Photochemical and Photobiological Studies of Tirapazamine (SR

Tirapazamine, 3-amino-1,2,4-benzotriazine 1,4-di-N-oxide (TPZ; SR 4233), ... photochemical properties of TPZ, and the related analogues 3-amino-2- ...
0 downloads 0 Views 162KB Size
Download PDF
164

Chem. Res. Toxicol. 2003, 16, 164-170

Photochemical and Photobiological Studies of Tirapazamine (SR 4233) and Related Quinoxaline 1,4-Di-N-oxide Analogues J. Johnson Inbaraj, Ann G. Motten, and Colin F. Chignell* Laboratory of Pharmacology & Chemistry, NIEHS, Research Triangle Park, North Carolina 27709 Received August 26, 2002

Tirapazamine, 3-amino-1,2,4-benzotriazine 1,4-di-N-oxide (TPZ; SR 4233), is currently undergoing phase II and III clinical trials as an antitumor agent. We have studied the photochemical properties of TPZ, and the related analogues 3-amino-2-quinoxalinecarbonitrile 1,4-di-N-oxide (TPZCN) and quinoxaline-1,4-di-N-oxide (quindoxin) with respect to their potential to photodamage DNA both oxidatively and reductively. We have found that TPZ, TPZCN, and quindoxin photosensitized the generation of singlet oxygen with quantum yields of 0.007, 0.19, and 0.02, respectively, in acetonitrile. Irradiation (λ > 300 nm) of TPZ at pH 9.4 in the presence of a reducing agent, NADH, generated the corresponding nitroxide radical. At pH 7.4, photoirradiation of either TPZ or TPZCN in the presence of NADH in air saturated buffer gave the superoxide radical, which was trapped by 5,5-dimethyl-1-pyrroline N-oxide (DMPO). In the absence of a reducing agent, singlet oxygen generated from TPZCN oxidized DMPO to 5,5-dimethyl-2-oxopyrrolin-1-oxyl (DMPOX). No spin adducts were detected during photoirradiation of TPZ, NADH, and DMPO in nitrogen-saturated buffer. However, when DMSO was also present, the DMPO/•CH3 adduct was observed, indicating the generation of the free hydroxyl radical. Both TPZ and TPZCN photooxidized reduced glutathione and azide to the glutathiyl and azidyl radicals, respectively. Under anaerobic conditions, NADH increased photoinduced strand breaks in pBR322 plasmid DNA caused by TPZ or TPZCN. For TPZ, the reactive species is probably the aforementioned nitroxide radical or the hydroxyl radical generated from its decomposition. In contrast, DNA damage by quindoxin was not affected by NADH, suggesting a different mechanism, possibly involving a photogenerated oxaziridine intermediate. These studies show that the photochemistry of TPZ, TPZCN, and quindoxin is complex and depends on the redox environment and whether oxygen is present.

Introduction Tirapazamine, 3-amino-1,2,4-benzotriazine 1,4-di-Noxide (TPZ;1 also known as SR 4233), is currently undergoing phase II and III clinical trials as an antitumor agent. Recently, it has been reported that TPZ and the related analogue 3-amino-2-quinoxalinecarbonitrile 1,4-di-N-oxide (TPZCN) cause light-dependent DNA damage in plasmid DNA and in oligonucleotides, respectively (1, 2).

Photodamage to DNA as mediated by TPZ or TPZCN could be either oxidative or reductive and could poten* To whom correspondence should be addressed. Fax: (919) 5415750. E-mail: [email protected]. 1 Abbreviations: DMPO, 5,5-dimethyl-1-pyrroline N-oxide; DMPOX, 5,5-dimethyl-2-oxopyrrolin-1-oxyl; EPR, electron paramagnetic resonance; GSH, reduced glutathione; hfsc, hyperfine splitting constant(s); SOD, superoxide dismutase; ssb single strand break(s); TPZ, 3-amino-1,2,4-benzotriazine 1, 4-N-dioxide; TPZCN, 3-amino-2-quinoxalinecarbonitrile 1,4-di-N-oxide; quindoxin, quinoxaline-1,4-di-Noxide.

tially involve several mechanisms. Oxidatively, a photosensitizer can react with DNA via direct electron or hydrogen abstraction (Type I) (3-6), or generate singlet oxygen from quenching of the sensitizer’s triplet state by molecular oxygen (Type II) (7, 8). In addition, the photolytic deoxygenation of heterocyclic N-oxides is a well-known process that may involve oxaziridine intermediates, potent oxidizing agents that might be capable of reacting with DNA (9). In the case of TPZCN, previous results have suggested the participation of singlet oxygen: when oligonucleotides were photolyzed in the presence of TPZCN, alkali-dependent strand scission was produced at all 2′-deoxyguanosines, and this photocleavage increased markedly when the photolysis reaction was performed in D2O (1). However, until now there has been no direct evidence for the generation of 1O2 by TPZCN. Reductive DNA damage and photodamage are more likely to take place in an anaerobic environment. In solid tumors, TPZ is thought to manifest its cytotoxic activity by selectively killing oxygen-poor (hypoxic) cells. Under hypoxic conditions, TPZ is known to undergo enzymatic one electron reduction in the presence of cellular reductases such as NADPH-cytochrome P450 reductase to give a radical anion which in its protonated form has been identified as a nitroxide radical (Scheme 1) (10). It has been proposed that TPZ metabolites arise from the bioreduction of tirapazamine (11, 12).

10.1021/tx0256073 CCC: $25.00 © 2003 American Chemical Society Published on Web 01/03/2003

Studies of Tirapazamine (SR 4233) Scheme 1

Our goal in the present work was to clearly define the photochemical properties of TPZ, TPZCN, and the related analogue quinoxaline-1,4-di-N-oxide (quindoxin) with respect to their potential to damage DNA both oxidatively and reductively. To that end, we have shown that these N-oxides generate 1O2 and have measured their quantum yields. We have also used electron paramagnetic resonance (EPR) spectroscopy to identify the free radicals generated photochemically from these compounds, radicals that may participate in their cytotoxic activity. Because the radical yield from TPZ and TPZCN increased in the presence of reducing agents, we also examined the effect of NADH on the photocleavage of plasmid DNA by these N-oxides.

Materials and Methods TPZ was purchased from Sigma (St. Louis, MO) and used without further purification. TPZCN and quindoxin were synthesized in our laboratory following literature procedures (13, 14). Unless otherwise specified, a saturated solution of TPZCN (∼2 mg/mL) in pH 7.4 50 mM phosphate buffer solution was used for all experiments. Plasmid supercoiled pBR322 DNA was purchased from Boehringer-Mannheim (Indianapolis, IN). Glutathione reduced form (GSH) was from Sigma Chemical Company (St. Louis, MO). 5,5-Dimethyl-1-pyrroline N-oxide (Aldrich Chemical Co., Milwaukee, WI) was purified by vacuum distillation and stored frozen until required. All other chemicals were of reagent grade or better. Superoxide dismutase (from bovine erythrocytes) was obtained from Boehringer Mannheim, Germany. Absorption Spectra. Absorption spectra were recorded using an HP diode array spectrophotometer model 8452A (Hewlett-Packard Co., Palo Alto, CA). The relative number of absorbed photons at the excitation wavelength was calculated using the Beer-Lambert law. Detection of Singlet Oxygen. Photosensitized production of 1O2 from TPZ, TPZCN, and quindoxin was calculated from its characteristic infrared (IR) phosphorescence at 1270 nm (15). Singlet oxygen phosphorescence was measured on a steady-state 1O spectrophotometer featuring an optimized optical system 2 as in our pulse 1O2 spectrophotometer. TPZ, TPZCN, and quindoxin were excited with a 500 W mercury lamp operating at 300 W through interference filters transmitting at 480, 480, and 366 nm, respectively. The 1O2-phosphorescence spectra were recorded over the range of 1200-1350 nm and were normalized to the same number of absorbed photons at the excitation wavelength.

Chem. Res. Toxicol., Vol. 16, No. 2, 2003 165 EPR Measurements. EPR spectra were recorded at 9.5 GHz on a Varian E-109B spectrometer equipped with a TM cavity using a quartz aqueous flat cell. Unless otherwise indicated, the following instrumental settings were used: microwave power, 10 mW; modulation, 0.33 G; time constant, 0.25 s; scan time, 4 min. For some experiments, the solutions were introduced into the EPR cell and gassed gently with nitrogen for 10 min. The samples were irradiated directly inside the microwave cavity of the spectrometer using a 1 kW xenon lamp whose light was passed through a cutoff filter transmitting above 300 nm. For air-saturated samples, only the initial spectra are presented so that the effect of any decrease in oxygen concentration was minimized. Spectra were accumulated on a PC and simulated using software described elsewhere (16). DNA Cleavage. Irradiations were performed on samples containing 15 ng/µL of pBR322 DNA (∼9 nM) and drug concentrations of 50 µM in phosphate buffer (50 mM) solution at pH 7.4. Samples were illuminated using a 500 W Hg-Xe lamp with a 320 nm cutoff filter. Samples were contained in a quartz cell (0.5 cm path length) that was placed at a distance of 12 cm from the irradiation source. The samples were purged with N2 or O2 for 10 min before and during irradiation. Instead of examining multiple samples after one irradiation time, we tested single samples after a series of irradiation times. After each irradiation time, 10 µL of the sample was withdrawn for analysis. Following photolysis, products resulting from the photocleavage of the pBR322 DNA (i.e., relaxed and supercoiled forms) were separated by electrophoresis in 1% agarose gels [highstrength analytical grade, Bio-Rad (Hercules, CA)] using a Horizon 58 horizontal gel electrophoresis system with a model 200 power supply from Life Technologies, Inc. (Gaithersburg, MD). Gels were electrophoresed for approximately 1 h at 100 V and then stained with aqueous ethidium bromide for 1-2 h. DNA in the gel was visualized by UV-transillumination. The gels were imaged using an IS-1000 digital imaging system from Alpha Innotech (San Leandro, CA). The integrated areas of the relaxed DNA were divided by 1.66 to correct for the higher fluorescence of the ethidium bromide when bound to this form as compared to the supercoiled DNA. The fractions of supercoiled DNA after each irradiation interval were then plotted as a function of time. Spin-Density Calculation. The spin-density calculations were performed using the Austin Method 1 (AM1) in the program MOPAC from QCPE (version 6.0).

Results and Discussion Photobleaching Studies. We first examined the photobleaching of TPZ, TPZCN, and quindoxin. When TPZ and TPZCN were irradiated at 480 nm in phosphate buffer solution (pH 7.4), the changes in their absorption spectra were negligible, indicating little or no decomposition of these two compounds during aerobic irradiation. However, when quindoxin was irradiated (366 nm), a new peak appeared at 306 nm (Figure 1). The latter was attributed to 2-hydroxyquinoxaline-4-oxide (1), formed by the tautomerization of quinoxaline-2-one 4-oxide (2) (17) via an oxaziridine intermediate (Scheme 2). It is wellknown that oxaziridines are efficient oxidizing agents (9) and might be capable of causing DNA cleavage. Singlet Oxygen Generation. To test for oxidizing mechanisms, we measured the production of singlet oxygen, which is known to cause DNA cleavage (7, 8). No 1O2 was detected during TPZ or TPZCN irradiation in aqueous buffer. However, we found that TPZ, TPZCN, and quindoxin photosensitized the generation of 1O2 with quantum yields of 0.007, 0.19, and 0.02, respectively, in acetonitrile. The higher quantum yield of TPZCN compared to TPZ is consistent with the observation that

166

Chem. Res. Toxicol., Vol. 16, No. 2, 2003

Inbaraj et al.

Figure 1. Absorption spectra of TPZCN and quindoxin (50 µM) in 50 mM phosphate buffer at pH 7.4 before and after irradiation at λmax ) 480 and 366 nm, respectively. The absorption maximum at 306 nm corresponds to 2-hydroxyquinoxaline-4oxide formed via quindoxin-derived oxaziridine intermediate (Scheme 2) and is marked with a dotted arrow.

Scheme 2 Figure 2. EPR spectrum of the DMPO radical adducts during the irradiation (λ >300 nm) of an air-saturated solution containing TPZ and DMPO. (A) Solution contained TPZ (4 mM) and DMPO (100 mM) in 50 mM phosphate buffer, pH 7.4, in the presence of NADH (1 mM). (B) Computer simulated spectrum, using parameters given in Table 1. (C) DMPOX radical generated during irradiation of TPZCN in air-saturated solution of phosphate buffer, pH7.4, containing 100 mM DMPO. (D) After light off. (E) Simulated spectrum of nitroxide radical generated from TPZCN using hfsc given in Table 2.

TPZCN damages DNA more efficiently than TPZ under aerobic conditions (1). The low quantum yield of 1O2 generation by TPZ suggests that the TPZ triplet lifetime is very short and/or the quantum yield of the TPZ triplet is low. This may explain why Poole and co-workers were unable to detect the TPZ triplet via phosphorescence or EPR spectroscopy (18). EPR Studies. EPR studies were carried out to investigate both oxidizing and reducing radical species. During aerobic photolysis of TPZ or TPZCN and NADH in the presence of 5,5-dimethyl-1-pyrroline N-oxide (DMPO) in phosphate buffer solution (pH 7.4), we observed a multiline spectrum characteristic of the DMPO/O2•- adduct together with the DMPO/•OH adduct (Table 1, Figure 2A). For quindoxin, only DMPO/•OH was observed (Table 1). The addition of superoxide dismutase (SOD; 50 µg/ mL) sharply reduced the DMPO/O2•- and/or DMPO/•OH

adducts from TPZ and quindoxin (not shown), demonstrating that these adducts were derived from the trapping of superoxide. It is well-known that the DMPO/O2•adduct decays to the DMPO/•OH adduct (19). Although Poole and co-workers have proposed that the TPZ radical decomposes to release the •OH radical (18) (Scheme 1), we were unable to detect any DMPO/•OH adduct during photoirradiation of an anaerobic solution of TPZ and DMPO in the presence of NADH. However, hydroxyl radical derived nitroxide adducts of nitrones are generally not as stable as the corresponding carboncentered adducts (20, 21). It is well-known that the •OH radical reacts with DMSO to generate the methyl radical at a rate that is close to that of its reaction with DMPO (22, 23). When we repeated the experiment in the presence of DMSO (1.25 M), we observed a six line EPR spectrum, which was identified as the DMPO/•CH3 adduct (Table 1). No DMPO/•CH3 adduct was seen when NADH was omitted. This finding suggests that the •OH radical is indeed generated during the photoirradiation of an anaerobic solution containing TPZ, DMPO, and NADH. In the case of TPZCN, during irradiation with DMPO in aerated phosphate buffer solution (pH 7.4), we observed the DMPO oxidation product 5,5-dimethyl-2oxopyrrolin-1-oxyl (DMPOX; Table 1, Figure 2C). In a related system containing Rose Bengal and DMPO, DMPOX was generated during photoirradiation via the reaction of singlet oxygen with the spin trap (24), suggesting that DMPOX formed during TPZCN irradia-

Studies of Tirapazamine (SR 4233)

Chem. Res. Toxicol., Vol. 16, No. 2, 2003 167

Table 1. Types of Radical Trapped and Hyperfine Coupling Constants of the Resultant DMPO Adducts Generated during the UV Irradiation of TPZ, TPZCN, and Quindoxin in Phosphate Buffer Solution (pH 7.4) irradiated compd TPZ

TPZCN

Quindoxin

addition

saturated with

none

air

none NADH NADH + DMSO NaN3

N2 N2 N2 air

GSH

air

none

air (light on) light off

NADH

air

GSH

air

NaN3 none GSH NaN3

air air air air

tion may also be derived from reaction of DMPO with 1 O2. Consistent with this mechanism, the DMPOX radical was completely eliminated by the addition of sodium azide (3 mM), a 1O2 scavenger (not shown). After the light was turned off, a weak multiline spectrum remained. The major component was identified as the TPZCN radical (83%) detected directly (Figure 2D; Table 2). In addition, minor contributions from DMPO/O2•- (5%), DMPO/•OH (2%), and an unidentified adduct (10%) were present with hfsc given in Table 1. The following mechanism, in which the reducing equivalents are supplied by DMPO itself, may explain our results:

[TPZCN]* + DMPO f TPZCN•- + DMPO•+

radical trapped O2••OH no adduct no adduct •CH 3 N3• O2•GS• •OH DMPOX TPZCN•(see Table 2) O2••OH unidentified O2••OH GS• •OH N3• •OH •OH N3• •OH

aN

hyperfine coupling constants (G) aH aX

14.25 15.19

11.39 15.00

16.55 15.00 14.35 15.40 15.16 7.40

23.71 14.45 11.54 16.3 14.95 4.10

14.08 15.35 14.12 14.40 15.35 15.40 15.19 14.95 15.19 15.19 14.95 15.24

11.61 14.85 8.67 11.45 15.00 16.30 15.00 14.40 15.00 15.00 14.40 15.19

1.24 (H)

3.17 (N) 1.28 (H)

1.97 (H) 1.22 (H)

3.17 (N) 3.17 (N)

nant aerobic mechanism appears to involve an oxaziridine intermediate, which is known to be a powerful oxidizing species (9). During photoirradiation of TPZ or TPZCN in the presence of reduced glutathione (GSH) and DMPO in air saturated phosphate buffer solution (pH 7.4), we observed a signal from the DMPO/GS• adduct and a minor contribution from the DMPO/•OH adduct (Table 1, Figure 3C). No adducts were observed when TPZ/TPZCN or GSH was omitted. Thus, both TPZ and TPZCN are capable of photooxidizing reduced GSH. At pH 7.4, no radicals were detected during anaerobic irradiation of TPZ in the absence of spin trap. However, Laderoute et al., have reported that the half-life of the radical generated from TPZ increased with increasing pH

DMPO•+ + H2O f DMPO/•OH + H+ TPZCN•- + O2 f TPZCN + O2•Photoirradiation of an aerated solution of TPZCN, DMPO, and NADH produced a multiline EPR spectrum characteristic of DMPO/O2•- together with the DMPO/•OH adduct (Table 1). Thus in the presence of a reducing agent, the electron-transfer process predominates (Scheme 1). To investigate the photooxidizing properties of the N-oxides, we examined the effects of added sodium azide and GSH in aerobic solution. When we irradiated a pH 7.4 solution containing quindoxin with NaN3 (3 mM) and DMPO, we obtained an EPR spectrum characteristic of the DMPO/•N3 adduct (Table 1, Figure 3A). The azidyl radical may be generated from azide by reaction with the • OH radical or some other oxidizing species. However, in the case of TPZCN or TPZ, no DMPO/•N3 adduct was observed until the concentration of NaN3 was increased to 100 mM (Table 1). No radical adducts were seen when TPZ or TPZCN were omitted. Under aerobic conditions, oxygen may compete more effectively than azide for the excited states of TPZ and TPZCN when the concentration of azide is low. By contrast, for quindoxin the predomi-

Figure 3. EPR spectrum of the DMPO radical adduct generated during the irradiation (>300 nm) of an air saturated 50 mM phosphate buffer solution (pH 7.4) in the presence of (A) quindoxin (5 mM) + 3 mM NaN3. (B) Simulated EPR spectrum of A using hfsc in Table 1. (C) TPZCN + GSH (1 mM). The instrumental condition as follows: microwave power, 10 mW; modulation amplitude, 0.33 G; time constant 0.25 G; scan time 4 min; scan range, 100 G.

168

Chem. Res. Toxicol., Vol. 16, No. 2, 2003

Inbaraj et al.

Table 2. Hyperfine Splittings of Radicals Derived from TPZ, TPZCN, and Quindoxin

(25). Therefore, we chose pH 9.4 (borate buffer) to investigate radicals produced under anaerobic conditions. Anaerobic TPZ photoirradiated in the presence of NADH or GSH produced a TPZ-derived free radical (Figure 4A) similar to that observed during microsomal reduction of TPZ (10). For GSH, the reducing species may be GSH itself or the GSSG•- radical anion. In the absence of reducing agents, no free radicals were detected. The spectrum in Figure 4A was attributed to the TPZ nitroxide radical (Scheme 1) and could be simulated using the hfsc in Table 2 (see Figure 4B). Our assignments are consistent with the calculated spin densities (Table 2). The two 3.0 G splittings, previously assigned to the 3-NH2 protons (10), are reassigned to the aromatic ring hydrogens based on our finding that the EPR spectrum of the TPZ radical was unchanged in D2O (not shown). No nitroxide radicals were detected for TPZCN, even in the presence of NADH or GSH. When quindoxin was photoirradiated in anaerobic borate buffer solution (pH 9.4) in the presence of the reducing agent NADH, a strong EPR spectrum was observed (Figure 4C). It is well-known that the photolysis of quinoxaline-di-N-oxides generates N-oxyl radicals via hydrogen abstraction from a solvent molecule (e.g., EtOH) by the photoexcited lowest triplet state (26), hence it is reasonable to assume that photoexcited quindoxin is reduced by NADH to generate the corresponding N-oxyl radical (Scheme 1, Table 2). However, chemical

reduction of quindoxin produced an entirely different radical. When reduced by fructose (0.1 M) in 0.1 M NaOH solution, the quindoxin-derived radical (Figure 4E) had hyperfine coupling constants markedly different from the photogenerated radical, but very similar to those obtained by Kubota et al. (27) during controlled-potential electrolysis of quindoxin in dimethylformamide (Table 2). The quindoxin radical may exist in two forms as shown below:

MOPAC calculations of spin-density distribution in these two forms may be used to distinguish between them. Table 2 clearly shows that the spin-density distribution of the protonated (neutral) quindoxin radical is different from the anionic form. From the spin-density and hfsc values, we conclude that the quindoxin radical exists in the protonated state at pH 9.4, while at pH 13, it is ionized. Simulated EPR spectra for the two forms, using the corresponding hfsc in Table 2, are shown in Figure 4, panels D and F. DNA Photocleavage. In view of our observation that the presence of NADH changed the photochemistry of the

Studies of Tirapazamine (SR 4233)

Chem. Res. Toxicol., Vol. 16, No. 2, 2003 169

Figure 5. Cleavage of plasmid DNA (pBR322) by the N-oxides (50 µM) in 50 mM phosphate buffer solution (pH 7.4). Samples were irradiated (λ > 320 nm) for 4 min.

Figure 4. EPR spectrum and computer simulation of radical generated from TPZ and quindoxin during photoirradiation (>300 nm) in the presence of NADH (1 mM) in borate buffer solution at pH 9.4. (A) TPZ (2 mM); (B) Simulation of panel A using hfsc in Table 2. (C) Quindoxin (2 mM). (D) Simulation of panel C using hfsc in Table 2. (E) EPR spectrum of radical obtained when quindoxin (2 mM) was reduced by fructose (0.1 M) in 0.1 M NaOH solution. (F) Simulated spectrum using hfsc given in Table 2. The instrumental condition as follows: microwave power, 10 mW; modulation amplitude, 0.33 G; time constant 0.25 G; scan time 4 min; scan range, 100 G.

N-oxides, we investigated the effect of NADH on the photocleavage of DNA, by measuring the conversion of supercoiled plasmid DNA to the relaxed form (open circular), which occurs upon single strand cleavage. In this experiment we observed single strand break (ssb) in pBR322 DNA during the photoirradiation (λ > 320 nm) of TPZ, TPZCN, and quindoxin in phosphate buffer solution (pH 7.4) under aerobic and anaerobic conditions in the presence and absence of NADH (Figure 5). Photodamage to DNA alone was the same under aerobic and anaerobic conditions, although ssb did increase slightly in the absence of oxygen when NADH was present (Figure 5). In contrast DNA damage caused by photolysis of TPZ under anaerobic conditions was considerably enhanced by the addition of NADH (Figure 5). This observation, together with the EPR data (vide supra) supports the involvement of the TPZ radical in DNA damage. Photocleavage by TPZCN was slightly greater under aerobic conditions (Figure 5), possibly due to singlet oxygen production. Under anaerobic conditions, the presence of NADH caused a large increase in DNA damage consistent with a switch from Type II to a Type I mechanism. Quindoxin photodamaged DNA significantly in both aerobic and anaerobic environments (Figure 5). Although there was more damage in the presence of NADH under anaerobic conditions, the increase was small. Photoirra-

diation of heterocyclic N-oxides such as quindoxin may generate oxaziridine intermediates (17, 28), which are known to be potent oxidizing agents (9) and might be capable of interacting with DNA.

Conclusions In summary, the photochemistry of the N-oxides is complex and depends on the redox environment and whether oxygen is present. This study has demonstrated that, under anaerobic conditions, TPZ mediates DNA damage through an intermediate generated by one electron reduction (11). The reactive intermediate may be the corresponding TPZ nitroxide radical and/or the hydroxyl radical generated from the TPZ radical (Scheme 1) as proposed by Daniels and Gates (29). In vivo, TPZ is relatively nontoxic to oxygenated cells (30). In the presence of oxygen, the TPZ radical is backoxidized to the parent compound, thereby producing the superoxide anion radical (Scheme 1) (10, 25, 31) and hydrogen peroxide, whose cytotoxicities are minimized by cellular enzymes such as superoxide dismutase, glutathione peroxidase and catalase (30, 32). DNA photodamage by TPZCN in the absence of a reducing agent is probably mediated via singlet oxygen. Although TPZCN generated singlet oxygen in acetonitrile with high quantum yield, we were unable to detect this active oxygen species in aqueous solution where its lifetime is very short. Indirect evidence for the production of singlet oxygen in aqueous buffer was provided by the conversion of DMPO to DMPOX. The enhancement of DNA damage in the presence of NADH suggests that, in the presence of a reducing agent, the pathway of photosensitization may shift from a Type II to a Type I mechanism involving electron transfer. For quindoxin, in contrast to TPZ and TPZCN, ssb produced under both aerobic and anaerobic conditions probably occur via an oxaziridine intermediate. The small increase in DNA damage in the presence of NADH under

170

Chem. Res. Toxicol., Vol. 16, No. 2, 2003

anaerobic conditions reveals that there may also be some contribution from a nitroxide radical.

Acknowledgment. The authors wish to thank Dr. Peter Wardman for suggesting the use of alkaline fructose for the generation of the quindoxin radical.

References (1) Fuchs, T., Gates, K. S., Hwang, J. T., and Greenberg, M. M. (1999) Photosensitization of guanine-specific DNA damage by a cyano- substituted quinoxaline di-N-oxide. Chem. Res. Toxicol. 12, 1190-1194. (2) Daniels, J. S., Chatterji, T., MacGillivray, L. R., and Gates, K. S. (1998) Photochemical DNA cleavage by the antitumor agent 3-amino-1,2,4-benzotriazine 1,4-dioxide (tirapazamine, WIN 59075, SR4233). J. Org. Chem. 63, 10027-10030. (3) Armitage, B. (1998) Photocleavage of nucleic acids. Chem. Rev. 98, 1171-1200. (4) Breslin, D. T., and Schuster, G. B. (1996) Anthraquinone photonucleases: Mechanisms for GG-selective and nonselective cleavage of double-stranded DNA. J. Am. Chem. Soc. 118, 2311-2319. (5) Saito, I., Takayama, M., Sugiyama, H., and Nakatani, K. (1995) Photoinduced DNA cleavage via electron-transfer - Demonstration that guanine residues located 5′ to guanine are the most electron-donating sites. J. Am. Chem. Soc. 117, 6406-6407. (6) Atherton, S. J., and Harriman, A. (1993) Photochemistry of intercalated methylene blue - Photoinduced hydrogen atom abstraction from guanine and adenine. J. Am. Chem. Soc. 115, 1816-1822. (7) Sies, H. (1993) Damage to plasmid DNA by singlet oxygen and its protection. Mutat. Res. 299, 183-191. (8) Blazek, E. R., and Peak, J. G., and Peak, M. J. (1989) Singlet oxygen induces Frank strand breaks as well as alkali-labile and piperidine-labile sites in supercoiled plasmid DNA. Photochem. Photobiol. 49, 607-613. (9) Davis, F. A., and Sheppard, A. C. (1989) Applications of oxaziridines in organic-synthesis. Tetrahedron 45, 5703-5742. (10) Lloyd, R. V., Duling, D. R., Rumyantseva, G. V., Mason, R. P., and Bridson, P. K. (1991) Microsomal reduction of 3-amino-1,2,4benzotriazine 1,4-dioxide to a free radical. Mol. Pharmacol. 40, 440-445. (11) Laderoute, K. R., and Rauth, A. M. (1986) Identification of 2 major reduction products of the hypoxic cell toxin 3-amino-1,2,4-benzotriazine-1,4-dioxide. Biochem. Pharmacol. 35, 3417-3420. (12) Fuchs, T., Chowdhury, G., Barnes, C. L., and Gates, K. S. (2001) 3-Amino-1,2,4-benzotriazine 4-oxide: Characterization of a new metabolite arising from bioreductive processing of the antitumor agent 3-amino-1,2,4-benzotriazine 1,4-dioxide (Tirapazamine). J. Org. Chem. 66, 107-114. (13) Coperet, C., Adolfsson, H., Khuong, T. A. V., Yudin, A. K., and Sharpless, K. B. (1998) A simple and efficient method for the preparation of pyridine N-oxides. J. Org. Chem. 63, 1740-1741. (14) Monge, A., Palop, J. A., Delcastillo, J. C., Caldero, J. M., Roca, J., Romero, G., Delrio, J., and Lasheras, B. (1993) Novel antagonists of 5-HT3 receptors - Synthesis and biological evaluation of piperazinylquinoxaline derivatives. J. Med. Chem. 36, 2745-2750. (15) Hall, R. D., and Chignell, C. F. (1987) Steady-state near-infrared detection of singlet molecular-oxygen - A Stern-Volmer quenching experiment with sodium azide. Photochem. Photobiol. 45, 459-464. (16) Duling, D. R. (1994) Simulation of multiple isotropic spin-trap EPR Spectra. J. Magn. Reson. Ser. B 104, 105-110.

Inbaraj et al. (17) Kawata, H., Kikuchi, K., and Kokubun, H. (1983) Studies of the photoreactions of heterocyclic N-dioxides - Identification of the oxaziridine intermediate of quinoxaline- 1,4-dioxide. J. Photochem. 21, 343-350. (18) Poole, J. S., Hadad, C. M., Platz, M. S., Fredin, Z. P., Pickard, L., Guerrero, E. L., Kessler, M., Chowdhury, G., Kotandeniya, D., and Gates, K. S. (2002) Photochemical electron-transfer reactions of tirapazamine. Photochem. Photobiol. 75, 339-345. (19) Finkelstein, E., Rosen, G. M., and Rauckman, E. J. (1982) Production of hydroxyl radical by decomposition of superoxide spin-trapped adducts. Mol. Pharmacol. 21, 262-265. (20) Kadiiska, M. B., Burkitt, M. J., Xiang, Q. H., and Mason, R. P. (1995) Iron supplementation generates hydroxyl radical in-vivo - An ESR spin trapping investigation. J. Clin. Invest. 96, 16531657. (21) Stoyanovsky, D. A., Melnikov, Z., and Cederbaum, A. I. (1999) ESR and HPLC analysis of the interaction of hydroxyl radical with DMSO: Rapid reduction and quantification of POBN and PBN nitroxides. Anal. Chem. 71, 715-721. (22) Ross, A. B., Mallard, W. G., Helman, W. P., Buxton, G. V., Huie, R. E., and Neta, P. (1998) NDRL-NIST Solution kinetic database, Ver. 3.0, Notre Dame Radiation Laboratory, Notre Dame, IN. (23) Buxton, G. V., Greenstock, C. L., Helman, W. P., and Ross, A. B. (1988) Critical-review of rate constants for reactions of hydrated electrons, hydrogen-atoms and hydroxyl radicals (.OH/.O-) in aqueous-solution. J. Phys. Chem. Ref. Data 17, 513-886. (24) Bilski, P., Reszka, K., Bilska, M., and Chignell, C. F. (1996) Oxidation of the spin trap 5,5-dimethyl-1-pyrroline N-oxide by singlet oxygen in aqueous solution. J. Am. Chem. Soc. 118, 13301338. (25) Laderoute, K., Wardman, P., and Rauth, A. M. (1988) Molecular mechanisms for the hypoxia-dependent activation of 3- amino1,2,4-benzotriazine-1,4-dioxide (SR-4233). Biochem. Pharmacol. 37, 1487-1495. (26) Lin, S. K. (1988) Radical mechanism in the photoreaction of organic N-oxides - An electron spin resonance observation of nitroxyl radicals from photoreaction systems involving benzofurazan N-oxide. J. Photochem. Photobiol., A 45, 243-247. (27) Kubota, T., Nishikida, K., Miyazaki, H., Iwatani, K., and Oishi, Y. (1968) Electron spin resonance and polarographic studies of the anion radicals of heterocyclic amine N-oxides. J. Am. Chem. Soc. 90, 5080-5090. (28) Kurasawa, Y., Takada, A., and Kim, H. S. (1995) Progress in the chemistry of quinoxaline N-oxides and N,N′-dioxides. J. Heterocycl. Chem. 32, 1085-1114. (29) Daniels, J. S., and Gates, K. S. (1996) DNA cleavage by the antitumor agent 3-amino-1,2,4-benzotriazine 1,4-dioxide (SR4233): Evidence for involvement of hydroxyl radical. J. Am. Chem. Soc. 118, 3380-3385. (30) Elwell, J. H., Siim, B. G., Evans, J. W., and Brown, J. M. (1997) Adaptation of human tumor cells to tirapazamine under aerobic conditions - Implications of increased antioxidant enzyme activity to mechanism of aerobic cytotoxicity. Biochem. Pharmacol. 54, 249-257. (31) Wardman, P., Priyadarsini, K. I., Dennis, M. F., Everett, S. A., Naylor, M. A., Patel, K. B., Stratford, I. J., Stratford, M. R. L., and Tracy, M. (1996) Chemical properties which control selectivity and efficacy of aromatic N-oxide bioreductive drugs. Br. J. Cancer 74, S70-S74. (32) Sies, H. (1986) Biochemistry of oxidative stress. Angew. Chem., Int. Ed. Engl. 25, 1058-1071.

TX0256073