1
Barrier-free proton transfer in anionic complex of thymine with glycine

2
We report the photoelectron spectrum of the thymine–glycine anionic complex (TG) recorded with low energy photons (2.540 eV).

3
The spectrum reveals a broad feature with a maximum between 1.6–1.9 eV.

4
The measured electron vertical detachment energy is too large to be attributed to a complex in which an anion of intact thymine is solvated by glycine, or vice versa.

5
The experimental data are paralleled by electronic structure calculations carried out at the density functional theory level with 6-31++G** basis sets and the B3LYP and MPW1K exchange–correlation functionals.

6
The critical structures are further examined at the second order Møller–Plesset level of theory.

7
The results of calculations indicate that the excess electron attachment to the complex induces an intermolecular barrier-free proton transfer from the carboxylic group of glycine to the O8 atom of thymine.

8
As a result, the four most stable structures of the thymine–glycine anionic complex can be characterized as a neutral radical of hydrogenated thymine solvated by an anion of deprotonated glycine.

9
The calculated vertical electron detachment energies for the four most stable anionic complexes lie in a range 1.6–1.9 eV, in excellent agreement with the maximum of the photoelectron peak.

Introduction

10
There are two major issues which we address in this study: (i) interaction of low energy electrons with the gas-phase thymine–glycine complex, and (ii) intermolecular proton transfer in this complex induced by an excess electron.

11
Low-energy electrons appear as secondary products of radiolysis of water, with the primary products being the OH and H radicals.1

12
Until now, the genotoxicity of radiation was primarily studied in the context of these radicals, and the connection between their presence and mutations of DNA is well documented.2,3

13
The results of recent experiments by Sanche and coworkers suggest that electrons with energies in a range 1–20 eV can induce DNA damage.4,5

14
The authors suggested that excess electrons trapped in temporary anionic states initiate chemical reactions leading to single- and double-strand breaks.

15
A few estimations of the activation barrier needed to rupture the sugar-phosphate bond have been made.6,7

16
However, the mechanism, leading from anionic states localized on nucleic acid bases to strand breaks, has been only briefly studied.

17
Electron trapping on nucleic acid bases has been an important topic in radiation biology for several decades.8–15

18
About ten years ago, it was realized that the high polarity of these molecules allows for the existence of dipole-bound anionic states as well.13

19
While our recent CCSD(T) results indicate that the valence anionic state of thymine (T) is vertically stable with respect to the neutral by 0.457 eV, our calculations also find the valence anionic state to be adiabatically unstable by 0.248 eV with respect to the dipole-bound anionic state and by 0.196 eV with respect to the neutral.16

20
The results of low-energy electron transmission spectroscopy experiments suggest that the valence anionic state is also unbound at the geometry of the neutral.17

21
The current view is that the valence anionic states of isolated nucleic acid bases are unbound, or at best, very weakly bound, although bound states dominate for solvated species.14,15

22
The intra- and intermolecular tautomerizations involving nucleic acid bases have long been suggested as critical steps in mutations of DNA.14,18

23
Intramolecular proton transfer reactions have been studied for both isolated and hydrated nucleic acid bases.8,19

24
Intermolecular single and double proton transfer reactions have been studied for the dimers of nucleic acid bases in both their ground and excited electronic states.20,21

25
Only small activation barriers were found for both the anionic and cationic GC pair, with the proton transfer reaction being favorable for the anion and slightly unfavorable for the cation.20

26
Recently we described a proton transfer process, which is induced by the attachment of an excess electron to the complex of uracil (U) with proton donors, such as glycine,22 alanine,23 formic acid,24 model inorganic acids,25,26 and alcohols.27

27
Denoting a proton donor as HA, the following process was identified:U⋯HA + e → UH˙⋯AOur ab initio calculations and photoelectron spectroscopy measurements (PES) strongly suggested that the electron attachment to complexes of uracil with some of these HA’s leads to a barrier-free proton transfer (BFPT) from the acid (HA) to the O8 atom of U, with the product dimer complex being a neutral radical of hydrogenated uracil (UH˙) and A.

28
The driving force for proton transfer is the stabilization of the excess electron onto the π* orbital of the base.

29
Thus, one may envision the process as one in which the excess electron first attaches itself to the nucleic acid base moiety of the dimer, forming an ultra-basic anion, which then extracts a proton from the proton donor portion of the dimer to yield the complex, UH˙⋯A.

30
Our previous studies were limited to complexes of uracil with various HA’s and we discussed the occurrence of BFPT as an outcome of the interplay between the deprotonation energy of HA, protonation energy of U, and the energy of intermolecular hydrogen bonds.22–27

31
Our results strongly suggested that alanine and phenylalanine act in a similar way as does glycine in its anionic dimeric complexes with uracil.22,23

32
Thus, the BFPT process was not very sensitive to the nature of the amino acid’s hydrophobic residual group.

33
Another intriguing question is whether other nucleic acid bases are also susceptible to BFPT.

34
We have recently found that BFPT also occurs in anionic complexes of thymine (T) with formic acid (F) though a difference in the photoelectron spectra of UF and TF suggested that the methyl group of thymine could make a difference in the intermolecular proton transfer to the O8 atom (see Fig. 1 for the numbering of atoms in T).24

35
In the present study we further explore the effect of methylation of the C5 atom of uracil.

36
We investigate anionic complexes of thymine–glycine (TG) and we compare the results with those for the UG complex.22

37
The photoelectron spectra of UG and TG as well as the results of ab initio calculations strongly suggest that both nucleic acid bases undergo BFPT in anionic complexes with glycine.

38
Moreover, a difference between U and T in the susceptibility to BFPT is much smaller in complexes with glycine than with formic acid.

39
An interesting point here is that attachment of low-energy electrons to complexes of RNA or DNA with proteins may also lead to mutations.

40
While UG and TG dimer complexes are only primitive models of such processes, our results demonstrate the possibility of electron-induced mutations in DNA-peptide complexes.

41
The formation of neutral radicals of hydrogenated pyrimidine nucleic acid bases could also play a role in damage to DNA and RNA by low energy electrons.

42
For instance, thymine hydrogenated at the O8 position cannot form a hydrogen bond with adenine, as dictated by the Watson–Crick pairing scheme.

43
We have also found that the radical, TH˙ has a significant electron affinity.16

44
Thus, the resulting anion, TH, might react with an adjacent deoxyribose molecule triggering strand-breaks in nucleic acids28.

Methods

Experimental

45
Negative ion photoelectron spectroscopy is conducted by crossing a mass-selected beam of negative ions with a fixed-frequency laser beam and energy-analyzing the resultant photodetached electrons.29

46
It is governed by the energy-conserving relationship: = EBE + EKE, where is the photon energy, EBE is the electron binding energy, and EKE is the electron kinetic energy.

47
One knows the photon energy of the experiment, one measures the electron kinetic energy spectrum, and then by difference, one obtains electron binding energies, which in effect are the transition energies from the anion to the various energetically-accessible states of its corresponding neutral.

48
Our apparatus has been described elsewhere.30

49
To prepare the species of interest, a mixture of thymine and glycine was placed in the stagnation chamber of a nozzle source and heated to ∼180 °C.

50
Argon gas at a pressure of 1–2 atm was used as the expansion gas.

51
The nozzle diameter was 25 μm.

52
Electrons were injected into the emerging jet expansion from a biased Th/Ir filament in the presence of an axial magnetic field.

53
The resulting anions were extracted and mass-selected with a magnetic sector mass spectrometer.

54
Electrons were then photodetached from the selected anions with ∼100 circulating Watts of 2.540 eV photons and then energy-analyzed with a hemispherical electron energy analyzer.

55
The temperature of mass-selected anions is not known, but based on the source conditions utilized, it likely lies between 100 K and 300 K.

Computational

56
As this computational effort is a continuation of our previous studies on the neutral31 and anionic22 complexes of uracil and glycine, we will use analogous notation for hydrogen-bonded structures and the same computational methodology.

57
The anionic structures characterized in the current study will be labeled as aTGn, indicating the parent neutral structure TGn the anionic structure is related to.

58
More precisely, an anionic structure aTGn is determined in the course of geometry optimization initialized from the optimal geometry for the neutral structure TGn.

59
The structures TGn for the neutral thymine–glycine complexes are analogous to the uracil–glycine UGn structures characterized in .ref. 31

60
The stability of the neutral (superscript = 0) or anionic (superscript = −) TG complexes is expressed in terms of Estab, Hstab, and Gstab.

61
Estab is defined as a difference in electronic energies of the monomers and the dimer:Estab = ETG(0,−)(GTG(0,−)) − ETG(0,−)(GT(0,−)) − EG(GG)with the electronic energy EX (X = T(0,−), G, TG(0,−)) computed for the coordinates determining the optimal geometry of X (i.e., the geometry where EX is at the minimum).

62
The values of Estab were not corrected for basis set superposition errors because our earlier results demonstrated that the values of this error in B3LYP/6-31++G** calculations for a similar neutral uracil–glycine complex did not exceed 0.06 eV.31

63
The stabilization enthalpy Hstab results from correcting Estab for zero-point vibration terms, thermal contributions to energy from vibrations, rotations, and translations, and the pV terms.

64
Finally, the stabilization Gibbs energy Gstab results from supplementing Hstab with the entropy term.

65
The values of Hstab and Gstab discussed below were obtained for T = 298 K and p = 1 atm in the harmonic oscillator-rigid rotor approximation.

66
The selection of T = 298 K was not dictated by the temperature of mass selected anions, because the latter is not known precisely.

67
It was chosen instead as a temperature which is reasonable for biological systems.

68
As our primary research method we applied density functional theory (DFT)32,33 with a Becke’s three parameter hybrid functional (B3LYP)34–36 and a modified Perdew-Wang 1-parameter-method for kinetics (MPW1K) designed by Truhlar et al.37

69
In both DFT approaches we used the same 6-31++G** basis set.38

70
Five d functions were used on heavy atoms.

71
The usefulness of the B3LYP/6-31++G** method to describe intra- and intermolecular hydrogen bonds has been demonstrated in recent studies through comparison with the second order Møller–Plesset (MP2) predictions.31,39

72
The ability of the B3LYP method to predict excess electron binding energies has recently been reviewed and the results were found to be satisfactory for valence-type molecular anions.40

73
We found that the value of electron vertical detachment energy (VDE) determined at the B3LYP/6-31++G** level for the valence π* anionic state of an isolated thymine is overestimated by 0.2 eV in comparison with the CCSD(T)/aug-cc-pVDZ result.

74
We will assume in the following that the same correction of 0.2 eV applies to the values of the VDE for all anionic TG complexes in which an excess electron occupies a π* orbital localized on thymine.

75
It is known that the B3LYP method underestimates barriers for proton transfer reactions,37 and thus lack of a barrier for the proton transfer reaction may be an artifact of the B3LYP method.

76
For this reason, we performed additional geometry optimizations using the MPW1K exchange–correlation functional, which was parameterized to reproduce barrier heights for chemical reactions.37

77
Finally, MP2 geometry optimizations have been performed for three anionic TG complexes to demonstrate consistency among different theoretical models and therefore to strengthen our conclusion.

78
The same 6-31++G** basis set was used in the B3LYP, MPW1K and MP2 calculations.

79
The presence of many low energy structures for neutral and anionic complexes prompted us to determine the populations of these structures in the gas phase equilibrium.

80
First, we selected a reference structure R for a given species (neutral or anionic).

81
Next, for every structure M other than the reference structure R we determined the equilibrium constant KM.KM = [M]/[R]from the difference in Gibbs free energies for M and R. The fraction of M in the equilibrated sample is given byxM = KM/(1 + K1 + K2 + …)where the sum in the denominator goes through all structures for a given species.

82
The fraction of R in the sample isxR = 1/(1 + K1 + K2 + …).

83
All calculations were carried out with the GAUSSIAN 9838 and NWChem41 codes on a cluster of 32 bit Xeon/SCI Dolphin processors, IBM SP/2, and SGI 2800, and Origin2000 numerical servers.

Results

PES spectra

84
The photoelectron spectra for TG and UG are quite similar, see Fig. 2.

85
Each spectrum exhibits a broad, structure-less feature.

86
For TG, its maximum occurs at EBE = 1.6–1.9 eV, while for UG, its maximum occurs at EBE = 1.7–2.0 eV.

87
The photoelectron spectrum of TG cannot be attributed to an intact T solvated by glycine.

88
As mentioned above, the valence π* and dipole-bound anionic states of thymine are characterized by a calculated value of the VDE of 0.457 and 0.055 eV, respectively (see Fig. 3 for the excess electron charge distributions in these systems).

89
Henceforth, only the valence π* anionic state will be considered further, since the experimental value of the VDE for TG is far too large for the dipole-bound anionic state of T solvated by G. However, the experimental value of the VDE is also too large for the valence π* anionic state of T solvated by G. The solvation energy by glycine would have to be larger by ∼1.4 eV, for the anion over its corresponding neutral, in order to be consistent with the maximum in the photoelectron spectrum.

90
This is rather improbable given that the VDE of U(H2O)1 is only 0.9 eV.9

91
Similarly, attributing the broad peak for TG to an anion of intact glycine solvated by thymine is inappropriate.

92
The reason is that the most stable conformer of canonical glycine (Fig. 1) does not bind an electron42 and the measured EA of glycine is ca. −1.8 eV.43

93
Our theoretical results indicate that glycine forms only weakly bound anions with the VDE values, determined at the CCSD(T) level, of 0.08 and 0.39 eV for the canonical (can) and zwitterionic (zwit) structure, respectively42 (see Fig. 3 for the excess electron charge distributions in these systems).

94
The electron binding energy shift induced by the interaction with thymine would have to be again approximately 1.4 eV to be consistent with the maximum of the photoelectron peak for TG, which is rather improbable.

95
Thymine in anionic complexes with glycine behaves very much like uracil in anionic complexes with glycine (see Fig. 2).

96
We expect that BFPT occurs in anionic complexes of glycine with thymine, in full analogy to the anionic uracil–glycine complexes.22

97
However, there is also a difference in the photoelectron spectrum of TG relative to that of UG.

98
When compared to the spectrum of UG, the photoelectron spectrum of TG appears to be shifted by about 0.1 eV towards smaller values of electron binding energy.

99
This can be partially attributed to the difference in the VDE for isolated anions of thymine and uracil, 0.46 and 0.51 eV, respectively.16

100
Interestingly, the difference between T and U is much more profound in anionic complexes with formic acid.24

101
Lastly, the widths of the main spectral features for TG and UG are much greater than of the photoelectron features for the valence anionic state of uracil solvated by either a single water molecule or a xenon atom.9

102
While this is mainly additional evidence that these complexes are not valence anions of T or U solvated by glycine, it may also indicate that several conformers of the anionic T(U)-glycine complexes coexist in the gas phase under our experimental conditions.

Computational results

Neutral thymine–glycine complexes

103
Thymine and uracil differ only by a methyl group at the C5 position, with the proton acceptor CO and proton donor N–H sites being the same.

104
When compared to uracil, the methyl group in thymine can create at most steric obstacles that have to be negotiated upon formation of hydrogen bonds.

105
In addition, a weak hydrogen bond with C5H acting as a weak proton donor, is operational only in complexes with uracil.

106
The topological space for the neutral TG complex is at least as complicated as for the neutral UG complex.

107
For the latter, we characterized twenty-three hydrogen-bonded structures formed by the lowest energy tautomers of U and G.31

108
In the current study only a limited subset of these twenty-three structures was explored.

109
We included the most stable structure of the neutral, which was labeled UG1 in ref. 31, plus all structures with the O8 atom involved in hydrogen bonding (labeled UGn, n = 2, 4, 14, 16, 18, and 20 in ref. 31), as we know from our earlier studies22–27 that these latter complexes become the most stable upon an excess electron attachment.

110
The neutral TGn complexes are displayed in Fig. 4 and their B3LYP/6-31++G** characteristics are given in Table 1.

111
The stability order follows the n index, which indicates that the stability order is the same for the UG and TG complexes, even though a weak hydrogen bond, with C5H acting as a weak proton donor, is not operational in the TG complexes.

112
The most stable complexes are TG1 and TG2 with the carbonyl (O9) and hydroxyl (O10H) groups of glycine interacting with the proton donor and acceptor centers of thymine.

113
The TG1 and TG2 structures have two strong hydrogen bonds and the values of Estab are −0.72 and −0.61 eV, respectively.

114
These stabilization energies are typical for dimers forming ring-like structures.31

115
The values of Gstab are negative for these structures indicating a thermodynamic preference to form the thymine–glycine dimer.

116
TG1 and TG2 are not connected by a simple reaction path as there is another ring-like structure, TG3, which separates them.

117
The glycine moiety in TG3 is oppositely oriented than in TG1 and TG2.

118
Thus, one can assume that a barrier along the path from TG1 to TG2 is approximately equal to the value of Estab for TG1, which is significant for a room temperature transformation.

119
The TG1 structure is important for gas phase studies of thymine as its percent share in equilibrium mixture at T = 298.15 K is 98% (see Table 1).

120
In DNA, however, the N1 atom is covalently bonded to the sugar-phosphate backbone and the N1 site is not available for hydrogen bonding.

121
The TG4 structure, with one O8⋯H–O10 hydrogen bond, and the carbonyl O9 atom interacting with the methyl group is stable by only −0.38 eV in terms of Estab and unstable in terms of Gstab, as are other TGn structures (n = 14, 16, 18, 20).

122
The diminished stability of TG14 results from the fact that the O10H hydroxyl group acts as a proton donor and acceptor.

123
The small stability of TG18 and TG20 results from a geometrical mismatch between the proton donor and acceptor sites of interacting monomers.31

124
Additionally, the presence of the CH3 group creates a steric obstacle that needs to be negotiated in the TGn (n = 4, 16, 20) structures.

125
In fact, we observed substantial geometrical reorganizations with respect to the corresponding UG structures upon geometry optimization.

126
The dipole moments of thymine, glycine, and the neutral TGn complexes provide insight into whether anionic dipole-bound states can contribute to the photoelectron spectrum of (TG) reported in Fig. 2.

127
For the most stable conformer of canonical glycine, the B3LYP/6-31++G** dipole moment of 1.23 D is smaller than the critical dipole moment of ca. 2.5 D required for excess electron binding,8 but a less stable canonical conformer is characterized by a larger dipole moment of 6.1 D. The calculated dipole moment of thymine is 4.7 D and the calculated dipole moments of the TGn complexes do not exceed 8.2 D. These dipole moments are too small to support a dipole-bound anionic state with a VDE of ca. 1.7 eV.44

128
However, the dipole-bound anionic states supported by TG2-TG20 might act as doorways to valence anionic states characterized in Section 3.2.2.

129
Finally, the electron binding energy at the maximum of the PES peak for TG is larger by one order of magnitude than the calculated value of the VDE for an excess electron solvated by glycine and uracil.45

130
Thus a structure with an excess electron solvated by thymine and glycine is not considered in this study.

Anionic thymine–glycine complexes

131
The results of B3LYP/6-31++G** calculations for anions of various hydrogen-bonded thymine–glycine complexes are summarized in Table 2, and representative structures are displayed in Fig. 5.

132
A common feature of anionic wavefunctions identified by us for the TG complexes is that the excess electron is localized on a π* orbital of thymine, in close resemblance to the valence anionic state of isolated thymine (see Figs. 3 and 5).

133
Occupation of the antibonding π orbital by an excess electron induces buckling of the ring of thymine because non-planar structures are characterized by a less severe antibonding interaction.11,22

134
The same kind of ring distortion takes place in all TG complexes upon excess electron attachment.

135
Our most important finding is that the most stable anionic TG structures are characterized by a BFPT from the carboxylic group of glycine to the O8 atom of thymine, see Table 2 and Fig. 5.

136
This behavior is analogous to that reported for the UG complexes22 and demonstrates that not only uracil but also thymine is susceptible to barrier-free intermolecular proton transfer.

137
This also explains why the PES spectra of UG and TG are so similar (see Fig. 2).

138
The driving force for proton transfer in the TG complexes is to stabilize the excess negative charge, which is primarily localized in the O8–C4–C5–C6 region.

139
In consequence of the extra stabilization of the excess electron provided by the transferred proton, the values of the VDE for the aTG2, aTG14, aTG4, aTG16 structures are larger by approximately 1.4 eV than for the valence anion of isolated thymine.

140
In fact, the calculated values of the VDE for these structures span a range of 2.1–1.8 eV.

141
After correcting downward by 0.2 eV, the resulting range of 1.9–1.6 eV coincides well with the broad peak in the photoelectron spectrum (see Fig. 2).

142
The products of the intermolecular proton transfer are the neutral radical TH, with the O8 atom hydrogenated, and the deprotonated glycine (see Fig. 5).

143
We found that deprotonation of glycine requires, in terms of electronic energy, 15.1 eV.

144
On the other hand, protonation of the valence anion of thymine provides, in terms of electronic energy, 14.7 eV.

145
Hence, the occurrence of a hypothetical reaction, which leads to noninteracting products,T + HOOC–CH2–NH2 → TH˙ + OOC–CH2–NH2requires 0.4 eV.

146
For the proton transfer to occur, the stabilizing interaction in the TH˙⋯OOC–CH2–NH2 system needs to: (i) compensate this barrier, and (ii) provide at least as much of the stabilization between the TH˙ and OOC–CH2–NH2 systems as the untransformed T and HOOC–CH2–NH2 moieties could provide.

147
Indeed, for the structures with BFPT, i.e., aTGn (n = 2, 4, 14, 16), Estab varies from −1.19 eV to −0.94 eV, whereas for the structures without BFPT, i.e., aTGn (n = 1, 18, 20), the values of Estab are smaller: −0.94 eV < Estab < −0.86 eV.

148
This confirms that occurrence of BFPT requires significant values of Estab and that these would compensate the endothermicity of reaction (6).

149
The most stable structure of the anionic complex results from an excess electron attaching to TG2 and its population amounts to 87% at standard conditions.

150
The neutral complex TG2 is less stable than TG1 by 0.1 eV, hence its population is only 2% at standard conditions (see Table 1).

151
Upon electron attachment to TG2, the carboxylic proton is transferred without a barrier to the O8 atom and the B3LYP value of the VDE for the optimal anionic structure is 1.83 eV (see Table 2 and Fig. 5).

152
The value of the VDE decreases to ca. 1.6 eV after correcting downward by 0.2 eV.

153
For comparison, the calculated value of the VDE for the aTG1 complex, which does not undergo intermolecular proton transfer, is only 1.00 eV (0.8 eV after the 0.2 correction).

154
For both structures the calculated values of adiabatic electron affinity are smaller by 0.6–1.2 eV than the VDE values.

155
The neutral structures TG4, TG14, and TG16, which in terms of Estab are less stable than TG1 by 0.4, 0.3, and 0.3 eV, respectively, are strongly stabilized upon an excess electron attachment as they undergo a significant geometrical relaxation that includes barrier-free proton transfer.

156
The resulting aTG14, aTG4, and aTG16 structures are more stable than the aTG1 structure and their B3LYP values of the VDE are 2.1 (1.9), 2.0 (1.8) and 2.0 (1.8) eV, respectively (see Table 2 and Fig. 5; the corrected values are given in parentheses).

157
These structures are less stable than aTG2 by only 0.2 eV.

158
Their adiabatic electron affinities, calculated with respect to the parent neutral structures, were found to be 0.5–0.7 eV.

159
What topologically discriminates TG1 from other structures considered here is the fact that the O7 rather than O8 atom is involved in hydrogen bonding with glycine.

160
The lack of BFPT for aTG1 may be related to the fact that an excess electron on the π* orbital is not localized in the neighborhood of the O7 atom.

161
This finding is important for gas phase studies of thymine, though in DNA the aTG1 structure is not operational because the N1 atom is covalently bonded to the sugar-phosphate backbone. aTG1 and aTG2 are not connected by a simple reaction path as there is another ring-like structure, aTG3, which separates them.

162
The glycine moiety in aTG3 is oppositely oriented than in aTG1 and aTG2.

163
Thus, one can assume that a barrier along the path from aTG1 to aTG2 is approximately equal to the value of Estab for aTG1, which is significant for a room temperature transformation.

164
Finally, we should point out that not every hydrogen bond O10H⋯O8 undergoes a BFPT upon attachment of an excess electron.

165
The two structures TG18 and TG20 undergo just a serious structural reorganization upon an excess electron attachment, but not BFPT.

166
The VDE values for aTG18 and aTG20 of 1.2 and 1.4 eV, respectively (after correcting downward by 0.2 eV), are remarkable for structures without a proton transferred to the ring of thymine and indicate an extra stabilization of the excess electron in T by 0.8–1.0 eV upon solvation by canonical glycine.

167
A smaller value of the VDE for the aTG1 structure of 0.8 eV illustrates that the solvation at the O8 site provides more stabilization to the excess electron than solvation at the O7 site.

168
Moreover, there is no local minimum on the potential energy surface of either aTG1, aTG18, or aTG20 with the O10H proton transferred to an oxygen atom of thymine.

169
The occurrence of BFPT (Yes/No in Table 2) proved to be consistent for the B3LYP and MPW1K methods.

170
As a final verification, MP2/6-31++G** geometry optimizations were performed for the aTGn (n = 1, 2, 4) complexes starting from the geometry of the corresponding neutral complex.

171
Again, BFPT occurred for n = 2 and 4 but not for n = 1.

172
We conclude that the occurrence of BFPT is consistently predicted within various theoretical models, which strengthens our findings and interpretations.

173
There are at least four anionic structures aTGn (n = 4, 14, 16, 20), which differ in terms of Gstab by less than 0.16 eV from the most stable structure aTG2.

174
The four structures susceptible to BFPT (n = 2, 4, 14, and 16) are characterized by large values of the VDE in a 1.6–1.9 eV range (after correcting downward by 0.2 eV).

175
Of particular importance might be the aTG2 and aTG14 structure as their populations amount at standard conditions to 87 and 12%, respectively, and their values of the VDE differ by 0.2 eV.

176
All these structures might contribute to the unusual width of the main feature in the photoelectron spectrum presented in Fig. 2.

177
Finally we comment on the differences in photoelectron spectra for TG and UG, see Fig. 2.

178
A difference by about 0.1 eV in the maximum of the main feature is also confirmed by the B3LYP/6-31++G** results (Table 2 for TG and Table 1 in ref. 22 for UG).

179
The calculated values of the VDE for the most stable anionic structure are 1.83 and 1.93 eV for TG and UG, respectively.

180
The width of the main spectral feature might be slightly larger for TG than for UG and might be affected by the distribution of different structures and their values of the VDE.

181
In addition there might be a contribution to the width of the PES feature of TG from the methyl group on thymine, which undergoes hindered rotations.

Summary

182
Our main findings are:

183
1.

184
The photoelectron spectrum of the thymine–glycine anionic complex, recorded with 2.540 eV photons, reveals a broad feature with its maximum between EBE = 1.6–1.9 eV.

185
The vertical electron detachment energy values are too large to be attributed to the complex of an anion of intact thymine solvated by glycine, or an anion of intact glycine solvated by thymine.

186
2.

187
The theoretical results, obtained at the density functional theory and second order Møller–Plesset levels, indicate that excess electron can induce a barrier-free proton transfer from the carboxylic group of glycine to the O8 atom of thymine.

188
The driving force for the proton transfer is to stabilize the negative excess charge localized primarily on the O8–C4–C5–C6 fragment of thymine.

189
3.

190
The anionic complexes with the O8 site protonated are the most stable.

191
These complexes can be viewed as the neutral radical of hydrogenated thymine solvated by the anion of deprotonated glycine and are characterized by the largest values of the VDE, which span a range of 1.9–1.6 eV.

192
These values of the VDE were obtained by shifting the B3LYP values down by 0.2 eV, as suggested by the CCSD(T) results for the valence anionic state of isolated thymine.

193
4.

194
The calculated values of the VDE for the most stable structures of TG are falling within the range observed in the PES spectrum.

195
The measured PES spectra and theoretical results are similar for the UG and TG species suggesting that thymine behaves very much like uracil in anionic complexes with glycine.

196
When compared to the spectrum of UG, the photoelectron spectrum of TG appears to be shifted by about 0.1 eV towards smaller values of electron binding energy.

197
This can be partially attributed to the difference in the VDE for isolated anions of thymine and uracil, 0.46 and 0.51 eV, respectively.

198
5.

199
The difference between T and U in anionic complexes with glycine is much less profound than in anionic complexes with formic acid.

200
To elucidate these differences we are currently exploring anionic complexes with acid molecules bound to the C5 and C6 atoms of uracil or thymine.

201
6.

202
The formation of neutral radicals of hydrogenated pyrimidine nucleic acid bases upon interaction with low energy electrons may be relevant to the damage of DNA and RNA.

203
The mechanism of strand breaks triggered by these radicals is currently being investigated in our group.