Production of active oxygen species (1O2 and O2⨪) by psoralens and ultraviolet radiation (320–400 nm)

Biochimica et Biophysica Acta, 798 (1984) 115-126 Elsevier

115

BBA 21692

PRODUCTION OF ACTIVE OXYGEN SPECIES ( t O 2 AND O~) BY PSORALENS AND ULTRAVIOLET RADIATION (320-400 nm) MADHU A. PATHAK and PKAKASH C. JOSHI Department of Dermatology, Harvard Medical School, Warren 5, Room 562, Massachusetts General Hospital, Boston, MA 02114 (U.S.A.) (Received May 19th, 1983) (Revised manuscript received October 10th, 1983)

Key words: Active oxygen," Psoralen; Ultraviolet radiation; Skin photosensitization

Furocoumarins (psoralens) are potent skin photosensitizing agents that are used in combination with long-wavelength ultraviolet radiation (320-400 nm) in the treatment of psoriasis and other skin diseases. Twelve linear and angular psoralens, capable of forming monofunctionai and bifunctional adducts with DNA, were examined with a view to elucidate the role of I O 2 and O~ in evoking skin photosensitization reactions and skin carcinogenesis. The results showed that both linear psoralens (capable of forming interstrand cross-links) and isopsoralens (angular, monofunctional type) and 3-carbethoxypsoralen (a linear and monofunctional type) produced tO 2 and O~, although at varying degrees. Psoralen and 3-carbethoxypsoralen produced tO 2 greater than isopsoralens (angelicins). However, nonphotosensitizing angelicin, 5-methylangelicin, and 4,8-dimethyl-5'-carboxypsoralen produced 102 greater than 8-methoxypsoralen and 5-methoxypsoralen. The three monofunctional angelicin derivatives (isopsoralens) produced more O~ than 8-methoxypsoralen, 5-methoxypsoralen, and 3,4'-dimethyi-8-methoxypsoralen. 3-Carbethoxypsoralen, a potent generator of t O2 and a moderate producer of O~, was highly photolabile. Until recently, skin photosensitization reactions (erythema, edema, damage to DNA or the membrane of cutaneous cells, the inhibition of scheduled DNA synthesis and skin carcinogenesis, etc.) were believed to involve photocyclo-addition of psoralens to DNA mediated by a type-I or anoxic reaction (a sensitizer-substrate interaction through the transfer of hydrogen atoms or electrons, but no direct involvement of molecular oxygen). Oxygen-dependent sensitized photodynamic reactions of type-ll, involving the production of reactive oxygen ( t O 2 and O~), were believed not to mediate psoralen photosensitization reactions. We suggest that I O 2 and O~ may also participate in skin photosensitization and cell membrane-damaging reactions. The fact that certain monofunctional isopsoralens produce 102 and O~ at rates comparable to or better than bifunctional psoralens suggests that these reactive moieties of oxygen could play a major role in explaining their recently observed carcinogenic property and cell membrane-damaging reactions (e.g., edema or inflammation, etc.).

Introduction Certain photoactive fucocoumarins such as 8methoxypsoralen and 4,5',8-trimethylpsoralen have important therapeutic uses. In combination with exposure to ultraviolet radiation (320-400 nm), these furocoumarins are effective in the treatment of disabling generalized psoriasis and other skin 0304-4165/84/$03.00 © 1984 Elsevier Science Publishers B.V.

diseases such as mycosis fungoides, vitiligo, etc. [1]. It has been suggested that the skin photosensitizing and photochemotherapeutic effectiveness of psoralen is related to the ability of the photoexcited psoralen (usually triplet state) to covalently bind to DNA producing photocyclo-addition reactions [2-6]. These reactions appear to involve a sensitizer-substrate interaction not mediated by

116 molecular oxygen in which the formation of stable covalent photoconjugation of furocoumarin-DNA photoadducts required: (a) ground state complexation of psoralens in the dark with DNA bases, (b) the formation of monoadducts in the first photochemical step involving the sensitizer and the pyrimidine bases, and (c) subsequent conversion of furan side monoadducts to interstrand crosslinks between the two pyrimidine bases belonging to opposite strands of DNA in the second photochemical reaction [2-5]. These reactions may involve initially a charge transfer process as suggested by Bensasson et al. [7] and Ronfard-Haret et al. [8] when oxygen is present in ultraviolet irradiated solutions. Studies involving the comparisons of the relative photosensitizing effects of different furocoumarin (both linear psoralen and angular isopsoralen) derivatives, based on erythema responses induced by exposing albino guinea pig skin or human skin to 320-400 nm radiation after topical application or oral adminis-. tration, indicated a causal relationship between skin erythema and the formation of DNA crosslinks. The non-DNA cross-linking psoralens were found to be nonphotosensitizing (i.e., induced no skin erythema or inflammation). Evidence has also been accumulating that photocyclo-addition of psoralens with pyrimidine bases in DNA leading to the formation of diadducts (i.e., interstrand cross-linkages) are more important than monoadducts (single strand cyclobutane adducts with pyrimidines) in producing these biological and therapeutic effects [5,6,9]. 3-Carbethoxypsoralen, a monoadduct forming psoralen in its photochemical reaction with DNA, was found to be nonerythemogenic and non-carcinogenic [10]. Studies on photosensitization of microorganisms by different furocoumarins also implicated monoadducts and cross-links in lethality, induction of mutation, inactivation, or inhibition of DNA synthesis [9,10]. Our studies, using both monofunctional and bifunctional psoralens, suggest that psoralens also undergo type-II reactions involving the formation of reactive singlet oxygen (102) and superoxide anion radicals (02- and HO~) through a photosensitization mechanism in which the photoexcited psoralens react with molecular oxygen to form reactive 102 and 0 2 and cause cellular damage by reaction pathways other than the formation of

mono- and bifunctional adducts. The significance of the formation of these reactive forms of oxygen is discussed in relation to the skin photosensitization and skin carcinogenesis. Materials and Methods

The chemicals used for the study were obtained as follows: Analytically pure grade 8-methoxypsoralen and 4,5',8-trimethylpsoralen from Elder Pharmaceuticals, Bryan, OH; calf thymus DNA, anthracene, hematoporphyrin derivatives, sodium azide, and methylene blue from Sigma Chemical Company, St. Louis, MO. 1,4-Diazabicyclo[2,2,2]octane was purchased from the Aldrich Chemical Company, Milwaukee, WI. Hematoporphyrin used is a mixture of porphyrins and appears to have a high content of hydroxyvinyldeuteroporphyrin. 4,5'-Dimethylangelicin and 5methylangelicin were generously provided by Dr. G. Rodighiero and Dr. F. Dall'Acqua, Padova, Italy. 3,4'-Dimethyl,8-methoxypsoralen, and 5'-diethylaminobutoxypsoralen were synthesized and kindly donated by Dr. S. Marciani, Padova, Italy. 5-Methoxypsoralen, 4,8-dimethyl,5'-carboxypsoralen, angelicin, and psoralen were obtained as synthetic compounds and purified in our laboratory. 3-Carbethoxypsoralen was kindly provided by Dr. J.K. Dunnick at the Department of Health and Human Services, Research Triangle Park, NC, and Dr. D. Averbeck at Institut Curie, Section de Biologie, France. Prior to use, the purity of all the psoralen derivatives was checked by thin-layer chromatography using cyclohexane/acetone (45 : 55, v/v) and chloroform/acetonitrile (3 : 1,

v/v). The ultraviolet irradiation system used in this study comprised a horizontal planar array of six 4-foot long ultraviolet radiation-emitting (320-400 nm) fluorescent tubes manufactured by GTE Sylvania Lighting Products Group, Peabody, MA. A 0.05 inch-thick mylar sheet was used to filter out wavelengths shorter than 320 nm. The irradiance of the emitted light was measured by an International Light IL-700 spectroradiometer (Newburyport, MA) equipped with a calibrated and cosinecorrected ultraviolet radiation-detecting probe. The system produced an ultraviolet radiation of 3.75 m W / c m 2 at 8-10 cm distance. Ultraviolet radia-

117 tion absorption spectra were recorded on a Hitachi l l 0 A spectrophotometer. Skin photosensiting activity of various linear and angular psoralens and of other chemicals were assessed in guinea-pig skin and human skin by the method previously published [11]. The interstrand cross-linking of psoralens was examined as follows: A stock solution of DNA, containing 10 mg/100 ml of 0.05 M phosphate buffer, pH 7, was prepared without stirring to prevent denaturation. Stock solutions of psoralens and other test compounds were prepared to contain 1 m g / m l of the chemical in ethanol. To a 4.9 ml DNA solution, 0.1 ml of the test compound was added and allowed to equilibrate at room temperature for 30 min. The mixture was irradiated in an open petri dish (60 × 15 mm) with varying doses of ultraviolet radiation ranging from 0.5 J / c m 2 up to 16 J / c m 2. After irradiation, the DNA solution was subjected to heat denaturation in a boiling water bath for 30 min followed by rapid cooling in an ice chilled water bath for 15 rain. 2 ml of the irradiated and denatured DNA sample was loaded in a thermally controlled (60-65°C) chromatography column (size 25 × 300 mm, Pharmacia Fine Chemicals) packed with 3.5-4.0 g of DNA grade Bio-Gel hydroxyapatite (Bio-Rad Laboratories, Richmond, CA). A linear gradient of 0.05-0.4 M phosphate buffer (pH 7) was used to separate the single-stranded (denatured, noncross-linked) and double-stranded (renatured, cross-linked) DNA using an LKB 11300 ultragrade gradient mixer [4,12]. About 2.3-2.5-ml fractions were collected from the column using an LKB 7000 Ultrorac fraction collector. Prior to separating the irradiated DNA samples, 2 ml of control samples of non-irradiated, native and heat denatured DNA samples were examined on a similar hydroxyapatite column under identical conditions. The cross-linked DNA was eluted between 45 to 65 ml volume which corresponded to the same retention volume as that of native double-stranded DNA. Percent cross-linked DNA was determined by measuring the absorbance of eluted samples at 260 nm. The meaning of the symbols + 1, +2, + 3, and + 4 are indicated in Table I. The singlet oxygen formation was detected by using two methods: (a) a specific method recently

proposed by Kraljic and Mohsni [13], and (b) the second, an indirect method, involving the photooxidative degradation of 2'-deoxyguanosine [14]. For the first method, a 10 ml solution of 0.35-0.4. 10- 5 M N, N-dimethyl-p-nitrosoaniline, used as a selective scavenger of 102, was prepared in 0.025 M phosphate buffer (pH 7); 10-2 M histidine was added as a selective acceptor of 102 and 0.1 ml psoralen solution to give 10 ttg/ml concentration of the sensitizer. 3 ml of this mixture were irradiated in a quartz cuvette for various time intervals to provide varying dose of ultraviolet radiation ranging from 0.5 to 10 J/cm2: The production of 102 was monitored by measuring the decrease in absorbance value of N,N-dimethyl-p-nitrosoaniline at 440 nm. The photooxidation of 2'-deoxyguanosine was studied by preparing a solution of a proper concentration (10, 1.0 and 0.1 /~g/ml) of the test compound in 0.01 M carbonate buffer (pH 10). An aliquot (5 ml) of this solution was adjusted to absorbance of 1.2-1.4 at 260 nm with 2'-deoxyguanosine. 3-ml samples with and without 2'deoxyguanosine were irradiated with 3.3 J / c m 2 of ultraviolet radiation. The degradation of 2'-deoxyguanosine was followed by monitoring the decrease of the absorption value at 260 nm. The formation of 0 2 radicals was studied by monitor.ing the photosensitized reduction of nitro blue tetrazolium [15]. The reduction of nitro blue tetrazolium by O~- leads to the formation of a blue-colored nitro blue formazan which can be quantitatively estimated spectrophotometrically at 560 nm. A 1.67.10 - 4 M solution of nitro blue tetrazolium was prepared in 0.01 M carbonate buffer (pH 10). The sensitizers, 100, 10 and 1 /~l, were added to about 9 ml nitro blue tetrazolium solution, and the final volume was brought up to 10 ml. 3-ml solutions were irradiated for varying periods up to a total of 0.5 J / c m 2 of ultraviolet radiation (320-400 nm). The production of reduced nitro blue tetrazolium was monitored by recording the increase in the absorbance of irradiated solutions at 560 nm. The generation of 102 and O 2 was further substantiated by carrying out quenching studies using sodium azide [16], 1,4-diazabicyclo[2,2,2]octane [17], fl-carotene [18], atocopherol (10 -2 M), and superoxide dismutase [19] as selective quenchers.

118

Results

Photosensitizing activity of various psoralens. In Table I, the relative photosensitizing activities of linear and angular psoralens are presented together with other chemicals such as hematoporphyrin derivatives, methylene blue, and riboflavin that are known to be very reactive photodynamic agents and exert their biologic effects in the presence of oxygen. It is apparent that linear psoralens exhibited a varying degree of skin photosensitizing activity. Some were extremely potent skin photosensitizing agents that induced marked erythema, edema, and blistering reaction; a few other linear psoralens (e.g., 3-carbethoxypsoralen, 4,8-dimethyl,5'-carboxypsoralen and 5'-di-

TABLE

I

SUMMARY FOR

ethylaminobutoxypsoralen), however, were inactive or nonphotosensitizing and induced no visible erythema or blistering reaction. This table also provides data, albeit qualitative in nature, about the extent of cross-linking in DNA. Although the data pertaining to the formation of monofunctional adducts and the formation of cross-links need to be presented in terms of the proportion of the damage they represent, we were unable to calculate absolute amounts of damage for lack of radiolabeled compounds. The relative rate constants for the formation of cross-linkages for 4,5',8-trimethylpsoralen was about two times greater than 8-methoxypsoralen or 5-methoxypsoralen. Hence, we have presented a rough estimate of the reaction in qualitative terms.

OF CHEMICALS

102 AND

Compounds

TESTED

FOR

SKIN

PHOTOSENSITIZATION,

PHOTOCONJUGATION

WITH

DNA,

AND

0 2 PRODUCTION

tested

Skin

A b i l i t y to f o r m

A b i l i t y to form

10 2

O~

photosensitizing

monofunctional

interstrand

production a

production a

activity a

adducts with DNA b

cross-links b

Bifunctionalpsoralens Psoralen

++++

++++

++

5-Methoxypsoralen

+ +

+ + +

+ +

++

++4-

+++

÷

+

8-Methoxypsoralen

+ + +

+ + +

+ 4-

+

+

4,5',8-Trimethylpsoralen

+ + + +

+ + + +

+ 4- + +

~-

+ +

+

+ +

+ 4-

+/-

+ +

+ + + +

+ + + +

+ + + +

+ +

not investigated

+ +

+ + +

+

+

+

-

+ +

+

+/+

3-Carbethoxypsoralen

-

+ +

-

+ + 4- +

Angelicin

-

+

-

+ + +

+ +

5-Methylangelicin

-

+ + +

-

+ +

+ +

4,5'-Dimethylangelicin

-

+ +

-

+

+

5-Diethylaminobutoxypsoralen 4'-Aminomethyl4,5',8-trimethylpsoralen 3,4'-Dimethyl8-methoxypsoralen 4,8-Dimethyl5'-carboxypsoralen

+

+ +

Monofunctionalpsoralens +

Other chemicals Hematoporphyrin

+ + +

-

-

+ + + +

+ +

Methylene blue

-

-

-

+ + + +

+

Anthracene

+ +

+/-

+/-

+ +

+ + +

Riboflavin

+/-

-

-

+ + +

+ + + +

a Key: -

derivatives

= absent; +/-

or highest. b Key: - = absent, +/-

= u n c e r t a i n o r v e r y w e a k ; + = m i l d o r w e a k ; + + ~ m o d e r a t e ; + + 4- = s t r o n g ; + + + + = v e r y s t r o n g =1-10%;

+ = > 10-25%;

+ + = > 25-50%;

+ + + = > 50-75%;

+ + + + = > 75-100%.

119

Ability of various linear and nonlinear psoralens to form singlet oxygen. Figure 1 shows the generation of 102 by both linear and angular psoralens using bleaching reaction of N,N-dimethyl-pnitrosoaniline at 440 nm induced by the presence of histidine as a selective acceptor. The figure also shows that the linear psoralens capable of forming interstrand cross-links in DNA (e.g., 5-methoxypsoralen, 8-methoxypsoralen, 4,5',8-trimethylpsoralen, and psoralen) and the nonlinear (angular) isopsoralens as well as linear psoralens, capable of forming only monofunctional adducts in DNA (e.g., 5-methylangelicin, angelicin, and 3carbethoxypsoralen), produced t o 2, although at varying degrees. Psoralen and 3-carbethoxypsoralen were the most effective producers of 102. Hematoporphyrin (1 /xg/ml), a well known producer of 10 2, generated a greater amount of t o 2 than many of the psoralens tested at 10 /~g/ml except 3-carbethoxypsoralen. The t O2-producing

activity based on the use of an equivalent concentration of the sensitizer, was found to be in the following order: 3-carbethoxypsoralen > hematoporphyrin > psoralen > angelicin > 5-methylangelicin > 4,8-dimethyl,5'carboxypsoralen > 3-amino, 4'-methylpsoralen > anthracene > 4,5',8-trimethylpsoralen > 4,5'-dimethylangelicin > 8-methoxypsoralen > 3,4'-dimethyl,8-methoxypsoralen > 5methoxypsoralen > 5'-diethylaminobutoxypsoralen.

Dose-response studies: effect of ultraoiolet radiation dose and concentration of psoralens on 102 production. The effect of varying the concentration of psoralens on 102 formation was studied with 8-methoxypsoralen, 5-methoxypsoralen, and 3carbethoxypsoralen. The results are shown in Fig. 2. A linear relationship was found between the 102 production and ultraviolet irradiation dose. It was interesting to observe that 0.1 # g / m l of 3carbethoxypsoralen produced more 102 (75 % more) than 10 # g / m l of 8-methoxypsoralen.

o

~

~ e. '..

""... "

o

5

fie-A,l/o¢

,s

....... "".. 6

~"-

"""c

"-..T

~.

"-9 o

B

"%

\

N

"'""o.. -.....

~8 ¢

,0

Fig. 1. The formation of singlet oxygen by various monofunctional and bifunctional psoralens is plotted as a function of ultraviolet radiation exposure dose. Reaction system: N,N-dimethyl-p-nitrosoaniline(3.5-4.10-s M)+histidine (1-10- 2 M) + sensitizer (10/~g/ml). The sensitizersused were: (1) 5-diethylaminobutoxypsoralen; (2) 5-methoxypsoralen; (3) 3,4'-dirnethyl-8-methoxypsoralen;(4) 8-methoxypsoralen;(5) 4,5'-dimethylangelicin; (6) 4,5',8-trimethylpsoralen; (7) anthracene; (8) 4'-aminomethyl-4,5',8-trimethylpsoralen;(9) 4,8-dimethyl5'-carboxypsoralen;(10) 5-methylangelicin;(11) angelicin; (12) psoralen; (13) hematoporphyrin derivatives; and (14) 3carbethoxypsoralen.

o~

2"s

~

7:s

UV-A, J/cM2 Fig, 2. A drug-dose response study of selected psoralen derivatives on the production o f 1 0 2. Reaction system: N,N-di-

methyl-p-nitrosoaniline(3.5-4.10- 5 M)+ histidine (1.10- 2 M) +sensitizer (0.1, 1.0 or 10/~g/ml). The sensitizers used were: (1) 8-methoxypsoralen,0.1 /zg/ml; (2) 8-methoxypsoralen,1.0 /xg/ml; (3) 5-methylangelicin0.1 #g/ml; (4) 5-methylangelicin, 1.0/Lg/ml; (5) 8-methoxypsoralen,10 t~g/ml; (6) 3-carbethoxypsoralen, 0.1 /~g/ml; (7) 5-methylangelicin,10 #g/ml; (8) 3-carbethoxypsoralen, 1.0 /tg/ml; and (9) 3-carbethoxypsoralen, 10 #g/ml.

120

Photooxidation of 2'-deoxyguanosine by psoralens. The production of 102 by psoralen and derivatives was also investigated by studying the photooxidative degradation of 2'-deoxyguanosine. Although the photodegradation of 2'-deoxyguanosine could involve radical processes (type-I mechanism) and hence may not be used as a specific method for monitoring the formation of 102 , we believe that this indirect method is useful in indicating the photooxidative damage to DNA that occurs preferentially in the presence of oxygen, and little or none in the absence of oxygen. Table II shows the percent degradation of 2'-deoxyguanosine with varying concentrations of photoreactive psoralens. The results obtained with hematoporphyrin derivatives are also given for comparison. Except for psoralen and hematoporphyrin derivatives, the t o 2 production and the photooxidation of 2'-deoxyguanosine with other psoralens did not follow the same order of efficiency for t o 2 production as was seen with the results of N, N-dimethyl-p-nitrosoaniline bleaching experiments. With 3-carbethoxypsoralen, the rate of 2'-deoxyguanosine photooxidation was slow initially up to 1.25 J / c m 2 ultraviolet radiation dose ( = 5.5% degradation) but was followed by a sharp increase to about 27.5% after exposure to 3.3 J / c m 2. This unusual characteristic of 3carbethoxypsoralen appears to be related to its TABLE 1I

Concentration ( # g/ml)

2'-Deoxyguanosine degradation

(%) Psoralen Psoralen 8-Methoxypsoralen 8-Methoxypsoralen 3-Carbethoxypsoralen 3-Carbethoxypsoralen Angelicin 5-Methylangelicin 5-Methylangelicin 4,5',8-Trimethylpsoralen Hematoporphyrin derivatives

Production of superoxide radicals by photoreactire psoralens. Another interesting and new finding of this study was that practically all the psoralen derivatives were found to generate O i and HO2 simultaneously along with the production of IO 2. Table III provides data for the nitro blue tetrazolium reduction test indicative of Oj formation by monofunctional and bifunctional psoralens. Riboflavin, hematoporphyrin derivative, and benzoyl peroxide were used as reference compounds for the selective generation of O£. Hematoporphyrin, which was earlier shown to be the most potent generator of ~O2, was found to be a weak producer of 03-. Psoralen, besides being an effecTABLE 111 PRODUCTION OF SUPEROXIDE RADICALS BY PHOTOSENSITIZED REACTIONS AND ITS QUENCHING BY SUPEROXIDE DISMUTASE Reaction system: sensitizer (10 #g/ml)+N,N-dimethyl-pnitrosoaniline (1.67.10 4 M, pH 10) and ultraviolet radiation (320-400 nm) = 0.5 J / c m 2. Compounds tested

PHOTODYNAMIC DEGRADATION OF 2'-DEOXYGUANOSINE BY SINGLET OXYGEN USING PSORALEN AND ITS DERIVATIVES AT 1 TO 10 ~ g / m l AND 3.3 J / c m 2 ULTRAVIOLET RADIATION Photosensitizing agents

photodecomposition to a new moiety as evidenced by the kinetics of this reaction and the spectroscopic changes in the absorption spectrum of 3carbethoxypsoralen. These observations appear to be in agreement with those recently reported by Ronfard-Haret et al. [8]. The results of this photochemical alteration of 3-carbethoxypsoralen will be published elsewhere.

1 10 1 10 1 10 10 1 10 10 10

17 40 3.1 12.7 11.5 27.3 6.5 7 15.25 3.4 86.5

O£ Effect of superoxide formation dismutase at 30-80 units (/~m) (% quenching)

Riboflavin 148.0 Psoralen 73.3 4,8-Dimethyl,5'carboxypsoralen 44.0 Angelicin 38.8 Benzoyl peroxide 37.0 4,5',8-Trimethylpsoralen 32.6 5-Methylangelicin 32.0 5-Diethylaminobutoxypsoralen 29.7 4,5'- Dimethylangelicin 25.3 3-Carbethoxypsoralen 24.5 8-Methoxypsoralen 23.0 Hematoporphyrin derivatives 20.6 3,4'-Dimethyl-8methoxypsoralen 19.1 5-Methoxypsoralen 10.6

83 91 94 88 100 100 93.5 88 100 76 97 97.2 97.5 97.5

121

tive generator of 102, was found once again to be the most potent O~- producer. Other psoralen compounds, however, did not follow the same order of efficiency for O 2 production as seen in the case of 102 production. 3-Carbethoxypsoralen, a very strong generator of 102, was found to be only a moderate producer of O~-.

Effect of various quenchers of 102 and O~ production. Additional evidence for the production of 102 and O~- was obtained by examining the bleaching of N,N-dimethyl-p-nitrosoaniline, the oxidation of 2'-deoxyguanosine, and the reduction of nitro blue tetrazolium in the presence of sodium azide (NAN3), 1,4-diazabicyclo[2,2,2]octane, flcarotene, a-tocopherol, and superoxide dismutase. In general, NaN 3 was found to be a specific quencher of 102, superoxide dismutase for O~, and 1,4-diazabicyclo[2,2,2]octane for both 102 and O 2. a-Tocopherol (10-3-10-2 M) and fl-carotene (10 -4 M) were also found to be good quenchers of 102. When NaN 3 (1 • 10 -2 M) or 1,4-diazabicyclo[2,2,2]octane (2.5- 10 -2 M) were added to the reaction mixture, about 80-100% inhibition in 102 production was observed with all the psoralens tested. Due to a strong absorption band of /3carotene in the 360 to 500-nm range, its quenching effect on 102 generation, although obviously noticeable, could not be quantitatively interpreted by our spectrophotometric procedure. Superoxide dismutase (30-80 units per ml), a known quencher of 0 2 and also an inhibitor of the formation of hydrogen peroxide, was found to cause 75-100% inhibition of O 2 formation (Table III). The addition of 1,4-diazabicyclo[2,2,2]octane also quenched the formation of O 2 as evidenced by the inhibition of nitro blue tetrazolium reduction. Azide ions, quenchers of 102, did not show any effect on the rate of nitro blue tetrazolium reduction caused by psoralens. Similarly, superoxide dismutase did not show any effect on the 102-specific N,N-dimethylp-nitrosoaniline bleaching test and the photooxidation of 2'-deoxyguanosine. It is of interest to note that the in vivo photosensitization reaction (erythema, edema) in the skin of five albino guinea pigs treated with topical application of 8-methoxypsoralen (10-15 ptg / c m 2) and subsequent exposure to ultraviolet radiation (320-400 nm, 6 J / c m 2) could be totally inhibited by pretreating the skin with singlet oxygen

quenchers like 1,4-diazabicyclo[2,2,2]octane (10-4-10 -5 M), sodium azide (10-2-10 -4 M) and superoxide anion quencher like superoxide dismutase (103-104 units). None of these agents was found to screen or absorb ultraviolet radiation in the 300-400 nm region. These observations strongly suggest that both 102 and O~- play a role in skin photosensitization response of psoralens.

Effect of 02, N2, and 21120 on the formation of 102 and 0 2. The generation of 102 was also studied by following the kinetics of N,N-dimethyl-pnitrosoaniline bleaching and 2'-deoxyguanosine photooxidation reactions under N 2 or in an 02purged system, and also by comparing the rate of 102 production in 2H20 and water. About 75-100% inhibition was observed in the rate of 102 production when the reaction was carried out under N2, and about 0-10% increase in the rate of 102 production was detected when pure 02 was bubbled through different photosensitizing systems. This suggested that the presence of molecular oxygen contributes to the generation of 102 in the photosensitized reaction induced by psoralens. When the reaction was carried out in 2H20 instead of H20, an increase in the rate of 102 production from 40 to 200% could be observed. 8-Methoxypsoralen, a weak generator of 102 in the presence of H20 and ultraviolet radiation (Figs. 1 and 2), showed a 40% increase in N,N-dimethylp-nitrosoaniline bleaching when 2H20 was used as a solvent.

Quenching study of interstrand cross-link formation by 8-methoxypsoralen. Bifunctional psoralens TABLE IV EFFECT OF QUENCHERS OF 102 AND 0 2 ON THE INTERSTRAND CROSS-LINKING REACTION OF 8METHOXYPSORALEN WITH DNA Reaction system: calf thymus DNA (0.1 mg/ml , pH 7.0, absorbance 1.2-1.4) + 8-methoxypsoralen (2 /~g/ml) + ultraviolet radiation (8 J / c m 2) with and without quenchers. Quencher used

Percent of cross-linking of DNA

NAN,- (10 "2 M) 1,4-Diazabicyclo[2,2,2]oetane (2.5.10- 2 M) Superoxide dismutase (80 urtits/ml) No quencher

72 71.5 72 74

122 are known to form interstrand cross-links in DNA upon irradiation [2-4]. To establish that the cross-linking of psoralen with DNA bases is a type-I, anoxic or oxygen-independent reaction, the interstrand cross-linking experiment was carried out in the presence of nitrogen, azide ions (10 -2 M), 1,4-diazabicyclo[2,2,2]octane (2.5.10-2 M), or superoxide dismutase (60 units/ml). The results are summarized in Table IV and show that: (a) Interstrand cross-linking of psoralen to DNA occurs in the anoxic condition (absence of oxygen) and the formation of cross-links, therefore, is a type-1 photosensitizing reaction; and (b) the triplet-excited state of psoralens is not affected by the quenchers of 102 and 0 2 .

Relationship between the production of lO: and Of and the ultraviolet absorption values of psoralens. To determine whether the production of 102 and 0 2 was related to the ultraviolet absorption values of the test chemicals, an absorption spectrum of 4.0.10 -5 M solution of each test chemical (3-carbethoxypsoralen, psoralen, 5-methoxypsoralen, 8-methoxypsoralen, 5-methylangelicin, isopsoralen, etc.) was recorded in the 320- to 400-nm range. Of the various test chemicals, the monofunctional 3-carbethoxypsoralen exhibited the highest absorption coefficient 1.85-10 4 M -~c m - I at 312 nm. To estimate the relative potency of 102 and 0 2 production for each test chemical, the area under the recorded absorption curve was calculated. The total absorbance value of 3carbethoxypsoralen in this range was arbitrarily equated to 100%. The calculated values for other test chemicals were: psoralen 60%, 5-methoxypsoralen, 5-methylangelicin, angelicin, 8-methoxypsoralen, 3-amino 4'-dimethylpsoralen, 3,4'-dimethyl-8-methoxypsoralen, 4,5',8-trimethylpsoralen (35 + 5%), 5-diethylaminobutoxypsoralen (25%) and 4,8-dimethyl-5'-carboxypsoralen (7.5%). A direct relationship between the production of 102 and O~- and the degree of ultraviolet radiation absorption was difficult to establish because of the weak optical absorptions in the 320-400 nm region and the low quantum yields. Nonetheless, the data on the ability of each chemical to produce 102 was to a certain extent related to the absorption value of each psoralen in the 320-400 nm range. The 3-carbethoxypsoralen, exhibiting the highest absorption value, was found to be the most

potent producer of 102 . It should be noted that 3-carbethoxypsoralen has been reported to be a good generator of 102 [7]. It was of interest to note that monofunctional psoralens (3-carbethoxypsoralen, 5-methylangelicin, and angelicin) produced more 102 than bifunctional psoralens (e.g., 4,5',8-trimethylpsoralen, 8-methoxypsoralen, etc.). 5-Methylangelicin and angelicin produced approximately two to four times more 102 than 8-methoxypsoralen and 4,5',8-trimethylpsoralen. These observations are of significance when the relative skin carcinogenic potential of 8-methoxypsoralen and psoralen are compared against 5methylangelicin and 4,5'-dimethylangelicin [20-22]. Compounds that showed a low absorption value in the 320-400-nm range tended to produce more O i than those exhibiting a high absorption value. 4,8-Dimethyl-5'-carboxypsoralen (4-10 -5 M) absorbed only 7% of 320-400-nm radiation in comparison to 3-carbethoxypsoralen (100%), yet it was found to generate O i almost 80% more than the equimolar concentration of 3-carbethoxypsoralen. Discussion

In mammalian skin, acute photosensitization reactions induced by bifunctional psoralens include an erythemal response (indicating damage to the endothelial and epidermal cells and is closely related to the dose of the drug and ultraviolet exposure), damage to DNA, RNA, cell membranes, proteins, and chromosomes and oxidation of lipids. The monofunctional psoralens, on the other hand, produce little or no visible erythema response, but do evoke damage to DNA, RNA, proteins and cell membranes of epidermal cells and promote increased sister chromatid exchanges in chromosomes. The chronic effects resulting from repeated exposures of the skin to psoralens and ultraviolet radiation are manifested in the form of actinically damaged skin (an indication of structural and functional alterations in proteins) and the development of basal or squamous cell carcinomas. In the past 20 years, several advances in our understanding of the skin photosensitization reaction by psoralens and other photosensitizers have emerged. The basic aspects of skin photosensitization reactions and the mode of therapeutic ef-

123 fectiveness of psoralens were recently reviewed [5,9,10,23,24]. The effects of psoralens on DNA, RNA, proteins, cell membranes, enzymes, lipids, etc., were believed to be fundamentally related to the primary mechanisms of skin photosensitization. It was shown that there were two distinct types of photoreactions which occurred independently of each other and concurrently when psoralen-treated skin was exposed to 320-400 nm radiation. The first, type-I, was an anoxic reaction (an oxygen-independent reaction) and involved the photocycloaddition reactions between pyrimidine DNA bases and the excited psoralen derivatives. The oxygen independent lethal effects of 8methoxypsoralen plus ultraviolet radiation in bacteria as reported by Oginsky et al. [25], or the primary events in skin reactions such as the inhibition of DNA synthesis, the killing of epidermal and dermal cells in psoriasis, or inhibition of recruitment of new cells in the G O and G 2 phases, etc., could be explained on the basis of photocyclo-addition reaction (type-I reaction) requiring the transfer of protons or electrons (sensitizer-substrate interaction) but no direct participation of oxygen in the transfer of energy to molecular oxygen. Oxygen-dependent type-II or sensitized reactions, involving the production of reactive oxygen (102 and O~-) were believed not to mediate psoralen photosensitization reactions (e.g., skin erythema and edema reaction or the cell membrane damaging reaction). Thus, until recently, it has been generally accepted that psoralens forming bifunctional adducts (cross-links) were only capable of evoking skin photosensitization, clearing psoriasis, and inducing skin cancer. Those psoralens forming monofunctional adducts were nonphotosensitizing (i.e., evoked no erythema or edema), noncarcinogenic, and therapeutically ineffective [6,17,18]. This generalization, however, is not strictly valid and exceptions are gradually emerging. 3-Carbethoxypsoralen, a monofunctional furocoumarin was found to be non-photosensitizing to human skin, did not exhibit carcinogenic activity in mice but was shown to possess therapeutic activity in psoriasis (Dubertret et al. [10]). Bifunctional psoralens (e.g., 4,8-dimethyl,5'carboxypsoralen and its methyl ester) and certain monofunctional psoralens (e.g., 5-methylangelicin, and 4,5'-dimeth-

ylangelicin) were also found to be nonphotosensitizing and yet were capable of clearing psoriasis [5,22,26,27]. Furthermore, the monofunctional 5methylangelicin and 4,5'-dimethylangelicin were found to be more carcinogenic than the bifunctional 8-methoxypsoralen [5,21,22]. Since 102 and O~- are highly reactive forms of oxygen and can damage DNA, proteins and membrane lipids [28,29], it was of interest for us to determine whether the skin photosensitization effects of bifunctional as well as monofunctional psoralens and the enhanced risk of cutaneous carcinoma resulting from bifunctional psoralens (8-methoxypsoralen, psoralen) and certain monofunctional psoralens (e.g., 5-methylangelicin and 4,5'-dimethylangelicin), could be better understood by sorting out the separate effects of type-I and type-II reactions. In this study, 12 linear and angular psoralens capable of forming monofunctional and bifunctional adducts with DNA were examined with a view to elucidate the role of 102 and O~ in photosensitization reactions and, if possible, in carcinogenic reactions. The production of 102 was established by observing the bleaching of p-nitrosodimethylaniline at 440 nm using histidine as a selective acceptor. Additional evidence for 102 involvement was obtained from quenching studies with azide ions and 1,4-diazabicyclo[2,2,2]octane or by carrying out reactions with 02, N2, 2H20 , and compounds such as hematoporphyrin and anthracene believed to be capable of generating 102 . The O~-generation was confirmed by using riboflavin and benzoyl peroxide as reference compounds in the nitro blue tetrazolium reduction reaction. The results showed that both the linear psoralens and angular isopsoralens produced 102 and O~ although at varying degrees. The linear psoralens such as psoralen and 3-carbethoxypsoralen produced more 102 than angular psoralens such as angelicin or 5-methylangelicin. It was of interest to note that nonphotosensitizing linear psoralen such as 3-carbethoxypsoralen and angular psoralens such as angelicin and 5-methylangelicin produced greater 1 0 2 than potent photosensitizing agents such as 8-methoxypsoralen, 4,5',8-trimethylpsoralen, and 5-methoxypsoralen. 3-Carbethoxypsoralen is a known agent for 10 2 production [8]. We have shown that skin photosensitization of

124

psoralens also involves a type-II sensitized reaction which was dependent on molecular oxygen and subsequent formation of ~O2 and 0 2. The formation of 102 and O~-, the reactive moieties of oxygen, can explain the damage of DNA (photooxidation of guanine bases), the inhibition of skin erythema reaction (indicating damage of cell membrane, cytoplasmic constituents such as lysosomes, mitochondria and increased sister chromatid exchanges in chromosomes), and the carcinogenic potential of monofunctional angelicins such as 5-methylangelicin and 4,5'-dimethylangelicin. Although O~- has been shown to be unreactive towards the various DNA components in aqueous solutions [30], we have reason to believe that the increased sister chromatid exchanges recently observed in our laboratory in chromosomes of mammalian skin fibroblasts treated with two angelicin derivatives and psoralen under conditions favorable for 0 2 generation but not for ~O2 production, suggest that O~- can induce damage to DNA. It is however, possible that the photo0xidative degradation of 2'-deoxyguanosine mediated by psoralens and increased sister chromatid exchanges in chromosomes may not involve type-II mechanism as suggested in this paper, but also radical processes of type-I mechanisms. Although we have little or no information on the binding sites and binding affinity constants of these different psoralens to cell membrane, we speculate that tO 2 and Oj may be involved in the induction of skin erythema reaction in the case of bifunctional psoralens and cell membrane-damaging responses manifested in the form of edema (swelling) reaction in the case of monofunctional psoralens. Clinical observations and examination of skin biopsies obtained after topical application of 5-methylangelicin or 3-carbethoxypsoralen and ultraviolet exposure has revealed evidence of edema reaction. In our opinion, psoralen-DNA cross-linkages have very little to do with the induction of erythema and skin photosensitization [5,24]. The psoralenDNA cross-link event represents a photochemical reaction which in most instances leads to the inhibition of DNA synthesis, cell death, and subsequent inhibition of cell proliferation, but not to an inflammatory reaction manifested in the form of skin erythema and edema reaction.

It is of interest to note that several other investigators have also recently reported the formation of 102 or 0 2. Poppe and Grossweiner [28,31] demonstrated 8-methoxypsoralen induced photoinactivation of hen lysozyme via the singlet oxygen pathway on the basis of six-fold higher efficiency of inactivation in 2HzO solution relative to H20. Cannistraro and Van de Vorst [32] presented electron spin resonance and optical absorption evidence for free radical involvement in photosensitizing action of furocoumarin derivatives and for their ability to produce singlet oxygen. Using 8methoxypsoralen and 4,5',8-trimethylpsoralen, Singh and Vadsaz [33] reported the involvement of singlet oxygen in the furocoumarin photosensitized inactivation of E. coli ribosomes. DeMol and Beijersbergen Van Hanegouwen [34] observed the formation of 102 by 8-methoxypsoralen in a study involving the photooxidation of 2-methyl, 2pentene and 3,4-dihydroxyphenylalanine (DOPA). Averbeck et al. [35,36] also reported oxygen dependent photodynamic killing of S. cereuisiae by 3-carbethoxypsoralen consistent with the relatively high yield of 102 . However, the biological significance of 102 and Oj production by these investigators has not been stressed in skin photosensitization and skin carcinogenesis reactions by psoralens. In view of the fact that the reactive products of oxygen reduction (O2, H202, and O H radicals) and excitation (102) have been implicated in tumor-generating capabilities of several chemicals, including the tumor-promotor phorbol ester or 4fl-phorbol, 12fl-myristate, 13-acetate [37], the question naturally arises as to whether psoralens promote the development of basal cell and squamous cell carcinomas by generating activated forms of oxygen, including 102, O~-, and HO2. The carcinogenic activity of bifunctional furocoumarins (psoralen, 8-methoxypsoralen, 4,5',8trimethylpsoralen) is easy to explain. It is related to: (a) their skin photosensitizing property of psoralens which causes proliferation of cells; (b) their ability to form monofunctional adducts and interstrand crosslinks that are difficult to repair but lead to error-prone repair. But we also support the view proposed by Hanawalt and his associates [38] that interstrand cross-links of 8-methoxypsoralen or psoralen with pyrimdine bases in DNA, by their very nature, must produce less repairable

125

than the monoadducts. Furthermore, persistence of cross-links most likely causes cell lethality, whereas the formation of monoadducts which are non-lethal to the cell and lead to error-prone DNA repair, appear to make monofunctional psoralens such as 5-methylangelicin and 4,5'-dimethylangelicin, more carcinogenic than bifunctional psoralens. The ability of monofunctional angleicin derivatives to generate more 10 2 and O~- than 8-methoxypsoralen and trimethylpsoralen, also contributes to the carcinogenic activity. We do not have a convincing explanation as to why 3carbethoxypsoralen, a monofunctional psoralen, is behaving differently than the monofunctional emthylangelicins; it is a potent generator of 10 2 and also produces O~- moderately, and yet it is non-photosensitizing to the skin and non-carcinogenic. Ronfard-Hart [8] recently reported that 3carbethoxypsoralen is a highly photolabile compound and undergoes rapid transformation to a cyclobutane dimer and to an ethanol (solvent) addition photoproduct. We have also confirmed that 3-carbethoxypsoralen is a non-carcinogenic agent in SKH : hr-1 hairless mice [21,22]. Recently, we have also observed that 3-carbethoxypsoralen is a highly photodegradable compound and the photoproducts (mostly mixed dimers) are less reactive moieties than 3-carbethoxypsoralen. Whether the differences in the skin penetration of these photoproducts of 3-carbethoxypsoralen or the differences in the binding capacity and cell transformation capacity of this monofunctional psoralen to cell nuclei, cell membrane, or cell type, play a major role in altering the reactivity of 3-carbethoxypsoralen, remains to be examined. The fact that monofunctional 5-methylangelicin, angelicin, and 4,5'-dimethylangelicin produce 102 and 0 2 at rates comparable to or better than bifunctional psoralens such as 8-methoxypsoralen suggests that the reactive moieties of oxygen and oxygen dependent sensitized type-II reactions play a definite role in understanding the carcinogenic property and perhaps DNA, and cell membrane damaging reactions. Studies involving the use of 10 2 and 0 2 quenchers are needed to obtain acceptable answers to these controversial issues.

Acknowledgements This investigation was supported by Grant No. 5-R01-CA-05003-24, awarded by the National Cancer Institute, Department of Health, Education, and Welfare. A partial support from the Paul B. Elder Company, Bryan, OH, is gratefully acknowledged.

References 1 Parrish, J.A., Stern, R.S. and Pathak, M.A. (1982) in The Science of Photomexticine (Regan, J.D. arid Parrish, J.A., eds.), pp. 595-623, Plenum Press, New York 2 Cole, R.S. (1971) Biochim. Biophys. Acta 254, 30-39 3 Dall'Acqua, F., Marciani, S., Zambon, F. and Rodighiero, G. (1979) Photochem. Photobiol. 29, 489-495 4 Cech, T., Patlaak, M.A. and Biswas, R.K. (1979) Biochim. Biophys. Acta 562, 342-360 5 Pathak, M.A. and Joshi, P.C. (1983) J. Invest. Dermatol. 80, 66-74 6 Pathak, M.A., Parrish, J.A. and Fitzpatrick, T.B. (1981) I1 Farmaco 36, 479-491 7 Bensasson, R.V., Land, E.J., Salet, C., Sloper, R.W. and Truscott, T.G. in Radiation Biology and Chemistry: Research Developments (Edwards, H.E., Nawaratnam, S., Parsons, B.J., and Phillips, G.O., eds.), pp. 431-439, Elsevier, Amsterdam 8 Ronfard-Haret, J.C., Averbeck, D., Benasson, R.W., Bisagni, E. and Land, E.J. (1982) Photochem. Photobiol. 35, 479-489 9 Averbeck, D. Fluitieme Congres International De Photobiologie et Colloque International Du CNRS, Strasbourg (France), 20-25, July 1980, Book of Resumes-Abstracts, p. 37 10 Dubertret, L., Averbeck, D., Zajdela, F., Bisagni, E., Moustacchi, R., Touraine, R. and Latarjet, R. (1979) Br. J. Dermatol. 101,379-389 11 Pathak, M.A., Worden, L.R. and Kaufman, K.D. (1967) J. Invest. Dermatol. 48, 103-118 12 Lawrey, P.D. and Brookes, P. (1967) J. Mol. Biol. 25, 143-166 13 Kraljic, I. and El Mohsni, S. (1978) Photochem. Photobiol. 28, 577-581 14 Rosenthal, I. and Pitts, J.N. (1971) Biophys. J. 11,963-966 15 Korycka-Dahl, M. and Richardson, T. (1977) J. Dairy Sci. 61, 400-407 16 Foote, C.S., Fugimoto, T.T. and Chang, Y.C. (1972) Tetrahedron Lett. 45-48 17 Ouannes, C. and Wilson, T. (1968) J. Am. Chem. Soc. 90, 6527-6528 18 Foote, C.S. and Denny, R.W. (1968) J. Am. Chem. Soe. 90, 6233-6235 19 McCord, J.M. and Fridovich, I. (1969) J. Biol. Chem. 244, 6049-6055 20 Pathak, M.A., Daniels, F., Hopkins, C.E. and Fitzpatrick, T.B. (1959) Nature 183, 728-730

126 21 Mullen, M.P., Pathak, M.A., West, J.D., Harrist, T.J. and DalrAcqua, F. (1983) J. Natl. Cancer Inst. in the press 22 West, J.D., Mullen, M.P. and Pathak, M.A. (1982) J. Invest. Dermatol. 78, 361 23 Musajo, L., Rodighiero, G., Caporale, G., Dall'Acqua, F., Marciani, S., Bordin, F., Baccichetti, F. and Bevilacqua, R. (1974) in Sunlight and Man: Normal and Abnormal Photobiologic Responses (Pathak, M.A. et al., eds.), pp. 369-387, University of Tokyo Press, Tokyo 24 Pathak, M.A. (1982) J. Natl. Cancer Inst. 69, 163-170 25 Ojinsky, I.L., Green, G.S., Griffith, D.G. and Fowlks, W.L. (1959) J. Bacteriol. 78, 821-833 26 Pathak, M.A. and Fitzpatrick, T.B. (1981) Abstr. 9th Annual Meeting, American Society for Photobiology, p. 109, Williamsburg, VA 27 Pathak, M.A., Marciani, S., Guiotto, A. and Rodighiero, G. (1983) J. Invest. Dermatol. 81, 533-539 28 Grossweiner, L.I. (1982) in Psoriasis (Farber, E.M., Cox, A.J., Nail, L. and Jacobs, P.H., eds.) pp. 135-141, Grune and Stratton, New York

29 Grossweiner, L.I., Patel, A.S. and Grossweiner, J.B. (1982) Photochem. Photobiol. 36, 159-167 30 Cadet, J. and Teoule, R. (1978) Photochem. Photobiol. 28, 661-667 31 Poppe, W. and Grossweiner, L.I. (1975) Photochem. Photobiol. 22, 217-219 32 Cannistraro, S. and Van De Vorst, A. (1977) Biochim. Biophys. Acta 476, 166-177 33 Singh, H. and Vadsaz, J.A. (1978) Photochem. Photobiol. 28, 539-545. 34 De Mol, N.J. and Beijersbergen Van Henegouwen, D.M.J. (1979) Photochem. Photobiol. 30, 331-335 35 Averbeck, D., Averbeck, S. and DalrAcqua, F. (1981) I1 Farmaco 36, 492-505 36 Averbeck, D. (1982) in Trends in Photobiology (Helene, C., Charlier, M., Montenay-Garestier, T. and Laustriat, G., eds.), pp. 295-308, Plenum Press, New York 37 Marx, J.L. (1983) Science 219, 518-519 38 Hanawalt, P.C., Su-Chin, L. and Parson, C.S. (1981) J. Invest. Dermatol. 77, 86-90.