The activation of N-glycosidic bond cleavage performed by base-excision repair enzyme hOGG1; theoretical study of the role of Lys 249 residue in activation of G, OxoG and FapyG

Abstract

The activation of N-glycosidic bond cleavage performed by the lysine 249 (Lys 249) residue of base-excision repair enzyme hOGG1 was calculated for 2′-deoxyguanosine (G), 8-oxo-2′-deoxyguanosine (OxoG) and N6-(2′-β-D-deoxyribofuranosyl)-2,6-diamino-4-hydroxy-5-formamidopyrimidine (FapyG). The interaction sites of Lys 249 included the C1′, N3, and N9 atoms of the nucleosides. The N9-pathway, specifically the attack of lone-pair electrons of glycosidic nitrogen N9 of a nucleoside on the proton of the Nε-ammonium group of Lys 249, resulted in effective activation of the C1′–N9 bond that was highly specific with respect to normal (G) and damaged (OxoG, FapyG) nucleosides. The specificity of the N9-pathway was because of the electrophilic (G) or nucleophilic (OxoG, FapyG) character of the glycosidic nitrogen and because of the specific interactions of the residues within the catalytic pocket with the substrate (particularly the Gly 42 hOGG1 residue) that enforced the displacement of G out of the interaction range of Lys 249. The chemical modifications of G owing to damage specifically affected a number of molecular properties, particularly the electrophilicity/nucleophilicity of N9 and the C1′–N9 bond order and the aromatic character of the nucleobases. The N9-pathway could be involved as a check-point mechanism during base-excision performed by hOGG1.

---

Introduction

The repair of DNA is necessary because genetic material may be altered owing to pollutants and some chemical compounds that occur normally in the cell. The damaged DNA must be repaired; otherwise, it may cause serious damage to organisms. The extreme efficiency and specificity of DNA repair with respect to particular kinds of damage is ensured by base-excision repair enzymes (BER enzymes).^1,2

The BER enzyme specifically targeting 8-oxo-2′-deoxyguanosine (OxoG) that is produced by the oxidation of normal 2′-deoxyguanosine (G), is called human 8-oxoguanine glycosylase 1 (hOGG1).^3–12 The repair mechanism of hOGG1 has been generally described.^13,14 The excision of the OxoG base is initiated by the formation of a DNA–hOGG1 conjugate involving interaction of the lysine 249 residue (Lys 249) with OxoG that is extruded out of the double helix and inserted into the hOGG1 catalytic pocket.^1,3,15 After OxoG elimination, the pathway continues by the formation of a Schiff base and the repair process terminates by the cleavage of the defective DNA strand. The mechanism of the excision of OxoG is currently unclear despite the fact that the structure of the catalytic pocket, including the substrate nucleoside, is known. The role of the residues within the hOGG1 catalytic pocket, including Lys 249, remained unknown.^16–19 The original base-excision S_N1- or S_N2-type mechanisms were proposed based on the crystallographic and biochemical data.^{3,13,14,17,18} The theoretical modeling of the excision pathways was aimed particularly at the role of hOGG1 residues during OxoG excision.^20–26

The recognition of OxoG lesions by hOGG1 is extremely specific.¹ Consecutive check of the individual nucleobases within the hOGG1 catalytic pocket is not assumed; however, the DNA–hOGG1 complex containing flipped out normal G can be formed transiently.²⁷ Moreover, the existence of an unknown check-point mechanism responsible for distinguishing OxoG from G was reported.²⁸ Although normal G was enforcedly inserted into the hOGG1 catalytic pocket, the base excision was not observed.²⁸ The peculiar check-point mechanism may therefore case-selectively recognize some local property of the substrate nucleoside. With that goal in mind, the interactions of Lys 249 with three model nucleosides were studied in this work.

The specific stabilization and activation of a substrate within the catalytic site is a typical strategy common to all enzymes, not excluding the BER enzymes.^1,29 In this respect, the pre-protonation of a nucleobase activates N-glycosidic bond cleavage because the negative charge of the nucleobase during excision is compensated. The proton addition to the O8 oxygen,²⁰ N3 nitrogen,³⁰ and N9 glycosidic nitrogen³¹ of OxoG were previously theoretically modeled.

The interaction of protonated Lys 249 with the glycosidic nitrogen of OxoG enforced a distinct pyramidal geometry of N9 (N9-pyramidalization) and shifted its electronic state from sp²-like toward sp³-like, which allowed substitution of the N-glycosidic bond with an N9–H bond.³¹ The pyramidal geometry of the glycosidic nitrogen was previously observed in crystal structures of normal DNA and RNA molecules.³² The N9-pyramidalization of G residues within the DNA G-quadruplex was enforced by surrounding molecules.³³ The N9-pyramidalization of model nucleosides was studied in this work as a key structural descriptor of one of the base-excision pathways.

The previous calculations of N-glycosidic bond cleavage indicated that Lys 249 may interact with OxoG and normal G differently.³¹ In particular, the electronic state of glycosidic nitrogen might allow or block its interaction with Lys 249. To analyze this check-point mechanism in detail we focused on the properties of the G, OxoG and N6-(2′-β-D-deoxyribofuranosyl)-2,6-diamino-4-hydroxy-5-formamidopyrimidine (FapyG) nucleosides (Fig. 1). Out of the three nucleosides only OxoG and FapyG are cleavable with hOGG1 and better cleavability was observed for FapyG.^17,18,34 The theoretical study on the activation of the N-glycosidic bond is therefore well supported by experimental observations.

Fig. 1 The chemical structures of 2′-deoxyguanosine (G), 8-oxo-2′-deoxyguanosine (OxoG) and N6-(2′-β-D-deoxyribofuranosyl)-2,6-diamino-4-hydroxy-5-formamidopyrimidine (FapyG) nucleoside. The atom numbering used in the text is depicted only for OxoG. The sugar-to-base orientation was measured with the glycosidic torsion angle χ = C4–N9–C1′–O4′ and the extent of N9-pyramidalization was measured employing torsion angle κ′ = C4–N9–C1′–C8–180° (G, OxoG) or κ′ = C4–N9–C1′–H9–180° (FapyG).

Results and discussion

Lys 249 is a key hOGG1 residue; however, its exact role during the base excision is currently unknown. Two base-excision pathways employing Lys 249 were proposed based on pioneering X-ray structural data.¹⁷ The classical scheme assumed for BER enzymes is, in the case of hOGG1, initiated by neutral Lys 249, specifically because of the interaction of the Nε-amino group of Lys 249 with the C1′ anomeric carbon of OxoG.^3,5,13,14,18 The reaction of Lys 249 with the nucleoside that initiates S_N2 reaction will be called the C1′-pathway (Fig. 2). It was suggested that the protonated Lys 249 stabilizes the leaving nucleobase after its excision owing to the interaction of the cationic Nε-ammonium group of Lys 249 with the anionic OxoG base.^1,17 The actual state of Lys 249 during excision is currently unknown; however, the occurrence of the protonated form was previously assumed based on the experimental NMR data.²⁰

Fig. 2 The initiation of reaction pathways performed by the Lys 249 hOGG1 residue as described employing small models (see Computational details): the N9-pathway assumes interaction of the Nε-ammonium group (Lys 249) with glycosidic nitrogen N9, the N3-pathway assumes interaction of Nε-ammonium with the N3 nitrogen, and the C1′-pathway assumes interaction of the Nε-amino group of Lys 249 with anomeric carbon C1′.

The pre-protonation of the nucleobase enhances its excision. Several reactions employing this strategy were previously proposed and calculated. The stabilization of the cleaved OxoG base by the Nε-ammonium group of Lys 249 was calculated by Schyman and coworkers.²¹ The proton addition to the O8 oxygen of OxoG was calculated by Osakabe and coworkers.²⁰ This scheme is not included in this work because: (a) the activation energy of OxoG excision (42 kcal mol⁻¹)²⁰ was significantly higher than values expected for BER enzymes (≈19 kcal mol⁻¹)¹ and (b) the direct contact of the O8 oxygen with the Nε-ammonium group of Lys 249 would be hardly possible because of the O8–Nε distance (4.74 Å and 4.34 Å, X-ray structures 1N3C and 2NOZ). The addition of a proton to the N3 nitrogen as a mode of N-glycosidic bond activation was proposed by Jang and coworkers.³⁵ The protonation of the N3 nitrogen of OxoG by deprotonation of the Nε-ammonium group of Lys 249 was calculated by Calvaresi and coworkers.³⁰ This initial reaction of Lys 249, resulting again in an S_N2 reaction pathway (after N3 protonation the reaction continues via the C1′-pathway) will be called the N3-pathway (Fig. 2). The effect of protonated glycosidic nitrogen on the hydrolytic cleavage of the N-glycosidic bond was calculated by Cysewski and coworkers.³⁶ The strong activation of the hydrolytic cleavage was concluded, but the protonation of N9 was considered practically impossible because of its very low basicity.³⁶ The attack of lone-pair electrons of the glycosidic nitrogen on the proton of the Nε-ammonium group of Lys 249 that allowed proton addition to N9 during the cleavage of the N-glycosidic bond of OxoG was calculated by Sebera and coworkers.³¹ The interaction of Lys 249 with the N9 nitrogen (N9-pathway, Fig. 2) is equally possible as the interactions with the C1′ and N3 atoms of OxoG. The distances between the Nε nitrogen of Lys 249 and the N9, N3, and C1′ atoms of OxoG ranged from 3.0 to 3.5 Å in the X-ray structures 1N3C and 2NOZ. The cleavage of the N-glycosidic bond of OxoG employing the C1′-pathway, N3-pathway, and N9-pathway were previously calculated and the lowest activation energy was obtained for the N9-pathway.³¹

The optimized geometries of G, OxoG, and FapyG nucleosides calculated employing small models were different. The lengths of the N-glycosidic bonds (r_C1′–N9) of G, OxoG, and FapyG were 1.468 Å, 1.456 Å, and 1.446 Å and the κ′ torsion angles were −3.1°, 10.5°, and 30.3°, respectively. The C1′–N9 bond of damaged nucleosides shortened compared to normal G, and N9-pyramidalization increased. The r_C1′–N9 shortened the most by 0.008 Å and the N9-pyramidalization slightly changed only owing to the C1′-pathway (Table 1). The r_C1′–N9 of FapyG lengthened by 0.018 Å and N9-pyramidalization of OxoG and FapyG increased owing to the N3-pathway (Table 1). The initiation of the N3-pathway may continue with proton transfer from Nε-ammonium to the N3 nitrogen that was calculated for all the three nucleosides. The r_C1′–N9 and N9-pyramidalization were practically unaffected by the proton transfer (Table 1).

Table 1 The calculated geometric parameters and bond orders of free nucleosides and nucleosides interacting with Lys 249^a

Parameter^b	Free nucleoside	C1′-pathway	N3-pathway		N9-pathway
a The interaction of Lys 249 with nucleosides is indicated in parentheses as follows: C1′-pathway (C1′), N3-pathway (N3), N3-pathway with proton transferred from Nε-ammonium to nitrogen N3 (N3H), N9-pathway (N9), N9-pathway with proton transferred from Nε-ammonium to nitrogen N9 (N9H).b The rC1′–N9 bond length in Å, the C1′–N9 bond order, and the κ′ torsion angle in degrees (the definition of κ′ is shown in Fig. 1). The calculations were performed employing the small models depicted in Fig. 2 and described in the Computational details. —: not calculated (see the text).
	G	G (C1′)	G (N3)	G (N3H)	G (N9)	G (N9H)
rC1′–N9	1.468	1.460	1.461	1.471	—	—
C1′–N9	0.898	0.908	0.913	0.897	—	—
κ′	−3.1	−3.8	1.5	2.3	—	—
	OxoG	OxoG (C1′)	OxoG (N3)	OxoG (N3H)	OxoG (N9)	OxoG (N9H)
rC1′–N9	1.456	1.454	1.458	1.464	1.479	—
C1′–N9	0.924	0.922	0.923	0.912	0.905	—
κ′	10.5	2.6	17.1	18.4	37.8	—
	FapyG	FapyG (C1′)	FapyG (N3)	FapyG (N3H)	FapyG (N9)	FapyG (N9H)
rC1′–N9	1.446	1.446	1.463	1.464	1.480	1.514
C1′–N9	0.971	0.969	0.957	0.947	0.929	0.867
κ′	30.3	31.6	41.9	38.3	49.6	58.0

---

The N9-pathway reaction was calculated only for OxoG and FapyG. The stationary point describing the respective interaction of G with Lys 249 was not calculated and all attempts to optimize the geometry resulted in the N3-pathway reaction (data not shown). The N9-pathway with proton transfer to nitrogen N9 was calculated only for FapyG. The r_C1′–N9 of OxoG and FapyG lengthened owing to the N9-pathway by 0.023 Å and by 0.034 Å, respectively, and by 0.068 Å owing to proton transfer to N9 of FapyG. The κ′ torsions of OxoG and FapyG increased owing to the N9-pathway by 27.3° and 19.6°, respectively, and by 27.7° owing to proton transfer to N9 of FapyG (Table 1). The N9-pathway affected key geometric parameters of OxoG and FapyG more than the other two pathways. The lengthening of the N-glycosidic bond and the increase of N9-pyramidalization were marked changes indicative of the activation of the N-glycosidic bond.

The calculations employing medium and large models revealed the effect of hOGG1 residues within the catalytic pocket on activation of the N-glycosidic bond (Fig. 3). The calculated glycosidic torsion angles χ of the nucleosides ranged from 248.7° to 290.5°, which was in agreement with the anti orientation of OxoG and G in X-ray structures (Table S4†). The r_C1′–N9 of nucleosides within the catalytic pocket (medium model) and the r_C1′–N9 of free nucleosides (small model) decreased in an order that was coherent with X-ray structural data (Table 1 and S4†). Damage to G affected the intrinsic properties of the N-glycosidic bond. The length of the glycosidic bond of damaged nucleosides is shorter.

Fig. 3 The positioning of OxoG nucleoside within the hOGG1 catalytic pocket as optimized employing the large model. The Nε-ammonium group of Lys 249 interacts with glycosidic nitrogen N9 of OxoG. The H-bond between the oxygen of Gly 42 and the H7 hydrogen of OxoG is a key for proper stabilization of OxoG within the catalytic site. For geometric parameters see Table S4.†

The QM/MM calculations employing the large model indicated that a relatively ordered catalytic core and Nε-ammonium group of Lys 249 is closer to atom N9 of OxoG than to the N3 or C1′ atoms of OxoG (Table S4†). The CAM-B3LYP calculations employing medium models revealed that a distinct initiation of the N9-pathway in the case of OxoG and FapyG, while in the case of G the N3-pathway was initiated. The Nε-N9 distances calculated for G, OxoG, and FapyG were 3.629 Å, 3.021 Å, and 2.777 Å and the Nε-N3 distances were 2.704 Å, 3.425 Å, and 3.260 Å, respectively. The X-ray structures revealed that the catalytic pocket containing G was more open compared to the pocket containing OxoG (Table S4†). The calculations and X-ray geometries consistently indicated initiation of the N9-pathway in the case of OxoG and the N3-pathway in the case of G.

The catalytic pocket containing G was corrupted and the G nucleoside was displaced from Lys 249. The activation of the N-glycosidic bond of G employing Lys 249 was therefore less possible. The adjustment of the catalytic core in response to the inserted nucleobase was substrate-specific. Among hOGG1 residues within the catalytic site, Gly 42 specifically interacted with normal and damaged nucleosides. The H-bond between the oxygen of Gly 42 and hydrogen H7 stabilized the positioning of OxoG and FapyG (Fig. 3). While in the case of G, repulsion between the oxygen of Gly 42 and the N7 nitrogen occurred. The calculations and X-ray structures coherently indicated a key role of Gly 42 in freezing the well-defined position of the damaged nucleoside within the catalytic pocket.

The N9-pyramidalizations of OxoG and FapyG calculated employing medium and small models were similar because the κ′ torsions differed by less than 1.4° (Table 1 and S4†). Contrastingly, the N9-pyramidalization of G within the catalytic pocket (medium model, κ′ = 29.9°) was notably larger than the N9-pyramidalizations of G calculated employing small models (κ′ = −3.1° to 2.3°, Table 1). The peculiar N9-pyramidalization of G within the catalytic pocket was rationalized. First, the deformation energy needed for enforcement of the N9-pyramidalization (κ′ ≈ 30°) should be only ca. 1.5 kcal mol⁻¹.³³ Second, the starting geometry of G employed in geometry optimization was derived from the X-ray structure 2NOZ that showed close proximity of OxoG to Lys 249 and a particularly effective H-bond of OxoG with Gly 42. The N9-pyramidalization of G was therefore most probably enforced by the displacement of G owing to Gly 42 and not as a result of an efficient N9-pathway reaction. The initiation of the N9-pathway is conditioned by close contact of Nε-ammonium with lone-pair electrons on the glycosidic nitrogen that is apparently ensured only for the damaged nucleosides (Table S4†).

The Wiberg bond orders were calculated employing small models. The bond orders of the C1′–N9 bonds of free G, OxoG and FapyG were 0.898, 0.924, and 0.971, respectively. The gradual strengthening of the N-glycosidic bond was coherent with the bond shortening (Table 1). The bond orders of the C1′–N9 bond of OxoG owing to the C1′-pathway, N3-pathway, and N9-pathway were 0.922, 0.923, and 0.905, respectively. The degree of activation of the C1′–N9 bond can be evaluated with the Wiberg bond orders because the activation energies of N-glycosidic bond cleavage previously calculated decreased in the same order as the bond orders.³¹ The activation of the C1′–N9 bond owing to the N9-pathway and particularly owing to the N9-pathway with proton transfer to N9 was more efficient than activations owing to other two pathways (Table S5†).

The NBO charges were calculated employing small models. The charges of nitrogen N9 of G, OxoG and FapyG ranged from −0.420 to −0.410 e, −0.474 to −0.544 e, and −0.745 to −0.645 e, respectively. The charge of the C1′ carbon ranged from 0.244 to 0.280 e and the charge of the N3 nitrogen ranged from −0.655 to −0.590 e (Table S6–S8†). The interaction of nitrogen N9 with the Nε-ammonium group of Lys 249 was more enhanced in the case of damaged nucleosides than in normal G. The N3-pathway seemed to be competitive in that regard, particularly in the case of OxoG. The effect of hOGG1 residues on the proper mutual arrangement of Lys 249 with respect to the nucleosides seemed indispensable.

The Fukui indices f² were calculated employing small models. The positive or negative value of f² indicated the electrophilic or nucleophilic character of atoms in the nucleosides. The C1′ carbon was slightly electrophilic (Table 2, S9–S11†). The N3 nitrogen was nucleophilic except for the N3 of G, which had a transferred proton owing to N3-pathway. The N9 nitrogen of G was electrophilic, but N9 of OxoG and FapyG were increasingly nucleophilic. Importantly, the glycosidic nitrogen atoms of the nucleosides interacting with Lys 249 maintained their inherent character, which was coherent with the specificity of the N9-pathway with respect to normal and damaged nucleosides. Particular characteristics of the glycosidic nitrogens determined their inherent propensity for interaction with the electrophile (the proton of Nε-ammonium) or the nucleophile (the Nε-amino group). The significant nucleophilicity of N9 of FapyG allowed proton addition. The electrophilicity of N9 of G caused ineffective initiation of the N9-pathway. A similar discriminative role of the N9 nitrogen was recently proposed for the excision of hypoxanthine by the alkyladenine DNA glycosylase repair enzyme.³⁷ The chemical hardness of G nucleoside was greater than the chemical hardness of OxoG and FapyG (Table S12†). The effect of the electronic character of N9 on activation of the N-glycosidic bond could therefore be amplified by increased reactivity of the damaged nucleosides.

Table 2 The Fukui indices f² calculated for the C1′, N9, and N3 atoms of G, OxoG and FapyG^a

Atom	Free nucleoside	C1′-pathway	N3-pathway		N9-pathway
a The numbering of atoms is depicted in Fig. 1. The three pathways are indicated in parentheses as follows: C1′-pathway (C1′), N3-pathway (N3), N3-pathway with proton transferred from Nε-ammonium to the N3 nitrogen (N3H), N9-pathway (N9), N9-pathway with proton transferred from Nε-ammonium to the N9 nitrogen (N9H). The calculations were performed employing the small models depicted in Fig. 2 and described in Computational details. —: not calculated (see the text).
	G	G (C1′)	G (N3)	G (N3H)	G (N9)	G (N9H)
C1′	0.000	0.007	0.006	0.001	—	—
N9	0.020	−0.003	0.011	0.011	—	—
N3	−0.104	−0.117	−0.068	0.015	—	—
	OxoG	OxoG (C1′)	OxoG (N3)	OxoG (N3H)	OxoG (N9)	OxoG (N9H)
C1′	0.003	0.007	0.008	0.005	0.004	—
N9	−0.013	−0.028	−0.029	−0.022	−0.017	—
N3	−0.048	−0.059	−0.046	−0.021	−0.080	—
	FapyG	FapyG (C1′)	FapyG (N3)	FapyG (N3H)	FapyG (N9)	FapyG (N9H)
C1′	0.009	0.010	0.008	0.008	0.004	0.005
N9	−0.063	−0.092	−0.086	−0.080	−0.036	−0.006
N3	−0.034	−0.038	−0.041	−0.018	−0.077	−0.078

---

Nucleus-independent chemical shielding (NICS) is an indicator of aromaticity.^38,39 Negative or positive NICS values computed at the ring center indicate its aromatic or anti-aromatic character (Fig. S1†). The character of the five-membered ring of G was more aromatic than the character of the five-membered ring of OxoG (Table 3). The six-membered ring of G possessed less aromatic character than the six-membered ring of OxoG. Interestingly, the decrease in aromaticity of the five-membered ring owing to damage was balanced by the aromaticity increase of the unmodified six-membered ring. The highest decrease in aromaticity of the five-membered ring of OxoG was calculated for the N9-pathway, which was coherent with the depletion of electron density near N9 owing to the less conjugated character of the N9–C8 and N9–C4 inner-ring bonds (Table S5†).

Table 3 The NICS values calculated for the five- and six-membered rings of G, OxoG and FapyG^a

NICS	Free nucleoside	C1′-pathway	N3-pathway		N9-pathway
a The nucleus independent chemical shielding NICS-5 and NICS-6 in ppm were calculated at the centre of mass of the five- and six-membered rings of the nucleobases.b The NICS-6 values calculated for the six-membered pseudo-ring of FapyG containing glycosidic nitrogen N9. The interaction of Lys 249 with the nucleosides are indicated in parentheses as follows: C1′-pathway (C1′), N3-pathway (N3), N3-pathway with proton transferred from Nε-ammonium to the N3 nitrogen (N3H), N9-pathway (N9), N9-pathway with proton transferred from Nε-ammonium to the N9 nitrogen (N9H). The calculations were performed employing small models (Fig. 2) as described in Computational details. —: not calculated (see the text).
	G	G (C1′)	G (N3)	G (N3H)	G (N9)	G (N9H)
NICS-5	−12.5	−11.6	−12.0	−12.8	—	—
NICS-6	−4.4	−4.2	−3.8	−3.9	—	—
	OxoG	OxoG (C1′)	OxoG (N3)	OxoG (N3H)	OxoG (N9)	OxoG (N9H)
NICS-5	−10.5	−10.9	−10.7	−10.9	−9.1	—
NICS-6	−4.9	−4.7	−4.0	−4.2	−4.3	—
	FapyG	FapyG (C1′)	FapyG (N3)	FapyG (N3H)	FapyG (N9)	FapyG (N9H)
NICS-6^b	−1.3	−1.3	−0.7	−0.7	−0.5	−0.8
NICS-6	−3.9	−3.2	−3.2	−4.0	−4.1	−3.8

---

The N3-pathway and N9-pathway were compared by means of calculated interaction energies. The absolute values of ΔE_int energies calculated for the N3-pathway were larger than the energies for the N9-pathway (Table 4), which was coherent with their respective bond orders. The H–N3 bond orders calculated for the H atom of Nε-ammonium and the N3 nitrogen of nucleosides owing to the N3-pathway were larger than the H–N9 bond orders owing to the N9-pathway (Table S5†). The interaction of Nε-ammonium with the N3 nitrogen was better energy-stabilized than the interaction with the N9 nitrogen. On the other hand, the N9-pathway was increasingly stabilized for the damaged nucleosides while the N3-pathway was more conservative owing to the distinct electrophilicity of N3. The interaction of Lys 249 with the N9 nitrogen depended on the character of the N9 nitrogen. The more distinct was the nucleophilic character of the glycosidic nitrogen the more stable was its interaction with Nε-ammonium. Importantly, the interaction of Nε-ammonium with atom N9 of OxoG is energetically stabilized owing to its negative interaction energy.

Table 4 The interaction energies calculated for G, OxoG and FapyG nucleosides interacting with Lys 249

Nucleoside^a (pathway)	ΔEint^b
a The N3-pathway and N9-pathway are depicted for OxoG in Fig. 2.b The interaction energy ΔEint in kcal mol^−1 was calculated as the energy of small models minus the energy of monomers (Lys 249, nucleoside) including the basis set superposition error (BSSE). The BSSE represented ca. 1 kcal mol^−1 and calculation details are described in Table S13.
G (N3)	−8.3
OxoG (N3)	−7.6
OxoG (N9)	−2.5
FapyG (N3)	−11.0
FapyG (N9)	−9.8

---

The specific activation of the N-glycosidic bond of G, OxoG, and FapyG was calculated because the normal and damaged nucleosides possess characteristic structural and electronic properties. When going from G to OxoG to FapyG the C1′–N9 bond shortened, the respective bond order increased, the charge of the N9 nitrogen decreased, the lone-pair electrons at N9 were more localized (Fig. S2†), the aromaticity of the five-membered ring decreased and the character of the glycosidic nitrogen changed from electrophilic to nucleophilic.

The activation of the N-glycosidic bond employing the C1′-pathway was relatively unspecific compared to the N3-pathway or the N9-pathway. The Gibbs free energies of the cleavage of the N-glycosidic bonds of G, OxoG, and FapyG calculated employing the C1′-pathway were 30.7, 32.0 and 35.2 kcal mol⁻¹, respectively (small models, Table S14 and Fig. S3†). The ΔG^# energies thus increased in the same order as increased the respective C′–N9 bond orders, which indicated rather dependence of cleavage on the bond strength and small activation role of Lys 249. Actually, the N-glycosidic bond of G is more labile toward hydrolytic cleavage than the N-glycosidic bond of OxoG.⁴⁰ The minor activation role of Lys 249 in calculations employing the C1′-pathway can be therefore assumed. The hydrolytic and enzymatic cleavage apparently proceeds according to different scenarios. The calculated activation employing the C1′-pathway was in conflict with the observed cleavability of G and OxoG by hOGG1.^17,18,28,34

The protonation of the N9 and N3 atoms owing to the N9-pathway and the N3-pathway depended on the degree of their nucleophilicity. The proton addition to the N3 nitrogen was calculated for all the three nucleosides while the proton addition to the N9 nitrogen was calculated only for FapyG. The Gibbs free activation and reaction energies calculated for the transfer of a proton from Nε-ammonium to the N3 or N9 nitrogen indicated that proton transfer to the N3 nitrogen of G should be easier and better stabilized than proton transfer to N3 or N9 of OxoG and FapyG (Table S14 and Fig. S4 and S5†). The activation energies calculated for proton transfer were nevertheless quite similar. The N3-pathway and N9-pathway reaction may thus probably result in pre-protonation of N3 nitrogen of nucleobases and N9 nitrogen of FapyG.

The activation performed by hOGG1 employs a mechanism that specifically activates the intrinsically more stable N-glycosidic bond of damaged nucleosides. The N9-pathway was specific in that regard, but the “net” interaction of Lys 249 with glycosidic nitrogen was only loosely stabilized. The workability of the relatively fragile N9-pathway was conditioned by the nucleophilic character of glycosidic nitrogen and by the hOGG1 catalytic site that ensured effective attack of the lone-pair electrons of the glycosidic nitrogen on the proton of Nε-ammonium by proper deposition of the damaged nucleosides (Fig. S6†). The N9-pathway may be regarded as a typical enzymatic strategy because its workability is strictly conditioned by arrangement of the catalytic core and the distinct character of the substrate nucleoside.

Conclusions

The activation of the N-glycosidic bond employing the Lys 249 residue of the hOGG1 enzyme was calculated for normal G, OxoG and FapyG nucleosides. The activation of the N-glycosidic bond performed by the attack of the lone-pair electrons of the glycosidic nitrogen on the Nε-ammonium group of Lys 249 was efficient and specific with respect to normal and damaged nucleosides. The activation of normal G initiated by the N9-pathway reaction was ineffective owing to the electrophilic character of the glycosidic nitrogen of G and a corrupted catalytic core that contained a displaced substrate. The activation of the damaged nucleosides was effective owing to the nucleophilic character of their glycosidic nitrogens and favorable mutual arrangement of the nucleosides with respect to Lys 249. The glycosidic nitrogen of both OxoG and FapyG can donate lone-pair electrons capable of interaction with the proton of the Nε-ammonium group of Lys 249, while the lone-pair electrons at N9 of G cannot interact efficiently with Nε-ammonium because of their delocalization within the five-membered ring.

The interaction of the protonated Lys 249 with electrophilic or nucleophilic glycosidic nitrogen triggered specific activation of the N-glycosidic bonds of normal and damaged nucleosides that could be involved in a check-point mechanism during base-excision performed by hOGG1. We therefore propose that Lys 249 is not only a key catalytic residue, but also the residue involved in recognition of damaged nucleobases.

Computational details

Three structural models called small, medium and large were employed. The small models that were derived employing the X-ray structure 1N3C¹⁷ consisted of nucleosides and Lys 249 (Fig. 2). Mutual positioning of the residues was constrained by fixed geometries of the C4′ and C3′ atoms of the nucleosides and the –OCCN– atoms of Lys 249. The B3LYP-D2 method including Grimme's dispersion corrections D2,^41–43 the 6-31G(d,p)^44,45 basis set, and the PCM solvent (diethylether) were employed in geometry optimizations. The PCM method was successfully used to mimic overall polarization within the catalytic site in similar studies.^46–48 The energy minima and transition states were confirmed by vibrational frequency analysis. The free nucleosides (not interacting with Lys 249) were optimized without constraints. The medium models that were derived employing the X-ray structure 2NOZ¹⁹ included nucleosides and hOGG1 residues within a 10 Å sphere centred at the glycosidic nitrogen of the nucleosides (Fig. 4). The Lys 249, Asp 268, Cys 253, Phe 319, W2979, W2908, W2907 water molecules, and nucleoside defined the inner part within the ONIOM QM/QM⁴⁹ method that was optimized employing the CAM-B3LYP DFT method,⁵⁰ while the outer part (650 atoms) was optimized employing the PM6 method keeping the heavy atoms fixed. The small and medium models of G and FapyG were derived using the original X-ray structures that included OxoG and by retaining the maximal overlap of nucleobases. The large model was derived employing the 2NOZ X-ray structure that described the hOGG1–DNA complex (Fig. 5). Only the DNA heptamer including OxoG in the middle was included in the large model. The OxoG and neighbouring nucleotides within the DNA strand and hOGG1 protein were geometry optimized, while the rest of the heavy atoms of the DNA heptamer were maintained in their X-ray geometries. The QM/MM AM1 ⁵¹/OPLS_2005 ⁵² and B3LYP(6-31G(d,p))/OPLS_2005 methods within the additive scheme were employed in geometry optimizations of the large model. The hydrogen cap and electrostatic treatment at the QM/MM interface using Gaussian charge distribution on a grid was employed. The QM part of the large model included the OxoG nucleoside and the Lys 249, Gly 42, and Gln 43 residues of hOGG1. The protonated form of Lys 249 was involved in both the medium and large models.

Fig. 4 The medium structural model.

Fig. 5 The large structural model.

The calculated N9-pyramidalizations (κ′ = 27.7° to 48.2°) were larger than the N9-pyramidalizations in X-ray structures (κ′ = −0.3° to 2.8°). The ultra-high resolution of the X-ray structures (resolution of 1 Å or better and R-factor < 0.2) is required because the lower resolution that is often linked with usage of an empirical force-field may result in underestimated N9-pyramidalization.³² The resolutions of the 3IH7,²⁸ 1YQK,⁵³ 1N3C,¹⁷ and 2NOZ¹⁹ X-ray structures ranged from 2.43 Å to 3.10 Å and the N9-pyramidalizations might be therefore underestimated.

The molecular properties calculated employing small models included geometry, natural bond orbital (NBO) atomic charges, natural localized molecular orbitals (NLMOs), Fukui functions,^54,55 chemical hardness,^56,57 interaction energies, bond orders, Wiberg bond indices,⁵⁸ nucleus independent chemical shifts (NICSs),^38,59,60 and Gibbs free activation and reaction energies. The NICSs were calculated employing the B3LYP/6-31G(d,p) method. The interaction energies were calculated employing the B3LYP-D2/6-311++G** method including the BSSE correction. All other properties were calculated employing the B3LYP-D2/6-31G(d,p) method. Further details can be found in the ESI.†

The Fukui index of an atom in a molecule f² = f⁺ − f⁻ was calculated as the difference of Fukui functions f⁻ = q(N − 1) − q(N) and f⁺ = q(N) − q(N + 1) where the q(N − 1), q(N), and q(N) were the atomic NBO charges of molecules containing N − 1, N, and N + 1 electrons, respectively. The global reactivity descriptor of a molecule called chemical hardness was calculated employing the electronic energies of the LUMO and HOMO molecular orbitals.

The small and medium models were geometry optimized using the Gaussian 09 program.⁶¹ The large model was optimized using the QSite program.⁶² The NBO atomic charges, NLMO orbitals and the Wiberg bond indices were calculated within the NBO analysis⁶³ using the NBO program 3.1.⁶⁴

Acknowledgements

This work was supported by the Czech Science Foundation GA ČR, grant number 13-27676S and by the Academy of Sciences, grant number M200551205. Y.T. and V.S. acknowledge the Young Investigator's Grant of the Human Frontier Science Program (HFSP). L.T. was supported by the project “CEITEC - Central European Institute of Technology” (CZ.1.05/1.1.00/02.0068) from European Regional Development Fund and from the European Social Fund and the state budget of the Czech Republic (CZ.1.07/2.3.00/20.0042).

References

1. P. J. Berti and J. A. B. McCann, Chem. Rev., 2006, 106, 506–555.
2. S. S. David and S. D. Wiliams, Chem. Rev., 1998, 98, 1221–1261.
3. H. M. Nash, S. D. Bruner, O. D. Scharer, T. Kawate, T. A. Addona, E. Sponner, W. S. Lane and G. L. Verdine, Curr. Biol., 1996, 6, 968–980.
4. M. Bjoras, L. Luna, B. Johnson, E. Hoff, T. Haug, T. Rognes and E. Seeberg, EMBO J., 1997, 16, 6314–6322.
5. H. M. Nash, R. Z. Lu, W. S. Lane and G. L. Verdine, Chem. Biol., 1997, 4, 693–702.
6. R. Z. Lu, H. M. Nash and G. L. Verdine, Curr. Biol., 1997, 7, 397–407.
7. S. Boiteux and J. P. Radicella, Arch. Biochem. Biophys., 2000, 377, 1–8.
8. M. D. Leipold, J. G. Muller, C. J. Burrows and S. S. David, Biochemistry, 2000, 39, 14984–14992.
9. K. Shinmura and J. Yokota, Antioxid. Redox Signaling, 2001, 3, 597–609.
10. H. Menoni, M. S. Shukla, V. Gerson, S. Dimitrov and D. Angelov, Nucleic Acids Res., 2012, 40, 692–700.
11. S. S. David, V. L. O'Shea and S. Kundu, Nature, 2007, 447, 941–950.
12. P. L. McKibbin, A. Kobori, Y. Taniguchi, E. T. Kool and S. S. David, J. Am. Chem. Soc., 2012, 134, 1653–1661.
13. Y. W. Kow and S. S. Wallace, Biochemistry, 1987, 26, 8200–8206.
14. M. L. Dodson, M. L. Michaels and R. S. Lloyd, J. Biol. Chem., 1994, 269, 32709–32712.
15. J. T. Stivers and Y. L. Jiang, Chem. Rev., 2003, 103, 2729–2759.
16. S. D. Bruner, D. P. G. Norman and G. L. Verdine, Nature, 2000, 403, 859–866.
17. D. P. G. Norman, S. J. Chung and G. L. Verdine, Biochemistry, 2003, 42, 1564–1572.
18. J. C. Fromme, S. D. Bruner, W. Yang, M. Karplus and G. L. Verdine, Nat. Struct. Biol., 2003, 10, 204–211.
19. C. T. Radom, A. Banerjee and G. L. Verdine, J. Biol. Chem., 2007, 282, 9182–9194.
20. T. Osakabe, Y. Fujii, M. Hata, M. Tsuda, S. Neya and T. Hoshino, Chem-Bio Inf. J., 2004, 4, 73–92.
21. P. Schyman, J. Danielsson, M. Pinak and A. Laaksonen, J. Phys. Chem. A, 2005, 109, 1713–1719.
22. J. A. B. McCann and P. J. Berti, J. Am. Chem. Soc., 2008, 130, 5789–5797.
23. A. L. Millen and S. D. Wetmore, Can. J. Chem., 2009, 87, 850–863.
24. E. J. Shim, J. L. Przybylski and S. D. Wetmore, J. Phys. Chem. B, 2010, 114, 2319–2326.
25. R. Rios-Font, J. Bertran, M. Sodupe and L. Rodriguez-Santiago, Theor. Chem. Acc., 2011, 128, 619–626.
26. J. L. Kellie and S. D. Wetmore, J. Phys. Chem. B, 2012, 116, 10786–10797.
27. A. A. Kuznetsova, N. A. Kuznetsov, A. A. Ishchenko, M. K. Saparbaev and O. S. Fedorova, Biochim. Biophys. Acta, Gen. Subj., 2014, 1840, 387–395.
28. C. M. Crenshaw, N. Kwangho, O. Kimberly, P. S. Kutchikian, B. Bowman, M. Karplus and G. L. Verdine, J. Biol. Chem., 2012, 287, 24916–24928.
29. P. J. O'Brien, Chem. Rev., 2006, 106, 720–752.
30. M. Calvaresi, A. Bottoni and M. Garavelli, J. Phys. Chem. B, 2007, 111, 6557–6570.
31. J. Sebera, L. Trantirek, Y. Tanaka and V. Sychrovsky, J. Phys. Chem. B, 2012, 116, 12535–12544.
32. V. Sychrovsky, S. Foldynova-Trantirkova, N. Spackova, K. Robeyns, L. Van Meervelt, W. Blankenfeldt, Z. Vokacova, J. Sponer and L. Trantirek, Nucleic Acids Res., 2009, 37, 7321–7331.
33. V. Sychrovsky, Z. S. Vokacova and L. Trantirek, J. Phys. Chem. A, 2012, 116, 4144–4151.
34. C. Dherin, J. P. Radicella, M. Dizdaroglu and S. Boiteux, Nucleic Acids Res., 1999, 27, 4001–4007.
35. Y. H. Jang, W. A. Goddard, K. T. Noyes, L. C. Sowers, S. Hwang and D. S. Chung, Chem. Res. Toxicol., 2002, 15, 1023–1035.
36. P. Cysewski, D. Bira and K. Bialkowski, J. Mol. Struct.: THEOCHEM, 2004, 678, 77–81.
37. X. Sun and J. K. Lee, J. Org. Chem., 2007, 72, 6548–6555.
38. Z. Chen, T. Heine, P. R. Schelyer and D. Sundholm, in Calculation of NMR and EPR Parameters, ed. M. B. M. Kaupp and V. G. Malkin, WILEY-VCH Verlag, 2004.
39. Z. F. Chen, C. S. Wannere, C. Corminboeuf, R. Puchta and P. V. Schleyer, Chem. Rev., 2005, 105, 3842–3888.
40. K. Bialkowski, P. Cysewski and R. Olinski, Z. Naturforsch., C: J. Biosci., 1996, 51, 119–122.
41. A. D. Becke, J. Chem. Phys., 1993, 98, 5648–5652.
42. C. T. Lee, W. T. Yang and R. G. Parr, Phys. Rev. B: Condens. Matter Mater. Phys., 1988, 37, 785–789.
43. S. Grimme, J. Comput. Chem., 2006, 27, 1787–1799.
44. W. J. Hehre, R. Ditchfie and J. A. Pople, J. Chem. Phys., 1972, 56, 2257–2261.
45. P. C. Harihara and J. A. Pople, Theor. Chim. Acta, 1973, 28, 213–222.
46. J. Tomasi, B. Mennucci and R. Cammi, Chem. Rev., 2005, 105, 2999–3093.
47. S. C. Harvey and P. Hoekstra, J. Phys. Chem., 1972, 76, 2987–2994.
48. G. King, F. S. Lee and A. Warshel, J. Chem. Phys., 1991, 95, 4366–4377.
49. T. Vreven, K. S. Byun, I. Komaromi, S. Dapprich, J. A. Montgomery Jr, K. Morokuma and M. J. Frisch, J. Chem. Theory Comput., 2006, 2, 815–826.
50. T. Yanai, D. P. Tew and N. C. Handy, Chem. Phys. Lett., 2004, 393, 51–57.
51. M. J. S. Dewar, E. G. Zoebisch, E. F. Healy and J. J. P. Stewart, J. Am. Chem. Soc., 1985, 107, 3902–3909.
52. W. L. Jorgensen, D. S. Maxwell and J. TiradoRives, J. Am. Chem. Soc., 1996, 118, 11225–11236.
53. A. Banerjee, W. Yang, M. Karplus and G. L. Verdine, Nature, 2005, 434, 612–618.
54. W. Yang and W. J. Mortier, J. Am. Chem. Soc., 1986, 108, 5708–5711.
55. C. Morell, A. Grand and A. Toro-Labbe, J. Phys. Chem. A, 2005, 109, 205–212.
56. P. D. P. Geerlings, F. De Proft and J. M. L. Martin, Theoretical and Computational Chemistry 4, Elsevier, 1996, pp. 773–809.
57. P. Geerlings, F. De Proft and W. Langenaeker, Chem. Rev., 2003, 103, 1793–1873.
58. K. B. Wiberg, Tetrahedron, 1968, 24, 1083–1096.
59. P. V. Schleyer, C. Maerker, A. Dransfeld, H. J. Jiao and N. Hommes, J. Am. Chem. Soc., 1996, 118, 6317–6318.
60. S. Pelloni, G. Monaco, P. Lazzeretti and R. Zanasi, Phys. Chem. Chem. Phys., 2011, 13, 20666–20672.
61. M. J. Frisch, G. W. Trucks, H. B. Schlegel, G. E. Scuseria, M. A. Robb, J. R. Cheeseman, G. Scalmani, V. Barone, B. Mennucci, G. A. Petersson, H. Nakatsuji, M. Caricato, X. Li, H. P. Hratchian, A. F. Izmaylov, J. Bloino, G. Zheng, J. L. Sonnenberg, M. Hada, M. Ehara, K. Toyota, R. Fukuda, J. Hasegawa, M. Ishida, T. Nakajima, Y. Honda, O. Kitao, H. Nakai, T. Vreven, J. A. Montgomery Jr, J. E. Peralta, F. Ogliaro, M. Bearpark, J. J. Heyd, E. Brothers, K. N. Kudin, V. N. Staroverov, R. Kobayashi, J. Normand, K. Raghavachari, A. Rendell, J. C. Burant, S. S. Iyengar, J. Tomasi, M. Cossi, N. Rega, J. M. Millam, M. Klene, J. E. Knox, J. B. Cross, V. Bakken, C. Adamo, J. Jaramillo, R. Gomperts, R. E. Stratmann, O. Yazyev, A. J. Austin, R. Cammi, C. Pomelli, J. W. Ochterski, R. L. Martin, K. Morokuma, V. G. Zakrzewski, G. A. Voth, P. Salvador, J. J. Dannenberg, S. Dapprich, A. D. Daniels, Ö. Farkas, J. B. Foresman, J. V. Ortiz, J. Cioslowski, and D. J. Fox, Gaussian 09, Revision B.02, Gaussian, Inc., Wallingford CT, 2009.
62. QSITE, v. 6.1, Schroedinger, LLC, New York, 2013.
63. A. E. Reed, L. A. Curtiss and F. Weinhold, Chem. Rev., 1988, 88, 899–926.
64. E. D. Glendening, A. E. Reed, J. E. Carpenter and F. Weinhold, NBO Program Version 3.1, Theoretical Chemistry Institute, University of Wisconsin.

---

Footnote

† Electronic supplementary information (ESI) available: Calculated molecular properties and geometric parameters in Tables S1–S15 and Fig. S1–S6. See DOI: 10.1039/c4ra08278h

---