Photochemistry und Photobiology Vol. 53, No. 4, pp. 481-491, 1991 Printed in Great Britain. All rights reserved

0031-8655191 $03.00+0.00 Copyright 0 1991 Pergamon Press pic

PHOTOSENSITIZED LIPID PEROXIDATION AND ENZYME INACTIVATION BY MEMBRANE-BOUND MEROCYANINE 540: REACTION MECHANISMS I N THE ABSENCE AND PRESENCE OF ASCORBATE* GARYJ. BACHOWSKI, THOMAS J. PINTAR and ALBERT W. GIROTW Department of Biochemistry, Medical College of Wisconsin, Milwaukee, WI 53226. U.S.A. (Received 15 August 1990; accepted 22 October 1990)

Abstract-The lipophilic photosensitizing dye merocyanine 540 (MC540) is being studied intensively as an antitumor and antiviral agent. Since plasma membranes are believed to be the principal cellular targets of MC540-mediated photodamage. we have studied membrane damage in a well characterized test system, the human erythrocyte ghost. When irradiated with white light, MC54O-sensitized ghosts accumulated lipid hydroperoxides (LOOHs derived from phospholipids and cholesterol) at a rate dependent on initial dye concentration. Neither desferrioxamine nor butylated hydroxytoluene inhibited LOOH formation, suggesting that Type I (iron-mediated free radical) chemistry is not important. By contrast, azide inhibited the reaction in a dose-dependent fashion, implicating a Type I1 (singlet oxygen, lo,) mechanism. Stern-Volmer analysis of the data gave a '0, quenching constant - 50 times lower than that determined for an extramembranous target, lactate dehydrogenase (the latter value agreeing with literature values). This suggests that '0, reacts primarily at its membrane sites of origin and that azide has limited access to these sites. Using [ i4C]cholesterol-labeledmembranes and HPLC with radiodetection, we identified 3~-hydroxy-5a-cholest-6-ene-5-hydroperoxide as the intermediacy. Irradiation of MC540-sensitized major cholesterol photoproduct, thereby confirming '02 membranes in the presence of added iron and ascorbate resulted in a large burst of lipid peroxidation, as shown by thiobarbituric acid reactivity and appearance of 7-hydroperoxycholesteroland 7-hydroxycholesterol as major oxidation products. Amplification of MC540-initiated lipid peroxidation by iron/ ascorbate (attributed to light-independent reduction of nascent photoperoxides, with ensuing free radical chain reactions) could prove useful in augmenting MC540's phototherapeutic effects.

a test system, the isolated plasma membrane of the

INTRODUCTION

Merocyanine 540 (MC540)$ is a photosensitizing dye of great therapeutic interest in connection with its ability to selectively inactivate neoplastic cells (e.g. leukemia and lymphoma cells) in autologous bone marrow grafts and also to eradicate pathogenic enveloped viruses in blood specimens (Sieber, 1987; Sieber et al., 1987). Since MC540 binds avidly to cell membranes, these structures are believed to be the primary targets of its photokilling activity (Sieber, 1987). Preliminary studies carried out with

*This paper is dedicated to the memory of Fumito Taketa, an esteemed colleague and friend. ?To whom correspondence should be addressed. $Abbreviations: ACE, acetylcholinesterase; AH-, ascorbate; BHT, butylated hydroxytoluene (2,6-di-t-butyl-4methylphenol; Ch, cholesterol; 5a-OH; 5a-cholest-6en-3P,5-diol; 7a/7P-OH, cholest-5-en-3P,7anp-diol; DCP, dicetylphosphate; DFO, desfemoxamine; DMPC, dimyristoylphosphatidylcholine; HPLC, high performance liquid chromatography; 5a-00H, 3P-hydroxy-5acholest-6-ene-5-hydroperoxide; 7a/7P-O0H, 3p-h~droxycholest-5-ene-7a/7P-hydroperoxide; LDH, lactate dehydrogenase; LOOH, lipid hydroperoxide; MC540, merocyanine 540; PBS, phosphate-buffered saline; PC, phosphatidylcholine; PE, phosphatidylethanolamine; PS, phosphatidylserine; SM, sphingomyelin; TBA, 2thiobarbituricacid; TBARS, thiobarbituric acid reactive substance(s); TLC, thin layer chromatography; TMPD, N,N,N',N'-tetramethyl-p-phenylenediamine. 481

human erythrocyte, have shown that MC540 can sensitize photodamage in the form of lipid peroxidation (Kalyanaraman et al., 1987). Singlet molecular oxygen (lo2) appeared to play a significant role in the reaction. In the present study, MC540's primary photochemistry in this natural membrane has been examined more rigorously by (i) performing competitive kinetic analyses with azide ion, an avid '02 quencher, and (ii) using endogenous radiolabeled cholesterol as a quantitative indicator of Type I1 (lo2)vs Type I (free radical) chemistry. An additional objective was to determine whether ascorbate, a widely distributed physiological reductant, might exacerbate the damaging effects of MC54O-mediated photoperoxidation as it does with other sensitizers (Girotti et al., 1985; Bachowski er al., 1988a). MATERIALS AND METHODS Materials. Freshly drawn human blood in citrate phosphate-glucose medium was obtained from the Blood Center of SoutheasternWisconsin. Sourcesof other materials were as follows: (a) MC540 from Eastman Kodak Co. (Rochester, NY); (b) Rose Bengal from Allied Chemical Co.(Morristown,NJ); (c) NADH. sodium pyruvate, acetylthiocholine, 5.5'-dithiobis-(2-nitrobenzoicacid), DMPC, egg PC. liver PE, Ch, 7-ketocholesterol,xanthine, xanthine oxidase, superoxide dismutase, and catalase (thymol-free)from Sigma Chemical Co. (St. Louis, MO); (d) sodium ascorbate from BDH Chemicals (Poole, England); (e) 2-thiobarbituric

GARYJ. BACHOWSKI et al.

482

acid and N,N,N’,N’-tetramethyl-p-phenylenediamine from Aldrich Chemical Co. (Milwaukee, W1); ( f ) desferrioxamine from Ciba-Geigy (Suffem, NY); (9) [4-1JC]cholesterol (50 mCimmol) from Research Products International (Mount Prospect, IL); (h) HPLC-grade organic solvents from Burdick and Johnson Corp. (Muskegon, MI). All aqueous solutions were prepared with deionized, glass-distilled water. Membrane preparations. Isolated erythrocyte membranes (unsealed ghosts) were prepared by conventional hypotonic lysis (Fairbanks et al., 1971). Membranes in PBS were stored under argon at 4°C and used within a fortnight. Membrane protein was determined by the method of Lowry el al. (1951), using serum albumin as the standard. Ghost membranes were exchange-radiolabeled with [ lJC]cholesterol by incubating with unilamellar [ 14C]cholesterol/eggPC liposomes (0.8: 1.O, mol/mol) as described previously (Bachowski et af., 1988b). After 48 h at 3 7 T , the ghosts were washed extensively with PBS to remove the liposomes and resuspended to a final concentration of 1.8 X lO’/mL (0.9 mg IipidlmL). Small unilamellar egg PC, egg PCICh, and DMPC/DCP liposomes were prepared by sonication, as described previously (Bachowski et al., 1988b). Photoreaction conditions. Stock solutions of MC540 ( 5 mM in ethanokwater ( l : l , vol/vol) were stored in the dark at 4°C. When added to membranes, the stored dye’s spectral properties were highly reproducible over at least a 2 week period. The typical reaction mixture consisted of ghost membranes (1.0 mg protein/mL or 0.67 mM phospholipid) and 25 pM MC540 in PBS (phospholipid to dye ratio 27:l mol/mol). Unless indicated otherwise, irradiation reactions were carried out aerobically at 25°C in thermostatted Stirrer Bath vials (Yellow Springs Instruments, Yellow Springs, OH). The light source was a Duro Test R40 Hg arc/fluorescent lamp positioned above the 4place Stirrer Bath. Incoming light was passed through pane glass to minimize wavelengths below 300 nm. The fluence rate was measured with a Yellow Springs radiometer (Model 65A). Since the probe detector could not be properly aligned in the reaction chambers, fluence rates were measured indirectly (externally), using MC540 bleaching as a convenient indicator. A stock suspension of 1 mM egg PC liposomes containing 25 p M MC540 in PBS was used for measuring dye bleaching as a function of fluence rate. This gave a linear calibration plot (slope 3.5 x W 4cmz/mW min-I), from which the fluence rate for bleaching in the reaction chambers could be determined. Experimental reactions were typically carried out at fluence rates of 30-40 mW/cm*. Samples were removed periodically for various determinations: lipid peroxidation, enzymatic activities, TLC and HPLC. Measurements of lipid peroxidation. Photogenerated LOOHs were determined by iodometric assay (Girotti et al., 1985; Thomas et al., 1990). In this approach, total peroxide content of extracted lipids is evaluated in terms of triiodide generated during anaerobic oxidation of iodide by LOOH. The triiodide (equimolar with starting LOOH) is measured spectrophotometrically at 353 nm. Quantitation is based on an extinction coefficient of 22.5 (mM)-lcm-I, which was obtained by using standardized t-butyl hydroperoxide in the assay (Thomas et al., 1990). MC540 did not interfere with the iodometric assay, since virtually all of the dye was extracted into the organic compartment and triiodide was determined in the aqueous compartment. Moreover, MC540 absorbs minimally in the 350-360 nm region. In some instances, e.g. when iron and ascorbate were present in the reaction system, lipid peroxidation was measured by TBA assay, in which TBA adducts of malonaldehyde and other carbonyl by-products of free radical peroxidation are detected spectrophotometrically at 532 nm. It was found that MC540 absorbance interferes severely with the one-phase TBA assay com-

-

-

monly used in this laboratory (Girotti et al., 1976). To circumvent this problem, we used an earlier approach (Girotti, 1979) in which samples are precipited with trichloroacetic acid, followed by TBA treatment of supernatant fractions. In this case, essentially all of the MC540 interference was removed in the precipitate. Absorbance readings at 532 nm were converted to TBARS values (nmol/mg protein), using an extinction coefficient obtained with authentic malonaldehyde, 157 (mM)-lcm-’. Enzyme assays. Activity of acetylcholinesterase in irradiated erythrocyte membranes was measured by coupled assay with acetylthiocholine and 5,5’-dithiobis-(2nitrobenzoic acid) as described previously (Girotti, 1976). Activity of extramembranous lactate dehydrogenase was measured by tracking NADH oxidation (rate of A,,,, decay) in the presence of pyruvate. Assay mixtures of 25°C contained 0.38 mM NADH, 1.5 mM pyruvate and 0.13 UlmL LDH. Kinetic analysis for ‘0,intermediacy. Involvement of lo2in various MCS40-sensitized reactions (lipid peroxidation, ACE inactivation, and LDH inactivation) was checked by kinetic analysis, using azide as a competitive quencher. Data were plotted according to a Stern-Volmer type relationship derived from steady-state kinetic considerations (Foote, 1979):

-

r,Jr = 1 + k,[Qll(kd +k,[Al) (1) where r,, and r are the reaction rates in the absence and presence of azide, respectively; [Q] and [A] are the azide and substrate concentrations; k,is the rate constant for lo, decay in water, 4.5 x los s-I (Wilkinson and Brummer, 1981); and k, and k,, are the rate constants for lo, interception by azide (physical quenching) and substrate (chemical and/or physical quenching), respectively. Eq. (1) simplifies to: r,Jr = 1

+ k,[Q]/k,

when kd k,,[A]. For the reaction systems described, this limitation is satisfied if one considers the “bulk phase” condition. Thin layer chromatography. TLC of phospholipid and cholesterol hydroperoxides was carried out on Silica Gel-60 plates (EM Science, Cherry Hill, NJ) using chloroform: rnethanokwater (75:25:4. voVvol) as the solvent system (Girotti et al., 1985; Thomas et a!., 1990). LOOHs were visualized by spraying with 1% (wtlvol) TMPD in methanol: wateracetic acid (50:50:l,vol/vol). TMPD is oxidized by peroxides to a vivid purple product (Wurster dye) (Smith and Hill, 1972). Immediately after spraying, the plates were clamped under glass to retard background autoxidation of TMPD and photographed as soon as possible. Higher resolution TLC of cholesterol hydroperoxides (e.g. 5 a - 0 0 H and 7-00H) and their corresponding diols (5a-OH, 7a-OH and 7P-OH) was accomplished by using heptane:ethyl acetate (l:l, vollvol) as the solvent system (Smith etal., 1973). In this system, phospholipid hydroperoxides remain at the origin, causing no interference with the detection of cholesterol hydroperoxides (R, 0.35-0.37). Subsequent to development, plates were either sprayed with TMPD or, in the case of [‘4C]cholesterol-labeled ghosts, scanned for radioactivity, using a Radiomatic RTLC Scanner (Model RS) equipped with data processing accessories (Bachowski et al., 1988b). Subsequent to TMPD treatment or radioscanning, the plates were sprayed with 50% H,SO, and warmed at 80°C to visualize cholesterol itself (R, -0.6) and reduction products of cholesterol hydroperoxides (diols: R,. 0.19-0.26). In some instances, extracted samples were reduced with sodium borohydride before chromatographing (Bachowski er al., 1988b). High performance liquid chromatography. The general methodology for HPLC separation of cholesterol photoproducts was adapted from Ansari and Smith (1979) and Smith et al. (1987). An IscolChemresearch HPLC system ))

Merocyanine 540 photosensitization equipped with a standard lsco Si column (4.6 X 250 mm; 5 pm particle size) was used. Injected 10 pL samples were eluted isocratically with hexane:isopropanol(96:4, vol/vol) at a flow rate of 1.5 mL/min. Cholesterol products were either detected optically at 212 nm or (for I4C-labeled products) by radioactivity, using a Radiomatic Flow-One Detector (Model A-120). The detector was equipped with a 2.5 mL flow cell and adjustable discriminator for minimization of background noise. A typical HPLC sample contained 200&3000 cpm (- 80 pg as starting cholesterol). Non-radioactive standards ( 5 a - 0 0 H , 7 - 0 0 H . 5a-OH, 7a-OH and 7P-OH) were obtained as a generous gift from Dr. J. I. Teng (University of Texas, Galveston).

483

.en F 1 '.

n C 0) 44

C

W

I 0

s

RESULTS

Dye-sensitized lipid peroxidation

10

0

Continuous aerobic irradiation of erythrocyte ghosts in the presence of 5 pM MC540 resulted in a linear accumulation of iodometrically determinable LOOH over a 2 h period [Fig. l(A)]. Increasing the MC540 concentration to 25 pM produced an approx. 5-fold increase in the initial rate of LOOH formation, showing that this rate is directly proportional to sensitizer concentration. (No reaction occurred in the absence of sensitizer.) Photoperoxidation was abolished when 0 2 was purged from the system, confirming that the reaction is photodynamic. Figure 1(B) shows a thin layer chromatogram of the TMPD-reactive LOOHs produced in the 25 pM MC540 experiment. Note that all major lipid classes are represented in the LOOH profile: phospholipids (PC, PE, PS, SM) comprising 60 wt% of the membrane lipid, and cholesterol (Ch) comprising 25 wt%. Each peroxide appeared as a purple spot, which intensified with light dose over a 90 min period, in keeping with the quantitative results shown in Fig. I(A). MC540, which extracted with the lipid fraction, migrated with a slightly lower RI than PEOOH. The intrinsic pink color of the dye disappeared during irradiation due to photobleaching (see below). Absorption spectra of MC.540 in erythrocyte ghosts before and after illumination ate shown in Fig. 2. The absorption maxima observed at 568 and 533 nm, previously described for liposomal systems (Lelkes and Miller, 1980; Aramendia et al., 1988), are characteristic of membrane-bound MC540, the former wavelength representing the monomeric form of the dye, and the latter wavelength the dimeric form. As seen in Fig. 2, the absorbance of membrane-bound MC540 decreased steadily during irradiation. This effect (like lipid peroxidation) was shown to be 02-dependent, indicative of an oxidative photobleaching process. Monomer and dimer bands decayed at approximately the same rate ( t L I 2 50 min under the conditions of Fig. 2) with no evidence of spectral shifts. It is curious that 25 pM MC540 had bleached by 80% after 90 min of irradiation (Fig. 2), yet this appeared to have little effect on the rate of LOOH formation up to that point [Fig. l(A)]. (Some attentuation in photoperoxidation was appar-

-

120

10

Time (min)

B -f -ChOOH

-PEW

MC -

-

-

-

-PCOOH -0 I

I

I

I

0'

45'

90'

Std

Figure 1. Merocyanine 540-sensitized photoperoxidation of membrane lipids. (A) Quantitation of lipid hydroperoxide (LOOH) formation. Ghost membranes (1 mg protein/mL) in PBS at 5°C were sensitized with 5 pM ( A ) and 25 pM (0)MC540 and irradiated continuously in air with white light (fluence rate 35 mW/cm*). At the indicated points, samples were removed, extracted, and total LOOH was determined by iodometric analysis. Also represented is a light control without dye (0)and a photoreaction (25 pM dye) carried out under argon ( 0 )or in the presence of 50 p M butylated hydroxytoluene (0). Points with error bars are means 2 SD of values from at least three experiments. (B) Thin layer chromatogram of different LOOH classes. Membranes were photoperoxidized with 25 pM MC540 as described above. Samples were analyzed after 45 and 90 min of irradiation and LOOHs were detected with TMPD. The spot in the 0 min lane is MC540, which gradually became bleached (oxidized) during irradiation. A mixture of Rose Bengal-photooxidized cholesterol (ChOOH), phosphatidylcholine (PCOOH) and phosphatidylethanolamine (PEOOH) was used as a standard (Std). SMOOH and PSOOH denote oxidized sphingomyelin and phosphatidylserine, respectively. Total lipid in each of first three lanes is 0.33 mg.

-

J. BACHOWSKI et nl. GARY

484

0.L -

3

0.3-

C

0

2 0.2 Q

0.1 -

mol% of the lipid concentration), inhibition of lipid peroxidation would not have been observed, except perhaps in the very early stages of the reaction before the antioxidant was totally consumed. As shown in Fig. 3(A), dye-sensitized lipid peroxidation was inhibited by sodium azide in a dosedependent fashion over the 5-40 mM range. A plot of the data expressed in terms of a Stern-Volmer relationship (Eq. 2: r,dr vs [N3-]) has a y-intercept close to 1.0 and is linear out to 40 mM [Fig. 3(A)], suggesting that azide is quenching only lo2,with no significant effect on other species, e.g. sensitizer triplet. From the slope of this plot, one obtains a quenching constant (k,) of 1.6 x lo7 M - l s - ' , which is 2 orders of magnitude lower than reported values of k, in water (Wilkinson and Brumrner, 1981). [This value is listed in Table 1, for comparison with other experimental values, cf. Figs. 3(B)-(D).] Photomodification of another membrane target, the exofacial enzyme ACE, was also inhibited by azide [Fig. 3(B)]. In this case, the k , value from the linear Stern-Volmer plot is 6.7 x lo7 M-Is-L , which is 4-times greater than the value determined with lipid peroxidation, but still 20times lower than published k , values. In an attempt to rationalize these discrepancies, we carried out an experiment in which an extramembranous non-lipid target, LDH, was photoinactivated via the excitation of membrane-bound MC540. The sensitizer was stationed either on ghost membranes [Fig. 3(C)] or on non-oxidizable DMPC/DCP liposomes [Fig. 3(D)]. As shown in Fig. 3(C), photoinactivation of LDH in the ghost system (first order rate 0.75 h-I) was strongly inhibited by azide, the linear Stern-Volmer plot (inset) revealing a k , of 6.3 x 1oX M - l s - ' , which is 40-times greater than the value obtained with lipid peroxidation and reasonably close to published k, values (Wilkinson and Brummer, 1981). We determined that 7.5% (3 pM) of the total MC540 in this system prior to illumination was in the medium, the remainder being membrane-bound. In PBS buffer this material was primarily in the dimeric and higher aggregate form (Lelkes and Miller, 1980), exhibiting two poorly resolved absorbance peaks at 501 and 533 nm. When LDH was irradiated in the presence of 3 p M MC540 in PBS, < 3% inactivation was observed after 1 h (data not shown). Therefore, the relatively large rate of LDH inactivation observed in the complete system [Fig. 3(C)] cannot be attributed to the small amount of free dye in the aqueous compartment. We conclude that membrane-bound MC540 was the principal source of the lo2that caused LDH damage and inactivation. Although '02generated on the erythrocyte membrane would be expected to have a relatively short lifetime due to the presence of oxidizable lipids and proteins (Kanofsky, 1990), it is clear that under the conditions described here, significant amounts of the oxidant were able to escape and react in the

-

0-

I

350 450 550 650 Wavelength (nml

Figure 2. Spectral characteristics of membrane-bound MC540. Visible absorption spectra of a freshly prepared mixture of MC540 (25 p M ) and erythrocyte membranes (1.O mg proteidml; 0.67 mM phospholipid) was recorded (a) before and (b) after irradiating for 30 min; (c) 45 min; (d) 70 min; and (e) 90 min. Refer to Fig. 1 for other details. Experimental samples and a membrane blank were diluted 7.5-fold with PBS for spectral measurements. ent after 120 min.) Similar findings applied to 5 pM dye. A possible explanation is based on the fact that MC540 photoproduct(s) absorbs rather strongly below 370 nm (cf. Fig. 2). Since wavelengths down to 300 nm were transmitted in our reaction system, photosensitization by these products could have at least partially compensated for any slowing of peroxidation due to bleaching. However, whether such chemistry is feasible remains to be established. An alternative (albeit trivial) explanation is that continuous sampling of reaction mixtures for LOOH analysis resulted in progressively less internal filtering by MC540, this factor tending to compensate for bleaching loss.

-

Mechanistic studies: kinetic analyses Neither the rate nor the extent of MC540-sensitized lipid peroxidation was affected by the free radical trap BHT [Fig. l(A)], suggesting that Type 1 (free radical) photochemistry and/or post-irradiation chain reactions are not important in this system. However, BHT was clearly reactive under the conditions described, since its hydroperoxide was detected on TLC (Rf 0.89 in the 1:l heptane:ethyl acetate system). In addition to scavenging radicals, BHT can react with ' 0 2 at fairly high rates (1-5 x lo6 W 1 s - l ) , the major product being the hydroperoxy-dienone (Thomas and Foote, 1978). Based on evidence for Type I1 chemistry (see below), BHT was probably oxidized to this product in the experiment described in Fig. l(A). However, since the BHT concentration was low (approx. 4

-

-

-

-

-

-

Merocyanine 540 photosensitization

485

n

.-W

C

n

:

v

w .b

.w

> *z

Y

LO

5 C v

I 0

3

(7,

.cC

/ O

30

100

.-

/ O

50

0

:

20

t Y

10

0 J

0

"

10

'

"

20

10

"

30

LO

50

€4

70

I

I

I

1

10

20

I

30

I

1

LO

50

1

60

I

70

Time (min )

Time (rnin )

w .-b

W

>

'3 100

Y 0 .c .-C

50

E W

I Y

"t

10 0

10

20

30

LO

50

60

70

D

0

1 0 2 0 ~ 1 ~ 0 s o w r n

Time (min ) Time (min) Figure 3. Effects of azide on MC54O-sensitized reactions. (A) Inhibition of lipid peroxidation. Ghost membranes (- 0.67 mM phospholipid; 0.56 mM cholesterol) were irradiated at 25°C in the presence of 25 pM MC540 alone (0) or MC540 plus sodium azide at the following concentrations: 5 mM (A), 20 m M (O), and 40 mM (V).At the indicated times, samples were analyzed iodometrically for total LOOH content: Inset: Stern-Volmer plot of rJr (rate ? azide) vs azide concentration. (Data from additional experiments using 15 and 30 mM azide are also represented.) (B) Inhibition of ACE inactivation. Separate membrane samples taken from the experiment shown in panel A were analyzed for ACE activity. Initial activity was 1.75 U/mg membrane protein. Quencher concentrations were as follows: 0 mM (0),5 mM (A), 20 mM (O), and 40 inM (V). Inset: Stern-Volmer plot showing the effect of these and three additional azide concentrations, 10, 15, and 30 mM. (C) Inhibition of lactate dehydrogenease inactivation (ghost system). Ghost membrane suspensions containing 40 pM MC540 and 5 U/mL LDH were irradiated at 25°C in the presence of 0 mM (0),0.4 mM (A), 2.5 mM (0). and 5.0 mM ( V ) sodium azide. Inset: Stern-Volmer plot; two additional azide concentrations are represented: 1.O and 3.0 mM. (D) Inhibition of lactate dehydrogenase inactivation (liposome system). Suspensions of small unilamellar DMPClDCP liposomes (lO:l,moymol; 1 mM DMPC) containing 0.2mM (A), 40 pM MC540 and 5 U/mL LDH were irradiated at 27°C in the presence of 0 mM (0), 1.0 mM (O), and 5.0 mM (V)azide. Inset: Stern-Volmer plot; two additional azide concentrations are represented: 0.5 and 2.0 mM. Linear correlation coefficents for the Stern-Volmer plots in (A), (B), (C) and (D) are 0.961, 0.987, 0.963 and 0.995, respectively.

-

medium. In the case of MC540 in DMPC/DCP liposomes [Fig. 3(D)], LDH underwent first order photoinactivation, but the rate constant, 3.2 h-lI was more than 4-times greater than that observed with MC540 in ghosts [Fig. 3(C)]. The rate enhancement could be explained by a longer lo2lifetime in the inert liposome system (see above). Not surprisingly, LDH inactivation in this system was strongly suppressed by azide, k, being 1.2 X lo9 M - l s - ' bY Stern-Volmer analysis.. In this instance, very close agreement with literature values was attained (Table l),clearly confirming that membrane-bound MC540

-

PAP 53:I-E

is generating '02of sufficiently long lifetime to escape its site of origin. The 40- to 70-fold lower quenching efficiency observed with lipid peroxidation [Fig. 3(A)] is ascribed to the fact that peroxidizable regions of the membrane are poorly accessible to azide and, therefore, competition with lipids for '02is unfavorable. The corollary is that localized reactivity and/or quenching limits the extent to which '02can escape from the membrane. The results with ACE (Table 1) suggest greater accessibility of azide to this target than to lipids, which is consistent with the fact that ACE'S active site is

GARY J. BACHOWSKI et al.

486

Table 1. Azide quenching of various MC54Osensitized reactions Membrane system

Target

(1) Ghosts (2) Ghosts

k,

Unsat. lipids ACE LDH LDH

(3) Ghosts (4) DMPClDCP liposomes

(W's-I)*

1.6 6.7 6.3 1.2

x i07 x 107 x I@ x

109

*Rate constants for azide quenching of '02were determined by Stern-Volmer analysis of competitive kinetic data [Figs. 3(A)-(D)], using the relationship provided in the Methods scction (Eq. 2). Range of published values for azide in H20: 0.7-2.2 X 109 M - ' s - l (Wilkinson and Brummer, 1981).

exposed to the polar medium on the external membrane face. Mechanistic studies: cholesterol product analyses

Involvement of lo2 in lipid peroxidation was established unequivocally by showing that 5 a - 0 0 H is formed, 5a-00H being a well-defined product of '02attack on cholesterol (Schenck et al., 1957; Kulig and Smith, 1973; Suwa et al., 1977). TLC

F-

analysis of extracted membrane lipids after 30 and 90 min of photooxidation revealed a stepwise accumulation of TMPD-detectable cholesterol hydroperoxide(s) (ChOOH) (Fig. 4(A), lanes 2 and 3). (Phospholipids and MC540 remained at the origin in this TLC system.) The ChOOH was mainly 5aOOH, since borohydride treatment gave the corresponding diol (5a-OH), which was well-separated from 5 a - 0 0 H (Fig. 4(B), lane 4). Assignment of 5 a - 0 0 H was based on co-migration with the Rose Bengal-generated product (lane 5), Rose Bengal being a well established photogenerator of (Foote, 1979). Relatively little, if any, of 7a-OH or 7P-OH (free radical-derived products) could be detected in this experiment (Fig. 4(B), cf. lanes 4 and 6). These findings corroborate earlier preliminary observations (Kalyanaraman et a f . , 1987). Quantitation of ChOOH formation was accomplished by using ghosts in which cholesterol was radiolabeled with carbon-14. The TLChadioscan in Fig. 5 shows that 3% of the membrane cholesterol was converted to ChOOH after 90 min of MC54O-sensitized photooxidation (lane c). Most of this material was Sa-OOH, since Sa-OH was the major product observed after borohydride treat-

-

B

A

-F

-Ch

ChOOH-

1@:8# -7a-OH

0-

-0 I

l

l

1

0' 30' 96 SO' Stbl StA2

-BH4-

0'

1

1

1

I

3 0 90 90' Stdl Std2

-BH4-

Figure 4. Thin layer chromatogram of cholesterol oxidation products. Ghost membranes (1 mg protein/mL or 0.55 mM cholesterol in bulk suspension) were sensitized with 25 p M MC540 and irradiated at 5°C (Ruence rate 20 mW/cm2). At the indicated time points (lanes 1-4). lipids were extracted and chromatographed, using heptane:ethyl acetate (1:l) as the solvent system. Reference standards are as follows: Std 1, membranes photooxidized in the presence of Rose Bengal; Std 2, reduced 7-ketocholesterol. Where indicated, samples were treated with sodium borohydride (BH,) before chromatographing. The plate was sprayed with TMPD (panel A) to visualize hydroperoxides, followed by H,SO, (panel B) to visualize parent cholesterol and its diol products. Sample load (as starting cholesterol): 80 pg (lanes 1-4).

-

-

Merocyanine 540 photosensitization A

a. 0 min b. 4 5 rain C.

90

miin

U. 90 min (red1

Ch

fl

ChOOR 5K-OR

1023(ch) Itf

I 1.0

Time (min)

1

I 0.5

0.0

Figure 5. Radioscan of TLC-separated products from photoperoxidized [~JC]cholesterol-labeledghosts. Membranes were irradiated at 5°C in the presence of 50 pM MC54O (Ruence rate 35 mW/cm2). At 45 min (lane b) and 90 min (lane c), lipids were extracted, chromatographed and detectedlquantitated by radioscanning. One of the 90 min samples was reduced with borohydride before analyzing (lane d). Starting material is shown in lane a. Load per lane: 0.32 mg total lipid (80 pg cholesterol, 2500 cpm). Product yields are as follows: lane c: ChOOH (3.0%); lane d: 5a-OH (2.3%); 7-OH (1.4%).

-

-

ment (lane d). The small amount of 7-OH in lane d is attributed to allylic rearrangement of 5 a - 0 0 H to 7 a - 0 0 H during photooxidation or sample workup rather than authentic Type I (radical) photochemistry, since it consisted mainly of 7a-OH rather than comparable amounts of 7a-OH and 7P-OH (Smith et al., 1973). The material at the origin in lanes b-d has not been identified. In addition to TLC, we used normal phase HPLC to identify and quantitate cholesterol photoproducts. HPLC has the advantage over TLC in affording a better separation of ChOOHs, e.g. 5aOOH from 7a-/7p-O0H as well as 7 a - 0 0 H from 7 P - 0 0 H (Smith et al., 1987). Photooxidation of [14C]cholesteroI-labeIedghosts under the conditions described in Fig. 6(A) produced 1% 5 a - 0 0 H (retention time 16-17 rnin), 0.2% of an unknown (12 min), and no significant 7 - 0 0 H (23-24 min) (panel b). BHT was present during the photoreaction to inhibit any rearrangement of 5 a - 0 0 H to 7 a - 0 0 H (Bachowski er al., 1988). Borohydride treatment (panel c) caused the 12 min peak to disappear, but had no apparent effect on the 16.5 min peak, suggesting that 5a-OH was not resolved from 5 a - 0 0 H in this system. [TLC with TMPD detection confirmed that reduction of ChOOH was complete; cf. Fig. 4(A).] Identification of 5 a - 0 0 H in Fig. 6(A) (panel b) was based on (i) recovery of the material, reduction, and identification of 5a-OH on TLC [cf. Fig. 5(B)]; (ii) comparison with authentic purified 5 a - 0 0 H (sample obtained from J. I. Teng); and (iii) comparison with material produced by Rose Bengal sensitization [Fig. 6(B)]. In the latter system, a small amount of 7 - 0 0 H was also evident (23.5 rnin), reduction of which gave 7-OH (35.5 min). Approximately 12-times more total

-

-

Figure 6. HPLC-radiodetection of cholesterol photooxidation products. (A) MC540 sensitization. A suspension of [ I4C] cholesterol-labeled ghosts containing 25 pM MC540 and 50 pM BHT was irradiated at 5°C under a stream of 0, (Ruence rate 35 mW/cm2). Samples were analyzed (a) before and (b,c) after photooxidation for 3 h; sample c was reduced with borohydride. Cholesterol and its oxidation products (- 2800 cpm per sample) were separated by normal phase HPLC, using hexane/isopropanol (24:l) as the solvent system and effluent Row radiodetection. Solute distribution: (b) cholesterol (6 min, 98%); 5alYo). (c) Cholesterol (6 min, 98%); OOH (16.5 min, 5a-OH (16 min, -1y0). (B) Rose Bengal sensitization. The same preparation of radiolabeled ghosts was irradiated for 90 min in the presence of 25 pM Rose Bengal and 50 pM BHT (fluence rate 80 mW/cm2). Extracted cholesterol products, (a) non-reduced and (b) borohydride-reduced, were chromatographed as described above. Distribution: (a) cholesterol (83%); 5 a - 0 0 H (1 lolo); 7OOH (3%). (b) Cholesterol (87%); 5a-OH (10%); 7-OH (2%).

-

-

-

ChOOH was generated in the Rose Bengal system than in the MC540 system normalized to the same light fluence (Fig. 6). Synergistic action of ironlaseorbate on lipid peroxidation

In previous studies with porphyrin sensitizers, we learned that AH- and catalytic iron could act synergistically with photodynamic action in driving erythrocyte membrane lipid peroxidation (Girotti et al., 1985; Bachowski et a[., 1988a). The reaction in the mixed system far exceeded the one induced by sensitizerllight alone or AH-/iron alone. It was of interest to determine whether similar effects could be demonstrated for MC540 sensitization. When MC54O-sensitized ghosts were irradiated in the pres-

GARY J. BACHOWSKI et al.

488

0

20

10

60

80

1w

120

Time (min )

B

ence of added Fe3+ and AH-, a large burst of TBAdetectable lipid peroxidation (TBARS formation) was observed, which was much greater than that observed in the absence of AH-/Fe'+ [Fig. 7(A)]. No TBARS could be detected in a light control (containing AH-/Fe3+, but lacking MC540) or a dark control (containing AH-/Fe'+, but lacking light). Ascorbate-stimulated peroxidation was abolished by BHT and also by DFO, a redox inhibiting iron chelator (Halliwell and Gutteridge, 1985). confirming that the reaction is iron-catalyzed and free radical-mediated. Free radical involvement was also established by examining cholesterol oxidation products. As can be seen in Figs. 7(B) and (C), 5 a - 0 0 H generated as a primary photooxidation product (lanes b and c) disappeared in the AH-/Fe"+-containing system and was replaced by 7 u - 0 0 H and 7P-O0H, the latter being predominant. Note that AH- caused a decrease in the 90 min ChOOH level while elevating the 7a-/7P-F OH level [cf. lanes d and b in Figs. 7(B) and (C)]. This effect was totally reversed by DFO, i.e. 7a/7p-OH disappeared and ChOOH increased to a level similar to that seen in the absence of AH- (cf. lanes h and b). Meanwhile, BHT had the opposite effect on ChOOH, causing its virtual disappearance, along with 7a-17P-OH (lane f). It is interesting that -ChOOH DFO and BHT had diametrically opposite effects on ChOOH (and presumably on other lipid peroxides) while inhibiting TBA activity to the same extent (Fig. 7). -0 When irradiated in the presence of AH-, certain 1 sensitizers (e.g. porphyrins) can undergo one-elecj tron photoreduction to their radical anions, which, upon autoxidation, yield superoxide radical, 0:(Felix et al., 1983; Bachowski el al., 1988a). Super-F oxide can either dismutate or be reduced to hydrogen peroxide, H202. Interaction of 02-and H z 0 2 via the iron-catalyzed Haber-Weiss pathway gives rise to hydroxyl radical (OH.), a potent initiator of free radical lipid peroxidation (Girotti, 1990). To study the importance of this mechanism in the -Ch MC540/AH-/ light system, we used two antioxidant enzymes, superoxide dismutase and catalase. As -ChOOH shown in Fig. 8(A), neither of these enzymes (in catalytic doses) had an inhibitory effect on AH-stimulated lipid peroxidation. By contrast, xanthine/xanthine oxidaseliron-induced lipid peroxi-

1

1

1

1

1

1

1

1

1

a

b

c

d

e

f

g

h

i

I

I

I

I

I

I

I

I

I

I

a

b

c

d

e

f

g

h

i

j

C

-0

Figure 7. Stimulation of MC54O-initiated lipid peroxidation by iron and ascorbate. (A) TBARS formation. A membrane suspension containing 25 pM MC540 and 75 p M FeCI, (basal mixture) was irradiated as such (0)or in the presence of the following: 1 mM AH- (A); 1 mM AH-10.05 mM BHT (0);1 mM AH-/0.15 mM DFO (0).A light control containing AH- and FeCI,, but no sensitizer (x), and a dark control containing sensitizer, AH-, and FeCI, (0)were analyzed alongside. Points with error bars are means t deviation of values from duplicate experiments. (BIC) TLC of cholesterol oxidation products (same experiment). Peroxide spots were visualized with TMPD (B). Samples were as follows: (a) basal mix, dark control; (b and c) 90 min irradiation; (d and e) irradiation in the presence of A H - ; (f and g) AH-/BHT; (h and i) AH-IDFO. Samples c, e, g and i were reduced with borohydride before chromatographing; j is reduced 7-ketocholesterol (7a-/7P-OH standard). Subsequent to peroxide detection, the plate was sprayed with 50% H,SO, (C) to visualize parent cholesterol and various diol products.

Merocyanine 540 photosensitization A

489

B

P /”

d’ V A/

J

1

0

20

1

LO

1

1

60

O

80

f

i

100 120

0

20

Time (min)

LO

60

80

100

Time (min)

Figure 8. Effects of catalase and superoxide dismutase on erythrocyte membrane lipid peroxidation. (A) MC540-sensitized/AH--sensitized lipid peroxidation. A membrane suspension containing 25 pM MC540 and 75 pM FeCI, was irradiated as such (0)or in the presence of the following: 1 mM AH(0); 1 mM AH- plus 50 pg (180 U)/mL superoxide dismutase (V); 1 mM AH- plus 50 pg (620 U)/mL catalase ( A ) . (B) Xanthine/xanthine oxidase/iron-induced lipid peroxidation. A membrane suspension (1 mg protein/mL) containing 2.5mM xanthine, 0.025 U/mL xanthine oxidase, and 0.1 mM Fe(NH,)2(S0,)2in PBS was incubated in air at 37°C (0). Additional constituents were as follows: 620 U/mL catalase ( 0 ) ;180 U/mL superoxide dismutase (A). Mixtures without added catalase contained 1 mM azide to inhibit any endogenous catalase in the membranes. Samples were removed for TBARS determinations at the indicated times.

be “broadcasted” to cytosolic or extracellular targets. Previous studies involving other sensitizers in model systems (micelles or lipid vesicles) have supported this possibility (Gorman et al., 1976; Rodgers and Bates, 1982). That LDH was photoinactivated at a much greater rate with MC540 in liposomes [Fig. 3(D)] relative to ghosts [Fig. 3(C)] is consistent with the demonstrated shorter lifetime of lo2 in the latter system (Kanofsky, 1990). When endogenous membrane targets were examined, azide quenched the reactions much less effectively DISCUSSION than it did LDH inactivation, the rate constant being In the present work we provide evidence that cell at least 40-fold lower in the case of lipid peroximembrane-bound MC540 not only sensitizes ‘ 0 2 dation and 10-fold lower in the case of ACE inactiformation, but that a target lipid, cholesterol, is vation. This discrepancy is not inconsistent with ‘02 oxidized to 5a-O0H, an unequivocal ‘02-derived intermediacy, but implies that lo2reacts predomiproduct. These results provide solid quantitative nantly at or near its membrane sites of origin, which confirmation of earlier preliminary findings are poorly accessible to azide. Accordingly, rela(Kalyanaraman et al., 1987; Valenzeno et al., 1987). tively little of the lipid peroxidation or ACE inactiUsing azide as a diagnostic quencher and LDH as vation can be attributed to ‘02that migrates from an extramembranous target, we verified that ‘ 0 2 one membrane to another. It is important to note causes enzyme inactivation by showing that the by way of contrast that when photoperoxidation was quenching constant (0.6-1.2 x loy M - l s - ’ ) lies carried out with uroporphyrin I as sensitizer, azide within a published range (Wilkinson and Brummer, quenched the reaction much more efficiently (data 1981). This was demonstrated for lo2generated on not shown), k, 5 X 10“ M-ls-I being nearly the oxidizable erythrocyte ghosts or on inert same as measured in the MC540lghostlLDH system. DMPC/DCP liposomes. In the ghost system, some Unlike MC540, UP does not bind to erythrocyte of the ’0, was evidently long-lived enough to escape membranes (Bachowski et al., 1988a); therefore, the membrane and react with LDH in the medium. ‘ 0 2 generated by UP in the aqueous medium shold This suggests that in MC54O-sensitized cells, ‘02 be intercepted by azide in kinetically predictable damage may not necessarily be restricted to the fashion (cf. Wilkinson and Brummer, 1981), as plasma membrane, but may (at least to some extent) observed. These results are consistent with the idea dation (serving as a positive control) was completely inhibited by catalase or superoxide dismutase [Fig. 8(B)], confirming that this reaction is mediated by 0,- and H z 0 2 (Thomas and Girotti, 1984). We infer from these results that AH--stimulated peroxidation in the photodynamic system is not mediated by OH., but rather by lipid-derived radicals, e.g. lipoxyl (LO-) arising via the light-independent reduction of LOOHs (Girotti, 1990).

-

GARY J. BACHOWSKI er al.

490

(Valenzeno, 1987) that the lifetime of I 0 2 (which determines its diffusion distance and probability of interception by polar quenchers) is considerably lower in oxidizable membranes than in aqueous solution. Action spectra for 0, consumption and quantum yields for lo2production by MC540 in liposomes and erythrocyte ghosts are reported in the accompanying paper (Singh et al.,1991). A particularly noteworthy aspect of the action spectra is their strong overlap with the absorption spectrum of dye monomer (A, 568 nm). The '02quantum yield for dye excitation at 568 nm was found to be 0.065, which is 20-times greater than the value obtained using ethanol or methanol as the solvent (Singh et al., 1991; Hoebeke et al., 1988). Rose Bengal, a dye with similar light absorbing properties, has a lo2quantum yield of 0.75 (Gandin et al., 1983). Compared with Rose Bengal, therefore, MC540 is a rather modest '0, generator, even in membranes. In agreement with this, we found that MC540 produced significantly less 5 a - 0 0 H than Rose Bengal under similar reaction conditions (Fig. 6). The relative inefficiency of MC540 as a lo2generator in simple solvents has suggested to Davila et al. (1989) that the dye's phototoxic effects are not mediated by '0,. However, other workers (Kalyanaraman et al., 1987; Gaffney et al., 1990) have shown that MC540-sensitized photoinactivation of tumor cells is not only 0,-dependent, but stimulated by D 2 0 , providing strong evidence for '0, intermediacy. The present findings add further support to this argument. We conclude that while the efficiency of '02 production is an important factor in sensitized photokilling, formation of lo2 in the vicinity of crucial membrane targets is an additional important consideration. Thus, localization of MC540 may tend to compensate for the relatively low '0, yield. When MC540-sensitized ghosts were irradiated in the presence of iron and AH-, lipid peroxidation switched from a 'Ormediated process to a predominantly free-radical-mediated one, as indicated by the large burst of TBA reactivity, large scale formation of 7-oxycholesterols (7a-/7P-OOH, 7u-/7POH), and disappearance of the primary '0, adduct, 5 a - 0 0 H . Lack of inhibition by catalase or superoxide dismutase suggests that OH. arising from Fenton chemistry (reduction of H202 by Fez+) is not an important initiator of AH--stimulated lipid peroxidation. This would rule out any Type I chemistry involving photoreduction of membrane-bound MC540 to the radical anion, with subsequent autoxidation to produce 0 2 - , H202, and thence OH. (Girotti, 1990). We feel that a more likely pathway is one in which nascent Type I1 photoperoxides undergo iron-catalyzed reduction to oxyl radicals [Eq. (3)], which then (either directly or indirectly; see below) trigger chain reactions [Eqs. (4-6)):

-

-

-

+ AH- -+ LO' + H20 + A' LO' + LH -+ LOH + L' L' + 0 2 -+ LOO' LOO' + LH L' + LOOH

LOOH

(3) (4) (5)

(6)

where LH denotes an unsaturated lipid, LO' a lipidoxyl radical, LOO' a lipid peroxyl radical, and A; ascorbate radical. The damaging effects of these reactions (which are light-independent) far exceed those o! photoperoxidation alone. A similar mechanism has been invoked for reactions carried out with other sensitizers, e.g. protoporphyrin, uroporphyrin, or hematoprophyrin derivative (Bachowski et al., 1988a,b). In each case, including the MC54O/ghost system, we found that peroxidation was stimulated by AH- in dose-dependent fashion up to 2 mM, but that the reaction subsided at higher concentrations, evidently because AH-'s antioxidant effects were overriding its prooxidant effects (Girotti, 1990). It is intriguing that little (if any) 5a-OH could be detected when photoperoxidation was carried out in the presence of AH- and iron [Fig. 7(C)]. A similar observation was made on pre-irradiated ghosts (containing 5 a - 0 0 H and phospholipid hydroperoxides) after incubation with AH-/iron in the dark (Bachowski et al., 1988b). If 5 a - 0 0 H were to undergo one-electron reduction according to Eq. (3), one would expect 5a-OH to accumulate (along with the observed 7-0x0 products) as 5a-oxyl radical abstracts hydrogen from an adjacent lipid. One possible explanation is that the 5a-oxyl radical rapidly rearranges to another species prior to abstracting hydrogen, e.g. an epoxyallylic radical, which (via a peroxyl intermediate) eventually ends up as an epoxy-diol (cholest-5a,6a-epoxy-3P,7-diol) or the keto analog (cholestJa,6a-epoxy-3P-ol,7one). Such rearrangements have been described for fatty acid-derived oxyl radicals (Gardner, 1989), but there is no published evidence as to whether the cholesterol 5a counterpart might behave similarly. We are currently attempting to identify the epoxydiol and/or the ketone in reaction systems such as described in Fig. 7. It remains to be seen whether membrane damage due to AH- stimulation of photoinitiated lipid peroxidation can also be realized in neoplastic cells or enveloped viruses. If so, this might constitute a novel approach for enhancing the therapeutic index of ex-vivo MC540llight treatment. Ascorbate-driven free radical reactions might amplify the effects of a relatively modest lo2quantum yield and translate subliminal or repairable damage in some tumor cells into full scale lethal damage.

-

Acknowledgements-We are grateful to H. W . Gardner and J. I. Teng for helpful discussions. This work was supported by USPHS Grant CA 49089 from the National Cancer Institute.

Merocyanine 540 photosensitization REFERENCES

Ansari, G. A. S. and L. L. Smith (1979) High performance liquid chromatography of cholesterol autoxidation products. J . Chromarogr. 175, 307-315. Aramendia, P. F., M. Krieg, C. Nitsch, E. Bittersmann and S . E. Braslavsky (198) The photophysics of merocyanine 540. A comparative study in ethanol and in liposomes. Phorochem. Photobiol. 48, 187-194. Bachowski, G. J., K. M. Morehouse and A. W. Girotti (1988a) Porphyrin-sensitized photoreactions in the presence of ascorbate: oxidation of cell membrane lipids and hydroxyl radical traps. Phorochem. Photobiol. 47, 635-645. Bachowski, G. J . , J . P. Thomas and A. W. Girotti (1988b) Ascorbate-enhanced lipid peroxidation in photooxidized cell membrane: cholesterol product analysis as a probe of reaction mechanism. Lipids 23, 580-586. Davila, J., K. S. Gulliya and A. Harriman (1989) Inactivation of tumors and viruses via efficient photoisomerization. J . Chem. SOC.Chem. Commun. 1215-1216. Fairbanks, G., T. L. Steck and D. F. H. Wallach (1971) Electrophoretic analysis of the major polypeptides of the human erythrocyte membrane. Biochemistry 10, 2 6 2 6 16. Felix, C. C., K. Reszka and R. C. Sealy (1983) Free radicals from photoreduction of hematoporphyrin in aqueous solution. Phorochem. Phorobiol. 37, 141-147. Foote, C. S . (1979) Quenching of singlet oxygen. In Singler Oxygen (Edited by H.H. Wasserman and R. W. Murray) pp. 139-171. Academic Press, New York. Gaffney, D. K., S . L. Schober and F. Sieber (1990) Merocyanine 540-sensitized photoinactivation of leukemia cells: role of oxygen and effects on plasma membrane integrity and mitochondria1 respiration. Exp. Hemarol. 18, 23-26. Gandin, E., Y. Lion and A . Van de Vorst (1983) Quantum yield of singlet oxygen production by xanthene derivatives. Phorochem. Phorobiol. 37, 271-278. Gardner, H. W. (1989) Oxygen radical chemistry of polyunsaturated fatty acids. Free Rad. Biol. Med. 7, 65-86. Girotti, A. W. (1976) Bilirubin-sensitized photomodification of enzymes in the isolated membrane of the human erythrocyte. Photochem. Photobiol. 24,525-532. Girotti, A. W. (1979) Porphyrin-sensitized photodamage in isolated membranes of human erythrocytes. Biochemisrry 18, 4403-4411. Girotti, A. W. (1990) Photodynamic lipid peroxidation in biological systems. Phorochem. Phorobiol. 51, 497-509. Girotti, A. W., J. P. Thomas and J. E. Jordan (1985) Prooxidant and antioxidant effects of ascorbate on photosensitized peroxidation of lipids in erythrocyte membranes. Photochem. Phorobiol. 41, 267-276. Girotti, A. W., J. P. Thomas and J. E. Jordan (1986) Xanthine oxidase-catalyzed cross-linking of cell membrane proteins. Arch. Biochem. Biophys. 251,639-653. Gorman, A. A,, G. Lovering and M. A. J. Rodgers (1976) The photosensitized formation and reaction of singlet oxygen in aqueous micellar systems. Phorochem. Photobiol. 23, 399-403. Halliwell, B. and J. M. C. Gutteridge (1985) Free Radicals in Biology and Medicine, Chap. 4. Clarendon Press, Oxford. Hoebeke, A,, A. Seret, J . Piette and A. Van de Vorst (1988) Singlet oxygen production and photoisomerization: Two competitive processes for merocyanine 540 irradiated with visible light. J . Phorochem. Photobiol. 1, 437-446.

49 1

Kalyanaraman, B., J. B. Feix, F. Sieber, J. P. Thomas and A. W. Girotti (1987) Photodynamic action of merocyanine 540 on artificial and natural cell membranes: involvement of singlet molecular oxygen. Proc. Narl. Acad. Sci. U S A . 84, 2999-3003. Kanofsky, J. R. (1990) Quenching of singlet oxygen by human red cell ghosts. Phorochem. Phorobiol. 51, 42s. Kulig, M. and L. L. Smith (1973) Sterol metabolism. XXV. Cholesterol oxidation by singlet molecular oxygen. J. Org. Chem. 38, 3639-3642. Lelkes, P. I. and I. R. Miller (1980) Perturbations of membrane structure by optical probes: 1. Location and structural sensitivity of merocyanine 540 bound to phospholipid membranes. J . Memb. Biol. 52, 1-15. Lowry, 0. H., N. L. Rosebrough, A. L. Farr and R. J Randall (1951) Protein measurement with the Folin phenol reagent. J . Biol. Chem. 193, 265-275. Rodgers, M. A. J. and A. L. Bates (1982) A laser flash kinetic spectrophotometric examination of the dynamics of singlet oxygen in unilamellar vesicles. Phorochem. Photobiol. 35, 47-77, Schenck, G. 0..K. Gollnick and 0. A. Neumueler (1957) eren autoxydation der steroide. Darstellung von steroid-hydroperoxyden mittels phototoxischer photosensibilisatoren. Ann. Chem. 603,4659. Sieber, F. (1987) Merocyanine 540. Phorochern. Phorobiol. 46, 1035-1042. Sieber, F., J . M. O’Brien, G. J. Krueger, S. L. Schober, W. H. Burns, S . J. Sharkis and L. Sensenbrenner (1987) Antiviral activity of mercyanine 540. Phorochem. Photobiol. 46, 707-7 11. Singh, R. J., J. B. Feix, T. J. Pintar, A. W. Girotti and B. Kalyanaraman (1991) Photodynamic action of merocyanine 540 in artificial bilayers and natural membranes: action spectra and quantum yields. Phorochem. Phorobiol. 53, 493-500. Smith, L. L. and F. L. Hill (1972) Determination of sterol hydroperoxides on thin layer chromatoplates by means of the Wurster dyes. J. Chromarogr. 27, 187-205. Smith, L. L., N. M. Made Gowda and J. I. Teng (1987) Sterol hydroperoxide metabolism by Salmonella Typhimurium. J . Sreroid Biochem. 26. 25S264. Smith, L. L., J. I. Teng, M. J. Kulig and F. L. Hill (1973) Steroid Metabol. XXIII. Cholesterol oxidation by radical-induced processes. J. Org. Chem. 38. 1763-1765. Suwa, K., T. Kinura and A. P. Schaap (1977) Reactivity of singlet molecular oxygen with cholesterol in a phospholipid membrane matrix. A model for oxidative damage in membranes. Biochem. Biophys. Res. Commun. 75, 785-792. Thomas, M. J. and C. S.Foote (1978) Chemistry of singlet oxygen. XXVI. Photooxygenation of phenols. Phorochem. Photobiol. 27, 683-693. Thomas, J . P., M. Maiorino, F. Ursini and A. W. Girotti (1990) Protective action of phospholipid hydroperoxide glutathione peroxidase against membrane-damaging lipid peroxidation: in situ reduction of phospholipid and cholesterol hydroperoxides. J. Biol. Chem. 265, 454-461. Valenzeno, D. P. (1987) Photomodification of biological membranes with emphasis on singlet oxygen mechanisms. Phorochem. Phorobiol. 46, 147-160. Valenzeno, D. P., J. Trudgen, A. Hutzenbuhler and M. Milne (1987) Singlet oxygen involvement in photohemolysis sensitized by merocyanine-540 and rose bengal. Phorochem. Phorobiol. 46, 985-990. Wilkinson, F. and J. G. Brummer (1981) Rate constants for the decay and reactions of the electronically excited singlet state of molecular oxygen. J. Phys. Chem. Ref. Data 10, 809-999.