyladenine in single crystals Table I Summary of experimental transition energies ( E, ev) of adenine, guanine, thymine, uracil, cytosine, and their derivatives. The f represents oscillator strength and represents transition moment direction ( ) according to the Devoe-Tinoco convention (Figure 1). Molecule/Transitions References Adenine Absorption Spectra E 4.92 5.99 Adenine, vapor (103) E 4.77 5.96 Adenine, TMP (103) E 4.81 5.85 9MA, MCH (103) E 4.77 5.90 9MA, TMP (103) E 4.77 5.99 Adenine, water (104) E 4.59 4.77 5.90 Adenine, water (130) E 4.63 4.77 6.05 Adenine, water (110) E 4.77 6.02 Adenosine, water (110) E 4.59 5.90 6.81 7.75 Adenine sublimed film (164) E 4.51 4.68 5.82 6.08 6.81 7.75 9MA, crystal (120) f 0.1 0.2 0.25 0.11 0.30 0.23 83 25-45 15 72 6 LD spectra E 4.55 4.81 5.38 5.80 5.99 9MA, stretched film (106) f 0.047 0.24 0.027 0.14 0.12 66 19-15 -21-64 CD spectra E 4.63 5.93 6.36 Adenines, water (115) E 4.77 5.74 6.36 6.63 Adenines, water (116) E 4.68 5.51 Adenosine, water (110) MCD spectra E 4.59 4.92 5.90 Adenine, water (110) E 4.56 4.90 5.77 Adenosine, water (110) Photo acoustic spectra E 4.28 4.59 6.20 6.89 Adenine, film (107) Electron scattering E 4.53 5.84 6.50 7.71 Adenine, film (116) Guanine Absorption Spectra E 4.46 5.08 6.20 6.57 Guanine, model (132) f 0.15 0.24 0.40 0.48-12 80 70-10 E 4.51 5.04 6.33 Guanine, water (104) E 4.56 5.04 6.19 6.67 Guanosine, water (132) f 0.15 0.24 0.40 0.48 a -24 88 86-8 to 44 E 4.56 4.98 6.02 6.63 9EtG, water (131) f 0.14 0.21 0.38 0.42 E 4.51 4.84 6.11 6.52 9EtG, TMP (114) E 4.51 4.92 6.05 6.59 9EtG, water (114) E 4.46 4.88 5.46 6.08 6.56 9EtG, crystal (131) f 0.16 0.25 <0.05 0.41 0.48-4 -75-75 -9 E 4.35 5.00 6.23 6.70 Guanine, sublimed film (164) LD Spectra E 4.43 5.00 Guanine, stretched film (109) 4-88 CD Spectra E 4.51 4.92 5.51 6.20 6.63 dgmp, water (116) E 5.06 5.77 Guanosine, water (110) MCD Spectra E 4.46 5.00 Guanosine, water (110)

(transition being at 275 nm) (105, 112), while it amounts to 9º in the film dichroism study (transition being at the 263 nm) (109). In the case of the protonated adenine, transtion moment of the strong transition near 257 nm (4.82 ev) makes an angle of 100º, while that of the weak transition near 273 nm (4.54 ev) is -28º with respect to the C4C5 direction (118). An extensive and elegant work in this regard was performed by Clark (119, 120) to model the electronic spectra of adenine in the UV and vacuum UV region. In this study, he has measured the polarized spectra of crystals of 9-methyladenine and 6-(methylamino)purine and assigned eight bands of adenine along with their transition moment directions and oscillator strengths. The strong transition (265 nm) of the main UV absorption band was shown to be polarized at 25º and weaker transition (near 275 nm) was found to be polarized close to the long molecular axis. The transition moment directions of several transitions of 9-methyl and 7-methyl adenine samples (9MA and 7MA) oriented in stretched polymer films were also measured (106). The existence of a new ππ* transition near 5.38 ev for 9MA (which had not previously been observed) was also revealed. The measured transition moment directions for the first two transitions are generally in agreement with those suggested by Clark (119, 120). However, the transition moment directions for higher energy transitions are different from those obtained by Clark (120). 99 Electronic Spectra, Excited State Structures, and Interactions of Nucleic Acid Bases and Base Assemblies Table I Continued Uracil Absorption Spectra E 5.08 6.05 6.63 Uracil, vapor (103) E 4.84 6.05 6.63 1,3-dimethyluracil, vapor (103) E 4.68 6.08 6.63 1,3-dimethyluracil, water (103) E 4.81 6.11 Uracil, water (110) E 4.75 6.05 Uridine, water (110) E 4.81 6.11 6.85 Uracil, TMP (114) E 4.79 6.14 6.85 Uracil, water (114) E 4.70 6.02 6.74 1,3-dimethyluracil, TMP (114) E 4.73 6.11 6.81 1,3-dimethyluracil, MCH (114) E 4.51 5.82 1-methyluracil, crystal (151) -9 59 E 4.66 6.08 6.97 7.90 Uracil, sublimed film (164) CD spectra E 4.73 5.77 6.36 7.00 Uridine, water (116) E 4.63 5.71 Uridine, water (110) E 4.68 5.82 6.26 Uridine, water (115) MCD spectra E 4.86 5.85 Uracil, water (110) E 4.77 5.71 Uridine, water (110) Electron Scattering E 4.70 5.93 6.93 Uracil, film (146) Thymine Absorption Spectra E 4.68 6.08 Thymine, water (110) E 4.64 6.05 Thymidine, water (110) E 4.54 5.99 1-methylthymine, water (105) f 0.19 0.28 E 4.64 5.88 7.04 Thymine, sublimed film (164) CD spectra E 4.68 5.77 6.36 7.00 Thymidine, water (116) E 4.54 5.69 Thymidine, water (110) E 4.63 5.85 6.42 Thymidine, water (115) MCD spectra E 4.71 5.77 Thymine, water (110) E 4.73 5.64 Thymidine, water (110) Photo Acoustic spectra E 4.59 5.90 7.08 Thymine, film (107) Electron scattering E 4.66 5.94 7.08 8.82 Thymine, film (146)

100 Shukla and Leszczynski Table I Continued Cytosine Absorption Spectra E 4.66 5.39 5.85 6.29 Cytosine, water (165) f 0.14 0.03 0.13 0.36 b 6-46 76-27 or 86 E 4.64 6.31 Cytosine, water (104) E 4.57 5.39 6.26 Cytidine, water (104) E 4.48 5.23 6.08 6.63 Cytosine, TMP (114) E 4.59 5.28 5.74 6.26 dcmp, water (116) E 4.64 5.21 5.83 6.46 Cytosine, water (167) f 0.096 0.100 0.211 0.639 E 4.57 5.34 5.77 6.26 Cytidine, water (166) E 4.57 6.17 Cytosine, sublimed film (164) E 4.54 5.40 6.07 6.67 7.35 Cytosine, sublimed film (168) f 0.058 0.073 0.115 0.072 0.072 LD spectra E 4.63 5.17 Cytosine, polymer film (109) 25±3 6±4 or or -46±4-27±3 CD spectra E 4.59 5.27 5.74 6.14 6.56 7.38 dcmp, water (116) E 4.59 5.02 5.64 6.36 Cytosine nucleosides (169) E 4.59 5.17 5.64 6.36 Cytidine c (169) TMP, Trimethylphosphate; 9MA, 9-methyladenine; MCH, Methylcyclohexane; Adenines, Adenine derivatives: For details see relevant references. a Based on polarized absorption spectra of crystalline guanosine (132); 9EtG, 9-ethylguanine; TMP, Trimethylphosphate: For details see relevant reference. b Based on polarized spectra of cytosine crystal (165); TMP, Trimethylphosphate: For details see relevant reference. c Based on CD and absorption measurements of cytosine nucleosides in different solvents (water, acetonitrile, dioxane, 1,2-dichloroethane). The existence of a transition near 230 nm (5.39 ev) was also indicated in the MCD (110) and CD (115, 116) spectra, but on the basis of the semiempirical calculations this transition was assigned as being of the nπ* type (11). The experimental electronic transitions of NABs and their derivatives are summarized in Table I. The tentative assignment of the existence of nπ* transitions near 244 and 204 nm (5.08 and 6.08 ev) in the crystal of 2ʹ-deoxyadenosine was made by Clark (121). The possibility of the existence of such nπ* transitions is also supported from a recent theoretical study (78). There are also some investigations suggesting the existence of an nπ* transition near the first singlet ππ* transition (94, 98, 99, 106, 122). The linear dichroism measurements of adenine derivatives partially oriented in stretched polymer poly(vinyl alcohol) films have yielded the existence of an nπ* transition near the first ππ* absorption transition in 9-methyl adenine (106). Similar results were also found in the molecular beam study of hydrated adenine clusters (98, 99). The existence of the nπ* transition as the first transition (the energy is very close to the first ππ* transition) in adenine in the gas phase is also predicted at the time dependent density functional theory (TDDFT) and the multi-reference perturbation configuration interaction method (known as CIPSI) (94). Kim et al. (122) have performed REMPI and fluorescence studies of jet-cooled adenine and have suggested that the first transition of adenine has nπ* character with the 0-0 band located at 35503 cm -1 (~281.7 nm, ~4.40 ev), while the corresponding band of the first ππ* transition is located at 36108 cm -1 (~276.9 nm, ~4.48 ev). Luhrs et al. (123) have performed a similar study of adenine and 9MA, but their results do not support the assignment of the nπ* transition suggested by Kim et al. (122). Luhrs et al. (123) have speculated that the nπ* peak observed by Kim et al. (122) may be due to the formation of other tautomers of adenine since the latter study involved the use of higher temperatures in heating the sample. Luhrs et al. (123) have observed the 0-0 band of the first ππ* transition of adenine and 9MA at 36105 cm -1 (~277 nm, ~4.48 ev) and 36136 cm -1 (~276.7 nm, ~4.48 ev), respectively,

and these results are in accordance with the observation made by Kim et al. (122). Similar results were also found from the REMPI study by Nir et al. (124) who used the laser desorption technique instead of heating the samples. The first ab initio calculations of the electronic transitions of adenine (and guanine) were performed at the multi-reference configuration interaction (MRCI) and random phase approximation (RPA) levels using the ground state self-consistent field orbitals with double-ζ/polarization/diffuse gaussian basis set utilizing the experimental molecular geometry and assuming its planarity (125). The computed transition energies were higher by 1.48-1.86 ev compared to the experimental transition energies, and linear scaling was needed for comparison with experimental data. Roos and coworkers (126) have used the CASSCF/CASPT2 level of theories applying a large ANO-type basis set to study the electronic transitions of the planar form of adenine. As expected, the CASPT2 correlation correction to the CASSCF energies yielded significant improvements in the CASSCF excitation energies and were found to be in reasonably good agreement with the corresponding experimental data. The TDDFT (94, 127-129) and configuration interaction singles (CIS) (13, 14, 93, 94, 96) methods were also used to study the excited state properties of adenine with reasonable success. The scaled [scaling factor 0.72 (13, 14, 78, 96)] CIS transition energies were found to be in good agreement with the experimental data and the corresponding CASPT2 transition energies. It should be noted that, unless otherwise stated, the discussed CIS computed transition energies of NABs in comparing with the corresponding experimental data and other theoretical results in the current manuscript correspond to the scaled values. 101 Electronic Spectra, Excited State Structures, and Interactions of Nucleic Acid Bases and Base Assemblies Table II shows the vertical singlet ππ* and nπ* transition energies, transition moment directions and dipole moments of the adenine tautomers (N9H and N7H), their hydrated forms obtained at the CIS/6-311G(d,p)//HF/6-311G(d,p) level (13, 14), along with the CASSCF and CASPT2 excitation energies (126), and some experimental data. The super molecular approach considering three water molecules in the first solvation shell of the adenine tautomers was used to model aqueous solvation. The first ππ* transition of the N9H tautomer is stronger, while the second ππ* transition is predicted to be much weaker. After hydration the transition energy of the weaker transition is decreased; therefore, the stronger transition becomes the second transition (Table II). Experimentally, a weak shoulder near 270 nm (4.59 ev) and a strong peak near 260 nm (4.77 ev) in the water solution are observed (130). Thus, the calculated transitions of the hydrated N9H tautomer are in a qualitative agreement with the experimental data (130), although the computed splitting between the two transitions is too small (Table II). Further, the experimental transition energies shown in Table II can be explained within an accuracy of 0.2 ev in terms of the scaled computed transition energies of the hydrated N9H tautomer. The CIS calculation predicts that the two scaled transitions computed at 6.18 and 6.24 ev for the isolated N9H tautomer and at 6.12 and 6.17 ev for its hydrated form (Table II) would contribute to the 6.2 ev experimental region of the molecule (Table I). The calculation predicts that the transition moment directions of these transitions would be approximately perpendicular to each other (Table II). The MCD results suggest that the UV-absorption band in the 200 nm (6.2 ev) region is composed of two transitions with non-parallel transition dipole moments (111). Therefore, the theoretical CIS results may correspond to the MCD observation in this regard. Although the predicted weak transition near 5.38 ev in the LD spectra of 9MA (106) is not obtained in the calculations; however, it was calculated for the planar form of adenine (78). The agreement between the CIS computed singlet ππ* transition energies of the N7H tautomer and those obtained by CASPT2 calculations (126) and the LD technique (106) is good for the first two transitions; however, such agreement is not reached for higher energy transitions (Table II). Due to the close proximity of the computed transition energies of the N7H tautomer and its hydrated form to those of the N9H tautomer and its hydrated form, contributions to the observed spectra of adenine

102 Shukla and Leszczynski from the N7H form cannot be ruled out. It is known from different experimental and theoretical studies that the N7H tautomer is present along with the N9H form under different environmental conditions (38-43). Three nπ* transitions near 5.18, 5.52, and 5.74 ev (scaled values) for the N9H tautomer of adenine are predicted at the CIS/6-311G(d,p) level. The corresponding values for the hydrated form are 5.38, 5.75, and 5.97 ev, respectively (Table II). The computed first nπ* transition may be related to that indicated in the MCD (110) and CD (115, 116) spectra in the 230 nm (5.39 ev) region as discussed earlier. Further, it can also be suggested as the possible source of the first nπ* transition located near 244 nm (5.08 ev) as indicated in 2ʹ-deoxyadenosine (121). Although, it is not possible to relate the second computed nπ* transition with experiment, the third computed transition near 5.97 ev (hydrated form) corresponds to the 204 nm (6.08 ev) transition of 2ʹ-deoxyadenosine (121). Table II Vertical singlet * and n * excitation energies ( E, ev), oscillator strengths (f), transition moment directions (, ), and dipole moments (μ, Debye) of the N9H and N7H tautomers of adenine in the isolated and hydrated forms at the CIS/6-311G(d,p)//HF/6-311G(d,p) level (13, 14). Isolated CIS Hydrated Experimental Data a CASPT2/CASSCF b Abs Crystal LD E f μ c E d E f E d E 1 / E 2 /f/ /μ E E/f/ E/f/ N9H * Transitions 6.61 0.394 60 2.85 4.76 6.61 0.440 50 4.76 5.20/6.48/0.37/37/2.30 4.77 4.68/0.2/25 4.81/0.24/19 6.65 0.024-6 3.40 4.79 6.59 0.038-66 4.74 5.13/5.73/0.07/23/2.37 4.59 4.51/0.1/83 4.55/0.047/66 8.20 0.398-38 0.83 5.90 8.09 0.342-31 5.82 6.24/7.80/0.851/-57/2.13 5.90 5.82/0.25/-45 5.80/0.14/-21 8.58 0.447 15 2.02 6.18 8.50 0.423 19 6.12 6.21 e 6.15 f 6.72/8.30/0.159/40/4.60 6.08/0.11/15 5.99/0.12/-64 8.67 0.547-87 3.14 6.24 8.57 0.589-77 6.17 9.39 0.232 29 2.65 6.76 9.41 0.375 23 6.78 6.99/8.77/0.565/27/3.42 6.81/0.30/72 n * Transitions 7.19 0.001-2.47 5.18 7.47 0.000-5.38 6.15/6.43/0.001/-/2.14 7.66 0.002-0.93 5.52 7.99 0.001-5.75 6.86/7.16/0.001/-/1.93 7.97 0.014-1.62 5.74 8.29 0.015-5.97 N7H * Transitions 6.38 0.162 35 6.83 4.59 6.36 0.175 28 4.58 4.61/5.12/0.050/23/5.95 4.54/0.11/45 6.78 0.051 3 6.51 4.88 6.84 0.103 16 4.92 4.97/6.63/0.187/-10/9.64 4.90/0.094/-16 8.06 0.766 81 6.87 5.80 7.99 0.765 77 5.75 6.02/7.81/0.363/3/8.68 5.28/0.052/-28 8.27 g 0.163-44 1.75 5.95 8.44 0.579-32 6.08 6.15/7.22/0.123/-49/6.70 5.68/0.16/76 8.47 0.377-12 6.00 6.10 8.55 0.207 69 6.16 6.32/8.12/0.077/52/6.72 5.91/0.19/-29 n * Transitions 6.87 0.012-4.73 4.95 7.24 0.004-5.21 7.33 0.002-6.32 5.28 7.61 0.001-5.48 7.78 0.014-4.24 5.60 8.11 0.015-5.84 a Abs, Absorption in aqueous medium (130); Crystal, based on the polarized spectra of single crystals of 6-(methylamino)purine and 9-methyladenine (120); LD, LD spectra of 9-methyladenine and 7-methyladenine oriented in stretch poly(vinyl alcohol) film (106); b E 1 corresponds to CASPT2 and E 2 corresponds to CASSCF transition energies (126); c Ground state dipole moments of the N9H and N7H tautomers at the HF/6-311G(d,p) level are 2.51 and 6.83 Debye, respectively; d Scaled (scaling factor 0.72) excitation energies; e Average of transitions at 6.18 and 6.24 ev; f Average of transitions at 6.12 and 6.17eV; g Rydberg contamination. Guanine: The existence of five electronic transitions in the UV region has been suggested in guanine (11, 126). The first transition lies near 275 nm (4.51 ev) and the second appears near 250 nm (4.96 ev); the intensity of the latter being larger than the former one (11, 110, 114, 126, 131-135). The third transition is located in the 225 nm (5.51 ev) region. It is a weak transition with the oscillator strength in the range of 0.01-0.03 and is not very often observed. Evidence for the existence of such a band is found in the CD spectra (116, 117), in the crystal spectra of guanine and 9-ethylguanine, and in aqueous solutions of protonated guanine (131). However, definite information could not be obtained from the CD spectra and this

transition was suggested to be due to a weak ππ* or nπ* transition (116). The fourth and fifth transitions are intense and located near 204 nm (6.08 ev) and 188 nm (6.59 ev), respectively (11, 114, 131, 132). The existence of three nπ* transitions near 238, 196, and 175 nm (5.21, 6.32, and 7.08 ev) in guanine has been suggested by Clark, but the assignment is not certain (132). The precise measurement of transition moment directions in the study of crystal spectra is complicated by the presence of crystal field (105, 136). Callis and coworkers (137) have estimated the angle between the I and II bands to be about 61 ± 10, while it was found to be 71º by Clark (131). However, it is now clear that the first band is polarized along the short axis (C4C5), while the second band (near 4.96 ev) is long axis polarized (108, 109, 131, 132). Recently, Clark (132) performed a very extensive and elegant study to determine the transition moment directions in guanine using polarized absorption spectra of a single crystal of guanosine dihydrate. Based on his investigations and by comparing with earlier results, he suggested that directions in guanine for transitions near 4.46, 5.08, 6.20, and 6.57 ev would be -12, 80, 70, and -10 degrees, respectively. Some advanced spectroscopic studies have been performed on guanine and substituted analogs, guanine-guanine, and guanine-cytosine base pairs (44, 45, 138-142). These investigations included: REMPI studies of guanosines (139) and guanine (140); REMPI and spectral hole burning (SHB) studies of guanine, methyl guanine (44), guanine-guanine base pairs, guanine-cytosine base pairs (138), guanine, and hydrated guanine (142); and REMPI and IR-UV depletion spectroscopic studies of guanine, and methyl guanine (45). In these studies (44, 45, 138-140, 142), the spectral origin (0-0 transition) of the S 1 excited state and some lower vibrational frequencies were determined, and the existence of different tautomers of guanine was investigated (44, 45, 142). The tautomeric distribution in guanine in low temperature was compounded by the recent experimental and theoretical investigation of Choi and Miller (46) by trapping guanine in the helium droplets and subsequent reassignment of R2PI spectra by Mons et al. (47) which showed the existence of imino tautomeric forms. However, it should be noted that according to this reassignment, the keto-n9h as well as keto-n7h tautomers have not yet been observed. This reassignment is also supported by other theoretical results (97, 43). Our recent detailed theoretical investigation on all guanine tautomers have predicted that the spectral origins of the keto-n9h and keto-n7h tautomers will be in between the spectral region covering the spectral origin of the keto-n7h- IMINO-cis and the enol-n9h-trans tautomers (143). 103 Electronic Spectra, Excited State Structures, and Interactions of Nucleic Acid Bases and Base Assemblies The CASSCF/CASPT2 investigation of the singlet electronic transitions of guanine was performed by Roos and coworkers (126) employing the ANO-type basis set and using the MP2/6-31G(d) optimized planar geometry. The effect of the aqueous solvent on electronic transitions was considered using the self-consistent reaction field (SCRF) model. The computed CASPT2 transition energies were found to be in reasonably good agreement (with an accuracy of 0.3 ev) with the experimental data, while the CASSCF transition energies were much larger. Electronic transition energies of guanine were also computed at the TDDFT (95, 127-129, 144) and CIS (13, 14, 92) methods and in one of the TDDFT calculation basis sets with several set of diffuse functions were used (127). In the case of CIS method the scaling factor 0.72 (13, 14, 96) was used to compare obtained energies with the experimental data and the corresponding CASPT2 transition energies. Mennucci et al. (95) have studied the photophysical properties of guanine tautomers (keto-n9h and keto-n7h) including the excited state tautomerization theoretically at the TDDFT, CIS, and multireference perturbation configuration interaction (CIPSI) methods both in the gas phase and in water solution modeled using the continuum model. The role of protonation in the excited state proton transfer from the keto-n9h to the keto-n7h form and the occurrence of fluorescence from the latter tautomer in the water solution have also been discussed. Table III shows the computed vertical singlet ππ* and nπ* transition energies, transition moment directions and dipole moments of the keto-n9h and keto-n7h tau-

104 Shukla and Leszczynski Table III Vertical singlet * and n * excitation energies ( E, ev), oscillator strengths (f), transition moment directions (, ), and dipole moments (μ, Debye) of the keto-n9h and keto- N7H tautomers of guanine in the isolated and hydrated forms at the CIS/6-311G(d,p)//HF/6-311G(d,p) level (13,14). CIS Abs 1 Range Experimental Data a Hydrated Abs 2 CD CASPT2/CASSCF b Isolated E f μ c E d E f E d E 1 / E 2 /f/ /μ E 3 /f/ E E/f/ E E keto-n9h * Transitions 6.39 0.282-42 5.96 4.60 6.45 0.245-25 4.64 4.76/6.08/0.113/-15/7.72 4.73/0.154/-4 4.51 4.56/0.15/-24 4.51 4.4-4.6 7.25 0.516 66 7.74 5.22 7.18 0.567 70 5.17 5.09/6.99/0.231/73/6.03 5.11/0.242/75 5.04 5.04/0.24/88 4.92 4.8-5.1 8.32 0.104 51 6.18 5.99 8.27 0.089 59 5.95 5.96/7.89/0.023/7/5.54 5.98/0.021/6 5.51 5.4-5.5 9.25 e 0.113 79 5.52 6.66 6.67 f 9.13 0.512-89 6.57 6.65/8.60/0.161/-80/10.17 6.49/0.287/-85 6.33 6.19/0.40/86 6.20 6.0-6.3 9.26 0.356 81 6.17 6.67 n * Transitions 6.97 0.001-4.71 5.02 7.28 0.001-5.24 5.79/6.22/10-4 /-/4.31 7.82 0.010-5.84 5.63 8.01 0.010-5.77 6.60/8.05/0.013/-/4.63 8.58 0.003-7.24 6.18 8.89 0.002-6.40 6.63/7.97/0.002/-/2.64 keto-n7h * Transitions 6.16 0.222-3 1.71 4.44 6.11 0.277 11 4.40 7.52 0.232 64 1.60 5.41 7.44 0.095 51 5.36 8.03 0.614 86 1.35 5.78 7.86 0.796 89 5.66 n * Transitions 7.05 0.001-4.66 5.08 7.53 0.001-5.42 8.06 0.058-3.07 5.80 8.18 0.004-5.89 a Abs 1, Absorption of guanine in water (104); Abs 2, Absorption of guanosine in water and values are based on polarized absorption spectra of crystalline guanosine (132); CD, CD spectra in aqueous solution of deoxy guanosine 5 -phosphate (dgmp) (116); b E 1 corresponds to CASPT2 and E 2 corresponds to CASSCF transition energies in the gas phase; E 3 /f/ corresponds to results in water (126); c Ground state dipole moments of the keto-n9h and keto-n7h tautomers at the HF/6-311G(d,p) level are 6.77 and 1.78 Debye, respectively; d Scaled (scaling factor 0.72) excitation energies; e Rydberg contamination; f Average of transitions at 6.66 and 6.67 ev. tomers of guanine, and their hydrated forms (with three water molecules) obtained at the CIS/6-311G(d,p)//HF/6-311G(d,p) level (13, 14) along with some experimental and CASSCF and CASPT2 results (126). It should be noted that the keto-n9h tautomer in the gas phase is found to be about 0.86 kcal/mol more stable than the keto-n7h form while under hydration with three water molecules the latter tautomer is found to be about 3.19 kcal/mol more stable than the former tautomer (at the HF/6-311G(d,p) level). Data shown in Table III suggest that the scaled CIS/6-311G(d,p)//HF/6-311G(d,p) and CASPT2/CASSCF results are in agreement with respect to the assignment of the first nπ* transition as being due to the excitation of the carbonyl group lone pair electron. Further, the order in transition intensity obtained at the CIS level agrees with the solution spectra of guanine and its derivatives in which the first transition (near 275 nm) appears as a weak peak in comparison to the stronger peak near the 250 nm region (11). Further, there is a good correspondence between the computed transitions (scaled) of the keto-n9h tautomer (and its hydrated form) and the CASPT2 results (solvation included), in particular when comparison is made with the transition of the hydrated tautomer. However, the agreement is better in the lower than in the higher energy region. The third ππ* transition computed at 5.95 ev of the hydrated keto-n9h tautomer has the lowest oscillator strength among all the ππ* transitions shown in Table III. This transition can be considered for an explanation of the 5.5 ev band in the experimental data (116, 117, 131). Therefore, the calculations favor this transition as a weak ππ* type. Two almost degenerate transitions near 6.66 and 6.67 ev (scaled values) of the keto-n9h tautomer in the gas phase correspond to a single transition at 6.57 ev for the hydrated form, which explains the 6.0-6.3 ev experimental region of guanine (Table III). This result may be related to that of the MCD observation of guanosine 5 -diphosphate which shows that the 200 nm (6.2 ev) band is composed of two transitions (111). The experimental measurement of 7- methylguanine shows that the first absorption peak is about 10 nm red-shifted

relative to that of guanosine monophosphate (GMP) (145). The predicted transition energy of the first ππ* transition of the keto-n7h tautomer and its hydrated forms is lower than the corresponding transition energies of the keto-n9h tautomer and its hydrated forms (Table III). Therefore, the calculated result is in agreement with the experimental data. Computed nπ* transitions of the hydrated keto-n9h tautomer are found to be at 5.24, 5.77, and 6.40 ev (Table IV). These results support the findings of Clark (132) with regard to the existence of the nπ* transitions near 5.21 and 6.32 ev in guanine. On the basis of theoretical predictions of a ππ* transition near 5.95 ev and an nπ* transition near 5.77 ev (Table III), it appears that in the 5.5 ev region of guanine, the weak ππ* and nπ* transitions are present and are responsible for the ambiguous assignment of transitions in that region. 105 Electronic Spectra, Excited State Structures, and Interactions of Nucleic Acid Bases and Base Assemblies Table IV Vertical singlet * and n * excitation energies ( E, ev), oscillator strengths (f), transition moment directions (, ), and dipole moments (μ, Debye) of the keto tautomer of uracil in the isolated and hydrated forms at the CIS/6-311G(d,p)//HF/6-311G(d,p) level (101). Isolated CIS Hydrated CASPT2/CASSCFb Experimental Dataa Abs 1 Abs 2 CD Crystal Range E f μ c E d E f E d E 1 / E 2 /f/ /μ E E E E/ E * Transitions 6.83 0.446-7 5.07 4.92 6.74 0.447-6 4.85 5.00/6.88/0.19/-7/6.3 5.08 4.79 4.73 4.51/-9 4.6-5.1 8.89 0.123 36 3.48 6.40 8.73 0.140 46 6.29 5.82/7.03/0.08/-29/2.4 6.05 6.14 5.77 5.82/59 5.8-6.1 9.29 0.386-66 4.99 6.69 9.12 0.439-57 6.57 6.46/8.35/0.29/23/6.9 6.63 6.36 6.3-6.6 10.0 0.322-14 2.43 7.20 9.93 0.251-15 7.15 7.00/8.47/0.76/-42/3.7 6.85 7.00 6.7-7.0 n * Transitions 6.51 0.000-2.82 4.69 6.79 0.001-4.89 4.54/4.78/-/-/3.4 7.98 0.000-5.10 5.75 8.11 0.000-5.84 6.00/6.31/-/-/4.8 9.96 e 0.006-7.06 7.17 9.97 0.009-7.18 6.37/7.80/-/-/8.7 a Abs 1, Absorption in the gas phase (103); Abs 2, Absorption in aqueous medium (114); CD, CD spectra of uridine in an aqueous medium (116); Crystal, Transition energy/transition moment direction (151); Range, Range of transitions observed in different experiments; b E 1 represents CASPT2 and E 2 represents CASSCF transition energies, for the f values of n * transitions see original paper (160); c Ground state dipole moment at the HF/6-311G(d,p) level is 4.67 Debye; d Scaled (scaling factor 0.72) excitation energies; e Rydberg contamination. Uracil and Thymine: The spectral features of uracil and thymine are generally similar and characterized by absorption bands near 260, 205, and 180 nm (4.77, 6.05, and 6.89 ev). It should be noted that with respect to the uracil, the first and third bands in thymine are generally slightly red- and blue-shifted, respectively (11, 103, 104, 107, 110, 114, 146, 147). The 205 nm band is found to be mixture of two peaks near 215 and 195 nm (5.77 and 6.36 ev) in the CD spectra while the presence of a band near 240 nm (5.17 ev) was also indicated in the CD measurement, which was assigned as nπ* type (11, 116). The existence of an nπ* transition near 250 nm (4.96 ev) was predicted by Hug and Tinoco (148) and this transition was suggested as the possible source of the 240 nm band observed in the CD spectra (11, 116). Based on the polarized absorption (11, 105, 149) and reflection experiments (11, 150) the transition moment direction for the first band is found to be close to 0º for uracil and -20º for thymine (Fig. 1). Although, Novros and Clark (151) have suggested -53º or +59º for the second absorption band, but latter was selected on the basis of agreement with the LD spectra of uracil (109). However, Anex et al. (152) have suggested it to be -31º. Eaton and Lewis (149) have estimated that polarization of the I and II bands are approximately perpendicular to each other. Holmen et al. (153) have found 35º for the second transition in 1,3- dimethyluracil. Several investigations of uracil, thymine, and their analogs have suggested the existence of an nπ* transition within the 260 nm envelope (11, 107, 147, 154). In the photoacoustic spectra of thymine the existence of another transition within the 270 nm (4.59 ev) envelope was also suggested (107). Backer and coworkers (147, 155-157) have performed a series of experiments on the absorption and emission properties of uracil, thymine, and their derivatives in polar protic and aprotic solvents at the low and room temperatures. It has been found that uracil, thymine, and thymidine exhibit strong phosphorescence in polar aprotic solvents [2-methyltetrahydrofuran (2-MTHF)], while a relatively stronger fluo-