Free Radical Reduction by Thioredoxin Reductase at the Surface of Normal and Vitiliginous Human Keratinocytes*

Karin U. Schallreuter, M .D ., Mark R. Pittelkow, M.D., and John M. Wood, Ph.D. Departm ents of Dermatology (KUS) an d Biochemistry (JMW) , University of Minneso ta School of Medicine, Minnea polis, and · Department of Dermatology, Mayo Clinic (MRP), Roches ter, Minnesota, U.S.A.

Cell cultures of human keratinocytes contain m embrane­ ulated by calcium concentrations of the cell culture m edium. associated thioredoxin reductase that is extremely active in Stratified keratinocytes are half as active in medium con­ reducin g radicals on the outer plasma membrane. This en­ taining 2 mM Ca + + compared with 0.1 mM Ca + + con­ zyme activity was confirmed by its purification from cul­ centration. (4) Product inhibition of the enzyme occurs tures of stratified human keratinocytes by affinity column with oxidized coenzyme NADP + (i. e., 87% inhibition of chro m atography. The enzym e was assayed both in vivo enzyme activity over 30 min). The enzyme is heat stable and in vitro usin g a spin-labeled quaternary ammonium at temperatures of70°C for 10 min. It is inactivated at 75°C . compound as the substrate, under saturating conditions in A comparative study of thioredoxin reductase activity on free radical substrate. Specific activities were determined stratified differentiated and undifferentiated rapidly grow­ by monitoring the sequential decrease in the amplitude of ing celI s was performed . Also, enzyme activity was quan­ the electron spin resonance signal per unit of cell protein . titated for cultured keratinocytes isolated from vitiliginous The following properties were found: (1) Cultures of adult and normal skin of the same donor. The results of this stratified cells have approximately twice the thioredoxin study, and the connection between this enzyme activity reductase activity of neonatal cells. (2) The enzyme is in­ and UV- generated free radicals are reconciled in terms of hibited by thioprotein inhibitors (i .e., parachloromecuri­ the m echanism of action and m etabolic activity of thio­ benzoate and ·dinitrochlorobenzen e). (3) The activity is reg- redoxin reductase. J Invest Dermatol 87:728-732, 1986

hioredoxin reductase and thioredoxin are ubiquitous preventing inactivation of sulfur-containing amino acids in pro­ electron-transfer systems common to a variety of or­ teins; (3) to function as electron donor for reduction of sulfate to ganisms and tissues [1,2]. In higher organisms these sulfite; (4) the reduction of protein disulfide linkages; and (5) to 2 proteins have been shown to function coordinately participate in insulin degradation [2] . in most cases, but sometimes they are found sepa­ Recently we synthesized a spin-labeled quaternary ammonium Trately in different organelles and tiss ues [3] . The widespread dis­ salt (quat) that serves as a free radical substrate for thioredoxin tribution of these 2 thioproteins, together with the observation reductase. We have shown that this enzyme acts as a free radical that they can exist separately, suggests a number of metabolic scavenger at the outer membrane surface of guinea pig and human functions. To date 5 independent functions have been described skin [4-6] and on the surface of cultured human keratinocytes for these electron-transfer systems: (1) thioredoxin/thioredoxin [7] . The enzyme can be purified from plasma membranes ofhu­ reductase as electron donors to the ribonucleotide reductases of man keratinocyte cultures by affinity column chromatography. bacteria, yeast, mammalian cell s, and plants; (2) to function as The recognition that thioredoxin reductase can reduce radicals at antioxidants in the reduction of methionine sulfoxide back to the membrane surface ofkeratinocytes documents a new function methionine, where they probabl y perform a protective role by for this fl avin-adenine dinucleotide containing thioprotein that utilizes NADPH as electron donor. The enzyme is ideally suited Manu scri pt received March 14, 1986; accep ted for publica tion June 17, for this purpose because both the thioprotein- and the fl avopro­ 1986. tein-active sites function as independent reductases [8,9]. The dual Supported by a grant from the National Institutes of Hea lth (AM 18101) function can catalyze the reduction of oxygen radicals to peroxide and by grants from th e Mayo and J. W. Kieckhefer Foundations. ions which are then reduced to water. · Part of this work was presented at the Joint International Meetin g of The spin-labeled substrate that we used to assay membrane­ the European Society for Dermatological Resea rch an d The Society for bound thioredoxin reductase has many advantages over super­ Investigative Dermatology, Inc., Geneva, Switzerland, June 22-26, 1986. oxide anion radicals . It is an excellent surfactant and reacts only Reprint requests to: Karin U. Schallreuter, M.D., Universi ty of Min­ at the outer membrane surface [4] . There it is reduced via its neso ta , P.O. Box 100, Navarre, Minnesota 55392. Abbreviations: hydroxylamine to a stable secondary amine product [5]. Hence, DNCB: dillitrochl orobenzene its reduction by the enzyme has identical stoichiometry to that EGF: epidermal growth factor expected for superoxide anion radical [5] . Confirmation that thio­ EPR: electron paramagnetic resonance spectroscopy redoxin reductase is active at the outer surface of cultured human PCMB: parachloromercuribenzoa te keratinocytes indica tes that this enzyme may function as a first quat: quatern ary am monium co mpound .line of defense against radical mediated cell damage.

0022-202X/86/S03.50 Copyright © 1986 by The Society for Investi ga tive Dermatology, In c.

728 VOL. 87, N O.6 DEC EMBER 1986 T HIOREDO X IN REDUCTASE IN HUMAN KERATINO CYTES 729

W e studied the in vivo and in vitro properties of this membrane­ bound free radical reductase in ce ll cultures from neonatal and adult human keratinocytes of healthy donors. In addition we examined thioredoxin reductase activity of cultured keratinocytes derived from vitiliginous and normal pigmented skin of 1 donor. Ker a tinocyte cultures have been grown in serum-free medium containing different Ca + + conce ntrations in the presence and ab­ sence of growth factors.

MATERIALS AND METHODS Chemicals Aceta m ido-2, 2, 6, 6-tetrameth yl pi perid ine-N-oxyl benzyl dimethyl ammonium bromide (spin labeled quat) was syn­ thesized by reacting 4-bromoacetamido-2,2,6,6-tetramethylpi­ peridine-~-oxyl with dimethyl benzylamine [4,5]. All other re­ agents used in this study were obtai ne·d from Sigma C hemica l Co. (St. Louis, Missouri) . Fetal bovine serum was purchased from K . C. Biologicals (Lenexa, Kansas). Keratinocyte Cell Cultures N eonatal and adult keratinocytes wer e isolated from fores kin and breast skin, respectively, ac­ cording to procedures previously described in deta il [10] . Briefly, specimens of skin were trypsinized overnight at 4°C and the epidermis was separated from dermis. Primary cultures of ke­ ratinocytes were initiated by seeding disaggregated and washed n cells at approximately 5 X 103 cells/cm 2 into 75-cm2 tissue culture flasks. 60 Serum-free cultures were established utilizin g MCDB 153 basa l medium containing supplements and growth factors, epidermal growth factor (EGF) 10 ng/ml, insulin 5 J.L g/ml, and bovine pi­ 50 tuitary extract 25 J.L g/ml. Medium ca lcium conce ntration was 0.1 mM. Cultures routinely displayed population doubling times of appr oximately 24 h and maintained a monola yer, undiffe rentiated morphology. In the presence of growth factor-containing me­ 40 w dium, increased ca lcium concentration of 2 mM ca used cells to 0 cluster more tightly within coloni es, but the monolayer, undif­ :::> ferentiated morphology persisted. However, removal of growth I- 30 -.J factors induced keratinocyte cultures to stratify and differentiate 0... as previously reported [10]. Confluent cultures of keratinocytes ~ were obtained by the addition of selected amino acids to the

band, bound at the top of the column, was eluted with 10 mM form stable surface complexes with spin-labeled quat, in contrast NADP+ at pH 7.2 anJ stored at -70°C prior to the enzyme to keratinocyte cell cultures or skin punch biopsies. assays. In Vivo Human Skin Assay A skin punch biopsy 3 mm 'in Enzyme Assays for Thioredoxin Reductase In Vivo and In diameter was performed under local anesthesia and the specimen Vitro The flavoprotein isolated from the affinity chromotog­ was placed in a 1 % (w/v) solution of spin-labeled quat for 7 min. raphy column was assayed by using the insulin disulfide reduction The residual spin label was washed from the sample with isotonic method [2]. Its activity was compared with pure thioredoxin saline solution for 2 min and then assayed directly in EPR quartz reductase from Escherichia coli provided by our colleague, Dr. cells. Since the enzyme activity was a direct function of the surface F. K. Gleason. area of the biopsy, the specific activity of the enzyme was deter­ Experiments with purified thioredoxin reductase from human mined per 3-mm punch biopsy per lO-min reaction [4,5]. keratinocyte membranes and from E. coli were conducted in quartz RESULTS capillary tubes (total volume 50 ILl) in 0.1 M KH2P04 buffer, pH 7.2. The reactions were initiated by adding 50 ILl of1 mM NADPH Earlier results from our laboratory showed that the outer surface to 10 ILl thioredoxin reductase with a total protein 0.1 mg/ml of the skin contains an extremely active enzyme that reduces and 50 ILl of a 1 mM spin-labeled quat. In these experiments the oxygen radicals [4,5]. Examination of the reaction products were nitroxide radical was totally reduced since the enzyme did not consistent with a thiyl reaction mechanism due to the reduction of the nitroxide radical to a secondary amine [5 ,13]. The enzyme activity was totally inhibited in vivo and in vitro by thioenzyme inhibitors such as parachloromercuribenzoate (PCMB), and di­ nitrochlorobenzene (DNCB). Experiments with guinea pig epi­ dermis in vivo showed that this thioenzyme was inhibited by its own reaction products in closed systems (i .e., 3-mm punch bi­ opsy) but that this inhibition was not observed in open systems (i .e. , after epicutaneous application on guinea pig skin before the biopsy) . Two representative thiocnzymes-glutathione reductase and thioredoxin reductase-were tested. Only thioredoxin re­ ductase actively reduced the spin label. Since thioredoxin reduc­ tase is known to be inhibited by its own product, NADP+, we tested the enzyme for specific inhibition by 1 mM NADP+ and found significant reduction in activity. Figure 1a shows the reduction of the spin-labeled quat by thio­ redoxin reductase from E. coli and Fig Ib illustrates the rate of this reaction in the presence and absence of the enzyme's natural electron acceptor thioredoxin. Next we examined cell cultures of human keratinocytes for enzyme activity. Figure 2a shows a typical bioassay with differ­ entiated keratinocytes from normal adult skin and Fig 2b shows inhibition of the activity by PCMB and by fetal bovine serum. At the present time the identity of the inhibitor(s) in the serum n remains unknown. NADP+ (1 mM) was a very potent product inhibitor of the enzyme activity at the surface of keratinocytes (Fig 3) . These results encouraged us to purify thioredoxin re-

50

40 • 40

Q) "0 :::J 30 +- W 30 0 0. :J E r <{ 20 :i 0.. 20 «~ 10 10

00 10 20 30 40 50 0 0 5 10 15 20 b Time (min) TIME (min) Figure 2. a, Reduction of spin-labeled quat (50 ILmol) by differentiated adult human keratinocytes cultured in 0.1 mM Ca + +. b, The rates of Figure 3. Effect of 1 mM NADP + on the rate of reduction of spin­ reduction of spin-labeled quat (50 p.mol) on adult human differentiated labeled quat on the surface of differentiated adult human keratinocytes keratinocytes cultured in 0.1 mM Ca + + (e-e), keratinocytes prein­ cultured in 2 mM Ca + + without NADP + (0-0) and with NADP + cubated with 50 JLmol of PCMB (A-A), and keratinocytes in the pres­ (6-6). Experimental conditions as described in "Materials and ence of 20% fetal bovine serum (0-0). Methods." VOL. 87. N O. (, DECEMBER 1986 THIO REDOXIN REDUCT ASE IN HUMAN KERATINOCYTES 731 ductase from plasma membranes of human keratinocytes. The id e ntity of the enzyme was confirmed by insulin disulfide re­ duc tion and by reduction of the spin-labeled quat substrate [4,5]. 60 However sin ce the enzyme concentrati on was low , from the small qua ntity of keratinocytes used for purifica tion, the time curve for the reaction was considerably slower than that found in E. co /i . B ecause thioredoxin reductase is known to be hea t resistant, we determined enzyme stability by exposing keratinocytes to dif fe rent temperatures for 10 min. Enzyme activity persisted at 70°C . Inactivation occurred at 75°C. Studies on neonatal and adult human keratinocytes demon­ strat ed reductase acti vity, although adult ce ll s yielded approxi­ a. m a tely twice the specific activity of neonatal ce lJ s. Because Ca + + E is known to control significantly epithelial proliferation and dif­

en z yme activity for cells grown in the presence of growth fa ctors OL-__~ ____~ ____~ ____~ __~L- __~~ cO ITlpared with cells cultured in their absence (Table I) . o 10 20 30 40 50 60 A survey of the enzyme ac ti vity from skin biopsies of patients Time (min) w ith vitiligo, without any other systemic disorders, showed that d e pig mented skin had an approximately 30% loss of the nitroxide Figure 4. Rate of reducti on of spin-labeled qu at (I %) by depi gmented radical reduction rate compared with normal pigmented skin (studies sk1l1 111 human untreated vitiligo O. M.) (0 - 0 ) and the rate of redu ction o n 11 cases) [14] . The in vivo res ults using 3-mm punch biopsies in pigmented skin (.6- .6). Assays were determined on 3-mm punch fro m 1 donor a.M .) are presented in Fi g 4. Based on these results, bIOpSIes as described in " Material s and Meth ods." keratinocyte cell cultures were es tablished from vitiliginous and no rmal epidermis isolated by the sucti on blister technique. We grew these cells at 0.1 mM and 2.0 mM Ca + + conce ntrations, ca ll y. Reactive radical s, O H ' and 0;, ca n be removed by small and in the presence and absence of growth fa ctors as described radical quenching molecules like vitamins A, C , -a nd E and by in M ateria/s and Meth ods. Surprisin gly, we found no signifi ca nt reduced glutathione. difference in thioredoxin reductase acti vity between vitiliginous Our spin-labeling experiments with human keratinocytes re­ and normal keratinocytes cultured at O. 1 mM Ca + + w ithout growth vealed that thioredoxin reductase is an extremely active free rad­ fact o rs (T able II). Ho wever, even though these cells had specifi c ical reducing enzyme on the plasma membrane. These results acti vities similar to control cultures of normal adult keratinocytes, es tablish a new fun ction for this electron transfer system that may cells grown at 2 mM Ca + + showed a signifi ca nt reduction in the be capable of removing radicals generated by UV radiation, burn­ activity compared with corres ponding control cultures. associated superoxide radical produced fro m xanthine oxidase, and from invading leukocytes that are res ponding to microor­ DISC USSION ganisms on the skin. Also this enzyme could provide a defense O x y gen radicals which ca n be generated enzymati ca ll y, photo­ mechanism that prevents membrane destruction through lipid ch e mically, or radiochemica lly have been implica ted in a number peroxidati on, interactions of free radicals with DNA ca usin g tu­ of disruptive biologic processes including lipid peroxidation, al­ mor induction and promotion, and radical additions to protein te r a tions of enzyme activity, cleavage of DNA, polymeriza tion receptor sites to produce antigens which stimulate the immune of p olysaccharides, and cell mortality [15] . It has been es tablished response. tha t these radi ca ls act as important mediators in the toxicity o f The experiments on cell cultures of human keratinocytes re­ ch e mical substances [161. Also, compounds that generate reacti ve ported herein document that thioredoxin reductase functions as oxy gen species have been shown to act as tumor promoters. a cell surface free radical reducing sys tem . T he thiol-acti ve site T h e refore multiple defense mechanisms have evolved in ce ll s to of this enzyme was confirmed by inhibition of reductase ac ti vity cope with oxidant stress. Free radical scavengers such as super­ with PC MB. The hea t stability and the product inhibitory effect oxide dismutase, catalase, and other peroxid ases (e.g., glutathione o fNADP + show n for the reductase on keratinocyte cultures and peroxidase) remove oxygen and superoxide radi ca ls enzymati- punch biopsies suggests that membrane-bound thioredoxin re-

Table I. Specifi c Activities for Thioredo xin Reductase at the Surface of Human Keratinocytes C ultured under Different Table II. Specific Activities for Thioredoxin Reductase at the C ulture Conditions Surfa ce o f Differentiated Adult Human Kcratinocytes from a Vitiligo Donor a .M.) Ratesl10 Min in Rates/lO Min in Differentiated Undifferentiated Rates /IO Min Medium Keratinocytes Keratinocytes in Di fferentiated Cell Type Ca++ [GF(-)]" IG F( +)I" Skin Site Medium Ca ++ Keratin ocytcs rGF( -) I" Adult 0.1 mM 85 12 Norm al 0. 1 111M 66 N eonatal 40 5 Vitiligin ous 70

Adult 2 111M 40 6 Norm al 2 mM 8 N eonatal 17 7 Vitiliginous 6

"GF = Growth r., ctors present ( +) or absent (-) in th e cultural mcdium . "G F = Grow th r.,ctorS abscnt (- ) in the culture medium . 732 SCHALLREUTER, PITTELKOW, AN D WOOD T H E JO URNAL OF INVESTIGATIVE DERMATOLOGY

ductase is the maj or free rad ica l reducin g enzy me. Cells of human 2. Holmgre. A: T hioredoxin. Annu Rev Biochem 54:237-271, 1985 adult and neonatal epidermis cultured under a va ri ety of condi­ 3. Holmgren A, Luthman M : Tissue distribution and subcellular lo­ ti ons showed that approximatel y double the enzyme activity was ca liza tion of bovine thioredoxin determined by radioimmunoas­ fo und in adult differentiated cell cultures grown at 0. 1 mM Ca + + say. Biochemistry 17:4071-4077, 1978 and in the absence of growth fac tors compared with cultures 4. Schallreuter KU, Wood JM: The allergenicity of complex cations. grown at 2 mM Ca + + . Undiffe rentiated, rapidly growing adult Biochem Biophys Res Commun 135:221-227, 1986 and neonatal keratinocytes examined under standard conditions 5. Schall reuter KU, Wood JM: T he role of thioredoxin reductase in yielded similar results but the specific activities were considerably the reduction offree radicals at the surface of the epidermis. Biochem lower. It is possible that thioredoxin reductase of rapidly dividing Biophys Res Commun 136:630- 637, 1986 cells is requi red to reduce ribonucleotide reductase that in turn 6. Schallreuter KU, Wood JM: Thioredoxin reductase for free radical provides the necessary deoxyribonucleotides fo r D NA synthesis. reducti on at the surface of the skin (a bstr). J In vest Dermatol T his explanati on would account for decreased enzyme ac tivity at 86:505, 1986 the cell surface. 7. Schall reuter KS, Pittclkow M, Wood JM: Free radi cal reduction by O ur study of the enzyme on keratinocytes from a patient with thioredoxin reductase at the surfa ce of human keratinocytes (a bstr). vi tili go provided the most interes ting res ults. We could not find J Invest Dermatol 87:164, 1986 any signi fica nt difference in re ducti on rates between the vitili­ 8. O 'Donnell ME, William s C H Jr: Proton stoichiometry in the re­ gino us and normal cultured cell s, although the in vi vo biopsy duction of the FAD and disulfide of Escherichia co li thioredoxin assay demonstrated markedl y reduced enzyme activity in vitili­ reductase. J Bioi C hem 258 :1 3795-13803, 1983 ginous skin . 9. O'Donnell ME, Willi ams C H Jr: Reaction of both active site thiols T his resul t indica tes that vitiligo might be environmentally of reduced thioredoxin reductase with n-ethylmaleimide. Bio­ induced and not due to irreversibly inhibited expression of re­ chemistry 24:7607-7621 , 1985 10. Wille JJ , PittcJkow Mi~ , Shi pley CD, Scott RE: Integrated control ductase. Clea rl y metaboli c alterations regul ated at the enzy me of growth and di ffe rentiation of normal human keratinocytes cul­ level appear to be responsibl e fo r the diminished enzyme ac ti vity tured in serum-free medium. C lonal analysis, growth kineti cs and observed in this case of vitiligo. However, sin ce we demonstrated cell cycl e studies . J Cell Phys iol 121:31-44, 1984 that reductase ac tivity is dependent on medium Ca + + concen­ 11. Pittelkow MIl., Wille JJ , Scott RE: T wo fun ctionall y distinct classes tration, we bel ieve that Ca + + regul ati on may play an important o f grow th arres t states in human keratinocytes that regulate clo­ role in the eti ology of vitiligo. Presently, we do not understand nogenic potential. J In vest D ermatol 86:4:41 0-418, 1986 the Ca + + dependent modulati on of thio redoxin reductase activ­ 12. Pittclkow MIl., Z insmeister A, Steinmuller D: Donor-recipient ity. Further studies on the effect of Ca + + and other cations are m atching w ith the mixed lymphocyte-epidermal cell reacti on, Re­ under investigation in our laboratory. cent Adva nces in Bone Marrow T ranspl antati on. Edited by R. T. Gale. N ew York, Al an R Liss , 1983, pp 227-241 13. Melhorn R, Packer L: Membrane surface potential measurements We are illdebted to Dr. F. K. C leaso ll fo r providillg I/S with pllrlfied thioredoxin w ith amphiphilic spin labels. Methods Enzymol 56:515, 1984 oll d tl/iOl'edoxill redll ctaseji-om E. coli . Steve M ic/II/rski (md Normo Nelle S"'OIl­ 14. Wood JM, Schallreuter KU, Hordinsky M : Thio redoxin reductase SO li are ockll ow ledg ed for excellell t tec/II/ ica l ossistoll ce. D . Bischke typed th is in different skin types (I- VI) (Fitzpatrick cl assifica ti on) and in m Ol II/ script . vitiligo skin (abstr) . J Inves t Dermatol 86:514, 1986 15. Bors W, Sa ran M, Czapski G: Biological and C linica l Aspects of Supcroxide and Superoxide D ismutase, C hapter 6. Edited by WH REFEREN CES Bannister, JW Bann ister. Amsterdam, Elsevier/ N orth Holl and, '1. Roze ll B, Hansson H A, Luthman M , Holmgren A: Immunohisto­ 1980, pp 38- 44 chemica l localiza tion o f thioredoxin and thioredoxin reductase in 16. Fridovich I: Oxygen and Li ving Processes, C hapter 5. Edited by adult rats. Eur J Cell Bioi 38:79-86, 1985 DC Gilbert. Springer-Verl ag, 1981, pp 250-284