ARCHIVES

OF BIOCHEMISTRY

Vol. 291, No.

1,

November

AND

BIOPHYSICS

15, pp. 43-51,199l

Production of Singlet Oxygen-Derived Hydroxyl Radical Adducts during Merocyanine-540-Mediated Photosensitization: Analysis by ESR-Spin Trapping and HPLC with Electrochemical Detection Jimmy

B. Feix’

and B. Kalyanaraman

Biophysics Section, Department of Radiology, Medical College of Wisconsin, 8701 Watertown Plunk Road, Milwaukee, mksconsin 53226

Received

April

26, 1991, and in revised

form

July 29, 1991

Activated oxygen species produced during merocyanine 640 (MC640)-mediated photosensitization have been examined by electron spin resonance (ESR) spin trapping and by trapping reactive intermediates with salicylic acid using HPLC with electrochemical detection (HPLC-EC) for product analysis. Visible light irradiation of MC540 associated with dilauroylphosphatidylcholine liposomes in the presence of the spin trap, S&dimethyl1-pyrroline-N-oxide (DMPO) gave an ESR spectrum characteristic of the DMPO-hydroxyl radical spin adduct (DMPO/ * OH). Addition of ethanol or methanol produced additional hyperfine splittings due to the respective hydroxyalkyl radical adducts, indicating the presence of free . OH. DMPO/ . OH formation was not significantly inhibited by Desferal, catalase, or superoxide dismutase (SOD). Production of DMPO/ . OH was strongly inhibited by azide and enhanced in samples prepared with deuterated phosphate buffer (PB-DzO), suggesting that singlet molecular oxygen (‘0,) was an important intermediate. When MC540-treated liposomes were irradiated in the presence of salicylic acid (SA), HPLC-EC analysis indicated almost exclusive formation of 2,5-dihydroxybenzoic acid (2,5-DHBA), with production of very little 2,3DHBA, in contrast to . OH generated by uv photolysis of HzOz, which gave nearly equimolar amounts of the two products. 2,5-DHBA production was enhanced in PBD20 and inhibited by azide, again consistent with ‘02 intermediacy. 2,5-DHBA formation was significantly reduced in samples saturated with Na or argon, and such samples showed no DzO enhancement. Ethanol had no effect on 2,6-DHBA production, even when present in large excess. Catalase and SOD also had no effect, and only a small inhibition was observed with Desferal. DMPO inhibited 2,6-DHBA production in a concentrar To whom

correspondence

should

0003s9861/91$3.00 Copyright 0 1991 by Academic Press, All rights of reproduction in any form

be addressed.

tion-dependant fashion and enhanced formation of 2.3DHBA. We propose that ‘Oz reacts with DMPO to give an intermediate which decays to form DMPO/. OH and free . OH, and that the reaction between ‘0% and SA preferentially forms the 2,5-DHBA isomer. This latter process may provide the basis for a sensitive analytical method to detect ‘Oz intermediacy. Singlet oxygen appears to be the principle activated oxygen species produced during MC540-mediated photosensitization. Q leoi Academic

Press,

Inc.

Merocyanine 540 (MC540)2 is a lipophilic dye under investigation for use in the photodynamic purging of bone marrow explants (1) and the elimination of viral contaminants from blood and blood products (2, 3). The mechanisms of MC540-mediated antiviral and antileukemic effects have been under intensive investigation. Kalyanaraman et al. proposed singlet molecular oxygen (‘0,) as the principle reactive intermediate (4) based on the observations of oxygen consumption in the presence of ‘02 traps, isolation of the reaction product of ‘02 with cholesterol, azide inhibition, and enhancement of cell killing in media prepared with deuterated water (D,O). In contrast, Davila et al. suggested a mechanism based on dye photoisomerization (5), which is independent of molecular oxygen. While photoisomerization clearly oc’ Abbreviations used: DLPC, dilauroylphosphatidylcholine; DCP, dicetylphosphate; DHBA, dihydroxybenzoic acid; DMPO, 5,5dimethyll-pyrroline-N-oxide; ESR, electron spin resonance; HPLC, high-performance liquid chromatography; EC, electrochemical detection; ‘Ox, singlet oxygen; . OH, hydroxyl radical; O;, superoxide anion; SA, salicylic acid; MC540, merocyanine 540, LUV, large unilamellar vesicles; PBDsO, deuterated phosphate buffer; SOD, superoxide dismutase; MLV, multilamellar vesicles. 43

Inc. reserved.

44

FEIX

AND

KALYANARAMAN

curs upon irradiation of MC540 (6,7), the demonstration by Gaffney et al. that MC540-mediated cell killing is O2 dependent (8) indicated that photoisomerization is not the basis for the antileukemic effects of this dye. This returned the focus of mechanistic studies to singlet oxygen and other activated oxygen species. The generation of singlet oxygen by MC540 bound to liposomes has been demonstrated by direct observation of ‘02 luminescence at 1268 nm (9) and the quantum yield for ‘02 production, based on oxygen consumption measurements, determined to be 0.068 in a membrane environment (10). The action spectrum for O2 consumption corresponds closely to the absorption spectrum of membrane-bound, monomeric dye (10). Action spectra of the antileukemic and antiviral effects of MC540 also correlate well with the absorption spectra of membrane-bound MC540, with the monomeric form of the dye accounting for most of the activity (11). These observations are consistent with ‘02 as the primary intermediate in MC540based photosensitization, although other mechanisms might be proposed. Relatively little work has been directed toward the identification of oxygen-derived radicals during MC540mediated photosensitization. Hoebeke et al. previously reported the failure to detect either superoxide or hydroxyl radical spin adducts, possibly due to their photochemical destruction (7). Sarna et al. reported spin trapping evidence for MCMO-sensitized formation of hydroxyl radical in methanol solutions containing chelated iron (12) and proposed a Type I (electron-transfer) mechanism based on this and other evidence for electron transfer in the presence of reductants. We have previously shown using a nitroxide spin label-reduction assay that Type I processes do occur under anaerobic conditions and are facilitated by physiological reductants such as NADPH and glutathione (13). The present study was directed at further examining the production of activated oxygen species other than ‘02 that might be produced during MC540-mediated photosensitization. We have utilized electron spin resonance (ESR) spin trapping (14, 15), as well as the trapping of reactive oxygen species with salicylic acid. ESR spin trapping has proven to be one of the most powerful analytical methods available for elucidating the production of 0; and 0OH in photochemical systems (e.g., (16, 17)). In contrast, there have been relatively few studies on the interaction of ‘02 with spin traps (18). Salicylic acid (SA) reacts with the hydroxyl radical to form predominantly 2,3- and 2,5-dihydroxybenzoic acid (2,3- and 2,5-DHBA). These DHBA products are stable and may be readily isolated and quantitated by reverse-phase HPLC with electrochemical detection (HPLC-EC) (19, 20). SA does not react with 0; at an appreciable rate compared to + OH (21), providing a degree of specificity for . OH. The reaction of salicylic acid with ‘02 has not been previously con-

sidered in the context of biological oxidative stress. In these studies we demonstrate the production of both DMPO spin adducts and DHBA products via a singlet oxygen-dependent mechanism. Our findings indicate that ‘02 is indeed the key activated oxygen species produced during MC540-mediated photosensitization and show that HPLC-EC is a valuable adjunct to ESR spin trapping in the characterization of activated oxygen intermediates. A preliminary account of this work has been given (41). MATERIALS

AND

METHODS

Merocyanine 540 was obtained from Sigma and was shown to be pure by HPLC and mass spectroscopy. Stock solutions of MC540 (0.5 mM in doubly distilled HrO) were freshly prepared the day of the experiment and protected from light. Dilauroylphosphatidylcholine (DLPC) and dicetylphosphate (DCP) were also obtained from Sigma. Catalase and superoxide dismutase were obtained from Boehringer-Mannheim (Indianapolis, IN), and Desferal mesylate was a generous gift from Ciba pharmaceuticals (Summit, NJ). Salicylic acid, 2,3-dihydroxybenzoic acid (2,3-DHBA), 2,5-dihydroxybenzoic acid (2,5-DHBA), and citric acid were from Aldrich Chemical Co. (Milwaukee, WI). 5,5-Dimethyl-1-pyrroline-N-oxide (DMPO) was obtained from Aldrich and further purified by high-vacuum distillation and treatment with activated charcoal (22). Perdeuterated 50 mM phosphate buffer (PB-DSO) was prepared by addition of perdeuterated phosphoric acid (DaPO,) to deuterated water (DaO) and titration with deuterated sodium hydroxide (NaOD) to pH 7.8 (pD 7.4). DzO, D3P01, and NaOD were obtained from Aldrich. Liposome preparation. Large unilamellar liposomes were prepared as described previously (13). Briefly, lipids (DLPC/DCP; molar ratio, 9/l) were dried down from stock solutions in chloroform under a stream of dry N2 gas and then further dried under vacuum for at least 1 h. Multilamellar vesicles (MLVs) were prepared by direct hydration in 50 mM phosphate buffer, pH 7.4 (PB-H,O), or (where noted) in PB-DrO, with periodic vortexing to form a homogeneous suspension. Large unilamellar vesicles (LUVs) were prepared by freeze-thawing the MLVs five times and then extruding the suspension through 0.2~pm polycarbonate filters (Nucleopore, Pleasanton, CA) a total of six times using an Extruder (Lipex Biomembranes Inc., Vancouver, B.C.) as described by Hope et al. (23). DLPC/DCP (9/l) liposomes in phosSpin trapping experiments. phate buffer were warmed to room temperature and mixed with MC540 in a molar ratio (lipid/dye) of 100/l. Buffer containing any desired additives (e.g., enzymes, Desferal, azide) and DMPO were then added to give final concentrations of 20 mM lipid, 0.20 mM MC540, and 45 mM DMPO. This sample was immediately transferred to either a capillary of the gas-permeable methylpentene polymer, TPX (Westlake Plastics, Lenni, PA) or a quarts flat cell and placed in the ESR cavity. An initial spectrum was then recorded to verify that there was no background signal present prior to irradiation. ESR spectra were recorded on a Varian E-line Century series spectrometer at room temperature (22 + 0.5”C) with 10 mW incident microwave power and a magnetic field modulation of 1.0 G. Computerized data acquisition and spectral integrations were performed using software developed at the National Biomedical ESR Center (Milwaukee, WI). Samples were irradiated in situ through the grid on the front of the (TEou rectangular) ESR cavity. Light from a 300-W Hg arc lamp source was passed first through 5 cm 1% CuSOl and then a long pass filter which cut off wavelengths below 345 nm (Oriel). Light intensity was measured at the front of the ESR cavity using a YSI Model 65A radiometer (24). Salicyhte trapping and HPLC-EC analysis. Salicylic acid was prepared aa a 32 mM stock solution in 30 mh4 citrate/27 mM acetate, pH

ACTIVATED

OXYGEN

SPECIES

FROM

MEROCYANINE

45

540

FeSO, (1 mM, prepared immediately before use) to a sample containing dye-treated liposomes and catalase that had been irradiated 30 min at 80 W/m’. Bovine erythrocyte superoxide dismutase (SOD), obtained from Boehringer-Mannheim, was prepared as a working stock solution of 0.5 mg/ml(l500 U/ml) in PB-H1O and used at a final concentrations of 60-180 pg/ml as noted in the legends. The activity of SOD and its stability during MC540-mediated sensitization was confirmed by its ability to inhibit the xanthine/xanthine oxidase mediated reduction of cytochrome c (37).

RESULTS

DARK

CONTROL

NI SATURATED

FIG. 1. DMPO spin adducts formed with MC540-treated liposomes. Samples of 20 mM DLPC/DCP liposomes, 0.20 mM MC540, and 45 mu DMPO were contained in a gas-permeable TPX capillary. The top sample (A) was irradiated with visible light at 140 W/m’ for 7 min under a continuous flow of air; (B) same conditions except without irradiation; (C) sample was purged continuously with NP. The bottom sample contained 2% (350 mM) ethanol, and additional hyperfine components are due to DMPO/.CH(OH)CH,.

4.75, and stored at 4’C? Liposomes were treated with MC540 as described above, and 50 mM phosphate buffer (t additives) and SA were added to give 5 mM lipid, 0.05 mM MC540, and 6.4 mM salicylate in a final volume of 0.5-1.0 ml. Samples were irradiated in disposable polycarbonate cuvettes with continuous bubbling of either air or (where noted) Nz. At desired time points, lOO-/.d aliquots were removed for HPLCEC analysis. HPLC was run on a Beckman system with a 25 cm X 4.6 mm Whatman Partisil 10 ODS-3 Cl8 reverse phase column and an Alltech macrosphere 300 Cl8 guard column. The electrochemical detector (EG & G Princeton Applied Research Model 400) was set at 600 mV in the oxidative mode and was interfaced to a Beckman 427 integrator. A total of 20 ~1 of sample was injected by overfilling a 20-1.11 sample loop and eluted at 1.0 ml/min with a mobile phase of 30 mM citrate/27 mM acetate, pH 4.75 (19). Elution times and the detector/integrator response to 2,3-DHBA and 2,5-DHBA were calibrated each day with authentic standards. Experiments with SOD and catahe. Beef liver catalasa (BoahringerMannheim) was freshly diluted to a working stock solution of 0.25 mg/ ml (approximately 16,250 U/ml) in PB-Hz0 and then added to MC540treated liposomes to give a final concentration of between 50 and 150 pg/ml. The stability of catalase during MC540-mediated photosensitization was tested by determining the ability of the enzyme to prevent hydroxylation of SA to DHBAs upon addition of H202 (1.4 mM) and

’ Salicylic acid was not soluble to 32 mM either in deionized water or in pH 7.4 phosphate buffer, but readily dissolves in the citrate/acetate (pH 4.75) running buffer. This stock solution gave extremely reproducible results over at least 4-6 weeks, but did decay with extended (e.g., 4 months) storage.

Spin trapping with DMPO. Irradiation of MC540treated liposomes in gas-permeable TPX capillaries under a continuous flow of air (or in a quartz flat cell) with visible light (>345 nm) in the presence of DMPO produced an ESR signal with hyperfine splittings (aN = aF = 15 G) characteristic of the hydroxyl radical adduct of DMPO (DMPO/ * OH) (Fig. 1A). No signal was observed in the dark (Fig. 1B). When samples contained in a TPX capillary were irradiated under a continuous flow of Nz, only a very weak signal was observed (Fig. 1C). The addition of either Desferal (lo-40 PM) or catalase (150 pg/ml) to the reaction mixture had negligible effects on the ESR signal, yielding integrated ESR signal intensities >90% of the control. This indicates that metal-catalyzed redox reactions, and particularly the metal-catalyzed decomposition of hydrogen peroxide, were not involved in formation of the observed DMPO spin adducts. No signal from the Desferal nitroxide free radical (25) was observed under our experimental conditions. In the presence of methanol a six-line spectrum (marked X in Fig. 2A) was detected which was assigned to the DMPO/ . CHzOH adduct (aN = 15.75 G; a? = 22.5 G), consistent with hydrogen abstraction from methanol by . OH and subsequent trapping of the * CHzOH radical by DMPO. Assignment of this adduct was confirmed by isotopic substitution. Replacement of [12C]methanol with [13C]methanol resulted in the production of the DMPO/ ‘13CH20H spin adduct (Fig. 2B; aN = 15.75 G, a: = 22.5 G, ak3” = 8.5 G) (26). This indicates that at least part of the DMPO/ * OH adduct observed in Fig. 1A was due to free *OH. The addition of ethanol during irradiation of MC540treated liposomes resulted in diminished production of the DMPO/ . OH adduct and appearance of a six-line ESR spectrum (Fig. 3A). Based on previous studies (27) the six-line spectrum was assigned to the two diastereomers of DMPO/ . CH(OH)CH3 (a N = 15.75 G, a: = 23.5 G and aN = 16.0 G, a,H - 22.5 G). Formation of a small amount of the DMPO-superoxide spin adduct (DMPO/ * OOH) was also observed in the presence of ethanol. This was sensitive to superoxide dismutase (Fig. 3B). Superoxide is presumably formed in a side-reaction from decomposition of the a-hydroxy peroxy radical of ethanol (28). CH,CHOH I 00.

--, CH3CH0

+ - OOH

46

FEIX (A) ’ 2C-Methanol

AND

KALYANARAMAN (A) Mc54O/Lipos/Ethanol

x

x

10G

(B) 13C-Methanol 0

IC) + Azide

FIG. 2. Effects of [“C]methanol and [‘%]methanol on DMPO spin trapping. Samples containing liposomes, MC540, and 45 mM DMPO, and 350 mM [“C]methanol (A) or [‘%]methanol (B) were irradiated at 140 W/m2 in a TPX capillary with continuous air purging. (0) DMPO/ . OH, (X) DMPO/‘sCHsOH, (0) DMPO/‘sCH,OH.

This is, however, a minor process in MCMO-mediated radical production as the total spin intensity in the presence of superoxide dismutase was decreased by only 12% relative to the control. No DMPO/ * OOH adduct was observed in samples lacking ethanol, and SOD had no effect on the intensity of the DMPO/ * OH adduct, indicating that DMPO/ * OH was not formed via the decomposition of DMPO/ . OOH (29, 30). By far the most significant inhibition of DMPO adduct formation was obtained with sodium azide (Fig. 3C), suggesting that ‘02 might be involved in the production of spin adducts. Low concentrations of azide (2 mM or less) were employed to avoid scavenging of +OH by azide, and no DMPO-azide spin adduct (31) was observed. To further test for ‘02 involvement, liposomes were prepared in PBDzO. Solvent deuteration (95% replacement of Hz0 with DzO) clearly enhanced the rate of DMPO/ * OH formation (Fig. 4). We also examined the effects of histidine, a commonly employed chemical quencher of ‘02, on DMPO/. OH production. Initial studies suggested that histidine completely blocked spin adduct formation. However, further examination of the reaction kinetics demonstrated that

FIG. 3. Effects of ethanol, SOD, and azide on DMPO spin trapping with MC540-sensitized liposomes. (A) A MC540-treated liposome suspension containing 45 mM DMPO and 350 mM ethanol after irradiation in a quartz flat cell for 8 min at 80 W/m’. Open triangles denote DMPO/ * OOH; other symbols are as in Fig. 2. (B) Same as A with the addition of 0.1 mg/ml SOD. (C) As in A with the addition of 2 mM sodium azide.

I

0

2

4

IRRADIATION

6

TIME

8

(min)

FIG. 4. Enhancement of DMPO-hydroxyl adduct formation in deuterated solvent. MCMO-treated liposomes were prepared in either PBHz0 or PB-D,O. After a 15-s irradiation to adjust the field to the maxima of the second low-field peak (inset), the field sweep was turned off, and the kinetics of adduct formation followed. Irradiation was at 40 W/m’, and the DMPO concentration was 22.5 mM.

ACTIVATED

OXYGEN

SPECIES

DMPO/ OH was transiently produced and then rapidly decayed with further irradiation (data not shown). We attribute this to photochemical destruction of DMPO/ * OH by a Type I process, which would be favored following depletion of O2 via the reaction of ‘Oz with histidine. Scavenging experiments with methanol and ethanol clearly indicated the formation of freely diffusing * OH during irradiation of MCMO-treated liposomes in the presence of DMPO. The experiments with PB-DBO, azide, and histidine indicated ‘Oz intermediacy in the formation of spin adducts. To further examine the origins of OH formation and to investigate the role of DMPO itself in spin adduct production, we employed an independent assay for hydroxyl radical which did not require a spin trap, i.e. analysis of salicylic acid reaction products by HPLC with electrochemical detection. HPLC analysis of salicylic acid trapping. As mentioned earlier, the reaction of salicylic acid with * OH produces as major products 2,3- and 2,5dihydroxybenzoic acids, which are readily analyzed by HPLC with electrochemical detection (19, 20). The HPLC chromatogram obtained with equimolar amounts of 2,3- and 2,5-DHBA is shown in Fig. 5A. Figure 5B shows the chromatogram obtained by SA trapping of free . OH produced by uv photolysis of H202 (20). Approximately equimolar amounts of 2,3DHBA and 2,5-DHBA are observed. In contrast, irradiation of MC540-treated liposomes in the presence of SA resulted predominantly in the production of the 2,5DHBA isomer (Fig. 5C). The ratio of 2,5-DHBA to 2,3DHBA produced during MC540-mediated photosensitization ranged from 5:l to 1O:l in samples prepared in PBHzO, to as high as 3O:l in samples prepared in PB-D20. Irradiation of a mixture of MCMO-treated liposomes with 2,3- and 2,5-DHBA standards demonstrated that both isomers were stable, indicating that selective destruction of 2,3-DHBA was not the cause of the 2,5-DHBA predominance. The rate of MCMO-mediated 2,5-DHBA formation was enhanced in air-saturated samples prepared with PB-DxO (Fig. 6). However, no DzO enhancement was observed in samples bubbled with Nz or argon, and Nz or argon saturation greatly diminished DHBA formation. No DHBA products were observed in the absence of either dye or light. To test whether the 2,5-DHBA was produced by SA trapping of the hydroxyl radical, we examined the effects of a variety of scavengers (Table I). Ethanol, a facile . OH scavenger, had a negligible effect on the production of DHBAs even when present in large excess. Catalase, Desferal, and SOD also had little effect of 2,5-DHBA production, in agreement with the DMPO spin trapping results. Also consistent with the spin trapping results, DHBA production was quite effectively inhibited by sodium azide even at relatively low concentrations. These results are all consistent with the absence of *OH and l

l

FROM

MEROCYANINE

47

540

with the intermediacy of ‘02 in salicylic acid hydroxylation. As a final assessment of the interaction of DMPO with ‘Oz and the possible involvement of DMPO in . OH production, we examined the effects of DMPO addition on DHBA formation. As shown in Table II, addition of excess DMPO inhibited the formation of 2,5-DHBA from SA in a concentration-dependant fashion and enhanced the production of 2,3-DHBA. We suggest these observations arise from trapping of ‘02 by DMPO, followed by decay of the DMPO-l0z intermediate to produce hydroxyl radical (discussed below). DISCUSSION There are four potential sources for the DMPO/ . OH spin adduct we observe during MCMO-mediated photosensitization of DMPC liposomes (Scheme 1): (I) direct interaction of the MC540 triplet excited state (MC*) with the spin trap, (II) production of superoxide anion followed by trapping of 0; by DMPO and decay of DMPO/ - OOH to DMPO/ OH (29, 30), (III) production of 0; and Haber-Weiss cycling leading to formation of OH, and (IV) trapping of ‘02 by DMPO and subsequent decay to DMPO/ * OH and free * OH. Similar reactions can be written for the production of DHBA products from salicylic acid by direct interaction with MC* (I), OH derived from Fenton chemistry (III), or the reaction of SA with lOz(IV). Salicylic acid, unlike DMPO, does not react at an appreciable rate with 0; (21). l

l

l

MC*

+ DMPO

“z

MC + DMPO/.

MC’YMC

4

MC + 0;

OMFO. n+ O? DMPO/*OOH

+

MC*r=?b,lCT 205 Fe(lll)

OH

DMPO/.

OH

2 MC + 0;

+ 2H+ +

H202 + O2

+ 0; +

Fe(N) + O2

H202 + Fe(ll) + Fe(lll)

+ OH-

+ . OH $ DMPO

DMPO/

* OH

M&MC+‘Oz ‘02 + DMPO [DMPO

- ‘02] :

+

DMPO/.

[DMPO

- ‘02]

OH + . OH 4 OMPO

DMPO/.OH SCHEME

1

Direct reaction of DMPO with MC* (reaction I) may occur to a small degree, but is unlikely to be the major source of DMPO/ . OH, as indicated by the greatly diminished spin adduct formation in the absence of oxygen (Fig. 1C). Direct interaction of MC* with DMPO also

48

FEIX

AND

KALYANARAMAN

W&JV

STANDARDS

L* 0

0

1

MC54O/visible

(I”“’

Time

20

‘lmin)

FIG. 5. HPLC with electrochemical detection of 2,3- and 2,klihydroxybenzoic acids. Samples containing 2,3-DHBA and 2,5-DHBA were subjected to reverse-phase HPLC as described in the text and detected by electrochemical oxidation on a glassy carbon electrode (600 mV against Ag/AgCl reference). (Left) Standard sample containing 200 pmol each 2,3- and 2,5-DHBA, (Middle) products from the uv photolysis of 0.75 mM HzOz in the presence of 6.4 mM salicylic acid; (Right) products from the visible irradiation of MC540-treated liposomes (5 mM DLPC/DCP, 50 pM MC540) and 6.4 mM salicylic acid.

does not explain the pronounced enhancement of spin adduct formation in PB-D20. Direct reaction of MC* with SA may give rise to the formation of 2,5-DHBA in samples purged with Nz or argon, and it is notable that there is

no DzO enhancement of 2,5-DHBA production in Napurged samples. SOD, catalase, and Desferal had little effect on the production of either DMPO spin adducts or 2,5-DHBA. The TABLE

400

Relative .

300

A z E a

2,5-DHBA 0 PB-H20, l PB-C&O, A PB-H20, A PB-D20,

PRODUCTION Nitrogen Nitrogen Air Air

Yields of 2,5-DHBA during Irradiation of MC540Treated Liposomes in the Presence of Inhibitors

/I

Control, PB-Hz0 Control, PB-D,O 700 mM ethanol Sodium azide (2.4 Catalase (50 pg/ml)’ SOD (60 pg/ml) * Desferal (20 pM)

s I D 100

0

10

20

30

Time FIG. 6. Effects of production. Samples (m, A.) and irradiated bubbling of either air indicated time points EC. Results are from

(rn;:)

2,5-DHBA (PM)

Sample

-200

0

I

50

60

solvent deuteration and N,-purging on 2,5-DHBA were prepared in either PB-Hz0 (Cl, A) or PB-DIO (80 W/m’) as described in the text with continuous (A, A) or Nz (Cl, n) in a disposable cuvette. At the aliquots were removed and analyzed by HPLCone of three similar experiments.

mM)

6.43 9.83 6.16 2.40 6.54 6.04 5.53

% Yield 100 152.8 95.9 37.2 101.7 93.9 86.9

Note. Samples containing 5 mM DLPC, 50 pM MC540, and 6.4 mM SA were irradiated for 30 min at 80 W/m” with visible light as described in the text. Similar relative yields were observed at all irradiation times examined (10-60 min, e.g., Fig. 6). ’ 3000 units/ml, where 1 unit decomposes 1 pmol HzOz per min. This concentration of catalase gave >80% inhibition of DHBA formation from uv photolysis of 1.4 mM H202, whereas denatured enzyme was without effect. * 300 units/ml. SOD activity was independently verified by ita ability to inhibit the xanthine/xanthine oxidase mediated reduction of cytochrome c (37).

ACTIVATED

OXYGEN

SPECIES

lack of inhibition by catalase shows that H202 is not an intermediate in the formation of these products, indicating that the production of . OH by Fenton chemistry (reaction III) is not occurring in this system. Similarly, the lack of inhibition by either Desferal or SOD indicates that superoxide-mediated reduction of Fe+3 (reaction III) is not serving to drive the production of the hydroxyl radical. The lack of significant inhibition by SOD also indicates that DMPO/ * OH formation is not due to trapping of 0; by DMPO and subsequent decay of DMPO/ . OOH (reaction II). There is a concern that these inhibitors (SOD, catalase, Desferal) could be ineffective due to their localization in the aqueous phase, since MC540 is intercalated into the lipid bilayer. However, DMPO partitions primarily into the aqueous phase (39), so that reactive species must leave the membrane bilayer before they can react with DMPO to any appreciable extent. We also note that sodium azide, located in the aqueous phase, very effectively inhibited both DMPO spin adduct and 2,5-DHBA formation. The enhancement of adduct formation observed in PBDzO and the effects of azide quenching are indicative of ‘02 intermediacy (reaction IV). Nitrones can interact with ‘02 by both p h y sical quenching (32) and chemical reaction (18). Harbour et al. have previously reported that ‘02, generated by methylene blue photosensitization, reacts chemically with several nitrone spin traps, including DMPO (18). Although the reaction products of ‘02 with several nitrones (including a-phenyl-N-tert-butylnitrone) were diamagnetic, reaction of DMPO with ‘02 was reported to give an ESR spectrum corresponding to the hydroxyl radical adduct (18). This is consistent with our studies indicating ‘02 intermediacy in the formation of DMPO/ . OH. Harbour et al. did not address the effects of OH scavengers in their study of ‘02 spin trapping. Our results with methanol and ethanol (Figs. 2 and 3) indicate that freely diffusing * OH is produced in the system containing MCMO-treated liposomes and DMPO. We suggest that free * OH could arise from decay of the DMPO-102 reaction intermediate. Consistent with the idea that the spin trap is intimately involved in *OH production are our HPLC-EC studies which show that generation of 2,5l

TABLE

II

Effect of DMPO on DHBA

Sample

2,3-DHBA formed bnol)

Control +50 mM DMPO +lOO mM DMPO +299 mM DMPO Note.

Experimental

Production 2,5-DHBA formed (pm4

25.3 46.8 48.2 34.3 conditions

were

248.4 224.0 195.4 131.0 as described

in Table

I.

FROM

MEROCYANINE

540

49

DHBA from SA, in the absence of DMPO, is completely insensitive to ethanol (Table I). An analogous process is known to occur in the trapping of 0; by DMPO, in which decay of the DMPO/ . OOH spin adduct produces DMPO/ -OH and releases a small fraction of free . OH that may be scavenged by ethanol to produce DMPO/ . CH(OH)CH3 (30). Although formation of DMPO/ . OH via superoxide is well known, the lack of significant inhibition by SOD indicates this is not the principle mechanism of adduct formation in the present study. A comparison of our experimental results with those predicted for mechanisms I-IV using the various diagnostic tests is summarized in Table III. Clearly mechanism IV, reaction of DMPO with ‘02 followed by decay to DMPO/ - OH and free * OH (which may be subsequently trapped by DMPO or scavenged, e.g., by ethanol), is the only proposed mechanism that is consistent with all of our experimental results. Our HPLC studies are also consistent with the production and trapping of ‘02 during MCMO-mediated photosensitization. Formation of 2,5-DHBA was sensitive to 02 concentration, enhanced in D20, and inhibited by azide. The inability of ethanol to inhibit 2,5-DHBA production by scavenging . OH (Table I) indicates that the free hydroxyl radical is not formed in systems lacking DMPO. The almost exclusive production of the 2,5DHBA isomer also indicates that this is not a . OH-mediated process.4 We interpret the effects of DMPO on DHBA formation (Table III) as follows: Reaction of DMPO with ‘02 competitively inhibits the SA-‘02 reaction which produces 2,5-DHBA, resulting in a decreased yield of 2,5-DHBA. Decay of the DMPO-102 intermediate gives rise to free * OH which reacts with SA to give the observed enhancement in the amount of 2,3-DHBA formed (as well as, presumably, to replace some 2,5-DHBA). The effect of DMPO concentration on 2,3-DHBA formation is expected to be complex, since . OH produced by DMPO-102 decay may also be trapped by DMPO, preventing enhancement of 2,3-DHBA formation (especially at high DMPO concentrations). These observations (Table II) were predicted by reaction mechanism IV (Scheme 1). We are currently investigating the basis for the selective production of 2,5-DHBA in preference to the 2,3 isomer. Initial studies utilizing SA to trap *OH found 2,3- and 2,5-DHBA to be produced in an approximately equimolar ratio (19, 20), and we verified this for uv photolysis of HzOz (Fig. 5). A recent study of SA trapping of - OH produced either radiolytically or by Fenton chemistry found 2,3-DHBA to be produced in excess of the 2,5 isomer (33). ’ During the preparation of this manuscript, we became aware of a similar observation, i.e., predominant 2,5-DHBA production and very little 2,3-DHBA formation, reported by Floyd et al. for hematoporphyrin derivative (38). We have now obtained similar results with rose bengal and aluminum phthalocyanine tetrasulfonate.

50

FEIX TABLE Predicted

Mechanism I 11 III IV Exptl.

and Observed on DMPO/ SOD

CAT

DzO

Azide

Hist

---+-+----+ -

Tests

C,H60H --

t -

+ -

KALYANARAMAN

III

Effects of Diagnostic - OH Formation

DFO

AND

+ +

Nz

+

-

t

+ +

t

+

t

t

+

t

+

Note. Mechanisms refer to those defined in the text. (+) Indicates an observable effect (enhancement, inhibition, or further splitting) in the production of DMPO/* OH at the concentrations used in the text; (-) indicates no observable effect. DFO, Desferal; Exptl., experimental results of this study.

The precise product ratio in that study was very sensitive to the presence of oxidants, but in no instance resulted in an excess of 2,5-DHBA, as was always the case for the (putative) ‘09-mediated reaction reported here. Floyd et al. reported an excess of the 2,5-DHBA isomer when SA was used to trap activated oxygen species (presumably . OH) in uiuo in adriamycin-treated rats (34). The basis for the increased yield of the 2,5 isomer in that study was not known, but may have been related to selective metabolism of 2,3-DHBA (34) or to metabolic production of 2,5-DHBA (40). As noted under Results, we verified that in our system both 2,3- and 2,5-DHBA were completely stable during the photochemical reaction. It has recently been suggested that 2,3-DHBA is the more reliable indicator of *OH production in uiuo or in metabolically active systems (40). We concur with this and propose that the almost exclusive production of 2,5DHBA may be a selective indicator of ‘02 in photochemical systems and in biochemical systems lacking metabolic activity. Preliminary studies with rose bengal and aluminum phthalocyanine tetrasulfonate, which are effective generators of ‘02, support this proposal. In contrast to addition of . OH, the reaction of ‘02 with substituted phenols may occur either by cycloaddition to form an unstable epoxide which decays to the peroxide or hydroxy adduct or by a charge-transfer reaction (35). Both mechanisms would likely lead to the 2,5-DHBA product. Direct interaction of the photosensitizer triplet state with the substrate has been reported to be important in the eosin-sensitized photooxidation of hydroxylated phenylalanines (36), and such an interaction between MC540 and salicylic acid may be the basis for DHBA production under anaerobic conditions-where no DzO effect is observed (Fig. 6). Previous work has focused on the use of SA as a selective trap for . OH (19-21, 35). Our studies suggest that SA may also be used to trap ‘02 and that it may be possible to differentiate between the trapping of singlet oxygen and the hydroxyl radical based on the relative molar ratio

of 2,3- and 2,5-DHBA produced. The stability of SA and its DHBA products, both in vitro and in uiuo, make this an attractive method for elucidating the production of activated oxygen species,and factors which influence the stoichiometry of isomers produced should be further investigated. In conclusion, formation of the hydroxyl adducts of DMPO and salicylic acid during MCMO-mediated photosensitization appears to arise from the trapping of singlet molecular oxygen. This represents, for the first time, a novel pathway for the production of products characteristic of a Type I reaction via a Type II mechanism. Superoxide and Hz02 have been shown not to be intermediates in these processes.Decay of the DMPO-‘02 intermediate apparently gives rise to some free * OH (which may be scavenged by ethanol or methanol), furthering the previous observations of the reaction between ‘02 and this nitrone spin trap (18). HPLC-EC is a valuable adjunct to spin trapping in the characterization of activated oxygen intermediates and may, based on the selective production of 2,5-DHBA, provide a sensitive analytical method for establishing ‘02 intermediacy. ACKNOWLEDGMENTS This work was supported The authors wish to thank preparation of this paper.

by USPHS Grants CA-49089 and RR-01008. Margaret Wold for her expert assistance in

REFERENCES 1. Sieber,

F. (1987)

Photo&em.

Photobid.

46,

71-76.

2. Sieber, F., O’Brien, J. M., Krueger, G. J., Schober, S. L., Burns, H., Sharkis, S. J., and Sensenbrenner, L. L. (1987) Photochem. Photobiol.

46,707-711. 3. Sieber, F., Krueger,

G. J., O’Brien, J. M., Schober, S. L., Sensnbrenner, L. L., and Sharkis, S. J. (1989) Blood 73,345-350.

4. Kalyanaraman, 5.

B., Feix, J. B., Sieber, F., Thomas, J. P., and Girotti, A. W. (1987) Proc. Natl. Acad. Sci. USA 84, 2999-3003. Davila, J., Gulliya, K. S., and Harriman, A. (1989) J. Chem. Sot. Chem. Commun. 1215-1216.

6. Aramendia, lavsky,

7. Hoebeke, Photo&m.

8. Gaffney,

P. F., Krieg, M., Nitsch, C., Bittermann, E., and BrasS. E. (1988) Photo&m. Photobiol. 48, 187-194. M., Seret, A., Piettee, J., and Van de Vorst, Photobid. B 1,437-446. D. K., Schober,

S. L., and Sieber,

F. (1990)

A. (1988)

J.

Exp. Hematol.

18.23-26.

9. Feix, J. B., Kalyanaraman, J. Biol.

Chem.

263,

B., Chignell, 17,247-17,250.

C. F., and Hall,

R. D. (1988)

10. Singh, R. J., Feix, J. B., Pintar, T. J., Girotti, A. W., and Kalyanaraman, B. (1991) Photo&em. Photobid. 53,493-500. 11. O’Brien, J. M., Singh, R. J., Feix, J. B., Kalyanaraman, B., and Sieber, F. (1991) Photo&em. Photobiol., in press. 12. Sarna, T., Pilas, B., Lambert, C., Land, E. J., and Truscott, T. G. (1990) Photo&m. Photobiol. 51, 42s. 13. Feix, J. B., and Kalyanaraman, B. (1991) Photochem. Photobiol. 63, 39-45. 14. Janzen, E. G. (1980) Free RadicaLv Bid. 4,115-154. 15. Mason, R. P. (1984) in Spin Labeling in Pharmacology J. L., Ed.), pp. 97-129, Academic Press, New York.

(Holtzman,

ACTIVATED 16. Chignell, C. F., Kalyanaraman, (1980) Photochem. Photobiol.

OXYGEN

B., Mason, 32,563-571.

R. P., and Sik, R. H.

17. Alegria, A. F., Murali Krishna, C., Elspuru, (1989) Photochem. Photobiol. 49, 257-265. 18. Harbour, J. R., Issler, Sot. 102.7778-7779. 19. Grootveld,

S. L., and Hair,

M., and Halliwell,

B. (1986)

21. Halliwell,

B. (1977)

22. Buettner, Commun.

G. R., and Oberley, 83,69-74.

Biochem.

J. 167,

&o&m.

24. Kalyanaraman, B., Felix, Photobiol. 36, 5-12. 25. Morehouse, Lett. 222,

K. M., Flitter, 246-250.

J. 237,

Biochem.

G., and Cullis,

C. C., and Sealy,

Biophys.

Res.

P. R. (1985) Biochim.

R. C. (1982)

W. D., and Mason,

Photo&em.

R. P. (1987)

B., and Kirk,

27. Towell, 119.

Anal.

B. (1991)

499-504.

J. B&hem.

317-320.

26. Hammel, K. E., Tien, M., Kalyanaraman, J. Biol. Chem. 260,8348-8353. J., and Kalyanaraman,

P.

J. Am. Chem.

P. K. (1984)

L. W. (1978)

23. Hope, M. J., Bally, M. B., Webb, Biophys. Acta 812,55-65.

R. K., and Riesz,

M. L. (1980)

20. Floyd, R. A., Watson, J. J., and Wong, Biophys. Methods 10, 221-235.

SPECIES

Biochem.

28. Bothe, E., Schuchmann, M. N., Schulte-Frohlinde, Sonntag, C. (1978) Photochem. Photobiol. 28,639-644.

FEBS

T. K. (1985) 196,

lll-

D., and Von

FROM

MEROCYANINE

540

51

29. Finkelstein, E., Rosen, G. M., and Raukman, E. J. (1982) Mol. Pharmacol. 2 1,262-265. 30. Lai, C.-S., and Piette, L. H. (1977) B&hem. Biophys. Res. Commun. 78,51-59. 31. Kalyanaraman, B., Janzen, E. G., and Mason, R. P. (1985) J. Biol. Chem. 260,4003-4006. 32. Ching, T.-Y., and Foote, C. S. (1975) Tetrahedron L&t. 44,37713774. 33. Maskos, Z., Rush, J. D., and Koppenol, W. H. (1990) Free Radicals Biol. Med. 8, 153-162. 34. Floyd, R. A., Henderson, R., Watson, J. J., and Wong, P. K. (1986) Free Radicals Biol. Med. 2, 13-18. 35. Thomas, M. J., and Foote, C. S. (1978) Photo&em. Photobiol. 27, 683-693. 36. Rizzuto, F., and Spikes, J. D. (1977) Photo&m. Photobiol. 26,465476. 37. McCord, J. M., and Fridovich, I. (1969) J. Biol. Chem. 244,60496055. 38. Floyd, R. A., West, M. S., Schneider, J. E., Watson, J. J., and Madidt, M. L. (1990) Free Radicals Biol. Med. Q(Supp1. l), 96. 39. Turner, M. J., and Rosen, G. M. (1986) J. Med. Chem. 29, 24392444. 40. Ingelman-Sundberg, M., Kaur, H., Terelius, Y., Persson, J.-O., and Halliwell, B. (1991) Biochem. J. 276, 753-757. 41. Feix, J. B., and Kalyanaraman, B. (1991) Photochem. Photobiol. 53, 104s.