Mechanism of unprecedented hydroxyl radical production and site-specific oxidative DNA damage by photoactivation of the classic arylhydroxamic acid carcinogens

Abstract

The carcinogenicity of N-hydroxy-2-acetamidofluorene (N-OHAAF), the major genotoxic metabolite of the classic model aromatic amine (AA) carcinogen 2-acetylaminofluorene, has been attributed mainly to the formation of DNA adducts via arylnitrenium upon enzymatic activation. Here, we show, unexpectedly, that exposure of N-OHAAF to UV or sunlight irradiation can not only induce the formation of the well-known covalent DNA adducts, but, more interestingly, simultaneous generation of oxidative DNA damage was also observed as measured by the formation of DNA single-/double-strand breaks (SSBs/DSBs) and 8-oxo-2′-deoxyguanosine (8-oxodG), which were partly inhibited by the typical hydroxyl radical (^•OH) scavengers. Electron spin resonance spin-trapping and fluorescent studies unequivocally confirmed that the highly reactive ^•OH was generated from photolysis of N-OHAAF. Further DNA sequencing investigations suggest that photoactivation of N-OHAAF caused preferential cleavage at guanine, thymine and cytosine sites. More importantly, the formation of 8-oxodG and DSBs were also observed when fibroblast Balb/c-3T3 cells were co-exposed to N-OHAAF/UV irradiation as measured by double immunofluorescence staining. Taken together, we propose that both ^•OH and amidyl radicals can be readily produced via N–OH homolysis in N-OHAAF by photoirradiation, which can induce both oxidative and covalent DNA damage. This represents the first report of ^•OH production and site-specific DNA damage via photoactivation of the genotoxic hydroxamic acid intermediate, which provides a new free radical perspective to better understand the molecular mechanism for the carcinogenicity of AAs.

Introduction

Aromatic amines (AAs) are widespread precursors and intermediates for the production of polymers, pesticides, dyes and pharmaceuticals. AAs are also produced in diesel exhaust, tobacco smoking and high-temperature cooking process. Intensive use and production of these compounds result in their ubiquitous occurrence in our environment, thereby creating a potential for human exposure (1,2). Epidemiological survey accompanied with extensive in vivo data have demonstrated that AAs can cause various cancers at different tissue sites in experimental animals and, unfortunately, also in humans (3,4). As a prototype carcinogen of AAs, 2-acetylaminofluorene (AAF) was anticipated to be potentially mutagenic and carcinogenic to humans based on suffcient evidence of induction of tumors at liver, bladder, mammary glands, small intestine, ear duct, skin and several other organs in various experimental animals through inhalation, ingestion and skin contact (5,6). 2-Nitrofluorene (NF), one of typical nitro-polycyclic aromatic hydrocarbons (nitro-PAHs) found most in particulate emissions from diesel engines and kerosene heaters, was classified as a group 2B carcinogen by the IARC for its carcinogenic potential to humans (7), which can be attributed mostly to the formation of its carcinogenic metabolite AAF (8).

AAs themselves are commonly biologically inert and require metabolic activation for carcinogenic activity. The most widely studied member of this class of carcinogens is AAF (9). The major activation pathway for AAF was N-hydroxylation to a proximate carcinogenic arylhydroxamic acid metabolite, N-hydroxy-2-acetamidofluorene (N-OHAAF) (10), followed by O-esterification. It has been proposed that the unstable ester metabolites may undergo N–O heterolysis to form a reactive electrophilic arylnitrenium ion, which exerts ultimate carcinogenic and mutagenic effects by covalently binding to critical sites on DNA, RNA and proteins (5,11). An alternative metabolic activation pathway of N-OHAAF has been proposed through one-electron oxidation to its corresponding nitroxide radical intermediates, which are known to produce two potent carcinogens 2-nitrosofluorene and N-acetoxy-AAF (N-OAcAAF) via disproportionation (12,13).

Although the enzymatic activation mechanism by which N-OHAAF induces cancer has been extensively studied, the photoactivation of N-OHAAF has received little attention. As the largest organ in the human body exposed to sunlight irradiation, skin is constantly suffering sunlight-induced skin lesions including sunburn, photoaging and skin cancers (14,15). Meanwhile, with the largest surface area in the body, skin is frequently exposed to various xenobiotics. Increasing evidence demonstrated that sunlight exposure combined with photoreactive environmental pollutions including PAHs and nitro-PAHs can synergistically induce the formation of 8-oxo-2′-deoxyguanosine (8-oxodG), DNA adducts and serious DNA double-strand breaks (DSBs), which significantly enhance visible photodamage and light-induced skin cancer in experimental animals (16,17). Arylhydroxamic acids are carcinogenic metabolites of nitro-PAHs (18), and it has been shown that they could be photoactivated to cause damage to DNA and other biomacromolecule via forming reactive radical intermediates including nitroxide radicals and amidyl radicals (19). In addition, as a model substance for nitro-PAHs, NF could induce covalent binding to RNA and protein in vitro and increase covalent binding in the rat skin in vivo via photoactivation (20), but the molecular mechanism underlying such damage is still not clear yet.

Therefore, as a typical carcinogenic arylhydroxamic acid and an important carcinogenic metabolite of NF, we need to pay attention to the potential phototoxicity of N-OHAAF. So, in this study, we address the following questions. (i) Can N-OHAAF be photoactivated? If so, (ii) can reactive radical species such as reactive oxygen species (ROS) and amidyl radicals be produced during irradiation of N-OHAAF? (iii) What kind of DNA damage will be induced by these radical species: oxidative DNA damage or covalent DNA adducts? (iv) What is the underlying molecular mechanism?

Materials and methods

Cell line and cell culture

Please refer to the supplementary materials available at Carcinogenesis Online.

DNA strand breakage

Plasmid DNA agarose gel electrophoresis was used to investigate DNA strand breakage induced by N-OHAAF irradiation.

Ultra high performance liquid chromatography-electrospray ionization mass spectrometry (UHPLC-ESI-MS)/ mass spectrometry (MS) analysis of 8-oxodG and dG adducts

8-OxodG, N-(deoxyguanosin-8-yl)-2-acetylaminofluorene (dG-C8-AAF) and N-(deoxyguanosin-8-yl)-2-aminofluorene (dG-C8-AF) were characterized and accurately quantified by UPLC-ESI-TSQ-MS/MS (Thermo Scientific TSQ Quantum Access MAX LC-MS) in selective reaction monitoring mode by the MS fragmentation of m/z 284.1→168.0, 489.2→373.1 and 447.2→331.1, respectively.

Double immunofluorescence analysis of γ-H2AX and 8-oxodG generation in Balb/c-3T3 cells induced by N-OHAAF/UV irradiation

For experimental procedure details, please refer to the supplementary materials available at Carcinogenesis Online.

Electron spin resonance (ESR) spin-trapping studies

ESR spectra were recorded in a Bruker EMX-plus-10/12 spectrometer at 9.88 GHz. For typical parameters, please refer to the supplementary materials available at Carcinogenesis Online.

Detection of ^• OH formation by terephthalic acid (TPA) fluorescent method

Fluorescent spectra were recorded on a Cary Eclipse (Varian) spectrofluorometer.

Detection of base specificity of DNA damage by irradiation of N-OHAAF

The reacted DNA samples were analyzed in the ChemiDoc™ MP Imaging System (Bio-Rad, Hercules, California) after treatment according to our previous report (21).

Results

Strand breakage in pBR322 DNA was induced by UV irradiation of N-OHAAF

It is well known that oxidative DNA damage will lead to strand breakage, intra- and inter-strand cross-links and DNA-protein cross-links. In our study, plasmid DNA (pBR322) agarose gel electrophoresis was used to investigate oxidative DNA damage induced by photolysis of N-OHAAF. Interestingly, we found that DNA single-strand breaks (SSBs) and a small amount of DSBs at higher concentration were induced by photolysis of N-OHAAF. As shown in Figure 1a and b, SSBs were observed when DNA was exposed to even 1 μM N-OHAAF or irradiation for only 10 s. In contrast, neither N-OHAAF nor irradiation alone induced appreciable DNA strand breakage. Besides, DNA strand breakage increased with increasing irradiation and concentration of N-OHAAF. These results showed that DNA cleavages can be induced by N-OHAAF plus UV irradiation.

Not only dG adducts but also 8-oxodG was produced from dG and calf thymus DNA (ct-DNA) by UV irradiation of N-OHAAF

DNA damage is an important first step during many pathological processes including carcinogenesis and ageing. In most cases, guanine is the main DNA target for its lowest ionization potential and being the most susceptible to oxidation (22). 8-OxodG is the most commonly measured biomarker of oxidative DNA damage. Evidence indicates that 8-oxodG residues in DNA will lead to G to T transversion unless repaired during DNA replication, and 8-oxodG plays a critical role in a broad range of pathophysiological process, such as carcinogenesis, aging and degenerative diseases (23). DNA adducts are another modification resulted from covalent interaction with carcinogens. Many mutagens and carcinogens are known to yield electrophiles, either directly or via metabolic activation, that form covalent adducts with DNA (11). Chemical modification of DNA may induce abnormal replication which is considered as the start of mutation and further cancer formation without proper DNA repair. Therefore, the level of DNA adducts in normal cells can serve as a biomarker for chemical hazard exposure or cancer therapy efficacy (24,25). The carcinogenicity of N-OHAAF has been attributed mostly to DNA adducts formation, the most abundant being the C8 adduct at guanine (26,27). Therefore, in our study, we examined whether 8-oxodG and dG adducts can be induced by photolysis of N-OHAAF with dG and ct-DNA, respectively.

Obviously, 8-oxodG was observed when dG was co-exposed to N-OHAAF/UV irradiation by comparison with the standard compound (Supplementary Figure 1, available at Carcinogenesis Online), while little 8-oxodG was detected without either irradiation or N-OHAAF (Figure 2a). The frequency of 8-oxodG was 2.2 lesions/10⁴ dG with 100 μM N-OHAAF irradiated for only 10 s and up to 59 lesions/10⁴ dG with continuing irradiation for 30 min. Besides, we found that 8-oxodG formation increased with increasing irradiation and concentration of N-OHAAF.

Unexpectedly, a major dG adduct (peak 1 in Supplementary Figure 2, available at Carcinogenesis Online) was also observed when dG was incubated with N-OHAAF/UV irradiation. The major adduct showed not only the same chromatographic retention time but also the same transition pair of selective reaction monitoring (m/z 489.2→373.1) (Supplementary Figure 2, available at Carcinogenesis Online) as the authentic dG-C8-AAF, which was isolated and purified from the reaction of dG with N-OAcAAF (28). We speculate that certain reactive intermediate produced by photolysis of N-OHAAF can covalently bind to dG, mostly at the C8 site of guanine. Quantitative results showed that the yield of dG-C8-AAF was in a concentration- and time-dependent manner (Figure 2c). The frequency of dG-C8-AAF reached up to 42 lesions/10⁴ dG when dG (0.5 mM) together with N-OHAAF (100 μM) were irradiated for only 10 s and reached maximum at 5 min. Surprisingly, continuous irradiation induced the decrease of dG-C8-AAF instead (for possible reasons, see below). Meanwhile, another minor dG adduct dG-C8-AF was also detected during irradiation of dG and N-OHAAF (Supplementary Figure 3a, available at Carcinogenesis Online) and increased with increasing irradiation. By contrast, little dG-C8-AF was observed in dG/N-OAcAAF reaction system, which can produce large amounts of dG-C8-AAF (Supplementary Figure 3a, available at Carcinogenesis Online). Further study showed that direct irradiation of standard dG-C8-AAF for 10 min induced nearly 60% conversion into dG-C8-AF (Supplementary Figure 3b and c, available at Carcinogenesis Online). Therefore, we speculated that dG-C8-AF might derive from deacetylation of dG-C8-AAF by UV irradiation (Supplementary Scheme 1, available at Carcinogenesis Online) and continuous irradiation may decrease the yield of dG-C8-AAF by the same way.

Formation of 8-oxodG and dG adducts was also observed when dG was substituted by ct-DNA (Figure 2b and d). It was more efficient to induce damage in dG than in ct-DNA, which might be partly due to the steric hindrance effect on the reaction of reactive species with DNA. These results indicate that photolysis of N-OHAAF can induce DNA damage not only through formation of the well-known covalent DNA adducts but also through oxidative damage to form DNA strand breaks and 8-oxodG.

Formation of 8-oxodG and DNA DSBs were also observed when fibroblast Balb/c-3T3 cells were co-exposed to N-OHAAF/UV irradiation as measured by double immunofluorescence staining

The above studies found that co-exposure of N-OHAAF/UV irradiation to isolated and purified DNA (plasmid DNA and ct-DNA) could lead to formation of oxidative DNA damage and DNA adducts. It is interesting to test whether similar effects can be observed in live cells. The Balb/c 3T3 neutral red uptake phototoxicity test has been widely used to identify the phototoxic potential of a test substance (29). Therefore, embryonic mouse fibroblast Balb/c-3T3 cells were used as a model cell for phototoxic examination in cellular system in our experiments.

Phosphorylated form of variant histone H2AX (γ-H2AX) is commonly considered to occur as a result of DSBs (30). So, monoclonal antibodies of γ-H2AX and 8-oxodG were used in immunofluorescence staining analysis to show whether DNA DSBs and 8-oxodG were induced in the cell nucleus by co-exposure to N-OHAAF/UV irradiation. As shown in Figure 3, neither N-OHAAF nor UV irradiation alone can induce DNA damage, while 8-oxodG was obviously produced after co-treatment with N-OHAAF/UV irradiation. Although the intensity of DSBs was not as strong as 8-oxodG, difference can still be observed between control groups and treated group (30 μM N-OHAAF plus UV irradiation) (Figure 3b and c). Since apoptosis can also significantly induce DSBs generation, we further examined whether apoptosis can be induced after co-treatment with N-OHAAF/UV irradiation by Annexin V-FITC/PI double staining. As shown in Supplementary Figure 4, available at Carcinogenesis Online, little apoptosis was observed after co-treatment with N-OHAAF/UV irradiation, which excluded the possibility of DNA DSBs generation from apoptosis. Taken together, these results showed that co-exposure of N-OHAAF/UV irradiation to Balb/c-3T3 cells can induce intracellular generation of large amount of 8-oxodG and small amount of DNA DSBs, suggesting that our findings might be physiologically relevant.

Methionyl-AAF adducts formation was also induced both in l-methionine and l-methionyl-glycine by photoirradiation of N-OHAAF

Previous studies indicated that besides DNA adducts, methionyl-AAF adducts, namely 1- and 3-methylmercapto-2-acetylaminofluorene, were also induced by N-OHAAF enzymatically (31). Interestingly, the same two methionyl-AAF adducts were also produced when l-methionine (Met) or l-methionyl-glycine (Met-Gly) co-exposed to N-OHAAF/UV irradiation. Both methionyl-AAF adducts were identified by comparison with the mimetic enzymatically activated system (N-OAcAAF system) (Supplementary Figure 5, available at Carcinogenesis Online). l-methionine was involved in the synthesis of bioactive peptides and proteins, such as glutathione and taurine. Furthermore, l-methionine was reported to be the precursor of S-adenosyl-l-methionine, an important endogenous methyl donor in various methylation reactions, including the methylation of DNA, RNA, proteins and other molecules (32). Binding of N-OHAAF residues to the methylmercapto groups in proteins was thought to disturb DNA methylation and synthesis of related protein, which may further contribute to tumorigenesis. Therefore, photoactivation of N-OHAAF can induce versatile biological damage (modification of both DNA and proteins), which should be considered as another activation pathway for the carcinogenic arylhydroxamic acid.

What role does the ROS play in oxidative DNA damage induced by photoactivation of N-OHAAF?

Oxidative DNA damage is generally produced when ROS are generated by endogenous cellular metabolism or by exogenous sources such as ionizing radiation or carcinogenic compounds. To test for the involvement of ROS [such as hydrogen peroxide (H₂O₂), superoxide anion radicals (^•O₂^-), ^•OH and singlet oxygen (¹O₂)] in the DNA strand breakage and 8-oxodG formation induced by photoactivation of N-OHAAF, the reactions were carried out for comparison in the presence of catalase (CAT), superoxide dismutase (SOD) and typical ^•OH scavengers. Both DNA strand breakage and 8-oxodG formation were significantly inhibited by the typical ^•OH scavengers [dimethyl sulfoxide (DMSO), ethanol and azide] and the spin-trapping agent 5,5-dimethyl-1-pyrroline N-oxide (DMPO) but not by CAT or SOD (Figure 1c and Supplementary Figure 6, available at Carcinogenesis Online). The inhibitory effects on 8-oxodG formation increased with increasing concentrations of ^•OH scavengers (Supplementary Figure 6a, available at Carcinogenesis Online), indicating that 8-oxodG formation should result mainly from ^•OH production during irradiation. Besides, no difference was observed when the reactions were carried out in D₂O in the buffer instead of H₂O or with N₂ or O₂ gas bubbled for 5 min before irradiation (Supplementary Figure 7, available at Carcinogenesis Online). These results suggest that it should be ^•OH but not H₂O₂ or ^•O₂^- or ¹O₂ that was involved in oxidative DNA damage induced by photoactivation of N-OHAAF.

In addition, traces amount of transition metal ion and H₂O₂ (i.e., Fenton reaction) does not play any significant role on the ^•OH formation, since no changes were observed when the classic iron and copper chelating agent diethylene triamine pentaacetic acid was added to the reaction mixture.

^•OH production during photoactivation of N-OHAAF was identified by both ESR secondary spin-trapping and fluorescent methods

^•OH is considered in biology as the most reactive and harmful ROS, which can cause DNA, protein and lipid damage. Cancer, Parkinson’s disease and arthritis are but a few of the ailments that are linked to ^•OH (33,34). One of the most widely accepted mechanisms for ^•OH production is through the transition metal-catalyzed Fenton reaction (33). To examine whether ^•OH can be produced by UV irradiation of N-OHAAF, ESR spin-trapping method using DMPO as spin-trapping agent was first employed. Apart from Fenton reaction, N-hydroxypyridine-2(1H)-thione (NHPT) (35) and H₂O₂ (36) are good ^•OH sources via UV irradiation. By comparison with the two positive control systems, a quartet signal with an intensity ratio of 1:2:2:1 (a^H = a^N = 14.9 G) was observed during UV irradiation of N-OHAAF with DMPO, which was identified as ^•OH adduct (DMPO/^•OH) (Figure 4a). Besides, ^•OH production by N-OHAAF photolysis increased with increasing irradiation and concentration of N-OHAAF (Figure 4c). As shown in Figure 4b, respective addition of DMSO, ethanol, formate and azide to the DMPO/N-OHAAF UV irradiation reaction system led to a marked decrease of the DMPO/^•OH signal and concomitant generation of DMPO/^•CH₃ (a^H = 23.2 G, a^N = 16.2 G), DMPO/^•CH(CH₃)OH (a^H = 22.9 G, a^N = 15.9 G), DMPO/^•COO^- (a^H = 18.0 G, a^N = 15.7 G) and DMPO/^•N₃ (a^H = 14.2 G, a^N1 = 14.7 G, a^N2 = 3.2 G), respectively, which also confirmed the ^•OH formation.

To further identify the ^•OH production during irradiation of N-OHAAF, fluorescent method was used in our following research, i.e. TPA was used as an ^•OH probe, since the TPA hydroxylation product 2-hydroxy-terephthalic acid (2-OH-TPA) showed strong fluorescence (37). As shown in Figure 4d, obvious fluorescence was observed from photolysis of N-OHAAF with TPA. Besides, we also found that intensity of fluorescence induced by UV irradiation of N-OHAAF/TPA increased with increasing irradiation and concentration of N-OHAAF (Figure 4e), and the fluorescence can be inhibited considerably by ^•OH scavengers including DMSO and ethanol (Figure 4f).

Taken together, both ESR spin-trapping and fluorescent methods demonstrated clearly that ^•OH was indeed produced during irradiation of N-OHAAF and should be responsible for the oxidative DNA damage caused by irradiation of N-OHAAF, which can be considered as a new activation pathway for the genotoxicity of N-OHAAF.

Oxidative DNA damage by N-OHAAF/UV irradiation might be due to ^•OH generated site-specifically near the binding sites between DNA and N-OHAAF

As shown in Figure 1c, the DNA strand breakages induced by irradiation of N-OHAAF was only partly inhibited by addition of excessive ^•OH scavengers. Besides, the protection by ^•OH scavengers on the photoinduced 8-oxodG formation was 15–30% less in ct-DNA system than in simple nucleoside dG system (Supplementary Figure 6b, available at Carcinogenesis Online). These results indicated that such photoinduced DNA damage might be partly site-specific because of interactions between DNA and N-OHAAF. To verify this hypothesis, we used the classic Fenton reaction [Fe(II)-EDTA/H₂O₂] for further comparison. Interestingly, we found that although DMSO can provide only partial protection against DNA strand breakage induced by irradiation of N-OHAAF, it can provide full protection induced by the Fenton system, even though the DNA damage in the latter case were much more serious (Figure 1d). It was reported that only very weak interactions exist between DNA and Fe(II)-EDTA (21), so the ^•OH produced by Fe(II)-EDTA/H₂O₂ system was equally dispersed in reaction solution, which would attack on DMSO and DNA randomly and non-discriminatively; and thus only a small number of these ^•OH can be quenched by DNA. However, in the photoactivated N-OHAAF system, the ^•OH might be produced in a site-specific manner due to stronger binding between N-OHAAF and DNA, so the ^•OH may be produced near the N-OHAAF-binding site of DNA, which could not be quenched effectively by DMSO.

Interestingly, it has been shown by Vlastimil’s group that there is a weak intercalative binding between ct-DNA and nitro derivatives of fluorene, namely NF and 2,7-dinitrofluorene by electrochemical DNA-based biosensors (38). Therefore, we speculate that N-OHAAF, as a close derivative of NF, may also interact with DNA in a similar manner. UV-visible absorption spectroscopy is one of the most convenient and common techniques for the study of the interaction between small molecules and nucleic acids. It was reported that binding commonly results in hypochromism and a wavelength shift of the intercalated molecules (39) and intercalation between the DNA base pairs is a common binding type for aromatic compounds with a planar moiety (40). Our results showed that a characteristic absorbance of N-OHAAF at 302 nm was reduced by the addition of ct-DNA (Supplementary Figure 8, available at Carcinogenesis Online). We speculated that π orbits stacking of N-OHAAF and the bases may partly reduce the π–π* transition probabilities, resulting in hypochromism. Besides, a little red-shift phenomenon was observed, indicating that partial intercalation may occur between N-OHAAF and DNA.

Would the proposed binding of N-OHAAF to DNA induce some site-specificity in the formation of DNA damage? To further answer this question, we investigated the site-specificity of DNA cleavage using both HEX-5′-end labeled 23mer single- and double-stranded DNA fragments (Figure 5). We found that after UV irradiation with N-OHAAF, followed by hot piperidine treatment, DNA cleavage occurred the most at guanine (G), less at thymine (T) and cytosine (C) and the least at adenine (A) residues. These results provided more direct experimental evidence to support that DNA damage induced by N-OHAAF irradiation was possibly or at least partially through a site-specific mechanism, in which ^•OH reacts immediately upon formation close to their site of generation, mainly G and T sites. This could readily explain why even excessive DMSO had only partial protective effect on N-OHAAF/irradiation-induced oxidative DNA damage in this study.

Analysis of photolysis products of N-OHAAF by high-performance liquid chromatography (HPLC) and HPLC–MS

The colorless N-OHAAF solution turned yellow after irradiation in phosphate buffer solution. HPLC analysis indicated that only 1% N-OHAAF was left after irradiation for 30 min. Complex photoproducts were separated by HPLC, among which AAF with yield around 50% (peak 2) and 7-hydroxy-2-acetylaminofluorene with yield less than 2% (peak 1) were identified with authentic standards (Supplementary Figure 9a, available at Carcinogenesis Online). HPLC–MS analysis indicated several isomers of N-OHAAF, i.e. x-hydroxy-2-AAFs (x = 1–9, where OH group should be on fluorene ring) were also formed (Supplementary Figure 9b, available at Carcinogenesis Online). Based on previous reports (7,41,42) and the UV-vis absorption spectra in our experiment, the other two isomers (peak 3 and 4) can be assigned as 5- and 3-OH-2-AAF, respectively. However, we were unable to fully identify other isomers of N-OHAAF due to their low yield and the lack of authentic standards.

Molecular mechanism for ^•OH production via N–OH homolysis by photoirradiation of N-OHAAF

The above results showed that N-OHAAF induced DNA damage when exposed to UV irradiation and ^•OH is directly responsible for oxidative DNA damage. Together with previous studies on photochemistry of hydroxamic acids and their derivatives (19,43), we speculated that ^•OH may be formed via the homolysis of N–OH in N-OHAAF upon irradiation. If so, we expect that the photoinduced ^•OH formation should be inhibited when the N–OH group in N-OHAAF is blocked. To further confirm this hypothesis, N-methoxy-AAF was synthesized (31). As expected, neither DNA strands breakage nor 8-oxodG was observed when DNA and dG were treated with N-methoxy-AAF plus irradiation (Supplementary Figure 10, available at Carcinogenesis Online). Meanwhile, our results also showed that AAF and 7-hydroxy-2-acetylaminofluorene, the main photoproducts of N-OHAAF, cannot induce DNA strand breakage or 8-oxodG formation under the same illumination conditions (Supplementary Figure 10, available at Carcinogenesis Online). In addition, fluorescence results showed that no fluorescence of 2-OH-TPA was observed during photolysis of N-methoxy-AAF, AAF or 7-hydroxy-2-acetylaminofluorene (Supplementary Figure 10c, available at Carcinogenesis Online). These results further confirmed that it was direct photolysis of N-OHAAF that induced ^•OH production and further DNA damage. Furthermore, the bond dissociation energies (BDEs) of the N–O bond and O–H bond for N-OHAAF were calculated with density functional method at the B3LYP/6-31G(d,p) level. Our computational results showed that BDE₂₉₈(N–O) and BDE₂₉₈(O–H) in N-OHAAF are 206.3 and 289.9 kJ/mol, respectively, which are well correlated with the reported data (44). These results also suggest that it should be much easier for homolysis of N–O bond than O–H bond upon irradiation.

These findings demonstrated that ^•OH was generated from homolytic scission of the N–OH bond in N-OHAAF upon irradiation. If so, another question arises: amidyl radicals intermediate should also be readily produced simultaneously along with ^•OH through homolysis of N–OH. Besides amidyl radicals, carbon-centered radicals on the fluorene ring are also expected to be formed via spin isomerization. To our surprise, neither amidyl radicals nor carbon-centered radicals were observed by ESR spin-trapping method under present experiment conditions. Amidyl radicals were reported to display remarkably electrophilic and nucleophilic characters (45,46), indicating that amidyl radicals, especially with an electron-rich fluorene group, are too unstable to be detected under normal experimental conditions. However, we speculated that these radical intermediates should be produced and might be responsible for the formation of DNA adducts and methionyl-AAF adducts based on the following lines of indirect evidence: (i) AAF was found to be a major photoproduct of N-OHAAF, which indicates that AAF should be formed via H-abstraction by the amidyl radicals (Path a in Supplementary Scheme 2, available at Carcinogenesis Online). (ii) Another radical intermediate, i.e. the nitroxide radical [N(^•O)AAF] was also detected by ESR method when N-OHAAF was irradiated under stronger irradiation in acetonitrile and was confirmed by comparison with the typical one-electron oxidation reaction of N-OHAAF by Cu(II)-bathocuproine disulfonate complex (Supplementary Figure 11, available at Carcinogenesis Online). We speculated that N(^•O)AAF may be produced via H-abstraction by amidyl radical when excessive N-OHAAF in the solution was the hydrogen donor (Path b in Supplementary Scheme 2, available at Carcinogenesis Online). (iii) Different isomers of N-OHAAF such as 7-, 5-, 3-OH-2-AAF were detected. These results indicate that hydroxylation isomers may be formed via the recombination of ^•OH with carbon-centered radicals on the fluorene ring (Path c in Supplementary Scheme 2, available at Carcinogenesis Online).

Discussion

Based on the above results, a unique mechanism for site-specific DNA damage and ^•OH production by UV irradiation of N-OHAAF was proposed (Scheme 1): UV irradiation of N-OHAAF in aqueous media induces homolysis of the N–OH bond in N-OHAAF, resulting in the formation of ^•OH and amidyl radicals. The highly reactive ^•OH could cause oxidative damage including DNA strand breaks and 8-oxodG formation. Due to interactions between DNA and N-OHAAF, ^•OH can be produced site-specifically, which could attack adjacent DNA constituents, mainly G, T and C sites, before being scavenged by ^•OH scavengers. The potential selectivity for reaction at certain sequences in DNA might partly contribute to the extraordinary genotoxicity of N-OHAAF. Besides, the amidyl radicals could be spin-isomerized to form carbon-centered radicals, both of which can interact with biological macromolecules including DNA and proteins to form the respective covalent adducts.

It is worth noting that our results showed that oxidative DNA damage induced by co-exposure of N-OHAAF/UV irradiation occurs not only in isolated and purified DNA (plasmid DNA and ct-DNA) but also in a skin cell model, namely fibroblast Balb/c-3T3 cells. Since N-OHAAF is one of the metabolites of NF, our results may partly explain the underlying mechanism for the phototoxicity of NF in animals (20). IARC reported that the concentration of NF in the air reached up to 210 pg/m³ in Tokyo urban and 186 μg/g in the exhaust of diesel passenger cars (7). It is expected that the concentration of NF and other nitro-PAHs in the air must be increasing with the increasing number of vehicles and worldwide serious air pollutions. Thus, the potential exposure of N-OHAAF via biotransformations of NF in vivo should also be increasing consistently. Therefore, even though the concentrations of N-OHAAF via direct exposure may be lower than that used in our study, the mechanism for the photoactivation of N-OHAAF we found in this study should provide a new perspective for better understanding the potential phototoxicity of NF and its carcinogenic derivatives. Further investigations are needed to study whether this is the case in animal model and in human beings.

More interestingly, ^•OH formation and DNA damage were observed when we expanded the light source from artificial xenon lamp to sunlight (Supplementary Figure 12, available at Carcinogenesis Online) and when N-OHAAF was substituted by other arylhydroxamic acids such as N-hydroxy-4-acetylaminobiphenyl (Supplementary Figure 13, available at Carcinogenesis Online), a carcinogenic metabolite of 4-aminobiphenyl (ABP) (47). Besides occupational exposure, the main sources of exposure to ABP for the general population are cigarette smoking and second-hand tobacco smoke, as ABP is formed during tobacco combustion. Besides, ABP has also been detected in hair dyes and fumes from cooking oils (48). NHPT has been reported as a well-known photochemical ^•OH source and therefore induced oxidative DNA damage (49,50). Herein, we compared the phototoxic effects of NHPT, N-OHAAF and N-hydroxy-4-acetylaminobiphenyl on the ^•OH production and oxidative DNA damage. Our results indicated that N-OHAAF showed more phototoxic potential than NHPT in ^•OH production and 8-oxodG formation (Supplementary Figure 13, available at Carcinogenesis Online), and therefore the phototoxicity of N-OHAAF should not be overlooked. This suggests that the long-term co-exposure to sunlight and carcinogenic AAs may lead to more serious adverse effects in the body, compared to single exposure. Based on these results and considering the ubiquitous occurrence of AAs in the environment, we suggested that photoactivation may play an important role in the carcinogenesis of AAs and special attention should be paid to the potential phototoxicity of the AAs carcinogens, especially with regard to skin cancers caused by AAs.

In summary, we proposed that in addition to classic enzymatic activation, photoactivation of carcinogenic arylhydroxamic acid intermediate may also play an important role in the carcinogenicity of AAs, especially in the skin. Arylhydroxamic acids can be activated by photoirradiation to cause multiple DNA damage including DNA strand breaks, 8-oxodG and dG adducts formation. Compared to the classical arylnitrenium ion formed via enzymatic activated N–O heterolysis, reactive free radical intermediates such as ^•OH and amidyl radicals are produced via photoactivated N–OH bond homolysis. Therefore, our research provides a new free radical perspective to understand the molecular mechanism for the carcinogenicity of AAs.

Funding

Strategic Priority Research Program of Chinese Academy of Sciences Grant (XDB01020300); National Science Foundation China Grants (21836005, 21577149, 21477139, 21621064, 21407163, 21777180); National Institutes of Health (ES11497, RR01008 and ES00210).

Conflict of Interest Statement: None declared.

Abbreviations

Abbreviations
 
• 2-OH-TPA

  2-hydroxy-terephthalic acid

 
• 8-oxodG

  8-oxo-2′-deoxyguanosine

 
• AA

  aromatic amine

 
• AAF

  2-acetylaminofluorene

 
• ABP

  4-aminobiphenyl

 
• CAT

  catalase

 
• ct-DNA

  calf thymus DNA

 
• dG-C8-AAF

  N-(deoxyguanosin-8-yl)-2-acetylaminofluorene

 
• dG-C8-AF

  N-(deoxyguanosin-8-yl)-2-aminofluorene

 
• DMPO

  5,5-dimethyl-1-pyrroline N-oxide

 
• DMSO

  dimethyl sulfoxide

 
• DSB

  double strand break

 
• ESR

  electron spin resonance

 
• H₂O₂

  hydrogen peroxide

 
• HPLC

  high-performance liquid chromatography

 
• MS

  mass spectrometry

 
• NF

  2-nitrofluorene

 
• NHPT

  N-hydroxypyridine-2(1H)-thione

 
• N-OAcAAF

  N-acetoxy-AAF

 
• N-OHAAF

  N-hydroxy-2-acetamidofluorene

 
• ^•OH

  hydroxyl radical

 
• PAH

  polycyclic aromatic hydrocarbon

 
• ROS

  reactive oxygen species

 
• SOD

  superoxide dismutase

 
• TPA

  terephthalic acid

References