**5. Oxidation of serine phosphate: Implications for DNA**

As mentioned in **section 2**, the exposure of DNA to high–energy radiation can also cause damage at its phosphate backbone. In particular, it can cause strand breaks via cleavage of the phosphoester bonds.(Lipfert et al. 2004) When the absorption of radiation causes the ionization of the phosphate it has been shown that it then abstracts a H• from the deoxyribose ring at either its C4' or C5' position. This is then followed by heterolytic cleavage of the phosphoester bond.(Lipfert et al. 2004; Steenken & Goldbergerova 1998) Alternatively, the strand break may be preceded by chemical modification of the nucleobase or deoxyribose ring or the phosphoester bond may simply undergo a direct cleavage.(Lipfert et al. 2004)

Experimentally it can be difficult to clearly observe and characterize damage within DNA due to its size. Thus, it is common to either use short fragments or model compounds that can reproduce or mimic the damage and associated processes that may occur. For example, serine phosphate contains a phosphate bond as well as a carboxylate that can act as an electron scavenger much like the bases within DNA itself. Thus, it is often used in experimental studies on the processes of DNA damage at its phosphate and subsequent bond cleavage reactions and in fact has led to a deeper understanding of those processes involved.

Fig. 7. The C2-centered radical **I** and C3-centered radicals **II** and **III** proposed to be formed upon the irradiation of non– and partially-deuterated single crystals of L–*O*–serine phosphate. (Sanderud & Sagstuen 1996)

In particular, evidence has been obtained suggesting the formation of several different radical species. Specifically, the irradiation of non– and partially-deuterated single crystals of L–*O*–serine phosphate has been suggested to produce the three radicals **I, II** and **III** shown in **Figure 7**.(Sanderud & Sagstuen 1996) In particular, it is thought that upon irradiation serine phosphate can take up a now free electron to form a C1-centered radical anion. This may then undergo deamination to form the C2-centered radical anion **I**. Alternatively, a serine phosphate may lose an electron to form a neutral C1-centered radical which then undergoes decarboxylation to give a C2-centered radical. This latter radical may then abstract a H• from the C3 position of another serine phosphate to generate the C3 centered radical anion **II**. In contrast, the neutral C3-centered radical **III** is proposed to be

Mechanisms of Mutagenic DNA Nucleobase Damages and

is common when the carbon is bonded to an oxygen.

calculated and experimentally measured HFCCs.(Eriksson et al. 1994)

using a redox mechanism catalysed by the AlkB family of enzymes.

HFCC studies on radicals **II** and **III**.

Their Chemical and Enzymatic Repairs Investigated by Quantum Chemical Methods 405

In contrast to radical **I**, the best agreement between computational and experimental HFCCs for radicals **II** and **III** was not obtained for their proposed structures shown in **Figure 7**. Rather, the structures that gave best agreement were found to be the radical-cationic forms of **II** and **III**; that is, structures in which their phosphate and carboxylic acid groups are neutral while the amino group is protonated. Hence, these latter structures were used in all

For radical-cationic **II** the calculated HFCCs for the α–hydrogen are generally in good agreement with the experimental values with Aiso differing by just 2.3 MHz (**Table 8**). However, for both β–hydrogens, less satisfactory agreement between theory and experiment is observed. Specifically, the calculated Aiso values of 15.4 and 12.7 MHz differ from their corresponding experimental values by 9.6 and 3.2 MHz, respectively. As noted for the α– hydrogen in **I**, these differences are likely due to bulk crystal effects not being modeled in the calculations. Similarly, for the γ–hydrogens on the amino group, a difference of 4.0 MHz is seen between the calculated and experimental values. However, it has been stated that the given experimental value is associated with a degree of uncertainty.(Sanderud & Sagstuen 1996) In the optimized structure of **II** the C3 centre is not completely planar where an angle between the hydrogen on C3 and the plane defined by C3–C2–O was found to be 22.0º.(Lipfert et al. 2004) This is in good agreement with the corresponding experimentally determined value of 19.5º.(Sanderud & Sagstuen 1996) It is noted that the C3 centre and adjacent oxygen have spin densities of 0.84 and 0.11 respectively, suggesting slight delocalization of the SOMO. Dobbs et al.(Dobbs et al. 1971; Dobbs et al. 1972) have suggested that this is due to incomplete rehybridization of the carbon–centered radical and

Indeed, in radical-cationic **III** in which the phosphate group on C3 has been lost, i.e., there is no C3—O bond, the radical is now fully localized on C3. Furthermore, from **Table 8** it can be seen that quite good agreement is obtained for α-H1 and the β-hydrogen with differences of 1.7 and 0.0 MHz respectively, while α-H2 differs slightly more by 8.5 MHz. It should be noted that in order to obtain the best agreement for the HFCCs of the α–hydrogens they had to be tilted out of the plane by 5º. That is, the planarity of the C3 centre had to be reduced. This is similar to that observed need in related studies on the methyl radical (•CH3) to slightly pyramidalize the carbon centre in order to obtain best agreement between the

**6. Oxidative repair of alkylated nucleobases: the catalytic mechanism of AlkB**  A common type of DNA damage is the alkylation of nucleobases at one or more of their oxygen and nitrogen centres.(Liu et al. 2009) This can be caused by endo– or exogenous agents such as S-adenosylmethionine or tobacco smoke, respectively.(Hecht 1999; Rydberg & Lindahl 1982) Consequently, cells have developed several methods by which to mediate or repair such damage. One approach is to use enzymes known as DNA glycolsylases to simply excise the damaged nucleobase via an acid-base cleavage of its N-glycosidic bond.(Mishina et al. 2006; Sedgwick et al. 2007) Alternatively, repair enzymes may use a non-redox mechanism to remove only the alkyl group. For example, O6–methylguanine– DNA methyltransferase transfers the methyl of O6–methylguanine onto an active site cysteinyl residue.(Lindahl et al. 1988) However, a third approach is alkyl group removal

formed via the uptake of a free electron by serine phosphate to give a P-centered radical anion which then simply undergoes loss of the phosphate group.

We investigated the nature of radicals **I**, **II** and **III** by obtaining optimized structures at the B3LYP/6–31G(d) level of theory. Hyperfine coupling constants (HFCCs) were then calculated via single point calculations at the B3LYP/6–311+G(2df,p) level of theory on the above structures and are listed for all three radicals in **Table 8**.(Lipfert et al. 2004)

In the optimized structure of radical **I** the largest spin density is located on C2 (0.71) while the carbonylic oxygen of the carboxylate group also has significant spin density (0.18). This delocalization of spin density is due to conjugation between the singly occupied molecular orbital (SOMO) on C2 and the C1=O *π*–bond. In addition, it is noted that near planarity was observed for C1 and C2 and their substituents, in agreement with experiment.(Sanderud & Sagstuen 1996) With respect to its calculated HFCCs, as can be seen in **Table 8** the values calculated for the α–hydrogen are in good agreement with experiment. The slight deviation in Aiso from experiment is possibly due to the fact that the calculations are performed in the gas-phase on an isolated molecule and hence effects due to crystal packing are ignored. The calculated HFCCs for the two β–hydrogens are also in good agreement with the experiment. However, a rotational scan was performed in order to obtain the dihedral angle between the β–hydrogens and SOMO that gave HFCCs in best agreement between theory and experiment. It was found that this occurred for dihedral angles of –1.7º and 117.2º.


Table 8. Comparison of calculated (see text) and experimental (Sanderud & Sagstuen 1996) isotropic and anisotropic components (MHz) of the hyperfine coupling tensor of radicals **I**, **II**, and **III**.

formed via the uptake of a free electron by serine phosphate to give a P-centered radical

We investigated the nature of radicals **I**, **II** and **III** by obtaining optimized structures at the B3LYP/6–31G(d) level of theory. Hyperfine coupling constants (HFCCs) were then calculated via single point calculations at the B3LYP/6–311+G(2df,p) level of theory on the

In the optimized structure of radical **I** the largest spin density is located on C2 (0.71) while the carbonylic oxygen of the carboxylate group also has significant spin density (0.18). This delocalization of spin density is due to conjugation between the singly occupied molecular orbital (SOMO) on C2 and the C1=O *π*–bond. In addition, it is noted that near planarity was observed for C1 and C2 and their substituents, in agreement with experiment.(Sanderud & Sagstuen 1996) With respect to its calculated HFCCs, as can be seen in **Table 8** the values calculated for the α–hydrogen are in good agreement with experiment. The slight deviation in Aiso from experiment is possibly due to the fact that the calculations are performed in the gas-phase on an isolated molecule and hence effects due to crystal packing are ignored. The calculated HFCCs for the two β–hydrogens are also in good agreement with the experiment. However, a rotational scan was performed in order to obtain the dihedral angle between the β–hydrogens and SOMO that gave HFCCs in best agreement between theory and

above structures and are listed for all three radicals in **Table 8**.(Lipfert et al. 2004)

experiment. It was found that this occurred for dihedral angles of –1.7º and 117.2º.

**Radical Atom Aiso Txx Tyy Tzz I** α–H(calcd) –43.1 28.2 –1.2 –29.5

**II** α–H(calcd) –53.3 –36.1 –1.4 37.5

**III** α–H1(calcd) –48.8 –37.3 –1.7 39.0

Table 8. Comparison of calculated (see text) and experimental (Sanderud & Sagstuen 1996) isotropic and anisotropic components (MHz) of the hyperfine coupling tensor of radicals **I**,

**II**, and **III**.

α–H(exp) –54.2 32.3 3.3 –29.0 β–H1(calcd) 105.6 8.7 –2.4 –6.3 β–H1(exp) 104.3 7.2 –2.0 –5.3 β–H2(calcd) 57.8 8.3 –3.6 –4.7 β–H2(exp) 57.4 8.1 –3.9 –4.2

α–H(exp) –51.0 –34.5 3.8 30.7 β–H(calcd) 15.4 10.8 –4.0 –6.8 β–H(exp) 25.0 14.0 –7.0 –8.0 β–N(calcd) 12.7 2.1 –0.9 –1.2 β–N(exp) 15.9 2.1 –0.3 –1.8 γ–H(calcd) 1.9 4.3 –1.5 –2.8 γ–H(exp) 5.9 10.1 –4.1 –5.8

α–H1(exp) –47.1 –22.0 –1.5 23.6 α–H2(calcd) –59.3 –37.4 –1.5 38.8 α–H2(exp) –50.8 –24.6 0.5 24.2 β–H(calcd) 73.0 8.9 –2.2 –6.7 β–H(exp) 73.0 (13.3) (–6.6) (–6.6)

anion which then simply undergoes loss of the phosphate group.

In contrast to radical **I**, the best agreement between computational and experimental HFCCs for radicals **II** and **III** was not obtained for their proposed structures shown in **Figure 7**. Rather, the structures that gave best agreement were found to be the radical-cationic forms of **II** and **III**; that is, structures in which their phosphate and carboxylic acid groups are neutral while the amino group is protonated. Hence, these latter structures were used in all HFCC studies on radicals **II** and **III**.

For radical-cationic **II** the calculated HFCCs for the α–hydrogen are generally in good agreement with the experimental values with Aiso differing by just 2.3 MHz (**Table 8**). However, for both β–hydrogens, less satisfactory agreement between theory and experiment is observed. Specifically, the calculated Aiso values of 15.4 and 12.7 MHz differ from their corresponding experimental values by 9.6 and 3.2 MHz, respectively. As noted for the α– hydrogen in **I**, these differences are likely due to bulk crystal effects not being modeled in the calculations. Similarly, for the γ–hydrogens on the amino group, a difference of 4.0 MHz is seen between the calculated and experimental values. However, it has been stated that the given experimental value is associated with a degree of uncertainty.(Sanderud & Sagstuen 1996) In the optimized structure of **II** the C3 centre is not completely planar where an angle between the hydrogen on C3 and the plane defined by C3–C2–O was found to be 22.0º.(Lipfert et al. 2004) This is in good agreement with the corresponding experimentally determined value of 19.5º.(Sanderud & Sagstuen 1996) It is noted that the C3 centre and adjacent oxygen have spin densities of 0.84 and 0.11 respectively, suggesting slight delocalization of the SOMO. Dobbs et al.(Dobbs et al. 1971; Dobbs et al. 1972) have suggested that this is due to incomplete rehybridization of the carbon–centered radical and is common when the carbon is bonded to an oxygen.

Indeed, in radical-cationic **III** in which the phosphate group on C3 has been lost, i.e., there is no C3—O bond, the radical is now fully localized on C3. Furthermore, from **Table 8** it can be seen that quite good agreement is obtained for α-H1 and the β-hydrogen with differences of 1.7 and 0.0 MHz respectively, while α-H2 differs slightly more by 8.5 MHz. It should be noted that in order to obtain the best agreement for the HFCCs of the α–hydrogens they had to be tilted out of the plane by 5º. That is, the planarity of the C3 centre had to be reduced. This is similar to that observed need in related studies on the methyl radical (•CH3) to slightly pyramidalize the carbon centre in order to obtain best agreement between the calculated and experimentally measured HFCCs.(Eriksson et al. 1994)
