Professor Dr Helmut Sies, Institut für Physiologische Chemie I, Heinrich-Heine-Universität, Postfach 101007, D-40001 Düsseldorf, Germany. Fax: + 49-211-811-3029; Tel: + 49-211-811-2707; E-mail: email@example.com
Ultraviolet A (UVA; 320–400 nm) radiation in human skin fibroblasts induces a pattern of mitogen-activated protein kinase (MAPK) activation consisting of a rapid and transient induction of p38 and c-Jun-N-terminal kinase (JNK) activity but not extracellular signal-regulated kinases (ERK). UVA activation of p38 can be inhibited by the singlet oxygen (1O2) quenchers azide and imidazole, but not by the hydroxyl radical scavengers mannitol or dimethylsulfoxide, pointing to the involvement of 1O2. The same effect has been shown for JNK. Like UVA, 1O2 generated intracellularly upon photoexcitation of Rose Bengal activates p38 and JNK but not ERK. p38 and JNK activation was also elicited by chemiexcitation for the intracellular generation of 1O2 by the lipophilic 1,4-endoperoxide of N,N′-di(2,3-dihydroxypropyl)-1,4-naphthalene dipropionamide. In contrast, extracellular generation of 1O2, by irradiation of Rose Bengal immobilized on agarose beads or by chemiexcitation employing the hydrophilic 1,4-endoperoxide of disodium 3,3′-(1,4-naphthylidene) dipropionate, was ineffective in activating p38 or JNK. These data suggest that the activation of p38 and JNK by 1O2 occurs only when the electronically excited molecule is generated intracellularly.
Electronically excited molecular oxygen, singlet oxygen (1O2), may be generated photochemically in the skin or by chemiexcitation in the inflammatory response in vivo[1,2]. 1O2 mimics the induction of the expression of genes in human dermal fibroblasts by ultraviolet A (UVA; 320–400 nm) and has been found to induce the expression of heme oxygenase-1 (HO-1) , matrix metalloproteinase-1 (MMP-1) , the interleukins IL-1α/β and IL-6  and intercellular adhesion molecule 1 (ICAM-1)  in human skin cells. These effects of UVA on gene expression may be crucial in photocarcinogenesis and photoaging ; e.g. the degradation of extracellular matrix by MMPs will favor tumor invasion/metastasis as well as wrinkle formation. UVA irradiation of human skin cells has been known to lead to the activation of transcription factors AP-1 (activator protein-1), NF-κB (nuclear factor-κB)  and AP-2 [6,10]. In addition, NF-κB [11,12] and AP-2 [6,10] have been shown to be activated by 1O2, in the latter case also mediating the activation by UVA.
Activation of AP-1-dependent gene expression is mediated by the activation of mitogen-activated protein kinases (MAPKs), a family of proline-directed Ser/Thr kinases activated by dual phosphorylation. Three MAPK groups have been intensively studied: the extracellular signal-regulated kinases (ERKs) responsive to mitogens such as growth factors; the c-Jun N-terminal kinases (JNKs) and p38 MAPKs, which are activated by proinflammatory cytokines and environmental stress [13,14]. Activated MAPKs phosphorylate and activate transcription factors such as ternary complex factors (TCFs, for ERK and p38), c-Jun (for JNK), or ATF-2 (for JNK and p38), leading to expression of c-Jun and c-Fos and finally inducing AP-1-dependent transcription [13,14]. A potent inducer of AP-1-dependent transcription is ultraviolet C (UVC) radiation (< 280 nm), which strongly induces activation of ERK, JNK and p38 [15–17]. Ultraviolet B (UVB) (280–320 nm) radiation activates JNKs [18,19], weakly enhancing ERK activity . Activation of JNK by UVA has been shown to be mediated by 1O2 in human skin fibroblasts . However, p38 and ERK activity upon treatment with UVA or 1O2 has not yet been investigated. The purpose of the present work was to study the MAPK (p38-, JNK-, ERK-) activation pattern induced by UVA and 1O2 in human skin fibroblasts.
Materials and methods
Human skin fibroblasts from foreskin biopsies were cultured in Dulbecco’s modified Eagle's medium (DMEM) supplemented with 10% fetal calf serum, l-glutamine (2 mm), streptomycin (0.02 g·L−1) and penicillin (20 000 IU·L−1).
Singlet oxygen and UVA sources
Chemical generation of 1O2 was achieved by incubating the cells with disodium 3,3′-(1,4-naphthylidene) dipropionate-1,4-endoperoxide (NDPO2) or the 1,4-endoperoxide of N,N′-di(2,3-dihydroxypropyl)-1,4-naphthalene dipropionamide (DHPNO2) in serum-free medium. NDPO2 and DHPNO2 were synthesized as described previously [21–23]. Control experiments were performed with solutions of preheated NDPO2 and DHPNO2 containing the decomposition products, NDP and DHPN, respectively. 1O2 was also produced by irradiating NaCl/Pi (137 mm sodium chloride, 3 mm potassium chloride, 8 mm disodium hydrogen phosphate, 2 mm potassium dihydrogen phosphate, pH 7.4) containing 0.3 µm of Rose Bengal (RB; Sigma, St Louis, MO, USA) or RB–agarose (Molecular Probes, Eugene, OR, USA) for 10 min with a commercially available 500-W halogen lamp from a fixed distance of 66 cm. Approximately 130 µm (cumulative concentration) of 1O2 was generated during 10 min of irradiation of 0.3 µm RB as determined by the bleaching of p-nitrosodimethylaniline , and no UVA was measurable under these conditions. UVA irradiation of cells covered with NaCl/Pi was performed at intensities of 40–44 mW·cm−2 for ≈12 min (yielding a dose of 300 kJ·m−2) using a broadband UVA700 illuminator with maximal output at 360–400 nm, from Waldmann (Villingen, Germany).
The treatments applied were nontoxic as determined by microscopy and by means of the reduction of 3-[4,5-dimethylthiazol-2-yl]-2,5-diphenyl-tetrazolium bromide (MTT).
Lipophilicity and delivery to cells of singlet oxygen sources
Lipophilicity of compounds was assessed by determining their 1-octanol/NaCl/Pi partition coefficients Kp. The compounds were dissolved in either 1-octanol or NaCl/Pi (pH 7.4). After addition of the same volume of either 1-octanol or NaCl/Pi, respectively, the samples were shaken vigorously and incubated at room temperature for at least 90 min with stirring or under rotation to ensure that partition equilibrium was reached. After separation of the phases by centrifugation, the concentrations of the compounds tested in the aqueous and organic phases were determined by spectrophotometry. Naphthalene derivatives were measured at 290 nm, and RB at 549 nm in NaCl/Pi and at 564 nm in octanol. The extinction coefficients assessed for RB in water and 1-octanol are ε549, NaCl/Pi = 96 mm−1·cm−1 and ε564,octanol = 124 mm−1·cm−1, respectively. The partition coefficient was then calculated according to Kp = [compound]organic phase/[compound]aqueous phase.
To assess the delivery of naphthalene derivatives and RB to cells, skin fibroblasts were grown to near confluence in 90 mm culture dishes and incubated with 6 mL of 5 mm NDP or DHPN in NaCl/Pi for 30 min at 37 °C or incubated with 6 mL of 0.3 µm RB or RB–agarose in NaCl/Pi at room temperature in the dark. After incubation, the cells were washed rapidly in NaCl/Pi and scraped off the plate in 300 µL NaCl/Pi. The suspension was sonicated for 1 min and centrifuged twice in a microfuge to remove particulate fractions. The supernatants were analysed by reversed-phase HPLC on a reverse phase C18 column for the presence of the tested compounds, which were quantitated by comparison with the signals elicited by standard solutions. Detection of eluting compounds was by absorbance at 290 nm. The mobile phase employed was methanol/50 mm ammonium acetate (30/70, pH 7.0) for NDP, methanol/50 mm ammonium acetate (40/60, pH 7.0) for DHPN and methanol/water (90/10) for RB.
Determination of singlet oxygen
Release of 1O2 from NDPO2 and DHPNO2 was determined by measuring 1O2 monomol emission at 1270 nm with a liquid-nitrogen-cooled germanium diode (Model EO-817 L, North Coast Scientific Co., Santa Rosa, CA, USA) sensitive in the 800–1800 nm spectral region using a bandpass filter for 1270 ± 10 nm. The diode was separated from the sample cuvette by an optical chopper operating at 30 Hz.
1O2 generated intracellularly was also assessed by the bleaching of 9,10-diphenylanthracene (DPA) fluorescence. The lipophilic DPA reacts with 1O2 to form a nonfluorescent and stable endoperoxide . Cells were loaded with DPA by incubation with 5 µm DPA in NaCl/Pi for 5 min. Following removal of nonincorporated DPA and washing with NaCl/Pi, cells were treated with NDP/NDPO2, DHPN/DHPNO2, RB ± light or RB–agarose ± light. After treatment, cells were scraped off and DPA was extracted with chloroform/methanol (2 : 1). Fluorescence was measured on an LS-5 luminescence spectrometer (Perkin-Elmer, Norwalk, CT, USA) with excitation and emission wavelengths of 376 nm and 410 nm, respectively. Results are reported as means ± SD (n = 3).
After irradiation, cells grown on 30 mm dishes in serum-free medium were incubated at 37 °C for 30 min, washed with NaCl/Pi and lysed by scraping in 2 × SDS/PAGE lysis buffer [125 mm Tris, 4% (w/v) SDS, 20% (v/v) g1ycero1, 100 mm dithiothreitol, 0.2% (w/v) bromophenol blue, pH 6.8]. The lysates were heated at 95 °C for 5 min and used for SDS/PAGE or frozen until use. Samples of 8–16 µL were subjected to gel electrophoresis on 12% SDS/polyacrylamide gels and blotted onto nitrocellulose membranes (ECL nitrocellulose, Amersham). Immunodetection of phosphorylated ERK 1/2, p38 and JNK was carried out using α-phospho-MAPK and α-phospho-p38 (New England Biolabs, Schwalbach, Germany) and α-phospho-JNK (Promega, Mannheim, Germany) antibodies, respectively. Incubation with an α-rabbit secondary antibody conjugated to horseradish peroxidase (Cappel/ICN, Eschwege, Germany) was followed by chemiluminescence detection (LumiGlo, New England Biolabs). After stripping, the membrane was reprobed with α-p38 (New England Biolabs) or α-JNK2 (Santa Cruz Biotechnology, Santa Cruz, CA, USA) antibodies which served as gel loading and protein controls.
JNK and ERK activity
JNK activity was determined as in Klotz et al.. Briefly, fibroblasts were grown to near confluence in 60 mm dishes, held in serum-free medium for at least 20 h and treated as desired. After irradiation, the cells were incubated at 37 °C for 30 min, washed with NaCl/Pi at room temperature and lysed with 300 µL of ice-cold RIPA-buffer [50 mm Tris pH 8.0, 150 mm NaCl, 1 mm dithiothreitol, 0.5 mm EDTA, 1% (w/v) Nonidet P-40, 0.5% (w/v) sodium desoxycholate, 0.1% (w/v) SDS, 0.2 mm Na3VO4, 0.8 mm PhCH2SO2F, 1 µg·mL−1 aprotinin, 2 µg·mL−1 leupeptin]. The lysed cells were scraped off and briefly centrifuged at 4 °C. Supernatants were normalized for protein content (DC-Protein Assay from BioRad) and JNKs immunoprecipitated from cell lysates (100 µg of protein in a total volume of 100 µL RIPA lysis buffer) using 5 µL of an antiserum (rabbit, diluted 1/10 in distilled water), which was a generous gift from Dr P. E. Shaw (Department of Biochemistry, University of Nottingham, UK).
Immune complexes were collected with Protein A–Sepharose, and washed with both ice-cold RIPA buffer and kinase buffer (10 mm Tris/Cl pH 7.4, 150 mm NaCl, 10 mm MgCl2, 0.5 mm dithiothreitol). The kinase assay was performed as in Klotz et al., and the phosphorylated substrate glutathione S-transferase (GST)–cJun (1–79; Alexis, Grünberg, Germany) was analysed by electrophoresis on a 12% SDS/polyacrylamide gel and identified by autoradiography.
ERK activity was determined analogously but employing an anti-ERK 1/2 antibody (Upstate Biotechnology, Lake Placid, NY, USA) as the precipitating agent (1 µg·sample−1) and myelin basic protein (MBP, Sigma, München, Germany) as the substrate (1 mg−1·mL−1·assay−1) for immunoprecipitated ERK.
Differential MAPK activation by UVA
UVA irradiation of serum-starved human skin fibroblasts resulted in a rapid and transient activation of p38-MAPKs and JNK-MAPKs; in contrast, ERK 1/2 activity was not increased (Fig. 1). Activation of p38 and JNK is maximal 15–30 min after irradiation.
UVA activation of p38 may be mediated by 1O2
Regarding induction of JNK activity by UVA, photochemically formed 1O2 is a primary mediator . To test whether the same holds for p38 activation, fibroblasts were irradiated in the presence of sodium azide (20 mm), present during irradiation only, or imidazole (40 mm), present during 30 min of preincubation and during irradiation. Both these compounds diminished the UVA effects on p38 phosphorylation (Fig. 2). In contrast, the hydroxyl radical scavengers mannitol (100 mm), present during 45 min of preincubation and during irradiation, or dimethylsulfoxide (270 mm), present during irradiation, did not affect the p38 response to UVA.
Experiments were also carried out in the presence of 90% deuterium oxide (D2O) in NaCl/Pi. D2O enhances the lifetime of 1O2, 64 µs vs. 4 µs in H2O .
JNK activation by UVA and RB + light was intensified (2.6-fold and 2.1-fold, respectively) in the presence of D2O . Surprisingly, there was no such D2O effect regarding p38 phosphorylation with UVA (n = 3) or RB plus light (n = 2; data not shown).
Intracellular or extracellular generation of singlet oxygen
To generate 1O2 independent of UVA, human skin fibroblasts were treated with 1O2 generated by RB + light or by thermodecomposition of two 1O2 carriers, DHPNO2 and NDPO2(Fig. 3A). The time course of 1O2 release by DHPNO2 was similar to that observed for NDPO2, as followed by 1O2 monomol emission (Fig. 3B).
RB and DHPN are lipophilic compounds, as reflected by their octanol/buffer partition coefficients (Table 1). These compounds are found inside cells after incubation as shown by HPLC analysis of RB-treated and DHPN-treated fibroblasts (Table 1), and they serve for intracellular generation of 1O2. In contrast, NDPO2 is a more hydrophilic 1O2 carrier. The octanol/NaCl/Pi partition coefficient of NDP is at least 40-fold lower than that of DHPN. Furthermore, NDP associated with cells after incubation is barely detectable (Table 1). Thus, similar to RB coupled to agarose, NDPO2 seems to generate 1O2 only outside cells.
Table 1. 1-Octanol/NaCl/Pi (pH 7.4) partition coefficients (Kp) and delivery to cells of naphthalene derivative 1O2 carriers, RB and 9,10-diphenylanthracene.
Amount found associated with cells (nmol/106 cells)b
The bleaching of DPA has been used previously for the detection of 1O2 generated by stimulated polymorphonuclear neutrophils . DPA is lipophilic and can penetrate cells (Table 1). To test whether 1O2 generated by RB/light, RB–agarose/light, DHPNO2 or NDPO2 can reach targets inside cells, we treated cells with these 1O2 sources and the intracellular DPA bleaching was measured. Figure 4 shows that RB + light or DHPNO2 but not RB–agarose + light or NDPO2 were able to bleach DPA, demonstrating intracellular 1O2 generation.
Effect of intracellularly and extracellularly generated 1O2 on p38-MAPK and JNK-MAPK activation
1O2 generated with RB + light shows the same pattern of MAPK activation as does UVA. It activates p38 and JNKs but not the ERKs (Fig. 5). Immobilizing RB on agarose, however, leads to the loss of its p38- and JNK-activating capacity upon irradiation with light (Fig. 6). Similar to RB + light, the incubation of human skin fibroblasts with DHPNO2, but not its decomposition product DHPN, leads to an activation of p38 and of JNK, as can be seen from their dual phosphorylation upon treatment (Fig. 6).
Unlike with RB + light and from DHPNO2, there is no increase of p38 or JNK activity with the hydrophilic NDPO2 tested at concentrations of 3–10 mm for incubation times of 30 min. Thus, 1O2 has to be generated intracellularly, by either RB + light or DHPNO2, for activation of p38 and JNK.
MAPK activation pattern
In this study, we demonstrate that UVA and 1O2 are capable of enhancing cellular p38 and JNK but not ERK activity (Figs 1 and 5). So far, this pattern of MAPK activation may be regarded as unique for UVA and 1O2, and distinct from reactive oxygen species such as hydroperoxides [27,28], superoxide generated by redox cycling , nitric oxide and related species , or species generated during treatment with UVC  or UVB , which also activate ERKs. The UVA/1O2 pattern has also been found in murine keratinocytes treated with benzoporphyrin derivative and light  and human keratinocytes treated with 5-aminolevulinate-derived porphyrins and light .
Topology of p38 and JNK activation by 1O2
It has been shown previously that the activation of JNK by UVA is mediated by 1O2. Here we show that the same may apply for p38 (Fig. 2). The 1O2 quenchers imidazole and azide, but not the hydroxyl radical scavengers mannitol or dimethylsulfoxide, inhibit the phosphorylation of p38 upon irradiation with UVA. However, phosphorylation upon irradiation is not enhanced when cells are irradiated in the presence of D2O, an enhancer of the lifetime of 1O2. It has been pointed out by Foote and Clennan  that solvent isotope effect ratios may vary from as high as 14 to 1.0, depending on the concentration of the substrate of the reaction with 1O2. The higher the concentration of the target of 1O2, the lower the isotope effect. Thus, even with no detectable D2O effect, 1O2 could act as a mediator of UVA activation of p38 if the primary target of 1O2 is different from that leading to the activation of JNK.
The activation of p38 and JNK can be elicited by 1O2 generated intracellularly but not extracellularly (Fig. 6). Intracellular 1O2 generation may be by UVA via endogenous photosensitizers such as porphyrins, flavins or certain quinones , by irradiation of membrane-permeant photosensitizers such as RB, or by employing membrane-permeant 1O2 carriers such as DHPNO2.
On the mechanism of activation
Many genes are induced upon treatment of cells with UVA . Regarding a possible mechanism, from the data described above it is concluded that gene expression upon exposure to UVA may be exerted not only via AP-2  or NF-κB , but also via the p38 and JNK pathways, mediated by 1O2 generated intracellularly. p38 and JNK, once activated, phosphorylate, and thereby activate, transcription factors such as c-Jun and ATF-2 and ternary complex factors such as Elk-1 [13,14], resulting in the activation of c-fos and c-jun transcription to form the AP-1 proteins, Fos and Jun. Although many of the UVA-induced genes contain functional AP-1 sites in their promoters, such as MMP-1  or HO-1 , evidence for transcriptional activation by UVA/1O2 via p38 or JNK and AP-1 is still missing.
Furthermore, it can be concluded that gene expression effects of UVA that were previously shown to be mimicked by NDPO2, such as the induction of ICAM-1 , MMP-1  or the interleukins IL-1 and IL-6 , may not be mediated by the MAPKs studied here. A pathway independent of AP-1 has indeed been identified, involving AP-2 for ICAM-1 .
Regarding the mechanism of activation of JNK and p38 by UVA or 1O2, it is known that the activation of JNK by sodium arsenite, a potent inducer of AP-1, is via inactivation of a JNK phosphatase . The tyrosine phosphatase inhibitor orthovanadate leads to an activation of JNK that can not be enhanced by additional UVA treatment (L.-O. Klotz et al., unpublished results). All protein tyrosine phosphatases known so far rely on the presence in their active site of a cysteine that serves as a nucleophile to accept the phosphate moiety of the phosphatase substrate, forming an intermediate phosphocysteine . Oxidation of this cysteine residue, e.g. by 1O2 generated during UVA irradiation, would lead to the inactivation of the phosphatase, thereby allowing kinase activities to become predominant. UVA/1O2 are known thiol depletors [37,38], and they may act on MAPKs by inactivating a corresponding phosphatase. In this regard, the inactivation of a membrane-bound phosphatase has been shown to be responsible for net phosphorylation and activation of membrane-bound receptors targeted by UVA, UVB and UVC .
In summary, activation of p38 and JNK MAPKs is a part of the UVA and 1O2 stress response of human skin fibroblasts. The pattern of p38 and JNK, but not the ERK MAP kinases, being activated by UVA/1O2 may so far be regarded as distinct from other reactive oxygen species that also activate ERKs.
We thank Dr Freimut Schliess for the ERK kinase assays and Annette Reimann for expert technical assistance. This study was supported by the Deutsche Forschungsgemeinschaft, SFB 503, Project B1.