Experimental and Theoretical Study of Deprotonation of DNA Adenine Cation Radical^{†}
Ⅰ. INTRODUCTION
Deoxyribonucleic acid (DNA) is an important substance that carries the genetic instructions used in the growth, development, functioning and reproduction of all known living organisms. Photoexcitation or potent radical oxidants can induce the DNA damage events [15] that are associated with degenerative diseases, cancer, and aging [6, 7]. Among all the components (nucleobase, deoxyribose, and phosphate) of DNA, the base is the most vulnerable to oxidative damage. Generally speaking, the electronloss center in DNA base will generate a hole, resulting in the base cation radical [1, 2, 813]. Extensive studies have shown that the base cation radical is more acidic than its natural structure, and easily deprotonates to form the base neutral radical [1, 2, 4, 813]. The deprotonation will lead to a variety of DNA lesion products, affecting structures and functions of DNA molecules and potential applications in electrochemical devices [13, 14]. In this context, knowing the deprotonation mechanism of the base cation radical is critically important to understand relevant processes in DNA oxidative damage and DNAbased molecular electronics devices.
With the lowest oxidation potential among the four bases, Guanine (G) ($E^\textrm{o}\hspace{0.20cm}^$=1.29 V vs. NHE) is extremely redoxactive [1, 2, 8, 9, 15]. The oxidation to G$^{+ \cdot}$ and the subsequent deprotonation process of G$^{+ \cdot}$ have been the subjects of considerable experimental and theoretical studies [1, 2, 4, 810, 13, 14, 16, 17]. As shown in FIG. 1, G$^{+ \cdot}$ has two deprotonation sites, the imino proton (N$_1$H, p$K_\textrm{a}$=3.9) and the amino proton (N$_2$H, p$K_\textrm{a}$=4.7) [1, 2, 8, 9, 12, 13]. Pulse radiolysis and ESR experiments have shown that G$^{+ \cdot}$ preferentially deprotonates N$_1$H to generate G(N$_1$H)$^{\cdot}$ with the rate constant of 1.8$\times$10$^7$ s$^{1}$ [10, 16]. Kobayashi et al. measured deprotonation rate constants of (3.33.6) $\times$10$^6$ s$^{1}$ in different duplex DNA, and suggested that the whole deprotonation process can be interpreted by the following steps: proton transfer from the basepair cation radical G$^{+ \cdot}$:C first gives deprotonated G radical and the cytosine base pair (G(H)$^{\cdot}$:C(+H$^+$)) and then the proton of C(+H$^+$) is expected to be finally picked up by a water molecule [10]. Recently, our group used nanosecond laser flash photolysis to investigate the deprotonation of 1methylguanine (mG), of which N$_1$H is substituted by methyl group, and thus obtained the rate constant and activation energy of deprotonation from N$_2$H [12]. Further, our group also examined the unique deprotonation behaviors of Gquadruplexes, revealing the unusual deprotonation site of amino proton N$_2$H and much slower rate constants of (1.82.1) $\times$10$^5$ s$^{1}$ in Gquadruplexes that are different from duplex DNA [13].
Adenine (A) owns a low oxidation potential ($E^\textrm{o}\hspace{0.20cm}^$=1.42 V vs. NHE) comparable to G, thus can readily undergo the oxidation to A$^{+ \cdot}$ [1, 11, 18]. However, different from intensive reports for G$^{+ \cdot}$, studies on deprotonation of A$^{+ \cdot}$ are limited. According to the structure of A$^{+ \cdot}$, there is merely one active proton N$_6$H that can be released to generate A(N$_6$H)$^{\cdot}$ (FIG. 1) [1, 11]. Pulse radiolysis experiments and low temperature ESR experiments had verified the deprotonation site [11, 19]. The pulse radiolysis experiments also showed that the deprotonation rate constant measured in duplex DNA (2.0$\times$10$^7$ s$^{1}$) was similar to that in free A base, which seemed to suggest that the N$_6$H proton of A$^{+ \cdot}$ in duplex DNA is directly released to the solvent, as in the case of free A (Scheme 1, pathway (ⅰ)) [11]. An alternative deprotonation pathway (Scheme 1, pathway (ⅱ)), that is, the proton transfer from N$_6$H of A to the O atom of Thymine (T) within the A$^{+ \cdot}$:T base pair, was ruled out by Steenken because the p$K_\textrm{a}$ values of A$^{+ \cdot}$ (p$K_\textrm{a}$=4.0) [11, 18, 20] and T(+H$^+$) (p$K_\textrm{a}$=5) make this pathway infeasible [1, 14]. However, a gas phase B3LYP/631G$^{**}$ calculation of relative stabilities of A$^{+ \cdot}$:T (reactant) and A(N$_6$H)$^\cdot$:T(+H$^+$) (product) showed that the pathway (ⅱ) is endothermic by 5.0 kJ/mol and there is only a small barrier of 6.7 kJ/mol [21]. The gas phase calculation results suggested that pathway (ⅱ) could possibly occur, which is in contradiction with the experimental results [14, 21]. It follows that further studies are required to clarify the A$^{+ \cdot}$ deprotonation process.
Herein, we performed joint experimental and theoretical investigations on the deprotonation reaction of A$^{+ \cdot}$. The oneelectron oxidation of A to A$^{+ \cdot}$ and its subsequent deprotonation product, A(N$_6$H)$^ {\cdot}$, were detected in the timeresolved UVVis absorption spectra. By monitoring the formation kinetics of A(N$_6$H)$^ {\cdot}$, the deprotonation rate constants were measured at different temperatures varying from 280 K to 300 K. Based on Arrhenius equation, the activation energy barrier for the N$_6$H deprotonation is thus determined to be 17.1 kJ/mol. Theoretically, we calculated potential energy surface (PES) of the deprotonation process of A$^{+ \cdot}$ under solvation models including both explicit waters and the polarized continuum model (PCM). The energy barrier obtained from the PES (18.2 kJ/mol) is coincident with our experimental value, implying the rationality of the current theoretical treatments. Based on the theoretical treatments, we also performed calculations for the pathways (ⅰ) and (ⅱ) of A$^{+ \cdot}$:T in duplex DNA, and discussed the A$^{+ \cdot}$ deprotonation pathway in duplex DNA from dynamics considerations. These experimental and theoretical results can provide valuable dynamics and mechanistic insights for oxidative DNA damage and potential application in DNAbased electrochemical devices.
Ⅱ. MATERIALS AND METHODS
A. Materials
2$'$Deoxyadenine (A, Alfa Aesar), sodium persulfate (Na_{2}S_{2}O_{8}, SigmaAldrich), and sodium phosphate buffer (50 mmol/L, Beijing Solarbio Science & Technology) were used as purchased. Ultrapure water obtained by Millipore filtration was used as solvent. The samples of A and Na_{2}S_{2}O_{8} were dissolved in 3 mL of sodium phosphate buffer in H$_2$O. Note that each kinetic measurement was performed with new sample.
B. Laser flash photolysis
Nanosecond timeresolved UVVis transient absorption spectra were measured using a flash photolysis setup Edinburgh LP920 spectrometer (Edinburgh Instruments Ltd.) combined with a Nd:YAG laser (SpectraPhysics Lab 170, Newport Corp.). Each measurement was performed in a 1 cm path length quartz cuvette that was put in the Oxford Instruments OptistatDN Cryostat and cooled to a certain temperature. The sample was excited by using a 355 nm laser pulse (1 Hz, 10 mJ/pulse, fwhm$\approx$7 ns). The analyzing light was from a 450 W pulsed xenon lamp. A monochromator equipped with a photomultiplier for collecting the spectral range from 300 nm to 700 nm was used to analyze transient absorption spectra. The signals from the photomultiplier were displayed and recorded as a function of time on a 100 MHz (1.25 Gs/s sampling rate) oscilloscope (TDS 3012C, Tektronix), and the data were transferred to a PC. Data were analyzed with online software of the LP920 spectrophotometer and instrument response function of Gaussian type was considered. The fitting quality was judged by weighted residuals and reduced $\chi^2$ value.
C. Computational methods
The geometries of the reactants, products, and transition states were optimized using B3LYP in connection with the 631++G$^{**}$ basis set in PCM model, which have been shown to be a sufficient and affordable computational method for studying DNA base [16, 22]. The harmonic frequency analysis was performed to identify the stationary points as either local minima (reactants and products) or first order saddle points (transition states) and to extract zeropoint vibrational energy (ZPE) corrections. Connections of the transition states between two local minima were confirmed by intrinsic reaction coordinate (IRC) calculations at the same level. The charges from natural bond orbital (NBO) results and the spin densities, were evaluated at the same level to characterize the proton transfer of the PES. All the calculations were carried out using the Gaussian 09 program package [23].
Ⅲ. RESULTS AND DISCUSSION
A. Transient absorption spectra of A$^{+ \cdot}$ and its deprotonation product A(N$_\textbf{6}$H)$^{\cdot}$
Sulfate radicals, SO$_4$$^{ \cdot}$, were produced by the 355 nm laser photodissociation of peroxodisulfate anions (S$_2$O$_8$$^{2 \cdot}$) (Eq.(1)) [12, 13, 17]. The instantaneously generated SO$_4$$^{ \cdot}$ further oxidizes A to A$^{+ \cdot}$ and A$^{+ \cdot}$ subsequently deprotonates to neural radical A(N$_6$H)$^{\cdot}$ (Eq.(2) and Eq.(3)) [11].
$\begin{eqnarray} &&\textrm{S}_2 \textrm{O}{_8} ^{2  }\frac{h\nu}{}\hspace{0.3cm}\rightarrow 2\textrm{ SO}_4 {^{  \cdot } } \end{eqnarray}$

(1) 
$\begin{eqnarray} &&\textrm{SO}_4 {^{  \cdot }} + \textrm{A} \rightarrow \textrm{SO}_4 {^{2  } }+ \textrm{A}^{ + \cdot} \end{eqnarray}$

(2) 
$\begin{eqnarray} &&\textrm{A}^{ + \cdot }\rightarrow \textrm{A}(\textrm{N}_6\textrm{}\textrm{H})^ \cdot + \textrm{H}^ + \end{eqnarray}$

(3) 
The transient absorption of sulfate radical SO$_4$$^{ \cdot}$ is featured with the resolved band at 450 nm and a shoulder at 330 nm (FIG. 2(a)), which agrees with previous spectra [4, 10, 12, 13]. The rapid photodissociation of S_{2}O_{8}$^{2}$ results in the generation of SO$_4$$^{ \cdot}$ within the ~14 ns laser pulse duration. As a result, the formation dynamics of SO$_4$$^{ \cdot}$ does not interfere with the detection for the subsequent oxidation reaction of SO$_4$$^{ \cdot}$+A (Eq.(2)). The concentration of SO$_4$$^{ \cdot}$ can be estimated by the absorbance at 450 nm using an extinction coefficient of 1600 (mol/L)$^{1}$cm$^{1}$ [4, 10, 12, 13], which is ~25 μmol/L and much smaller than the concentration of A (6 mmol/L). The large excess concentration of A ensures the oxidation reaction of A by SO$_4$$^$ is a pseudofirst order reaction, and excludes the possibilities of sequential oxidation events of A.
For the oxidation of A+SO$_4$$^{ \cdot}$ at room temperature (FIG. 2(b)), one strong peak at 330 nm and a weak and flat absorption from 380 nm to 700 nm (absorption maxima around 500 nm), are exhibited in the early time (30 ns) transient spectrum, which are essentially similar to those reported before and can be identified as those of A$^{+ \cdot}$ [11]. It is noticeable that the 355 nm photolysis of A alone does not generate any signal.
As the reaction proceeds with time, the absorbance of the peak at 330 nm and the broad band from 550 nm to 700 nm increases, while the absorbance around 450 nm decays in the 300 ns spectrum (FIG. 2(b)). This transient spectrum is ascribed to the A(N$_6$H)$^{ \cdot}$ radical, which is featured with a strong peak at 330 nm and a weak absorption from 450 nm to 700 nm, with absorption maximum around 550 nm [11]. Furthermore, the inset of FIG. 2(b) also shows that the decay at 450 nm corresponds to the absorbance increase at 600 nm, meaning that the two absorbance changes should originate from the same reaction process, that is, the deprotonation of A$^{+ \cdot}$ to A(N$_6$H)$^{ \cdot}$.
B. The measurement of deprotonation rate constant of A$^{+ \cdot}$
The oneelectron oxidation to form A $^{+ \cdot}$ is accompanied by fast deprotonation to A(N $_6$ H) $^{ \cdot}$ . To discern the deprotonation of A $^{+ \cdot}$ , the A concentration dependence of the reaction kinetics was studied. As shown FIG. 2(b), the absorption spectrum of A(N $_6$ H) $^{ \cdot}$ exhibits a stronger absorption from 550 nm to 700 nm than that of A $^{+ \cdot}$ , the absorbance changes at 600 nm were thus measured. FIG. 3(a) shows the dependence of the apparent rate constants on the concentration of A. The plot of rate constant versus the concentration of A exhibits a linear relationship below 5 mmol/L, indicating that the ratedetermining step is the bimolecular reaction of A with SO $_4$ $^{ \cdot}$ . From the slope, the secondorder rate constant of the reaction with SO $_4$ $^{ \cdot}$ is obtained to be (4.0 $\pm$ 0.2) $\times$ 10 $^9$ (mol/L) $^{1}$ s $^{1}$ . The rate constant reached a plateau above 5 mmol/L, and thus the saturated rate constant, (2.0 $\pm$ 0.1) $\times$ 10 $^7$ s $^{1}$ , is ascribed to the firstorder deprotonation process of A $^{+ \cdot}$ which should be independent of the A concentration. The deprotonation rate constant measured here in laser flash photolysis agrees well with that measured by pulse radiolysis (2.0 $\times$ 10 $^7$ s $^{1}$ ) [11]. By performing experiments at high A concentration, reactions (Eq.(2) and Eq.(3)) can be kinetically resolved.
C. The measurement of the activation energy of A$^{+ \cdot}$ deprotonation
The measurement of the activation energy of A$^{+ \cdot}$ deprotonation} The better assessment of the deprotonation processes of A$^{+ \cdot}$ to A(N$_6$H)$^{ \cdot}$ requires the knowledge of the dynamics data, not only the rate constant but also the activation energy barrier. To obtain the activation energy of the A$^{+ \cdot}$ deprotonation, the deprotonation rate constant as a function of temperature in the range of 280300 K was measured (FIG. 3(b)). The concentration of A of 6 mmol/L was employed at each temperature to ensure that deprotonation is the ratedetermining step. As shown in FIG. 3(b), the deprotonation rate constants decrease with the decreasing temperature, and the rate constants exhibit an Arrhenius behavior.
$\begin{eqnarray} {\rm ln}k = \textrm{ln}A  \frac{E_\textrm{a}}{RT} \end{eqnarray}$

(4) 
The Arrhenius plot yields an $E_\textrm{a}$ value equal to (17.1$\pm$1.0) kJ/mol (Eq.(4), FIG. 3(b)), which corresponds to the activation energy barrier of the deprotonation processes of A$^{+ \cdot}$ to A(N$_6$H)$^{ \cdot}$. This small energy barrier, (17.1$\pm$1.0) kJ/mol, can also interpret the rapid deprotonation process of A$^{+ \cdot}$ to A(N$_6$H)$^{ \cdot}$.
In addition, we also tried to measure the activation energy barrier of the deprotonation of A$^{+ \cdot}$ in duplex DNA. However, the present experimental conditions cannot fulfill the requirement of high concentration of duplex DNA (above 5 mmol/L), and the activation energy of A$^{+ \cdot}$ in duplex DNA thus was not obtained.
D. Theoretical study of deprotonation process of A$^{+ \cdot}$ in free base
Here, the experimentally determined energy barrier for the A$^{+ \cdot}$ deprotonation (17.1$\pm$1.0 kJ/mol) is quite different from the value (6.7 kJ/mol) predicted by gas phase theoretical calculations [21]. The gas phase calculations obviously ignored the fact that the reaction takes place in aqueous solution. So the solvent effect has to be considered carefully to model the reaction process [22, 24, 25]. Therefore, we mimicked the solvation of A$^{+ \cdot}$ by using a combined explicit water and bulk water approach which has been successfully testified in the G$^{+ \cdot}$ system [22, 24]. The five explicit water molecules are used to reflect the specific shortrange interactions around the deprotonation site while the PCM is used to treat the remaining bulk water.
Under this solvation model, we first calculated the PES for the deprotonation process of A$^{+ \cdot}$ in free A base. The geometries of A$^{+ \cdot}$ (reactant), A(N$_6$H)$^ {\cdot}$(product) and the transition state (TS) [A$^{\cdot}$…H$^+$] for the deprotonation process of A$^{+ \cdot}$ in the presence of 5 explicit water molecules were fully optimized using density functional theory (DFT) B3LYP and the 631++G$^{**}$ basis set. Frequency analysis was also performed, at the same level of theory and basis set, to ensure the existence of reactant and product as local energy minimum, and for the TS, a negative frequency was found.
FIG. 4 displays the PES for the deprotonation of A$^{+ \cdot}$ to A(N$_6$H)$^ {\cdot}$ in free base. Initially, the reactant A$^{+ \cdot}$ and the five water molecules form a complex A$^{+ \cdot}$$\cdots$5H$_2$O. Accompanying the cleavage of N$_6$H bond, a transition state (TS) with an energy barrier of 18.2 kJ/mol was located. NBO results show that the proton H$^+$ is finally stabilized at the H$_2$O near N$_7$ site of A, leading to the formation of the deprotonation product A(N$_6$H)$^{\cdot}$ (FIG. 4). The PES clearly describes how the proton H$^+$ is released from A$^{+ \cdot}$ to the first hydration shell as represented by the explicit water, revealing the deprotonation mechanism of A$^{+ \cdot}$. As for the proton further migrating to the outer hydration shell, it has been proved to be very efficient [26, 27]. Therefore, the transport of H$^+$ to the first hydration shell is supposed to be the ratedetermining step for A$^{+ \cdot}$ deprotonation. The calculated reaction energy barrier of 18.2 kJ/mol thus can be approximately regarded as the activation energy for the overall deprotonation process of A$^{+ \cdot}$, which is in good agreement with the experimental value of 17.1 kJ/mol.
It should be noted that we also tested the effect of the number and the position of the explicit waters. First, increasing more waters far away from the A$^+$$^{\cdot}$ deprotonation site does not change the energy barrier, indicating the influence of the water molecules far from the deprotonation site is not significant for deprotonation process of A$^+$$^{\cdot}$. Second, with the use of less explicit waters, the energy barrier becomes much lower than the experimental value. For example, the energy barrier under the three explicit water model is only 10.9 kJ/mol. Hence, it can be deduced that the use of five waters is suitable for elaborating deprotonation mechanism of A$^+$$^{\cdot}$. The agreement with the experiment also demonstrates the rationality of the current theoretical treatments.
E. Theoretical study of deprotonation process of A$^{+ \cdot}$:T in duplex DNA
Based on the theoretical treatments of A$^{+ \cdot}$ deprotonation, we also performed DFT calculations for the two possible deprotonation pathways (Scheme 1) of A$^{+ \cdot}$:T in duplex DNA. The AT base pair is used here as a simplified duplex unit to save the computational costs. In this base pair, N$_1$ and N$_2$ of A separately form two hydrogen bonds with H$_3$ and O$_4$ of T, three explicit waters are thus used in the following calculations.
The PES for the deprotonation pathway (ⅰ) (Scheme 1) that involves the direct release of N$_6$H proton of A$^{+ \cdot}$ to the solvent is shown in FIG. 5. The reactant A:T$^+$$^\cdot$ and the three explicit water molecules form an expected complex A$^{+ \cdot}$:T$\cdots$3H$_2$O. Starting from the initial structure, the deprotonation of A$^{+ \cdot}$:T to A(N$_6$H)$^{\cdot}$:T requires a reaction energy barrier of 18.0 kJ/mol, which agrees with the activation energy for the deprotonation of A$^{+ \cdot}$ in free base (experimental value 17.1 kJ/mol, theoretical value 18.2 kJ/mol). Hence, by incorporating the solvent effect based on explicit water and PCM model, our calculations can well interpret the experimental results [11] that similar deprotonation rate constants of A$^{+ \cdot}$ were observed in the free A base and in duplex DNA, suggesting that the N$_6$H proton of A$^{+ \cdot}$ in duplex DNA should be directly released to the solvent, as in the case of free A.
For the deprotonation pathway (ⅱ) (Scheme 1) that occurs through the proton transfer within the AT base pair, the previous calculation in the gas phase suggested it could possibly occur [21]. However, the calculation ignored the solvent effect, which might not accurately describe the deprotonation of A$^+$$^\cdot$ in duplex DNA [14]. In this work, we attempted to locate the structures of the product A(N$_6$H)$^\cdot$:T(+H$^+$) and the transition state [A^{.}…H ^{+}…T] with inclusion of three or more explicit waters based on the optimized structures in Ref.[21]. However, it was found that the product and the transition state could not be located and always converged to the structure of reactant A$^+$$^\cdot$:T, no matter with or without PCM model. In this case, we further scanned the proton transfer coordinate N$_6$H bond length and obtained corresponding energy variation (FIG. 6). It is found that the energy increases with the increasing of N$_6$H bond length monotonically. These results indicate that the N$_6$H proton transfer along the hydrogen bond within AT base pair (pathway (ⅱ)) hardly occurs, supporting the postulation [1, 14] that the p$K_\textrm{a}$ values of A$^{+}$$^{\cdot}$ (p$K_\textrm{a}$=4.0) and T(+H$^+$) (p$K_\textrm{a}$=5) make the proton transfer within the AT base pair infeasible. Considering quite low energy barrier (18.0 kJ/mol) of the pathway (ⅰ), it can thus be deduced that the deprotonation process of A$^+$$^{\cdot}$:T in duplex DNA follows the pathway (ⅰ): the N$_6$H proton of A$^{+\cdot}$ in duplex DNA is directly released to the solvent, as in the case of free A (Scheme 1, pathway (ⅰ)).
Ⅳ. CONCLUSION
In this work, we have experimentally and theoretically investigated the deprotonation behavior of A$^{+ \cdot}$. The highly potent radical SO$_4$$^{ \cdot}$ was used to oxidize A, generating A$^{+ \cdot}$. The timeresolved UVVis absorption spectra showed the formation of deprotonation product of A(N$_6$H)$^ {\cdot}$. The rate constants of A+SO$_4$$^{ \cdot}$ oxidation and A$^{+ \cdot}$ deprotonation were measured to be 4.0$\times$10$^9$ (mol/L)$^{1}$ s$^{1}$ and 2.0$\times$10$^7$ s$^{1}$ at room temperature, respectively. Further, the formation kinetics of A(N$_6$H)$^ {\cdot}$ at various temperatures were measured, from which the activation energy barrier for A$^{+ \cdot}$ deprotonation was determined to be 17.1 kJ/mol. To gain mechanistic insights, we performed DFT calculations for describing the A$^{+ \cdot}$ deprotonation under the solvation models including both explicit waters and PCM. The calculated PES reveals an energy barrier (18.2 kJ/mol) that matches with our experimental value, suggesting the reliability of the current theoretical treatment for A$^{+ \cdot}$ deprotonation. Based on this strategy, we also calculated and assessed the A$^{+ \cdot}$ deprotonation pathways in duplex DNA. For the deprotonation pathway (ⅰ) of A$^{+ \cdot}$:T in duplex DNA (Scheme 1, pathway (ⅰ)), the calculated energy barrier (17.8 kJ/mol) agrees with both experimental and calculated values of A$^{+ \cdot}$ in free base, suggesting that similar deprotonation pathway of A$^{+ \cdot}$ occurs in duplex DNA to that in free base, undergoing direct release of the N$_6$H to aqueous solvent. Whereas, the deprotonation pathway (ⅱ) of A$^{+ \cdot}$:T involving the proton transfer along the hydrogen bond within AT base pair was found to be energetically infeasible, confronting an energy increase monotonically with the increase of the N$_6$H bond distance (Scheme 1, pathway (ⅱ)). These results have provided the kinetics energy barrier data and clarified the deprotonation behavior of A$^{+ \cdot}$ in both free A base and in AT base pair of duplex DNA, which should be of value for understanding the oxidative DNA damage and guiding related applications in adenine DNAbased electronics devices.
Ⅴ. Acknowledgments
This work was supported by the National Natural Science Foundation of China (No.21425313, No.21333012, No.21373233, and No.91441108) and the Chinese Academy of Sciences (No.XDB12020200).