14

Biochimica et Biophysica Acta, 1054 (1990) 14-20 Elsevier

BBAMCR 12739

The mechanism of action of the nitrosourea anti-tumor drugs on thioredoxin reductase, glutathione reductase and ribonucleotide reductase K a r i n U. S c h a l l r e u t e r 1, F l o r e n c e K. G l e a s o n 2 a n d J o h n M. W o o d 3 1 Department of Dermatology, University of Hambur~ Hamburg (F..R.G.), 2 Department of Botany, College of Biological Sciences, University of Minnesota, St. Paul and 3 Department of Biochemistry, Gray Freshwater Biological lnstitute, University of Minnesota, Navarre, MN (U.S.A.) (Received 26 February 1990)

Key words: Antitumor drug; Thioredoxinreductase; (Human); (Melanotic melanoma)

The nitrosoureas BCNU, CCNU, ACNU, and Fotemustine covalently deactivate thioredoxin reductase, glutathione reductase and ribonucleotide reductase by alkylating their thiolate active sites. Since thioredoxin reductase and glutathione reductase function as alternative electron donors in the biosynthesis of deoxyribonucleotides, catalyzed by ribonucleotide reductase, the inhibition of these electron transfer systems by the nitrosoureas could determine the cytostatic property of this homologous series of drugs. A detailed study of the kinetics and mechanism for the inhibition of purified thioredoxin reductases from human metastatic melanotic and amelanotic melanomas by the nitrosoureas showed significantly different inhibitor constants. This difference is due to the regulation of these proteins by calcium. Calcium protects thioredoxin reductase from deactivation by the nitrosoureas. In addition, it has been shown that reduced thioredoxin displaces the nitrosourea-inhibitor complex from the active site of thioredoxin reductase to fully reactivate enzyme purified from human metastatic amelanotic melanoma. It has been possible to label the active sites of thioredoxin reductase and glutathione reductase by using chloro[14C]ethyl Fotemustine, resulting in the alkylation of the thiolate active sites to produce chloro[14C]ethyl ether-enzyme inhibitor complexes. These complexes can be reactivated via reduced thioredoxin and reduced glutathione, respectively, by a t-elimination reaction yielding I t4Clethylene and chloride ions as reaction products.

Introduction Since 1975, the nitrosourea anti-tumor agents have been broadly used against brain tumors, Hodgldn's lymphomas, melanoma, lung cancer, breast cancer and colorectal cancer [1]. Pharmacokinetic studies have indi-

Abbreviations: TRM, human metastatic melanotic melanoma thioredoxin reductase; TRAM , human metastatic amelanotic melanoma thioredoxin reductase; TRE. coti, Escherichia coli thioredoxin reductase; Tox, oxidized thioredoxin; Tr,,a, reduced thioredoxin; DTNB, 5,5'-dithio-bis(2-nitrobenzoic acid); NADPH, reduced nicotinamide adenine dinucleotide phosphate; CDP, cytidine diphosphate; dCDP, deoxyribose cytidine diphosphate; BCNU, 1,3-bis(2chloroethyl)-2-nitrosourea; CCNU, 1,2-chloroethyl-3-cyclohexyl-1nitrosourea; ACNU, 1-(4-amino-2-methyl-5-pyrimidyl)methyl-2-chloroethyl-3-nitrosourea; Fotemustine, diethyl(-1-[3-(2-chloroethyl)-3nitrosoureido) ethylphosphonate; DTIC, dacarbazin. Correspondence: K.U. Schallreuter, Department of Dermatology, University of Hamburg, Martinistrasse 52, D-2000 Hamburg 20, F.R.G.

cated that these drugs have been highly reactive undergoing rapid transport across the blood-brain barrier with concomitant rapid degradation [1]. It has been suggested that this instability of the nitrosoureas primarily resides in their susceptibility to nucleophilic attack causing the displacement of, and alkylation by, a chloroethyl-carbonium ion and the release of this ion from the nitrosoureas has been proposed as the common mechanism of action for this homologous series of drugs [2]. However, the nitrosoureas are so reactive that it has proved difficult to determine specific mechanisms of action, especially in relation to the common cytostatic property of preventing DNA synthesis [1,2]. Recently, it has been shown that the nitrosoureas, and other chemotherapeutic drugs acting as alkylating agents, inhibit hepatic O6-alkylguanine-DNA alkyltransferase [3]. It has been proposed that this metabolic block may represent a common mechanism for the cytotoxicity/mutagenicity of this class of drugs [4-6] since a deficiency in DNA repair of O-alkylguanine ha: been shown to correlate with increased susceptibility t(

0167-4889/90/$03.50 © 1990 Elsevier Science Pubfishers B.V. (Biomedical Division)

15 malignant transformation and the sensitivity to alkylating drugs [3-6]. Initially, treatment protocols with the nitrosoureas were conducted with BCNU, CCNU and methyl-CCNU, but more recently, ACNU has been introduced. At the present time, phase III clinical trials are underway with Fotemustine in the treatment of disseminated metastatic melanoma. The role of the thioredoxin reductase/thioredoxin system as electron donors to the ribonucleotide reductases has been well-established and the general antioxidant properties of these thioproteins have suggested more than one functionality [7]. More recently, the importance of this electron transport system has been verified in free radical defense and in the regulation of melanin biosynthesis [8-10]. Thioredoxin reductase activity has been shown to differ with skin types and is decreased in depigmentation disorders such as vitiligo and in tyrosinase positive albinism (Hermansky-Pudlak Syndrome) [11-13]. Only recently, it has been discovered that melanotic murine and human melanoma cells, established in synthetic culture medium, yielded abnormally high membrane associated thioredoxin reductase activities compared to melanocytes cultured from normal healthy adult donors [14]. Earlier immunohistochemical experiments on the distribution of thioredoxin reductase in the rat, indicated that melanocytes have moderate to high levels of this enzyme both in the cell cytosol and on plasma membranes [15,16]. Also, it has been shown that thioredoxin reductase, thioredoxin and ribonucleotide reductase levels are increased in rapidly dividing cells underlining the importance of these thioproteins to deoxyribonucleotide biosynthesis [16,17]. The determination of thioredoxin reductase activities in human metastatic melanoma tissues yielded increased activity compared to normal skin controis in all cases tested so far [18]. A study of thioredoxin reductase activities on 30 primary melanotic melanomas yielded two different groups with one group (n = 19) revealing more than double the activity over each individual's normal skin control, and a second group (n = 11) with equal or less enzyme activity compared to normal skin [19]. Only recently, these differences in thioredoxin reductase activity have been explained by the influence of calcium status [20-22]. The thioredoxin reductase/thioredoxin electron transport system represents only one pathway for ribonucleotide reduction, whereas the glutathione reductase/glutaredoxin system functions as an alternative source of electrons for DNA biosynthesis (Scheme I). For instance, mutants of Escherichia coli, lacking thioredoxin reductase, still synthesize DNA and divide slowly by using the glutathione reductase/glutaredoxin pathway [23]. As a consequence, the inhibition of DNA synthesis at the stage of ribonucleotide reduction very likely requires the inactivation of both systems or ribonucleotide reductase itself (Scheme I). Based on this

NADPH +H

/

.SH

" y " ' y " , \, , y ' A

A

NAOe*J % r ~ J

\~

/sA

I jr NT

S I

II

Thloredo~n reducfmm OR) Olul~tdorm lxoeemlc onx~, e~ AD /

.S II \

/ \

SH

!y'.y

/sHA

/sA

J ~nR I J ' ~

s II I 11'MorodoxlnRlbonucleo~e 03 nm~cme (RR) /s

.SH ,¢

+,,.y y%y°,.__%y

NAOPH

l~

S

Rilx~e unit

/SH

unit

R , - - untl

$

unl I

II elum'e(~xJn reductue (OR)

II I elutamdo3dn Rlbonucleotlde (O) rmuct=~ (RR)

Scheme I. The alternative electron transfer pathways for ribonucleotide reduction by ribonucleotide reductase.

knowledge, we have examined the direct effect of the nitrosoureas on the activities of thioredoxin reductase, glutathione reductase and ribonucleotide reductase.

Materials and Methods Enzymes Thioredoxin reductase has been purified from metastatic amelanotic and melanotic melanoma tissues by the method of Schallreuter and Wood [14,18]. Enzyme purity (>95%) has been established by FPLC and polyacrylamide gel electrophoresis [18,22,24]. Both thioredoxin reductases yielded a significant difference in activity. Thioredoxin reductase from amelanotic melanoma (TR-AM) has been isolated completely calcium-free, whereas the enzyme from melanotic melanoma (TR-M) contained bound calcium. These enzymes could be distinguished by FPLC and by kinetic analyses (i.e., for TR-AM normal Michaelis-Menten kinetics and for TR-M sigmoidal kinetics) [22]. It should be recognized that earlier studies with thioredoxin reductase showed allosteric regulation by calcium for this enzyme [21,22]. A single EF-hands binding site for calcium has been discovered on both the human melanoma and the E. coli thioredoxin reductases. Glutathione reductase from human erythrocytes was purchased from Sigma Chemical Company, St. Louis, MO. Ribonucleotide reductase was purified by the method of Gleason and Holmgren [25]. Crystalline Bovine Serum Albumin was obtained from British Drug Houses, U.K. Chemicals The structures of the four nitrosoureas used in this study are presented in Fig. 1. Each compound has the chloroethylnitrosourea functional group differing only with respect to the different non-polar R-groups. Fotemustine has been a generous gift from the Servier

16 O II CI--CH2 CH2--N--C--N--R I

I

N---O H OH3 I R

=

jOC2H5

-- CH--,P.,.O

O

C2 H5

FOTEMUSTINE R

= --CH2CH2CI BCNU

CCNU NH2 R

=

-CH2~-CH

3

ACNU Fig. l. Thestructuresofthenitrosoureas.

Company, Paris, France. B C N U (Carmubris) was from Bristol Labs (Arzneimittel), C C N U was from H a m b u r g Medac, A C N U was from Asta Pharma AG, F r a n k f u r t / Main. All other reagents were purchased from Sigma Chemical Company, St. Louis, MO. [14C]Fotemustine (53 m C i / m m o l ) was provided by Servier Company, France. The radiochemical purity ( > 98%) determined by (1) H P L C on a Zorbax ODS column eluted with aqueous acetonitrile (10%, v / v ) , and (2) by T L C on Silica-gel using dichloromethane/ethanol (95:5, v / v ) and toluene dichloromethane methanol (50 : 40 : 5, v / v ) . Structural analysis was confirmed by mass spectrometry.

nitrosoureas on T R - A M , Fotemustine, as one representative, has been used for all other experiments. A comparison of the inhibition of T R - A M and T R - M by Fotemustine clearly shows that T R - M is more resistant to this drug. This result can be explained by the calcium-dependent protection of the active site of this enzyme by the EF-hands induced conformational change (Fig. 2). Fig. 3a shows that h u m a n glutathione reductase is inhibited by Fotemustine and Fig. 3b presents the concentration-dependent inhibition of C D P reduction by ribonucleotide reductase. A comparison of the inhibition of the three thioproteins by Fotemustine showed that glutathione reductase is more resistant than T R - A M and ribonucleotide reductase. In order to examine the mechanism of action of the nitrosoureas, a kinetic analysis for the inhibition of T R - A M in the presence and absence of 10 3 M BCNU, C C N U and Fotemustine has been performed (Fig. 4). An equivalent decrease in the m a x i m u m velocity (Vmax) of T R - A M occurred with all three nitrosoureas proving covalent deactivation through alkylation of the active site. In order to test the effect of reduced/oxidized thioredoxin on the stability of the TRAM inhibitor complex, the D T N B reduction assay has been performed. Fig. 5 presents the stoichiometry for the reactivation of T R - A M after complete inhibition with 10 -3 M Fotemustine. Enzyme activity has been fully restored by the addition of a 12 : 1 ratio of reduced thioredoxin over the TR-AM-Fotemustine-inhibitor complex. These data unambiguously prove that the reactive thiolate group on reduced thioredoxin can dealkylate the TR-AM-Fotemustine-inhibitor complex. Fotemustine-labeled 14C in the chloroethyl group has been used to gain more

Enzyme assays Mammalian thioredoxin reductases and glutathione reductases reduce the disulfide bond of D T N B using N A D P H as the electron donor. D T N B reduction was measured at 412 nm by the method of Luthman and Holmgren [20]. Ribonucleotide reductase was assayed by d C D P formation from C D P by the method described by Gleason and Holmgren [25].

~EE0.8

~

0.4

Results

In order to examine the effect of the nitrosoureas on the thioredoxin reductases T R - A M and TR-M, the rates for the reduction of D T N B were assayed over a concentration range of 0-7.5 • 10 -3 M nitrosourea (Fig. 2). Each of these drugs inhibits T R - A M to approximately the same extent indicating a c o m m o n inhibitory mechanism for this homologous series of compounds. Due to the common inhibition mechanism observed for the

0

" O.5 J nitrosoureas X 10-5 M

Fig. 2. Inhibition of TR-AM and TR-M by the nitrosoureas. Reactions contained 1.0 ml of DTNB reaction mixture, 30 #1 of TR-AM (1.67 mg/ml) or 60 #1 of TR-M (0.8 mg/mi) and the different concentrations of the nitrosoureas as indicated. All reactions were started by the addition of enzyme. Fotemustine TR-AM (o ©), Fotemustine TR-M (~ ~), BCNU (A A), CCNU (zx zx), and ACNU (O O).

17

0.4 la o

!ii

% ~4E o 3_ 1:3 0

"5

-0

~2-

rr rn Z 0.1

E °

t-

1-

0 o

,.~ Fotemustine,X10-3 M

£5

0

~ ~

~

O ;, ~ Fotemustine,X 10-3M

Fig. 3. (a) The inhibition of human erythrocyte glutathione reductase by Fotemustine. Reactions contained 1.0 ml of DTNB reaction mixture with 50 /tl of glutathione reductase (2.0 mg/ml) in the presence and absence of 0-2.5.10 -3 M Fotemustine. (b) The inhibition of ribonucleotide reductase by Fotemustine (0-7.5.10 -3 M) reactions contained 0.5.10 6 M of thioredoxin, 10 -3 M dithiothreitol (DTr), and 5.0 /~g of ribonucleotide reductase. Enzyme activity is expressed in nmol of dCDP formed from saturating CDP/10 rain.

detailed information in the inhibition and reactivation of TR and GR. For these experiments, T R (1.0 mg enzyme) and GR (2.0 mg enzyme) were preincubated with a mixture of 50.10 -6 M NADPH and 24- 10 -6 M DTNB. After 5 min, the enzymes were fully reduced as determined by the deep yellow color of the reaction mixtures. Next, 10/~1 of [14C]Fotemustine (53 m C i / m M) were added to each incubation mixture and dialysed for 24 h in Tris-EDTA buffer, pH 7.5 (50.10 -3 M Tris,

with 10.10 -3 M EDTA). After dialysis, radioactive incorporation into TR and GR was determined on the 14C-labeled channel of a Hewlett-Packard scintillation

0.3-

E

tC~

0.6

/

ojO

o

-

tO

0.2-

"0

II1 z ka

0.4H /

0.1-

i /

12:1 6:1

~0.2

t~

z

I

/

Y _3

~-'~_~*

time,minutes o

dl

o'2 o'.a X 10-5MDTNB

o)4

Fig. 4. V vs. S plot with DTNB as the substrate in the presence of 10 -3 M Fotemustine (m m), BCNU (zx zx), CCNU (o o) and control ( o o).

Fig. 5. The reactivation of Fotemustine-inhibited TR-AM by reduced thioredoxin. Reactions contained 1.0 ml of DTNB reaction mixture plus 60/11 of TR-AM (1.67 mg/ml) in the presence (m m) and absence (© o) of 10.10 -3 M Fotemustine. At 2, 3 and 4 min, increment increases of reduced thioredoxin were added showing reactivation of TR-AM.

18

14J a

o

i2 [

,?

""°"'°""~

°'" °

I

¢0 4 -

~ I

&20

0!

20

0

4'0

0

60

Time (minutes)

~

~

2

4

,

,

,

6 8 10 Time (minutes)

~

12

14

Fig. 6. (a) The volatilization of 14C-label from the GR-[14C]Fotemustine inhibited system by the addition of 10-3 M reduced glutathione. (b) The volatilization of 14C-label from the TR-[14C]Fotemustine inhibitor complex by the addition of a 20-fold excess of reduced thioredoxin (6.6 mg) over TRAM (1.67 mg).

spectrometer in 10 ml Ecolume scintillation fluid/ sample. TR was labeled with 16 605 cpm and GR with 31464 cpm [14C]Fotemustine. Both TR- and GR[t4C]Fotemustine-inhibitor complexes were unstable to their respective reaction products reduced thioredoxin and reduced glutathione. Fig. 6a shows the volatilization of the chloro[14C]ethyl label from the GR[14C]Fotemustine-inhibitor complex upon the addition of 10-3 M reduced glutathione and Fig. 6b presents the rapid volatilization of the chloro[14C]ethyl group from

0.4-

.=~a3-

the TRAM-[14C]Fotemustine complex by a 20-fold excess of reduced thioredoxin (i.e., 1.67 mg of TRAM[14C]Fotemustine complex (27146 cpm) plus 6.6 mg of reduced thioredoxin). (TRAM and thioredoxin have molecular mass of 58 kD and 11.4 kD, respectively.) The TRAM-[~4C]Fotemustine complex spontaneously loses its radioactivity upon oxidation of the enzyme to its disulfide form. In addition, a study of other cytostatic drugs on the activity of TRAM has been performed. Adriablastin, Bleomycin, Endoxan, Melphalan, Cisplatin, DTIC, Vinblastin and Holoxan showed no reaction with TRAM at concentrations used to inhibit this enzyme by the nitrosoureas, whereas Mithramycin, Methotrexate and 5-Fluorouracil were very weak inhibitors of this enzyme (Fig. 7). Discussion

0.2-

no

~ 0.1

.

0

2

4

.

6 8 Time, minutes

.

.

10

12

Fig. 7. The inhibition of TR-AM by 10 -3 M 4-fluorouracil (o o), methotrexate (A) and mithramycin (1:3 [3). Reactions contained 1.0 ml of DTNB reaction mixture and 30/~l of TR-AM (1.67 mg/ml) (O O).

To date, the nitrosoureas have been of limited use due to their cumulative renal, bone marrow and pulmonary toxicity. It has been proposed that this toxicity is largely due to the inhibition of glutathione reductase with specific recognition that this enzyme maintains a sufficient concentration of reduced glutathione in cells. Reduced glutathione is known to function in detoxification of alkylating agents and by forming Michael-addition complexes with xenobiotics [27,28]. Fotemustine has been unique in its lack of toxicity among the nitrosoureas and this has been shown to be due to its weak inhibition of tissue and cellular glutathione reductase [29]. In this report, we have performed a detailed study on the interaction between 14C-labeled Fotemustine with the key enzymes in the ribonucleotide reduction pathway TR, GR and RR. Several general conclusions can be drawn from the reaction between the nitrosoureas and the thioproteins

19

oH2\ S

/0.2 \

SH / r.,~A~."~----'8~'~[9 S®

SH / ÷CICH2CH2-NsC--N-I~

Therefore, the calcium status of human melanoma cells may be very important in the proliferation, differentiation and susceptibility of these cells to anti-tumor drugs such as the nitrosoureas.

Acknowledgements We kindly acknowledge Louise Mohn for typing the manuscript and Doris Lewandowski for drawing the figures. This research has been supported by Servier Company and the University of Hamburg. (DEACTIVATEDTR) Scheme II. Proposed reactivation pathway for [14C]Fotemustine-inhibited TRAM by reduced thioredoxin, fl-elimination produced 14Clabelled CH2=CH 2, oxidized thioredoxin and C1-. Spontaneous elimination may occur when deactivated TR is oxidized to the disulfide. A similar mechanism can bo proposed for the reactivation of Fotemustine inhibited GR by GSH.

thioredoxin reductase, glutathione reductase and ribonucleotide reductase: (i) A common mechanism for deactivation of these thioproteins occurs by alkylation of their respective thiolate active sites. (ii) The direct alkylation of TR-AM and GR by dechlorination of the nitrosoureas can be ruled out since the other chloroethyl-containing drugs Endoxan and Melphalan do not inhibit this enzyme. (iii) The thioethers formed upon alkylation of the thiolate active sites of TRAM and GR are susceptible to nucleophilic attack by reduced thioredoxin and reduced glutathione respectively. However, in the case of TRAM, spontaneous volatilization of the chloro[lac]ethyl group can occur slowly upon the oxidation of the TR active site from thiolate to disulfide. (iv) The alkyl group donated to TRAM and GR by the nitrosoureas must contain the electron-withdrawing atom (i.e., chlorine) to allow the volatilization of the chloro[lac]ethyl group as 14C-labelled ethylene and chloride ions by a fl-elimination mechanism (Scheme 2). (v) The other fragment from the nitrosoureas could only carbamoylate TR-AM or GR to form an unstable thioester-inhibitor complex. If carbamoylation did occur, this inhibitor should be unstable at pH 8.0 to yield reversible (competitive) enzyme kinetics and not the observed covalent deactivation mechanism (Fig. 4). (vi) The allosteric inhibition of TR by calcium, as represented by TR M, highlights the protective role played by this ion in the resistance of this enzyme to the nitrosoureas. Since calcium has been shown to play a major role in cell division and differentiation [30,31], its protection of the active site of thioredoxin reductase will depend on the calcium status in the cell cycle.

References 1 Wassermann, T.H., Slavik, M. and Carter, S.K. (1975) Cancer 36, 1258-1266. 2 Goodman, L.S. and Gilman, S.G. (1985) in The Pharmacological Basis of Therapeutics, Vol. 7, pp. 1260-1262, MacMillan, New York. 3 Meer, L., Schold, S.C. and Klethufs, M. (1989) Biochem. Pharmacol. 34 : 6, 929-934. 4 Shiloh, Y. and Becker, Y. (1981) Cancer Res. 41, 5114-5120. 5 Bodell, W.J., Alda, T., Barger, M.S. and Rosenblum, M.L. (1985) Environ. Health Perspect. 62, 119-126. 6 Dolan, M.E., Young, G.S. and Pegg, A.R. (1985) Cancer Res. 46, 4500-4505. 7 Holmgren, A. (1985) Annu. Rev. Biochem. 54, 237-271. 8 Schallreuter, K.U. and Wood, J.M. (1986) Biochim. Biophys. Res. Commun. 136, 630-637. 9 Wood, J.M. and Schallreuter, K.U. (1988) Inorg. Claim. Acta 151, 7. 10 Schallreuter, K.U. and Wood, J.M. (1989) Free Rad. Biol. Med. 6, 519-532. 11 Schallreuter, K.U. and Witkop, C.J., Jr. (1988) J. Invest. Dermatol. 90, 3, 372-377. 12 Schallreuter, K.U. and Pittelkow, M.R. (1988) Arch. Dermatol. Res. 280, 137-139. 13 Schallreuter, K.U. and Pittelkow, M.R. (1989) Arch. Dermatol. Res. 281, 40-44. 14 Schallreuter, K.U. and Wood, J.M. (1987) Cancer Lett. 36, 297305. 15 Rozell, B., Hansson, H.A., Luthman, M. and Holmgren, H.A. (1985) Eur. J. Cell Biol. 38, 79-86. 16 Hansson, H.A., Holmgren, A., Rozell, B. and Stemme, S. (1986) in Thioredoxin and Glutaredoxin Systems (Holmgren, A., Branden, C.-I., Jornvall, H. and Sjoberg, B.-M., eds.), pp. 134-141, Raven Press, New York. 17 SchaUreuter, K.U. and Pittelkow, M.R. (1987) Arch. Dermatol. 123, 1492-1498. 18 Schallreuter, K.U. and Wood, J.M. (1988) Biochim. Biophys. Acta 967, 103-109. 19 Schallreuter, K.U. and Wood, J.M. (1990) Cancer Res. (submitted). 20 Schallreuter, K.U., Pitteikow, M.R. and Wood, J.M. (1986) J. Invest. Dermatol. 87, 728-732. 21 Schallretuer, K.U., Pittelkow, M.R., Gleason, F.K. and Wood, J.M. (1986) Inorg. Biochem. 28, 227-238. 22 Schallreuter, K.U. and Wood, J.M. (1989) Biochim. Biophys. Acta 997, 242-247. 23 Fuchs, J. (1977) J. Bacteriol. 129, 967-974. 24 Schallreuter, K.U. and Wood, J.M. (1989) Biochim. Biophys. Res. Commun. 160, 573-579. 25 Gleason, F.K. and Holmgren, A. (1981) J. Biol. Chem. 256, 8306-8309.

20 26 Luthman, M. and Holmgren, A. (1982) Biochemistry 21, 66286633. 27 Moore, M.L. and Dahl, R.H. (1978) Nature 271,650-662. 28 Kosower, M.S. and Kosower, E.M. (1983) in Functions of Glutathione (Larson et al., eds.), pp. 307-322, Raven Press, New York. 29 Boutin, J.A., Norbeck, K., Moldens, P., Gerton, A., Paraire, M.,

Bizzari, J.P., Lavielle, G. and Cudenner, C.A. (1989) Eur. J. Cancer Clin. Oncol. 23 : 9, 1311-1316. 30 Wille, J.J., Jr., Pittelkow, M.R. and Scott, R.E. (1984) J. Cell. Physiol. 121, 31-44. 31 Pittelkow, M.R., Wille, J.J., Jr., and Scott, R.E. (1986). J. Invest. Dermatol. 86, 410-417.