Riboflavin activated by ultraviolet A1 irradiation induces oxidative DNA damage-mediated mutations inhibited by vitamin C

Abstract

An increasingly popular theory ascribes UVA (>320–400 nm) carcinogenicity to the ability of this wavelength to trigger intracellular photosensitization reactions, thereby giving rise to promutagenic oxidative DNA damage. We have tested this theory both at the genomic and nucleotide resolution level in mouse embryonic fibroblasts carrying the lambda phage cII transgene. We have also tested the hypothesis that inclusion of a cellular photosensitizer (riboflavin) can intensify UVA-induced DNA damage and mutagenesis, whereas addition of an antioxidant (vitamin C) can counteract the induced effects. Cleavage assays with formamidopyrimidine DNA glycosylase (Fpg) coupled to alkaline gel electrophoresis and ligation-mediated PCR (LM-PCR) showed that riboflavin treatment (1 μM) combined with UVA1 (340–400 nm) irradiation (7.68 J/cm²) or higher dose UVA1 irradiation alone induced Fpg-sensitive sites (indicative of oxidized and/or ring-opened purines) in the overall genome and in the cII transgene, respectively. Also, the combined treatment with riboflavin and UVA1 irradiation gave rise to single-strand DNA breaks in the genome and in the cII transgene determined by terminal transferase-dependent PCR (TD-PCR). A cotreatment with vitamin C (1 mM) efficiently inhibited the formation of the induced lesions. Mutagenicity analysis showed that riboflavin treatment combined with UVA1 irradiation or high-dose UVA1 irradiation alone significantly increased the relative frequency of cII mutants, both mutation spectra exhibiting significant increases in the relative frequency of G:C → T:A transversions, the signature mutations of oxidative DNA damage. The induction of cII mutant frequency was effectively reduced consequent to a cotreatment with vitamin C. Our findings support the notion that UVA-induced photosensitization reactions are responsible for oxidative DNA damage leading to mutagenesis.

A large body of evidence exists regarding the association between solar UV irradiation and human skin carcinogenesis (1, 2). Sunlight UV wavelengths that reach the surface of the earth are UVA (>320–400 nm) and UVB (280–320 nm), with shorter wavelengths (UVC) being completely absorbed by stratospheric oxygen (O₂) (1, 3). The biologically relevant UVA and UVB have been extensively studied as the etiologic factors for skin cancer (4). The mechanistic involvement of UVB in carcinogenesis rests upon the ability of this wavelength to induce promutagenic cis-syn cyclobutane pyrimidine dimers (CPDs), pyrimidine (6-4) pyrimidone photoproducts ((6-4)PPs), and Dewar valence photoisomers (4). However, the underlying mechanism of action for UVA carcinogenicity is not fully delineated (4). Despite the weak absorbance of UVA by DNA (3), a genotoxic mode of action for UVA has been demonstrated (1). Yet, the exact process through which UVA exerts genotoxicity remains elusive (4).

A widely recognized theory ascribes UVA genotoxicity to its ability to trigger intracellular photosensitization reactions, thereby giving rise to promutagenic DNA lesions (5). In fact, UVA has been shown to induce CPDs (6–10) and oxidative DNA damage (5, 6, 11–14), as well as mutagenesis (12, 15–20). However, the correlation between UVA-induced DNA damage and mutations has not been straightforward inasmuch as the spectrum of mutations produced by UVA has inconsistently matched the mutagenic potential of the various induced lesions (4). For the most part, however, the apparent discrepancies are attributable to a failure to investigate UVA-induced DNA damage and mutagenesis simultaneously (4). Ideally, the correlative nature of events pertaining to each aspect of UVA mutagenicity should be established by using comparative model systems under uniform experimental conditions.

Recently, we have shown distinctive UVA-induced oxidative DNA damage and mutagenesis in the cII transgene of Big Blue mouse embryonic fibroblasts (11, 12). In the present study, we have expanded our investigation to determine the underlying mechanism of induced mutagenesis in the same model system. To gain insights into the mechanism of DNA damage-targeted mutagenesis, we have studied the effects of UVA irradiation alone and in combination with a cellular photosensitizer, riboflavin (21), introduced into our model system. To verify the role of oxidative DNA damage in mutagenesis, we have also incorporated an antioxidant, vitamin C (22), into our experimental system. Subsequently, we have investigated the modulation of DNA damage and mutagenesis consequent to irradiation with UVA or treatment with UVA-activated riboflavin in the presence and absence of vitamin C.

The relevance of vitamin C for the present study lies on the fact that unlike other antioxidants, e.g., α-tocopherol or β-carotene, vitamin C does not absorb radiation in the UVA range (23). Thus, the inability of vitamin C to absorb UVA rules out the possibility of interference with our experimental design, which was tailored to investigate the antioxidative but not filtrating effects of a biologically relevant compound against oxidative stress imposed by riboflavin treatment combined with UVA irradiation or intense UVA irradiation alone. In addition, vitamin C is a major hydrophilic low-molecular weight nonenzymatic compound that belongs to a network of antioxidant defense mechanisms (22). It is known that vitamin C is widespread throughout the body, especially in the skin (24), the target organ for UV-associated cancers (22). On the other hand, riboflavin is relevant for the herein study because as an endogenous photosensitizer, this compound is ubiquitously present throughout the body, including in various highly UV-exposed organs, e.g., eye lens (21, 25).

Results

Cytotoxicity Examination.

In the absence of UVA1 irradiation, riboflavin treatment at none of the tested concentrations (0.01, 0.1, 1, 10, and 100 μM) was cytotoxic to mouse embryonic fibroblasts. In combination with UVA1 irradiation, however, riboflavin treatment caused concentration dependent cytotoxicity at radiation doses that were nonlethal per se. Irradiation with UVA1 alone resulted in appreciable cytotoxicity only at doses exceeding 15.36 J/cm² (Fig. 1). As a prerequisite for our mutagenicity experiments, we performed a series of preliminary tests to determine a riboflavin concentration together with UVA1 irradiation dose, which was minimally cytotoxic, yet, significantly mutagenic. This choice of dosing is crucial for establishing viable and replicating cells in which the induced DNA damage can be efficiently translated into mutations. As shown in Fig. 1, a treatment with 1 μM concentration of riboflavin together with 7.68 J/cm² UVA1 irradiation was marginally cytotoxic but significantly mutagenic in the cII transgene (Table 1). Treatment with riboflavin at concentrations ≥10 μM combined with UVA1 irradiation (7.68 J/cm²) was extremely lethal, thus, precluding inclusion of this dose range in mutagenicity testing. In keeping with the objective of our study, we have therefore used the combination treatment of 1 μM riboflavin and 7.68 J/cm² UVA1 irradiation in all DNA damage and mutagenicity experiments.

Fig. 1.

Cytotoxicity of riboflavin and/or UVA1 irradiation in the presence and absence of vitamin C to Big Blue mouse embryonic fibroblasts. Cell cultures were treated with increasing concentrations of riboflavin or solvent in the presence or absence of vitamin C at different concentrations for 20 min at 37°C in the dark. Subsequently, the treated cultures were irradiated with various doses of UVA1, and afterward cell viability was determined by the trypan blue dye exclusion assay. Viability is expressed as a percentage of total cell number. Error bars = SD.

Table 1.

Mutant frequency of the cII transgene in Big Blue mouse embryonic fibroblasts treated with riboflavin and/or UVA irradiation in the presence and absence of vitamin C

Treatment	Total no. of plaques, pfu	Verified mutant plaques	Mutant frequency, × 10−5*
Nontreatment	6,235,000	161	2.8 ± 0.7
Riboflavin (1 μM)	2,865,000	81	3.0 ± 0.7
UVA (7.68 J/cm2)	1,882,500	75	3.9 ± 1.5
UVA (15.36 J/cm2)	3,595,000	304	8.4 ± 0.9†
Riboflavin (1 μM) + UVA (7.68 J/cm2)	4,462,500	569	12.1 ± 2.2‡
Vitamin C (1 mM)	2,387,500	68	2.8 ± 0.1
Riboflavin (1 μM) + vitamin C (1 mM)	1,900,000	69	3.3 ± 2.1
UVA (7.68 J/cm2) + vitamin C (1 mM)	1,272,500	46	3.7 ± 1.4
UVA (15.36 J/cm2) + vitamin C (1 mM)	1,665,500	63	4.0 ± 1.0
Riboflavin (1 μM) + UVA (7.68 J/cm2) + vitamin C (1 mM)	3,937,500	167	4.1 ± 1.8

*Results are expressed as median ± SD.

^†Statistically significant as compared with control, P < 0.01.

^‡Statistically significant as compared with control, P < 0.01.

Vitamin C at concentrations between 0.001 and 10 mM alone or in combination with UVA1 irradiation was not cytotoxic in mouse embryonic fibroblasts. However, vitamin C at a minimum concentration of 1 mM could completely protect against the cytotoxic effects of combined treatment with riboflavin and UVA1 irradiation or of UVA1 irradiation alone at high doses. Supporting information (SI) Fig. 4 illustrates the protective effects of vitamin C against cytotoxicity of UVA1-activated riboflavin treatment and of UV irradiation at a high dose of 30.72 J/cm². Next, we repeated all cytotoxicity examinations 24 h posttreatment to determine possible delayed effects on cell viability consequent to various treatments. In all cases, there was a slight decrease (3–8%) in cell viability in cultures harvested 24 h posttreatment relative to counterpart cultures harvested immediately. Nonetheless, the overall pattern of cytotoxicity established immediately after all treatments or 24 h afterward was very comparable.

DNA Damage in the Genome.

Cleavage assay with formamidopyrimidine DNA glycosylase (Fpg) digestion and alkaline gel electrophoresis showed that neither riboflavin treatment nor UVA1 irradiation (up to 7.68 J/cm²) alone produced Fpg-sensitive sites in the genomic DNA of mouse embryonic fibroblasts. However, the combination treatment of riboflavin and UVA1 irradiation gave rise to Fpg-sensitive sites. Irradiation with UVA1 alone at doses ≥15.36 J/cm² also yielded considerable Fpg-sensitive sites. Treatment with vitamin C (0.001–10 mM) alone or in combination with UVA1 irradiation did not generate Fpg-sensitive sites. However, vitamin C treatment (1 mM) efficiently prevented the formation of Fpg-sensitive sites consequent to riboflavin treatment combined with UVA1 irradiation or UVA1 irradiation alone at high dose (Fig. 2a). Conversely, cleavage assay with T4 endonuclease V (T4 Endo V) digestion and alkaline gel electrophoresis showed no formation of CPDs in the genomic DNA in cells treated with riboflavin or irradiated with UVA1, individually or combined.

Fig. 2.

Quantification of induced DNA lesions. (a) DNA damage in the genome. Cleavage assay with Fpg digestion and alkaline gel electrophoresis of the genomic DNA of mouse embryonic fibroblasts treated with riboflavin or irradiated with UVA1, individually or combined, in the presence and absence of vitamin C. (b and c) LM-PCR mapping of DNA damage in the cII transgene. DNA footprinting in mouse embryonic fibroblasts treated with riboflavin or irradiated with UVA1, individually or combined: in the presence and absence of vitamin C with (+) and without (−) predigestion with Fpg enzyme (b) and with (+) and without (−) pretreatment with T4 Endo V plus CPD photolyase reactivation (c). bp, base pair; M, molecular size marker; nt, nucleotide position.

DNA Damage in the cII Transgene.

Ligation-mediated PCR (LM-PCR) of the Fpg-digested DNA showed that there was no lesion formation in the cII transgene in mouse embryonic fibroblasts treated with either riboflavin or UVA1 irradiation (up to 7.68 J/cm²) alone. However, the combination treatment of riboflavin and UVA1 irradiation caused a significant formation of lesions throughout the cII transgene. Irradiation with UVA1 alone at doses equal to or higher than 15.36 J/cm² also produced detectable level of lesions in the cII transgene (Fig. 2b). The specificity of Fpg-sensitive sites induced by riboflavin treatment combined with UVA1 irradiation or high-dose UVA1 irradiation alone was verified by showing the absence of these lesions in the respective samples not predigested with Fpg (Fig. 2b). Treatment with vitamin C (0.001–10 mM) alone or in combination with UVA1 irradiation did not result in lesion formation in the cII transgene. However, vitamin C treatment (1 mM) effectively inhibited the formation of lesions induced by riboflavin treatment combined with UVA1 irradiation or high-dose UVA1 irradiation alone (Fig. 2b). LM-PCR of the pretreated DNA with T4 Endo V plus CPD photolyase reactivation showed no detectable level of lesions in the cII transgene in cells treated with riboflavin or UVA1 irradiation, individually or combined (Fig. 2c). For LM-PCR analysis, CPDs can be converted to ligatable 5′ ends only by successive enzymatic digestion with T4 endo V and Escherichia coli CPD photolyase reactivation. The T4 endo V cleaves the glycosidic bond of the 5′ pyrimidine in a CPD and breaks the sugar-phosphate backbone between the two dimerized pyrimidines. The resulting dissociated 3′ pyrimidine retains an overhang dimer, which makes it unligatable until the E. coli CPD photolyase reactivation step detaches the dimer, yielding a single-stranded DNA with a normal base on the 5′ sugar-phosphate terminus (26).

Likewise, terminal transferase-dependent PCR (TD-PCR) analysis revealed no substantial formation of lesions in the cII transgene in mouse embryonic fibroblasts treated with either riboflavin or UVA1 irradiation alone (SI Fig. 5a). The combined treatment with riboflavin and UVA1 irradiation, however, produced significant levels of lesions throughout the cII transgene. Treatment with vitamin C (0.001–10 mM) alone or in combination with UVA1 irradiation did not yield any detectable level of lesions in the cII transgene. However, vitamin C treatment (1 mM) efficiently prevented the formation of lesions consequent to combined treatment with riboflavin and UVA1 irradiation. We presume the TD-PCR quantified lesions to be single-strand DNA breaks because photo-dimeric (6-4)PPs are not produced by UVA1 irradiation (6), and because the T4 Endo V cleavage assay and LM-PCR could rule out the possibility of formation of CPDs at the genomic or nucleotide resolution level, respectively. We additionally excluded the formation of (6-4)PPs by confirming that digestion with UV damage endonuclease, which is known to cleave both CPDs and (6-4)PPs (27), did not produce any cleavage in the genomic DNA in cells treated with riboflavin or UVA1 irradiation, individually or combined. Thus, by way of elimination, the lesions detected by TD-PCR can be considered as single-strand DNA breaks. We verified this assumption by subjecting the nondigested genomic DNA to alkaline gel electrophoresis, which showed clear smears indicative of strand breaks in the genomic DNA in cells treated with riboflavin and UVA1 irradiation combined (SI Fig. 5b).

cII Mutation Analysis.

In the absence of UVA1 irradiation, riboflavin treatment at none of the tested concentrations was mutagenic in the cII transgene in mouse embryonic fibroblasts. In combination with UVA1 irradiation (7.68 J/cm²), however, riboflavin treatment at a concentration of 1 μM was significantly mutagenic, raising the cII mutant frequency 4.3-fold over the background (12.1 ± 2.2 vs. 2.8 ± 0.7 × 10⁻⁵; P < 0.001). Although the UVA1 dose (7.68 J/cm²) administered in the combination treatment was nonmutagenic per se, higher doses of UVA1 irradiation alone significantly increased the cII mutant frequency relative to background. Treatment with vitamin C (0.001–10 mM) alone or in combination with UVA1 irradiation was not mutagenic in the cII transgene. However, vitamin C treatment (1 mM) substantially reduced the cII mutant frequency induced by the combination treatment with riboflavin and UVA1 irradiation or UVA1 irradiation alone at high doses (≥15.36 J/cm²) (Table 1).

For mutation spectrometry analysis, we randomly selected one hundred cII mutant plaques induced by riboflavin treatment combined with UVA1 irradiation as compared with ninety-seven spontaneously arisen control plaques. Of these plaques, 94 and 92 plaques, respectively, contained a minimum of one mutation along the cII transgene verified by DNA sequencing. The vast majority of the induced and spontaneous mutations were single base substitutions and less frequently single deletions/insertions, with one multiple mutations only in the control (SI Table 2). Detailed mutation spectra induced by riboflavin treatment combined with UVA1 irradiation or derived spontaneously are presented in SI Fig. 6. Overall, the spectrum of induced mutations was significantly different from the spontaneously arisen mutation spectrum (P < 0.04, 95% confidence interval = 0.03–0.05). To underscore the difference(s) between induced and spontaneous mutation spectra, we compared the frequency of each type of mutation, e.g., transitions, transversions, etc. in the respective mutation spectra. Because the cII transgene in the Big Blue system is a nontranscribed gene (28), the strand bias of mutagenesis, a phenomenon caused by transcription-coupled DNA repair (29), is unlikely to affect the spectrum of mutations in this transgene (28). Hence, it is justified to combine the strand mirror counterparts of all transitions (e.g., G → A + C → T) and transversions (e.g., G → T + C → A and G → C + C → G) when comparing the specific types of mutation between different treatment groups. As shown in Fig. 3, the spectrum of mutations induced by riboflavin treatment combined with UVA1 irradiation was different from the spontaneous mutation spectrum in that there was a significant increase in the relative frequency of G:C → T:A transversions (52.1% vs. 14.0%; P < 0.001). In confirmation, we have previously shown that UVA1 irradiation alone induced predominantly G:C → T:A transversions in the cII (11, 12) and lacI (30) transgenes of the same model system. Obviously, the hallmark of G:C → T:A transversion mutations induced by UVA1 irradiation alone was more pronounced in the mutation spectrum produced by riboflavin treatment combined with UVA1 irradiation (see Fig. 3). The frequent occurrence of G → T transversions, which are the signature mutation of oxidative DNA damage, e.g., 8-oxo-7,8-dihydro-2′-deoxyguanosines (8-oxo-dG) (31, 32), was concurrent with a lower frequency of C → T transitions targeted to dipyrimidine sites, which are typically induced by photo-dimeric CPDs or (6-4)PPs (4), in the induced mutation spectra relative to control (Fig. 3).

Fig. 3.

Comparative mutation spectra of the cII transgene in Big Blue mouse embryonic fibroblasts treated with riboflavin (1 μM) combined with UVA1 irradiation (7.68 J/cm²), UVA1 irradiation alone (18.0 J/cm²), or control. Ins, insertion; Del, deletion. ∗, As compared with “nontreated control”; P < 0.001. †, as compared with “nontreated control”; P < 0.05.

Discussion

Our cytotoxicity examination revealed that riboflavin treatment alone was not cytotoxic to mouse embryonic fibroblasts. However, in combination with UVA1 irradiation, riboflavin treatment could significantly impair cellular viability. It has been shown that UVA irradiation of riboflavin generates reactive intermediate species, such as singlet oxygen, superoxide anions, triplet state riboflavin radicals, and hydrogen peroxide (33–35). High-dose UVA irradiation alone has also been proven to produce reactive oxygen species with known cytotoxic properties (36). In our experiments, vitamin C at a physiologically relevant concentration of 1 mM was completely protective against the cytotoxic effects of combined treatment with riboflavin and UVA1 irradiation or of UVA1 irradiation alone at high doses. To rule out the possibility of filtration of UVA1 by vitamin C, we spectrophotometrically determined the wavelength absorption of vitamin C and riboflavin in our experimental system. The UV absorption spectrum for vitamin C was only in the UVC range, reaching a peak at ≈265 nm. However, the wavelength scan of riboflavin confirmed an absorption in the UVA1 range with a peak at 375 nm, which was well within the emission spectrum of our UVA source. Thus, the observed protective effects of vitamin C can essentially be ascribed to its antioxidant and free radical scavenging capacities (22).

Our quantification of DNA damage both at the genomic and nucleotide resolution level verified that riboflavin treatment in combination with UVA1 irradiation or high-dose UVA1 irradiation alone induced oxidized (ring opened) purines in the overall genome and in the cII transgene in mouse embryonic fibroblasts. DNA footprinting mapped the formation of most induced lesions to guanine residues along the cII transgene. It is known that the lower ionization potential of guanine relative to other DNA bases makes it a favorable target for type-I photosensitized one-electron oxidation and/or type-II photooxidation, mediated through singlet oxygen generation (37, 38). The resulting radical cations of guanine can undergo hydration to produce 8-oxo-dG and 2,6-diamino-4-hydroxy-5-formamidopyrimidine (39), both lesions being substrates of the DNA repair protein, Fpg (40). In our hands, cleavage assays coupled to alkaline gel electrophoresis and LM-PCR showed that Fpg-sensitive sites were formed conspicuously in the overall genome and in the cII transgene in cells treated with riboflavin combined with UVA1 irradiation or irradiated with high-dose UVA1 alone.

We and others have previously shown that intense UVA irradiation can cause single-strand DNA breaks in various experimental systems (6, 41–43). The formation of these lesions has been ascribed to hydroxyl radical generation, e.g., by means of conversion of hydrogen peroxide in the presence of transition metals through Fenton-like reaction (44). Formation of hydrogen peroxide has also been reported in various test systems consequent to treatment with UVA-activated riboflavin or irradiation with high-dose UVA alone (34, 45). In our experiments, single-strand DNA breaks were formed in cells treated with riboflavin combined with UVA1 irradiation. However, irradiation with UVA1 alone at a moderately mutagenic dose did not produce detectable levels of single-strand DNA breaks. These findings imply that in our experimental system, the presence of endogenous photosensitizers enables UVA1 to exert its DNA damaging effects primarily through type-II photosensitization reactions involving singlet oxygen generation. However, the excess riboflavin introduced into our system may have prompted UVA1 to cause additional DNA damage by producing other reactive oxygen species such as hydrogen peroxide. In both cases, the formation of DNA damage was effectively inhibited by a cotreatment with vitamin C. The protective effect of vitamin C against oxidative stress-induced DNA damage has been demonstrated by others (43) and ascribed to its antioxidant and/or free radical scavenging properties (22), as well as its ability to regenerate other antioxidants, e.g., α-tocopherol and urate from their respective radicals (46). Vitamin C can also directly react with singlet oxygen, and the resulting peroxide may be less toxic owing to efficient peroxide removing systems (47).

Our mutagenicity analysis showed that riboflavin treatment in combination with UVA1 irradiation or high-dose UVA1 irradiation alone significantly increased the cII mutant frequency relative to background in mouse embryonic fibroblasts. A cotreatment with vitamin C, however, substantially reduced the cII mutant frequency induced by the combination treatment with riboflavin and UVA1 irradiation or high-dose UVA1 irradiation alone. DNA sequencing of the cII mutants revealed a significant increase in the relative frequency of G:C → T:A transversions induced by riboflavin treatment combined with UVA1 irradiation, which is consistent with our previously established mutation spectra induced by UVA1 irradiation alone in the cII (11, 12) and lacI (30) transgenes of the same experimental model system. As mentioned earlier, G → T transversions are the hallmark mutations induced by oxidative DNA damage, including 8-oxo-dG and 2,6-diamino-4-hydroxy-5-formamidopyrimidine (11, 12, 31, 48). The established signature mutation of G:C → T:A transversions induced by UVA1 irradiation alone was more pronounced in the mutation spectrum produced by riboflavin treatment combined with UVA1 irradiation (25.2% vs. 52.1% of all mutations in Fig. 3). These observations confirm that the addition of an endogenous photosensitizer, such as riboflavin, to our experimental system magnifies the oxidative-DNA damage-mediated mutagenicity of UVA1 alone. The induced mutation spectra did not display the signature mutations of photodimeric CPDs or (6-4)PPs, i.e., single or tandem C → T transitions targeted to dipyrimidine sites (4). The latter finding is in good agreement with the absence of photodimeric lesions in the genomic DNA consequent to various UVA1 treatments.

The spectrum of mutations produced by UVA irradiation in different studies has not always matched the mutagenic potential of the various induced lesions (4). The discrepancies have mostly manifested in studies where the two endpoints, i.e., DNA damage and mutation, have not been determined simultaneously and/or within a single test system (4). We have shown that cell cultures of varying species and types are differently resistant toward cytotoxic and genotoxic effects of UVA radiation (6, 11, 12). Such variability might arise from the unique DNA repair capacity and diverse content of intracellular photosensitizers, specific for each species and cell type (4). In addition, experimental variables such as UVA source and dose may also lead to some discrepancies (4). For instance, irradiation sources, which emit contaminating UV wavelengths, e.g., in the UVB range, could cause distorted mutation spectra by inducing UVB-specific DNA lesions. This latter concern is of significance because per joule basis, UVB radiation is up to 50,000 times more genotoxic than UVA radiation (49). Ikehata et al. (50) have used an irradiation source emitting a minute but appreciable fraction of UVB radiation, and reported an induced mutation spectrum predominated by single and tandem C → T transitions at dipyrimidine sites in the lacZ transgene in mouse skin epidermis. Also, Woollons et al. (51) have demonstrated that the 0.8% UVB component of a UVA sunlamp accounted for 75% of the CPDs induced in human keratinocytes irradiated in vitro.

The fairly high detection limit of most available techniques for quantification of photo-induced DNA damages (particularly at the nucleotide resolution level) has necessitated intense UVA irradiation, enabling sufficient production of DNA lesions (6, 9, 10, 52). Because high-dose UVA irradiation causes severe cytotoxicity, the induced DNA damage is unlikely to be translated into mutation in cells with no or low proliferation capacity (4). We have established that the type of induced DNA lesions depends on UVA irradiation dose, i.e., at low irradiation dose, of relevance for mutagenicity experiments, oxidative DNA damage is formed predominantly (11, 12). However, at high-dose UVA irradiation, both CPDs and oxidative DNA damage are induced (6). It is imperative that UVA-induced DNA damage and mutagenesis be investigated simultaneously and at a biologically relevant dose. Our studies are unique in that they compare DNA damage-targeted mutagenicity of UVA at a relevant dose in a single test system, therefore, accounting for the above-mentioned potential confounding variables.

Under special circumstances, vitamin C has been shown to exhibit pro-oxidant activities (53). For example, Fan et al. (25) have reported that overexpression of the human vitamin C transporter SVCT2 in a mouse model accelerated the modification of lens crystalline as a consequence of vitamin C glycation by the Maillard reaction. In humans, high concentrations of vitamin C have been found in the stratum corneum, and viable epidermis and dermis, with the greatest levels being in the deeper layers (24) wherein penetrating UVA causes oxidative stress (22). Millimolar concentrations of vitamin C (up to 10 mM) have been reported in various tissues, including hippocampus and the adrenal glands. The presence of vitamin C and glutathione in the millimolar range in the human lens takes on additional significance with the knowledge that the riboflavin concentration in this tissue reaches 4.5 μM (25). In our experiments, vitamin C at a physiologic concentration exerted antioxidant effects in response to oxidative stress imposed by UVA1-activated riboflavin treatment or high-dose UVA1 irradiation alone. The implication of these findings for public health is the formulation of sunscreens with optimal concentrations of antioxidants.

In conclusion, we have shown a characteristic oxidative DNA damage-targeted mutagenicity of UVA1 in the cII transgene of Big Blue mouse embryonic fibroblasts, augmented by the addition of the photosensitizer riboflavin, and diminished by the inclusion of the antioxidant vitamin C. Our findings reaffirm the notion that intracellular photosensitization reactions, which generate promutagenic oxidative DNA damage, are responsible for UVA-induced genotoxicity.

Materials and Methods

Cell Culture and Treatment.

Early passage Big Blue mouse embryonic fibroblasts were grown to monolayer ≈30% confluence in DMEM supplemented with 10% FBS. Before treatment, the culture media were removed, and the cells were washed thoroughly with PBS. Riboflavin and vitamin C (Sigma–Aldrich, St. Louis, MO) were dissolved in PBS. The cells were treated with increasing concentrations of riboflavin (0.01–100 μM) or solvent in the presence or absence of vitamin C at different concentrations (0.001–100 mM) for 20 min at 37°C in the dark. After multiple washes with PBS, the culture dishes were placed on ice and irradiated (in PBS) with various doses of UVA1 (0.96, 1.92, 3.84, 7.68, 15.36, and 30.72 J/cm²). The UVA source was a Sellas Sunlight System (Medizinische Geräte GmbH; Gevelsberg, Germany) with an average fluence rate of 60 mW/cm². The source exclusively emits long wave UVA (UVA1: 340–400 nm) and not UVA at the borderline with UVB (see SI Fig. 7). Immediately after irradiation, the cells were harvested by trypsinization for determination of cell survival, and for detection of DNA damage. Alternatively, the cells were cultured in complete growth medium for an additional 4 days, and afterward were analyzed for mutant frequency and mutational spectrum of the cII transgene. The 4-day growing period is essential for the fixation of all mutations into the genome (26). At the time of harvesting, all cultured cells had undergone 3–4 population doublings and reached full confluence. All experiments were conducted in triplicates.

Lesion-Specific Cleavage Assays with Fpg and T4 Endo V.

The enzymatic digestion assays use Fpg for cleavage of oxidized (ring-opened) purines, and T4 Endo V for nicking at CPDs (26). The digested DNA is then subjected to alkaline gel electrophoresis for assessment of damage in the overall genome. Alternatively, the DNA digest can be analyzed by footprinting methodologies, e.g., LM-PCR, enabling the detection of lesions at the level of nucleotide resolution (26). Briefly, DNA was digested with an excess amount of Fpg (Trevigen, Gaithersburg, MD) or T4 Endo V (Epicentre, Madison, WI) in special buffers supplied by the respective manufacturers. After ethanol precipitation, the digests were loaded onto a 1.5% alkaline agarose gel, and run for 4 h at 40 V. As standard controls, genomic DNAs treated with methylene blue plus white light or irradiated with UVB, which contained known amounts of 8-oxo-dG or CPDs (6), respectively, were used in all runs.

LM–PCR and TD–PCR.

LM–PCR was used for DNA footprinting of specific UVA-induced lesions, including oxidized (ring-opened) purines and CPDs along the entire length of the cII transgene. Methodologically, LM–PCR is based on the concept that DNA polymerase cannot synthesize DNA past certain type of lesions, i.e., single-strand breaks. The principles of TD-PCR are essentially similar to LM-PCR with the only differences being the primer extension and ligation steps (54). Unlike LM-PCR, TD-PCR does not require chemical or enzymatic digestion of DNA to produce single-strand DNA breaks with 5′ phosphate groups at the lesion formation sites. Thus, TD-PCR can be readily applied to all 5′ phosphorylated or unphosphorylated DNA templates that contain polymerase blocking lesions, i.e., single-strand breaks and/or bulky lesions, e.g., CPDs or (6-4)PPs.

cII Mutant Frequency and Mutation Spectrometry Determination.

The λLIZ shuttle vectors were rescued from the genomic DNA, and packaged into viable phage particles by using the “lambda LIZ transpack packaging extract” according to the manufacturer's instructions (Stratagene, La Jolla, CA). The phages were preadsorbed to G1250 E. coli, and the bacterial culture was plated on TB1 agar plates. The plates were incubated for 48 h at 24°C or overnight at 37°C. Putative cII mutant plaques were all verified after being replated under the selective condition on a second TB1 agar plate. The cII mutant frequency was expressed as the ratio of the number of verified mutant plaques to the total number of screened plaques. For quality assurance, control phage solutions containing mutant and wild-type lambda cII having known mutant frequencies were assayed in all runs. As recommended by the manufacturer, a minimum of 3 × 10⁵ rescued phages were screened for each experimental condition. For mutation spectrometry analysis, the verified plaques were cored, and subjected to PCR amplification by using the “lambda select-cII sequencing primers” according to the manufacturer's recommended protocol. The PCR products were purified with a QIA quick PCR-purification kit (Qiagen GmbH, Hilden, Germany), and sequenced by using a Big Dye terminator cycle sequencing kit and an ABI-377 DNA Sequencer (ABI Prism; PE Applied BioSystems, Foster City, CA). For more details, see SI Materials and Methods.

Abbreviations

(6-4)PP

pyrimidine (6-4) pyrimidone photoproducts

8-oxo-dG

8-oxo-7,8-dihydro-2′-deoxyguanosine

CPD

cis-syn cyclobutane pyrimidine-dimer

Fpg

formamidopyrimidine DNA glycosylase

LM-PCR

ligation-mediated PCR

TD-PCR

terminal transferase-dependent PCR

T4 Endo V

T4 endonuclease V.