Oxidative Stress Resulting from Ultraviolet A Irradiation of Human Skin Fibroblasts Leads to a Heme Oxygenase-dependent Increase in Ferritin*

Heme oxygenase-1 mRNA levels increase following exposure of many mammalian cell lines to oxidative stress such as ultraviolet A (UVA) irradiation. Here we demonstrate a 4-fold increase in microsomal heme ox- ygenase activity and a 40% decrease in microsomal heme content 14 h after treatment of human skin fibroblasts (FEKJ with 260 kJ m-2 of UVA radiation. Paralleling this was a 2-fold increase in ferritin levels that was sustained for at least 46 h after UVA irradi- ation. Treatment of fibroblasts with the iron chelating agent desferrioxamine, after the UVA-dependent induction of heme oxygenase, prevented the increase in ferritin levels. Treatment of fibroblasts with Sn-pro-toporphyrin IX (an inhibitor of heme oxygenase) also prevented the effect of UVA radiation on ferritin levels. Thus we conclude that the effect of UVA radiation on ferritin levels is via the heme oxygenase-dependent release of iron from endogenous heme sources. We propose that the increase in ferritin that follows UVA irradiation would decrease intracellular free iron such that iron-catalyzed free radical reactions would be restricted during periods of subsequent oxidative stress.


Oxidative Stress Resulting from Ultraviolet A Irradiation of Human Skin Fibroblasts Leads to a Heme Oxygenase-dependent Increase in Ferritin*
(Received for publication, September 11, 1992, and in revised form, January 18, 1993) Glenn F. Vile$ and Rex M. Tyrrell Paralleling this was a 2-fold increase in ferritin levels that was sustained for at least 46 h after UVA irradiation. Treatment of fibroblasts with the iron chelating agent desferrioxamine, after the UVA-dependent induction of heme oxygenase, prevented the increase in ferritin levels. Treatment of fibroblasts with Sn-protoporphyrin IX (an inhibitor of heme oxygenase) also prevented the effect of UVA radiation on ferritin levels. Thus we conclude that the effect of UVA radiation on ferritin levels is via the heme oxygenase-dependent release of iron from endogenous heme sources.
We propose that the increase in ferritin that follows UVA irradiation would decrease intracellular free iron such that iron-catalyzed free radical reactions would be restricted during periods of subsequent oxidative stress.
Many procaryotic and eucaryotic cells synthesize specific proteins in response to oxidative stress. In some cases these stress proteins have been shown to have an antioxidant role, e.g. treatment of Escherichia coli with hydrogen peroxide results in expression of at least 30 proteins including catalase (1). The protein most consistently activated by oxidative stress in a wide variety of eucaryotic cells is the heme degrading enzyme, heme oxygenase-1 (2)(3)(4)(5). UVA radiation (320-380 nm), hydrogen peroxide, and glutathione depleting compounds all enhance heme oxygenase-1 mRNA synthesis (6) and result in accumulation of heme oxygenase-1 mRNA in cultured human skin fibroblasts (FEK,) and other mammalian cell lines (7).
UVA irradiation of biological molecules gives rise to superoxide and hydrogen peroxide (8-lo), species that may be involved not only in cell death but also in the carcinogenic effects of UVA irradiation (11). There is good evidence that the biological damage attributed to superoxide and hydrogen peroxide is dependent on the presence of iron (12)(13)(14). It has been proposed that there is a small intracellular pool of free ~~ * This study was supported in part by the Swiss League Against Cancer and The Swiss National Science Foundation (31-30880-91). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertkement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. iron that can react with hydrogen peroxide and superoxide, giving rise to the very reactive hydroxyl radical ('OH) via Reactions 1 and 2, and it is the hydroxyl radical that is the initiator of biological damage (15)(16)(17).

REACTIONS 1 AND 2
Intracellularly most of the iron that is not metabolized is sequestered in ferritin as a crystalline core of ferric (Fe3+) ions (18). To catalyze oxidative reactions, the iron must first be released from the core. Although superoxide is able to do this, it is very inefficient (19). Thus ferritin is able to restrict the availability of iron to participate in Reaction 1.
It has been shown that a hemin-dependent increase in heme oxygenase protein synthesis activates ferritin mRNA translation in rat fibroblasts (20). Whether this led to an increase in heme oxygenase activity or ferritin levels was not examined. Elevated levels of newly synthesized ferritin would result in an enhancement of cellular iron sequestering capacity that may confer increased resistance to oxidative stress. This study tests the hypothesis that levels of ferritin in human skin fibroblasts are effected by UVA irradiation and that this occurs via heme oxygenase.

MATERIALS AND METHODS
All biochemicals were from Sigma except where indicated. Cell Culture-Monolayers of the normal human skin fibroblast line (FEKJ were grown to 100% confluence in 15-cm dishes over 7 days in minimum essential media supplemented with penicillin, streptomycin, glutamine, sodium carbonate, and 15% fetal calf serum. At day 7 each dish contained approximately 8 X lo6 fibroblasts. For some experiments fibroblasts were grown to 40% confluence over 3 days. Fibroblasts were passaged twice a week and used between passages 9 and 16. Cell culture materials were from Life Technologies (Paisley, Scotland), except fetal calf serum, which was from Biological Industries (Haemek, Israel).
UVA Irradiation-Fibroblasts were irradiated with 250 kJ rn-' of broad spectrum UVA light using a Uvasun 3000 lamp (Mutzhas, Munich, Germany). The UVA dose was measured using an IL 1700 radiometer (International Light, Newbury, USA). Irradiation was done at 25 "C. Prior to irradiation, media were removed and retained, and the fibroblasts were covered with Ca2+/Mg2+ (0.01% each) enriched phosphate-buffered saline (PBS)' as described previously (3). After irradiation the original media were added back to the fibroblasts. Control fibroblasts were treated in the same manner except that they were not irradiated.
Hemin Treatment-A stock solution of hemin (1 mM) was prepared in KOH (8 mM) and phosphate buffer (100 mM, pH 7.4). After removal of the media fibroblasts were treated with hemin (4 P M ) in PBS for 1 h. After treatment the fibroblasts were washed thoroughly and the The abbreviations used are: PBS, phosphate-buffered saline; MOPS, 4-morpholinepropanesulfonic acid. Desferrioxamine Treatment of Fibroblasts-To bind low molecular weight intracellular iron fibroblasts were treated with desferrioxamine (Desferal) (500 p~) (Ciba Geigy, Basel, Switzerland) in PBS at 37 "C after removal and storage of the media. After incubation for 1.5 h the desferrioxamine was removed and fibroblasts were washed thoroughly with PBS before adding back the original media.
Heme Oxygenase-1 mRNA-Total RNA was isolated by the guanidinium thiocyanate-phenol-chloroform method 3 h after irradiation (21). RNA (15 pg/well) was electrophoresed in a MOPS/HCHO 1.3% agarose gel (22), transferred onto a Genescreen nylon membrane (NEN Research Products, Regensdorf, Switzerland), and hybridized with the 1000 base pair EcoRI fragment of the human heme oxygenase cDNA clone 2/10 (3). After autoradiography blots were rehybridized with the PstI fragment (1300 base pairs) of rat glyceraldehyde-3phosphate dehydrogenase cDNA. The glyceraldehyde-3-phosphate dehydrogenase RNA signal was used as an internal control for the loading error between samples.
Fibroblast Extracts-Fibroblasts (8 X 10') were washed thoroughly with ice-cold PBS, harvested with a rubber policeman and homogenized with a Potter Elvehjem homogenizer (Bellco, Fetham, United Kingdom) at 4 "C. Cell debris was removed by centrifugation at 5000 X g, and an aliquot of supernatant was retained for ferritin analysis. The remaining supernatant was spun at 15,000 X g to remove mitochondria, and the new supernatant was centrifuged at 105,000 X g for 60 min. The resulting microsomal pellet was resuspended in phosphate buffer (100 mM, pH 7.4). All centrifugation steps were performed at 4 "C. The protein content of extracts was determined using the method of Bradford (23) standardized with bovine serum albumin.
Heme Oxygenase Determination-The method of Shibihara et al. (24) was used to determine heme oxygenase activity. Microsomes (100-200 pg of protein) were incubated with hemin (20 p~) , bovine serum albumin (0.064%), NADPH (100 p~) , and crude biliverdin reductase extract (0.29 mg ml-'), prepared according to the method of Tenhunen et al. (25). Reactions took place in phosphate buffer (100 mM, pH 7.4), and samples were gently mixed in the dark at 37 "C for 30 min. The reaction was stopped by placing on ice, tubes were centrifuged at 5000 X g for 1 min at 4 "C, and the absorbance of bilirubin at 465 nm was measured against a base-line absorbance at 520 nm (e4% = 40,000 M" cm-'; Ref. 26).
Ferritin Content-The ferritin assays were performed with a polyclonal enyme-linked immunosorbent assay kit (Boehringer, Mannheim, Germany). Supernatants (10-20 pg of protein) from the 5000 X g centrifugation step after homogenization of the fibroblasts were analyzed for ferritin according to the procedure supplied with the kit.
Heme Content-Mitochondria1 and microsomal pellets, prepared from fibroblasts as described above, were resuspended in 1 ml of concentrated formic acid (Fluka, Buchs, Switzerland), and the heme content of the solution was measured at 398 nm (27). The concentration of heme was calculated from a standard curve constructed from treatment of cytochrome c with concentrated formic acid.

RESULTS AND DISCUSSION
This study shows that treatment of human skin fibroblasts (FEK4) with 250 kJ m-' of UVA irradiation elevated heme oxygenase activity 4-fold, 14 h after irradiation (Fig. 1). The increase in heme oxygenase activity after UVA irradiation we show here extends previous results from this laboratory that have shown induction of heme oxygenase protein levels and increased rate of heme oxygenase-1 RNA accumulation after UVA irradiation of FEK, fibroblasts (3,6). We now show that the UVA-dependent increase in heme oxygenase activity of after UVA irradiation and were still maximal 46 h after irradiation (Table I). Heme oxygenase activity had returned to control levels 46 h after irradiation (Fig. 1). The UVA dose used (250 kJ m-') was equivalent to less than 30 min of exposure to a typical tanning lamp. Thus the UVA-dependent increase in heme oxygenase activity and ferritin levels we show here represent a response to a physiological level of oxidative stress. Heme oxygenase-1 mRNA levels were increased 13 f 5- fold (mean f S.D. of 10 determinations) following irradiation of fibroblasts with 250 kJ m-2 of UVA 3 days after seeding (Fig. 2). Fibroblasts irradiated 7 days after seeding (100% confluent) showed an even greater increase in heme oxygenase-1 mRNA (Fig. 2). Previous work from this laboratory has shown accumulation of heme oxygenase-1 mRNA in fibroblasts irradiated 2-4 days after seeding (30-50% confluent) . For all investigations in the present study we used fibroblasts 7 days after seeding because they were not undergoing so rapid a rate of growth as that seen in 3-day fibroblasts (data not shown) and therefore more closely represent the growth characteristics of cells in uiuo. The apparent discrep-

HO -1
that is degraded upon UVA induction of heme oxygenase. Intracellular free iron has been shown to be involved in the UVA-dependent induction of heme oxygenase-1 mRNA accumulation (34). Therefore, in order to test the effect of iron desferrioxamine a t a time when the rate of heme oxygenaseoxygenase-1 mRNA content of FEK, fibroblasts. Fibroblasts   FIG. 2. Effect of UVA irradiation (250 kJ m-2) on heme RNA synthesis had reached a maximum, i.e. 1 h after UVA were irradiated either 3 or 7 days after seeding and probed with heme irradiation (6). Chelation of iron at this time blocked the oxygenase-1 mRNA then glyceraldehyde-3-phosphate dehydrogenase increase in ferritin levels attributed to the release O f iron by mRNA. The autoradiograph shown is of a typical experiment. the induced heme oxygenase (Table I). This implies a role for free iron in the regulation of ferritin levels. For the purposes  ancy between the -fold increase in mRNA and the -fold increase in enzyme activity after UVA irradiation is likely to be because the total cellular heme oxygenase activity consists of the inducible heme oxygenase-1 isozyme and the constitutive heme oxygenase-2 isozyme (28). In the rat liver there is a 2:l ratio of heme oxygenase-2 to heme oxygenase-1 activity (29), and in rat testes almost all the heme oxygenase activity has been attributed to heme oxygenase-2 (30). Therefore the -fold increase in heme oxygenase-1 activity of FEK4 fibroblasts following UVA irradiation is almost certainly greater than we show here due to the contribution (currently unknown) of heme oxygenase-2 activity to the basal levels.
T o examine the effect of increased heme oxygenase activity on cellular heme, we measured total heme content of microsomes (endoplasmic reticulum) and mitochondria prepared from UVA-irradiated fibroblasts. The heme content of microsomes of irradiated fibroblasts was decreased to approximately 60% of the level in non-irradiated fibroblasts (Table  11), at the time when the UVA-dependent increase in heme oxygenase activity was maximal, i.e. 14 h after UVA irradiation (Fig. 1). The heme content of mitochondria from irradiated fibroblasts was unchanged 14 h after irradiation (Table  11). The decrease in microsomal heme content was transitory, since the heme content of microsomes increased from 60% of the level in non-irradiated fibroblasts 14 h after irradiation to 90% of the level in non-irradiated fibroblasts 22 h after irradiation (Table 11). We did not examine the type of microsomal heme that was decreased after the UVA-dependent induction of heme oxygenase activity. In addition to cytochrome P450 and bs (31), there may also be a small amount of free heme associated with microsomes (32). Cytochrome P450 has been shown to be degraded to biliverdin by a reconstituted heme oxygenase-1 system (33) and liver microsomes prepared from rats treated with cobalt show an increase in heme oxygenase activity and a decrease in cytochrome Pd50 (26). Cytochrome b5 levels were unchanged (26). The unsuitability of cytochrome bs as a substrate for heme oxygenase-2 has been demonstrated in a reconstituted system (30). The decrease in heme content we show in microsomes with enhanced heme oxygenase activity is much greater than could be ac-enhancement on ferritin protein levels is modest compared to the effect of free iron on translation (36). It has also been proposed that the iron needs to be associated with porphyrin in order to activate ferritin mRNA translation (38). However our results support the alternative proposal that the principal effect of heme on ferritin levels is via free iron (20, 36).
The possibility that UVA irradiation was stimulating ferritin levels by a heme oxygenase-independent mechanism was tested by treating UVA-irradiated fibroblasts with Sn-protoporphyrin IX. This porphyrin irreversibly inhibits heme oxygenase. Sn-protoporphyrin IX treatment of fibroblasts inhibited heme oxygenase activity almost completely (Table I).
Under these conditions UVA irradiation had no effect on ferritin levels. This rules out the possibility that iron, released directly from hemes or ferritin by UVA irradiation (39), was stimulating ferritin synthesis. Thus it can be concluded that the effect of UVA on ferritin levels is via heme oxygenase. Treatment of fibroblasts with PBS for the same length of time as fibroblasts were treated with Sn-protoporphyrin IX (2 h) did not effect the UVA-dependent increase in heme oxygenase activity and ferritin levels. Treatment of nonirradiated fibroblasts with Sn-protoporphyrin IX for 2 h inhibited heme oxygenase activity but did not affect basal ferritin levels. T o demonstrate that the relationship between heme oxygenase and ferritin was not a specific effect of oxidative stress, we used hemin instead of UVA irradiation as an inducer of heme oxygenase. Fibroblasts treated with hemin showed an increase in heme oxygenase activity of 6-fold 14 h after treatment (from 0. This study has demonstrated that the inducibility of heme oxygenase by oxidative stress plays an important role in the regulation of the major intracellular iron-binding protein, ferritin. It has been hypothesized that an increase in heme oxygenase activity may lead to an increase in the antioxidant potential of cells, thereby enhancing cell survival under oxidative stress (34,40). We now show that a heme oxygenase-Induction of Ferritin by UVA Occurs via Heme Oxygenase 14681 dependent increase in antioxidant potential may actually be mediated by ferritin. We propose that the increased levels of ferritin that result from the UVA-dependent induction of heme oxygenase will further decrease intracellular free iron levels and therefore may limit iron-catalyzed oxidative reactions that would occur during subsequent periods of oxidative stress.