Occurrence, Biological Consequences, and Human Health Relevance of Oxidative Stress-Induced DNA Damage (original) (raw)

. Author manuscript; available in PMC: 2017 Dec 19.

Abstract

A variety of endogenous and exogenous agents can induce DNA damage and lead to genomic instability. Reactive oxygen species (ROS), an important class of DNA damaging agents, are constantly generated in cells as a consequence of endogenous metabolism, infection/inflammation, and/or exposure to environmental toxicants. A wide array of DNA lesions can be induced by ROS directly, including single-nucleobase lesions, tandem lesions, and hypochlorous acid (HOCl)/hypobromous acid (HOBr)-derived DNA adducts. ROS can also lead to lipid peroxidation, whose byproducts can also react with DNA to produce exocyclic DNA lesions. A combination of bioanalytical chemistry, synthetic organic chemistry, and molecular biology approaches have provided significant insights into the occurrence, repair, and biological consequences of oxidatively induced DNA lesions. The involvement of these lesions in the etiology of human diseases and aging was also investigated in the past several decades, suggesting that the oxidatively induced DNA adducts, especially bulky DNA lesions, may serve as biomarkers for exploring the role of oxidative stress in human diseases. The continuing development and improvement of LC-MS/MS coupled with the stable isotope-dilution method for DNA adduct quantification will further promote research about the clinical implications and diagnostic applications of oxidatively induced DNA adducts.

Graphical abstract

graphic file with name nihms906582u1.jpg

1. Oxidative Stress and ROS

The human genome is constantly exposed to a variety of endogenous and exogenous agents that can generate DNA damage, which may compromise genomic integrity.1 Reactive oxygen species (ROS) constitute an important class of DNA damaging agents, and they are continuously generated in cells as a consequence of endogenous metabolism and/or exposure to environmental toxicants.2 ROS encompass a variety of chemical species, e.g., superoxide anion radical (O2−•), hydrogen peroxide (H2O2), hydroxyl radical (•OH), and singlet oxygen (1O2). In this vein, mitochondrion is considered a major source of ROS production in cells, where electrons leaking from the electron transport chain during mitochondrial respiration can combine with molecular oxygen to generate O2−•, which can be subsequently converted to H2O2 by superoxide dismutase (SOD).3 H2O2, which diffuses freely in the cellular environment, may react with the reduced-state transition metal ions to give •OH via the Fenton-type reactions:4

Cu+/Fe2++H2O2→Cu2+/Fe3++•OH+OH−

Oxidation of biomolecules depends on the location of ROS production and the redox potential of the biomolecules. Some of the aforementioned ROS, such as O2−• and •OH, are extremely unstable, whereas others, like H2O2, are relatively long-lived.2,5,6

Infection and inflammation activate inflammatory cells, which induce and activate various oxidant-generating enzymes.7 Activated inflammatory cells produce O2−• through nicotinamide adenine dinucleotide phosphate (NADPH) oxidase complexes or xanthine oxidase, and these cells release high concentrations of oxidant-generating enzymes such as myeloperoxidase and eosinophil peroxidase through degranulation.711 These enzymes produce high concentrations of reactive oxygen, nitrogen, and halogen species such as superoxide anion, nitric oxide, peroxynitrite, hydrogen peroxide, hypochlorous acid, and hypobromous acid at sites of inflammation.7,12 Although intended to neutralize invading pathogens, these reactive chemical species can result in collateral DNA damage of host cells.

Aside from damaging DNA directly, ROS may also lead to DNA damage indirectly, through reaction with lipids, proteins, and other cellular components to produce electrophilic species that can react with DNA.13,14 In particular, peroxidation of polyunsaturated fatty acids (PUFA) can give rise to a multitude of reactive aldehydes that can conjugate with DNA to yield DNA adducts.13,14 In this vein, DNA is susceptible to electrophilic attack because it contains many nucleophilic sites, including the _N_1, _N_2, _N_3, _N_7, and _O_6 of guanine; the _N_1, _N_3, _N_6, and _N_7 of adenine; the _O_2, _N_3, and _O_4 of thymine; and the _O_2, _N_3, and _N_4 of cytosine.15,16

In this review, we will discuss common types of oxidatively induced DNA lesions, including single-nucleobase lesions and tandem lesions that arise from direct ROS attack, as well as indirect ROS-induced DNA damage, such as those induced by inflammation and byproducts of lipid peroxidation. The emphasis is placed on their chemical mechanisms of formation, biological consequences, and human health relevance. In addition, we will discuss cellular replication and transcription studies of these lesions as well as their repair pathways and detection.

2. Chemistry of Oxidative Stress-Induced DNA Damage

2.1. Direct ROS-Induced DNA Lesions

2.1.1. Single-Nucleobase Lesions

•OH is highly reactive toward DNA; it can readily abstract a hydrogen atom from 2-deoxyribose or methyl group on nucleobases or be added to double bonds of purine and pyrimidine bases. Addition of hydroxyl radical to guanine leads to the formation of adduct radicals on the C4, C5, and C8 atoms, which have been previously reviewed.17 One-electron oxidation of the resulting C8-OH adduct radical gives rise to the formation of 8-oxo-7,8-dihydro-2′-deoxyguanosine (8-oxo-dG, Figure 1), which was first reported in the 1980s.18 The C8-OH adduct radical of guanine can also undergo an intramolecular ring opening of its imidazole moiety by the cleavage of the C8–N7 bond, followed by one-electron reduction, to yield 2,6-diamino-4-hydroxy-5-formamidopyrimidine 2′-deoxynucleoside (Fapy-dG, Figure 1). Alternatively, one-electron reduction may also occur prior to the imidazole ring-opening process, generating 7-hydro-8-hydroxy-2′-deoxyguanosine, which subsequently undergoes ring-opening to form Fapy-dG.19,20 Similarly, the reaction between 2′-deoxyadenosine (dA) and hydroxyl radical gives 8-oxo-7,8-dihydro-2′-deoxyadenosine (8-oxo-dA)21 and 4,6-diamino-5-formamidopyrimidine 2′-deoxynucleoside (Fapy-dA).20 Apart from hydroxyl radical attack, singlet oxygen (1O2) may also oxidize dG to yield 8-oxo-dG. In this vein, 18O-labeled 8-oxo-dG could be detected in DNA isolated from cells incubated with water-soluble 18O-labeled nonionic 1,4-endoperoxide N,_N_′-di(2,3-dihydroxypropyl)-1,4-naphthalene-dipropanamide (DHPN18O2), whose thermal decomposition gives 18O-labeled singlet oxygen.22 The above-mentioned DNA adducts have already been detected at appreciable levels in vivo and summarized in previous reviews.2328

Figure 1.

Figure 1

ROS-induced primary and secondary oxidation products of dG. [H] and [O] represent reduction and oxidation, respectively.

Guanine is the most easily oxidized among the four nucleobases in DNA, and a reduction potential of 1.29 V vs NHE was reported for guanosine.29 Compared to dG, 8-oxo-dG has an even lower reduction potential (0.74 V vs NHE).30 Indeed, it has been demonstrated that 8-oxo-dG can be readily oxidized by various oxidizing agents including Na2IrCl6, γ rays, peroxynitrite, Fenton-like reagents, hypochlorous acid (HOCl), etc.3136 One-electron oxidation of 8-oxo-dG can give rise to a radical cation, which can undergo hydration, deprotonation, and another one-electron oxidation to produce 5-hydroxy-substituted derivative of 8-oxo-dG (5-OH-8-oxo-dG).31,34 Subsequent oxidation of 5-OH-8-oxo-dG leads to the formation of guanidinohydantoin 2′-deoxynucleoside (dGh), spiroiminodihydantoin 2′-deoxynucleoside (dSp), and various other oxidation products.31,34 The distributions of these products depend on the reaction context (e.g., nucleoside, single-stranded DNA, double-stranded DNA, and base pairing),37 oxidizing agents,32 and reaction conditions (e.g., pH and temperature).38 Generally speaking, the formation of dGh is favored in duplex DNA, while the formation of dSp is preferred in nucleosides, single-stranded DNA, and G-quadruplex DNA.39 dGh is the predominant product at acidic pH, while dSp is the major product at higher pH.40 Under physiological pH, especially within cells where nucleosides, ssDNA, and dsDNA are available, the formation of both dGh and dSp are feasible.

The pyrimidine bases are also prone to attack by free radicals.15 For instance, the hydroxyl radical can be added to the C5═C6 double bond of thymine and cytosine to yield C5-OH- and C6-OH-adduct radicals (Figure 2).17 The C5-OH- and C6-OH-adduct radicals can lead to the formation of 5,6-dihydroxy-5,6-dihydrothymidine (thymidine glycol) via different reaction pathways in the presence or absence of O2 (Figure 2), which was reviewed by Dizdaroglu and Jaruga41 and Teoule.42 Similar mechanisms may account for the formation of 5-methyl-2′-deoxycytidine (5-mdC) glycol, which can undergo deamination to yield thymidine glycol.4347 The formation of 5-mdC glycol may be involved in the C → T transition mutations occurring at CpG dinucleotide sites, a type of mutation ubiquitously found in human cancers.48,49 Exposure to ionizing radiation50,51 and reaction with oxidizing agents, including KMnO4, OsO4,52 and Fenton reagents,47 can result in the formation of thymidine glycol in DNA.

Figure 2.

Figure 2

Pathways for hydroxyl radical-mediated oxidation of dT.

The hydroxyl radical can also abstract a hydrogen atom from the 5-methyl group of thymine and 5-methylcytosine to produce the 5-methyl radical of the two pyrimidine bases (Figures 2 and 3).17 In addition, the 5-methyl radical may also form from one-electron oxidation of the pyrimidine bases followed by deprotonation.53,54 The 5-methyl radical can be transformed to produce 5-hydroxymethyl-2′-deoxyuridine (5-hmdU) and 5-formyl-2′-deoxyuridine (5-fdU),41,42,55 5-hydroxymethyl-2′-deoxycytidine (5-hmdC), and 5-formyl-2′-deoxycytidine (5-fdC).53 Along this line, it was found recently that the oxidation of 5-mdC could also be catalyzed by Fe(II)- and 2-oxoglutarate (2-OG)-dependent ten-eleven translocation (TET) family dioxygenases.5658 The resulting 5-hmdC, 5-fdC, and 5-carboxyl-2′-deoxycytidine (5-cadC) can be considered as epigenetic marks (Figure 3a).5863 In addition, the removal of 5-fdC and 5-cadC by the base excision repair (BER) machinery is thought to play an important role in active cytosine demethylation in mammalian systems.60,64 Aside from being an oxidation product of thymidine, 5-hmdU may also arise from the deamination of 5-hydroxymethyl-2′-deoxycytidine (5-hmdC).65,66 In addition, Pfaffeneder et al.67 reported that TET enzymes could catalyze directly the formation of 5-hmdU from thymidine in the DNA of mouse embryonic stem cells (mESCs), and they also found that the deamination of 5-hmdC did not contribute significantly to the 5-hmdU level in mESCs (Figure 3b).67 5-fdU can be formed from thymidine upon exposure to ionizing radiation, one-electron photooxidation, and Fenton-type reactions, and its yield was similar or somewhat lower than that of 8-oxo-dG.55,6871

Figure 3.

Figure 3

Oxidation pathways of 5-mdC. (a) ROS-induced oxidation of 5-mdC; (b) TET-mediated oxidation of 5-mdC.

2.1.2. Tandem Lesions

Apart from the above-mentioned single-nucleobase lesions, ROS may also induce the formation of bulky DNA lesions. In this context, exposure to ROS from a variety of experimental systems was found to induce CC → TT and mCG → TT tandem base substitutions, suggesting that ROS may induce the formation of intrastrand cross-link lesions.7275 In addition, Randerath et al.,76,77 by using thin-layer chromatography (TLC) analysis of 32P-postlabeled DNA digestion products, demonstrated the existence of I (indigenous)-compounds, the bulky DNA modifications which increase markedly with aging in tissues of healthy laboratory animals and are derived from DNA-reactive intermediates arising from nutrient and oxygen metabolism. Especially, the type II I-compounds include several bulky DNA lesions, which are enhanced in kidney DNA of rodents treated with pro-oxidant carcinogen ferric nitrilotriacetate (Fe-NTA) and are identical to these lesions generated in DNA or oligodeoxyribonucleotides (ODNs) treated with Fenton reagents in vitro.76

Later, a modified 32P-postlabeling assay demonstrated that four type II I-compounds in mammalian tissue DNA are dinucleotides containing the bulky 5′S diastereomer of 8,5′-cyclo-2′-deoxyadenosine (cdA) as the 3′ nucleoside.78 The dinucleotides arise from the incomplete hydrolysis of the phosphodiester bond on the 5′ side of the modified nucleoside with the enzymes used in the 32P-postlabeling assay.78

The formation of purine cyclonucleosides (cPus) was proposed to arise from a single hydroxyl radical attack via a two-step mechanism (Figure 4).79,80 In this respect, the hydroxyl radical abstracts a hydrogen atom from the C5′ of 2-deoxyribose, yielding a carbon-centered radical, which attacks the C8 of adenine or guanine to form a new C–C bond. The resulting conjugate can lose an electron and a proton to give cdA and cdG. Molecular oxygen can inhibit this reaction by directly reacting with the C5′ radical, thereby preventing intramolecular cyclization.81,82 The above cyclization reaction yields two diastereomers at similar frequencies in calf thymus DNA exposed to ionizing radiation under anaerobic conditions.83 However, the 5′R diastereomers of cdA and cdG were induced in calf thymus DNA by Fenton-type reagents at markedly higher levels than the 5′S counterparts, with cdG being produced at a higher yield than cdA.84 Moreover, the cPus could be detected at appreciable levels in cells and animal tissues.23,8591

Figure 4.

Figure 4

Proposed mechanism for the formation of 5′R_- and 5′_S diastereomers of 8,5′-cyclo-2′-deoxyadenosine.

The ROS-induced tandem DNA lesions with the adjacent nucleobases in the same DNA strand being covalently bonded were also investigated in the past few decades. Earlier studies by Box et al.9295 showed the formation of intrastrand nucleobase–nucleobase cross-link lesions with guanine being covalently bonded with its adjacent thymine or cytosine when aqueous solutions of synthetic di- or tetranucleotides were exposed to X or γ rays under anaerobic conditions.

By introducing a photolabile precursor of the 5-methyl radical of thymine and 5-methylcytosine in synthetic dinucleoside monophosphates and ODNs, it was later found that the 5-methyl radical of the two pyrimidine bases can couple with the C8 position of its neighboring guanine and/or adenine to yield intrastrand cross-link lesions.9699 Additionally, Zhang et al.100,101 revealed that an independently generated 5-hydroxy-5,6-dihydrothymidin-6-yl radical can conjugate with the C8 of guanine to give an intrastrand cross-link lesion in dinucleoside monophosphates and ODNs, though the formation of this lesion in duplex DNA was minimal. These studies provided important mechanistic insights into the ROS-induced formation of the intrastrand cross-link lesions. Furthermore, Zeng et al.102104 and Hong et al.105 found that the UVB irradiation of duplex DNA containing a site-specifically inserted 5-bromocytosine or 5-bromouracil could give rise to efficient formation of intrastrand cross-link products with the C5 position of the pyrimidine base being covalently bonded with the C8 or _N_2 position of its neighboring guanine or with the C2, _N_6, and C8 position of its adjacent adenine. Together, these photochemical approaches offered a facile synthetic route for the generation of ODNs harboring site-specifically inserted and structurally defined intrastrand cross-link lesions, which are necessary for the characterizations of the repair of these lesions as well as their impact on DNA replication and transcription.

Further investigations were conducted about the formation of nucleobase–nucleobase intrastrand cross-links in vitro and in vivo. Along this line, Gu et al.106 showed that exposure of synthetic duplex DNA with γ rays under anaerobic conditions could give rise to the formation of the d(G[8–5]C) intrastrand cross-link. In addition, Zhang et al.98 observed that treatment of d(5mCG) with γ rays under anoxic conditions could lead to the formation of the d(5mC[5m-8]G) intrastrand cross-link. By using LC-MS/MS, Hong et al.71 further observed a dose dependent induction of d(G[8–5m]T) in calf thymus DNA upon treatment with the Fenton reagent, with a yield that is 2–3 orders of magnitude lower than that of common single-nucleobase lesions like 8-oxo-dG,5-hmdU and 5-fdU. Moreover, d(G[8–5]C) and d(G[8–5m]T) (Figure 5) could be detected in HeLa S3 cells upon exposure to γ rays, and the yields for these two lesions increase with the dose of γ rays.107 It was also observed that d(G[8–5m]5mC) (Figure 5)was formed at a higher yield than d(5mC[5m-8]G) in synthetic double-stranded DNA upon treatment with Fenton-type reagents.108 In addition to the above-mentioned intrastrand cross-links involving two adjacent nucleobases, Crean et al.109 demonstrated the induction of a nonadjacent intrastrand cross-link lesion between guanine and thymine bases separated by a cytosine in the single-stranded 5′-d(GpCpT)-3′ ODN exposed to a CO3•− radical. Similarly, the generation of nonadjacent and adjacent cross-link lesions between the C8 of guanine and the _N_3 of thymidine (d(G[8-_N_3]T)) (Figure 5) was observed in 5′-d(GpT)-3′ and 5′-d(GpCpT)-3′ ODN or calf thymus DNA treated with peroxynitrite/carbon dioxide/bicarbonate, in addition to the nitration/oxidation products of guanine such as 8-nitro-2′-deoxyguanosine (8-nitro-dG), 5-guanidino-4-nitroimidazole 2′-deoxynucleoside (dNIm), 8-oxo-dG, and dSp.110 Further study also demonstrated the formation of these two d(G[8-_N_3]T) lesions in HeLa cells upon one-electron oxidation initiated by intense nanosecond 266 nm laser irradiation.111

Figure 5.

Figure 5

Structures of ROS-induced intrastrand cross-link lesions discussed in this review.

2.2. Inflammation-Induced DNA Damage

Chronic inflammation is an established risk factor for different types of cancers.112,113 Inflammatory responses protect human bodies from adverse effects inflicted by pathogens and damaged cells through the generation of reactive oxygen, nitrogen, and halogen species.12,114 These reactive chemical entities damage proteins and DNA of invaders as well as nearby healthy cells and tissues.12,112114 The heme enzyme myeloperoxidase, which is secreted by activated neutrophils and monocytes, employs hydrogen peroxide (H2O2) and chloride ion as substrates to yield hypochlorous acid (HOCl) as the initial product (Figure 6).115,116 Likewise, eosinophil peroxidase, a structurally related heme protein released by activated eosinophils, preferentially oxidizes bromide to give hypobromous acid (HOBr, Figure 6).117 In addition, myeloperoxidase-induced production of HOCl is also involved in the formation of brominating species (Figure 6).117,118

Figure 6.

Figure 6

Inflammation-induced formation of hypochlorous acid (HOCl) and hypobromous acid (HOBR) and major nucleobase halogenation products. Myeloperoxidase can induce the formation of both HOCl and HOBr. The mechanism for the formation of HOBr by myeloperoxidase is different from that by eosinophil peroxidase.

HOCl and HOBr may lead to mutagenesis by damaging the nucleotide pool or reacting directly with DNA. In this vein, reactions of HOCl and HOBr with uracil produce 5-chlorouracil (5-ClU) and 5-bromouracil (5-BrU, Figure 6),117,118 respectively, which can be further converted to 5-chloro-2′-deoxyuridine (5-Cl-dU) and 5-bromo-2′-deoxyuridine (5-Br-dU) by thymidine phosphorylase.119,120 Both 5-Cl-dU and 5-Br-dU are dT analogues, and they can be converted to their corresponding nucleoside triphosphates and incorporated into DNA.121,122 In addition, bromination of 2′-deoxycytidine leads to the formation of 5-bromo-2′-deoxycytidine (5-Br-dC, Figure 6), which can undergo deamination before being incorporated into DNA as 5-Br-dU.117,118 Major products arising from the reaction of HOCl with DNA include 5-chloro-2′-deoxycytidine (5-Cl-dC), 5-Cl-dU, 8-chloro-2′-deoxyguanosine (8-Cl-dG), and 8-chloro-2′-deoxyadenosine (8-Cl-dA).118,123127 Reaction of HOBr with DNA can lead to the formation of 8-bromo-2′-deoxyguanosine (8-Br-dG), 8-bromo-2′-deoxyadenosine (8-Br-dA), and 5-Br-dU (Figure 6).118,128130 Among these halogenated nucleosides, 5-Cl-dC has been the most extensively studied and is considered a biomarker for chronic inflammation.131,132

During immune response, activated neutrophils and macrophages can also secrete other reactive chemical species, such as nitric oxide (•NO), which can further react with superoxide (O2−•), leading to the formation of peroxynitrite (ONOO−).12,133136 Peroxynitrite is highly reactive toward DNA and may contribute to the cytotoxicity and carcinogenesis associated with excess generation of •NO and O2−• during chronic inflammation.137 ONOO− was found to react preferentially with dG, at a reaction rate that is at least 9 times higher than that of dA, dC, and dT in the nucleoside form.138 Along this line, previous in vitro experiments demonstrated that ONOO− induced the formation of dNIm (Figure 1) in synthesized ODNs and calf thymus DNA.110,139 Other in vitro studies also illustrated that the nitrogen dioxide radical (•NO2), produced from photolysis of nitrate with 308 nm nanosecond XeCl laser, could react with guanine neutral radicals (G(-H)•) in aqueous solution of ODNs and calf thymus DNA, leading to the formation of dNIm.140

2.3. DNA Damage Formed from Byproducts of Lipid Peroxidation

ROS can also attack biomolecules other than DNA. Specifically, the hydroxyl radical initiates the peroxidative degradation of lipids by abstracting a hydrogen atom from polyunsaturated fatty acids (PUFA).13,141148 The resulting lipid radical (L•)is first converted to a lipid peroxyl radical (LOO•) in the presence of O2, leading to the formation of a lipid hydroperoxide (LOOH) via hydrogen atom abstraction, and finally to an alkoxyl radical (LO•) by the transition metal ion-catalyzed Fenton-type reaction. Further fragmentations of peroxyl and alkoxyl radicals give rise to reactive aldehydes, including malondialdehyde (MDA), acrolein, crotonaldehyde, 2-hexenal, 4-hydroxy-2-hexenal (HHE), 4-hydroxy-2-nonenal (HNE), 4-oxo-2-nonenal (ONE), 4-hydroperoxy-(2_E_)-nonenal (HPNE), 9,12-dioxo-(10_E_)-dodecenoic acid (DODE), 5,8-dioxo-(10_E_)-octenoic acid (DOOE), 2,4-decadienal (DDE), 4,5-epoxy-(2_E_)-decenal (EDE), etc. (Figure 7)13,14,148

Figure 7.

Figure 7

Structures of representative lipid peroxidation (LPO) byproducts and DNA adducts induced by these byproducts.

The aldehydes formed from lipid peroxidation can react with DNA to generate a variety of DNA adducts (Figure 7).13,14,148 In this vein, malondialdehyde reacts with guanine, adenine, and cytosine in DNA to form exocyclic pyrimido-[1,2-α_]purine-10(3_H)-one-2′-deoxyribose (M1dG), linear _N_6-(3-oxopropenyl)-2′-deoxyadenosine (M1dA), and _N_4-(3-oxopropenyl)-2′-deoxycytidine (M1dC), respectively (Figure 7).13 The α,_β_-unsaturated LPO products, such as acrolein, crotonaldehyde, and HNE, can lead to the formation of different diastereomers of exocyclic six-membered ring propano adducts in DNA with or without substituted alkyl side chains. The α,_β_-unsaturated aldehydes can undergo Michael addition with the _N_2-amino group of dG to give _N_2-(3-oxopropyl)-dG adducts, followed by cyclization of _N_1 with the aldehyde moiety to generate the corresponding exocyclic 1,_N_2-propano-dG products (Acr-dG, Cro-dG, and HNE-dG, etc.).149 This cyclization is reversible, and the ensuing release of the aldehyde functionality can induce the generation of DNA interstrand cross-links and DNA–protein cross-links.14,150,151 Along this line, it is worth noting that the unsubstituted Cro-dG can also arise from consecutive reactions of guanine with two molecules of acetaldehyde.152

Another type of LPO-induced DNA adduct, etheno adduct, possesses unsaturated five-membered exocyclic rings fused with heterocyclic nucleobases. Previous studies proposed a putative mechanism for the formation of etheno adducts from HNE-derived epoxide intermediate, where the α,_β_-unsaturated HNE can be converted to reactive intermediate 2,3-epoxy-4-hydroxynonanal by auto-oxidation or by oxidation with H2O2.143,144,153,154 The resulting reactive epoxy aldehyde can further react with DNA to yield the etheno adducts with or without substituted alkyl side chains, such as 1,_N_6-etheno-2′-deoxyadenosine (_ε_dA), 3,_N_4-etheno-2′-deoxycytidine (_ε_dC), 1,_N_2-etheno-2′-deoxyguanosine (1,_N_2-_ε_dG), and _N_2,3-etheno-2′-deoxyguanosine (_N_2,3-_ε_dG), etc.143,144,153,154 _N_2,3-_ε_dG can also be induced in DNA from exposure to carcinogen vinyl chloride via a similar epoxide initiation mechanism.155 Additionally, Lee et al.156,157 demonstrated possible HNE-independent pathways for the formation of 1,_N_2-_ε_dG initiated from either another LPO product EDE or HNE precursor HPNE, suggesting the existence of probable controversies concerning the mechanism for the formation of etheno DNA adducts in vivo. The above-mentioned DNA adducts emanating from products of LPO could be detected at appreciable levels in vivo.88,152,158174

3. Repair and Biological Consequences of Oxidative Stress-Induced DNA Lesions

To minimize mutation induction and to maintain genome integrity, cells are equipped with multiple DNA repair systems to enable efficient removal of DNA lesions from the genome. Unrepaired DNA lesions may elicit cytotoxic and mutagenic effects by perturbing the accuracy and efficiency of DNA replication and transcription. Chemical synthesis of ODNs harboring site-specifically inserted and structurally defined DNA lesions, along with in vitro biochemical assay and shuttle vector-based cellular experiments, has provided important insights into how the oxidative stress-induced DNA lesions are repaired and how they compromise the flow of genetic information by inhibiting DNA replication and transcription and inducing mutations in these processes.175,176 Tables 1 and 2 summarize the key findings made from studies about how oxidative stress-induced DNA lesions perturb the efficiencies and fidelities of DNA replication and transcription in cells. Tables 3 and 4 provide information about the detection and repair mechanisms of oxidatively induced DNA modifications.

Table 1. Summary of Findings Made from Cellular Replication Studies of Oxidative Stress-Induced DNA Lesions.

DNA lesions assay system bypass efficiency mutation type (frequency) refererces
8-oxo-dG M13 phage in E. coli 50–90% G T (∼0.5–l%) 179
M13 phage in E. coli G →T (0.72%) 178
8-oxo-dA double-stranded vector in NIH 3T3 cells A → G + A → C (∼1% in total) 180
Fapy-dG single-stranded vector in COS-7 simian kidney cells G → T (∼8–30%) 181
G → C (∼2%)
Fapy-dA single-stranded vector in COS-7 cells A → C (∼0.4%) 181
dGh single-stranded phage in wild-type AB1157 E. coli cells ∼75 ± 5% G → C (98%) 191
G → T (2%)
single-stranded M13 phage in AB1157 E. coli cells (with MufY+ or Mut Y−) ∼20% (MutY+) ∼30% (MutY−) for MutY+ E. coli 192
G → C (∼57%)
G → T (∼40%)
G → A (∼3%)
no significant changes in mutation type or frequency were found in MutY− E. coli compared to MutY+ E. coli
dSp single-stranded M13 phage in wild-type AB1157 E. coli cells stereoisomer 1: ∼9 ± 3% stereoisomer 1: 191
G → C (72%)
G → T (27%)
stereoisomer 2: ∼9 ± 4% stereoisomer 2:
G → C (57%)
G → T (41%)
single-stranded M13 phage in AB1157 E. coli cells (with MutY+ or MutY−) MutY+ strain: <20% for both stereoisomers; MutY− strain: ∼30% for both stereoisomers stereoisomer 1 (for MutY+ E. coli): G → C (∼19%) 192
G → T (∼78%)
G → A (∼1%)
stereoisomer 2 (for MutY+ E. coli):
G → C (∼48%)
G → T (∼49%)
G → A (∼3%)
no significant changes in mutation type or frequency were found in MutY− E. coli compared to MutY+ E. coli
thymidine glycol single-stranded M13 phage in E. coli lethal not detectable 202
5-hmdU _ϕ_X-174am3 phage in E. coli spheroplasts not detectable 214
5-fdU double-stranded vectors in COS-7 simian kidney cells not blocking T → G + T → A (0.01–0.04% in total) 226
5-hmdC single-stranded M13 phage in E. coli cells ∼90–110% C → G + C → T + C → A (0.17–1.12% in total) 240
double-stranded vector in HEK-293T human kidney epithelial cells ∼100% not detectable 242
5-fdC single-stranded M13 phage in E. coli cells ∼90–110% C → G + C → T + C → A (0.17–1.12% in total) 240
double-stranded vector in COS-7 cells 39–90% C → G + C → T + C → A (0.03–0.28% in total) 241
double-stranded vector in HEK-293T cells ∼70% not detectable 242
5-cadC single-stranded M13 phage in E. coli cells ∼ 90–110% C → G + C → T + C→A (0.17–1.12% in total) 240
double-stranded vector in HEK-293T cells ∼70% not detectable 242
_S_-cdA single-stranded M13 phage in E. coli ∼10–31% A → T (∼11%) 253
double-stranded vector in Pol _η_-deficient XP30RO cells and Pol η -complemented XP30RO cells Pol η -deficient XP30RO cells: ∼3%; Pol η -complemented XP30RO cells: ∼5% Pol _η_-deficient XP30RO cells: A →T (∼S%) Pol _η_-complemented XP30RO cells: A → T (∼9%) 254
_S_-cdG single-stranded plasmid in E. coli <1% without SOS induction; <5.5% with SOS induction G→A+G→T + deletion of 5′C (∼34% in total) 252
single-stranded M13 phage in E. coli ∼4–11% G→A (∼40% without SOS, ∼20% with SOS) 253
double-stranded vector in Pol η -deficient XP30RO cells and Pol η -complemented XP30RO cells Pol η -deficient XP30RO cells: ∼_2%;_ Pol η -complemented XP30RO cells: ∼4% Pol _η_-deficient XP30RO cells: G→A (∼3%) 254
G →T (∼27%);
Pol _η_-complemented XP30RO cells: G→A (∼11%)
G →T (∼32%)
d(G[8-S]C) single-stranded M13 phage in wild-type AB11S7 E. coli 20% G →T (8.7%) 107
G→C (1.2%)
d(G[8-Sm]T) single stranded pMS2 vector in E. coli (wild type and polymerase-deficient cells) without SOS induction: 1.2–25%; with SOS induction: 3.1–35% G→T (2.5% without SOS, 6.2% with SOS in wild-type cells) 258
5-Cl-dC single-stranded M13 phage in E. coli 82–102% C → T (∼5%) 264
dNIm single stranded M13mp7L2 bacteriophage genome in AB1157 E. coli without SOS induction: 7%; with SOS induction: 57% G → C (8.9%) 272
G → A (19%)
G → T (22%)
εdA single-stranded pMS2 vector in E. coli and COS-7 cells in E. coli: very limited mutations; in COS-7 cells: 275
A → G (63%)
A → T (6%)
A → C (1%)
single-stranded pMS2 vector and double-stranded pSBK vector in HeLa and HCT116 cells ssDNA in HeLa cells: 277
A → G (2%)
A → T (8%)
A → C (1%)
dsDNA in HeLa cells (leading strand):
A → G (2%)
A → T (7%)
A → C (5%)
dsDNA in HCT116 cells (leading strand):
A → G (5%)
A → C (2%)
dsDNA in HeLa cells (lagging strand):
A → G (5%)
A → T (4%)
A → C (1%)
double-stranded M13mp2S_Vori_L vectors in E. coli A → C (1.6 X (10−4) 276
A → G (2.9 X (10−4)
A → T (2.0 X 10−4)
_ε_dC single-stranded pMS2 vector in E. coli or COS-7 cells uninduced E. coli cells: C → A + C → T (2% in total) 278
SOS-induced cells: C → A + C → T (32% in total)
COS-7 cells: C → A + C → T (81% in total)
l,_N_2-_ε_dG single-stranded M13MB19 phage in uvrA− E. coli G → A (2.05%) 279
G → T (0.74%)
G → C (0.09%)
_N_2,3-_ε_dG single-stranded M13G*1 phage in E. coli G → A (0.5%) 280
_M_1G doubled stranded M13MB102 phage in wild-type LM102 cells E. coli cells 20% for the (–)-strand G → A (0.35%) 284
G → T (0.4%)
G → C (0.12%)
single-stranded pS189 vector and double-stranded M13MB102-1 vector in E. coli and COS-7 cells −1 or −2 frameshift in E. coli and COS-7 with reiterated (CpG)4 sequence (≤1% in total) G→A+G → T + G → C (≤2% in total) 286

Table 2. Summary of Findings Made from Cellular Transcription Studies of Oxidative Stress-Induced DNA Lesions.

DNA lesions assay system transcription bypass efficiency mutagenic properties refs
8-oxo-dG pBESTluc-fl luciferase reporter in E. coli no detectable pausing or arrest C → A (33%) 184,185
deletion of the first base (26%)
pcDNA3.l(+) expression vector in mouse embryonic fibroblasts Ogg+/+ cells: C → A + one-nucleotide deletion (∼2.6%) 187
_Ogg_−/− cells: C → A + one-nucleotide deletion: (∼10.8–13.9%)
5-hmdC double-stranded vector in HEK293T cells ∼90–100% not detectable 242
5-fdC double-stranded vector in HEK293T cells ∼69% G → A (∼1%) 242
5-cadC double-stranded vector in HEK293T cells ∼55% G → A (∼1%) 242
S-cdA double-stranded vector in SV40-transformed NER-deficient XP12BE cells 5′A mutation (∼30%); multiple nucleotide deletion (−7, −13, and −21 nt deletions, ∼12.5%) 255
double-stranded vectors in NER-proficient (GM00637) and NER-deficient (GM04429) human skin fibroblasts increase with time, and up to 45% after 24 h in NER-proficient cells; no significant increase in bypass efficiency with time in NER-deficient cells in NER-deficient cells: 5′A mutation (21%) 250
_S_-cdG double-stranded vectors in NER-proficient (GM00637) and NER-deficient (GM04429) human skin fibroblasts increase with time, and up to 45% after 24 h in NER-proficient cells; no significant increase in bypass efficiency with time in NER-deficient cells in NER-deficient cells: 5′A mutation (32%) 250

Table 3. Levels of Oxidative Stress-Induced DNA Lesions in Cellular and Tissue DNA.

DNA lesions DNA sources levels refs
8-oxo-G/8-oxo-dG mouse ∼2.5–4/106 nucleosides in liver; 24,25
∼3–4/106 nucleosides in kidney;
∼3/106 nucleosides in brain
mouse ∼1–3/106 nucleosides in liver 27,28
mouse ∼ 1.5/106 nucleosides in brain; 23,26
∼3–4/106 nucleosides in liver;
∼1–2/106 nucleosides in spleen
8-oxD-A/8-oxo-dA mouse ∼0.3–0.5/106 nucleosides in brain; 23,26
∼0.2–0.4/106 nucleosides in liver;
∼0.4–0.7/106 nucleosides in spleen
Fapy-G/Fapy-dG mouse ∼0.6–1.2/106 nucleosides in liver; 24,25
∼1–2/106 nucleosides in kidney;
∼0.7–1.4/106 nucleosides in brain
mouse ∼6–14/106 nucleosides in liver 28
Fapy-A/Fapy-dA mouse ∼0.2–0.7/106 nucleosides in liver; 24,25
∼0.2–1.2/106 nucleosides in kidney;
∼0.2–0.6/106 nucleosides in brain
mouse ∼1–3/106 nucleosides in liver 27,28
dGh mouse 1.08–3.14/108 nucleosides in colon; 131
3.09–7.99/108 nucleosides in liver
dSp E. coli ∼200–600/106 dG 197
mouse 0.99–4.94/108 nucleosides in colon; 131
2.74–20.8/108 nucleosides in liver
thymine glycol/thymidine glycol human urine 3.1 pmol/μmol of creatinine 293
rat urine 4.8–8.9 nmol/kg per day 293
mouse urine 8.62 nmol/kg per day 294
monkey urine 2.07 nmol/kg per day 294
human white blood cells 2.16 fmol/μg DNA (control); 2.83 fmol/μg DNA (ovarian cancer patients) 301
5-hmU/5-hmdU HeLa cells 3.0/106 nucleosides 306
WM-266–4 3.4/106 nucleosides 306
human brain 3.9/106 nucleosides 306
mouse brain 8.3/106 nucleosides 306
mouse skin (red head) 6.4/106 nucleosides 306
mouse skin (albino) 6.0/106 nucleosides 306
LEA rat 2.6/106 nucleosides in liver; 85
20.5/106 nucleosides in brain
LEC rat 3.2–8.6/106 nucleosides in liver; 85
12–28.3/106 nucleosides in brain
human brain ∼2–3/106 nucleosides (control); 332
∼3–6/106 nucleosides (XPA-deficient)
mouse (control and Ercc1_−/c)_ for control, ∼6/106 nucleosides in liver, ∼1–2/106 nucleosides in kidney and ∼10/106 nucleosides in brain; 332
For _Ercc1_−/c, -4-20/106 nucleosides in liver, ∼1–4/106 nucleosides in kidney and ∼5–10/106 nucleosides in brain
5-fU/5-fdU Hela-S3 cells (exposed with edrays) formation rate: 0.022 lesion/106 nucleosides/Gy 308
LEA rat 7/106 nucleosides in liver; 85
18.8/106 nucleosides in brain
LEC rat 7.8–31.4/106 nucleosides in liver; 85
19.7–36.3/106 nucleosides in brain
human brain ∼7–55/106 nucleosides (control); 332
−10–45/106 nucleosides (XPA-deficient)
mouse (control and _Erccl_−/c) for control, ∼10–25/106 nucleosides in liver, ∼5–20/106 nucleosides in kidney and ∼9–10/106 nucleosides in brain; 332
for Erccl_−/_c, ∼10–95/106 nucleosides in liver, ∼10–35/106 nucleosides in kidney and ∼10–12/106 nucleosides in brain
5-hmC/5-hmdC mouse ES cells 1300/106 C 57
LEA rat 339/106 nucleosides in liver; 88
619/106 nucleosides in brain
LEC rat 176/106 nucleosides in liver; 88
654/106 nucleosides in brain
human lung 0.078–0.182% dG in normal lung; 322
0.033–0.096% dG in Stage-I lung squamous cell carcinoma (SCC)
human brain 0.817–1.175% dG in normal brain; 322
0.028-0.753% dG in stage II/III astrocytomas
HeLa cells 31.2/106 nucleosides 306
WM-266–4 12.2/106 nucleosides 306
mouse ES cells 163/106 nucleosides 306
human brain 1550/106 nucleosides 306
mouse brain 560/106 nucleosides 306
mouse skin (red head) 277/106 nucleosides 306
mouse skin (albino) 217/106 nucleosides 306
5-fC/5-fdC mouse ES cells 20/106 dC 57
HeLa cells 0.67/106 nucleosides 306
WM-266–4 0.69/106 nucleosides 306
mouse ES cells 3.5/106 nucleosides 306
human brain 1.7/106 nucleosides 306
mouse brain 1.4/106 nucleosides 306
mouse skin (red head) 1.2/106 nucleosides 306
mouse skin (albino) 0.7/106 nucleosides 306
5-caC/5-cadC mouse ES cells 3/106 dC 57
HeLa cells 0.27/106 nucleosides 306
WM-266–4 0.29/106 nucleosides 306
mouse ES cells 0.83/106 nucleosides 306
human brain 0.15/106 nucleosides 306
mouse brain 0.12/106 nucleosides 306
mouse skin (red head) 0.21/106 nucleosides 306
mouse skin (albino) 0.19/106 nucleosides 306
_S_-cdA mouse ∼0.1–0.2/106 nucleosides in brain; 23
∼0.2–0.35/106 nucleosides in liver; 26
∼0.15–0.2/106 nucleosides in spleen
LEA rat 0.11/106 nucleosides in liver; 85
0.088/106 nucleosides in brain
LEC rat 0.14–0.56/106 nucleosides in liver; 85
0.08–0.26/106 nucleosides in brain
mouse (control and _Csb_−/−) for control, ∼0.05/106 nucleosides in brain, ∼0.02/106 nucleosides in kidney and ∼0.04/106 nucleosides in liver; 87
for _Csb_−/−, ∼0.09/106 nucleosides in brain, ∼0.06/106 nucleosides in kidney and ∼0.08/106 nucleosides in liver
LEA rat 1.2/106 nucleosides in liver; 88
1.54/106 nucleosides in brain
LEC rat 2.68/106 nucleosides in liver; 88
1.41/106 nucleosides in brain
mouse (control and _Erccl_−/c) for control, 0.21–0.42/106 nucleosides in liver, 0.26–1.25/106 nucleosides in kidney and 0.12–0.22/106 nucleosides in brain; 86
for _Erccl_−/c, 0.96–4.09/106 nucleosides in livers, 0.19–1.81/106 nucleosides in kidney and 0.1–0.21/106 nucleosides in brain
mouse (albino and red head) ∼0.15/106 nucleosides (albino); ∼0.35/106 nucleosides (red head); 89
_R_-cdA mouse ∼0.025–0.035/106 nucleosides in brain; 23
∼0.02–0.05/106 nucleosides in liver; 26
∼0.04-0.05/106 nucleosides in spleen
LEA rat 0.1/106 nucleosides in liver; 85
0.15/106 nucleosides in brain
LEC rat 0.18–0.48/106 nucleosides in liver; 85
0.13–0.56/106 nucleosides in brain
mouse (control and Ercc1−/c) for control, 0.17–0.95/106 nucleosides in liver, 0.2–0.65/106 nucleosides in kidney and 0.08–0.16/106 nucleosides in brain; 86
for _Ercd_−/c, 2.54–8.37/106 nucleosides in liver, 0.27–0.72/106 nucleosides in kidney and 0.09–0.17/106 nucleosides in brain
mouse skin (albino and red head) ∼0.15/106 nucleosides (albino); ∼0.3/106 nucleosides (red head); 89
_S_-cdG mouse ∼2–3.5106 nucleosides in brains 23
∼1–1.5/106 nucleosides in liver; 26
∼2.5–3.5/106 nucleosides in spleen
LEA rat 0.19/106 nucleosides in liver; 85
0.16/106 nucleosides in brain
LEC rat 0.20–1.08/106 nucleosides in liver; 85
0.14–0.43/106 nucleosides in brain
LEA rat 2.02/106 nucleosides in liver; 88
2.31/106 nucleosides in brain
LEC rat 4.45/106 nucleosides in liver; 88
2.21/106 nucleosides in brain
mouse (control and _Ercc_−/c) for control, 0.32–1.05/106 nucleosides in liver, 0.91–2.86/106 nucleosides in kidney and 0.37–0.53/106 nucleosides in brain; 86
for _Ercd_−/c, 2.03–5.64/106 nucleosides in liver, 0.53–2.81/106 nucleosides in kidney and 0.25–0.66/106 nucleosides in brain
mouse skin (albino and red head) ∼0.35/106 nucleosides (albino); ∼0.75/106 nucleosides (red head); 89
_R_-cdG mouse ∼0.5–0.8/106 nucleosides in brain; 23
∼0.5–0.55/106 nucleosides in liver; 26
∼0.5–0.6/106 nucleosides in spleen
LEA rat 0.13/106 nucleosides in liver; 85
0.14/106 nucleosides in brain
LEC rat 0.16–0.54/106 nucleosides in liver; 85
0.13–0.52/106nucleosides in brain
mouse (control and _Erccl_−/c) for control, 0.14–1.01/106 nucleosides in liver, 0.35–0.73/106 nucleosides in kidney and 0.11–0.17/106 nucleosides in brain; 86
for _Ercrt_−/c, 2.43-7.31/106 nucleosides in liver, 0.28–0.73/106 nucleosides in kidney and 0.11–0.21/106 nucleosides in brain
mouse skin (albino and red head) ∼0.15/106 nucleosides (albino); ∼0.3/106 nucleosides (red head); 89
d(G[8-5]C) Hela-S3 cells (with y irradiation) formation rate: ∼0.037 lesions/109 nucleosides per Gy 106
d(G[8-5m]T) Hela-S3 cells (with y irradiation) formation rate: 0.05 lesion/109 nucleosides/Gy 107
LEA rat ∼0.005/106 nucleosides in liver; 332
∼0.02/106 nucleosides in brain
LEC rat ∼0.01–0.04/106 nucleosides in liver; 332
∼0.01–0.08/106 nucleosides in brain
mouse liver (control and XPA-deficient) ∼0.005/106 nucleosides (control); 332
∼0.01/106 nucleosides (XPA-deficient)
human brain (control and XPA-deficient) ∼0.005–0.04/106 nucleosides (control); 332
∼0.015–0.07/106 nucleosides (XPA-deficient)
mouse (control and _Erccl_−/c) for control, ∼0.007∼0.01/106 nucleosides in liver, ∼0.015–0.02/106 nucleosides in kidney and ∼0.001–0.005/106 nucleosides in brain; 332
for _Erccl_−/c, ∼0.008–0.035/106 nucleosides in liver, ∼0.01–0.045/106 nucleosides in kidney and 0.001–0.007/106 nucleosides in brain
d(G[8-_N_3]T) HeLa cells (with 266 nm laser pulse irradiation) 0.21–1.19/106 nucleosides 111
5–Cl-C/5-Cl-dC mouse 3.86–7.06/108 nt in colon; 131
4.33–16.61/108 nt in liver
human leukocyte 0.06–0.4/106 nucleosides 127
human endothelial cell (treated with 300 _μ_M HOC1) 40/106 dC 343
human colon (from patients with inflammatory bowel disease) 0.002–0.294/106 nucleobases 132
mouse colon (_H. hepaticus-_infected) 0.031–0.129/106 nucleobases 132
5–Br-C/5-Br-dC human eosinophils −60–225/106 cells 117
5–C1-U/5-CU1U rat 20–80 pg/pouch 341
human atherosclerotic tissue ∼0.1 pmol/g tissue (normal); ∼0.7 pmol/g tissue (atherosclerotic) 342
5–Br-U/5-Br-dU human atherosclerotic tissue ∼0.08 pmol/g tissue (normal); ∼0.18 pmol/g tissue (atherosclerotic) 342
M1G/M1dG human 0.5–1.2/106 nucleosides in liver; 13
0.05–2.8/106 nucleosides in white cells;
0.001–0.5/106 nucleosides in pancreas; 0.002–0.56/106 nucleosides in breast
human leukocyte 64.9 fmol/mg DNA in smokers; 171
56.5 fmol/mg DNA in nonsmokers
rat liver 0.52/106 nucleosides 172
human colorectal mucosa 0.43/106 nucleosides for men; 4.6/106 nucleosides for women 174
ε_A/ε_dA human asymptomatic colon epithelia ∼0.025–0.065/106 nucleosides 154
rats exposed to 600 ppm vinyl chloride (4 h/day for 5 day) 0.21/106 nucleosides in liver; 155
0.65/106 nucleosides in lung;
0.04/106 nucleosides in kidney
human placenta 2.3–2.5/106 nucleosides 161
human 0.282/106 nucleosides in placenta 163
0.162/106 nucleosides in leukocyte;
human saliva 0.22–2.1/106 nucleosides 165
LEC rat liver ∼0.002–0.1/106 nucleosides 170
LEA rat 0.23/106 nucleosides in liver; 0.39/106 nucleosides in brain 88
LEC rat 0.3/106 nucleosides in liver; 0.44/106 nucleosides in brain 88
1, _N_2-_ε_G/l, _N_2-_ε_dG human IMR-90 cells ∼0.02/106 nucleosides 152
human urine 95 pg/mL for smokers;
50–68 pg/mL for nonsmokers
human 0.085/106 nucleosides in placenta 163
0.086/106 nucleosides in leukocyte;
human saliva 0.68–7.52/106 nucleosides 165
Wistar rat 2.47/108 dG in livers; 168
0.87/108 dG in lungs;
2.96/108 dG in brains
LEA rat 0.11/106 nucleosides in livers; 0.15/106 nucleosides in brains 88
LEC rat 0.16/106 nucleosides in livers; 0.13/106 nucleosides in brains 88
_ε_C/_ε_dC human asymptomatic colon epithelia ∼0.015–0.035/106 nucleosides 154
rats exposed to 600 ppm vinyl chloride (4 h/day for 5 days) 0.98/106 nucleosides in liver; 155
0.3/106 nucleosides in lung;
0.29/106 nucleosides in kidney
human 0.441/106 nucleosides in placenta 163
0.111/106 nucleosides in leukocyte;
human urine 104–105 pg/mL 164
human saliva 0 –1.39/106 nucleosides 165
human urine 0.45 nM for smokers; 166
0.16 nM for nonsmokers
LEC rat liver ∼0.035–0.25/106 nucleosides 170
_N_2,3-_ε_G/_N_2,3-_ε_dG rats exposed to 600 ppm vinyl chloride (4 h/day for 5 days) 1.81/106 nucleosides in liver: 155
0.21/106 nucleosides in lung;
0.31/106 nucleosides in kidney
Acr-dG human 0.78/106 nucleosides in leukocyte; 159
1.08/106 nucleosides in placenta
human saliva 0.13/106 nucleosides 165
human brain 5.15/106 nucleosides (with Alzheimer's disease); 336
2.8/106 nucleosides (control)
Cro-dG human IMR-90 cells ∼0.05/106 nucleosides 152
human 0.06/106 nucleosides in leukocyte; 159
0.26/106 nucleosides in placenta
human saliva 0–0.485/106 nucleosides 165
human 14.57/109 dG in livers; 19.99/109 dG in lung 167
Wistar rat 4.61/108 dG in liver; 168
2.25/108 dG in lung;
5.66/108 dG in brain

Table 4. Repair Mechanisms of Oxidative Stress-Induced DNA Lesions.

DNA lesions/modifications repair pathways refs
8-oxo-dG BER 25,27,188190
8-oxo-dA BER 25,27,188190
Fapy-dG BER 25,27,188190
Fapy-dA BER 25,27,188190
dGh BER 193,195198
dSp BER 193,195198
thymidine glycol BER 206,207,210
NER 211
5-hmdU BER 188,219,220
5-fdU BER 228,230236
5-fdC BER 65,220,246,247
5-cadC BER 65,220,246,247
R/_S_-cdA NER 86,91,249,250
R/_S_-cdG NER 86,91
d(G[8-5]C) NER 257
d(G[8-5m]T) NER 256,257,332
d(G[8-_N_3]T) NER 259
BER 260
8-Cl-dG BER 262
5-Cl-dU BER 263
dNIm BER 199
_ε_dA AlkB/ALKBH 281,282
BER 281
_ε_dC AlkB/ALKBH 281,282
BER 281
1,_N_2-_ε_G/1,_N_2-_ε_dG AlkB/ALKBH 281,282
BER 281
_N_2,3-_ε_G/_N_2,3-_ε_dG AlkB/ALKBH 282
281
BER 281
M1dG AlkB/ALKBH 287
NER 14
Acr-dG AlkB/ALKBH 287
NER 14
Cro-dG NER 14

3.1. Direct ROS-Induced DNA Lesions

3.1.1. Single-Nucleobase Lesions

It has been illustrated that owing to the 8-oxo-dG:dA mispairing, replicative bypass of 8-oxo-dG leads to G → T transversion and that misincorporation of 8-oxodGTP formed in the nucleotide pool into DNA gives rise to A → C substitutions.177179 8-oxo-dA induces A → G transitions and A → C transversions in mammalian cells.180 Fapy-dG can mispair with dA, which can induce G → T transversion (∼8–30% frequency).181,182 Fapy-dA is weakly mutagenic (0.4% frequency) and induces A → C transversion in vivo.181

In vitro experiments showed that 8-oxo-dG only slightly perturbed T7 RNA polymerase (T7RNAP)-mediated transcription, with bypass efficiency being up to 95%.183 E. coli RNA polymerase can efficiently bypass 8-oxo-dG in vivo, where the lesion induces C → A transversion and single-nucleotide deletion at frequencies of 33% and 26%, respectively.184,185 In addition, this lesion transiently paused transcription mediated by mammalian RNA polymerase II and led to a similar C → A transversion mutation,183,186 and Saxowsky et al.187 found that 8-oxo-dG induced one-nucleotide deletion and C → A transversion in transcripts in mouse embryonic fibroblasts (MEFs). Moreover, elevated levels of mutant transcripts were generated in the _Ogg1_−/− MEFs and _Csb_−/−_Ogg1_−/− MEFs, indicating the involvement of transcription-coupled repair and DNA glycosylase in the removal of 8-oxo-dG.187 The major known proteins and pathways for the repair of the aforementioned oxidatively induced nucleobase lesions were previously reviewed.188190

A number of replication and repair studies have been conducted for dGh and dSp. These two lesions strongly block DNA polymerases, and once bypassed, they can be highly mutagenic and yield G → C and G → T transversions,191,192 and the frequencies of the dGh-induced mutations are pH-dependent.39 In addition, dGh and dSp adducts were found to be substrates for both BER and NER pathways. While neither lesion could be repaired by human hOGG1,193,194 both were found to be substrates for E. coli DNA glycosylases MutM, Nth and Nei,195197 yeast yOGG1, yOGG2,193 murine NEIL1 and NEIL2,197 and human hNEIL1,198 where the repair mediated by hNEIL1 seems to be stereoselective.198 McKibbin et al.194 demonstrated the excision of dGh, dSp, and dSp-amine adducts by Nei, Fpg, NEIL1 (BER glycosylases), as well as the UvrABC system (bacterial NER pathway). It is worth noting that the authors demonstrated that bulky dSp-amine adducts (including dSp-Lys, dSp-GlcN, and dSp-GPRPGP) can be repaired by the BER pathway, indicating the presence of overlapping mechanisms for the removal of hydantoin and hydantoin-amine adducts.194 Shafirovich et al.199 also observed the formation of both BER and NER products when the dGh- and dSp-containing ODNs were treated with human cell extracts. The involvement of NER was further substantiated by the observation that the lack of XPA or XPC diminished the repair activity, whereas complementation of extracts of XPC-deficient cells with XPCRAD23B restored the NER activity.199 Comprehensive investigations into how BER and NER pathways are involved in the removal of dGh and dSp lesions in vivo will paint a more complete picture about the repair of these lesions.

Thymidine glycol effectively blocks DNA polymerases in vitro,200 while resulting in cell death in vivo.189,201 Although generally regarded as not highly mutagenic on its own,202 thymidine glycol was found to modulate the mutagenic properties of other closely placed DNA lesions. One example is that thymine glycol can form as part of clustered DNA damage with a neighboring 8-oxo-dG, where the presence of a neighboring thymidine glycol significantly increased the mutagenic potential of 8-oxo-dG.203 As noted elsewhere in this review, thymidine glycol can also arise from the deamination of 5-methyl-2′-deoxycytidine glycol;46,47 thus, the thymidine glycol/8-oxo-dG tandem lesion may be induced at methylated CpG sites upon ROS attack, thus contributing to CpG mutagenesis.47,48 When thymidine glycol is located opposite to an apurinic/apyrimidinic (AP) site, DNA double-strand breaks (DSBs) can be formed through a BER mechanism or at replication fork.204 When thymidine glycol is opposite to 8-oxo-dG, the DSB formation decreased, but mutation frequency of 8-oxo-dG increased compared to that found for 8-oxo-dG present as an isolated lesion.204 In addition, Almohaini and co-workers205 reported that the presence of a thymidine glycol at the first or second position from one 3′ terminus of a blunt-end DSB significantly impeded nonhomologous end-joining, while BER of thymidine glycol located at the fifth position from the blunt end interfered with the DSB ligation.

Thymidine glycol is mainly repaired via the BER pathway, and the lesion was found to be a substrate for a number of DNA glycosylases. These include endonucleases III (Endo III; Nth) and VIII (Endo VIII; Nei) in E. coli, yNTG1 (Ntg1) and yNTG2 (Ntg2) in S. cerevisiae, as well as NTH1, NEIL1, NEIL2, and NEIL3 in mammalian cells.189,201,206 In this context, it is worth noting that ADAR1-dependent adenosine-to-inosine editing of pre-mRNA of human NEIL1 gene yields a different form of NEIL1 protein with lysine 242 being converted to an arginine, and the edited form exhibits differential activity from its corresponding unedited form toward the removal of thymine glycol in duplex DNA.207 Both forms of NEIL1 can promote tautomerization of thymine glycol, thereby facilitating the recognition and removal of the lesion.208 Besides BER, some NER activity was also observed for the removal of thymidine glycol in E. coli and human systems,209212 though the lesion is unlikely a substrate for the mismatch repair pathway in E. coli.213 5-hmdU formed from dT oxidation pairs with dA, while 5-hmdU produced by oxidation and deamination of 5-mdC pairs with dG. Earlier studies of 5-hmdU focused on its role as an oxidatively induced DNA lesion from dT and revealed that the modified nucleoside is weakly mutagenic,214 does not block DNA polymerases,215 pairs with dA in Watson–Crick geometry,216 and even replaces thymidine in bacteriophage DNA.217 The excision activity of 5-hmU in 5-hmU:A pair was much lower than that in 5-hmU:G pair by human cell extracts.218 5-hmU:G can be excised by TDG, SMUG1, MBD4,219222 and with some weak activity by NEIL1.221 SMUG1 was also found to remove 5-hmU:A, albeit with lower activity.219 These attributes of 5-hmdU are in agreement with the recent hypothesis that this modified nucleoside may assume an epigenetic role.67,223 The occurrence, repair, and biological consequences of 5-hmdU were previously reviewed.189,224

5-fdU does not strongly block DNA polymerases.225 Although high-fidelity DNA polymerases can incorporate any of the four dNTPs opposite the lesion in vitro, 5-fdU is weakly mutagenic (0.01–0.04% mutation frequencies in double-stranded vectors), and it induces T → G and T → A transversions in simian COS-7 cells.226 5-fdU is predominantly repaired by the BER pathway. AlkA in E. coli227229 and SMUG1 in mammalian cells230232 are the major enzymes for the removal of 5-fdU from DNA. Nth, Fpg, and Nei in E. coli, SpNth1 in Schizosaccharomyces pombe (a homologue of E. coli endonuclease III), human Nth1 and Mbd4, and mouse Nth1 and Tdg can also excise 5-fdU from DNA.230,233236 Similar to 5-hmU, 5-fdU in DNA exists as 5-fU:A (formed from A:T base pair) or 5-fU:G (formed from 5mC:G base pair). The former form can be removed by E. coli AlkA protein,227,228 and the latter may be repaired by AlkA protein and a MutHLS mismatch repair system (e.g., E. coli mismatch uracil DNA glycosylase, Mug),229,233 suggesting the involvement of the MMR pathway in the repair of 5-fdU. Meanwhile, repair initiated by SMUG1 can excise 5-fdU opposite any of the four nucleobases, with the highest activity toward 5-fU:C and 5-fU:T.232 The subsequent repair process will lead to T → G and T → A transversions, which are in agreement with the T → G and T → A transversions induced by this lesion in mammalian cells. In addition, KsgA was recently reported to remove the 5-fU:C mispair in E. coli, and mutation in ksgA resulted in increased spontaneous mutations in the mutM mutY and nth nei background.237 The repair of 5-fdU was also found to be modulated by a nearby apurinic/apyrimidinic (AP) site. When an AP site is located directly opposite 5-fdU, it is repaired through the long-patch BER pathway; by contrast, when the AP site is shifted it is primarily repaired by the short-patch BER pathway.238

Previous studies have provided significant insights about how the oxidized 5-mdC derivatives influence the efficiency and accuracy of DNA replication and transcription. An in vitro mutagenesis assay illustrated that, among the oxidized 5-mdC derivatives, only 5-fdC is marginally mutagenic, leading to 1–2% C → T transitions.239 5-hmdC, 5-fdC, and 5-cadC are slightly mutagenic in E. coli cells, where the C → T transition mutation occurs at frequencies of 0.17–1.12%.240 In this vein, 5-fdC was found to block DNA replication and lead to mutation in simian COS-7 cells, with bypass efficiencies and mutation frequencies being 39–90% and 0.03–0.28%, respectively.241 Moreover, Ji et al.242 demonstrated that, in HEK293T human embryonic kidney epithelial cells, 5-fdC and 5-cadC constituted modest blocks to DNA replication (with a 30% reduction in bypass efficiencies) without inducing detectable mutations in human cells, whereas replicative bypass of 5-hmdC is highly accurate and efficient.

An in vitro assay demonstrated that the yeast and mammalian RNA polymerase II (Pol II)-mediated polymerization rates and specificity constants for GTP incorporation against 5-fC and 5-caC were reduced significantly compared with those for unmodified C template, whereas no changes were observed for 5-mC and 5-hmC templates.243 Additionally, the substrate specificity was reduced by ∼30-fold for the 5-fC-containing template in comparison with the C template.243 Later, You et al.242 revealed that 5-fC and 5-caC displayed marginal mutagenic (∼0.7–1.7%) and modest inhibitory (31–50%) effects on transcription mediated by T7RNAP or human RNA polymerase II (hRNAPII) in vitro and in HEK293T cells. In addition, 5-hmC did not compromise appreciably the efficiency or accuracy of transcription in vitro or in HEK293T cells.242 The lack of pronounced deleterious effects of 5-hmC, 5-fC, and 5-caC on replication or transcription is in keeping with the potential roles of these oxidized 5-mC derivatives in epigenetic regulation.

Different from the passive DNA cytosine demethylation where 5-mdC is diluted during replication,244,245 it has been proposed that the TET-mediated oxidation of 5-mdC to 5-fdC and 5-cadC may play an important role in active cytosine demethylation in mammals, a process that results in the loss of 5-mdC independent of DNA replication. Along this line, the excision of 5-fdC and 5-cadC from DNA by thymine DNA glycosylase (TDG) and the following action through the BER pathway may result in the restoration with unmodified cytosine.59,246 It was found that genetic depletion or catalytic inactivation of TDG leads to embryonic lethality in mice,220,247 indicating the significant roles of TDG in maintaining epigenetic stability during embryonic development. An alternative active cytosine demethylation pathway was also proposed: The AID (activation-induced cytidine deaminase)/APOBEC (apolipoprotein B mRNA editing enzyme, catalytic polypeptide-like) proteins can deaminate 5-hmdC to give 5-hmdU (Figure 3b), which is then removed by TDG or SMUG1 (single-stranded-selective monofunctional uracil DNA glycosylase 1) and finally restored to unmodified cytosine through BER.65,220 Nevertheless, this mechanism is still controversial because the purified AID/APOBEC proteins exhibit no detectable activity toward 5-hmC deamination in vitro.248

3.1.2. Tandem Lesions

Multiple lines of evidence support that cdA and cdG are repaired by the NER pathway. First, cleavage products of cdA-containing duplex DNA emanating from NER activity in human cell nuclear extracts was observed, and depletion of XPA protein led to a pronounced decrease in such cleavage.249 By contrast, no correction of the lesion by direct damage reversal or base excision repair was detected.249 Second, genetic depletion of ERCC1 gave rise to elevated accumulation of cdA and cdG in mouse tissues.86,91 Third, a transcriptional bypass assay revealed that cdA and cdG are substrates for the transcription-coupled NER pathway in mammalian cells.250

cdA and cdG were found to be strong blockades to DNA replication, and replicative bypass of these lesions are highly mutagenic. With the use of DNA containing a site-specifically inserted cdA as template, both the 5′_R_- and 5′_S_-diastereomers of cdA were observed to block primer extension by replicative polymerases, including human DNA polymerase δ and T7 DNA polymerase.249,251 _S_-cdA and _S_-cdG were strongly blocking to DNA replication in E. coli and human cells and induced substantial frequencies of mutations at the lesion sites, where cdA induces A → T mutation, and cdG induces mainly G → A and G → T mutations.252254 Additionally, Pol η, Pol ι, and Pol ζ, but not Pol κ, assume crucial roles in promoting replication across _S_-cdA and _S_-cdG in human cells,254 and Pol V plays a major role in bypassing these lesions in E. coli252,253

It was found that both _S_-cdA and _S_-cdG strongly inhibited transcription and induced transcriptional mutagenesis in vitro and in mammalian cells.250,255 Different from the observations made from replication studies, _S_-cdA and _S_-cdG primarily induce a 5′-A mutation during transcription in human cells, where the lesions direct human RNA polymerase II to misincorporate an adenosine nucleotide opposite the template base situated on the 5′ side of the lesions. Furthermore, when placed on the template strand of an actively transcribed gene, both _S_-cdA and _S_-cdG were primarily repaired by transcription-coupled NER pathways in mammalian cells.250

For the oxidatively induced intrastrand nucleobase-nucleobase cross-link lesions, d(G[8–5]C), d(G[8–5m]mC), and d(G[8–5m]T) could be recognized by E. coli UvrABC nuclease, suggesting the possible involvement of the NER pathway in the repair of these lesions in vivo.256,257 S. cerevisiae DNA polymerase η (Pol η) was able to bypass d(G[8–5]C) and d(G[8–5m]T), with the 5′ guanine portion of the lesion markedly reducing the efficiency and the fidelity of nucleotide incorporation.106,107 In line with in vitro replicative bypass studies, d(G[8–5]C) was found to block considerably DNA replication in E. coli cells, as reflected by a 20% bypass efficiency, and the lesion was significantly mutagenic in vivo, inducing G → T (8.7%) and G → C (1.2%) transversion mutations. In addition, Pol V was found to be responsible for the error-prone bypass of d(G[8–5]C).107 Similar observations were made for d(G[8–5m]T), where the guanine portion of the lesion induces G → T mutation, and among the three SOS-inducible DNA polymerases in E. coli, Pol V is the most efficient in bypassing the lesion, and it is required for most targeted G → T transversions.258 In vitro studies performed in HeLa cell extracts have demonstrated that the d(G[8-_N_3]T) lesions could be potential substrates of both NER and BER pathways.259,260 Another recent primer extension experiment revealed that the d(G[8-_N_3]T) lesions could strongly block the A-family BF polymerase from Bacillus stearothermophilus, Y-family polymerases Dpo4 from Sulfolobus sulfataricus P2 and human Pol κ, with bypass efficiencies being <1–2%, ∼8%, and 9–11%, respectively.261 In addition, the primer extension catalyzed by Pol η was also partially inhibited (with bypass efficiency being 28–45%) by the d(G[8-_N_3]T) cross-links, and more efficient bypass of nonadjacent d(G[8-_N_3]T) lesions in the GCT sequence context was observed than the adjacent counterpart in the GT sequence context.261,243

3.2. Inflammation-Induced DNA Damage

To date, not much is known about the repair of halogenated nucleobases. Notwithstanding the activity of the human MutT homologue (hMTH1) toward 8-Cl-dGTP,262 human 8-oxoguanine DNA glycosylase 1 (hOGG1), endonuclease VIII-like 1 (hNEIL1), alkyladenine DNA glycosylase (hAAG), E. coli formamidopyrimidine DNA glycosylase (FPG), or endonuclease V (EndoV) could not cleave 8-Cl-G when paired with a C in duplex DNA.130

5-Cl-dU can be incorporated into DNA as a thymidine analogue. It is more readily cleaved by hSMUG1 when paired with dG than dA.263 Relative to the T:G mispair, 5-Cl-U, when mispaired with G, is more efficiently repaired by TDG and hSMUG1.263 No specific repair pathway has yet been established for 5-Cl-C. On the other hand, DNA glycosylase MBD4-mediated excision repair of 5-Cl-U, 5-Br-U, and 5-Br-C within CpG and mCpG sites has been demonstrated.264,265 The lack of specific repair pathway may account for the accumulation of halogenated nucleobases in DNA.

Base-pairing energy of 5-Cl-C:G is only slightly lower than that of the C:G pair,266 suggesting that 5-Cl-C is likely a persistent DNA lesion. It was recently reported that all families of DNA polymerases predominantly decode 5-Cl-C as C in vitro; meanwhile, 5-Cl-C, when placed on single-stranded M13 plasmid and replicated in E. coli cells, induced 3–9% C → T transition, which is as mutagenic as 8-oxo-dG in similar assays.264 On the other hand, the 5-halogenated derivatives of cytosine may also perturb epigenetic signaling. In this vein, 5–Cl-C and 5–Br-C in a CpG sequence context can, similar to 5-mC, direct DNMT1-mediated maintenance DNA cytosine methylation and can bind to methyl-CpG-binding proteins.265,267,268 Such interactions may result in aberrant cytosine methylation and alteration of epigenetic signaling. Hence, 5-Cl-dC may contribute to inflammation-driven cancers through both epigenetic and mutagenic mechanisms.

The accumulation of 5-Cl-dU has been shown to cause mutations, such as T:A → C:G transition, and sister chromatid exchange.122,263,269,270 5-Cl-dU can be incorporated into DNA as a dT analogue, and it codes as a dT in an oligonucleotide template.263 Similar to 5-Cl-dU, 5-Br-dU is also a mutagenic analogue of thymidine, and it can mispair with guanine in DNA.117 Pols α, κ, and η were shown to incorporate predominantly a dG opposite 5-Br-dC, indicating that 5-Br-dC itself is not a mutagenic lesion. However, 5-Br-dC can be deaminated to 5-Br-dU and further lead to mutation.271 8-CldG is a mutagenic adduct; Sassa and co-workers130 showed that Pol α and Pol κ were slightly retarded at the 8-Cl-dG site, while Pol η readily bypassed the lesion. 8-Br-dG is a mutagenic lesion, and it may produce a broad spectrum of mutations at the site of inflammation. Pols α, κ, and η all incorporated the correct base opposite 8-Br-dA, indicating a low mutagenic potential of this lesion.271 Further investigations regarding the repair of halogenated nucleosides in DNA and the impact of these lesions on the efficiency and fidelity of DNA replication in mammalian cells are needed.

Recently, Shafirovich et al.199 revealed that dNIm was a substrate of the human BER pathway but was resistant to excision by the NER machinery when incubated with cell-free HeLa S3 cell extracts. Previous primer extension assays demonstrated that dNIm blocked significantly replication mediated by calf thymus polymerase α and E. coli polymerase I, but not human polymerase β.139 In addition, replicative bypass of dNIm by these two polymerases could induce G → T and G → C transversions.139 Along this line, by conducting cellular replication studies with the use of a single-stranded M13mp7L2 bacteriophage genome in E. coli AB1157 cells, Neeley et al.272 showed that dNIm strongly blocked DNA replication, with bypass efficiency being only (7.0 ± 1.6)% in uninduced wild-type cells. However, the bypass efficiency of dNIm markedly increased to (57 ± 1)% in SOS-induced cells. In wild-type AB1157 cells, dNIm induced (8.9 ± 0.5)% G → C mutations and roughly equal frequencies of G → A and G → T mutations, at (19 ± 2)% and (22 ± 3)%, respectively.272 Nonetheless, much lower frequencies of G → A and G → C mutations, at (13 ± 2)% and (2.5 ± 0.6)%, respectively, were found in SOS-induced cells.272 Later, Dimitri et al.273 found that dNIm displayed modest inhibitory effects, with bypass efficiency being (87 ± 5)% during transcription mediated by T7RNAP. However, dNIm strongly blocked transcription mediated by human RNA polymerase II (hRNAPII) in HeLa nuclear extract, with a bypass efficiency of (9 ± 5)%.273 Lesion bypass by T7RNAP induced base misinsertions and deletions opposite the dNIm (22% A, 13% –1 deletion, 7% > G and 1% U), while hRNAPII exhibited error-free nucleotide incorporation opposite the lesion.273

3.3. DNA Damage Formed from Byproducts of Lipid Peroxidation

Previous in vitro primer extension assays showed that both human Pol α and Pol β were primarily blocked by LPO-induced etheno DNA adduct _ε_dA with minimal extension.274 Pol η was capable of catalyzing a substantial amount of bypass across the lesion, where the polymerase incorporated all four nucleotides opposite _ε_dA with different preferences. Human Pol ι, a paralogue of Pol η, was blocked by _ε_dA with a very small amount of synthesis past _ε_dA, which results in insertion of dCMP and, to a much lesser extent, dTMP, opposite _ε_dA.274 The mutagenic potential of _ε_dA was also investigated using a single-stranded shuttle vector system in E. coli and in COS7 simian kidney cells. A nonmutagenic dTMP incorporation opposite _ε_dA was found as the nearly exclusive event in E. coli; the lesion is, however, highly mutagenic in COS7 cells, which leads to a very high frequency of A → G transition (63%), followed by A → T (6%) and A → C (1%) transversions.275 Different from what was observed in COS7 cells, _ε_dA induced all three possible base substitutions at similar frequencies (1.5–3% each) in HeLa human cervical cancer cells.276 Levine et al.277 also reported that, when placed on the leading strand, _ε_dA induces A → T (7%),A → C (5%), and A → G (2%) mutations in HeLa human cervical cancer cells and A → G (5%) and A → C (2%) mutations in HCT-116 human colorectal carcinoma cells. Similar to what was observed for _ε_dA, the mutagenic properties of _ε_dC were found to be strikingly different in E. coli and COS7 cells (2% in uninduced E. coli cells, 32% in SOS-induced E. coli cells, and 81% in COS7 cells).278 1,_N_2-_ε_dG is moderately mutagenic and directs the incorporation of the correct nucleotide (dCMP) in >80% of the replication events in E. coli.279 _N_2,3-_ε_dG specifically induces a very low frequency (0.5%) of G → A transition during DNA replication in E. coli.280 The above-mentioned etheno adducts can be repaired by multiple DNA repair pathways, including BER and AlkB/ALKBH family dioxygenases-mediated direct damage reversal, which were previously reviewed.281,282

The mutagenicity of malondialdehyde (MDA)-induced DNA adducts was measured in the lacZα forward mutation assay in E. coli.283 The most common type of mutations induced by MDA was base-pair substitution (76%), though frameshift mutations were detected in 16% of the induced mutants, and they comprised mainly single-nucleotide additions in runs of reiterated bases.283 Modified genomes containing a C opposite M1G resulted in roughly equal frequencies of G → A and G → T mutations with few G → C mutations. The (−)-strand was replicated only 20% of the time when M1dG was present. M1dG was also found to be a substrate for the NER pathway in E. coli.284 In addition, MDA-induced mutations, such as large insertions and deletions, were found after lesion-carrying shuttle vectors undergo replication in human cells.285 Furthermore, replication studies also demonstrated that M1dG can induce −1 and −2 frameshift mutations when positioned in a reiterated (CpG)4 sequence but not when positioned in a nonreiterated sequence in E. coli or COS-7 cells.286 Recently, Singh et al.287 revealed that AlkB could repair Acr-dG and M1dG in vitro, suggesting an important role for the AlkB family of dioxygenases in protecting against the deleterious biological consequences of acrolein- and MDA-induced DNA adducts in vivo.

For more detailed discussion about the mutagenic consequences, replication bypass and repair of DNA lesions induced by lipid peroxidation byproducts, the readers should consult a recent review by Minko et al.14

4. Implications of Oxidative Stress-Induced DNA Lesions in Human Diseases

The oxidative stress-induced DNA lesions may have significant impact on human health, including the natural processes of aging, neurodegeneration, and carcinogenesis. Recently, the development of LC-MS, coupled with the isotope-dilution method for the unambiguous identification and accurate quantification of multiple DNA lesions, has provided profound insights into the involvement of DNA lesions in different pathological conditions.15,16,288

4.1. Direct ROS-Induced DNA Lesions

4.1.1. Single-Nucleobase Lesions

Previous data from the investigation of Chinese and Japanese patients indicated that low BER activity arising from inactivating mutations of the NEIL1 gene may be involved in the pathogenesis of a subset of gastric cancers.289 In addition, elevated levels of Fapy-dA and Fapy-dG were observed in the liver, kidney, and brain tissues of _Neil1_−/− mice relative to the wild-type animals.24 These findings, along with the high incidence of pulmonary and hepatocellular tumors in _Nth1_−/−_Neil1_−/− mice, suggest the importance of DNA glycosylase NEIL1 in maintaining genomic stability.24

As discussed above, the dGh and dSp lesions are highly mutagenic.290 In the viewpoint that 8-oxo-dG is produced at high frequencies and that it is more readily oxidized than dG, the major oxidation products of 8-oxo-dG, i.e., dGh and dSp, may bear a significant impact on cellular functions. For example, the presence of dGh and dSp could influence the thermal stability and folding of the G-quadruplex,291 and dSp lesions could disturb the structure of duplex DNA and affect nucleosome positioning.292 dGh and dSp lesions have been detected in E. coli197 and in mice,131 though further studies are needed for systematically assessing their formation and repair in mammals.

Thymidine glycol has been suggested as a biomarker of oxidative stress and detected in urine samples of mammals.293295 Thymidine glycol in DNA inhibits the nuclease P1-mediated hydrolysis of its neighboring 3′ phosphodiester bond,296298 rendering the release of the lesion as a dinucleotide. The lesion-containing dinucleotide was thus utilized for the quantification of thymidine glycol in DNA using LC-MS/MS coupled with the stable isotope-dilution method.47,299,300 This method has been applied for probing oxidative stress in white blood cell DNA of ovarian cancer patients301 and BRCA mutation carriers302 as well as for examining the effect of smoking cessation303 and antioxidant usage304 on levels of oxidatively induced DNA damage. Rather than being highly mutagenic, thymidine glycol strongly inhibits DNA replication.203,305

As mentioned above, 5-hmdU can be produced from oxidative stress as well as epigenetic machinery (i.e., TET-mediated oxidation of thymidine). 5-hmdU has been detected in murine and human tissues,306 although the physiological implications are less well explored.

A number of studies have been conducted in measuring 5-fdU. Earlier quantification of 5-fdU in DNA was performed using GC-MS analysis of the modified nucleobase released from DNA with the use of formic acid or 70% (w/w) hydrogen fluoride in pyridine.68,69 Stable isotope dilution coupled with LC-MS/MS was later developed for the quantification of 5-fdU but with a relatively poor detection limit.307 Derivatization with Girard reagent T significantly improved the detection limit and has been applied for the detection of 5-fdU in DNA of HeLa-S3 cells308 and Trypanosoma brucei.309 5-fdU has also been detected in various tissues of LEA and LEC rats with LC-MS/MS/MS in the negative-ion mode.85 Higher levels of 5-fdU were observed in the liver of 3-month old LEC rats,85 which model Wilson disease, a disease characterized by hepatitis and hepatocellular carcinoma.310 Wilson's disease arises from mutations in ATP7B, which encodes a transporter protein required for hepatic excretion of copper ions.310 This results in the accumulation of copper ion and increased ROS production.311,312 This finding provides a vivid illustration of the role of oxidative DNA damage in transition metal-induced diseases such as Wilson's disease.

Global 5-hmC levels are lower in a variety of human cancers including breast, liver, lung, pancreatic, and prostate cancers than in normal tissues.245 One mechanism to explain this is that cancer-related gain-of-function mutations in IDH1 and IDH2 genes cause an increase in production of the oncometabolite (R)-2-hydroxyglutarate ((R)-2-HG) instead of production of the normal product 2-oxogluterate (2-OG). In addition, cancer-related mutations in two Krebs cycle genes, fumarate hydratase (FH) and succinate dehydrogenase (SDH), led to the accumulation of their substrates, fumarate and succinate, respectively. (R)-2-HG, succinate, and fumarate, which are structurally similar to 2-OG, act as competitive inhibitors of 2-OG-dependent TET activity, leading to diminished levels of 5-hmC in some tumors313321 Additionally, Jin et al.322 reported pronounced depletion of 5-hmdC in multiple human cancers in an IDH mutation-independent manner, indicating the existence of alternative mechanism(s) involved in the loss of 5-hmdC. Together, the data suggest that 5-hmdC levels may serve as a useful molecular biomarker for cancer detection and diagnosis.322

It has been shown that 5-hmC, 5-fC, and 5-caC may serve as epigenetic marks in addition to being intermediates for active cytosine demethylation in mammals.62,63,88,323,324 In this vein, 5-hmC, 5-fC, and 5-caC are recognized by some specific cellular proteins critical for chromatin remodeling and transcriptional regulation.325331 Thus, the homeostasis of these oxidized 5-mdC derivatives is crucial for maintaining normal cellular function, whereas the loss or aberrant accumulation of these epigenetic marks may lead to deleterious biological consequences and diseases. For instance, LC-MS/MS results revealed a significantly lower level of 5-hmdC in the liver of diseased LEC rats compared to that of control LEA rats, though no difference was found in the levels of 5-mdC.88 In vitro biochemical assays showed that Cu2+ ions could directly inhibit the activity of TET enzymes, suggesting that in LEC rats perturbation of 5-hmdC-mediated epigenetic signaling contributes to the etiology of Wilson's disease.88

4.1.2. Bulky Lesions

The implication of bulky DNA lesions, including cPus and the d(G[8–5m]T) intrastrand cross-link, were also investigated. Markedly elevated levels of 5′_R_- and 5′_S_-diastereomers of cdA and cdG were found in the liver of the LEC rat model of Wilson's disease85 Moreover, the levels of these lesions increased with age in the liver and brain of LEC rats, illustrating a correlation with disease progression.85 cPus also accumulate in the genomic DNA of wild-type mice with age, providing further evidence of the accumulation of endogenous lesions over the lifespan of mammals and a potential contribution of these lesions to aging.86 Mitra et al.89 detected greater levels of cPus lesions in the skin of a murine model of human melanoma. The increase in cPus levels is ultraviolet radiation-independent but pheomelanin pigment-dependent, suggesting that pigment-driven oxidative stress and the resultant oxidative DNA damage contribute to melanoma in humans even in the absence of UV exposure.89

Results from a similar LC-MS3 method revealed that the levels of d(G[8–5m]T) are elevated in tissues of LEC rats relative to LEA rats.332 Additionally, XPA-deficient human brain and mouse liver as well as various types of tissues of ERCC1-deficient mice contain higher levels of d(G[8–5m]T).332 In an earlier review, Brooks81 proposed that chemically stable bulky DNA lesions, such as cPus, may play a crucial role in neurodegeneration in XP patients.333 Since the brain is not directly exposed to sunlight, the accumulation of ROS-induced bulky DNA lesions in brains of XP patients may contribute to neuron loss in these patients. In support of this, in _Ercc1_−/Δ mice, with defective nucleotide excision repair of cPus and progressive neurodegeneration,334 the lesions accumulate more rapidly in the brain than in normal mice. However, as neurodegeneration progresses, the number of cPus decreases significantly in the brain, suggesting a direct causal role of oxidative DNA damage in the loss of neurons in the brain.86

A recent technique improvement by employing nanoflow liquid chromatography–nanoelectrospray ionization coupled with tandem mass spectrometry provided much better sensitivity in measuring cPus as well as LPO-induced _ε_dA and _ε_dG lesions in the liver and brain tissues of LEA and LEC rats.88 Simultaneous quantification of these two different types of lesions demonstrated a preferential accumulation of direct ROS-induced cPus in vivo.88 This result indicates that cPus may contribute to the etiology of oxidative stress-induced diseases to a greater extent than the etheno adducts arising from byproducts of lipid peroxidation.88

Earlier reviews discussed the application of LPO-induced DNA lesions as potential biomarkers for cancer risk assessment in humans with cancer-prone diseases such as chronic pancreatitis, ulcerative colitis, or Crohn's disease, as well as in patients with alcohol abuse-related chronic hepatitis, fatty liver, fibrosis, or cirrhosis.158,335 Results from an isotope dilution-capillary LC-MS/MS method revealed a significant increase in levels of Acr-dG in Alzheimer's disease (AD) patients compared to controls in DNA isolated from the hippocampus/parahippocampal gyrus.336 This, along with the existence of potential acrolein-derived DNA-peptide cross-links,149 provides evidence to support the correlation between acrolein-induced DNA damage and AD pathogenesis. The LPO-induced bulky exocyclic 1,_N_2-propano-dG adducts may play a significant role in driving congenital abnormalities, myelodysplasia, acute myeloid leukemia (AML), and certain solid tumors in patients suffering from Fanconi anemia (FA).337339 FA is an autosomal recessive disease caused by deficiency in the repair of DNA interstrand cross-links. As discussed previously, the DNA interstrand cross-links generated from the ring-open form of 1,_N_2-propano-dG are likely substrates of the FA repair pathway, and if left unrepaired, these cross-links impede replication and transcription, causing chromosomal breaks and translocations that can drive disease in FA patients.340

4.2. Inflammation-Induced DNA Damage

Halogenated nucleobases are commonly associated with inflammation, and 5–Cl-C is often used as a biomarker for inflammation.131,132 Various halogenation products have been detected in mammalian cells, tissues, and fluids. Chlorinated DNA lesions such as 5-Cl-dC, 8-Cl-dG, 8-Cl-dA, and chlorinated ribonucleo-sides including 5-chlorocytidine (5-Cl-rC), 8-chloroguanosine (8-Cl-rG), and 8-chloroadenosine (8-Cl-rA) have been detected with an LC-MS/MS-based method in HOCl-treated cells and freshly isolated human white blood cells. 5-Cl-dC was the predominant DNA lesion, and 8-Cl-G and 5–Cl-C were present at higher levels in RNA than DNA.127 Jiang et al.341 found, from GC-MS analysis, increased levels of 5-Cl-dU in DNA from cells treated with HOCl, as well as in exudate fluid from carrageenan-induced inflammation sites, but not in the DNA from tissues at inflammation sites in rats, possibly due to the lack of proliferation in the isolated tissues.341 5-Cl-U has been detected in neutrophil-rich inflammatory sites118 as well as aortic tissue342 with isotope dilution coupled with GC-MS methods. The 5-Cl-U level in atherosclerotic aortic tissue was reported to be 10-fold higher than that in normal tissue, implying a potential role of halogenated nucleobases in atherogenesis. 8-Br-dG and 8-Cl-dG were quantitatively measured with stable isotope-dilution coupled with the LC-MS/MS method in rat liver, rat urine, and human urine samples, and elevated levels of 8-Br-dG and 8-Cl-dG were observed in urine samples of diabetic patients compared to those in healthy controls.129 Noyon et al.343 used LC-MS/MS for the quantification of 5-Cl-dC, 5-Cl-C, and 8-Cl-G and detected the presence of 5-Cl-C (1.0 ± 0.2 nM) in healthy human plasma. 5-Cl-C and 8-Cl-G could be detected in the cytoplasmic nucleotide pool and RNA, and 5-Cl-C is present in DNA from endothelial cells treated with HOCl.343

5. Conclusions and Perspectives

In this review, we summarized the chemical formation and biological consequences of a series of oxidatively induced DNA lesions. We also discussed the involvement of these lesions in the etiology of human diseases. Over the past several decades, much has been learned about the mechanisms of formation and repair of oxidative stress-induced DNA lesions. A number of shuttle vector-based studies have provided important molecule-level information about the degrees to which some of the oxidatively induced DNA lesions perturb the transmission of genetic information by inhibiting DNA replication and transcription as well as inducing mutations during these processes (Tables 1 and 2). Nevertheless, work remains to be done on assessing how other oxidatively induced DNA lesions, particularly the secondary oxidation products of dG (i.e., the dGhand dSp adducts) as well as the halogenated nucleosides induced by inflammatory processes, alter the efficiency and fidelity of DNA replication and transcription in mammalian cells.

The available data suggest that these mutagenic DNA adducts, especially the bulky DNA lesions such as exocyclic propano adducts, cPus, and nucleobase-nucleobase intrastrand cross-links, could be potential biomarkers for investigating the role of oxidative stress in human diseases. Along this line, future work must be done to further elucidate the relationship between the levels of oxidatively induced DNA lesions and the stages of disease (e.g., cancer), which will provide important knowledge for future diagnostic applications of DNA adduct measurements.

During the past two decades, the sensitivities of mass spectrometers for measuring DNA adducts have greatly improved, and the LC-MS coupled with the stable isotope-dilution method is considered the most powerful technique for the unequivocal identification and accurate quantification of oxidatively induced DNA lesions in cells and tissues. We envision that further development of the method for high-throughput and simultaneous quantification of multiple oxidatively induced DNA lesions is necessary. Adapting methods to the measurement of lesions in ever increasingly smaller quantities of genomic DNA will also improve translation of these techniques, for example, to tissue biopsies like bronchoalveolar lavage. Improvement of LC-MS instrument performance as well as sample preparation and cleanup are both needed. For instance, the use of small internal diameter columns coupled with nLC-nESI-MS systems has already facilitated the quantitative analysis of multiple oxidatively induced DNA lesions with the use of low microgram quantity of DNA.88,159,160,163,165 Recently, high-resolution Orbitrap mass spectrometers have been used for the detection of alkylated DNA lesions induced by carcinogens present in tobacco smoke.344,345 Owing to their accurate mass measurement capability and high sensitivity, we expect more applications of this type of instruments for the quantification of low levels of oxidatively generated DNA lesions in cellular and tissue samples in the future. The improvement in LC-MS-based quantification techniques will also enable systematic adductomics research,346348 which may facilitate the discovery of novel oxidatively induced DNA lesions involved in the pathogenesis of human diseases.

Apart from the improvement in MS instrumentation, new sample preparation and cleanup techniques are also crucial for future high-throughput analysis of DNA lesions in vivo. For example, offline HPLC enrichment permits sensitive quantification of the 5′_R_- and 5′_S_- diastereomers of cdA and cdG.85,86,88,89 Although offline HPLC provides effective sample cleanup and removal of the excess amount of unmodified nucleosides in the DNA digestion mixture, it is relatively time-consuming and not conducive to rapid analysis. Because of the high separation ability, ultraperformance liquid chromatography (UPLC),349351 nanoflow UPLC (nUPLC)-nESI coupled with high-sensitivity and high-resolution MS method may enable simultaneous and efficient quantification of multiple oxidatively induced DNA lesions in a DNA digestion mixture without offline HPLC enrichment.

Last, but not least, it is important to note that caution needs to be exerted during sample preparation (i.e., DNA extraction, enzymatic digestion of DNA, etc.) so that artificial generation of oxidatively induced DNA lesions can be minimized. In this vein, the levels of 8-oxo-dG in the DNA of pig liver or HeLa cells could vary by 2 orders of magnitude when the same samples were analyzed by a network of laboratories.352 Thus, extreme precautions should always be taken while measuring this and other direct ROS-induced DNA lesions as well as those DNA adducts arising from byproducts of lipid peroxidation. It is crucial that the final levels of DNA lesions measured reflect the levels of DNA adducts present in cellular and tissue DNA, not a combination of those present in initial DNA and artificially formed during sample preparation.

Acknowledgments

Funding: This work was supported by the National Institutes of Health (R01 CA101864 and P01 AG043376).

Abbreviations

Acr-dG

acrolein _N_2-(3-oxopropyl)-dG adduct

AD

Alzheimer's disease

AID/APOBEC proteins

activation-induced cytidine deaminase/apolipoprotein B mRNA editing enzyme, catalytic polypeptide-like proteins

AML

acute myeloid leukemia

AP site

apurinic/apyrimidinic site

BER

base excision repair

cdA

8,5′-cyclo-2′-deoxyadenosine

Cro-dG

crotonaldehyde _N_2-(3-oxopropyl)-dG adduct

dA

2′-deoxyadenosine

dCMP

2′-deoxycytidine 5′-monophosphate

DDE

2,4-decadienal

dGh

guanidinohydantoin 2′-deoxynucleoside

dNIm

5-guanidino-4-nitroimidazole 2′-deoxynucleoside

DODE

9,12-dioxo-(10_E_)dodecenoic acid

DOOE

5,8-dioxo-(10_E_)-octenoic acid

DSB

double strand break

dSp

guanidinohydantoin 2′-deoxynucleoside

DHPN18O2

18O-labeled 1,4-endoperoxide of N,_N_′-di(2,3-dihydroxypropyl)-1,4-naphthalene-dipropanamide

EDE

4,5-epoxy-(2_E_)-decenal

EndoV

endonuclease V

FA

Fanconi anemia

Fapy-dA

4,6-diamino-5-formamidopyrimidine 2′-deoxynucleoside

Fapy-dG

2,6-diamino-4-hydroxy-5-formamidopyrimidine 2′-deoxynucleoside

Fe-NTA

ferric nitrilotriacetate

FH

fumarate hydratase

Fpg

formamidopyrimidine DNA glycosylase

G(-H)•

guanine neutral radicals

HHE

4-hydroxy-2-hexenal

hAAG

alkyladenine DNA glycosylase

hMTH1

human MutT homologue

hNEIL1

human endonuclease VIII-like 1

hOGG1

human 8-oxoguanine DNA glycosylase 1

HNE

4-hydroxy-2-nonenal

HNE-dG

HNE _N_2-(3-oxopropyl)-dG adduct

HPNE

4-hydroperoxy-(2_E_)nonenal

HOBr

hypobromous acid

HOCl

hypochlorous acid

H2O2

hydrogen peroxide

hRNAPII

human RNA polymerase II

M1dA

_N_6-(3-oxopropenyl)-2′-deoxyadenosine

M1dC

_N_4-(3-oxopropenyl)-2′-deoxycytidine

M1dG

pyrimido[1,2-α_]purine-10(3_H)-one-2′-deoxyribose

MDA

malondialde-hyde

MEFs

mouse embryonic fibroblasts

mESCs

mouse embryonic stem cells

_N_23-_ε_dG

_N_2,3-etheno-2′-deoxyguano-sine

NADPH

nicotinamide adenine dinucleotide phosphate

NER

nucleotide excision repair

ODN

oligodeoxyribonucleo-tide

ONE

4-oxo-2-nonenal

ONOO−

peroxynitrite

Pol η

DNA polymerase η

PUFA

polyunsaturated fatty acids

(R)-2-HG

(R)-2-hydroxyglutarate

ROS

reactive oxygen species

SDH

succinate dehydrogenase

SOD

superoxide dismutase

T7RNAP

T7 RNA polymerase

TDG

thymine DNA glycosylase

TET

ten-eleven translocation

thymidine glycol

5,6-dihydroxy-5,6-dihydro-2′-deoxythymidine

TLC

thin-layer chromatography

UPLC

ultraperformance liquid chromatography

1_N_2-_ε_dG

1,_N_2-etheno-2′-deoxyguanosine

2-OG

2-oxoglutarate

5-Br-dC

5-bromo-2′-deoxycytidine

5-Br-dU

5-bromo-2′-deoxyuridine

5-Br-U

5-bromouracil

5-cadC

5-carboxyl-2′-deoxycytidine

5-Cl-C

5-chlorocytosine

5-Cl-dC

5-chloro-2′-deoxycytidine

5-Cl-dU

5-chloro-2′-deoxyuridine

5-Cl-rC

5-chlorocytidine

5-Cl-U

5-chlorouracil

5-fdC

5-formyl-2′-deoxycytidine

5-fdU

5-formyl-2′-deoxyuridine

5-hmdC

5-hydroxymethyl-2′-deoxycytidine

5-hmdU

5-hydroxymethyl-2′-deoxyuridine

5-mdC

5-methyl-2′-deoxycytidine

5-OH-8-oxo-dG

5-hydroxy-substituted derivative of 8-oxo-dG

8-Br-dA

8-chloro-2′-deoxyadenosine

8-Br-dG

8-chloro-2′-deoxyadenosine

8-Cl-dA

8-chloro-2′-deoxyadenosine

8-Cl-dG

8-chloro-2′-deoxyguanosine

8-Cl-G

8-chloroguanine

8-Cl-rA

8-chloroadenosine

8-Cl-rG

8-chloroguanosine

8-nitro-dG

8-nitro-2′-deoxyguanosine

8-oxo-dA

8-oxo-7,8-dihydro-2′-deoxyadenosine

8-oxo-dG

8-oxo-7,8-dihydro-2′-deoxyguanosine

_ε_dA

1,_N_6-etheno-2′-deoxyadenosine

_ε_dC

3,_N_4-etheno-2′-deoxycytidine

Biographies

Biographies

Yang Yu received his B.S. (2009) and M.S. (2012) degrees in Biological Technology and Microbiology from Nankai University, China. He is currently a Ph.D. candidate in the Environmental Toxicology Graduate Program at the University of California Riverside under the supervision of Professor Yinsheng Wang. At the time of writing, Yang was close to completing his dissertation research project, which involves the development of mass spectrometric methods for the quantification of different types of structurally modified DNA and the application of these methods in assessing the biological implications of DNA adducts/modifications in various living cells/organisms.

Yuxiang Cui obtained her B.S. degree in Chemistry at Nankai University and M.S. degree in Food Safety and Toxicology at the University of Hong Kong. She joined the Environmental Toxicology Graduate Program at the University of California Riverside in 2014 and is currently a Ph.D. candidate under the guidance of Professor Yinsheng Wang. Her research projects mainly focus on the development and application of LC-MS methods for the quantitative measurement of DNA adducts.

Laura Niedernhofer is an Associate Professor in the Department of Metabolism and Aging at The Scripps Research Institute in Florida. She has a Bachelor of Sciences in Chemistry from Duke University and completed the Medical Scientist Training Program at Vanderbilt University, earning her Ph.D. in Biochemistry. She followed this with postdoctoral training in mouse genetics at the Erasmus Medical Center in Rotterdam, The Netherlands. Her area of expertise is DNA damage and repair. Dr. Niedernhofer's current research program is focused on discovering the mechanism by which DNA damage contributes to aging and age-related diseases using genetic and pharmacologic approaches.

Yinsheng Wang obtained his Ph.D. in Chemistry from Washington University in St. Louis in 2001. He is currently a professor of Chemistry and the Director for the Environmental Toxicology Graduate Program at the University of California Riverside. Yinsheng's research concentrates on the use of mass spectrometry, along with molecular biology and/or synthetic organic chemistry, to understand the occurrence and biological consequences of DNA adducts and post-translational modifications of proteins.

Footnotes

Notes: The authors declare no competing financial interest.

Special Issue: Mass Spectrometry and Emerging Technologies for Biomarker Discovery in the Assessment of Human Health and Disease

References