Introduction

Measurement of oxidatively base damaged DNA

The measurement of oxidized bases and nucleosides in cellular DNA is still a challenging analytical issue due to the complexity of the partly identified oxidation reactions in model compounds and the difficulties of detecting low amounts of oxidatively formed DNA damage, typically within the range of a few lesions per 10⁷ to 10⁹ normal bases. The assessment of the level of oxidized bases has been hampered during more than three decades by the existence of several major drawbacks in the designed assays [5]. Thus self-radiolysis processes associated with the use of radiolabeled isotope have led to an overestimation of the base damage [45]. This concerns the earlier assay that was designed for the determination of 5,6-dihydroxy-5,6-dihydrothymine (3), the so-called “thymine glycol” [46] and the more recently developed ³²P-postlabeling assay for monitoring the formation of 8-oxo-7,8-dihydro-2′-deoxyguanosine (13) [47]. Immunoassays aimed at detecting 13 have also given rise to conflicting data due to the lack of specificity of the antibodies that cross-react with the overwhelming normal guanine bases [5] and [48]. Other misleading data that prevent any biological conclusions to be drawn were also provided by the gas chromatography measurements whose values have been overestimated by factors varying from two to four orders of magnitude [5]. The origin of the main drawback that is associated with the use of the GC-MS method [49] and [50], introduced about 25 years ago, has been identified [51]. Thus, spurious oxidation of the normal bases has been shown to occur with an efficiency of about 0.1% during the derivatization step that is required to make the samples volatile. This leads to an artefactual generation of oxidized purine and pyrimidine bases such as 13, 8-oxo-7,8-dihydroadenine (21) and 5-(hydroxymethyl)uracil (9) [5]. A second matter of concern that is shared by the GC-MS assay and the other chromatographic methods requiring an acidic hydrolysis steps for the release of the bases, is the lack of stability of several modifications including 2,6-diamino-4-hydroxy-5-formamidopyrimidine (14) and 4,6-diamino-5-formamidopyrimidine (22) under hot acid formic treatment [52]. A third source of artefacts, although usually of lower amplitude, that may occur for all HPLC and GC assays deals with adventitious Fenton type oxidation reactions during the DNA extraction and digestion steps [5]. A general consensus now exists on improved chromatographic methods aimed at measuring 8-oxo-7,8-dihydro-2′-deoxyguanosine (13) through the cooperative efforts of 25 laboratories involved in the European Standard Committee on Oxidative DNA Damage (ESCODD) network [53] and [54]. Recommended protocols that include suitable conditions of DNA extraction in order to minimize artefactual oxidation [55] followed by suitable high performance liquid chromatography analysis of the DNA digestion products are now available [4] and [56]. It should be pointed out that an inverse correlation between the amount of extracted DNA and the level of adventitious oxidation of the overwhelming normal bases has been observed [57], thus requiring extraction of at least 30 µg of DNA. Two main detection techniques of the oxidized bases or nucleosides at the output of the HPLC column are available. The frequently used electrochemical detection technique (HPLC-ECD) which was introduced more than 20 years ago [58], is a robust method whose application in the oxidative detection mode is, however, restricted to only a few electroactive DNA lesions including 13,21, related formamidopyrimidine lesions 14,22 and 5-hydroxy derivatives of cytosine and uracil [59]. The most recently available electrospray ionization-tandem mass spectrometry (ESI/MS/MS) method [5], [39], [56] and [60] operating in the multiple reaction monitoring (MRM) mode is more versatile and, on the average, more sensitive than HPLC-ECD, allowing the accurate measurement of up to 11 oxidatively modified bases and nucleosides in cellular DNA of the about 80 identified so far in model compounds. This is not the case when HPLC-MS is used leading to overestimated values of radiation-induced 13 and 21 by 50- and 100-fold respectively [61] and [62]. This is likely due to the lack of accuracy of the selective ion monitoring mode when sensitivity close to a few femtomoles is required. It should be added that HPLC-MS³ is even required for the detection of modified nucleosides whose frequency is around a few lesions per 10⁹ nucleosides as illustrated by the measurement of two radiation-induced tandem base modifications in cellular DNA [63] and [64]. Therefore the review article will mostly focus on available data that were obtained by HPLC-MS/MS (eventually MS³) and/or HPLC-ECD measurements. Since it appears difficult to prevent any spurious oxidation reactions to occur in chromatographic assays despite a recent claim [4], the measurement of background levels of oxidatively base damage that can be accurately achieved by associating either the comet assay [65] or the alkaline elution technique [66] with DNA repair glycosylases will not be addressed in the present review. Cells were exposed to suitable oxidative reactions mediated by either physical or chemical agents that led to significant increases in the levels of base damage with respect to background levels that involve a significant contribution of oxidative metabolism. It may also be added that under the acute oxidation conditions used, repair of DNA damage was kept minimum.

Single base damage

Hydroxyl radical reactions

The highly reactive hydroxyl radical (^•OH) which is known to react at diffusion controlled rates with the four main purine and pyrimidine bases [29] and [67] was generated by the predominant indirect effects of ionizing radiation. The main ^•OH reaction with nucleobase consists in addition to unsaturated bonds with, however, two major exceptions that consist in competitive hydrogen abstraction reactions from the methyl group of thymine [6] and [19] and the 2-amino group of guanine [68].

Thymine

HPLC-MS/MS analysis [69] of suitably hydrolyzed DNA from γ–irradiated THP-1 human monocytes using the accurate isotopic dilution technique has led to the detection of six thymidine (1) oxidation products [70] and [71]. They consist of the four cis and trans diastereomers of 5,6-dihydroxy-5,6-dihydrothymidine (8) and two methyl oxidation products including 5-(hydroxymethyl)-2′-deoxyuridine (9) and 5-formyl-2′-deoxyuridine (10) (Fig. 1) The generation of radiation-induced 8-10 was found to be linear with the applied dose of γ–rays within the dose range 90-450 Gy. This allows the determination of the formation yield per each lesion that varies between 29 and 97 modifications per Gy and 10⁹ nucleosides (Table 1). Exposure of the monocytes to ¹²C⁶⁺particles that exhibit a linear energy transfer (LET) value of 31.5 keV/µm led to the formation of the 6 oxidized nucleosides 8-10, however with decreased yields. The reduced efficiency in ¹²C⁶⁺-ion induced formation of 8-10 can be explained by the decrease in ^•OH yield formation with the increase in the LET value of heavy particles with respect to γ-rays. This provides strong indirect support for the major role played by the indirect effects of ionizing radiation through the intermediary of ^•OH in the decomposition of cellular DNA.

A large body of information is available for the ^•OH-mediated degradation of thymidine (1) in aerated aqueous solution as either the free 2′-deoxyribonucleoside [6] or when inserted into DNA [72]. The formation of the four cis and trans diastereomers of 8 in cellular DNA may be accounted in the initial step of the sequence of reactions by preferential addition of ^•OH at C5 and to a lesser extent at C6 of thymine (1) [67]. This is suggested from the results of redox titration pulse radiolysis experiments that were performed with free thymine [67], [73] and [74]. Subsequently fast addition of molecular oxygen to reducing 6-yl 2 and oxidizing 5-yl 3 radicals is expected to give rise to 6-hydroperoxides 5 and 5-hydroperoxides 6 respectively upon reduction of transient peroxyl radicals, likely by O₂^•-[75]. The two sets of four diastereomers of 5 and 6 have been isolated, characterized and their absolute configuration assigned [76] and [77]. One of the main decomposition pathways of 5 and 6 is the stereospecific reduction of the O-O bond of the peroxide group that gives rise to 8 with retention of the initial cis or trans configuration [78]. The formation of 9 and 10 may be rationalized in terms of initial ^•OH-mediated hydrogen abstraction from the methyl group of 1. Addition of O₂, likely controlled by diffusion, to 5-(2′-deoxyuridilyl) methyl radical (4) leads to the formation of 5-(hydroperoxymethyl)-2′-deoxyridine (7) through reduction and protonation of the transient corresponding peroxyl radical. Loss of a water molecule from the peroxide function of 7 gives rise to 5-formyl-2′-deoxyuridine (10) while competitive reduction of the O-OH bond explains the formation of 5-hydroxymethyl)-2′-deoxyuridine (9) (Fig. 1). The overall radiation-induced formation yield of 9 and 10, the two methyl oxidation products, with respect to that of thymidine glycols 8 is much higher in cellular DNA than in free nucleoside 1. This may be explained by the location of the methyl group of 1 in the major groove of DNA that is accompanied by a relative increase in the susceptibility of the CH₃ group to ^•OH at the expense of that of the 5,6-ethylenic bond.

Guanine

8-Oxo-7,8-dihydro-2′-deoxyguanosine (13), the ubiquitous marker of oxidative stress to cells since it can be formed indifferently by ^•OH, one-electron oxidants, ¹O₂, peroxynitrite [1], [2], [6] and [19] and upon intrastrand addition with thymine 5(6)-hydroxy-6(5)-hydroperoxides [79] and [80] has been shown to be generated in cellular DNA upon exposure to gamma rays and high LET-heavy ions [70] and [71]. A second radiation-induced degradation product of guanine that has been identified as 2,6-diamino-4-hydroxy-5-formamidopyrimidine (14), has been found to be produced more efficiently than 13 (Table 1). It should be pointed out that a suitable protocol that takes into account the lability of the N-glycosidic bond of formamidopyrimidine 2′-deoxyribonucleoside derivatives 14,22 has been set-up in order to get a quantitative release of 14 and related adenine derivative compound 22 prior to HPLC-MS/MS measurements [52]. It was also found that increase in LET of ¹²C⁶⁺ (24.5 keV/µm) and ³⁶Ar¹⁸⁺ (250 keV/µm) heavy ions with respect to gamma photons led to a decrease in the yield of both 13 and 14 (Table 1). As discussed earlier for the radiation-induced degradation of 1, this is also strongly supportive of the predominant involvement of ^•OH in the formation of 13 and 14 in cellular DNA upon exposure to ionizing radiation. Therefore a reasonable mechanism based on previous model studies [15], [16] and [20] may be postulated for the radiation-induced generation of 13 and 14. Thus addition of ^•OH to the purine ring at C8 of 11 that is a relatively minor process in free nucleoside (17%) gives rise to reducing 8-hydroxy-7,8-dihydro-7-yl radical (12) [29] as the precursor of both 13 and 14. One-electron oxidation of 12 which is likely to involve O₂ as the oxidant in the nucleus leads to the formation of 13 (Fig. 2). It should be noted that competitive one-electron reduction of 12 giving rise to 14 is a major pathway in cells (Table 1) while relatively minor in aerated aqueous solution. The formation of 14 implies the cleavage of the C8-N9 imidazole bond that has been shown to occur on the basis of pulse radiolysis experiments of model compounds with a high unimolecular rate (k = 2 × 10⁵ s^-1) [81]. Interestingly 2,2-diamino-4-[(2-deoxy-β-D-erythro-pentofuranosyl)amino]-5(2H)-oxazolone (18) [82] that is the main ^•OH -mediated degradation product of 11 in aerated aqueous solution has been measured by HPLC-MS/MS in the DNA of diabetic rats [83]. It was found that 18 is formed in about 10-fold lower yield than 13 as the result of oxidative stress induced by diabetis with concomitant generation of ^•OH via O₂^•-. A likely mechanism for the formation of 18 involves the main ^•OH reaction pathway that was initially proposed to occur by addition at C4 followed by dehydration of the resulting adduct to give rise to oxyl radical 24. In fact it was shown [68] that ^•OH is able to abstract one hydrogen atom from the 2-amino group of 11 (Fig. 2). The resulting aminyl radical 15 would then undergo tautomerization into carbon C5 centered radical 16. In the next step, addition of O₂^•- , as shown in nucleosides and oligonucleotides |7] and not of O₂ as initially proposed, gives rise to the corresponding unstable 5-hydroperoxide. Subsequently, the latter intermediate is converted through a rather complicate pathway that involves successively water molecule addition, loss of CO₂ and then of formamide together with a final rearrangement of the purine ring [16], [20] and [82] into 2-amino-5-[2-deoxy-β-D-erythro-pentofuranosyl)amino]-4H-imidazol-4-one (17). The latter compound that is unstable in aqueous solution with a half-life of 10 h at neutral pH and 20 °C [84] is gradually converted into 14 (Fig. 2).

Adenine

8-Oxo-7,8-dihydro-2′-deoxyadenosine (21) and 4,6-diamino-5-formamidopyrimidine (22) which are two adenine analogs of 13 and 14, the related ^•OH-mediated degradation products of guanine (11) have been shown to be formed by HPLC-ESI/MS-MS measurement in the DNA of human monocytes following exposure to either gamma rays or heavy particles [70] and [71]. Similar degradation pathways than those proposed for 11 apply to 2′-deoxyadenosine (19). Thus the radiation-induced formation of 21 and 22 in cellular DNA may be rationalized in terms of initial ^•OH addition at C8 of 19 giving rise to 20 that then undergoes either one-electron oxidation [85] or one-electron reduction [86] as shown in Fig. 3. It should be however pointed out that 21 and 22 are generated in about 10-fold lower yield than 13 and 14, the related guanine degradation products. Another relevant piece of information concerning oxidative chemistry of 19 is the lack of detection of 2-hydroxy-2′-deoxyadenosine in γ–irradiated human monocytes up to doses of 200 Gy [87]. This contrasts with previous GC-MS measurements which have shown the formation of 2-hydroxyadenine as one of the main ^•OH-mediated base degradation products in the DNA of γ–irradiated cells and mice [88] and [89].

One-electron oxidation of nucleobases

The bases and the 2-deoxyribose moiety of DNA may be ionized through the direct effect of ionizing radiation [90] and [91]. In addition the purine bases and to a lesser extent the pyrimidine bases may be one-electron oxidized by type I photosensitizers [2] and [20], high intensity UV laser pulses [92] and [93] and oxidants including KBrO₃ [94] and CO3^•- [95], a decomposition product of nitrosoperoxycarbonate [96]. Among these different oxidizing systems, UVC nanosecond laser irradiation [92] and [93] has been found to be a suitable system for generating radical cations through bi-photonic ionization of the purine and pyrimidine bases in cellular DNA [71] despite the formation of dimeric pyrimidine photoproducts including cyclobutane pyrimidine dimers and pyrimidine (6-4) pyrimidone photoadducts as the result of monophotonic excitation [97]. Thus, several oxidized nucleosides including predominant 8-oxo-7,8-dihydro-2′-deoxyguanosine (13) and relatively minor 5,6-dihydroxy-5,6-dihydrothymidine (8), 5-(hydroxymethyl)-2′-deoxyuridine (9) and 5-formyl-2′-deoxyuridine (10) were found to be generated in the DNA of THP1 neoplastic human monocytes upon nanosecond UVC irradiation [71]. The formation of reducing radical 16 may be accounted for by hydration of the guanine radical cation 23 [98] leading to transient 12 (Fig. 2), the precursor of 13 through one-electron oxidation (Fig. 4). Hydration of the radical cation 25 arising from bi-photonic ionization of 1 has been shown to give rise quantitatively to the oxidizing radical 3 [99] (Fig. 5) which is further converted into thymidine glycols 8 through the intermediacy of the four diastereomers of 6-hydroperoxide 6 (Fig. 1). Competitive deprotonation of 25 leads to radical 4 (Fig. 5) which is transformed into 5-hydroperoxymethyl)-2′-deoxyuridine (7) after reduction of the transient peroxyl radical thus formed (Fig. 1). Subsequently 7 may undergo two competitive reactions that consist of reduction and loss of water molecule leading to the formation of 9 and 10 respectively (Fig. 1). It is interesting to note that the overall formation of thymidine oxidation products including 8, 9 and 10 represent only 10% of the yield of 13 upon one-electron oxidation reactions of the DNA bases. This distribution is quite different from the degradation product pattern generated by ^•OH, providing further support for the predominant contribution of indirect effects in the radiation-induced degradation of cellular DNA. The preferential formation of 14 upon bi-photonic ionization of nucleobases strongly suggests occurrence of charge transfer from initially generated purine and pyrimidine radical cations to guanine sites that have been shown to act as sinks for positive holes in model studies. Several mechanisms including phonon-assisted polaron-like hopping, coherent super-exchange hopping and multistep hopping have been proposed for the transfer of positive hole in double-stranded DNA [23], [100], [101] and [102].This constitutes to our best knowledge the first evidence for the occurrence of positive hole migration within cellular DNA.