Theoretical studies are very helpful to understand aspects related to the chemical and biochemical processes (Warshel, 1991; Müeller et al., 1992; Cramer and Truhlar, 1994; Bała et al., 1996; Náray-Szabó and Warshel, 1997; Tapia and Bertrán, 2002; Ferrario et al., 2006; Kotz et al., 2009). So, the processes that involve the DNA and RNA nucleic acids are frequently studies at molecular level and they are object of many investigations using computational methods (Watson and Crick, 1953).

The interconversion of nitrogenous bases in DNA and RNA in genetic mutations will depend on the medium in which the reaction takes place, the stacking in DNA, and RNA acid strands (Jacquemin et al., 2014; Cerón-Carrasco and Jacquemin, 2015) or the explicit water molecules assisting the mechanism of the process (Cerón-Carrasco et al., 2009a,b,c; Brovarets and Hovorun, 2010; Tolosa et al., 2018a). Also, nucleic acids undergo alterations of their structures from attacks by various agents, like ionizing radiation (Cerón-Carrasco et al., 2010).

Solvation methods (Cramer and Truhlar, 1999; Tomasi et al., 2005; Marenich et al., 2009) are usually employed to consider solvent effects as a continuous medium. Classical molecular dynamic (MD) simulations (Alder and Wainwright, 1959) are a useful method to study the processes at the molecular level, but now the solvent is considered as a discrete medium. The steered molecular dynamic (SMD) technique (Izrailev et al., 1999; Isralewitz et al., 2001) allows us to follow the changes in Gibbs free energy over the course of any elementary process.

The nitrogenous bases: adenine, guanine, cytosine, thymine, and uracil can be associate to form adenine–thymine, guanine–cytosine, and adenine–uracil canonical pairs, although other rare bases may participate in genetic mutation processes (Cerón-Carrasco and Jacquemin, 2015). Adenine and guanine are associated to form the adenine–thymine and guanine-cytosine base pairs. The most stable structures for the adenine base are the amine and imine forms, whereas in guanine there are the keto and enol tautomeric forms (Figure 1). Both bases have been studied in gas and solution phases. The interaction of adenine base with water and oxygen donor molecules has been the subject of studies, because this base can be converted into guanine base and can participate in spontaneous mutation which will be the subject of this work.

As shown in Figure 2 this A→G transition occurs through three processes: first, hydrolytic deamination of adenine base; second, oxidation of the hipoxanthine intermediate; and finally, the animation of the xanthine intermediate to guanine. Specifically, one water molecule attacks the carbon linked to amine group of adenine base to form hypoxanthine and ammonia molecules. Then hipoxanthine oxidation occurs assisted by an oxygen donor molecule (formic acid in this work) to give a xanthine molecule. Finally, the NH₃ molecule released in the first process attacks by the carbonyl group new to give the guanine base.

The hydrolytic deamination reaction mechanism of adenine has been theoretically investigated with the DFT method by Zhang et al. (2007), considering the solvent by the PCM model. They found that the deamination may proceed in a stepwise mechanism. A tetrahedral intermediate is formed first by the hydrolysis reaction, and this is followed by the deamination reaction where the C-N link is broken and C = O group is formed. The study shows that the deamination with one water molecule does not take place because of a higher barrier. However, it may be possible if several water molecules are considered, because one molecule can act as a catalyst and the others act as assistant molecules. This is agreement with the experimental results.

In other similar work, the adenine deamination was studied by Zhu and Meng (2009) by four different mechanisms using the DFT method. The most favorable pathway describes a tetrahedral intermediate formed through a hydrolysis reaction on the C-N bond, then the deamination reaction takes place breaking the C-N bond and forming a C = O bond. The activation barriers are high, and it is difficult for this process to happen with such high activation barriers.

Later, Zheng and Meng (2009) studied the hydrolytic deamination mechanism of adenine using several water-assistant molecules and the density functional method at the DFT method. When the first water molecule attack to adenine base a tetrahedral intermediate is formed. Then, two different intermediates were studied. In A-pathway, the second water molecule acts as a bridge to transfer the hydrogen atom. In B-pathway, the second water molecule is not involved in the formation of the transition structure and only acts as a aqueous medium. Energy barriers of the two processes are 23.40 and 37.17 kcal·mol⁻¹, respectively.

Deamination of protonated adenine to produce hypoxanthine has also been studied by Wang and Meng (2010) in acid medium and assisted with a water molecule, using the DFT method. Because adenine could be protonated on different nitrogen atoms, four pathways were investigated. The pathway where the nitrogen closer to the NH₂ amine group is protonated presents the lowest energy barrier of 28.9 kcal·mol⁻¹.The first step and rate-determining is the nucleophilic attack of water molecule to form a tetrahedral structure. The results show that adenine deamination under acidic medium is more favorable than in neutral conditions.

Alrawashdeh et al. (2013) studied several mechanisms for the deamination of adenine with up to four H₂O molecules in basic medium and at different levels of theory. The effect of aqueous solvent was computed using the PCM model. Deamination with a water molecule presents a high activation barrier (190 kJ mol⁻¹). The addition of 1 s water molecule reduces this barrier by 68 kJ mol⁻¹. The most favorable mechanism with three water molecules presents an activation energy of 139 (in gas phase) and 137 kJ mol⁻¹ (in solution phases).

The oxidation mechanism of hypoxanthine has been investigated by Tafasse (2015). This theoretical study aimed to predict the transition state structure in the mechanism of oxidation with xanthine oxidase (XO) enzyme. The DFT/B3LYP method was used in the geometry optimization and it revealed that oxidation of hypoxanthine by xanthine oxidase follows a stepwise mechanism: (a) abstraction of a proton by Glu1226 from the hydroxyl hydrogen of XO followed by nucleophilic attack on the carbon of the substrate, and (b) then a proton transfer from substrate to enzyme to produce the hypoxanthine linked to the enzyme complex. The process was exothermic and with a high energy barrier.

The catalytic mechanism of guanine amination was investigated by a combination of QM calculations using the ONIOM method and MD simulations (Yao et al., 2007). The process was assisted by a Zn-metalloenzyme and two residues (Glutamate 55 and Aspartate 114). The proposed mechanism is initiated by a proton transfer from a Zn-bound water molecule to protonate Asp114. The protonated Asp114 can perform the proton transfer to the guanine, facilitating the nucleophilic attack on the nitrogenous base. The residue Glu55 then perform the proton transfer from the Zn-hydroxide to the amino group of the intermediate. The C-NH₃ bond breaks forming ammonia which leaves the active site and xanthine is freed with a barrier about 8 kcal·mol⁻¹.

The deamination of guanine with H₂O and OH⁻ has been realized with ab initio calculations by Uddin et al. (2011). Optimized geometries of all species were determined at different levels of theory and the Gibbs free energies were also determined. Deamination with OH⁻ radical was found to have an activation barrier of 155 kJ mol⁻¹ compared to 187 kJ mol⁻¹ for the reaction with H₂O. The lowest overall activation energy, 144 kJ mol⁻¹, was obtained using the H₂O/OH⁻ pair.

No theoretical studies have been realized for this A→G transition, although the separate processes of deamination, oxidation, and animation have been performed with model systems above mentioned. Here, we present the first study that uses the SMD technique to calculate molecular properties of this A→G process.

The formalism used in this work has been the same as the one used in previous studies by our research group, so for more detailed information it can be consulted in the work of cytosine to thymine conversion (Tolosa et al., 2019). Here we will only mention the methods used and some specific considerations of the simulated system.

The LJ (12-6-1) potential was used to describe the solute-solvent interaction, where van der Waals parameters are taken from the AMBER force field (Cornell et al., 1993) and the charge on each solute are the initially obtained by Mulliken method (Mulliken, 1955) and, subsequently, recalculated with the RESP method (Damm et al., 1997). The solvent charges were the TIP3P charges (Jorgensen and Tirado-Rives, 1988; Kaminski et al., 1994; Jorgensen et al., 1996).

The initial geometry of the Adenine-Water-Formic complex (denoted hereafter as A-W-F system), with the formic acid and water reactants about 2.0 Å from adenine, was obtained with the AMBER12 software (Case et al., 2012) at 298.15 K.

The reaction coordinates used to carry out the SMD simulations can be expressed as:

Long-range electrostatic interactions were considered by the Ewald summation (Ewald, 1921) and the Jarzynski's equality (Jarzynski, 1997) was used to calculate Gibbs energy differences between two equilibrium states. The activation and reaction energies were determined through the evolution of the process, as has been shown in several works (Tolosa et al., 2014, 2016, 2017a,b, 2018a,b; Tolosa et al., 2019).

Simulations were performed with the QM/MM method and the semi-empirical Hamiltonian AM1 method (Dewar et al., 1985). The system was partitioned, applying quantum calculations to the A-W-F (QM subsystem) an a classical way to the H₂O solvent (MM subsystem). HCOOH and H₂O molecules that assisted proton transfer were included in the QM part.

In the A→G transition none of the simulated processes were both thermodynamically and kinetically favorable, except for the deamination which was slightly exergonic. The result, considering all processes, leads us to conclude that the A→G transition is not spontaneous, although somewhat more favorable in vacuum. Also, the lifetimes of guanine show that this base could participate in genetic mutation processes.

The reaction profiles are similar for both phases, although the final result shows lower energies when the transition is simulated in gas phase, justified by the amination process. The effect of the medium can influence the mechanism that is followed depending on whether there are molecules surrounding the system that hinder or favor the transfer, as happens for the deamination (more exergonic in solution) or oxidation and amination (more endergonic in solution) processes.

Transition states that describe nucleophilic attacks present high barriers. The highest the barrier to overcome is the TS4 structure that describes the break between the base and the formic acid and the formation of a formaldehyde molecule. The protonation of N7 (2A-step) or N10 (2B-step) atoms are the only exergonic steps in all simulations. On the other hand, the nucleophilic attack of the water molecules on the base (1A-step) is the most endergonic step of the conversion.

We also want to highlight that SMD simulations allows us to investigate reaction mechanisms of complex systems in solution and follow the evolution of processes at molecular level, which provides information on every step about structures, and thermodynamics and kinetics properties through energy profiles.

All authors listed have made a substantial, direct and intellectual contribution to the work, and approved it for publication.