Glutathione Conjugates - ACS Publications - American Chemical Society

University of Maryland Medical School, 655 West Baltimore Street, Baltimore, Maryland 21201 .... system described below to follow the formation of con...
0 downloads 0 Views 1MB Size
Chem. Res. Toxicol. 1996, 8, 455-464

465

Diaziquone -Glutathione Conjugates: Characterization and Mechanisms of Formation Peter L. Gutierrez* and Sadhish Siva University of Maryland Cancer Center and Department of Biological Chemistry, University of Maryland Medical School, 655 West Baltimore Street, Baltimore, Maryland 21201 Received October 28, 1994@

The antitumor agent diaziquone ( A Z Q ) reacts with reduced glutathione (GSH) in aqueous solutions and under aerobic conditions to give rise to the glutathionyl and hydroxyl free radicals, as well a s the AZQ semiquinone. Under anaerobic conditions, the only radical observed was the glutathionyl radical. These radicals are quickly abrogated when superoxide dismutase and catalase are coincubated. Separately, superoxide dismutase favors the formation of thiyl radicals while catalase favors the formation of hydroxyl radicals and AZQ semiquinone. The metal chelator diethylenetriaminepentaacetic acid favors the production of hydroxyl radicals and AZQ semiquinone. The reaction of AZQ with GSH a t pH 7.2 and 5.5 results in a variety of conjugates. These conjugates include addition of glutathione t o both aziridines, displacement of the aziridines by GSH, and a combination of both. The majority of the conjugates are formed by nucleophilic attack of GSH to the AZQ aziridines or by 1,6Michael addition to the AZQ quinone or a combination of both. There may be a small free radical component in conjugate formation, but the majority of the free radicals observed are from redox reactions that involve the oxidation of glutathione and the reduction and autoxidation of AZQ to produce oxygen radicals and hydrogen peroxide, a process that is enhanced by trace metal ions.

Introduction Diaziquone [2,5-diaziridinyl-3,6-bis(carboethoxyamino)1,4-benzoquinone] (AZQY is a bioreductive alkylating agent which has shown clinical activity in brain tumors ( I ) and acute nonlymphocytic leukemia (2, 3). Studies on the mechanism of action of AZQ have suggested that its cytotoxicity is due to two pathways: (a) redox cycling which produces oxidative stress via oxygen radicals and hydrogen peroxide, and (b) aziridine alkylation that is enhanced by the 2-electron reduction of the quinone. Consistent with the first pathway is the ability of whole cells to reduce AZQ with the production of oxygen radicals and the semiquinone (4,5),the ability of reduced AZQ to increase the cytotoxicity of AZQ against cells in culture (61,and the ability of catalase to abrogate this activity (7). Consistent with the alkylating pathway is the 3-fold increase in alkylating activity shown by the (nitrobenzy1)pyridine assay when AZQ is reduced with borohydride (2 e- reduction) or with DT-diaphorase-rich cytosol (2 e- reduction) from MCF-7 cells (8). Similarly, M Q reduced with borohydride results in DNA alkylation (9-11).

Reduced glutathione is an important detoxifying sulfhydryl tripeptide that protects the cell against a variety of toxic challenges such as ionizing radiation, cytotoxins, hyperthermia, and the reactive products of reduced oxygen (12). Of particular interest to oncopharmacology is that GSH prevents myocardial toxicity in mice treated with cardiotoxic doses of doxorubicin (13).

* Author to whom correspondence should be addressed. FAX: 410328-6559; EMAIL: [email protected]. Abstract published in Advance ACS Abstracts, March 15, 1995. Abbreviations: AZQ, diaziquone; DMPO, 5,5-dimethyl-l-pyrroline

N-oxide; FABMS, fast atom bombardment mass spectrometry; SOD, superoxide dismutase; DETAPAC, diethylenetriaminepentaacetic acid; CID, collisionally induced dissociation.

The effect of GSH on the activity of antitumor agents has been documented (12,14)and implicated in de novo multidrug resistance (15). Mitomycin C-GSH adducts are not active against cells in culture (14), and GSH diminishes the interaction of this drug with DNA (16). In the case of AZQ, it has been shown that this drug depletes GSH in hepatocytes (17 ) and in reactions mediated by DT-diaphorase (18). In both cases, a concomitant increase of oxidized glutathione dimer GSSG was observed (17,181. In the case of hepatocytes, GSSG is not reduced back to GSH because AZQ reversibly inhibits glutathione reductase (17). Glutathione has been shown to react with a variety of quinones (reviewed in ref 19). The majority of these reactions result in hydroquinone-GSH conjugates, which are the products of a 1,4-reduced Michael addition of the nucleophile GSH to the quinone. Of interest is the mono-, di-, tri-, and tetraconjugates found in the reaction of benzoquinone with GSH which are nephrotoxic (20). When benzoquinone was incubated with rat hepatocytes, both the bensosemiquinone and the semiquinone of its GSH conjugate (2-(S-glutathionyl)-2,4~benzosemiquinone) were observed, indicating that quinone-GSH conjugates can redox cycle (21). Similar results were obtained when menadione and related naphtoquinones were reacted nonenzymatically with GSH (22-26). These reactions resulted in the generation of the semiquinone (22, 24, 25) as well as oxygen radicals (e.g., ref 22) suggesting redox cycling (26). The presence (27) and absence (18,281 of AZQ-GSH conjugates have been reported in different systems. During DT-diaphorase catalysis, the adducts were absent (18),but they were found upon the reduction of AZQ with borohydride (27). In the absence of catalyst, and at controlled neutral pH, no adducts were reported (28).

0893-228x/95/2708-0455$09.00/00 1995 American Chemical Society

Gutierrez and Siva

456 Chem. Res. Toxicol., Vol. 8, No. 3, 1995

Because of our interest in the mechanism of action of AZQ, the importance of GSH, and the discrepancies in the literature, we continue to study here the chemical reaction of diaziquone with reduced glutathione, a t physiological pH (7.4/7.2) and at lower pH (5.5).

Materials and Methods Chemicals. Diaziquone was supplied by the Drug Synthesis and Chemistry Branch, Division of Cancer Treatment, National Cancer Institute (Bethesda, MD). GSH, superoxide dismutase (SOD), and catalase were purchased from Sigma Chemical Co. N-oxide (DMPO) was (St. Louis MO). 5,5-Dimethyl-1-pyrroline purchased from Aldrich Chemical Co. (Milwaukee, WI). Chemical Reaction Conditions. To investigate the formation of conjugates, diaziquone and glutathione were reacted i n a test tube at 1:1, lO:l, or 30:l ratios of GSH to AZQ. GSH was dissolved in water on the day of the experiment and the pH adjusted to 7 with dilute ammonium hydroxide. The use of this base was to minimize sodium ions which lead to large contamination in fast atom bombardment mass spectrometry ( F U N S )used to identify the conjugates. To further minimize this contaminant, all glassware was acid washed. In experiments where the pH was to be controlled more carefully, the reaction was conducted in acetate buffer (pH 5.5) used in the HPLC assay. AZQ stock solutions were prepared in advance at 30-45 mM concentrations in dimethylacetamide. We found t h a t the reaction is faster when carried out at 37 "C. Upon starting the reaction, aliquots were injected into the HPLC system described below to follow the formation of conjugates. When collecting fractions for FAB/MS analysis, the reaction was allowed to go on for 3-5 hours and in some cases up to 22 hours. HPLC Analysis. A modular HPLC Waters Associates system (Milford, MA) with a constant flow pump and a gradient module was used to separate the conjugates. Injection volumes varied from 100 pL (used to determine the position of the peaks) to 3 mL (when collecting fractions for FAB/MS measurements). The pump delivered the eluent at a rate of 1m u m i n to a n amino column (220 x 4.6 cm, AS-224, Brownlee Labs, S a n t a Clara, CA). The solvent system is that of Reed et al. (29). Solvent A was a n 80% methanol solution. Solvent B was a 4:l mixture of solvent A with acetate buffer composed of 15.4 g of ammonium acetate, 12.2 mL of water, and 37.8 mL of glacial acetic acid, pH 5.5. The best solvent gradient was as follows. From the time of sample injection u p to 5 min, 100% solvent A was used. At this time, there was a step increase of solvent B from 0% to 20%. This 20% level was maintained for 5 min, whereupon there was a linear increase of solvent B from 20% to 70% over a time span of 5 min. The 70% level of buffer B was maintained for 5 min and then increased linearly to 100% over 1 min. At 30 min from the sample injection, the chromatography was terminated with all the fractions from the reaction mixture having been eluted. Before each HPLC run, the column was equilibrated by running the system with 100% solvent A for a t least 15 min. Conjugates were detected with a Waters 440 absorbance detector, using the quinone absorbance at 340 nm. When a fraction was isolated, lyophilized, and stored at -77 "C, a clean reproducible chromatogram was obtained. Mass Spectrometry. Before performing FABMS, the fractions collected from the HPLC were subjected to rotary evaporation to remove all organic solvents and to concentrate the products. The condensed fractions were then lyophilized in a VirTis flash evaporator. The conjugates were now i n powder form from which 1-10 n g was dissolved in methanol (Fisher Scientific Co., Fair Lawn, NJ) or a 1 : l methanovwater mixture and added to monothioglycerol (the matrix) (Sigma Chemical Co.) on the probe tip. Measurements were performed with a JEOL HXllO/HX110 (Tokyo, J a p a n ) four sector mass spectrometer (EBEB, where E represents an electric sector and B a magnetic sector). Conventional magnetic scans were obtained with the first two

sectors (EB) using a JEOL fast atom bombardment (FAB) gun and a JEOL DA 5000 data system. The FAB gun was operated a t 6 kV with xenon. The resolution was 3000. Tandem measurements were obtained by utilizing all four sectors (EBEB) of the instrument. Product ions were analyzed with MS-I1 following mass selection of the precursor ions by MS-I and collisional activation. The resolution in both MS-I and MSI1 was 1000. Helium was used for collisionally induced dissociation (CID) in the third field-free region, a t pressures sufficient to attenuate the precursor ion by 80%. The accelerating voltage in MS-I was 10 kV, and the collision cell was floated to 4 kV. Spectra were plotted with the most abundant fragment ion at or near 100% relative abundance. CID is necessary to unequivocally identify glutathione conjugates (30).

Electron Paramagnetic Resonance Spectroscopy (EPR). Electron paramagnetic resonance spectroscopy was carried out using a Varian Century Series X-band (9.3 GHz) spectrometer (Varian Instruments, Palo Alto, CAI. A dual rectangular cavity was used which contained strong pitch (g = 2.0028) i n the back section and the sample in a n ESR flat cell i n the other. Spin trapping experiments were carried out using DMPO under conditions similar to those described for the conjugation reactions or in phosphate buffer (pH 7.4). The DMPO/SG adduct was separated by HPLC and submitted to mass spectrometry measurements as described above. In some experiments, the whole mixture was lyophilized and submitted directly to mass spectral analysis. Anaerobic conditions were achieved by passing nitrogen gas through the 200 pL sample for 3 min before adding AZQ (generally 5 pL) and then quickly passing some more nitrogen for a few seconds. Oxygen Consumption Measurements. Oxygen uptake was determined with a Clark-type electrode in a biological oxygen monitor (Model 53, Yellow Springs Instrument Co., Yellow Springs, OH). The 4 mL phosphate buffered solution (40 mM) was aerated for 3 min in the cuvette of the Model 53 oxygen monitor before adding GSH (10 mM). GSH oxygen consumption was monitored for 5 min before adding AZQ (1 mM). Conversely, AZQ (1mM), which does not consume oxygen, was monitored for 5 min before adding glutathione (10 mM). Ferric iron (FeC13)(50 pM) was then added as well as superoxide dismutase (600 units/mL) or catalase (500 units/mL). Because we were interested in the relative rates of oxygen consumption, these were assessed in terms of percent oxygen consumed per minute.

Results Electron Paramagnetic Resonance Spectroscopy. To understand the reaction of GSH with AZQ in terms of intermediates, EPR spectroscopy was employed. We found that in this case, as well as in reactions to isolate conjugates, excess GSH rather than excess AZQ leads to more dramatic results. The EPR spectrum shown in Figure 1A was obtained 3 min after the onset of the aerobic reaction of AZQ with GSH a t 10-fold excess GHS in the presence of 100 mM DMPO and phosphate buffer (pH 7.4). Twenty minutes into the reaction (Figure 1B) the 5-line spectrum of the AZQ semiquinone (AZQ-) (5) became stronger and was surrounded by two sets of quartets with hyperfine couplings ofAN= 15.0 G andAH = 16.2 G a t g = 2.0051 for one set andAN=AH=14.7G at ~ 2 . 0 0 5 for 9 the other. The former quartet has hyperfine couplings in the range reported for the trapping of the glutathionyl radical by DMPO (DMPO/GS) (31-34). The latter quartet is the better known DMPO/OH adduct (33). The spectrum became stronger and the DMPOIOH signals became more prominent if oxygen was passed through the solution (Figure 1C). EPR spectra similar to that recorded on Figure 1A-C have been recently

Chem. Res. Toxicol., Vol. 8, No. 3, 1995 457

Glutathione Reactions with Aziridine Quinone

A

" ouPeo0

C

D

Figure 1. Electron paramagnetic resonance spectra: (A) Aerobic incubation of GSH (10 mM), AZQ (1 mM), and DMPO (100 mM) i n 40 mM phosphate buffer, pH 7.4, 3 min after the onset of the reaction. (B) Same as (A), 20 min after the onset of the reaction. (C) Oxygenated solution of (A), 30 min after oxygen saturation. (D) Anaerobic incubation of (A). The EPR parameters at room temperature a r e as follows: 9.15 GHz, 15 mW incident microwave power, 1.25 G modulation amplitude, and 1.25 x lo5 receiver gain.

reported in a system catalyzed by DT-diaphorase under aerobic conditions where GSH was used on a 125-fold excess over AZQ (18). Spectra, with both thiyl and hydroxyl radicals, were also reported in experiments that investigated the autoxidation of cysteine (35). Anaerobic incubations yield a sextet with hyperfine couplings ofAN= 15.16 G and AH = 17.93 at g = 2.0056 (Figure 1D). These hyperfine couplings are compatible with reported DMPO-cysteinyl radicals (33-35). The small line near the high field line on the sextet is not understood at this time. Aeration diminishes the hyperfine couplings of this cysteinyl-like radical in favor of those better known for the glutathionyl radical (Figure 1C). To investigate the role of metal ions in the reaction of AZQ with GSH, experiments with the chelator DETAPAC and ferric iron were undertaken. DETAPAC at 3

Figure 2. Electron paramagnetic resonance spectra: (A) DETAPAC (3 mM) containing aerobic incubation of GSH (100 mM), AZQ (1mM), and DMPO (100 mM) in 40 mM phosphate buffer, pH 7.4,20 min after the onset of the reactions. (B) Fe3+ (50 yM) containing aerobic incubation of (A) in the absence of DETAPAC 20 min after the onset of the reaction. (C) Glutathione (10 mM) and Fe3+(50pM) alone. EPR parameters were the same as in Figure 1.

mM concentration favors the production of the AZQ semiquinone and hydroxyl radicals (Figure 2A). In the absence of DETAPAC and in the presence of 50 pM Fe3+, the glutathionyl radical prevails over hydroxyl radicals, and the AZQ semiquinone is present (Figure 2B). In the absence of AZQ, GSH is oxidized to the thiyl radical by metal ions and/or oxygen, traces of which are trapped by DMPO (Figure 2C). Thiols have been known to oxidize to the disulfide (autoxidation) by mild oxidizers such as oxygen (reviewed in ref 36). The antioxidant enzymes SOD and catalase influenced the reaction and the EPR spectra in different ways. Shortly after the start of the reaction, SOD (728 units) favors the DMPO/cysteinyl-like EPR spectrum with no AZQ- (Figure 3A). This spectrum was also observed under anaerobic conditions (Figure 1D). The radical diminishes quickly so that by 20 min into the reaction only traces of thiyl radical remain (Figure 3A'). Catalase (821 units), on the other hand, initially produces a quickly decaying DMPO/glutathionyl radical with traces of AZQ(Figure 3B,B'). In the spectrum of Figure 3B (8 min sweep), the signal decayed considerably by the end of the sweep judging by the low intensity of the high field line. The hyperfine couplings of this EPR spectrum, although not the line intensities, are consistent with DMPO/SG adduct. Adding both SOD and catalase to the reaction resulted in quenching all free radicals after a transient DMPOkysteinyl-like spectrum (Figure 3C and 3C'). Confirmation of the presence of the DMPO/SG adduct was obtained from the mass spectrum of a mixture of the a reaction of GSH and AZQ at 10-fold excess GSH. The strong ion a t 421 amu (Figure 4) indicates the presence of DMPO/SG. Hence the DMPO/cysteinyl-like spectrum must be that of DMPO/SG with slightly different couplings due to anaerobic conditions (see below). No DMPO/OH adducts were observed in this mass spectrum

Gutierrez and Siva

458 Chem. Res. Toxicol., Vol. 8, No. 3, 1995

Figure 3. Electron paramagnetic resonance spectra: (A) SOD (728 units), GSH (10 mM), AZQ (1mM), and DMPO (100 mM) in 40 mM phosphate buffer (pH 7.41, 3 min after adding AZQ to start the reaction. (A)Same a s (A), 20 min later. (B) Same as (A), except that CAT (821 units) was added instead of SOD, 3 min after mixing. (B)Same as (B), 20 min after the onset of the reaction. (C) CAT (821 units) and SOD (728 units) both were added to the system of (A) 3 min after mixing. (C') Same a s (C), 20 min after the onset of the reaction. EPR parameters were the same as in Figure 1. DMP0-W

M+W' 421.1

IS

-

L a

10-

P

f

3

.

'

I

5-

0-

mlz

Figure 4. FAB/MS of a solution of AZQ (1 mM), GSH (10 mM), and DMPO (100 mM). The mixture was kept at pH 7.2 with

NH40H, lyophilized, resuspended i n methanovwater and added to monothioglycerol. The ion at m/z 421 indicates the [M + HI+ of the DMPO/SG adduct.

due to the adduct's low molecular weight for the type of mass spectrometry used. No DMPO/GSSG conjugates were detected, although ions with [M HI' of 613 and 614 amu were routinely observed (data not shown). The ion with a [M HI+ of 614 indicates a protonated disulfide, probably arising from its free radical anion. Oxygen Consumption. Oxygen consumption experiments were important in trying to understand the role of oxygen in the reactions under study. Diaziquone alone does not consume oxygen, but in the presence of glutathione, oxygen consumption is substantial (1.48 0.24%0z/min), and it is further increased by ferric iron (2.77 f: 1.0% Oz/min) (Figure 5). Glutathione alone consumes oxygen a t a small but detectable rate (0.64 f 0.17%0z/min); this rate is enhanced by ferric iron (1.03 f 0.45%0z/min). (All experiments performed a minimum of three times.) These results are consistent with reported GSH oxidation by air and the catalytic effect of ferric iron (reviewed in ref 36). Separation and Identification of Conjugates. Conjugates were separated by HPLC and analyzed by mass spectrometry. The reaction of AZQ with GSH at pH 7.2 or 5.5 resulted in slow bleaching of AZQ to its hydroquinone followed by a gradual oxidation which depended

+

+

\

*

5 min. '

-

\

\

Figure 5. Oxygen consumption of AZQ (1mM) and GSH (10 mM) augmented by Fe3+ (50 pM) in 40 mM phosphate buffer (pH 7.4). GSH (10 mM) consumes oxygen at a small rate (not shown, see text).

on the concentration of GSH. This reaction invariably resulted in the formation of AZQ-GSH conjugates (Figure 6). These conjugates were separated by HPLC (Figure 7) and analyzed by mass spectrometry (e.g., Figure 8). HPLC analysis of solutions with 10-fold excess GSH to AZQ held at 37 "C resulted in the chromatograms

Chem. Res. Toxicol., Vol. 8, No. 3, 1995 469

Glutathione Reactions with Aziridine Quinone

(M+H]*

Peak

k mln

pH

- NH (CH,),SG

979'

A

16

7.2,5.5

= -NH (CH,kSG, Y = -SG

936'

C

25

7.2

895

B

22"

5.5

691'

A

16

5.5

CONJUGATE I

X

I1

X

111

X = Y =-SG

IV

X

-NH(CH,),OH, Y

V

X

H, Y = -NH(CH,),SG

VI

X

Y

=

-NH(CH,),SG

-"a

, Y = -NH(CHJ2SG

631,

7.2

673,

7.2

OH

OH CONJUGATE

[MtH]'

Peak

1, mln

pH

VI1

X

= -NH,, Y -SG

823

8

22

5.5

Vlll

X

= -NHCO,CH,CH,, Y = -NH(CH,),SG

865

B

22

5.5

'From rcfcrencc 30 'Minor conjugate

Figure 6. Chemical structures of AZQ-GSH conjugates isolated in this study. The structure of glutathione indicated in each conjugate

as SG is SC10H16N306.

shown in Figure 7. The rise of three main peaks labeled A ( t =~16.4 m i d , B ( t R = 22.0 min), and C (tR = 24.9 min) was observed as a function of time with a concomitant decrease in the peak corresponding to AZQ ( t R = 3.7 m i d . After 24 h, changes in the chromatograms indicated the possible breakdown of the products, and no further collection was made. Fractions for mass spectrometry analysis were collected, pooled, concentrated in a rotary evaporator and submitted to FAB/MS as described in Materials and Methods. Figure 6 shows the structure of the AZQ-GSH conjugates found in the different fractions. By far the most abundant and consistently found adduct was diconjugate I (MW 9781, present in HPLC peak A with retention time of 16.4 min. This conjugate was observed both at pH 7.2 and at pH 5.5 and represents the nucleophilic addition of glutathione to both aziridine groups of AZQ (30). Diconjugate I1 (MW 935) was found a t pH 7.2 in HPLC peak C ( t R = 25 min) and represents the displacement of an aziridine group and the alkylation of aziridine to GSH. Displacement of both aziridines by GSH resulted in conjugate I11 (MW 894). It was found mostly at pH 5.5 and eluted at 22 min (peak B). These three compounds represent most of the possible conjugates between AZQ and GSH. At pH 5.5 and eluting in the first peak along with I was the alcohol monoconjugate adduct IV (MW 690). In addition, in reactions carried

out at pH 5.5 minor hydroquinone compounds VI1 and VI11 eluted at 22 min (peak B). Compound M I 1 is the hydroquinone form of conjugate 11. If a slower gradient is used, one can also observed small amounts of compounds V and VI along with I and I1 at pH 7.2. The retention times are different from those shown in Figure 7 (data not shown). All the conjugates described above have a characteristic mass spectrum that involves the loss of OCHzCH3 (M - 45) and COOCH2CH3 (M - 73) from the carbethoxyamino groups. In addition, an ion at [M - 2741 is found corresponding to the loss of G = C10H16N306, the desulfurinated ion of glutathione. Furthermore, decompositions of the glutathione moiety that permits recognition of an unknown as a glutathione conjugate can also be observed. Details of this fragmentation have been published for I, 11,and IV (30).Figure 8 shows the mass spectrum of 111, as an example of the mass spectra of the conjugates encountered here. The loss of 45, 73 and 274 amu is shown in prominent ions with mlz 850, 822, and 621 for this conjugate. It is important to point out that when AZQ was reduced by borohydride, no products in the HPLC chromatograms of Figure 7 were observed. This result is consistent with that recently reported in reactions mediated by DT-diaphorase (18).

Gutierrez and Siva

460 Chem. Res. Toxicol., Vol. 8, No. 3, 1995

0

p

t = O h

Conjugate I was observed at both pH 7.417.2 and 5.5. It is formed by nucleophilic attack of GS- on the aziridines of AZQ as shown:

k r Nu:- (-SG,-OH, H 2 0 )

CI

B This is also the mechanism for the formation of conjugates IV and VI. In the case of these two conjugates, the pH effect is clearly seen. The formation of conjugate IV,which is a monoalcohol, is favored a low pH (381, while at neutral pH, one of the aziridines in not open. Glutathione is one of a family of compounds with an active hydrogen. These compounds can add to a or p unsaturated bonds containing electron-withdrawinggroups such as C=O in quinones (37,39):

1

A

t = 23 h

1

(2)

1

RETENTION TI ME Figure 7. HPLC chromatograms at various times of the reaction of GSH (12 mM, pH 7.2) with AZQ (1 mM). Similar chromatograms were obtained when the reaction was carried out in water and the pH adjusted to 7.2/7.4 with NH40H before AZQ was added. Peaks A, B, and C lacking sodium were collected, pooled, and analyzed by FABNS as described in Materials and Methods.

Discussion The data presented here provide evidence that glutathione and diaziquone can undergo reactions leading to a variety of conjugates. These conjugates are not formed when AZQ is reduced with borohydride. ESR data provide evidence for the presence of glutathionyl and hydroxyl radicals as well of the AZQ semiquinone, demonstrating oxidation-reduction reactions. These data, along with bleaching of AZQ, are consistent with the reduction of AZQ by GSH. Delineation of all the steps involved in the formation of the conjugates observed (including free radical involvement) is difficult. We attempted to control the conditions and used two pH values: 7.417.2 and 5.5. The neutral pH will favor anions (GS-), while at the more acidic pH, the equilibrium GSH * GS- +Hf will shift to the left to favor GSH CpK, ca. 9). At these two pH conditions, both GS- and GSH are present at different concentrations, resulting in a mixture of reactions. Thus all the conjugates were inevitably formed at both pH 7.4 and 5.5, but their abundance was modified by the pH condition. In all cases, it is the anion state that is the active nucleophile (37).

R = NHC02CH2CH3

The conjugates formed indicate that aziridines are the best leaving groups in reaction 2. This is understood in terms of protonation of the aziridine nitrogen. Glutathione is a stronger acid than RNH; thus GS- should be a better leaving group than RN-. However, protonation of the nitrogen on the aziridine ring (40)will provide a better leaving group than GS-. This implies that the bond breaking step is the kinetically significant step in reaction 2 (39). Conjugate I1 represents the product of a 1,4-Michael addition (reaction 2) and a nucleophilic attack of GS- to one aziridine group of AZQ (reaction 1). Reaction 2 can also explain conjugate 111. The product of reaction 2 can undergo another GS- Michael addition reaction with the other oxygen and nitrogen to form conjugate 111. Again, these two conjugates represent the mixture of reactions taking place and how pH can influence the outcome. Conjugate I11 was almost exclusively found at pH 5.5, while conjugate I1 was found a t pH 7.2. Displacement of aziridine by GSH has also been observed in the conversion of melphalan to 44glutathiony1)phenylalanine catalyzed by glutathione S-transferase (41).

Conjugate V was observed at pH 7.2 in small quantities. It is a bit more difficult t o explain. We have observed the formation of monoaziridine quinone from AZQ by the elimination of an aziridine group during the 2-electron oxidation of the AZQ hydroquinone (42). We explained this by invoking a reverse Michael 1,4-elimina-

Chem. Res. Toxicol., Vol. 8, No. 3, 1995 461

Glutathione Reactions with Aziridine Quinone

0

m/z

Figure 8. FABMS conjugate 111. Experimental details are described in the text.

tion in preference of tautomerization to the hydroquinone. The steps in this proposed reaction required a semiquinone which was observed then (42)and now. The monoaziridine compound represents a net 2-electron oxidation of the hydroquinone. Thus, conjugate V would start out as the hydroquinone of conjugate VI. During the collection of our HPLC fractions for mass spectral analysis, no effort was made to keep the solution anaerobic. Thus we are confident that oxidation of any reduced AZQ-GSH conjugates took place. The hydroquinone conjugates observed (VI1 and VIII) are evidently not readily oxidized, which indicates the presence of an equilibrium reaction. Conjugate VI11 is the hydroquinone of conjugate 11. Hydrolysis and/or fragmentation of the carbethoxyamino group in compound I11 is most likely responsible for conjugate VII. Equation 3 shows the way to reduce AZQ to its semiquinone while GSH is oxidized to the thiyl radical. In this reaction, an 0,s-hemiacetal intermediate is formed after adding GS-. This intermediate then dissociates into the semiquinone and the glutathionyl radical by a bond breaking step. The AZQ semiquinone can then undergo a nucleophilic attack at the aziridines by GSH to form compound I. Most of the unpaired electron resides on the carbons of the quinone attached to the aziridine (43),rendering the aziridine nitrogen slightly positive, a condition that favors nucleophilic attack (42) (reaction 4). Reactions 3 and 4 can contribute to the formation of conjugate I.

increase in DMPO/GS signal and the lack of DMPO/OH signal upon adding ferric iron to the reaction (Figure 2B). In this case, reaction 5 for Fe3+produces Fez+which can participate in Fenton chemistry with the production of O H through a Fe2+/thiolcomplex (46) (reactions 6-9). G 8 can further react with oxygen to produce O H from peroxysulfenic acid (reactions 10- 12) (47). Oxygen consumption experiments indicate that reactions 7 and 10 are very important. When DETAPAC is present, it competes for iron, decreasing the formation of the thiol complex, resulting in traces of GB radical concentrations (Figure 2A). The O H and AZQ- observed (Figure 2A) are products of normal redox reactions.

Fe2+(GS),

Fe2'(RS),

+ 0,

+ H,O, GS'

[Fe3+(GS),-0i-1

- Fe3+(RS),+ + 0,

O'H

+ OH-

GSOO'

+ GSH - GSOOH + GS' GSOOH GSO' + O'H

GSOO'

-

(4) "

0

R= NHCOzCHzCH3

There are other sources of the glutathionyl radical in addition to reaction 3 that must be explored. Ferric iron catalyzes the autoxidation of glutathione (35,44,45).In the present experiments, we found that iron is catalytic in generating the glutathionyl radical and did not increase hydroxyl radicals. Evidence for this is the

(7)

(9) (10)

(11) (12)

The fact that the combination of SOD and catalase (reactions 13 and 14) abrogated all free radicals (Figure 3C,C'), suggested that H20z and 02.- have an important role in free radical propagation (reactions 17-23) (46,48). The reduction of AZQ by GSH (reactions 4 and 15) followed by oxidation (reaction 16) acts as initiator. Superoxide radical anion is produced during the oxidation of AZQ- and AZQH2 (reaction 16), which dismutates t o hydrogen peroxide. Alternatively, 0 2 ' - may react with GSH (present in excess) to form hydrogen peroxide

462 Chem. Res. Toxicol., Vol. 8, No. 3, 1995

Gutierrez and Siva

(reacti0nl8).~Hydrogen peroxide in turn is able to generate O'H through Fenton chemistry which in turn can generate GS' (reaction 21). 2H,O,

CAT

2H,O

+ 0,

(13)

explanation. What is the origin of AZQ- in this catalasetreated reaction? The presence of hydroxyl radicals indicates that AZQ- originates from redox cycling brought about by the oxidation of AZQH2 and by the reaction of AZQHz or AZQ with 0 2 ' - to give AZQ- (AZQH2+ 02'-+ HAZQH H202) (511. Ordoiiez and Cadenas (18) studied a system similar to ours, except that they used DT-diaphorase. When this enzyme is included, superoxide dismutase decreases significantly the DMPO/SG signal and, to a lesser extent, the DMPO/OH signal. Catalase, on the other hand, had no effect on free radical production. Both enzymes quenched all free radicals. These facts lead these authors to conclude that hydrogen peroxide and hydroxyl radicals had no role in the formation of thiyl radicals (18). The elimination of all free radicals with the addition of SOD and catalase in the presence of DT-diaphorase (18) is consistent with our findings. Our results with the individual treatment of the SOD and catalase are also consistent. With the exception that, with catalase, we observed A Z Q - and minor amounts of O H . Another discrepancy is that under anaerobic conditions Ordoiiez and Cadenas did not observe GS' radicals perhaps due to the large excess of GSH used and to the presence of DT-diaphorase. It would be interesting to explore how much of the effect observed is enzymatic and how much chemical. In terms of conjugate production, our results are in agreement in that AZQHZ formed either by reduction with DT-diaphorase (18)or borohydride is not able to conjugate with GSH. Are any or some of the free radicals detected important in the production of conjugates? This question is rather difficult to answer due to the complicated system under study. The definitive experiment, that of quenching free radicals and assaying for conjugates, is technically difficult and equivocal. Suppression of GSH oxidation with HCN (52) will only eliminate some thiyl and hydroxyl radicals arising from the autoxidation of GSH, but not necessarily those from the AZQ mediated GSH oxidation (17, 19). Nevertheless, we conclude that the majority of the adducts are formed by 1,CMichael addition and nucleophilic reactions with the aziridines of AZQ. We conclude also that it is important for GSH to participate in the reduction of AZQ by reaction 4 because if AZQ is reduced by borohydride, no conjugates arise. Thus, there is a free radical component in adduct formation. In terms of the mechanism of action of AZQ as an antitumor agent, GSH can partially neutralize its alkylating activity. The majority of the conjugates, except VI, result in aziridine loss as GSH reacted with them. These conjugates are expected to be inactive or much less active than AZQ. In the case of mitomycin C, mono- and bis-GSH conjugates were found to be inactive against human colon carcinoma cell lines (14).It is interesting that GSH changed the alkylating profile of mitomycin C alkylation to DNA from monoadducts to bisadducts upon a one-electron reduction (16) and that mitomycin C must be reduced by one electron to form conjugates with GSH +

+ 2H' SOD H202+ 0, H+ + AZQ- + GSH - AZQH, + GS' AZQ-/AZQH, + 0, 0,'- + AZQ/AZQHO', + 0;- + H+ - H,O, + 0, GSH + 0,'- + H+ - GS' + H,O, Fe2++ H,02 - Fe3+ + -OH + O'H O'H + H,O, - 0, + H,O + O'H GSH + O'H - GS' + H,O GS' + GS- - GSSGGSSG- + 0, GSSG + 0,'20,'-

-

-

(14) (15) (16)

(17)

(18) (19)

(20) (21)

(22) (23)

SOD alone prevents the formation of AZQ semiquinone. SOD pulls reaction 16 to the right, rendering the solution anaerobic. Because there is excess GSH, it can further react with AZQ- to produce GS' and AZQHz (reaction 15). This reaction was invoked to explain the quenching of AZQ- by GSH in an enzyme system (18). Evidence for AZQHz is the absence of semiquinone in the EPR spectrum and the colorless reaction mixture. Reaction 15 explains the absence of AZQ- upon SOD incubation under aerobic conditions (Figure 3Aj and also under anaerobic conditions in the absence of any antioxidant enzymes (Figure 1D). The DMPO/SG EPR spectra under anaerobic conditions (Figure 1Dj and under aerobic conditions in the presence of SOD (Figure 3A) are the same and are like DMPO/cysteinyl adducts. Oxygen broadens the sextet of lines and the AN coupling decreases, resulting in the better known DMPO/SG quartet. The AH coupling also decreases but not as significantly as AN. Further evi- . dence that we have trapped the glutathionyl radical and not the cysteinyl or the G S S W radicals comes from mass spectrometry (Figure 4). Therefore, under anaerobic conditions, the DMPO/SG adduct gives a DMPO/cysteinyl-like EPR spectrum. Trapping GSSG- from eqs 22 and 23 can only be ruled out a t this time on the grounds of mass spectrometry and size, but reactions 22 and 23 are most likely present (17-19). Furthermore, the fact that GS' is detected anaerobically validates eqs 4 and 15 as a source of GS.. Catalase results in a quick decay of GS' radicals in the initial minutes of the reaction. After 20 min, AZQpersists along with traces of hydroxyl radicals (Figures 3B,B'). The quick decay of GS' implicates reactions 182 1 in the production of GS' (50).The absence of GS' 20 min after treating with catalase is consistent with this *Reaction 18 is somewhat controversial, but it is suggested as evidence for the production of thiyl radicals in the xanthineixanthine oxidase and quinol autoxidation systems (48, 49).

+

(16).

The reactions depicted here are chemical reactions likely to occur in vivo especially since GSH is present at such high concentrations in cells (up to 5 mM) (50). It is probable that the reduction of AZQ by GSH can lead not only to deleterious effects in terms of oxidative stress but also to detoxification pathways.

Glutathione Reactions with Aziridine Quinone

Acknowledgment. This publication was supported by Grant CA53491 from the National Cancer Institute. Its contents are solely the responsibility of the authors and do not necessarily represent the official views of the National Cancer Institute. The authors thank Drs. Constance Murphy and Catherine Fensleau for mass spectra obtained in the Structural Biochemistry Center at the University of Maryland Baltimore County. We are indebted to Dr. Philip J. Wilder who conducted preliminary studies leading us to this investigation, and to Jamilah Borjac for her assistance during one phase of this project. We thank Dr. George Sosnovsky for fruitful discussions during the preparation of the manuscript.

References Schold, S. C., Jr., Friedman, H. S., Bjornsson, T. D., and Falletta, J . M. (1984)Treatment of patients with recurrent primary brain tumors with AZQ. Neurology 34,615-619. Lee, E., Van Echo, D., Egorin, M., Nayar, M. S. B., Shulman, P., and Schiffer, C. (1986)Diaziquone given as a continuous infusion is an active agent for relapsed adult acute nonlymphocytic leukemia. Blood 67,182-187. Schulman, P., Davis, R., Lee, E., Ellerton, J., and Staszweski, H. (1987) Phase I1 study of continuous-infusion diaziquone in relapsedlrefractory acute nonlymphocytic leukemia. Cancer Treat. Rep. 71, 755-157. Gutierrez, P. L., Egorin, M. J., Fox, B. M., Friedman, R., and Bachur, N. R. (1985) Cellular activation of diaziquone 2,5diaziridinyl-3,6-bis(carboethoxyamino~-l,4-benzoquinone to its free radical species. Biochem. Pharmacol. 34, 1449-1455. Gutierrez, P. L. (1989)Mechanism(s1 of bioreductive activation. The example of diaziquone (AZQ).Free Radical Biol.Med. 6,405445. Nguyen, B., Biswal, S., and Gutierrez, P. L. (1988)The influence of diaziquone free radicals on the in vitro activity of diaziauone. Xenobiokca 18,593-602. (7) Doroshow, J. (1986)Role of hydrogen peroxide and hydroxyl radical formation on the killing of Ehrlich tumor cells by anticancer quinones. Proc. Natl. Acad. Sci. U S A . 83,4514-4518. (8) Fisher, G., Donis, J., and Gutierrez, P. L. (1992)Reductive metabolism of diaziquone (AZQ) in the S9 fraction of MCF-7 cells: 11. Enhancement of the alkylating activity of AZQ by NAD(P)H: Quinone-acceptor oxidoreductase (DT-Diaphorase). Biochem. Pharmacol. 44,1625-1635. (9)Szmigiero, L., and Kohn, K. W. (1984)Mechanisms of DNA strand breakage and interstrand cross-linking by diaziridinylbenzoquinone (Diaziquone) in isolated nuclei from human cells. Cancer Res. 44,4453-4457. (10)Szmigiero, L., Erickson, L. C., Ewig, R. A., and Kohn, K. K. (1984) DNA strand scission and cross-linking by diaziridinylbenzoquinone (Diaziquone) in human cells and relation to cell killing. Cancer Res. 44,4447-4452. (11)King, C. L., Wong, S.-K., and Loo, T. L. (1984)Alkylation ofDNA by the new anticancer agent 3,6-diaziridinyl-2,5-bis(carboethoxyamino)-1,4-benzoquinone(AZQ). Eur. J . Cancer Clin. Oncol. 20,261-264. (12)Arrick, B. A,, and Nathan, C. F. (1984)Glutathione metabolism as a determinant of therapeutic efficacy: A review. Cancer Res. 44,4224-4232. (13)Yoda, Y., Nakazawa, M., Abe, T., and Kawakami, Z. (1986) Prevention of doxorubicin myocardial toxicity in mice by reduced glutathione. Cancer Res. 46,2551-2556. (14)Sharma, M.,and Tomasz, M. (1994)Conjugation of glutathione and other thiols with bioreductively activated mitomycin C. Effects of thiols on the reductive activation rate. Chem. Res. Toxicol. 7,390-400. (15)Gamer, R. A., Zakher, J., and Kim. G. (1988) Role of the glutathione redox cycle in acquired and de novo multidrug resistance. Science 241,694-697. (16)Sharma, M., He, Q.-Y., and Tomasz, M. (1994)Effects of glutathione on alkylation and cross-linking of DNA by mitomycin C. Isolation of a ternary glutathione-mitomycin-DNA adduct. Chem. Res. Toricol. 7,401-407. (17)Silva, J. M., and O'Brien, P. J. (1989) Diaziquone-induced cytotoxicity in isolated rat hepatocytes. Cancer Res. 49,55505554.

Chem. Res. Toxicol., Vol. 8, No. 3, 1995 463 Ordofiez, I. D., and Cadenas, E. (1992)Thiol oxidation coupled to DT-diaphorase-catalyzed reduction of diaziqubne. Reductive and oxidative pathways of diaziquone semiquinone modulated by glutathione and superoxide dismutase. Biochem. J. 286, 481490. OBrien, P.J. (1991)Molecular mechanisms of quinone cytotoxicity. Chem.-Biol. Interact. 80,1-41. Lau, S. S.,Hill, B. A., Highet, R. J., and Monks, T. J. (1988) Sequential oxidation of glutathione addition to l,4-benzoquinone: Correlation of toxicity with increased glutathione. Mol. Pharmacol. 34,829-836. Ramakrishna Rao, D.N., Takhashi, N., and Mason, R. P. (1988) Characterization of glutathione conjugate of the 1,4-benzosemiquinone-free radical formed in rat hepatocytes. J . Biol. Chem. 263, 17981-17986. Miura, T., Muraoka, S., and Ogiso, T. (1992)Generation of semiquinone and oxygen radicals by the reaction of menadione with reduced glutathione at various pH. Chem. Pharm. Bull. 40, 709-712. Nickerson, W. J.,Falcone, G., and Straws, G. (1963)Studies on quinone-thioethers. I. Mechanism of formation and properties of thiodione. Biochemistry 2, 537-543. Takahashi, N., Schreiber, J.,Fischer, V., and Mason, R. P. (1987) Formation of glutathione-conjugated semiquinones by the reaction of quinones with glutathione: An ESR study. Arch. Biochem. Biophys. 252,41-48. Gant, T. W., Doherty, M. D., Odowole, D., Sales, K. D., and Cohen, G. M. (1986)Semiquinone anion radicals formed by the reaction of quinones with glutathione or amino acids. FEBS Lett. 201, 296-300. Wefers, H., and Sies, H. (1983)Hepatic low-level chemiluminescence during redox cycling of menadione and the menadioneglutathione conjugate: Relation to glutathione and NAD(P)H: Quinone reductase (DT-diaphorase) activity. Arch. Biochem. Biophys. 224,568-578. Ross, D., Seigel, D., Gibson, N. W., Pacheco, D., Thomas, D. J., Reasor, M., and Wierda, D. (1990)Activation and deactivation of quinones catalyzed by DT-diaphorase. Evidence for bioreductive activation of diaziquone (AZQ) in human tumor cells and detoxification of benzene metabolites in bone marrow stroma. Free Radical Res. Commun. 8,373-381. Lusthof, K. J., de Mol, N. J., Janssen, L. H., Prins, B., Verboom, W., and Reinhoudt, D. N. (1990)Interactions between potential anti-tumour 2,5-bis(l-aziridinyl)-1,4-benzoquinone derivatives and glutathione: Reductive activation, conjugation and DNA damage. Anti-Cancer Drug Des. 5,283-290. Reed, D. J., Babson, J. R., Beatty, P. W., Ellis, W. W., and Potter, D. W. (1980)High performance liquid chromatography analysis of nanomole levels of glutathione, glutathione disulfide, and related thiols and disulfides. Anal. Biochem. 106,55-62. Murphy, C. M., Fenselau, C., and Gutierrez, P. L. (1992) Fragmentation characteristic of glutathione conjugates activated by high-energy collisions. J . Am. SOC.Mass Spectrom. 3, 815822. Harman, L. S., Carver, D., Schreiber, J., and Mason, R. (1986) One- and two-electron oxidation of reduced glutathione by peroxides. J . Biol. Chem. 261, 1642-1648. Mottley, C., Toy, K., and Mason, R. (1987)Oxidation ofthiol drugs and biochemicals by the lactoperoxidasehydrogen peroxide system. Mol. Pharmacol. 31,417-421. Buettner, G. R. (1987)Spin trapping: ESR parameters of spin adducts. Free Radical Biol. Med. 3,359-303. Ross, D., Norbeck, K., and Moldeus. P. (1985)The eeneration and subsequent fate of glutathionyl radicals in biologiial systems. J . Biol. Chem. 260, 15028-15032. (35)Saez, G.,Thornalley, P. J., Hill, H. A. O., Hems, R., and Bannister, J. V. (1982)The production of free radicals during the autoxidation of cysteine and their effect on isolated rat hepatocytes. Biochim. Biophys. Acta 719,24-31. (36)Tarbell, D. S. (1961)The Mechanism of oxidation of thiols to disulfides. In Organic Sulfur Compounds (Kharasch, N., Ed.) pp 97- 102, Pergamon Press, Symposium Publications Division, New York. (37) Douglas, K. T. (1988)Reactivity of glutathione in model systems for glutathione S-transferases and related enzymes. In Glutathione Conjugation (Sies, H., and Ketterer, B., Eds.) pp 2-37, Academic Press Limited, London. (38)Poochikian, G. K., and Kelley, J. A. (1981)2,5-Diaziridinyl-3,6bis(carboethoxyamino)-1,4-benzoquinone11: Isolation and characterization of degradation products. J . Pharm. Sci. 70,162-167. (39)Parker, A. J. (1961)Sulfur nucleophiles in aromatic SN reactions. In Organic Sulfur Compounds (Kharasch, N., Ed.) pp 103-111, Pergamon Press, Symposium Publications Division, New York.

Gutierrez and Siva

464 Chem. Res. Toxicol., Vol. 8, No. 3, 1995 (40)Mossoba, M. M., Alizadeh, M., and Gutierrez, P. L. (1985) Mechanism for the reductive activation of diaziquone (AZQ). J. Pharm. Sci. 74, 1249-1254. (41)Dulik, D. M.,and Fenselau, C. (1987)Conversion of melphalan to 4-(glutathiony1)phenylalanine.A novel mechanism for conjugation by glutathione-S-transferases. Drug Metab. Dispos. 15, 195199. (42)Gutierrez, P. L., Nayar, M. S. B., Nardino, R., and Callery, P. S. (1987)The chemical reduction of diaziquone: Products and free radical intermediates. Chem.-BioL Interact. 64,23-37. (43)Gutierrez, P. L., Fox, B. M., Mossoba, M. M., Egorin, M. J., Nakazawa, H., and Bachur, N. R. (1985)Electron spin resonance of electrochemically generated free radicals from diaziquone and its derivatives. Biophys. Chem. 22, 115-123. (44)Searle, A. J. F., and Tomasi, A. (1982)Hydroxyl free radical production in iron-cysteine solutions and protection by zinc. J. Inorg. Biochem. 17, 161-166. (45) Cecil, R., and McPhee, J. R. (1959)The sulfur chemistry of proteins. Adu. Protein Chem. 14, 255-389. (46)Wardman, P. (1988)Conjugation and oxidation of glutathione via thiyl free radicals. In Glutathione Conjugation (Sies, H., and Ketterer, B., Eds). pp 43-72, Academic Press Limited, London.

(47)Sevilla, M.D., Becker, D., Swarts, S., and Herrington, J. (1987) Sulfinyl radical formation from the reaction of cysteine and glutathione thiyl radicals with molecular oxygen. Biochem. Biophys. Res. Commun. 144, 1037-1042. (48)Willson, R. L. (1978)Free radicals and tissue damage: Mechanistic evidence from radiation studies. In Biochemical Mechanisms of Liver Injury (Slater, T. F., Ed.) pp 123-224, Academic Press, London. (49)Ross, D., Cotgreave, I., and Moldeus, P. (1985)The interaction of reduced glutathione with active oxygen species generated by xanthine-oxidase-catalyzed metabolism and xanthine. Biochim. Biophys. Acta 841, 278-282. (50) Kosower, N. S., and Kosower, E. M. (1976)The glutathioneglutathione disulfide system. In Free Radicals in Biology, Vol. 11, pp 55-84, Academic Press, New York. (51) Gutierrez, P. L., Egorin, M. E., Davis, T. A., and Bachur, N. R. (1985)Effects of ascorbic acid on biologically obtained diaziquone free radicals. Biochem. Pharmacol. 34,2394-2397. (52)Colowick, S.,Lazarow, A., Racker, E., Schwartz, D. R., Stadtman, E., and Waelsch, H., Eds. (1954)Glutathione, p 51, Academic Press, New York.

TX9401472